diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,35153 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 15.0,
+  "global_step": 585540,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99914608737234e-05,
+      "loss": 8.1433,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982921747446805e-05,
+      "loss": 5.3387,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99743826211702e-05,
+      "loss": 4.6505,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99658434948936e-05,
+      "loss": 4.5935,
+      "step": 400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9957304368617e-05,
+      "loss": 4.1372,
+      "step": 500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9948765242340405e-05,
+      "loss": 4.2621,
+      "step": 600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9940226116063806e-05,
+      "loss": 4.2878,
+      "step": 700
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.993168698978721e-05,
+      "loss": 4.3196,
+      "step": 800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99231478635106e-05,
+      "loss": 4.0911,
+      "step": 900
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991460873723401e-05,
+      "loss": 4.2543,
+      "step": 1000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9906069610957406e-05,
+      "loss": 4.2923,
+      "step": 1100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.989753048468081e-05,
+      "loss": 4.3948,
+      "step": 1200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.988899135840421e-05,
+      "loss": 4.1853,
+      "step": 1300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.988045223212761e-05,
+      "loss": 3.8924,
+      "step": 1400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.987191310585101e-05,
+      "loss": 4.4658,
+      "step": 1500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9863373979574414e-05,
+      "loss": 3.9246,
+      "step": 1600
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.985483485329781e-05,
+      "loss": 4.2261,
+      "step": 1700
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.984629572702122e-05,
+      "loss": 4.5428,
+      "step": 1800
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.983775660074461e-05,
+      "loss": 3.9668,
+      "step": 1900
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.982921747446801e-05,
+      "loss": 4.412,
+      "step": 2000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9820678348191415e-05,
+      "loss": 4.2224,
+      "step": 2100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9812139221914816e-05,
+      "loss": 4.4448,
+      "step": 2200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.980360009563821e-05,
+      "loss": 4.1606,
+      "step": 2300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.979506096936162e-05,
+      "loss": 4.335,
+      "step": 2400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9786521843085014e-05,
+      "loss": 3.945,
+      "step": 2500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.977798271680842e-05,
+      "loss": 4.0949,
+      "step": 2600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.976944359053182e-05,
+      "loss": 4.4423,
+      "step": 2700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.976090446425522e-05,
+      "loss": 3.9127,
+      "step": 2800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.975236533797862e-05,
+      "loss": 3.8504,
+      "step": 2900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.974382621170202e-05,
+      "loss": 4.1752,
+      "step": 3000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9735287085425417e-05,
+      "loss": 4.2515,
+      "step": 3100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9726747959148825e-05,
+      "loss": 4.1318,
+      "step": 3200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.971820883287222e-05,
+      "loss": 3.993,
+      "step": 3300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.970966970659562e-05,
+      "loss": 3.8842,
+      "step": 3400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.970113058031902e-05,
+      "loss": 4.1074,
+      "step": 3500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9692591454042424e-05,
+      "loss": 4.203,
+      "step": 3600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9684052327765826e-05,
+      "loss": 3.884,
+      "step": 3700
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.967551320148923e-05,
+      "loss": 3.9207,
+      "step": 3800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.966697407521262e-05,
+      "loss": 3.9057,
+      "step": 3900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.965843494893603e-05,
+      "loss": 4.1704,
+      "step": 4000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9649895822659425e-05,
+      "loss": 3.9624,
+      "step": 4100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.964135669638283e-05,
+      "loss": 4.2685,
+      "step": 4200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.963281757010623e-05,
+      "loss": 4.0752,
+      "step": 4300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.962427844382963e-05,
+      "loss": 4.2952,
+      "step": 4400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.961573931755303e-05,
+      "loss": 3.9692,
+      "step": 4500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.960720019127643e-05,
+      "loss": 4.1518,
+      "step": 4600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.959866106499983e-05,
+      "loss": 4.0681,
+      "step": 4700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9590121938723236e-05,
+      "loss": 4.0123,
+      "step": 4800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.958158281244663e-05,
+      "loss": 4.1172,
+      "step": 4900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.957304368617003e-05,
+      "loss": 3.9204,
+      "step": 5000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9564504559893434e-05,
+      "loss": 4.104,
+      "step": 5100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9555965433616835e-05,
+      "loss": 4.1304,
+      "step": 5200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.954742630734023e-05,
+      "loss": 4.1386,
+      "step": 5300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.953888718106364e-05,
+      "loss": 4.0127,
+      "step": 5400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.953034805478703e-05,
+      "loss": 3.9139,
+      "step": 5500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9521808928510435e-05,
+      "loss": 3.9059,
+      "step": 5600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9513269802233836e-05,
+      "loss": 3.8485,
+      "step": 5700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.950473067595724e-05,
+      "loss": 4.0259,
+      "step": 5800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.949619154968064e-05,
+      "loss": 3.9548,
+      "step": 5900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.948765242340404e-05,
+      "loss": 4.1649,
+      "step": 6000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9479113297127436e-05,
+      "loss": 3.9737,
+      "step": 6100
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9470574170850844e-05,
+      "loss": 4.2423,
+      "step": 6200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.946203504457424e-05,
+      "loss": 4.0339,
+      "step": 6300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.945349591829764e-05,
+      "loss": 4.0146,
+      "step": 6400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.944495679202104e-05,
+      "loss": 4.097,
+      "step": 6500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9436417665744444e-05,
+      "loss": 3.8055,
+      "step": 6600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9427878539467845e-05,
+      "loss": 4.1866,
+      "step": 6700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.941933941319125e-05,
+      "loss": 4.2445,
+      "step": 6800
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.941080028691464e-05,
+      "loss": 3.9666,
+      "step": 6900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.940226116063805e-05,
+      "loss": 3.9317,
+      "step": 7000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9393722034361445e-05,
+      "loss": 3.6663,
+      "step": 7100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9385182908084846e-05,
+      "loss": 3.7571,
+      "step": 7200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.937664378180825e-05,
+      "loss": 3.772,
+      "step": 7300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.936810465553165e-05,
+      "loss": 3.5315,
+      "step": 7400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.935956552925505e-05,
+      "loss": 3.9508,
+      "step": 7500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.935102640297845e-05,
+      "loss": 4.0053,
+      "step": 7600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.934248727670185e-05,
+      "loss": 3.8292,
+      "step": 7700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9333948150425255e-05,
+      "loss": 3.9688,
+      "step": 7800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.932540902414865e-05,
+      "loss": 3.8298,
+      "step": 7900
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.931686989787205e-05,
+      "loss": 3.8508,
+      "step": 8000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.930833077159545e-05,
+      "loss": 3.6922,
+      "step": 8100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9299791645318855e-05,
+      "loss": 3.8841,
+      "step": 8200
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.929125251904225e-05,
+      "loss": 3.7595,
+      "step": 8300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.928271339276566e-05,
+      "loss": 3.6968,
+      "step": 8400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.927417426648905e-05,
+      "loss": 3.7239,
+      "step": 8500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9265635140212454e-05,
+      "loss": 3.7007,
+      "step": 8600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9257096013935856e-05,
+      "loss": 3.762,
+      "step": 8700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.924855688765926e-05,
+      "loss": 3.912,
+      "step": 8800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.924001776138266e-05,
+      "loss": 3.8828,
+      "step": 8900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.923147863510606e-05,
+      "loss": 3.7351,
+      "step": 9000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9222939508829455e-05,
+      "loss": 3.7754,
+      "step": 9100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9214400382552864e-05,
+      "loss": 4.0024,
+      "step": 9200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.920586125627626e-05,
+      "loss": 3.946,
+      "step": 9300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.919732212999966e-05,
+      "loss": 3.7708,
+      "step": 9400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.918878300372306e-05,
+      "loss": 3.7317,
+      "step": 9500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.918024387744646e-05,
+      "loss": 3.9509,
+      "step": 9600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9171704751169864e-05,
+      "loss": 3.5076,
+      "step": 9700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9163165624893266e-05,
+      "loss": 3.9101,
+      "step": 9800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.915462649861666e-05,
+      "loss": 3.9417,
+      "step": 9900
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.914608737234007e-05,
+      "loss": 3.9119,
+      "step": 10000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9137548246063464e-05,
+      "loss": 3.9257,
+      "step": 10100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9129009119786865e-05,
+      "loss": 4.0386,
+      "step": 10200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.912046999351027e-05,
+      "loss": 3.942,
+      "step": 10300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.911193086723367e-05,
+      "loss": 3.7424,
+      "step": 10400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.910339174095706e-05,
+      "loss": 3.5835,
+      "step": 10500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.909485261468047e-05,
+      "loss": 3.9946,
+      "step": 10600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9086313488403866e-05,
+      "loss": 3.9583,
+      "step": 10700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.907777436212727e-05,
+      "loss": 3.6759,
+      "step": 10800
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.906923523585067e-05,
+      "loss": 3.8039,
+      "step": 10900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.906069610957407e-05,
+      "loss": 3.8028,
+      "step": 11000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.905215698329747e-05,
+      "loss": 3.6921,
+      "step": 11100
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9043617857020874e-05,
+      "loss": 3.5227,
+      "step": 11200
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.903507873074427e-05,
+      "loss": 3.6822,
+      "step": 11300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.902653960446768e-05,
+      "loss": 3.5773,
+      "step": 11400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.901800047819107e-05,
+      "loss": 3.7838,
+      "step": 11500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9009461351914474e-05,
+      "loss": 3.7591,
+      "step": 11600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9000922225637875e-05,
+      "loss": 3.8082,
+      "step": 11700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.899238309936128e-05,
+      "loss": 3.7714,
+      "step": 11800
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.898384397308468e-05,
+      "loss": 3.7589,
+      "step": 11900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.897530484680808e-05,
+      "loss": 3.8622,
+      "step": 12000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8966765720531475e-05,
+      "loss": 3.7458,
+      "step": 12100
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.895822659425488e-05,
+      "loss": 3.5666,
+      "step": 12200
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.894968746797828e-05,
+      "loss": 3.773,
+      "step": 12300
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.894114834170168e-05,
+      "loss": 3.792,
+      "step": 12400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.893260921542508e-05,
+      "loss": 3.8636,
+      "step": 12500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.892407008914848e-05,
+      "loss": 3.9016,
+      "step": 12600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8915530962871884e-05,
+      "loss": 3.5966,
+      "step": 12700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8906991836595285e-05,
+      "loss": 3.7381,
+      "step": 12800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.889845271031868e-05,
+      "loss": 3.8202,
+      "step": 12900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.888991358404209e-05,
+      "loss": 3.7938,
+      "step": 13000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.888137445776548e-05,
+      "loss": 3.668,
+      "step": 13100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8872835331488885e-05,
+      "loss": 3.598,
+      "step": 13200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8864296205212286e-05,
+      "loss": 3.8545,
+      "step": 13300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.885575707893569e-05,
+      "loss": 3.7405,
+      "step": 13400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.884721795265908e-05,
+      "loss": 3.7744,
+      "step": 13500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.883867882638249e-05,
+      "loss": 3.7296,
+      "step": 13600
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8830139700105886e-05,
+      "loss": 3.7762,
+      "step": 13700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.882160057382929e-05,
+      "loss": 3.6065,
+      "step": 13800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.881306144755269e-05,
+      "loss": 3.8279,
+      "step": 13900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.880452232127609e-05,
+      "loss": 3.7685,
+      "step": 14000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.879598319499949e-05,
+      "loss": 3.7034,
+      "step": 14100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8787444068722893e-05,
+      "loss": 3.963,
+      "step": 14200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.877890494244629e-05,
+      "loss": 3.6013,
+      "step": 14300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8770365816169697e-05,
+      "loss": 3.6733,
+      "step": 14400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.876182668989309e-05,
+      "loss": 3.4549,
+      "step": 14500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.875328756361649e-05,
+      "loss": 3.5551,
+      "step": 14600
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8744748437339894e-05,
+      "loss": 3.6137,
+      "step": 14700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8736209311063296e-05,
+      "loss": 3.9688,
+      "step": 14800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.87276701847867e-05,
+      "loss": 3.7047,
+      "step": 14900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.87191310585101e-05,
+      "loss": 3.5083,
+      "step": 15000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8710591932233494e-05,
+      "loss": 3.7918,
+      "step": 15100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.87020528059569e-05,
+      "loss": 3.7162,
+      "step": 15200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.86935136796803e-05,
+      "loss": 3.6648,
+      "step": 15300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.86849745534037e-05,
+      "loss": 3.5669,
+      "step": 15400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.86764354271271e-05,
+      "loss": 3.659,
+      "step": 15500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.86678963008505e-05,
+      "loss": 3.9466,
+      "step": 15600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8659357174573896e-05,
+      "loss": 3.4854,
+      "step": 15700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8650818048297305e-05,
+      "loss": 3.7612,
+      "step": 15800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.86422789220207e-05,
+      "loss": 3.7588,
+      "step": 15900
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.86337397957441e-05,
+      "loss": 3.4317,
+      "step": 16000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.86252006694675e-05,
+      "loss": 3.7622,
+      "step": 16100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8616661543190904e-05,
+      "loss": 3.5345,
+      "step": 16200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8608122416914306e-05,
+      "loss": 3.7285,
+      "step": 16300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.859958329063771e-05,
+      "loss": 3.5887,
+      "step": 16400
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.85910441643611e-05,
+      "loss": 3.4669,
+      "step": 16500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.858250503808451e-05,
+      "loss": 3.535,
+      "step": 16600
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.8573965911807905e-05,
+      "loss": 3.6845,
+      "step": 16700
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.856542678553131e-05,
+      "loss": 3.6075,
+      "step": 16800
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.855688765925471e-05,
+      "loss": 3.7537,
+      "step": 16900
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.854834853297811e-05,
+      "loss": 3.6431,
+      "step": 17000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.853980940670151e-05,
+      "loss": 3.4635,
+      "step": 17100
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.853127028042491e-05,
+      "loss": 3.5606,
+      "step": 17200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.852273115414831e-05,
+      "loss": 3.9068,
+      "step": 17300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.8514192027871716e-05,
+      "loss": 3.6972,
+      "step": 17400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.850565290159511e-05,
+      "loss": 3.6389,
+      "step": 17500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.849711377531851e-05,
+      "loss": 3.6478,
+      "step": 17600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.8488574649041914e-05,
+      "loss": 3.8012,
+      "step": 17700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.848003552276531e-05,
+      "loss": 3.5439,
+      "step": 17800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.847149639648872e-05,
+      "loss": 3.44,
+      "step": 17900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.846295727021211e-05,
+      "loss": 3.6266,
+      "step": 18000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.845441814393551e-05,
+      "loss": 3.2828,
+      "step": 18100
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.8445879017658915e-05,
+      "loss": 3.2803,
+      "step": 18200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.8437339891382316e-05,
+      "loss": 3.7721,
+      "step": 18300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.842880076510571e-05,
+      "loss": 3.7414,
+      "step": 18400
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.842026163882912e-05,
+      "loss": 3.6759,
+      "step": 18500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.8411722512552514e-05,
+      "loss": 3.6668,
+      "step": 18600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.8403183386275916e-05,
+      "loss": 3.6371,
+      "step": 18700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.839464425999932e-05,
+      "loss": 3.3712,
+      "step": 18800
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.838610513372272e-05,
+      "loss": 3.5995,
+      "step": 18900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.837756600744612e-05,
+      "loss": 3.6567,
+      "step": 19000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.836902688116952e-05,
+      "loss": 3.626,
+      "step": 19100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.836048775489292e-05,
+      "loss": 3.7768,
+      "step": 19200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.8351948628616325e-05,
+      "loss": 3.2899,
+      "step": 19300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.834340950233972e-05,
+      "loss": 3.6196,
+      "step": 19400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.833487037606312e-05,
+      "loss": 3.4995,
+      "step": 19500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.832633124978652e-05,
+      "loss": 3.5778,
+      "step": 19600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.8317792123509924e-05,
+      "loss": 3.6503,
+      "step": 19700
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8309252997233326e-05,
+      "loss": 3.5208,
+      "step": 19800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.830071387095673e-05,
+      "loss": 3.5284,
+      "step": 19900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.829217474468012e-05,
+      "loss": 3.4265,
+      "step": 20000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.828363561840353e-05,
+      "loss": 3.597,
+      "step": 20100
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8275096492126925e-05,
+      "loss": 3.7741,
+      "step": 20200
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.826655736585033e-05,
+      "loss": 3.8414,
+      "step": 20300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.825801823957373e-05,
+      "loss": 3.5326,
+      "step": 20400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.824947911329713e-05,
+      "loss": 3.7442,
+      "step": 20500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8240939987020525e-05,
+      "loss": 3.5901,
+      "step": 20600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.823240086074393e-05,
+      "loss": 3.8139,
+      "step": 20700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.822386173446733e-05,
+      "loss": 3.5049,
+      "step": 20800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.821532260819073e-05,
+      "loss": 3.4308,
+      "step": 20900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.820678348191413e-05,
+      "loss": 3.5181,
+      "step": 21000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.819824435563753e-05,
+      "loss": 3.6863,
+      "step": 21100
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8189705229360934e-05,
+      "loss": 3.4282,
+      "step": 21200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.8181166103084336e-05,
+      "loss": 3.4732,
+      "step": 21300
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.817262697680773e-05,
+      "loss": 3.6055,
+      "step": 21400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.816408785053114e-05,
+      "loss": 3.4187,
+      "step": 21500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.8155548724254534e-05,
+      "loss": 3.4109,
+      "step": 21600
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.8147009597977935e-05,
+      "loss": 3.5652,
+      "step": 21700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.813847047170134e-05,
+      "loss": 3.5527,
+      "step": 21800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.812993134542474e-05,
+      "loss": 3.5741,
+      "step": 21900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.812139221914814e-05,
+      "loss": 3.6138,
+      "step": 22000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.811285309287154e-05,
+      "loss": 3.5562,
+      "step": 22100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.8104313966594936e-05,
+      "loss": 3.5844,
+      "step": 22200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.8095774840318344e-05,
+      "loss": 3.4481,
+      "step": 22300
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.808723571404174e-05,
+      "loss": 3.547,
+      "step": 22400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.807869658776514e-05,
+      "loss": 3.5046,
+      "step": 22500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.807015746148854e-05,
+      "loss": 3.5041,
+      "step": 22600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.8061618335211944e-05,
+      "loss": 3.5439,
+      "step": 22700
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.8053079208935345e-05,
+      "loss": 3.5554,
+      "step": 22800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.804454008265875e-05,
+      "loss": 3.517,
+      "step": 22900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.803600095638214e-05,
+      "loss": 3.5763,
+      "step": 23000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.802746183010555e-05,
+      "loss": 3.5727,
+      "step": 23100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8018922703828945e-05,
+      "loss": 3.6164,
+      "step": 23200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8010383577552346e-05,
+      "loss": 3.5565,
+      "step": 23300
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.800184445127575e-05,
+      "loss": 3.2922,
+      "step": 23400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.799330532499915e-05,
+      "loss": 3.4609,
+      "step": 23500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7984766198722544e-05,
+      "loss": 3.6773,
+      "step": 23600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.797622707244595e-05,
+      "loss": 3.4789,
+      "step": 23700
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.796768794616935e-05,
+      "loss": 3.9325,
+      "step": 23800
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.795914881989275e-05,
+      "loss": 3.3577,
+      "step": 23900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.795060969361615e-05,
+      "loss": 3.4149,
+      "step": 24000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.794207056733955e-05,
+      "loss": 3.6172,
+      "step": 24100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.7933531441062953e-05,
+      "loss": 3.4819,
+      "step": 24200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.7924992314786355e-05,
+      "loss": 3.4195,
+      "step": 24300
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.791645318850975e-05,
+      "loss": 3.3149,
+      "step": 24400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.790791406223316e-05,
+      "loss": 3.3554,
+      "step": 24500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.789937493595655e-05,
+      "loss": 3.5358,
+      "step": 24600
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.7890835809679954e-05,
+      "loss": 3.4567,
+      "step": 24700
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.7882296683403356e-05,
+      "loss": 3.5939,
+      "step": 24800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.787375755712676e-05,
+      "loss": 3.3348,
+      "step": 24900
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.786521843085016e-05,
+      "loss": 3.5004,
+      "step": 25000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.785667930457356e-05,
+      "loss": 3.5305,
+      "step": 25100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.7848140178296955e-05,
+      "loss": 3.3193,
+      "step": 25200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.7839601052020364e-05,
+      "loss": 3.4184,
+      "step": 25300
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.783106192574376e-05,
+      "loss": 3.4644,
+      "step": 25400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.782252279946716e-05,
+      "loss": 3.5732,
+      "step": 25500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.781398367319056e-05,
+      "loss": 3.2855,
+      "step": 25600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.780544454691396e-05,
+      "loss": 3.464,
+      "step": 25700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.779690542063736e-05,
+      "loss": 3.5029,
+      "step": 25800
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.7788366294360766e-05,
+      "loss": 3.4364,
+      "step": 25900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.777982716808416e-05,
+      "loss": 3.3555,
+      "step": 26000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.777128804180756e-05,
+      "loss": 3.6529,
+      "step": 26100
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.7762748915530964e-05,
+      "loss": 3.537,
+      "step": 26200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.7754209789254366e-05,
+      "loss": 3.4883,
+      "step": 26300
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.774567066297777e-05,
+      "loss": 3.6947,
+      "step": 26400
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.773713153670117e-05,
+      "loss": 3.457,
+      "step": 26500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7728592410424563e-05,
+      "loss": 3.4468,
+      "step": 26600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.772005328414797e-05,
+      "loss": 3.5284,
+      "step": 26700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7711514157871367e-05,
+      "loss": 3.221,
+      "step": 26800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.770297503159477e-05,
+      "loss": 3.4848,
+      "step": 26900
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.769443590531817e-05,
+      "loss": 3.4874,
+      "step": 27000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.768589677904157e-05,
+      "loss": 3.435,
+      "step": 27100
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.767735765276497e-05,
+      "loss": 3.5241,
+      "step": 27200
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.7668818526488374e-05,
+      "loss": 3.6542,
+      "step": 27300
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.766027940021177e-05,
+      "loss": 3.3834,
+      "step": 27400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.765174027393518e-05,
+      "loss": 3.6834,
+      "step": 27500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.764320114765857e-05,
+      "loss": 3.2529,
+      "step": 27600
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7634662021381974e-05,
+      "loss": 3.3623,
+      "step": 27700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7626122895105375e-05,
+      "loss": 3.4693,
+      "step": 27800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.761758376882878e-05,
+      "loss": 3.6151,
+      "step": 27900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.760904464255218e-05,
+      "loss": 3.4634,
+      "step": 28000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.760050551627558e-05,
+      "loss": 3.2379,
+      "step": 28100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.7591966389998975e-05,
+      "loss": 3.5468,
+      "step": 28200
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.758342726372238e-05,
+      "loss": 3.3572,
+      "step": 28300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.757488813744578e-05,
+      "loss": 3.4741,
+      "step": 28400
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.756634901116918e-05,
+      "loss": 3.5941,
+      "step": 28500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.755780988489258e-05,
+      "loss": 3.2845,
+      "step": 28600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.754927075861598e-05,
+      "loss": 3.3763,
+      "step": 28700
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.754073163233938e-05,
+      "loss": 3.4421,
+      "step": 28800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.7532192506062786e-05,
+      "loss": 3.3501,
+      "step": 28900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.752365337978618e-05,
+      "loss": 3.4032,
+      "step": 29000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.751511425350958e-05,
+      "loss": 3.1186,
+      "step": 29100
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.7506575127232983e-05,
+      "loss": 3.5095,
+      "step": 29200
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.7498036000956385e-05,
+      "loss": 3.1795,
+      "step": 29300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.7489496874679787e-05,
+      "loss": 3.5747,
+      "step": 29400
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.748095774840319e-05,
+      "loss": 3.6188,
+      "step": 29500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.747241862212658e-05,
+      "loss": 3.3566,
+      "step": 29600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.746387949584999e-05,
+      "loss": 3.547,
+      "step": 29700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.7455340369573386e-05,
+      "loss": 3.1518,
+      "step": 29800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.744680124329679e-05,
+      "loss": 3.466,
+      "step": 29900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.743826211702019e-05,
+      "loss": 3.2771,
+      "step": 30000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.742972299074359e-05,
+      "loss": 3.5729,
+      "step": 30100
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.742118386446699e-05,
+      "loss": 3.4516,
+      "step": 30200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7412644738190394e-05,
+      "loss": 3.3993,
+      "step": 30300
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.740410561191379e-05,
+      "loss": 3.4406,
+      "step": 30400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.73955664856372e-05,
+      "loss": 3.5511,
+      "step": 30500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.738702735936059e-05,
+      "loss": 3.2725,
+      "step": 30600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.737848823308399e-05,
+      "loss": 3.4044,
+      "step": 30700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7369949106807395e-05,
+      "loss": 3.6242,
+      "step": 30800
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7361409980530796e-05,
+      "loss": 3.6762,
+      "step": 30900
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.735287085425419e-05,
+      "loss": 3.4278,
+      "step": 31000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.73443317279776e-05,
+      "loss": 3.3284,
+      "step": 31100
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7335792601700994e-05,
+      "loss": 3.5034,
+      "step": 31200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.73272534754244e-05,
+      "loss": 3.4058,
+      "step": 31300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.73187143491478e-05,
+      "loss": 3.3646,
+      "step": 31400
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.73101752228712e-05,
+      "loss": 3.4112,
+      "step": 31500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.73016360965946e-05,
+      "loss": 3.475,
+      "step": 31600
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7293096970318e-05,
+      "loss": 3.6318,
+      "step": 31700
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7284557844041397e-05,
+      "loss": 3.2754,
+      "step": 31800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7276018717764805e-05,
+      "loss": 3.3146,
+      "step": 31900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.72674795914882e-05,
+      "loss": 3.37,
+      "step": 32000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.72589404652116e-05,
+      "loss": 3.4869,
+      "step": 32100
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7250401338935e-05,
+      "loss": 3.4892,
+      "step": 32200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7241862212658404e-05,
+      "loss": 3.217,
+      "step": 32300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7233323086381806e-05,
+      "loss": 3.4166,
+      "step": 32400
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.722478396010521e-05,
+      "loss": 3.4837,
+      "step": 32500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.72162448338286e-05,
+      "loss": 3.399,
+      "step": 32600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.720770570755201e-05,
+      "loss": 3.6814,
+      "step": 32700
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7199166581275405e-05,
+      "loss": 3.5863,
+      "step": 32800
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.719062745499881e-05,
+      "loss": 3.3289,
+      "step": 32900
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.718208832872221e-05,
+      "loss": 3.2249,
+      "step": 33000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.717354920244561e-05,
+      "loss": 3.6567,
+      "step": 33100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.716501007616901e-05,
+      "loss": 3.471,
+      "step": 33200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.715647094989241e-05,
+      "loss": 3.2949,
+      "step": 33300
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.714793182361581e-05,
+      "loss": 3.3271,
+      "step": 33400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7139392697339216e-05,
+      "loss": 3.2935,
+      "step": 33500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.713085357106261e-05,
+      "loss": 3.4567,
+      "step": 33600
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.712231444478601e-05,
+      "loss": 3.5018,
+      "step": 33700
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7113775318509414e-05,
+      "loss": 3.462,
+      "step": 33800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7105236192232816e-05,
+      "loss": 3.4016,
+      "step": 33900
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.709669706595621e-05,
+      "loss": 3.3442,
+      "step": 34000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.708815793967962e-05,
+      "loss": 3.4873,
+      "step": 34100
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7079618813403013e-05,
+      "loss": 3.3928,
+      "step": 34200
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7071079687126415e-05,
+      "loss": 3.6974,
+      "step": 34300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7062540560849816e-05,
+      "loss": 3.316,
+      "step": 34400
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.705400143457322e-05,
+      "loss": 3.3236,
+      "step": 34500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.704546230829662e-05,
+      "loss": 3.53,
+      "step": 34600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.703692318202002e-05,
+      "loss": 3.5239,
+      "step": 34700
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7028384055743416e-05,
+      "loss": 3.1612,
+      "step": 34800
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7019844929466824e-05,
+      "loss": 3.3604,
+      "step": 34900
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.701130580319022e-05,
+      "loss": 3.259,
+      "step": 35000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.700276667691362e-05,
+      "loss": 3.3522,
+      "step": 35100
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.699422755063702e-05,
+      "loss": 3.3189,
+      "step": 35200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.698568842436042e-05,
+      "loss": 3.457,
+      "step": 35300
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6977149298083825e-05,
+      "loss": 3.3757,
+      "step": 35400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.696861017180722e-05,
+      "loss": 3.3143,
+      "step": 35500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.696007104553062e-05,
+      "loss": 3.329,
+      "step": 35600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.695153191925402e-05,
+      "loss": 3.4637,
+      "step": 35700
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6942992792977425e-05,
+      "loss": 3.3483,
+      "step": 35800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.693445366670082e-05,
+      "loss": 3.4662,
+      "step": 35900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.692591454042423e-05,
+      "loss": 3.2127,
+      "step": 36000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.691737541414762e-05,
+      "loss": 3.2416,
+      "step": 36100
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.690883628787103e-05,
+      "loss": 3.1665,
+      "step": 36200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6900297161594426e-05,
+      "loss": 3.371,
+      "step": 36300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.689175803531783e-05,
+      "loss": 3.2884,
+      "step": 36400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.688321890904123e-05,
+      "loss": 3.3204,
+      "step": 36500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.687467978276463e-05,
+      "loss": 3.2915,
+      "step": 36600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6866140656488025e-05,
+      "loss": 3.2848,
+      "step": 36700
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.685760153021143e-05,
+      "loss": 3.2604,
+      "step": 36800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.684906240393483e-05,
+      "loss": 3.1758,
+      "step": 36900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.684052327765823e-05,
+      "loss": 3.3302,
+      "step": 37000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.683198415138163e-05,
+      "loss": 3.5299,
+      "step": 37100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.682344502510503e-05,
+      "loss": 3.3508,
+      "step": 37200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6814905898828434e-05,
+      "loss": 3.3845,
+      "step": 37300
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6806366772551836e-05,
+      "loss": 3.4054,
+      "step": 37400
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.679782764627523e-05,
+      "loss": 3.4561,
+      "step": 37500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.678928851999864e-05,
+      "loss": 3.4069,
+      "step": 37600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6780749393722034e-05,
+      "loss": 3.2533,
+      "step": 37700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6772210267445435e-05,
+      "loss": 3.4234,
+      "step": 37800
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.676367114116884e-05,
+      "loss": 3.1182,
+      "step": 37900
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.675513201489224e-05,
+      "loss": 3.2555,
+      "step": 38000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.674659288861564e-05,
+      "loss": 3.5934,
+      "step": 38100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.673805376233904e-05,
+      "loss": 3.3511,
+      "step": 38200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6729514636062436e-05,
+      "loss": 3.3294,
+      "step": 38300
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6720975509785845e-05,
+      "loss": 3.2316,
+      "step": 38400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.671243638350924e-05,
+      "loss": 3.2372,
+      "step": 38500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.670389725723264e-05,
+      "loss": 3.43,
+      "step": 38600
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.669535813095604e-05,
+      "loss": 3.2801,
+      "step": 38700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6686819004679444e-05,
+      "loss": 3.4245,
+      "step": 38800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.667827987840284e-05,
+      "loss": 3.3985,
+      "step": 38900
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.666974075212625e-05,
+      "loss": 3.2582,
+      "step": 39000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.666120162584964e-05,
+      "loss": 3.445,
+      "step": 39100
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.665266249957304e-05,
+      "loss": 3.3439,
+      "step": 39200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6644123373296445e-05,
+      "loss": 3.3569,
+      "step": 39300
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6635584247019846e-05,
+      "loss": 3.2744,
+      "step": 39400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.662704512074325e-05,
+      "loss": 3.4693,
+      "step": 39500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.661850599446665e-05,
+      "loss": 3.2883,
+      "step": 39600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6609966868190044e-05,
+      "loss": 3.3045,
+      "step": 39700
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.660142774191345e-05,
+      "loss": 3.4343,
+      "step": 39800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.659288861563685e-05,
+      "loss": 3.2372,
+      "step": 39900
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.658434948936025e-05,
+      "loss": 3.4615,
+      "step": 40000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.657581036308365e-05,
+      "loss": 3.4103,
+      "step": 40100
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.656727123680705e-05,
+      "loss": 3.5328,
+      "step": 40200
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6558732110530454e-05,
+      "loss": 3.4379,
+      "step": 40300
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6550192984253855e-05,
+      "loss": 3.3792,
+      "step": 40400
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.654165385797725e-05,
+      "loss": 3.5508,
+      "step": 40500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.653311473170066e-05,
+      "loss": 3.3502,
+      "step": 40600
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.652457560542405e-05,
+      "loss": 3.4316,
+      "step": 40700
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.6516036479147455e-05,
+      "loss": 3.1812,
+      "step": 40800
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.6507497352870856e-05,
+      "loss": 3.348,
+      "step": 40900
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.649895822659426e-05,
+      "loss": 3.3069,
+      "step": 41000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.649041910031766e-05,
+      "loss": 3.2655,
+      "step": 41100
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.648187997404106e-05,
+      "loss": 3.1583,
+      "step": 41200
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.6473340847764456e-05,
+      "loss": 3.3218,
+      "step": 41300
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.6464801721487864e-05,
+      "loss": 3.4447,
+      "step": 41400
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.645626259521126e-05,
+      "loss": 3.2127,
+      "step": 41500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.644772346893466e-05,
+      "loss": 3.1546,
+      "step": 41600
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.643918434265806e-05,
+      "loss": 3.5458,
+      "step": 41700
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.643064521638146e-05,
+      "loss": 3.3776,
+      "step": 41800
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.642210609010486e-05,
+      "loss": 3.3053,
+      "step": 41900
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.6413566963828266e-05,
+      "loss": 3.4147,
+      "step": 42000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.640502783755166e-05,
+      "loss": 3.308,
+      "step": 42100
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.639648871127506e-05,
+      "loss": 3.1021,
+      "step": 42200
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.6387949584998464e-05,
+      "loss": 3.4419,
+      "step": 42300
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.6379410458721866e-05,
+      "loss": 3.1076,
+      "step": 42400
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.637087133244527e-05,
+      "loss": 3.2782,
+      "step": 42500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.636233220616867e-05,
+      "loss": 3.43,
+      "step": 42600
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.6353793079892064e-05,
+      "loss": 3.4279,
+      "step": 42700
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.634525395361547e-05,
+      "loss": 3.2666,
+      "step": 42800
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.633671482733887e-05,
+      "loss": 3.3548,
+      "step": 42900
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.632817570106227e-05,
+      "loss": 3.378,
+      "step": 43000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.631963657478567e-05,
+      "loss": 3.2853,
+      "step": 43100
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.631109744850907e-05,
+      "loss": 3.3193,
+      "step": 43200
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.630255832223247e-05,
+      "loss": 3.4065,
+      "step": 43300
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.6294019195955875e-05,
+      "loss": 3.2967,
+      "step": 43400
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.628548006967927e-05,
+      "loss": 3.3564,
+      "step": 43500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.627694094340268e-05,
+      "loss": 3.4165,
+      "step": 43600
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.626840181712607e-05,
+      "loss": 3.2119,
+      "step": 43700
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.6259862690849474e-05,
+      "loss": 3.2635,
+      "step": 43800
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.6251323564572875e-05,
+      "loss": 3.2283,
+      "step": 43900
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.624278443829628e-05,
+      "loss": 3.2392,
+      "step": 44000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.623424531201967e-05,
+      "loss": 3.274,
+      "step": 44100
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.622570618574308e-05,
+      "loss": 3.1144,
+      "step": 44200
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.6217167059466475e-05,
+      "loss": 3.2448,
+      "step": 44300
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.6208627933189876e-05,
+      "loss": 3.4071,
+      "step": 44400
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.620008880691328e-05,
+      "loss": 3.2883,
+      "step": 44500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.619154968063668e-05,
+      "loss": 3.3453,
+      "step": 44600
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.618301055436008e-05,
+      "loss": 3.237,
+      "step": 44700
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.617447142808348e-05,
+      "loss": 3.4076,
+      "step": 44800
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.616593230180688e-05,
+      "loss": 3.2343,
+      "step": 44900
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.6157393175530286e-05,
+      "loss": 3.0996,
+      "step": 45000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.614885404925368e-05,
+      "loss": 3.2876,
+      "step": 45100
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.614031492297708e-05,
+      "loss": 3.3522,
+      "step": 45200
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.6131775796700484e-05,
+      "loss": 3.274,
+      "step": 45300
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.6123236670423885e-05,
+      "loss": 3.168,
+      "step": 45400
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.611469754414729e-05,
+      "loss": 3.3319,
+      "step": 45500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.610615841787069e-05,
+      "loss": 3.2365,
+      "step": 45600
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.609761929159408e-05,
+      "loss": 3.208,
+      "step": 45700
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.608908016531749e-05,
+      "loss": 3.0745,
+      "step": 45800
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.6080541039040886e-05,
+      "loss": 3.1878,
+      "step": 45900
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.607200191276429e-05,
+      "loss": 3.3813,
+      "step": 46000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.606346278648769e-05,
+      "loss": 3.2259,
+      "step": 46100
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.605492366021109e-05,
+      "loss": 3.355,
+      "step": 46200
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.604638453393449e-05,
+      "loss": 3.3909,
+      "step": 46300
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.6037845407657894e-05,
+      "loss": 3.2687,
+      "step": 46400
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.602930628138129e-05,
+      "loss": 3.2326,
+      "step": 46500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.60207671551047e-05,
+      "loss": 3.3135,
+      "step": 46600
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.601222802882809e-05,
+      "loss": 3.5051,
+      "step": 46700
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.600368890255149e-05,
+      "loss": 3.009,
+      "step": 46800
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.5995149776274895e-05,
+      "loss": 3.1782,
+      "step": 46900
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.5986610649998296e-05,
+      "loss": 3.0154,
+      "step": 47000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.597807152372169e-05,
+      "loss": 3.3357,
+      "step": 47100
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.59695323974451e-05,
+      "loss": 3.2224,
+      "step": 47200
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.5960993271168494e-05,
+      "loss": 3.1395,
+      "step": 47300
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.5952454144891896e-05,
+      "loss": 3.4303,
+      "step": 47400
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.59439150186153e-05,
+      "loss": 3.1074,
+      "step": 47500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.59353758923387e-05,
+      "loss": 3.3658,
+      "step": 47600
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.59268367660621e-05,
+      "loss": 3.3173,
+      "step": 47700
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.59182976397855e-05,
+      "loss": 2.9707,
+      "step": 47800
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.59097585135089e-05,
+      "loss": 3.3756,
+      "step": 47900
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.5901219387232305e-05,
+      "loss": 3.0612,
+      "step": 48000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.58926802609557e-05,
+      "loss": 3.1551,
+      "step": 48100
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.58841411346791e-05,
+      "loss": 3.3508,
+      "step": 48200
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.58756020084025e-05,
+      "loss": 3.3484,
+      "step": 48300
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.5867062882125904e-05,
+      "loss": 3.2109,
+      "step": 48400
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.5858523755849306e-05,
+      "loss": 3.2551,
+      "step": 48500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.584998462957271e-05,
+      "loss": 3.1801,
+      "step": 48600
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.58414455032961e-05,
+      "loss": 3.2943,
+      "step": 48700
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.583290637701951e-05,
+      "loss": 3.1567,
+      "step": 48800
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.5824367250742905e-05,
+      "loss": 3.6534,
+      "step": 48900
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.581582812446631e-05,
+      "loss": 3.2714,
+      "step": 49000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.580728899818971e-05,
+      "loss": 3.2884,
+      "step": 49100
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.579874987191311e-05,
+      "loss": 3.0826,
+      "step": 49200
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.5790210745636505e-05,
+      "loss": 3.2807,
+      "step": 49300
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.578167161935991e-05,
+      "loss": 3.3775,
+      "step": 49400
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.577313249308331e-05,
+      "loss": 3.0698,
+      "step": 49500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.576459336680671e-05,
+      "loss": 3.299,
+      "step": 49600
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.575605424053011e-05,
+      "loss": 3.1239,
+      "step": 49700
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.574751511425351e-05,
+      "loss": 3.2043,
+      "step": 49800
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.5738975987976914e-05,
+      "loss": 3.3373,
+      "step": 49900
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.5730436861700316e-05,
+      "loss": 3.6232,
+      "step": 50000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.572189773542371e-05,
+      "loss": 3.1563,
+      "step": 50100
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.571335860914712e-05,
+      "loss": 3.2558,
+      "step": 50200
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.5704819482870514e-05,
+      "loss": 3.2061,
+      "step": 50300
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.5696280356593915e-05,
+      "loss": 3.1413,
+      "step": 50400
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.568774123031732e-05,
+      "loss": 3.1439,
+      "step": 50500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.567920210404072e-05,
+      "loss": 3.4335,
+      "step": 50600
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.567066297776412e-05,
+      "loss": 3.3769,
+      "step": 50700
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.566212385148752e-05,
+      "loss": 3.2077,
+      "step": 50800
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.5653584725210916e-05,
+      "loss": 3.2767,
+      "step": 50900
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.5645045598934324e-05,
+      "loss": 3.3502,
+      "step": 51000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.563650647265772e-05,
+      "loss": 3.2704,
+      "step": 51100
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.562796734638112e-05,
+      "loss": 3.3187,
+      "step": 51200
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.561942822010452e-05,
+      "loss": 3.1367,
+      "step": 51300
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.5610889093827924e-05,
+      "loss": 3.3311,
+      "step": 51400
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.5602349967551325e-05,
+      "loss": 3.2221,
+      "step": 51500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.559381084127473e-05,
+      "loss": 3.3227,
+      "step": 51600
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.558527171499812e-05,
+      "loss": 3.2998,
+      "step": 51700
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.557673258872153e-05,
+      "loss": 3.3506,
+      "step": 51800
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.5568193462444925e-05,
+      "loss": 3.3649,
+      "step": 51900
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.5559654336168326e-05,
+      "loss": 3.3714,
+      "step": 52000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.555111520989173e-05,
+      "loss": 3.0725,
+      "step": 52100
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.554257608361513e-05,
+      "loss": 3.2297,
+      "step": 52200
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.5534036957338524e-05,
+      "loss": 3.3677,
+      "step": 52300
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.552549783106193e-05,
+      "loss": 3.2325,
+      "step": 52400
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.551695870478533e-05,
+      "loss": 3.142,
+      "step": 52500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.550841957850873e-05,
+      "loss": 3.1671,
+      "step": 52600
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.549988045223213e-05,
+      "loss": 3.285,
+      "step": 52700
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.5491341325955525e-05,
+      "loss": 3.0934,
+      "step": 52800
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.5482802199678934e-05,
+      "loss": 3.3774,
+      "step": 52900
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.547426307340233e-05,
+      "loss": 3.4387,
+      "step": 53000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.546572394712573e-05,
+      "loss": 3.2799,
+      "step": 53100
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.545718482084913e-05,
+      "loss": 3.4177,
+      "step": 53200
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.544864569457253e-05,
+      "loss": 3.248,
+      "step": 53300
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.5440106568295934e-05,
+      "loss": 3.2275,
+      "step": 53400
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.5431567442019336e-05,
+      "loss": 3.1949,
+      "step": 53500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.542302831574273e-05,
+      "loss": 3.3461,
+      "step": 53600
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.541448918946614e-05,
+      "loss": 3.0244,
+      "step": 53700
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.5405950063189534e-05,
+      "loss": 3.3828,
+      "step": 53800
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.5397410936912935e-05,
+      "loss": 3.2586,
+      "step": 53900
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.538887181063634e-05,
+      "loss": 3.2267,
+      "step": 54000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.538033268435974e-05,
+      "loss": 3.0411,
+      "step": 54100
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.537179355808313e-05,
+      "loss": 3.4893,
+      "step": 54200
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.536325443180654e-05,
+      "loss": 3.1938,
+      "step": 54300
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.5354715305529936e-05,
+      "loss": 3.0352,
+      "step": 54400
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.534617617925334e-05,
+      "loss": 3.1959,
+      "step": 54500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.533763705297674e-05,
+      "loss": 3.2033,
+      "step": 54600
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.532909792670014e-05,
+      "loss": 3.2207,
+      "step": 54700
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.532055880042354e-05,
+      "loss": 3.1523,
+      "step": 54800
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.5312019674146944e-05,
+      "loss": 3.099,
+      "step": 54900
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.530348054787034e-05,
+      "loss": 3.3705,
+      "step": 55000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.529494142159375e-05,
+      "loss": 3.1773,
+      "step": 55100
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.528640229531714e-05,
+      "loss": 3.076,
+      "step": 55200
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.5277863169040544e-05,
+      "loss": 3.1966,
+      "step": 55300
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.5269324042763945e-05,
+      "loss": 3.1723,
+      "step": 55400
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.526078491648735e-05,
+      "loss": 3.3688,
+      "step": 55500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.525224579021075e-05,
+      "loss": 3.241,
+      "step": 55600
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.524370666393415e-05,
+      "loss": 3.3047,
+      "step": 55700
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.5235167537657545e-05,
+      "loss": 2.9904,
+      "step": 55800
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.522662841138095e-05,
+      "loss": 3.1969,
+      "step": 55900
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.521808928510435e-05,
+      "loss": 3.1953,
+      "step": 56000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.520955015882775e-05,
+      "loss": 3.1532,
+      "step": 56100
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.520101103255115e-05,
+      "loss": 3.2045,
+      "step": 56200
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.519247190627455e-05,
+      "loss": 3.2981,
+      "step": 56300
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.5183932779997954e-05,
+      "loss": 3.3037,
+      "step": 56400
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.5175393653721355e-05,
+      "loss": 3.1982,
+      "step": 56500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.516685452744475e-05,
+      "loss": 3.2365,
+      "step": 56600
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.515831540116816e-05,
+      "loss": 3.2118,
+      "step": 56700
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.514977627489155e-05,
+      "loss": 3.2382,
+      "step": 56800
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.5141237148614955e-05,
+      "loss": 3.1484,
+      "step": 56900
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.5132698022338356e-05,
+      "loss": 3.255,
+      "step": 57000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.512415889606176e-05,
+      "loss": 3.3042,
+      "step": 57100
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.511561976978515e-05,
+      "loss": 3.1508,
+      "step": 57200
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.510708064350856e-05,
+      "loss": 3.1394,
+      "step": 57300
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.5098541517231956e-05,
+      "loss": 3.1828,
+      "step": 57400
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.509000239095536e-05,
+      "loss": 3.2893,
+      "step": 57500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.508146326467876e-05,
+      "loss": 2.9791,
+      "step": 57600
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.507292413840216e-05,
+      "loss": 3.1324,
+      "step": 57700
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.506438501212556e-05,
+      "loss": 3.1465,
+      "step": 57800
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.5055845885848963e-05,
+      "loss": 3.172,
+      "step": 57900
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.504730675957236e-05,
+      "loss": 3.2903,
+      "step": 58000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.5038767633295767e-05,
+      "loss": 3.2377,
+      "step": 58100
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.503022850701916e-05,
+      "loss": 3.0679,
+      "step": 58200
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.502168938074256e-05,
+      "loss": 3.3194,
+      "step": 58300
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.5013150254465964e-05,
+      "loss": 3.2289,
+      "step": 58400
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.5004611128189366e-05,
+      "loss": 3.1449,
+      "step": 58500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.499607200191277e-05,
+      "loss": 3.3433,
+      "step": 58600
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.498753287563617e-05,
+      "loss": 3.3692,
+      "step": 58700
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.4978993749359564e-05,
+      "loss": 3.2493,
+      "step": 58800
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.497045462308297e-05,
+      "loss": 3.2536,
+      "step": 58900
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.496191549680637e-05,
+      "loss": 3.2131,
+      "step": 59000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.495337637052977e-05,
+      "loss": 2.9688,
+      "step": 59100
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.494483724425317e-05,
+      "loss": 3.3731,
+      "step": 59200
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.493629811797657e-05,
+      "loss": 3.2793,
+      "step": 59300
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.4927758991699966e-05,
+      "loss": 3.1691,
+      "step": 59400
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.4919219865423375e-05,
+      "loss": 3.3579,
+      "step": 59500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.491068073914677e-05,
+      "loss": 3.3206,
+      "step": 59600
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.490214161287017e-05,
+      "loss": 3.1517,
+      "step": 59700
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.489360248659357e-05,
+      "loss": 3.3422,
+      "step": 59800
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.4885063360316974e-05,
+      "loss": 3.2501,
+      "step": 59900
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.4876524234040376e-05,
+      "loss": 3.0996,
+      "step": 60000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.486798510776378e-05,
+      "loss": 3.4043,
+      "step": 60100
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.485944598148717e-05,
+      "loss": 3.2396,
+      "step": 60200
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.485090685521058e-05,
+      "loss": 3.0377,
+      "step": 60300
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.4842367728933975e-05,
+      "loss": 3.2877,
+      "step": 60400
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.483382860265738e-05,
+      "loss": 3.2888,
+      "step": 60500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.482528947638078e-05,
+      "loss": 3.2973,
+      "step": 60600
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.481675035010418e-05,
+      "loss": 3.2159,
+      "step": 60700
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.480821122382758e-05,
+      "loss": 3.3014,
+      "step": 60800
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.479967209755098e-05,
+      "loss": 3.0788,
+      "step": 60900
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.479113297127438e-05,
+      "loss": 3.1098,
+      "step": 61000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.4782593844997786e-05,
+      "loss": 3.3015,
+      "step": 61100
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.477405471872118e-05,
+      "loss": 3.1981,
+      "step": 61200
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.476551559244458e-05,
+      "loss": 3.1165,
+      "step": 61300
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.4756976466167984e-05,
+      "loss": 3.1384,
+      "step": 61400
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.4748437339891385e-05,
+      "loss": 3.2794,
+      "step": 61500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.473989821361479e-05,
+      "loss": 3.3021,
+      "step": 61600
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.473135908733819e-05,
+      "loss": 3.4168,
+      "step": 61700
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.472281996106158e-05,
+      "loss": 3.1338,
+      "step": 61800
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.471428083478499e-05,
+      "loss": 3.3508,
+      "step": 61900
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.4705741708508386e-05,
+      "loss": 3.1625,
+      "step": 62000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.469720258223179e-05,
+      "loss": 3.3888,
+      "step": 62100
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.468866345595519e-05,
+      "loss": 3.2584,
+      "step": 62200
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.468012432967859e-05,
+      "loss": 3.3002,
+      "step": 62300
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.4671585203401986e-05,
+      "loss": 3.173,
+      "step": 62400
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.4663046077125394e-05,
+      "loss": 3.0448,
+      "step": 62500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.465450695084879e-05,
+      "loss": 3.0071,
+      "step": 62600
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.464596782457219e-05,
+      "loss": 3.3659,
+      "step": 62700
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.463742869829559e-05,
+      "loss": 3.1854,
+      "step": 62800
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.4628889572018993e-05,
+      "loss": 3.3044,
+      "step": 62900
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.4620350445742395e-05,
+      "loss": 3.2739,
+      "step": 63000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.4611811319465797e-05,
+      "loss": 3.2855,
+      "step": 63100
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.460327219318919e-05,
+      "loss": 3.0983,
+      "step": 63200
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.45947330669126e-05,
+      "loss": 3.1033,
+      "step": 63300
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.4586193940635994e-05,
+      "loss": 3.0957,
+      "step": 63400
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.4577654814359396e-05,
+      "loss": 3.0914,
+      "step": 63500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.45691156880828e-05,
+      "loss": 3.1328,
+      "step": 63600
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.45605765618062e-05,
+      "loss": 3.1602,
+      "step": 63700
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.45520374355296e-05,
+      "loss": 3.1097,
+      "step": 63800
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.4543498309253e-05,
+      "loss": 3.0696,
+      "step": 63900
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.45349591829764e-05,
+      "loss": 3.0489,
+      "step": 64000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.4526420056699805e-05,
+      "loss": 3.027,
+      "step": 64100
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.45178809304232e-05,
+      "loss": 3.2089,
+      "step": 64200
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.45093418041466e-05,
+      "loss": 3.1203,
+      "step": 64300
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.450080267787e-05,
+      "loss": 3.0265,
+      "step": 64400
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.4492263551593405e-05,
+      "loss": 3.2565,
+      "step": 64500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.4483724425316806e-05,
+      "loss": 2.9759,
+      "step": 64600
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.447518529904021e-05,
+      "loss": 3.3455,
+      "step": 64700
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.44666461727636e-05,
+      "loss": 3.1068,
+      "step": 64800
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.445810704648701e-05,
+      "loss": 3.1791,
+      "step": 64900
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.4449567920210406e-05,
+      "loss": 3.1656,
+      "step": 65000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.444102879393381e-05,
+      "loss": 3.235,
+      "step": 65100
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.443248966765721e-05,
+      "loss": 3.0505,
+      "step": 65200
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.442395054138061e-05,
+      "loss": 3.2848,
+      "step": 65300
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.4415411415104005e-05,
+      "loss": 3.0479,
+      "step": 65400
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.440687228882741e-05,
+      "loss": 3.2758,
+      "step": 65500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.439833316255081e-05,
+      "loss": 3.0013,
+      "step": 65600
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.438979403627421e-05,
+      "loss": 3.1324,
+      "step": 65700
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.438125490999761e-05,
+      "loss": 3.0407,
+      "step": 65800
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.437271578372101e-05,
+      "loss": 3.2393,
+      "step": 65900
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.4364176657444414e-05,
+      "loss": 3.1361,
+      "step": 66000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.4355637531167816e-05,
+      "loss": 3.2417,
+      "step": 66100
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.434709840489121e-05,
+      "loss": 3.2056,
+      "step": 66200
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.433855927861462e-05,
+      "loss": 3.2976,
+      "step": 66300
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.4330020152338014e-05,
+      "loss": 3.2442,
+      "step": 66400
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.4321481026061415e-05,
+      "loss": 3.1343,
+      "step": 66500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.431294189978482e-05,
+      "loss": 3.213,
+      "step": 66600
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.430440277350822e-05,
+      "loss": 3.1263,
+      "step": 66700
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.429586364723162e-05,
+      "loss": 3.2514,
+      "step": 66800
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.428732452095502e-05,
+      "loss": 3.1529,
+      "step": 66900
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.4278785394678416e-05,
+      "loss": 3.1256,
+      "step": 67000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.4270246268401825e-05,
+      "loss": 3.3028,
+      "step": 67100
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.426170714212522e-05,
+      "loss": 2.9583,
+      "step": 67200
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.425316801584862e-05,
+      "loss": 3.3054,
+      "step": 67300
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.424462888957202e-05,
+      "loss": 3.3126,
+      "step": 67400
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.4236089763295424e-05,
+      "loss": 3.0507,
+      "step": 67500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.422755063701882e-05,
+      "loss": 3.1097,
+      "step": 67600
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.421901151074223e-05,
+      "loss": 2.932,
+      "step": 67700
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.421047238446562e-05,
+      "loss": 3.1756,
+      "step": 67800
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.4201933258189023e-05,
+      "loss": 3.126,
+      "step": 67900
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.4193394131912425e-05,
+      "loss": 3.154,
+      "step": 68000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.4184855005635827e-05,
+      "loss": 3.1391,
+      "step": 68100
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.417631587935923e-05,
+      "loss": 3.2438,
+      "step": 68200
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.416777675308263e-05,
+      "loss": 3.1666,
+      "step": 68300
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.4159237626806024e-05,
+      "loss": 3.1894,
+      "step": 68400
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.415069850052943e-05,
+      "loss": 3.1682,
+      "step": 68500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.414215937425283e-05,
+      "loss": 3.1575,
+      "step": 68600
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.413362024797623e-05,
+      "loss": 2.9601,
+      "step": 68700
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.412508112169963e-05,
+      "loss": 3.2788,
+      "step": 68800
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.411654199542303e-05,
+      "loss": 3.3421,
+      "step": 68900
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.4108002869146434e-05,
+      "loss": 3.0026,
+      "step": 69000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.4099463742869835e-05,
+      "loss": 3.2379,
+      "step": 69100
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.409092461659323e-05,
+      "loss": 3.2624,
+      "step": 69200
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.408238549031664e-05,
+      "loss": 3.0439,
+      "step": 69300
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.407384636404003e-05,
+      "loss": 3.2517,
+      "step": 69400
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.4065307237763435e-05,
+      "loss": 3.259,
+      "step": 69500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.4056768111486836e-05,
+      "loss": 3.1286,
+      "step": 69600
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.404822898521024e-05,
+      "loss": 3.2362,
+      "step": 69700
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.403968985893364e-05,
+      "loss": 3.1902,
+      "step": 69800
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.403115073265704e-05,
+      "loss": 3.4516,
+      "step": 69900
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.4022611606380436e-05,
+      "loss": 3.0569,
+      "step": 70000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.4014072480103844e-05,
+      "loss": 3.2138,
+      "step": 70100
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.400553335382724e-05,
+      "loss": 3.1757,
+      "step": 70200
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.3996994227550633e-05,
+      "loss": 3.2162,
+      "step": 70300
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.398845510127404e-05,
+      "loss": 3.2975,
+      "step": 70400
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.3979915974997437e-05,
+      "loss": 3.1081,
+      "step": 70500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.397137684872084e-05,
+      "loss": 3.1195,
+      "step": 70600
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.396283772244424e-05,
+      "loss": 3.2665,
+      "step": 70700
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.395429859616764e-05,
+      "loss": 3.125,
+      "step": 70800
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.394575946989104e-05,
+      "loss": 3.2652,
+      "step": 70900
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3937220343614444e-05,
+      "loss": 3.0792,
+      "step": 71000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.392868121733784e-05,
+      "loss": 3.2215,
+      "step": 71100
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.392014209106125e-05,
+      "loss": 2.9989,
+      "step": 71200
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.391160296478464e-05,
+      "loss": 3.2054,
+      "step": 71300
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.3903063838508044e-05,
+      "loss": 3.1994,
+      "step": 71400
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.3894524712231445e-05,
+      "loss": 3.0047,
+      "step": 71500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.388598558595485e-05,
+      "loss": 2.9717,
+      "step": 71600
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.387744645967825e-05,
+      "loss": 3.1778,
+      "step": 71700
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.386890733340165e-05,
+      "loss": 3.1838,
+      "step": 71800
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.3860368207125045e-05,
+      "loss": 3.0509,
+      "step": 71900
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.385182908084845e-05,
+      "loss": 3.0945,
+      "step": 72000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.384328995457185e-05,
+      "loss": 3.2435,
+      "step": 72100
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.383475082829525e-05,
+      "loss": 3.134,
+      "step": 72200
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.382621170201865e-05,
+      "loss": 3.4882,
+      "step": 72300
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.381767257574205e-05,
+      "loss": 3.4011,
+      "step": 72400
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.380913344946545e-05,
+      "loss": 3.1124,
+      "step": 72500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.3800594323188856e-05,
+      "loss": 3.2966,
+      "step": 72600
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.379205519691225e-05,
+      "loss": 3.3746,
+      "step": 72700
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.378351607063565e-05,
+      "loss": 3.085,
+      "step": 72800
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.3774976944359053e-05,
+      "loss": 3.1161,
+      "step": 72900
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.3766437818082455e-05,
+      "loss": 3.0355,
+      "step": 73000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.3757898691805857e-05,
+      "loss": 3.1345,
+      "step": 73100
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.374935956552926e-05,
+      "loss": 3.2653,
+      "step": 73200
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.374082043925265e-05,
+      "loss": 3.0472,
+      "step": 73300
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.373228131297606e-05,
+      "loss": 3.0984,
+      "step": 73400
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.3723742186699456e-05,
+      "loss": 3.2532,
+      "step": 73500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.371520306042286e-05,
+      "loss": 3.2685,
+      "step": 73600
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.370666393414626e-05,
+      "loss": 3.2064,
+      "step": 73700
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.369812480786966e-05,
+      "loss": 2.9865,
+      "step": 73800
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.368958568159306e-05,
+      "loss": 3.0411,
+      "step": 73900
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.3681046555316464e-05,
+      "loss": 3.0619,
+      "step": 74000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.367250742903986e-05,
+      "loss": 3.271,
+      "step": 74100
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.366396830276327e-05,
+      "loss": 3.2532,
+      "step": 74200
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.365542917648666e-05,
+      "loss": 3.18,
+      "step": 74300
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.364689005021006e-05,
+      "loss": 3.3295,
+      "step": 74400
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.3638350923933465e-05,
+      "loss": 3.1967,
+      "step": 74500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.3629811797656866e-05,
+      "loss": 3.2007,
+      "step": 74600
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.362127267138027e-05,
+      "loss": 3.2682,
+      "step": 74700
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.361273354510367e-05,
+      "loss": 2.9575,
+      "step": 74800
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.3604194418827064e-05,
+      "loss": 3.2221,
+      "step": 74900
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.359565529255047e-05,
+      "loss": 2.9996,
+      "step": 75000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.358711616627387e-05,
+      "loss": 3.0046,
+      "step": 75100
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.357857703999727e-05,
+      "loss": 3.1096,
+      "step": 75200
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.357003791372067e-05,
+      "loss": 3.3285,
+      "step": 75300
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.356149878744407e-05,
+      "loss": 3.1116,
+      "step": 75400
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.3552959661167467e-05,
+      "loss": 3.2307,
+      "step": 75500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.3544420534890875e-05,
+      "loss": 3.3776,
+      "step": 75600
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.353588140861427e-05,
+      "loss": 3.2924,
+      "step": 75700
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.352734228233767e-05,
+      "loss": 3.117,
+      "step": 75800
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.351880315606107e-05,
+      "loss": 3.2263,
+      "step": 75900
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.3510264029784474e-05,
+      "loss": 3.0552,
+      "step": 76000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.3501724903507876e-05,
+      "loss": 3.142,
+      "step": 76100
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.349318577723128e-05,
+      "loss": 3.3654,
+      "step": 76200
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.348464665095467e-05,
+      "loss": 3.1482,
+      "step": 76300
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.347610752467808e-05,
+      "loss": 3.1302,
+      "step": 76400
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.3467568398401475e-05,
+      "loss": 3.2023,
+      "step": 76500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.345902927212488e-05,
+      "loss": 3.0512,
+      "step": 76600
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.345049014584828e-05,
+      "loss": 3.2809,
+      "step": 76700
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.344195101957168e-05,
+      "loss": 3.1384,
+      "step": 76800
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.343341189329508e-05,
+      "loss": 3.2299,
+      "step": 76900
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.342487276701848e-05,
+      "loss": 3.0812,
+      "step": 77000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.341633364074188e-05,
+      "loss": 3.2408,
+      "step": 77100
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.3407794514465286e-05,
+      "loss": 3.1539,
+      "step": 77200
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.339925538818868e-05,
+      "loss": 3.1098,
+      "step": 77300
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.339071626191208e-05,
+      "loss": 3.066,
+      "step": 77400
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.3382177135635484e-05,
+      "loss": 2.984,
+      "step": 77500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.3373638009358886e-05,
+      "loss": 3.2029,
+      "step": 77600
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.336509888308228e-05,
+      "loss": 3.1193,
+      "step": 77700
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.335655975680569e-05,
+      "loss": 3.193,
+      "step": 77800
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.3348020630529083e-05,
+      "loss": 3.2723,
+      "step": 77900
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.3339481504252485e-05,
+      "loss": 3.3588,
+      "step": 78000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.3330942377975886e-05,
+      "loss": 3.0575,
+      "step": 78100
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.332240325169929e-05,
+      "loss": 2.9386,
+      "step": 78200
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.331386412542269e-05,
+      "loss": 3.1232,
+      "step": 78300
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.330532499914609e-05,
+      "loss": 2.9355,
+      "step": 78400
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.3296785872869486e-05,
+      "loss": 3.0254,
+      "step": 78500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.3288246746592894e-05,
+      "loss": 3.2184,
+      "step": 78600
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.327970762031629e-05,
+      "loss": 3.0451,
+      "step": 78700
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.327116849403969e-05,
+      "loss": 3.085,
+      "step": 78800
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.326262936776309e-05,
+      "loss": 2.9988,
+      "step": 78900
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.3254090241486494e-05,
+      "loss": 3.0037,
+      "step": 79000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.3245551115209895e-05,
+      "loss": 3.1751,
+      "step": 79100
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.32370119889333e-05,
+      "loss": 3.0905,
+      "step": 79200
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.322847286265669e-05,
+      "loss": 3.0897,
+      "step": 79300
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.32199337363801e-05,
+      "loss": 3.1822,
+      "step": 79400
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.3211394610103495e-05,
+      "loss": 3.2876,
+      "step": 79500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.3202855483826896e-05,
+      "loss": 3.1157,
+      "step": 79600
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.31943163575503e-05,
+      "loss": 3.1676,
+      "step": 79700
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.31857772312737e-05,
+      "loss": 3.2256,
+      "step": 79800
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.31772381049971e-05,
+      "loss": 3.162,
+      "step": 79900
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.31686989787205e-05,
+      "loss": 3.0349,
+      "step": 80000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.31601598524439e-05,
+      "loss": 3.0997,
+      "step": 80100
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.3151620726167305e-05,
+      "loss": 3.1964,
+      "step": 80200
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.31430815998907e-05,
+      "loss": 3.1396,
+      "step": 80300
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.31345424736141e-05,
+      "loss": 3.1143,
+      "step": 80400
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.31260033473375e-05,
+      "loss": 3.0063,
+      "step": 80500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.3117464221060905e-05,
+      "loss": 3.2515,
+      "step": 80600
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.31089250947843e-05,
+      "loss": 3.1233,
+      "step": 80700
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.310038596850771e-05,
+      "loss": 2.9936,
+      "step": 80800
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.30918468422311e-05,
+      "loss": 2.9814,
+      "step": 80900
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.3083307715954504e-05,
+      "loss": 3.2333,
+      "step": 81000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.3074768589677906e-05,
+      "loss": 3.2694,
+      "step": 81100
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.306622946340131e-05,
+      "loss": 3.0663,
+      "step": 81200
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.305769033712471e-05,
+      "loss": 3.4595,
+      "step": 81300
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.304915121084811e-05,
+      "loss": 3.2051,
+      "step": 81400
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.3040612084571505e-05,
+      "loss": 3.2371,
+      "step": 81500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.3032072958294914e-05,
+      "loss": 3.1418,
+      "step": 81600
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.302353383201831e-05,
+      "loss": 3.3268,
+      "step": 81700
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.301499470574171e-05,
+      "loss": 3.1883,
+      "step": 81800
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.300645557946511e-05,
+      "loss": 3.2198,
+      "step": 81900
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.299791645318851e-05,
+      "loss": 3.3429,
+      "step": 82000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.2989377326911915e-05,
+      "loss": 3.0335,
+      "step": 82100
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2980838200635316e-05,
+      "loss": 3.1086,
+      "step": 82200
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.297229907435871e-05,
+      "loss": 3.1151,
+      "step": 82300
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.296375994808212e-05,
+      "loss": 3.1687,
+      "step": 82400
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2955220821805514e-05,
+      "loss": 2.9901,
+      "step": 82500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.2946681695528916e-05,
+      "loss": 3.3057,
+      "step": 82600
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.293814256925232e-05,
+      "loss": 3.1803,
+      "step": 82700
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.292960344297572e-05,
+      "loss": 3.0191,
+      "step": 82800
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.292106431669911e-05,
+      "loss": 3.1051,
+      "step": 82900
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.291252519042252e-05,
+      "loss": 3.2292,
+      "step": 83000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.2903986064145916e-05,
+      "loss": 3.0433,
+      "step": 83100
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.289544693786932e-05,
+      "loss": 3.1835,
+      "step": 83200
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.288690781159272e-05,
+      "loss": 3.3649,
+      "step": 83300
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.287836868531612e-05,
+      "loss": 3.0549,
+      "step": 83400
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.286982955903952e-05,
+      "loss": 3.159,
+      "step": 83500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.2861290432762924e-05,
+      "loss": 3.0583,
+      "step": 83600
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.285275130648632e-05,
+      "loss": 3.1267,
+      "step": 83700
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.284421218020973e-05,
+      "loss": 3.1198,
+      "step": 83800
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.283567305393312e-05,
+      "loss": 3.261,
+      "step": 83900
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.2827133927656524e-05,
+      "loss": 2.9281,
+      "step": 84000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.2818594801379925e-05,
+      "loss": 3.0939,
+      "step": 84100
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.281005567510333e-05,
+      "loss": 3.0363,
+      "step": 84200
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.280151654882673e-05,
+      "loss": 3.068,
+      "step": 84300
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.279297742255013e-05,
+      "loss": 3.129,
+      "step": 84400
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.2784438296273525e-05,
+      "loss": 3.2133,
+      "step": 84500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.277589916999693e-05,
+      "loss": 3.2536,
+      "step": 84600
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.276736004372033e-05,
+      "loss": 2.9967,
+      "step": 84700
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.275882091744373e-05,
+      "loss": 2.9866,
+      "step": 84800
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.275028179116713e-05,
+      "loss": 3.1255,
+      "step": 84900
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.274174266489053e-05,
+      "loss": 2.9926,
+      "step": 85000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.2733203538613934e-05,
+      "loss": 3.0661,
+      "step": 85100
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.2724664412337335e-05,
+      "loss": 3.1502,
+      "step": 85200
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.271612528606073e-05,
+      "loss": 2.8778,
+      "step": 85300
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.270758615978414e-05,
+      "loss": 3.2147,
+      "step": 85400
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.269904703350753e-05,
+      "loss": 2.9854,
+      "step": 85500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.2690507907230935e-05,
+      "loss": 3.1181,
+      "step": 85600
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.2681968780954336e-05,
+      "loss": 2.9929,
+      "step": 85700
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.267342965467774e-05,
+      "loss": 3.1671,
+      "step": 85800
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.266489052840113e-05,
+      "loss": 3.0857,
+      "step": 85900
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.265635140212454e-05,
+      "loss": 3.1893,
+      "step": 86000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.2647812275847936e-05,
+      "loss": 3.1835,
+      "step": 86100
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.263927314957134e-05,
+      "loss": 3.3165,
+      "step": 86200
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.263073402329474e-05,
+      "loss": 3.0212,
+      "step": 86300
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.262219489701814e-05,
+      "loss": 3.0893,
+      "step": 86400
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.261365577074154e-05,
+      "loss": 3.0834,
+      "step": 86500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.2605116644464944e-05,
+      "loss": 3.0804,
+      "step": 86600
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.259657751818834e-05,
+      "loss": 3.133,
+      "step": 86700
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.258803839191175e-05,
+      "loss": 3.3209,
+      "step": 86800
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.257949926563514e-05,
+      "loss": 3.1159,
+      "step": 86900
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.257096013935854e-05,
+      "loss": 3.0212,
+      "step": 87000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.2562421013081945e-05,
+      "loss": 3.0374,
+      "step": 87100
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.2553881886805346e-05,
+      "loss": 2.8671,
+      "step": 87200
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.254534276052875e-05,
+      "loss": 3.211,
+      "step": 87300
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.253680363425215e-05,
+      "loss": 3.2994,
+      "step": 87400
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.2528264507975544e-05,
+      "loss": 3.1808,
+      "step": 87500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.251972538169895e-05,
+      "loss": 3.0728,
+      "step": 87600
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.251118625542235e-05,
+      "loss": 3.1494,
+      "step": 87700
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.250264712914574e-05,
+      "loss": 3.0919,
+      "step": 87800
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.249410800286915e-05,
+      "loss": 3.1287,
+      "step": 87900
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.2485568876592545e-05,
+      "loss": 3.0724,
+      "step": 88000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.2477029750315946e-05,
+      "loss": 3.1844,
+      "step": 88100
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.246849062403935e-05,
+      "loss": 2.9444,
+      "step": 88200
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.245995149776275e-05,
+      "loss": 3.2252,
+      "step": 88300
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.245141237148615e-05,
+      "loss": 2.9772,
+      "step": 88400
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.244287324520955e-05,
+      "loss": 3.1618,
+      "step": 88500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.243433411893295e-05,
+      "loss": 3.2035,
+      "step": 88600
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.2425794992656356e-05,
+      "loss": 3.03,
+      "step": 88700
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.241725586637975e-05,
+      "loss": 3.0816,
+      "step": 88800
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.240871674010315e-05,
+      "loss": 3.0276,
+      "step": 88900
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.2400177613826554e-05,
+      "loss": 3.1817,
+      "step": 89000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.2391638487549955e-05,
+      "loss": 2.8854,
+      "step": 89100
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.238309936127336e-05,
+      "loss": 2.99,
+      "step": 89200
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.237456023499676e-05,
+      "loss": 3.2029,
+      "step": 89300
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.236602110872015e-05,
+      "loss": 3.1559,
+      "step": 89400
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.235748198244356e-05,
+      "loss": 3.1818,
+      "step": 89500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.2348942856166956e-05,
+      "loss": 3.1181,
+      "step": 89600
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.234040372989036e-05,
+      "loss": 3.1893,
+      "step": 89700
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.233186460361376e-05,
+      "loss": 3.1009,
+      "step": 89800
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.232332547733716e-05,
+      "loss": 3.2049,
+      "step": 89900
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.231478635106056e-05,
+      "loss": 3.0272,
+      "step": 90000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.2306247224783964e-05,
+      "loss": 3.2394,
+      "step": 90100
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.229770809850736e-05,
+      "loss": 2.8205,
+      "step": 90200
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.228916897223077e-05,
+      "loss": 3.2412,
+      "step": 90300
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.228062984595416e-05,
+      "loss": 2.9937,
+      "step": 90400
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.227209071967756e-05,
+      "loss": 3.2042,
+      "step": 90500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.2263551593400965e-05,
+      "loss": 3.0972,
+      "step": 90600
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.2255012467124366e-05,
+      "loss": 3.2973,
+      "step": 90700
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.224647334084776e-05,
+      "loss": 3.0852,
+      "step": 90800
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.223793421457117e-05,
+      "loss": 3.0435,
+      "step": 90900
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.2229395088294564e-05,
+      "loss": 3.1896,
+      "step": 91000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.2220855962017966e-05,
+      "loss": 3.2066,
+      "step": 91100
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.221231683574137e-05,
+      "loss": 3.0411,
+      "step": 91200
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.220377770946477e-05,
+      "loss": 3.1446,
+      "step": 91300
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.219523858318817e-05,
+      "loss": 2.9866,
+      "step": 91400
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.218669945691157e-05,
+      "loss": 2.9629,
+      "step": 91500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.217816033063497e-05,
+      "loss": 3.0371,
+      "step": 91600
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.2169621204358375e-05,
+      "loss": 3.1809,
+      "step": 91700
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.216108207808177e-05,
+      "loss": 3.2357,
+      "step": 91800
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.215254295180517e-05,
+      "loss": 3.298,
+      "step": 91900
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.214400382552857e-05,
+      "loss": 3.0594,
+      "step": 92000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.2135464699251974e-05,
+      "loss": 2.8645,
+      "step": 92100
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.2126925572975376e-05,
+      "loss": 3.1693,
+      "step": 92200
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.211838644669878e-05,
+      "loss": 3.0464,
+      "step": 92300
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.210984732042217e-05,
+      "loss": 3.0673,
+      "step": 92400
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.210130819414558e-05,
+      "loss": 2.959,
+      "step": 92500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.2092769067868975e-05,
+      "loss": 3.0101,
+      "step": 92600
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.208422994159238e-05,
+      "loss": 3.13,
+      "step": 92700
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.207569081531578e-05,
+      "loss": 3.0754,
+      "step": 92800
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.206715168903918e-05,
+      "loss": 3.0374,
+      "step": 92900
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.2058612562762575e-05,
+      "loss": 3.2083,
+      "step": 93000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.205007343648598e-05,
+      "loss": 2.9163,
+      "step": 93100
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.204153431020938e-05,
+      "loss": 3.0933,
+      "step": 93200
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.2032995183932786e-05,
+      "loss": 3.125,
+      "step": 93300
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.202445605765618e-05,
+      "loss": 3.1055,
+      "step": 93400
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.201591693137958e-05,
+      "loss": 3.0669,
+      "step": 93500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.2007377805102984e-05,
+      "loss": 3.0477,
+      "step": 93600
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.1998838678826386e-05,
+      "loss": 3.1347,
+      "step": 93700
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.199029955254978e-05,
+      "loss": 3.1189,
+      "step": 93800
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.198176042627319e-05,
+      "loss": 3.2261,
+      "step": 93900
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.1973221299996584e-05,
+      "loss": 3.0995,
+      "step": 94000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.1964682173719985e-05,
+      "loss": 3.0714,
+      "step": 94100
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.195614304744339e-05,
+      "loss": 3.2601,
+      "step": 94200
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.194760392116679e-05,
+      "loss": 3.1525,
+      "step": 94300
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.193906479489019e-05,
+      "loss": 3.0394,
+      "step": 94400
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.193052566861359e-05,
+      "loss": 3.2386,
+      "step": 94500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.1921986542336986e-05,
+      "loss": 3.239,
+      "step": 94600
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.1913447416060394e-05,
+      "loss": 3.0516,
+      "step": 94700
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.190490828978379e-05,
+      "loss": 3.0337,
+      "step": 94800
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.189636916350719e-05,
+      "loss": 2.9627,
+      "step": 94900
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.188783003723059e-05,
+      "loss": 3.0803,
+      "step": 95000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.1879290910953994e-05,
+      "loss": 2.9824,
+      "step": 95100
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.1870751784677395e-05,
+      "loss": 3.1304,
+      "step": 95200
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.18622126584008e-05,
+      "loss": 3.0516,
+      "step": 95300
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.185367353212419e-05,
+      "loss": 2.9625,
+      "step": 95400
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.18451344058476e-05,
+      "loss": 2.9388,
+      "step": 95500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.1836595279570995e-05,
+      "loss": 3.1363,
+      "step": 95600
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.1828056153294396e-05,
+      "loss": 2.9823,
+      "step": 95700
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.18195170270178e-05,
+      "loss": 3.108,
+      "step": 95800
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.18109779007412e-05,
+      "loss": 3.1888,
+      "step": 95900
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.1802438774464594e-05,
+      "loss": 3.1066,
+      "step": 96000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.1793899648188e-05,
+      "loss": 2.9461,
+      "step": 96100
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.17853605219114e-05,
+      "loss": 3.0784,
+      "step": 96200
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.17768213956348e-05,
+      "loss": 3.0233,
+      "step": 96300
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.17682822693582e-05,
+      "loss": 3.0728,
+      "step": 96400
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.17597431430816e-05,
+      "loss": 3.0897,
+      "step": 96500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.1751204016805004e-05,
+      "loss": 3.0342,
+      "step": 96600
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.1742664890528405e-05,
+      "loss": 2.9922,
+      "step": 96700
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.17341257642518e-05,
+      "loss": 3.0786,
+      "step": 96800
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.172558663797521e-05,
+      "loss": 2.9265,
+      "step": 96900
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.17170475116986e-05,
+      "loss": 3.109,
+      "step": 97000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.1708508385422004e-05,
+      "loss": 3.229,
+      "step": 97100
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.1699969259145406e-05,
+      "loss": 2.9595,
+      "step": 97200
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.169143013286881e-05,
+      "loss": 2.9587,
+      "step": 97300
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.168289100659221e-05,
+      "loss": 3.0812,
+      "step": 97400
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.167435188031561e-05,
+      "loss": 3.1367,
+      "step": 97500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.1665812754039005e-05,
+      "loss": 3.0375,
+      "step": 97600
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.1657273627762414e-05,
+      "loss": 3.1682,
+      "step": 97700
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.164873450148581e-05,
+      "loss": 3.0156,
+      "step": 97800
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.164019537520921e-05,
+      "loss": 3.1416,
+      "step": 97900
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.163165624893261e-05,
+      "loss": 3.1544,
+      "step": 98000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.162311712265601e-05,
+      "loss": 3.1956,
+      "step": 98100
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.1614577996379415e-05,
+      "loss": 3.0813,
+      "step": 98200
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.1606038870102816e-05,
+      "loss": 3.0339,
+      "step": 98300
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.159749974382621e-05,
+      "loss": 3.0538,
+      "step": 98400
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.158896061754962e-05,
+      "loss": 3.152,
+      "step": 98500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.1580421491273014e-05,
+      "loss": 3.1042,
+      "step": 98600
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.1571882364996416e-05,
+      "loss": 2.8913,
+      "step": 98700
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.156334323871982e-05,
+      "loss": 3.0919,
+      "step": 98800
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.155480411244322e-05,
+      "loss": 3.0475,
+      "step": 98900
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.1546264986166614e-05,
+      "loss": 2.9587,
+      "step": 99000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.153772585989002e-05,
+      "loss": 3.0237,
+      "step": 99100
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.152918673361342e-05,
+      "loss": 2.9342,
+      "step": 99200
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.152064760733682e-05,
+      "loss": 3.0741,
+      "step": 99300
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.151210848106022e-05,
+      "loss": 3.3034,
+      "step": 99400
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.150356935478362e-05,
+      "loss": 2.997,
+      "step": 99500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.149503022850702e-05,
+      "loss": 3.1982,
+      "step": 99600
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.1486491102230424e-05,
+      "loss": 3.0697,
+      "step": 99700
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.147795197595382e-05,
+      "loss": 3.111,
+      "step": 99800
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.146941284967723e-05,
+      "loss": 3.1279,
+      "step": 99900
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.146087372340062e-05,
+      "loss": 3.1502,
+      "step": 100000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.1452334597124024e-05,
+      "loss": 2.8996,
+      "step": 100100
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.1443795470847425e-05,
+      "loss": 3.2088,
+      "step": 100200
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.143525634457083e-05,
+      "loss": 3.1219,
+      "step": 100300
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.142671721829423e-05,
+      "loss": 3.0994,
+      "step": 100400
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.141817809201763e-05,
+      "loss": 3.076,
+      "step": 100500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.1409638965741025e-05,
+      "loss": 3.0914,
+      "step": 100600
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.140109983946443e-05,
+      "loss": 2.9685,
+      "step": 100700
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.139256071318783e-05,
+      "loss": 3.3519,
+      "step": 100800
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.138402158691123e-05,
+      "loss": 3.0838,
+      "step": 100900
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.137548246063463e-05,
+      "loss": 3.0116,
+      "step": 101000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.136694333435803e-05,
+      "loss": 2.9562,
+      "step": 101100
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.135840420808143e-05,
+      "loss": 3.1637,
+      "step": 101200
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.1349865081804836e-05,
+      "loss": 3.1111,
+      "step": 101300
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.134132595552823e-05,
+      "loss": 2.9523,
+      "step": 101400
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.133278682925163e-05,
+      "loss": 2.9791,
+      "step": 101500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.1324247702975033e-05,
+      "loss": 3.0496,
+      "step": 101600
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.1315708576698435e-05,
+      "loss": 3.0005,
+      "step": 101700
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.1307169450421837e-05,
+      "loss": 3.242,
+      "step": 101800
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.129863032414524e-05,
+      "loss": 2.9777,
+      "step": 101900
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.129009119786863e-05,
+      "loss": 3.0187,
+      "step": 102000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.128155207159204e-05,
+      "loss": 2.9164,
+      "step": 102100
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.1273012945315436e-05,
+      "loss": 3.1537,
+      "step": 102200
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.126447381903884e-05,
+      "loss": 3.0327,
+      "step": 102300
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.125593469276224e-05,
+      "loss": 3.3472,
+      "step": 102400
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.124739556648564e-05,
+      "loss": 3.0513,
+      "step": 102500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.123885644020904e-05,
+      "loss": 3.0413,
+      "step": 102600
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.1230317313932444e-05,
+      "loss": 3.041,
+      "step": 102700
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.122177818765584e-05,
+      "loss": 3.0476,
+      "step": 102800
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.121323906137925e-05,
+      "loss": 3.0221,
+      "step": 102900
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.120469993510264e-05,
+      "loss": 3.0639,
+      "step": 103000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.119616080882604e-05,
+      "loss": 3.1602,
+      "step": 103100
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.1187621682549445e-05,
+      "loss": 2.9889,
+      "step": 103200
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.1179082556272846e-05,
+      "loss": 3.0175,
+      "step": 103300
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.117054342999625e-05,
+      "loss": 2.9935,
+      "step": 103400
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.116200430371965e-05,
+      "loss": 3.168,
+      "step": 103500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.1153465177443044e-05,
+      "loss": 3.0671,
+      "step": 103600
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.114492605116645e-05,
+      "loss": 3.1425,
+      "step": 103700
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.113638692488985e-05,
+      "loss": 3.0317,
+      "step": 103800
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.112784779861325e-05,
+      "loss": 3.0031,
+      "step": 103900
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.111930867233665e-05,
+      "loss": 3.0538,
+      "step": 104000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.111076954606005e-05,
+      "loss": 2.9468,
+      "step": 104100
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.110223041978345e-05,
+      "loss": 2.9273,
+      "step": 104200
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.1093691293506855e-05,
+      "loss": 2.8708,
+      "step": 104300
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.108515216723025e-05,
+      "loss": 3.0507,
+      "step": 104400
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.107661304095365e-05,
+      "loss": 3.059,
+      "step": 104500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.106807391467705e-05,
+      "loss": 2.9688,
+      "step": 104600
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.1059534788400454e-05,
+      "loss": 3.0545,
+      "step": 104700
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.1050995662123856e-05,
+      "loss": 3.1936,
+      "step": 104800
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.104245653584726e-05,
+      "loss": 3.158,
+      "step": 104900
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.103391740957065e-05,
+      "loss": 3.0972,
+      "step": 105000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.102537828329406e-05,
+      "loss": 3.2179,
+      "step": 105100
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.1016839157017455e-05,
+      "loss": 3.1017,
+      "step": 105200
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.100830003074086e-05,
+      "loss": 3.0019,
+      "step": 105300
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.099976090446426e-05,
+      "loss": 3.1353,
+      "step": 105400
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.099122177818765e-05,
+      "loss": 3.1709,
+      "step": 105500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.098268265191106e-05,
+      "loss": 2.9705,
+      "step": 105600
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.0974143525634456e-05,
+      "loss": 3.1053,
+      "step": 105700
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.096560439935786e-05,
+      "loss": 3.0786,
+      "step": 105800
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.095706527308126e-05,
+      "loss": 3.1098,
+      "step": 105900
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.094852614680466e-05,
+      "loss": 2.8744,
+      "step": 106000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.0939987020528056e-05,
+      "loss": 2.9382,
+      "step": 106100
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.0931447894251464e-05,
+      "loss": 2.9395,
+      "step": 106200
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.092290876797486e-05,
+      "loss": 3.053,
+      "step": 106300
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.091436964169826e-05,
+      "loss": 3.1599,
+      "step": 106400
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.090583051542166e-05,
+      "loss": 3.0938,
+      "step": 106500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0897291389145063e-05,
+      "loss": 2.9966,
+      "step": 106600
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0888752262868465e-05,
+      "loss": 3.1077,
+      "step": 106700
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0880213136591867e-05,
+      "loss": 3.1119,
+      "step": 106800
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.087167401031526e-05,
+      "loss": 3.1429,
+      "step": 106900
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.086313488403867e-05,
+      "loss": 3.0915,
+      "step": 107000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0854595757762064e-05,
+      "loss": 3.324,
+      "step": 107100
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.0846056631485466e-05,
+      "loss": 3.1729,
+      "step": 107200
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.083751750520887e-05,
+      "loss": 2.9374,
+      "step": 107300
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.082897837893227e-05,
+      "loss": 3.3158,
+      "step": 107400
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.082043925265567e-05,
+      "loss": 3.0624,
+      "step": 107500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.081190012637907e-05,
+      "loss": 3.199,
+      "step": 107600
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.080336100010247e-05,
+      "loss": 3.0148,
+      "step": 107700
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.0794821873825875e-05,
+      "loss": 3.0849,
+      "step": 107800
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.078628274754927e-05,
+      "loss": 2.9066,
+      "step": 107900
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.077774362127267e-05,
+      "loss": 3.0127,
+      "step": 108000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.076920449499607e-05,
+      "loss": 2.9849,
+      "step": 108100
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.0760665368719475e-05,
+      "loss": 2.8326,
+      "step": 108200
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.0752126242442876e-05,
+      "loss": 3.1002,
+      "step": 108300
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.074358711616628e-05,
+      "loss": 2.8001,
+      "step": 108400
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.073504798988967e-05,
+      "loss": 2.9754,
+      "step": 108500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.072650886361308e-05,
+      "loss": 3.1234,
+      "step": 108600
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.0717969737336476e-05,
+      "loss": 3.0748,
+      "step": 108700
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.070943061105988e-05,
+      "loss": 3.0435,
+      "step": 108800
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.070089148478328e-05,
+      "loss": 3.0546,
+      "step": 108900
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.069235235850668e-05,
+      "loss": 3.2162,
+      "step": 109000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.0683813232230075e-05,
+      "loss": 2.8097,
+      "step": 109100
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.067527410595348e-05,
+      "loss": 2.949,
+      "step": 109200
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.066673497967688e-05,
+      "loss": 3.0555,
+      "step": 109300
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.065819585340028e-05,
+      "loss": 2.9504,
+      "step": 109400
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.064965672712368e-05,
+      "loss": 3.1909,
+      "step": 109500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.064111760084708e-05,
+      "loss": 2.7929,
+      "step": 109600
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.0632578474570484e-05,
+      "loss": 3.0875,
+      "step": 109700
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.0624039348293886e-05,
+      "loss": 3.0052,
+      "step": 109800
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.061550022201728e-05,
+      "loss": 2.9049,
+      "step": 109900
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.060696109574069e-05,
+      "loss": 3.0253,
+      "step": 110000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.0598421969464084e-05,
+      "loss": 3.1229,
+      "step": 110100
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.0589882843187485e-05,
+      "loss": 2.993,
+      "step": 110200
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.058134371691089e-05,
+      "loss": 3.04,
+      "step": 110300
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.057280459063429e-05,
+      "loss": 3.0167,
+      "step": 110400
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.056426546435769e-05,
+      "loss": 3.3009,
+      "step": 110500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.055572633808109e-05,
+      "loss": 3.1617,
+      "step": 110600
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.0547187211804486e-05,
+      "loss": 3.1479,
+      "step": 110700
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.0538648085527895e-05,
+      "loss": 3.1471,
+      "step": 110800
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.053010895925129e-05,
+      "loss": 3.046,
+      "step": 110900
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.052156983297469e-05,
+      "loss": 3.1228,
+      "step": 111000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.051303070669809e-05,
+      "loss": 2.9721,
+      "step": 111100
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.0504491580421494e-05,
+      "loss": 2.9042,
+      "step": 111200
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.049595245414489e-05,
+      "loss": 3.1024,
+      "step": 111300
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.04874133278683e-05,
+      "loss": 3.2672,
+      "step": 111400
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.047887420159169e-05,
+      "loss": 3.0472,
+      "step": 111500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.0470335075315093e-05,
+      "loss": 3.0819,
+      "step": 111600
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.0461795949038495e-05,
+      "loss": 3.0193,
+      "step": 111700
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.0453256822761897e-05,
+      "loss": 2.9185,
+      "step": 111800
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.04447176964853e-05,
+      "loss": 3.3109,
+      "step": 111900
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.04361785702087e-05,
+      "loss": 2.9213,
+      "step": 112000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.0427639443932094e-05,
+      "loss": 2.9948,
+      "step": 112100
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.04191003176555e-05,
+      "loss": 2.936,
+      "step": 112200
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.04105611913789e-05,
+      "loss": 3.2933,
+      "step": 112300
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.04020220651023e-05,
+      "loss": 3.0799,
+      "step": 112400
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.03934829388257e-05,
+      "loss": 3.0602,
+      "step": 112500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.03849438125491e-05,
+      "loss": 3.1894,
+      "step": 112600
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.0376404686272504e-05,
+      "loss": 2.8404,
+      "step": 112700
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.0367865559995905e-05,
+      "loss": 2.8567,
+      "step": 112800
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.03593264337193e-05,
+      "loss": 2.9545,
+      "step": 112900
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.035078730744271e-05,
+      "loss": 3.1068,
+      "step": 113000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.03422481811661e-05,
+      "loss": 3.0017,
+      "step": 113100
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.0333709054889505e-05,
+      "loss": 2.9809,
+      "step": 113200
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.0325169928612906e-05,
+      "loss": 2.9555,
+      "step": 113300
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.031663080233631e-05,
+      "loss": 2.7379,
+      "step": 113400
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.030809167605971e-05,
+      "loss": 2.882,
+      "step": 113500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.029955254978311e-05,
+      "loss": 2.8831,
+      "step": 113600
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.0291013423506506e-05,
+      "loss": 3.051,
+      "step": 113700
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.0282474297229914e-05,
+      "loss": 3.0809,
+      "step": 113800
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.027393517095331e-05,
+      "loss": 2.9474,
+      "step": 113900
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.026539604467671e-05,
+      "loss": 3.083,
+      "step": 114000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.025685691840011e-05,
+      "loss": 3.0755,
+      "step": 114100
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.024831779212351e-05,
+      "loss": 3.2383,
+      "step": 114200
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.023977866584691e-05,
+      "loss": 2.9592,
+      "step": 114300
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.0231239539570316e-05,
+      "loss": 2.841,
+      "step": 114400
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.022270041329371e-05,
+      "loss": 2.9557,
+      "step": 114500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.021416128701711e-05,
+      "loss": 3.0037,
+      "step": 114600
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.0205622160740514e-05,
+      "loss": 2.8829,
+      "step": 114700
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.0197083034463916e-05,
+      "loss": 3.0496,
+      "step": 114800
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.018854390818732e-05,
+      "loss": 3.0789,
+      "step": 114900
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.018000478191072e-05,
+      "loss": 2.9696,
+      "step": 115000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.0171465655634114e-05,
+      "loss": 2.9391,
+      "step": 115100
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.016292652935752e-05,
+      "loss": 3.196,
+      "step": 115200
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.015438740308092e-05,
+      "loss": 2.8523,
+      "step": 115300
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.014584827680432e-05,
+      "loss": 3.1546,
+      "step": 115400
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.013730915052772e-05,
+      "loss": 2.8606,
+      "step": 115500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.012877002425112e-05,
+      "loss": 3.0395,
+      "step": 115600
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.012023089797452e-05,
+      "loss": 2.8805,
+      "step": 115700
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.0111691771697925e-05,
+      "loss": 3.0013,
+      "step": 115800
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.010315264542132e-05,
+      "loss": 3.0258,
+      "step": 115900
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.009461351914473e-05,
+      "loss": 2.9218,
+      "step": 116000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.008607439286812e-05,
+      "loss": 3.0788,
+      "step": 116100
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.0077535266591524e-05,
+      "loss": 3.0068,
+      "step": 116200
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.0068996140314926e-05,
+      "loss": 3.1724,
+      "step": 116300
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.006045701403833e-05,
+      "loss": 3.103,
+      "step": 116400
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.005191788776172e-05,
+      "loss": 3.1552,
+      "step": 116500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.004337876148513e-05,
+      "loss": 3.036,
+      "step": 116600
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.0034839635208525e-05,
+      "loss": 3.2177,
+      "step": 116700
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.0026300508931927e-05,
+      "loss": 3.1162,
+      "step": 116800
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.001776138265533e-05,
+      "loss": 3.0141,
+      "step": 116900
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.000922225637873e-05,
+      "loss": 2.9341,
+      "step": 117000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.000068313010213e-05,
+      "loss": 2.9165,
+      "step": 117100
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.999214400382553e-05,
+      "loss": 3.046,
+      "step": 117200
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.998360487754893e-05,
+      "loss": 2.8234,
+      "step": 117300
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.9975065751272336e-05,
+      "loss": 3.1521,
+      "step": 117400
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.996652662499573e-05,
+      "loss": 3.014,
+      "step": 117500
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.995798749871913e-05,
+      "loss": 3.1095,
+      "step": 117600
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.9949448372442534e-05,
+      "loss": 3.0437,
+      "step": 117700
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.9940909246165935e-05,
+      "loss": 2.9899,
+      "step": 117800
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.993237011988934e-05,
+      "loss": 2.9003,
+      "step": 117900
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.992383099361274e-05,
+      "loss": 3.0027,
+      "step": 118000
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.991529186733613e-05,
+      "loss": 3.1018,
+      "step": 118100
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.990675274105954e-05,
+      "loss": 2.9672,
+      "step": 118200
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.9898213614782936e-05,
+      "loss": 2.9003,
+      "step": 118300
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.988967448850634e-05,
+      "loss": 2.8893,
+      "step": 118400
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.988113536222974e-05,
+      "loss": 3.148,
+      "step": 118500
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.987259623595314e-05,
+      "loss": 2.9243,
+      "step": 118600
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.986405710967654e-05,
+      "loss": 2.9773,
+      "step": 118700
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.9855517983399944e-05,
+      "loss": 2.8417,
+      "step": 118800
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.984697885712334e-05,
+      "loss": 3.0683,
+      "step": 118900
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.983843973084675e-05,
+      "loss": 2.9911,
+      "step": 119000
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.982990060457014e-05,
+      "loss": 2.8986,
+      "step": 119100
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.982136147829354e-05,
+      "loss": 3.2045,
+      "step": 119200
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.9812822352016945e-05,
+      "loss": 2.8607,
+      "step": 119300
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.9804283225740346e-05,
+      "loss": 2.9648,
+      "step": 119400
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.979574409946374e-05,
+      "loss": 2.9939,
+      "step": 119500
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.978720497318715e-05,
+      "loss": 3.1878,
+      "step": 119600
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.9778665846910544e-05,
+      "loss": 2.9334,
+      "step": 119700
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.9770126720633946e-05,
+      "loss": 2.9575,
+      "step": 119800
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.976158759435735e-05,
+      "loss": 2.8912,
+      "step": 119900
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.975304846808075e-05,
+      "loss": 2.7881,
+      "step": 120000
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.974450934180415e-05,
+      "loss": 3.0064,
+      "step": 120100
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.973597021552755e-05,
+      "loss": 2.963,
+      "step": 120200
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.972743108925095e-05,
+      "loss": 3.2183,
+      "step": 120300
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.9718891962974355e-05,
+      "loss": 2.8965,
+      "step": 120400
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.971035283669775e-05,
+      "loss": 2.8146,
+      "step": 120500
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.970181371042115e-05,
+      "loss": 2.9657,
+      "step": 120600
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.969327458414455e-05,
+      "loss": 3.0492,
+      "step": 120700
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.9684735457867955e-05,
+      "loss": 3.0475,
+      "step": 120800
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.9676196331591356e-05,
+      "loss": 2.8418,
+      "step": 120900
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.966765720531476e-05,
+      "loss": 2.9023,
+      "step": 121000
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.965911807903815e-05,
+      "loss": 2.9436,
+      "step": 121100
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.965057895276156e-05,
+      "loss": 3.025,
+      "step": 121200
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.9642039826484956e-05,
+      "loss": 3.0259,
+      "step": 121300
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.963350070020836e-05,
+      "loss": 3.0025,
+      "step": 121400
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.962496157393176e-05,
+      "loss": 2.8463,
+      "step": 121500
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.961642244765516e-05,
+      "loss": 3.0121,
+      "step": 121600
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.9607883321378555e-05,
+      "loss": 2.9603,
+      "step": 121700
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.959934419510196e-05,
+      "loss": 2.9895,
+      "step": 121800
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.959080506882536e-05,
+      "loss": 2.9055,
+      "step": 121900
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.9582265942548766e-05,
+      "loss": 2.9311,
+      "step": 122000
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.957372681627216e-05,
+      "loss": 2.9778,
+      "step": 122100
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.956518768999556e-05,
+      "loss": 2.9807,
+      "step": 122200
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.9556648563718964e-05,
+      "loss": 3.1473,
+      "step": 122300
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.9548109437442366e-05,
+      "loss": 3.1106,
+      "step": 122400
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.953957031116576e-05,
+      "loss": 2.8529,
+      "step": 122500
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.953103118488917e-05,
+      "loss": 3.0184,
+      "step": 122600
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.9522492058612564e-05,
+      "loss": 3.0022,
+      "step": 122700
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.9513952932335965e-05,
+      "loss": 3.1624,
+      "step": 122800
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.950541380605937e-05,
+      "loss": 2.8583,
+      "step": 122900
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.949687467978276e-05,
+      "loss": 2.9685,
+      "step": 123000
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.948833555350617e-05,
+      "loss": 3.1809,
+      "step": 123100
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.9479796427229565e-05,
+      "loss": 3.116,
+      "step": 123200
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.9471257300952966e-05,
+      "loss": 3.0942,
+      "step": 123300
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.946271817467637e-05,
+      "loss": 3.0704,
+      "step": 123400
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.945417904839977e-05,
+      "loss": 3.0432,
+      "step": 123500
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.944563992212317e-05,
+      "loss": 2.9057,
+      "step": 123600
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.943710079584657e-05,
+      "loss": 2.9233,
+      "step": 123700
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.942856166956997e-05,
+      "loss": 2.9568,
+      "step": 123800
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.9420022543293375e-05,
+      "loss": 2.9892,
+      "step": 123900
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.941148341701677e-05,
+      "loss": 3.1241,
+      "step": 124000
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.940294429074017e-05,
+      "loss": 3.1531,
+      "step": 124100
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.939440516446357e-05,
+      "loss": 2.8468,
+      "step": 124200
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.9385866038186975e-05,
+      "loss": 3.1295,
+      "step": 124300
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.937732691191037e-05,
+      "loss": 2.959,
+      "step": 124400
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.936878778563378e-05,
+      "loss": 2.9155,
+      "step": 124500
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.936024865935717e-05,
+      "loss": 3.0969,
+      "step": 124600
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.9351709533080574e-05,
+      "loss": 2.978,
+      "step": 124700
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.9343170406803976e-05,
+      "loss": 3.0562,
+      "step": 124800
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.933463128052738e-05,
+      "loss": 2.9286,
+      "step": 124900
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.932609215425078e-05,
+      "loss": 2.9911,
+      "step": 125000
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.931755302797418e-05,
+      "loss": 2.9229,
+      "step": 125100
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.9309013901697575e-05,
+      "loss": 3.0515,
+      "step": 125200
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.9300474775420984e-05,
+      "loss": 3.0484,
+      "step": 125300
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.929193564914438e-05,
+      "loss": 2.9142,
+      "step": 125400
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.928339652286778e-05,
+      "loss": 3.0388,
+      "step": 125500
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.927485739659118e-05,
+      "loss": 2.9048,
+      "step": 125600
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.926631827031458e-05,
+      "loss": 2.8863,
+      "step": 125700
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.9257779144037985e-05,
+      "loss": 2.9024,
+      "step": 125800
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.9249240017761386e-05,
+      "loss": 2.828,
+      "step": 125900
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.924070089148478e-05,
+      "loss": 2.8643,
+      "step": 126000
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.923216176520819e-05,
+      "loss": 3.0604,
+      "step": 126100
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.9223622638931584e-05,
+      "loss": 3.0527,
+      "step": 126200
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.9215083512654985e-05,
+      "loss": 3.0711,
+      "step": 126300
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.920654438637839e-05,
+      "loss": 2.9576,
+      "step": 126400
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.919800526010179e-05,
+      "loss": 2.8507,
+      "step": 126500
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.918946613382518e-05,
+      "loss": 2.9436,
+      "step": 126600
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.918092700754859e-05,
+      "loss": 3.0334,
+      "step": 126700
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.9172387881271986e-05,
+      "loss": 3.071,
+      "step": 126800
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.9163848754995395e-05,
+      "loss": 2.9778,
+      "step": 126900
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.915530962871879e-05,
+      "loss": 3.2001,
+      "step": 127000
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.914677050244219e-05,
+      "loss": 3.0323,
+      "step": 127100
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.913823137616559e-05,
+      "loss": 2.9486,
+      "step": 127200
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.9129692249888994e-05,
+      "loss": 2.965,
+      "step": 127300
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.912115312361239e-05,
+      "loss": 2.7533,
+      "step": 127400
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.91126139973358e-05,
+      "loss": 2.985,
+      "step": 127500
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.910407487105919e-05,
+      "loss": 2.9597,
+      "step": 127600
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.9095535744782594e-05,
+      "loss": 3.1081,
+      "step": 127700
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.9086996618505995e-05,
+      "loss": 2.8903,
+      "step": 127800
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.90784574922294e-05,
+      "loss": 2.6809,
+      "step": 127900
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.90699183659528e-05,
+      "loss": 3.1497,
+      "step": 128000
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.90613792396762e-05,
+      "loss": 3.027,
+      "step": 128100
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.9052840113399595e-05,
+      "loss": 3.0816,
+      "step": 128200
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.9044300987123e-05,
+      "loss": 2.9045,
+      "step": 128300
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.90357618608464e-05,
+      "loss": 3.1785,
+      "step": 128400
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.90272227345698e-05,
+      "loss": 2.9222,
+      "step": 128500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.90186836082932e-05,
+      "loss": 3.013,
+      "step": 128600
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.90101444820166e-05,
+      "loss": 2.863,
+      "step": 128700
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.9001605355740004e-05,
+      "loss": 3.0084,
+      "step": 128800
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.8993066229463405e-05,
+      "loss": 2.9316,
+      "step": 128900
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.89845271031868e-05,
+      "loss": 2.8692,
+      "step": 129000
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.897598797691021e-05,
+      "loss": 2.8845,
+      "step": 129100
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.89674488506336e-05,
+      "loss": 2.9282,
+      "step": 129200
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.8958909724357005e-05,
+      "loss": 2.9483,
+      "step": 129300
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.8950370598080406e-05,
+      "loss": 3.0091,
+      "step": 129400
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.894183147180381e-05,
+      "loss": 2.9191,
+      "step": 129500
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.89332923455272e-05,
+      "loss": 3.0748,
+      "step": 129600
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.892475321925061e-05,
+      "loss": 2.9653,
+      "step": 129700
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.8916214092974006e-05,
+      "loss": 2.7149,
+      "step": 129800
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.890767496669741e-05,
+      "loss": 3.0052,
+      "step": 129900
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.889913584042081e-05,
+      "loss": 3.1336,
+      "step": 130000
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.889059671414421e-05,
+      "loss": 2.8969,
+      "step": 130100
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.888205758786761e-05,
+      "loss": 2.817,
+      "step": 130200
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.8873518461591014e-05,
+      "loss": 2.995,
+      "step": 130300
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.886497933531441e-05,
+      "loss": 2.8422,
+      "step": 130400
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.885644020903782e-05,
+      "loss": 2.9654,
+      "step": 130500
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.884790108276121e-05,
+      "loss": 2.793,
+      "step": 130600
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.883936195648461e-05,
+      "loss": 3.042,
+      "step": 130700
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.8830822830208015e-05,
+      "loss": 3.0508,
+      "step": 130800
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.8822283703931416e-05,
+      "loss": 2.8977,
+      "step": 130900
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.881374457765482e-05,
+      "loss": 2.8101,
+      "step": 131000
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.880520545137822e-05,
+      "loss": 2.9051,
+      "step": 131100
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.8796666325101614e-05,
+      "loss": 2.9152,
+      "step": 131200
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.878812719882502e-05,
+      "loss": 2.9743,
+      "step": 131300
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.877958807254842e-05,
+      "loss": 3.0049,
+      "step": 131400
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.877104894627182e-05,
+      "loss": 3.0142,
+      "step": 131500
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.876250981999522e-05,
+      "loss": 2.9883,
+      "step": 131600
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.875397069371862e-05,
+      "loss": 2.9499,
+      "step": 131700
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.874543156744202e-05,
+      "loss": 2.8573,
+      "step": 131800
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.8736892441165425e-05,
+      "loss": 2.9085,
+      "step": 131900
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.872835331488882e-05,
+      "loss": 2.8178,
+      "step": 132000
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.871981418861223e-05,
+      "loss": 2.8433,
+      "step": 132100
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.871127506233562e-05,
+      "loss": 2.8564,
+      "step": 132200
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.8702735936059024e-05,
+      "loss": 2.9895,
+      "step": 132300
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.8694196809782426e-05,
+      "loss": 3.0891,
+      "step": 132400
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.868565768350583e-05,
+      "loss": 2.7649,
+      "step": 132500
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.867711855722922e-05,
+      "loss": 3.0051,
+      "step": 132600
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.866857943095263e-05,
+      "loss": 2.9748,
+      "step": 132700
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.8660040304676025e-05,
+      "loss": 2.8772,
+      "step": 132800
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.865150117839943e-05,
+      "loss": 3.038,
+      "step": 132900
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.864296205212283e-05,
+      "loss": 2.9276,
+      "step": 133000
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.863442292584623e-05,
+      "loss": 2.8391,
+      "step": 133100
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.862588379956963e-05,
+      "loss": 2.8938,
+      "step": 133200
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.861734467329303e-05,
+      "loss": 2.5677,
+      "step": 133300
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.860880554701643e-05,
+      "loss": 2.934,
+      "step": 133400
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.8600266420739836e-05,
+      "loss": 2.9272,
+      "step": 133500
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.859172729446323e-05,
+      "loss": 3.0463,
+      "step": 133600
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.858318816818663e-05,
+      "loss": 3.1404,
+      "step": 133700
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.8574649041910034e-05,
+      "loss": 2.9115,
+      "step": 133800
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.8566109915633435e-05,
+      "loss": 2.9446,
+      "step": 133900
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.855757078935684e-05,
+      "loss": 2.7483,
+      "step": 134000
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.854903166308024e-05,
+      "loss": 3.0783,
+      "step": 134100
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.854049253680363e-05,
+      "loss": 2.9553,
+      "step": 134200
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.853195341052704e-05,
+      "loss": 2.896,
+      "step": 134300
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.8523414284250436e-05,
+      "loss": 3.1303,
+      "step": 134400
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.851487515797384e-05,
+      "loss": 3.0356,
+      "step": 134500
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.850633603169724e-05,
+      "loss": 2.8436,
+      "step": 134600
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.849779690542064e-05,
+      "loss": 2.9615,
+      "step": 134700
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.8489257779144036e-05,
+      "loss": 2.8318,
+      "step": 134800
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.8480718652867444e-05,
+      "loss": 3.1643,
+      "step": 134900
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.847217952659084e-05,
+      "loss": 3.0151,
+      "step": 135000
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.846364040031424e-05,
+      "loss": 3.0796,
+      "step": 135100
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.845510127403764e-05,
+      "loss": 2.7781,
+      "step": 135200
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.8446562147761044e-05,
+      "loss": 3.1224,
+      "step": 135300
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.8438023021484445e-05,
+      "loss": 2.6578,
+      "step": 135400
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.842948389520785e-05,
+      "loss": 2.9786,
+      "step": 135500
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.842094476893124e-05,
+      "loss": 2.8441,
+      "step": 135600
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.841240564265465e-05,
+      "loss": 2.7938,
+      "step": 135700
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.8403866516378044e-05,
+      "loss": 3.0032,
+      "step": 135800
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.8395327390101446e-05,
+      "loss": 2.8721,
+      "step": 135900
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.838678826382485e-05,
+      "loss": 2.8084,
+      "step": 136000
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.837824913754825e-05,
+      "loss": 2.8603,
+      "step": 136100
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.836971001127165e-05,
+      "loss": 3.0363,
+      "step": 136200
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.836117088499505e-05,
+      "loss": 2.9738,
+      "step": 136300
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.835263175871845e-05,
+      "loss": 2.9322,
+      "step": 136400
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.8344092632441855e-05,
+      "loss": 3.1189,
+      "step": 136500
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.833555350616525e-05,
+      "loss": 2.8642,
+      "step": 136600
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.832701437988865e-05,
+      "loss": 2.8171,
+      "step": 136700
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.831847525361205e-05,
+      "loss": 3.0146,
+      "step": 136800
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.8309936127335455e-05,
+      "loss": 2.9867,
+      "step": 136900
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.8301397001058856e-05,
+      "loss": 2.8355,
+      "step": 137000
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.829285787478226e-05,
+      "loss": 2.7731,
+      "step": 137100
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.828431874850565e-05,
+      "loss": 2.9714,
+      "step": 137200
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.827577962222906e-05,
+      "loss": 2.9709,
+      "step": 137300
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.8267240495952456e-05,
+      "loss": 3.0123,
+      "step": 137400
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.825870136967586e-05,
+      "loss": 2.9678,
+      "step": 137500
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.825016224339926e-05,
+      "loss": 2.7263,
+      "step": 137600
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.824162311712266e-05,
+      "loss": 2.9291,
+      "step": 137700
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.8233083990846055e-05,
+      "loss": 2.848,
+      "step": 137800
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.8224544864569463e-05,
+      "loss": 2.8997,
+      "step": 137900
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.821600573829286e-05,
+      "loss": 3.0481,
+      "step": 138000
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.820746661201626e-05,
+      "loss": 3.079,
+      "step": 138100
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.819892748573966e-05,
+      "loss": 2.9975,
+      "step": 138200
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.819038835946306e-05,
+      "loss": 3.0374,
+      "step": 138300
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.8181849233186464e-05,
+      "loss": 2.912,
+      "step": 138400
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.8173310106909866e-05,
+      "loss": 3.0378,
+      "step": 138500
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.816477098063326e-05,
+      "loss": 2.9073,
+      "step": 138600
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.815623185435667e-05,
+      "loss": 2.9153,
+      "step": 138700
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.8147692728080064e-05,
+      "loss": 2.8937,
+      "step": 138800
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.8139153601803465e-05,
+      "loss": 2.9533,
+      "step": 138900
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.813061447552687e-05,
+      "loss": 2.923,
+      "step": 139000
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.812207534925027e-05,
+      "loss": 2.9185,
+      "step": 139100
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.811353622297367e-05,
+      "loss": 3.0708,
+      "step": 139200
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.810499709669707e-05,
+      "loss": 3.0239,
+      "step": 139300
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.8096457970420466e-05,
+      "loss": 3.0284,
+      "step": 139400
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.8087918844143875e-05,
+      "loss": 2.762,
+      "step": 139500
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.807937971786727e-05,
+      "loss": 2.975,
+      "step": 139600
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.807084059159067e-05,
+      "loss": 3.1,
+      "step": 139700
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.806230146531407e-05,
+      "loss": 2.9565,
+      "step": 139800
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.8053762339037474e-05,
+      "loss": 2.8751,
+      "step": 139900
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.804522321276087e-05,
+      "loss": 2.9852,
+      "step": 140000
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.803668408648428e-05,
+      "loss": 2.8591,
+      "step": 140100
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.802814496020767e-05,
+      "loss": 2.8044,
+      "step": 140200
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8019605833931073e-05,
+      "loss": 3.0422,
+      "step": 140300
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8011066707654475e-05,
+      "loss": 2.9013,
+      "step": 140400
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.800252758137787e-05,
+      "loss": 2.9534,
+      "step": 140500
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.799398845510128e-05,
+      "loss": 2.9336,
+      "step": 140600
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.798544932882467e-05,
+      "loss": 3.0396,
+      "step": 140700
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7976910202548074e-05,
+      "loss": 2.8578,
+      "step": 140800
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7968371076271476e-05,
+      "loss": 3.0548,
+      "step": 140900
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.795983194999488e-05,
+      "loss": 2.7049,
+      "step": 141000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.795129282371828e-05,
+      "loss": 2.9947,
+      "step": 141100
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.794275369744168e-05,
+      "loss": 2.8426,
+      "step": 141200
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7934214571165075e-05,
+      "loss": 2.8844,
+      "step": 141300
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7925675444888484e-05,
+      "loss": 2.7712,
+      "step": 141400
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.791713631861188e-05,
+      "loss": 2.8845,
+      "step": 141500
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.790859719233528e-05,
+      "loss": 2.8524,
+      "step": 141600
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.790005806605868e-05,
+      "loss": 2.6535,
+      "step": 141700
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.789151893978208e-05,
+      "loss": 2.9005,
+      "step": 141800
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.7882979813505485e-05,
+      "loss": 2.9857,
+      "step": 141900
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.7874440687228886e-05,
+      "loss": 2.7747,
+      "step": 142000
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.786590156095228e-05,
+      "loss": 3.0335,
+      "step": 142100
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.785736243467569e-05,
+      "loss": 2.82,
+      "step": 142200
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.7848823308399084e-05,
+      "loss": 3.0067,
+      "step": 142300
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.7840284182122486e-05,
+      "loss": 2.9935,
+      "step": 142400
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.783174505584589e-05,
+      "loss": 2.9691,
+      "step": 142500
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.782320592956929e-05,
+      "loss": 2.9787,
+      "step": 142600
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.7814666803292684e-05,
+      "loss": 3.0089,
+      "step": 142700
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.780612767701609e-05,
+      "loss": 3.0469,
+      "step": 142800
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.779758855073949e-05,
+      "loss": 3.0362,
+      "step": 142900
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.778904942446289e-05,
+      "loss": 2.9551,
+      "step": 143000
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.778051029818629e-05,
+      "loss": 2.6902,
+      "step": 143100
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.777197117190969e-05,
+      "loss": 2.8595,
+      "step": 143200
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.776343204563309e-05,
+      "loss": 2.8926,
+      "step": 143300
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.7754892919356494e-05,
+      "loss": 2.8702,
+      "step": 143400
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.774635379307989e-05,
+      "loss": 2.9157,
+      "step": 143500
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.77378146668033e-05,
+      "loss": 2.8203,
+      "step": 143600
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.772927554052669e-05,
+      "loss": 2.9764,
+      "step": 143700
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.7720736414250094e-05,
+      "loss": 3.0338,
+      "step": 143800
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.7712197287973495e-05,
+      "loss": 2.8205,
+      "step": 143900
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.77036581616969e-05,
+      "loss": 2.9368,
+      "step": 144000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.76951190354203e-05,
+      "loss": 3.0391,
+      "step": 144100
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.76865799091437e-05,
+      "loss": 2.9394,
+      "step": 144200
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.7678040782867095e-05,
+      "loss": 3.0125,
+      "step": 144300
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.76695016565905e-05,
+      "loss": 2.9793,
+      "step": 144400
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.76609625303139e-05,
+      "loss": 3.0127,
+      "step": 144500
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.76524234040373e-05,
+      "loss": 2.9295,
+      "step": 144600
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.76438842777607e-05,
+      "loss": 2.926,
+      "step": 144700
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.76353451514841e-05,
+      "loss": 2.8679,
+      "step": 144800
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.76268060252075e-05,
+      "loss": 2.874,
+      "step": 144900
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.7618266898930906e-05,
+      "loss": 2.8243,
+      "step": 145000
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.76097277726543e-05,
+      "loss": 2.6775,
+      "step": 145100
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.76011886463777e-05,
+      "loss": 2.9936,
+      "step": 145200
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.7592649520101103e-05,
+      "loss": 2.87,
+      "step": 145300
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.7584110393824505e-05,
+      "loss": 2.6478,
+      "step": 145400
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.7575571267547907e-05,
+      "loss": 3.1228,
+      "step": 145500
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.756703214127131e-05,
+      "loss": 2.9155,
+      "step": 145600
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.75584930149947e-05,
+      "loss": 2.9365,
+      "step": 145700
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.754995388871811e-05,
+      "loss": 2.9418,
+      "step": 145800
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.7541414762441506e-05,
+      "loss": 2.8428,
+      "step": 145900
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.753287563616491e-05,
+      "loss": 2.9287,
+      "step": 146000
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.752433650988831e-05,
+      "loss": 2.7329,
+      "step": 146100
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.751579738361171e-05,
+      "loss": 2.9242,
+      "step": 146200
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.750725825733511e-05,
+      "loss": 3.0612,
+      "step": 146300
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.7498719131058514e-05,
+      "loss": 2.8653,
+      "step": 146400
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.749018000478191e-05,
+      "loss": 2.989,
+      "step": 146500
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.748164087850532e-05,
+      "loss": 2.9249,
+      "step": 146600
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.747310175222871e-05,
+      "loss": 2.9209,
+      "step": 146700
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.746456262595211e-05,
+      "loss": 2.7595,
+      "step": 146800
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.7456023499675515e-05,
+      "loss": 2.8333,
+      "step": 146900
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.7447484373398916e-05,
+      "loss": 2.886,
+      "step": 147000
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.743894524712232e-05,
+      "loss": 2.806,
+      "step": 147100
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.743040612084572e-05,
+      "loss": 2.8587,
+      "step": 147200
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.7421866994569114e-05,
+      "loss": 2.8406,
+      "step": 147300
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.741332786829252e-05,
+      "loss": 2.7372,
+      "step": 147400
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.740478874201592e-05,
+      "loss": 3.1244,
+      "step": 147500
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.739624961573932e-05,
+      "loss": 2.8076,
+      "step": 147600
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.738771048946272e-05,
+      "loss": 2.7125,
+      "step": 147700
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.737917136318612e-05,
+      "loss": 2.8492,
+      "step": 147800
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.737063223690952e-05,
+      "loss": 2.9047,
+      "step": 147900
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.7362093110632925e-05,
+      "loss": 2.7274,
+      "step": 148000
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.735355398435632e-05,
+      "loss": 2.96,
+      "step": 148100
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.734501485807972e-05,
+      "loss": 2.8808,
+      "step": 148200
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.733647573180312e-05,
+      "loss": 2.749,
+      "step": 148300
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.7327936605526524e-05,
+      "loss": 2.6918,
+      "step": 148400
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.7319397479249926e-05,
+      "loss": 3.0696,
+      "step": 148500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.731085835297333e-05,
+      "loss": 2.9101,
+      "step": 148600
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.730231922669672e-05,
+      "loss": 3.0157,
+      "step": 148700
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.729378010042013e-05,
+      "loss": 3.138,
+      "step": 148800
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.7285240974143525e-05,
+      "loss": 2.9909,
+      "step": 148900
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.727670184786693e-05,
+      "loss": 2.6905,
+      "step": 149000
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.726816272159033e-05,
+      "loss": 2.7996,
+      "step": 149100
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.725962359531373e-05,
+      "loss": 2.8014,
+      "step": 149200
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.725108446903713e-05,
+      "loss": 2.914,
+      "step": 149300
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.724254534276053e-05,
+      "loss": 2.9544,
+      "step": 149400
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.723400621648393e-05,
+      "loss": 2.7879,
+      "step": 149500
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.7225467090207336e-05,
+      "loss": 2.741,
+      "step": 149600
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.721692796393073e-05,
+      "loss": 2.9305,
+      "step": 149700
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.720838883765413e-05,
+      "loss": 2.8774,
+      "step": 149800
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.7199849711377534e-05,
+      "loss": 2.8693,
+      "step": 149900
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.7191310585100936e-05,
+      "loss": 2.8799,
+      "step": 150000
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.718277145882433e-05,
+      "loss": 2.9987,
+      "step": 150100
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.717423233254774e-05,
+      "loss": 3.0406,
+      "step": 150200
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.7165693206271133e-05,
+      "loss": 2.8638,
+      "step": 150300
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.7157154079994535e-05,
+      "loss": 3.0982,
+      "step": 150400
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.7148614953717937e-05,
+      "loss": 2.8802,
+      "step": 150500
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.714007582744134e-05,
+      "loss": 2.8833,
+      "step": 150600
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.713153670116474e-05,
+      "loss": 2.9561,
+      "step": 150700
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.712299757488814e-05,
+      "loss": 2.853,
+      "step": 150800
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.7114458448611536e-05,
+      "loss": 3.0224,
+      "step": 150900
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.7105919322334944e-05,
+      "loss": 3.076,
+      "step": 151000
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.709738019605834e-05,
+      "loss": 2.7872,
+      "step": 151100
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.708884106978174e-05,
+      "loss": 2.9373,
+      "step": 151200
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.708030194350514e-05,
+      "loss": 3.0687,
+      "step": 151300
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.7071762817228544e-05,
+      "loss": 2.9908,
+      "step": 151400
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.7063223690951945e-05,
+      "loss": 3.005,
+      "step": 151500
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.705468456467535e-05,
+      "loss": 2.8163,
+      "step": 151600
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.704614543839874e-05,
+      "loss": 2.804,
+      "step": 151700
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.703760631212215e-05,
+      "loss": 2.8771,
+      "step": 151800
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.7029067185845545e-05,
+      "loss": 2.7535,
+      "step": 151900
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.7020528059568946e-05,
+      "loss": 3.0275,
+      "step": 152000
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.701198893329235e-05,
+      "loss": 3.0363,
+      "step": 152100
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.700344980701575e-05,
+      "loss": 2.7471,
+      "step": 152200
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.699491068073915e-05,
+      "loss": 2.9191,
+      "step": 152300
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.698637155446255e-05,
+      "loss": 3.0247,
+      "step": 152400
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.697783242818595e-05,
+      "loss": 2.8533,
+      "step": 152500
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.6969293301909356e-05,
+      "loss": 2.8196,
+      "step": 152600
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.696075417563275e-05,
+      "loss": 2.9538,
+      "step": 152700
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.695221504935615e-05,
+      "loss": 2.973,
+      "step": 152800
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.694367592307955e-05,
+      "loss": 2.8075,
+      "step": 152900
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.6935136796802955e-05,
+      "loss": 2.8664,
+      "step": 153000
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.692659767052635e-05,
+      "loss": 3.0199,
+      "step": 153100
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.691805854424976e-05,
+      "loss": 2.7264,
+      "step": 153200
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.690951941797315e-05,
+      "loss": 2.7767,
+      "step": 153300
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.6900980291696554e-05,
+      "loss": 2.8702,
+      "step": 153400
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.6892441165419956e-05,
+      "loss": 2.8619,
+      "step": 153500
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.688390203914336e-05,
+      "loss": 2.7582,
+      "step": 153600
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.687536291286676e-05,
+      "loss": 2.8249,
+      "step": 153700
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.686682378659016e-05,
+      "loss": 2.9421,
+      "step": 153800
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.6858284660313555e-05,
+      "loss": 2.9268,
+      "step": 153900
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.6849745534036964e-05,
+      "loss": 2.9982,
+      "step": 154000
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.684120640776036e-05,
+      "loss": 2.8282,
+      "step": 154100
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.683266728148376e-05,
+      "loss": 2.6538,
+      "step": 154200
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.682412815520716e-05,
+      "loss": 2.9332,
+      "step": 154300
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.681558902893056e-05,
+      "loss": 2.9436,
+      "step": 154400
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.6807049902653965e-05,
+      "loss": 2.9657,
+      "step": 154500
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.6798510776377366e-05,
+      "loss": 2.8455,
+      "step": 154600
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.678997165010076e-05,
+      "loss": 3.0626,
+      "step": 154700
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.678143252382417e-05,
+      "loss": 3.0584,
+      "step": 154800
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.6772893397547564e-05,
+      "loss": 3.0656,
+      "step": 154900
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.6764354271270966e-05,
+      "loss": 2.9965,
+      "step": 155000
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.675581514499437e-05,
+      "loss": 3.1522,
+      "step": 155100
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.674727601871777e-05,
+      "loss": 2.9915,
+      "step": 155200
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.6738736892441163e-05,
+      "loss": 2.927,
+      "step": 155300
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.673019776616457e-05,
+      "loss": 3.0846,
+      "step": 155400
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.6721658639887967e-05,
+      "loss": 2.9297,
+      "step": 155500
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.6713119513611375e-05,
+      "loss": 2.8472,
+      "step": 155600
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.670458038733477e-05,
+      "loss": 2.7157,
+      "step": 155700
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.669604126105817e-05,
+      "loss": 2.8304,
+      "step": 155800
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.668750213478157e-05,
+      "loss": 2.9214,
+      "step": 155900
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.6678963008504974e-05,
+      "loss": 2.8294,
+      "step": 156000
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.667042388222837e-05,
+      "loss": 2.7344,
+      "step": 156100
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.666188475595178e-05,
+      "loss": 2.9135,
+      "step": 156200
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.665334562967517e-05,
+      "loss": 2.7938,
+      "step": 156300
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.6644806503398574e-05,
+      "loss": 2.7882,
+      "step": 156400
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.6636267377121975e-05,
+      "loss": 2.7811,
+      "step": 156500
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.662772825084538e-05,
+      "loss": 2.8213,
+      "step": 156600
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.661918912456878e-05,
+      "loss": 2.8907,
+      "step": 156700
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.661064999829218e-05,
+      "loss": 2.8295,
+      "step": 156800
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.6602110872015575e-05,
+      "loss": 2.7955,
+      "step": 156900
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.659357174573898e-05,
+      "loss": 2.9084,
+      "step": 157000
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.658503261946238e-05,
+      "loss": 2.7326,
+      "step": 157100
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.657649349318578e-05,
+      "loss": 2.8411,
+      "step": 157200
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.656795436690918e-05,
+      "loss": 2.9072,
+      "step": 157300
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.655941524063258e-05,
+      "loss": 2.7981,
+      "step": 157400
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.6550876114355984e-05,
+      "loss": 2.7779,
+      "step": 157500
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.6542336988079385e-05,
+      "loss": 3.0451,
+      "step": 157600
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.653379786180278e-05,
+      "loss": 3.1684,
+      "step": 157700
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.652525873552618e-05,
+      "loss": 2.9367,
+      "step": 157800
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.651671960924958e-05,
+      "loss": 2.8674,
+      "step": 157900
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.650818048297298e-05,
+      "loss": 2.9975,
+      "step": 158000
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.6499641356696386e-05,
+      "loss": 3.1238,
+      "step": 158100
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.649110223041978e-05,
+      "loss": 2.8777,
+      "step": 158200
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.648256310414318e-05,
+      "loss": 2.7382,
+      "step": 158300
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.6474023977866584e-05,
+      "loss": 3.0051,
+      "step": 158400
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.6465484851589986e-05,
+      "loss": 2.9901,
+      "step": 158500
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.645694572531339e-05,
+      "loss": 2.9462,
+      "step": 158600
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.644840659903679e-05,
+      "loss": 3.0967,
+      "step": 158700
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.6439867472760184e-05,
+      "loss": 2.8192,
+      "step": 158800
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.643132834648359e-05,
+      "loss": 3.0004,
+      "step": 158900
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.642278922020699e-05,
+      "loss": 2.8679,
+      "step": 159000
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.641425009393039e-05,
+      "loss": 2.6969,
+      "step": 159100
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.640571096765379e-05,
+      "loss": 2.8129,
+      "step": 159200
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.639717184137719e-05,
+      "loss": 2.9315,
+      "step": 159300
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.638863271510059e-05,
+      "loss": 2.8286,
+      "step": 159400
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.6380093588823995e-05,
+      "loss": 2.8636,
+      "step": 159500
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.637155446254739e-05,
+      "loss": 2.8053,
+      "step": 159600
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.63630153362708e-05,
+      "loss": 2.8558,
+      "step": 159700
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.635447620999419e-05,
+      "loss": 2.8891,
+      "step": 159800
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.6345937083717594e-05,
+      "loss": 2.8303,
+      "step": 159900
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.6337397957440996e-05,
+      "loss": 2.8755,
+      "step": 160000
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.63288588311644e-05,
+      "loss": 2.9814,
+      "step": 160100
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.632031970488779e-05,
+      "loss": 2.9474,
+      "step": 160200
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.63117805786112e-05,
+      "loss": 2.7752,
+      "step": 160300
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.6303241452334595e-05,
+      "loss": 2.9008,
+      "step": 160400
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.6294702326058e-05,
+      "loss": 2.8035,
+      "step": 160500
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.62861631997814e-05,
+      "loss": 2.899,
+      "step": 160600
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.62776240735048e-05,
+      "loss": 3.0576,
+      "step": 160700
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.62690849472282e-05,
+      "loss": 2.9951,
+      "step": 160800
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.62605458209516e-05,
+      "loss": 2.7915,
+      "step": 160900
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.6252006694675e-05,
+      "loss": 3.0475,
+      "step": 161000
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.6243467568398406e-05,
+      "loss": 2.8174,
+      "step": 161100
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.62349284421218e-05,
+      "loss": 2.8561,
+      "step": 161200
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.62263893158452e-05,
+      "loss": 2.9166,
+      "step": 161300
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.6217850189568604e-05,
+      "loss": 2.8769,
+      "step": 161400
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.6209311063292005e-05,
+      "loss": 2.9089,
+      "step": 161500
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.620077193701541e-05,
+      "loss": 2.9342,
+      "step": 161600
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.619223281073881e-05,
+      "loss": 2.7861,
+      "step": 161700
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.61836936844622e-05,
+      "loss": 2.9156,
+      "step": 161800
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.617515455818561e-05,
+      "loss": 2.8686,
+      "step": 161900
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.6166615431909006e-05,
+      "loss": 2.8429,
+      "step": 162000
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.615807630563241e-05,
+      "loss": 2.917,
+      "step": 162100
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.614953717935581e-05,
+      "loss": 2.9653,
+      "step": 162200
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.614099805307921e-05,
+      "loss": 2.9225,
+      "step": 162300
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.613245892680261e-05,
+      "loss": 2.9669,
+      "step": 162400
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.6123919800526014e-05,
+      "loss": 2.9289,
+      "step": 162500
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.611538067424941e-05,
+      "loss": 2.8101,
+      "step": 162600
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.610684154797282e-05,
+      "loss": 2.7668,
+      "step": 162700
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.609830242169621e-05,
+      "loss": 2.6915,
+      "step": 162800
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.608976329541961e-05,
+      "loss": 2.8424,
+      "step": 162900
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.6081224169143015e-05,
+      "loss": 2.9561,
+      "step": 163000
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.6072685042866416e-05,
+      "loss": 2.8571,
+      "step": 163100
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.606414591658981e-05,
+      "loss": 3.0197,
+      "step": 163200
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.605560679031322e-05,
+      "loss": 2.8519,
+      "step": 163300
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.6047067664036614e-05,
+      "loss": 2.6463,
+      "step": 163400
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.6038528537760016e-05,
+      "loss": 2.6629,
+      "step": 163500
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.602998941148342e-05,
+      "loss": 2.8949,
+      "step": 163600
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.602145028520682e-05,
+      "loss": 2.709,
+      "step": 163700
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.601291115893022e-05,
+      "loss": 2.7052,
+      "step": 163800
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.600437203265362e-05,
+      "loss": 2.7292,
+      "step": 163900
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.599583290637702e-05,
+      "loss": 2.7062,
+      "step": 164000
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.5987293780100425e-05,
+      "loss": 2.8184,
+      "step": 164100
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.597875465382382e-05,
+      "loss": 2.8715,
+      "step": 164200
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.597021552754722e-05,
+      "loss": 2.9527,
+      "step": 164300
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.596167640127062e-05,
+      "loss": 2.9549,
+      "step": 164400
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.5953137274994025e-05,
+      "loss": 2.7693,
+      "step": 164500
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.5944598148717426e-05,
+      "loss": 2.847,
+      "step": 164600
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.593605902244083e-05,
+      "loss": 2.9589,
+      "step": 164700
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.592751989616422e-05,
+      "loss": 2.7985,
+      "step": 164800
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.591898076988763e-05,
+      "loss": 2.7499,
+      "step": 164900
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.5910441643611026e-05,
+      "loss": 2.8699,
+      "step": 165000
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.590190251733443e-05,
+      "loss": 2.9339,
+      "step": 165100
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.589336339105783e-05,
+      "loss": 2.7762,
+      "step": 165200
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.588482426478123e-05,
+      "loss": 2.6134,
+      "step": 165300
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.587628513850463e-05,
+      "loss": 2.8161,
+      "step": 165400
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.586774601222803e-05,
+      "loss": 2.9087,
+      "step": 165500
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.585920688595143e-05,
+      "loss": 2.7699,
+      "step": 165600
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.5850667759674836e-05,
+      "loss": 2.9153,
+      "step": 165700
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.584212863339823e-05,
+      "loss": 2.9153,
+      "step": 165800
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.583358950712163e-05,
+      "loss": 2.8115,
+      "step": 165900
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.5825050380845034e-05,
+      "loss": 2.6892,
+      "step": 166000
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 3.5816511254568436e-05,
+      "loss": 2.7988,
+      "step": 166100
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 3.580797212829183e-05,
+      "loss": 2.9257,
+      "step": 166200
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 3.579943300201524e-05,
+      "loss": 2.8666,
+      "step": 166300
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 3.5790893875738634e-05,
+      "loss": 2.7994,
+      "step": 166400
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 3.5782354749462035e-05,
+      "loss": 2.9492,
+      "step": 166500
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 3.577381562318544e-05,
+      "loss": 2.8635,
+      "step": 166600
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 3.576527649690884e-05,
+      "loss": 2.868,
+      "step": 166700
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 3.575673737063224e-05,
+      "loss": 2.887,
+      "step": 166800
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 3.574819824435564e-05,
+      "loss": 2.8005,
+      "step": 166900
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 3.5739659118079036e-05,
+      "loss": 2.905,
+      "step": 167000
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 3.5731119991802444e-05,
+      "loss": 2.7722,
+      "step": 167100
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 3.572258086552584e-05,
+      "loss": 2.836,
+      "step": 167200
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.571404173924924e-05,
+      "loss": 2.9579,
+      "step": 167300
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.570550261297264e-05,
+      "loss": 2.7944,
+      "step": 167400
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.5696963486696044e-05,
+      "loss": 2.7011,
+      "step": 167500
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.5688424360419445e-05,
+      "loss": 2.8905,
+      "step": 167600
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.567988523414285e-05,
+      "loss": 2.9255,
+      "step": 167700
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.567134610786624e-05,
+      "loss": 2.919,
+      "step": 167800
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.566280698158965e-05,
+      "loss": 2.9211,
+      "step": 167900
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.5654267855313045e-05,
+      "loss": 2.7088,
+      "step": 168000
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.5645728729036446e-05,
+      "loss": 2.6552,
+      "step": 168100
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.563718960275985e-05,
+      "loss": 2.9929,
+      "step": 168200
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.562865047648325e-05,
+      "loss": 2.8072,
+      "step": 168300
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.5620111350206644e-05,
+      "loss": 2.9567,
+      "step": 168400
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.561157222393005e-05,
+      "loss": 2.8903,
+      "step": 168500
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.560303309765345e-05,
+      "loss": 2.8801,
+      "step": 168600
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.559449397137685e-05,
+      "loss": 2.749,
+      "step": 168700
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.558595484510025e-05,
+      "loss": 2.8069,
+      "step": 168800
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.557741571882365e-05,
+      "loss": 2.8033,
+      "step": 168900
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.5568876592547054e-05,
+      "loss": 2.6476,
+      "step": 169000
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.5560337466270455e-05,
+      "loss": 3.0062,
+      "step": 169100
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.555179833999385e-05,
+      "loss": 2.9767,
+      "step": 169200
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.554325921371726e-05,
+      "loss": 2.8588,
+      "step": 169300
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.553472008744065e-05,
+      "loss": 2.8502,
+      "step": 169400
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.5526180961164055e-05,
+      "loss": 3.108,
+      "step": 169500
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.5517641834887456e-05,
+      "loss": 2.7552,
+      "step": 169600
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.550910270861086e-05,
+      "loss": 3.0094,
+      "step": 169700
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.550056358233426e-05,
+      "loss": 2.8699,
+      "step": 169800
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.549202445605766e-05,
+      "loss": 2.8333,
+      "step": 169900
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.5483485329781055e-05,
+      "loss": 2.8573,
+      "step": 170000
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.5474946203504464e-05,
+      "loss": 2.8919,
+      "step": 170100
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.546640707722786e-05,
+      "loss": 3.0407,
+      "step": 170200
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.545786795095126e-05,
+      "loss": 2.8344,
+      "step": 170300
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.544932882467466e-05,
+      "loss": 2.8432,
+      "step": 170400
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.544078969839806e-05,
+      "loss": 2.646,
+      "step": 170500
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.5432250572121465e-05,
+      "loss": 2.7221,
+      "step": 170600
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.5423711445844866e-05,
+      "loss": 2.8708,
+      "step": 170700
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.541517231956826e-05,
+      "loss": 2.8989,
+      "step": 170800
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.540663319329167e-05,
+      "loss": 2.9317,
+      "step": 170900
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.5398094067015064e-05,
+      "loss": 2.7789,
+      "step": 171000
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.5389554940738466e-05,
+      "loss": 2.9044,
+      "step": 171100
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.538101581446187e-05,
+      "loss": 2.8208,
+      "step": 171200
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.537247668818527e-05,
+      "loss": 2.7788,
+      "step": 171300
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.5363937561908664e-05,
+      "loss": 2.7944,
+      "step": 171400
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.535539843563207e-05,
+      "loss": 2.837,
+      "step": 171500
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.534685930935547e-05,
+      "loss": 2.9127,
+      "step": 171600
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.533832018307887e-05,
+      "loss": 2.847,
+      "step": 171700
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.532978105680227e-05,
+      "loss": 2.853,
+      "step": 171800
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.532124193052567e-05,
+      "loss": 2.6868,
+      "step": 171900
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.531270280424907e-05,
+      "loss": 3.0012,
+      "step": 172000
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.5304163677972474e-05,
+      "loss": 3.0031,
+      "step": 172100
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.529562455169587e-05,
+      "loss": 2.8709,
+      "step": 172200
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.528708542541928e-05,
+      "loss": 2.707,
+      "step": 172300
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.527854629914267e-05,
+      "loss": 2.6456,
+      "step": 172400
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.5270007172866074e-05,
+      "loss": 2.6917,
+      "step": 172500
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.5261468046589475e-05,
+      "loss": 2.7597,
+      "step": 172600
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.525292892031288e-05,
+      "loss": 2.5973,
+      "step": 172700
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.524438979403628e-05,
+      "loss": 2.7756,
+      "step": 172800
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.523585066775968e-05,
+      "loss": 2.8136,
+      "step": 172900
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.5227311541483075e-05,
+      "loss": 2.7833,
+      "step": 173000
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.521877241520648e-05,
+      "loss": 2.7723,
+      "step": 173100
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.521023328892988e-05,
+      "loss": 2.8283,
+      "step": 173200
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.520169416265328e-05,
+      "loss": 3.0486,
+      "step": 173300
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.519315503637668e-05,
+      "loss": 2.9591,
+      "step": 173400
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.518461591010008e-05,
+      "loss": 2.7472,
+      "step": 173500
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.517607678382348e-05,
+      "loss": 2.8585,
+      "step": 173600
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.5167537657546886e-05,
+      "loss": 2.9055,
+      "step": 173700
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.515899853127028e-05,
+      "loss": 2.8869,
+      "step": 173800
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.515045940499368e-05,
+      "loss": 2.9449,
+      "step": 173900
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.5141920278717084e-05,
+      "loss": 2.7024,
+      "step": 174000
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.5133381152440485e-05,
+      "loss": 2.931,
+      "step": 174100
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.512484202616389e-05,
+      "loss": 2.8525,
+      "step": 174200
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.511630289988729e-05,
+      "loss": 2.9064,
+      "step": 174300
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.510776377361068e-05,
+      "loss": 2.6678,
+      "step": 174400
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.509922464733409e-05,
+      "loss": 2.9454,
+      "step": 174500
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.5090685521057486e-05,
+      "loss": 2.7788,
+      "step": 174600
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.508214639478089e-05,
+      "loss": 2.949,
+      "step": 174700
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.507360726850429e-05,
+      "loss": 2.8295,
+      "step": 174800
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.506506814222769e-05,
+      "loss": 2.8091,
+      "step": 174900
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.505652901595109e-05,
+      "loss": 2.8725,
+      "step": 175000
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.5047989889674494e-05,
+      "loss": 2.9462,
+      "step": 175100
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.503945076339789e-05,
+      "loss": 2.9779,
+      "step": 175200
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.50309116371213e-05,
+      "loss": 2.9337,
+      "step": 175300
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.502237251084469e-05,
+      "loss": 2.7945,
+      "step": 175400
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.501383338456809e-05,
+      "loss": 2.8266,
+      "step": 175500
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.5005294258291495e-05,
+      "loss": 2.7534,
+      "step": 175600
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.499675513201489e-05,
+      "loss": 2.9093,
+      "step": 175700
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.49882160057383e-05,
+      "loss": 2.8192,
+      "step": 175800
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.497967687946169e-05,
+      "loss": 2.7516,
+      "step": 175900
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.4971137753185094e-05,
+      "loss": 2.8898,
+      "step": 176000
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.4962598626908496e-05,
+      "loss": 2.9933,
+      "step": 176100
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.49540595006319e-05,
+      "loss": 3.0046,
+      "step": 176200
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.494552037435529e-05,
+      "loss": 2.8737,
+      "step": 176300
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.49369812480787e-05,
+      "loss": 2.9585,
+      "step": 176400
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.4928442121802095e-05,
+      "loss": 2.7224,
+      "step": 176500
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.49199029955255e-05,
+      "loss": 2.9317,
+      "step": 176600
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.49113638692489e-05,
+      "loss": 2.896,
+      "step": 176700
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.49028247429723e-05,
+      "loss": 2.7501,
+      "step": 176800
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.48942856166957e-05,
+      "loss": 2.8642,
+      "step": 176900
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.48857464904191e-05,
+      "loss": 3.2009,
+      "step": 177000
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3.48772073641425e-05,
+      "loss": 2.8458,
+      "step": 177100
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3.4868668237865906e-05,
+      "loss": 2.7384,
+      "step": 177200
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3.48601291115893e-05,
+      "loss": 2.9411,
+      "step": 177300
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3.48515899853127e-05,
+      "loss": 2.8195,
+      "step": 177400
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3.4843050859036104e-05,
+      "loss": 2.8032,
+      "step": 177500
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3.4834511732759505e-05,
+      "loss": 2.8678,
+      "step": 177600
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3.482597260648291e-05,
+      "loss": 2.7897,
+      "step": 177700
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3.481743348020631e-05,
+      "loss": 2.8494,
+      "step": 177800
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.48088943539297e-05,
+      "loss": 2.9871,
+      "step": 177900
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.480035522765311e-05,
+      "loss": 2.8696,
+      "step": 178000
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.4791816101376506e-05,
+      "loss": 2.8582,
+      "step": 178100
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.478327697509991e-05,
+      "loss": 2.8457,
+      "step": 178200
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.477473784882331e-05,
+      "loss": 2.8258,
+      "step": 178300
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.476619872254671e-05,
+      "loss": 2.6707,
+      "step": 178400
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.4757659596270106e-05,
+      "loss": 2.8959,
+      "step": 178500
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.4749120469993514e-05,
+      "loss": 2.7047,
+      "step": 178600
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.474058134371691e-05,
+      "loss": 2.8022,
+      "step": 178700
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.473204221744031e-05,
+      "loss": 2.9514,
+      "step": 178800
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.472350309116371e-05,
+      "loss": 2.7032,
+      "step": 178900
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4714963964887114e-05,
+      "loss": 2.7697,
+      "step": 179000
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4706424838610515e-05,
+      "loss": 2.6671,
+      "step": 179100
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4697885712333917e-05,
+      "loss": 2.9532,
+      "step": 179200
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.468934658605731e-05,
+      "loss": 2.895,
+      "step": 179300
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.468080745978072e-05,
+      "loss": 2.6902,
+      "step": 179400
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.4672268333504114e-05,
+      "loss": 2.9458,
+      "step": 179500
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.4663729207227516e-05,
+      "loss": 2.8755,
+      "step": 179600
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.465519008095092e-05,
+      "loss": 2.7573,
+      "step": 179700
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.464665095467432e-05,
+      "loss": 2.971,
+      "step": 179800
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.463811182839772e-05,
+      "loss": 2.7879,
+      "step": 179900
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.462957270212112e-05,
+      "loss": 2.742,
+      "step": 180000
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.462103357584452e-05,
+      "loss": 2.8963,
+      "step": 180100
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.4612494449567925e-05,
+      "loss": 2.8663,
+      "step": 180200
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.460395532329132e-05,
+      "loss": 2.8793,
+      "step": 180300
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.459541619701472e-05,
+      "loss": 2.8008,
+      "step": 180400
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.458687707073812e-05,
+      "loss": 2.766,
+      "step": 180500
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.4578337944461525e-05,
+      "loss": 2.9717,
+      "step": 180600
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.4569798818184926e-05,
+      "loss": 2.7027,
+      "step": 180700
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.456125969190833e-05,
+      "loss": 2.6709,
+      "step": 180800
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.455272056563172e-05,
+      "loss": 2.7249,
+      "step": 180900
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.454418143935513e-05,
+      "loss": 2.7697,
+      "step": 181000
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.4535642313078526e-05,
+      "loss": 2.771,
+      "step": 181100
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.452710318680193e-05,
+      "loss": 2.6931,
+      "step": 181200
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.451856406052533e-05,
+      "loss": 2.8809,
+      "step": 181300
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.451002493424873e-05,
+      "loss": 2.9175,
+      "step": 181400
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.4501485807972125e-05,
+      "loss": 2.838,
+      "step": 181500
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.4492946681695533e-05,
+      "loss": 2.7583,
+      "step": 181600
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.448440755541893e-05,
+      "loss": 2.7249,
+      "step": 181700
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.447586842914233e-05,
+      "loss": 2.9048,
+      "step": 181800
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.446732930286573e-05,
+      "loss": 2.6823,
+      "step": 181900
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.445879017658913e-05,
+      "loss": 2.8402,
+      "step": 182000
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.4450251050312534e-05,
+      "loss": 2.8399,
+      "step": 182100
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.4441711924035936e-05,
+      "loss": 2.8171,
+      "step": 182200
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.443317279775933e-05,
+      "loss": 2.7929,
+      "step": 182300
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.442463367148274e-05,
+      "loss": 2.8139,
+      "step": 182400
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.4416094545206134e-05,
+      "loss": 2.6094,
+      "step": 182500
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.4407555418929535e-05,
+      "loss": 2.8512,
+      "step": 182600
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.439901629265294e-05,
+      "loss": 2.9651,
+      "step": 182700
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.439047716637634e-05,
+      "loss": 2.8471,
+      "step": 182800
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.438193804009974e-05,
+      "loss": 2.7212,
+      "step": 182900
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.437339891382314e-05,
+      "loss": 2.7698,
+      "step": 183000
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.4364859787546536e-05,
+      "loss": 2.7476,
+      "step": 183100
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.4356320661269945e-05,
+      "loss": 3.0972,
+      "step": 183200
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.434778153499334e-05,
+      "loss": 2.8557,
+      "step": 183300
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.433924240871674e-05,
+      "loss": 2.9168,
+      "step": 183400
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.433070328244014e-05,
+      "loss": 2.893,
+      "step": 183500
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.4322164156163544e-05,
+      "loss": 2.8344,
+      "step": 183600
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.431362502988694e-05,
+      "loss": 2.7075,
+      "step": 183700
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.430508590361035e-05,
+      "loss": 2.7563,
+      "step": 183800
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.429654677733374e-05,
+      "loss": 2.7989,
+      "step": 183900
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.4288007651057143e-05,
+      "loss": 2.8689,
+      "step": 184000
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.4279468524780545e-05,
+      "loss": 2.7793,
+      "step": 184100
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.4270929398503947e-05,
+      "loss": 2.935,
+      "step": 184200
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.426239027222735e-05,
+      "loss": 2.8373,
+      "step": 184300
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.425385114595075e-05,
+      "loss": 2.7941,
+      "step": 184400
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.4245312019674144e-05,
+      "loss": 2.8139,
+      "step": 184500
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.423677289339755e-05,
+      "loss": 2.9945,
+      "step": 184600
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.422823376712095e-05,
+      "loss": 2.8065,
+      "step": 184700
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.421969464084435e-05,
+      "loss": 2.7373,
+      "step": 184800
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.421115551456775e-05,
+      "loss": 2.9016,
+      "step": 184900
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.420261638829115e-05,
+      "loss": 2.7924,
+      "step": 185000
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.4194077262014554e-05,
+      "loss": 2.6757,
+      "step": 185100
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.4185538135737955e-05,
+      "loss": 2.98,
+      "step": 185200
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.417699900946135e-05,
+      "loss": 2.6377,
+      "step": 185300
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.416845988318476e-05,
+      "loss": 2.7265,
+      "step": 185400
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.415992075690815e-05,
+      "loss": 3.0404,
+      "step": 185500
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.4151381630631555e-05,
+      "loss": 2.854,
+      "step": 185600
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.4142842504354956e-05,
+      "loss": 2.8406,
+      "step": 185700
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.413430337807836e-05,
+      "loss": 2.8111,
+      "step": 185800
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.412576425180176e-05,
+      "loss": 2.713,
+      "step": 185900
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.411722512552516e-05,
+      "loss": 2.9278,
+      "step": 186000
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.4108685999248556e-05,
+      "loss": 2.7635,
+      "step": 186100
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.4100146872971964e-05,
+      "loss": 2.7555,
+      "step": 186200
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.409160774669536e-05,
+      "loss": 2.7749,
+      "step": 186300
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 3.408306862041876e-05,
+      "loss": 2.7755,
+      "step": 186400
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 3.407452949414216e-05,
+      "loss": 2.8913,
+      "step": 186500
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 3.4065990367865563e-05,
+      "loss": 2.9017,
+      "step": 186600
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 3.405745124158896e-05,
+      "loss": 2.6779,
+      "step": 186700
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 3.4048912115312367e-05,
+      "loss": 2.8349,
+      "step": 186800
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 3.404037298903576e-05,
+      "loss": 3.0744,
+      "step": 186900
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 3.403183386275916e-05,
+      "loss": 2.8017,
+      "step": 187000
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 3.4023294736482564e-05,
+      "loss": 2.6702,
+      "step": 187100
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 3.4014755610205966e-05,
+      "loss": 2.7281,
+      "step": 187200
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 3.400621648392937e-05,
+      "loss": 2.7488,
+      "step": 187300
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 3.399767735765277e-05,
+      "loss": 2.7947,
+      "step": 187400
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 3.3989138231376164e-05,
+      "loss": 2.8185,
+      "step": 187500
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.398059910509957e-05,
+      "loss": 2.8457,
+      "step": 187600
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.397205997882297e-05,
+      "loss": 2.9416,
+      "step": 187700
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.396352085254637e-05,
+      "loss": 2.7752,
+      "step": 187800
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.395498172626977e-05,
+      "loss": 2.6326,
+      "step": 187900
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.394644259999317e-05,
+      "loss": 2.7806,
+      "step": 188000
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.393790347371657e-05,
+      "loss": 2.8613,
+      "step": 188100
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.3929364347439975e-05,
+      "loss": 2.9939,
+      "step": 188200
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.392082522116337e-05,
+      "loss": 2.9626,
+      "step": 188300
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.391228609488678e-05,
+      "loss": 2.8134,
+      "step": 188400
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.390374696861017e-05,
+      "loss": 2.6093,
+      "step": 188500
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.3895207842333574e-05,
+      "loss": 2.837,
+      "step": 188600
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.3886668716056976e-05,
+      "loss": 2.8663,
+      "step": 188700
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.387812958978038e-05,
+      "loss": 2.9026,
+      "step": 188800
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.386959046350377e-05,
+      "loss": 2.7949,
+      "step": 188900
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.386105133722718e-05,
+      "loss": 2.7179,
+      "step": 189000
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.3852512210950575e-05,
+      "loss": 2.6774,
+      "step": 189100
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.384397308467398e-05,
+      "loss": 2.7785,
+      "step": 189200
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.383543395839738e-05,
+      "loss": 2.9502,
+      "step": 189300
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.382689483212078e-05,
+      "loss": 2.7294,
+      "step": 189400
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.381835570584418e-05,
+      "loss": 2.7518,
+      "step": 189500
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.380981657956758e-05,
+      "loss": 2.6775,
+      "step": 189600
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.380127745329098e-05,
+      "loss": 2.7685,
+      "step": 189700
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.3792738327014386e-05,
+      "loss": 2.6363,
+      "step": 189800
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.378419920073778e-05,
+      "loss": 2.8164,
+      "step": 189900
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.377566007446118e-05,
+      "loss": 2.8068,
+      "step": 190000
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.3767120948184584e-05,
+      "loss": 2.8893,
+      "step": 190100
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.3758581821907985e-05,
+      "loss": 3.1823,
+      "step": 190200
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.375004269563139e-05,
+      "loss": 2.8149,
+      "step": 190300
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.374150356935479e-05,
+      "loss": 2.8773,
+      "step": 190400
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.373296444307818e-05,
+      "loss": 2.8124,
+      "step": 190500
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.372442531680159e-05,
+      "loss": 2.6555,
+      "step": 190600
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.3715886190524986e-05,
+      "loss": 2.5526,
+      "step": 190700
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.370734706424839e-05,
+      "loss": 2.6868,
+      "step": 190800
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.369880793797179e-05,
+      "loss": 2.7514,
+      "step": 190900
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.369026881169519e-05,
+      "loss": 2.8321,
+      "step": 191000
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.368172968541859e-05,
+      "loss": 2.6201,
+      "step": 191100
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.3673190559141994e-05,
+      "loss": 2.8815,
+      "step": 191200
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.366465143286539e-05,
+      "loss": 2.8936,
+      "step": 191300
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.36561123065888e-05,
+      "loss": 3.011,
+      "step": 191400
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.364757318031219e-05,
+      "loss": 2.8954,
+      "step": 191500
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.363903405403559e-05,
+      "loss": 2.7834,
+      "step": 191600
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.3630494927758995e-05,
+      "loss": 2.6221,
+      "step": 191700
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.3621955801482396e-05,
+      "loss": 2.842,
+      "step": 191800
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.361341667520579e-05,
+      "loss": 2.7935,
+      "step": 191900
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.36048775489292e-05,
+      "loss": 2.9066,
+      "step": 192000
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.3596338422652594e-05,
+      "loss": 2.8521,
+      "step": 192100
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.3587799296375996e-05,
+      "loss": 2.7344,
+      "step": 192200
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.35792601700994e-05,
+      "loss": 2.7969,
+      "step": 192300
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.35707210438228e-05,
+      "loss": 2.8141,
+      "step": 192400
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.35621819175462e-05,
+      "loss": 2.8277,
+      "step": 192500
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.35536427912696e-05,
+      "loss": 3.0706,
+      "step": 192600
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.3545103664993e-05,
+      "loss": 2.6699,
+      "step": 192700
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.3536564538716405e-05,
+      "loss": 2.7326,
+      "step": 192800
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.35280254124398e-05,
+      "loss": 2.9995,
+      "step": 192900
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.35194862861632e-05,
+      "loss": 2.6497,
+      "step": 193000
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.35109471598866e-05,
+      "loss": 2.7172,
+      "step": 193100
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.350240803361e-05,
+      "loss": 2.6651,
+      "step": 193200
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.3493868907333406e-05,
+      "loss": 2.6778,
+      "step": 193300
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.34853297810568e-05,
+      "loss": 2.9511,
+      "step": 193400
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.34767906547802e-05,
+      "loss": 2.8123,
+      "step": 193500
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.3468251528503604e-05,
+      "loss": 2.7655,
+      "step": 193600
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.3459712402227006e-05,
+      "loss": 2.7086,
+      "step": 193700
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.345117327595041e-05,
+      "loss": 3.0242,
+      "step": 193800
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.344263414967381e-05,
+      "loss": 2.7939,
+      "step": 193900
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.3434095023397203e-05,
+      "loss": 2.7467,
+      "step": 194000
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.342555589712061e-05,
+      "loss": 2.789,
+      "step": 194100
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.3417016770844007e-05,
+      "loss": 2.9062,
+      "step": 194200
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.340847764456741e-05,
+      "loss": 2.9673,
+      "step": 194300
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.339993851829081e-05,
+      "loss": 2.7477,
+      "step": 194400
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.339139939201421e-05,
+      "loss": 2.7192,
+      "step": 194500
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.3382860265737606e-05,
+      "loss": 2.88,
+      "step": 194600
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.3374321139461014e-05,
+      "loss": 2.7896,
+      "step": 194700
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.336578201318441e-05,
+      "loss": 2.6271,
+      "step": 194800
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.335724288690781e-05,
+      "loss": 2.864,
+      "step": 194900
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.334870376063121e-05,
+      "loss": 2.8348,
+      "step": 195000
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.3340164634354614e-05,
+      "loss": 2.8911,
+      "step": 195100
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.3331625508078015e-05,
+      "loss": 2.7296,
+      "step": 195200
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.332308638180142e-05,
+      "loss": 2.6448,
+      "step": 195300
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.331454725552481e-05,
+      "loss": 2.948,
+      "step": 195400
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.330600812924822e-05,
+      "loss": 2.8449,
+      "step": 195500
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.3297469002971615e-05,
+      "loss": 2.5992,
+      "step": 195600
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.3288929876695016e-05,
+      "loss": 2.7582,
+      "step": 195700
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.328039075041842e-05,
+      "loss": 2.842,
+      "step": 195800
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.327185162414182e-05,
+      "loss": 2.8697,
+      "step": 195900
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.326331249786522e-05,
+      "loss": 2.8471,
+      "step": 196000
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.325477337158862e-05,
+      "loss": 2.8698,
+      "step": 196100
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.324623424531202e-05,
+      "loss": 2.6487,
+      "step": 196200
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.3237695119035425e-05,
+      "loss": 2.7105,
+      "step": 196300
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.322915599275882e-05,
+      "loss": 2.8269,
+      "step": 196400
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.322061686648222e-05,
+      "loss": 2.7231,
+      "step": 196500
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.321207774020562e-05,
+      "loss": 2.7965,
+      "step": 196600
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.3203538613929025e-05,
+      "loss": 2.7904,
+      "step": 196700
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.319499948765242e-05,
+      "loss": 2.9175,
+      "step": 196800
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.318646036137583e-05,
+      "loss": 2.6893,
+      "step": 196900
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.317792123509922e-05,
+      "loss": 2.6112,
+      "step": 197000
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.3169382108822624e-05,
+      "loss": 2.7459,
+      "step": 197100
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.3160842982546026e-05,
+      "loss": 2.8013,
+      "step": 197200
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.315230385626943e-05,
+      "loss": 2.6266,
+      "step": 197300
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.314376472999283e-05,
+      "loss": 3.0153,
+      "step": 197400
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.313522560371623e-05,
+      "loss": 2.8443,
+      "step": 197500
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.3126686477439625e-05,
+      "loss": 2.9788,
+      "step": 197600
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.3118147351163034e-05,
+      "loss": 2.8516,
+      "step": 197700
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.310960822488643e-05,
+      "loss": 2.7007,
+      "step": 197800
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.310106909860983e-05,
+      "loss": 2.8234,
+      "step": 197900
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.309252997233323e-05,
+      "loss": 2.8521,
+      "step": 198000
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.308399084605663e-05,
+      "loss": 2.7973,
+      "step": 198100
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 3.3075451719780035e-05,
+      "loss": 2.8427,
+      "step": 198200
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 3.3066912593503436e-05,
+      "loss": 2.7293,
+      "step": 198300
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 3.305837346722683e-05,
+      "loss": 2.7858,
+      "step": 198400
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 3.304983434095024e-05,
+      "loss": 2.7012,
+      "step": 198500
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 3.3041295214673634e-05,
+      "loss": 2.5829,
+      "step": 198600
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 3.3032756088397036e-05,
+      "loss": 2.7131,
+      "step": 198700
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 3.302421696212044e-05,
+      "loss": 2.8127,
+      "step": 198800
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 3.301567783584384e-05,
+      "loss": 2.8022,
+      "step": 198900
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 3.300713870956724e-05,
+      "loss": 2.5661,
+      "step": 199000
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 3.299859958329064e-05,
+      "loss": 2.697,
+      "step": 199100
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 3.2990060457014037e-05,
+      "loss": 2.8766,
+      "step": 199200
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 3.2981521330737445e-05,
+      "loss": 2.7826,
+      "step": 199300
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 3.297298220446084e-05,
+      "loss": 2.7409,
+      "step": 199400
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 3.296444307818424e-05,
+      "loss": 2.773,
+      "step": 199500
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 3.295590395190764e-05,
+      "loss": 2.9256,
+      "step": 199600
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 3.2947364825631044e-05,
+      "loss": 2.817,
+      "step": 199700
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 3.293882569935444e-05,
+      "loss": 2.7419,
+      "step": 199800
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 3.293028657307785e-05,
+      "loss": 2.7595,
+      "step": 199900
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 3.292174744680124e-05,
+      "loss": 2.7349,
+      "step": 200000
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 3.2913208320524644e-05,
+      "loss": 2.6855,
+      "step": 200100
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 3.2904669194248045e-05,
+      "loss": 2.8252,
+      "step": 200200
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 3.289613006797145e-05,
+      "loss": 2.8531,
+      "step": 200300
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 3.288759094169485e-05,
+      "loss": 2.7198,
+      "step": 200400
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 3.287905181541825e-05,
+      "loss": 2.8435,
+      "step": 200500
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 3.2870512689141645e-05,
+      "loss": 2.7236,
+      "step": 200600
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 3.286197356286505e-05,
+      "loss": 2.7964,
+      "step": 200700
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 3.285343443658845e-05,
+      "loss": 2.8058,
+      "step": 200800
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 3.284489531031185e-05,
+      "loss": 2.6538,
+      "step": 200900
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 3.283635618403525e-05,
+      "loss": 2.8613,
+      "step": 201000
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 3.282781705775865e-05,
+      "loss": 2.8554,
+      "step": 201100
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 3.2819277931482054e-05,
+      "loss": 2.6762,
+      "step": 201200
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3.2810738805205455e-05,
+      "loss": 2.772,
+      "step": 201300
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3.280219967892885e-05,
+      "loss": 2.5641,
+      "step": 201400
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3.279366055265226e-05,
+      "loss": 2.8266,
+      "step": 201500
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3.278512142637565e-05,
+      "loss": 3.0291,
+      "step": 201600
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 3.2776582300099055e-05,
+      "loss": 2.648,
+      "step": 201700
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 3.2768043173822456e-05,
+      "loss": 2.8076,
+      "step": 201800
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 3.275950404754586e-05,
+      "loss": 2.7914,
+      "step": 201900
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 3.275096492126925e-05,
+      "loss": 2.8401,
+      "step": 202000
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 3.274242579499266e-05,
+      "loss": 2.6394,
+      "step": 202100
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 3.2733886668716056e-05,
+      "loss": 2.8558,
+      "step": 202200
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 3.272534754243946e-05,
+      "loss": 2.7711,
+      "step": 202300
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 3.271680841616286e-05,
+      "loss": 2.9304,
+      "step": 202400
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 3.270826928988626e-05,
+      "loss": 2.6178,
+      "step": 202500
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 3.269973016360966e-05,
+      "loss": 2.8751,
+      "step": 202600
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 3.2691191037333064e-05,
+      "loss": 2.7741,
+      "step": 202700
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 3.268265191105646e-05,
+      "loss": 2.7994,
+      "step": 202800
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 3.267411278477987e-05,
+      "loss": 2.7062,
+      "step": 202900
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 3.266557365850326e-05,
+      "loss": 2.804,
+      "step": 203000
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 3.265703453222666e-05,
+      "loss": 2.6745,
+      "step": 203100
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.2648495405950065e-05,
+      "loss": 2.8282,
+      "step": 203200
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.2639956279673466e-05,
+      "loss": 2.6831,
+      "step": 203300
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.263141715339687e-05,
+      "loss": 2.7802,
+      "step": 203400
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.262287802712027e-05,
+      "loss": 2.7884,
+      "step": 203500
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.2614338900843664e-05,
+      "loss": 2.6893,
+      "step": 203600
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.260579977456707e-05,
+      "loss": 2.8855,
+      "step": 203700
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.259726064829047e-05,
+      "loss": 2.6815,
+      "step": 203800
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.258872152201387e-05,
+      "loss": 2.7475,
+      "step": 203900
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.258018239573727e-05,
+      "loss": 2.7349,
+      "step": 204000
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.257164326946067e-05,
+      "loss": 2.6756,
+      "step": 204100
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.256310414318407e-05,
+      "loss": 2.6423,
+      "step": 204200
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.2554565016907475e-05,
+      "loss": 2.9113,
+      "step": 204300
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 3.254602589063087e-05,
+      "loss": 2.797,
+      "step": 204400
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 3.253748676435428e-05,
+      "loss": 2.7763,
+      "step": 204500
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 3.252894763807767e-05,
+      "loss": 2.8551,
+      "step": 204600
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 3.2520408511801074e-05,
+      "loss": 2.7685,
+      "step": 204700
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.2511869385524476e-05,
+      "loss": 2.6483,
+      "step": 204800
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.250333025924788e-05,
+      "loss": 2.7825,
+      "step": 204900
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.249479113297127e-05,
+      "loss": 2.5934,
+      "step": 205000
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.248625200669468e-05,
+      "loss": 2.7495,
+      "step": 205100
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.2477712880418075e-05,
+      "loss": 2.629,
+      "step": 205200
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.246917375414148e-05,
+      "loss": 2.8097,
+      "step": 205300
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.246063462786488e-05,
+      "loss": 2.8791,
+      "step": 205400
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.245209550158828e-05,
+      "loss": 2.8065,
+      "step": 205500
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.244355637531168e-05,
+      "loss": 2.5965,
+      "step": 205600
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.243501724903508e-05,
+      "loss": 2.8874,
+      "step": 205700
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.242647812275848e-05,
+      "loss": 2.5646,
+      "step": 205800
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.2417938996481886e-05,
+      "loss": 2.8783,
+      "step": 205900
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 3.240939987020528e-05,
+      "loss": 2.8532,
+      "step": 206000
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 3.240086074392868e-05,
+      "loss": 2.7758,
+      "step": 206100
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 3.2392321617652084e-05,
+      "loss": 2.9672,
+      "step": 206200
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 3.2383782491375485e-05,
+      "loss": 2.6679,
+      "step": 206300
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3.237524336509889e-05,
+      "loss": 2.68,
+      "step": 206400
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3.236670423882229e-05,
+      "loss": 2.9546,
+      "step": 206500
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3.235816511254568e-05,
+      "loss": 2.665,
+      "step": 206600
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.234962598626909e-05,
+      "loss": 2.6909,
+      "step": 206700
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.2341086859992486e-05,
+      "loss": 2.733,
+      "step": 206800
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.233254773371589e-05,
+      "loss": 2.7725,
+      "step": 206900
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.232400860743929e-05,
+      "loss": 2.975,
+      "step": 207000
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.231546948116269e-05,
+      "loss": 2.8213,
+      "step": 207100
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.2306930354886086e-05,
+      "loss": 2.6112,
+      "step": 207200
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.2298391228609494e-05,
+      "loss": 2.8263,
+      "step": 207300
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.228985210233289e-05,
+      "loss": 2.7241,
+      "step": 207400
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.228131297605629e-05,
+      "loss": 2.7141,
+      "step": 207500
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.227277384977969e-05,
+      "loss": 2.7914,
+      "step": 207600
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.2264234723503094e-05,
+      "loss": 2.7239,
+      "step": 207700
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.2255695597226495e-05,
+      "loss": 2.8226,
+      "step": 207800
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.22471564709499e-05,
+      "loss": 2.7109,
+      "step": 207900
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.223861734467329e-05,
+      "loss": 2.764,
+      "step": 208000
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.22300782183967e-05,
+      "loss": 2.7809,
+      "step": 208100
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.2221539092120095e-05,
+      "loss": 2.7088,
+      "step": 208200
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.2212999965843496e-05,
+      "loss": 2.6485,
+      "step": 208300
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.22044608395669e-05,
+      "loss": 2.7716,
+      "step": 208400
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.21959217132903e-05,
+      "loss": 2.7955,
+      "step": 208500
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.21873825870137e-05,
+      "loss": 2.7065,
+      "step": 208600
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.21788434607371e-05,
+      "loss": 2.7836,
+      "step": 208700
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.21703043344605e-05,
+      "loss": 2.7758,
+      "step": 208800
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.2161765208183905e-05,
+      "loss": 2.7706,
+      "step": 208900
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.21532260819073e-05,
+      "loss": 2.6964,
+      "step": 209000
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.21446869556307e-05,
+      "loss": 2.9809,
+      "step": 209100
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.21361478293541e-05,
+      "loss": 2.6947,
+      "step": 209200
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.2127608703077505e-05,
+      "loss": 2.7077,
+      "step": 209300
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.2119069576800906e-05,
+      "loss": 2.8423,
+      "step": 209400
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.211053045052431e-05,
+      "loss": 2.7861,
+      "step": 209500
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.21019913242477e-05,
+      "loss": 2.81,
+      "step": 209600
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.209345219797111e-05,
+      "loss": 2.8115,
+      "step": 209700
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.2084913071694506e-05,
+      "loss": 2.6927,
+      "step": 209800
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.207637394541791e-05,
+      "loss": 2.8274,
+      "step": 209900
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.206783481914131e-05,
+      "loss": 2.8401,
+      "step": 210000
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.205929569286471e-05,
+      "loss": 2.9757,
+      "step": 210100
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.2050756566588105e-05,
+      "loss": 2.5407,
+      "step": 210200
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3.2042217440311513e-05,
+      "loss": 2.6253,
+      "step": 210300
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3.203367831403491e-05,
+      "loss": 2.6943,
+      "step": 210400
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3.202513918775831e-05,
+      "loss": 2.8777,
+      "step": 210500
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 3.201660006148171e-05,
+      "loss": 2.6651,
+      "step": 210600
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 3.2008060935205106e-05,
+      "loss": 2.6112,
+      "step": 210700
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 3.1999521808928514e-05,
+      "loss": 2.7957,
+      "step": 210800
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 3.199098268265191e-05,
+      "loss": 2.8197,
+      "step": 210900
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 3.198244355637531e-05,
+      "loss": 2.8579,
+      "step": 211000
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 3.197390443009871e-05,
+      "loss": 2.8535,
+      "step": 211100
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 3.1965365303822114e-05,
+      "loss": 2.813,
+      "step": 211200
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 3.1956826177545515e-05,
+      "loss": 2.8407,
+      "step": 211300
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 3.194828705126892e-05,
+      "loss": 2.6361,
+      "step": 211400
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 3.193974792499231e-05,
+      "loss": 2.9215,
+      "step": 211500
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 3.193120879871572e-05,
+      "loss": 2.8109,
+      "step": 211600
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 3.1922669672439115e-05,
+      "loss": 2.7735,
+      "step": 211700
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 3.1914130546162516e-05,
+      "loss": 2.5616,
+      "step": 211800
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 3.190559141988592e-05,
+      "loss": 2.8323,
+      "step": 211900
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 3.189705229360932e-05,
+      "loss": 2.6473,
+      "step": 212000
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 3.1888513167332714e-05,
+      "loss": 2.6523,
+      "step": 212100
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 3.187997404105612e-05,
+      "loss": 2.6755,
+      "step": 212200
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 3.187143491477952e-05,
+      "loss": 2.6132,
+      "step": 212300
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 3.186289578850292e-05,
+      "loss": 2.573,
+      "step": 212400
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 3.185435666222632e-05,
+      "loss": 2.8654,
+      "step": 212500
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 3.184581753594972e-05,
+      "loss": 2.7064,
+      "step": 212600
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 3.1837278409673124e-05,
+      "loss": 2.9272,
+      "step": 212700
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 3.1828739283396525e-05,
+      "loss": 2.8052,
+      "step": 212800
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 3.182020015711992e-05,
+      "loss": 2.6925,
+      "step": 212900
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 3.181166103084333e-05,
+      "loss": 2.7437,
+      "step": 213000
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 3.180312190456672e-05,
+      "loss": 2.7572,
+      "step": 213100
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 3.1794582778290125e-05,
+      "loss": 2.7736,
+      "step": 213200
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 3.1786043652013526e-05,
+      "loss": 2.6393,
+      "step": 213300
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.177750452573693e-05,
+      "loss": 2.8257,
+      "step": 213400
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.176896539946033e-05,
+      "loss": 2.6962,
+      "step": 213500
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.176042627318373e-05,
+      "loss": 2.7065,
+      "step": 213600
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.1751887146907125e-05,
+      "loss": 2.7483,
+      "step": 213700
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.1743348020630534e-05,
+      "loss": 2.7294,
+      "step": 213800
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.173480889435393e-05,
+      "loss": 2.8052,
+      "step": 213900
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.172626976807733e-05,
+      "loss": 2.643,
+      "step": 214000
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.171773064180073e-05,
+      "loss": 2.8264,
+      "step": 214100
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.170919151552413e-05,
+      "loss": 2.8191,
+      "step": 214200
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.1700652389247535e-05,
+      "loss": 2.7863,
+      "step": 214300
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.1692113262970936e-05,
+      "loss": 2.6596,
+      "step": 214400
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.168357413669433e-05,
+      "loss": 2.6472,
+      "step": 214500
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.167503501041774e-05,
+      "loss": 2.6921,
+      "step": 214600
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.1666495884141134e-05,
+      "loss": 2.5574,
+      "step": 214700
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.1657956757864536e-05,
+      "loss": 2.664,
+      "step": 214800
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.164941763158794e-05,
+      "loss": 2.6997,
+      "step": 214900
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.164087850531134e-05,
+      "loss": 2.64,
+      "step": 215000
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.1632339379034734e-05,
+      "loss": 2.5975,
+      "step": 215100
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.162380025275814e-05,
+      "loss": 2.875,
+      "step": 215200
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.161526112648154e-05,
+      "loss": 2.7729,
+      "step": 215300
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.160672200020494e-05,
+      "loss": 2.5971,
+      "step": 215400
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.159818287392834e-05,
+      "loss": 2.9298,
+      "step": 215500
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.158964374765174e-05,
+      "loss": 2.5974,
+      "step": 215600
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.158110462137514e-05,
+      "loss": 2.7801,
+      "step": 215700
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.1572565495098544e-05,
+      "loss": 2.6669,
+      "step": 215800
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.156402636882194e-05,
+      "loss": 2.7291,
+      "step": 215900
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.155548724254535e-05,
+      "loss": 2.9889,
+      "step": 216000
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.154694811626874e-05,
+      "loss": 2.783,
+      "step": 216100
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.1538408989992144e-05,
+      "loss": 2.7246,
+      "step": 216200
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.1529869863715545e-05,
+      "loss": 2.6294,
+      "step": 216300
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.152133073743895e-05,
+      "loss": 2.8448,
+      "step": 216400
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 3.151279161116235e-05,
+      "loss": 2.4837,
+      "step": 216500
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 3.150425248488575e-05,
+      "loss": 2.8523,
+      "step": 216600
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 3.1495713358609145e-05,
+      "loss": 2.4857,
+      "step": 216700
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 3.148717423233255e-05,
+      "loss": 2.7957,
+      "step": 216800
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 3.147863510605595e-05,
+      "loss": 2.5642,
+      "step": 216900
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 3.147009597977935e-05,
+      "loss": 2.8443,
+      "step": 217000
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 3.146155685350275e-05,
+      "loss": 2.8137,
+      "step": 217100
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 3.145301772722615e-05,
+      "loss": 2.6857,
+      "step": 217200
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 3.144447860094955e-05,
+      "loss": 2.519,
+      "step": 217300
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 3.1435939474672956e-05,
+      "loss": 2.5561,
+      "step": 217400
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 3.142740034839635e-05,
+      "loss": 2.6158,
+      "step": 217500
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 3.141886122211975e-05,
+      "loss": 2.7535,
+      "step": 217600
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 3.1410322095843154e-05,
+      "loss": 2.5762,
+      "step": 217700
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 3.1401782969566555e-05,
+      "loss": 2.7406,
+      "step": 217800
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 3.139324384328996e-05,
+      "loss": 2.6391,
+      "step": 217900
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 3.138470471701336e-05,
+      "loss": 2.6371,
+      "step": 218000
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 3.137616559073675e-05,
+      "loss": 2.5723,
+      "step": 218100
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 3.136762646446016e-05,
+      "loss": 2.6901,
+      "step": 218200
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 3.1359087338183556e-05,
+      "loss": 2.4828,
+      "step": 218300
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 3.135054821190696e-05,
+      "loss": 2.7638,
+      "step": 218400
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 3.134200908563036e-05,
+      "loss": 2.7363,
+      "step": 218500
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 3.133346995935376e-05,
+      "loss": 2.6146,
+      "step": 218600
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 3.132493083307716e-05,
+      "loss": 2.7168,
+      "step": 218700
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 3.1316391706800564e-05,
+      "loss": 2.7791,
+      "step": 218800
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 3.130785258052396e-05,
+      "loss": 2.8559,
+      "step": 218900
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 3.129931345424737e-05,
+      "loss": 2.6361,
+      "step": 219000
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 3.129077432797076e-05,
+      "loss": 2.7913,
+      "step": 219100
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 3.128223520169416e-05,
+      "loss": 2.7234,
+      "step": 219200
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 3.1273696075417565e-05,
+      "loss": 2.9367,
+      "step": 219300
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 3.1265156949140966e-05,
+      "loss": 2.8618,
+      "step": 219400
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 3.125661782286437e-05,
+      "loss": 2.7544,
+      "step": 219500
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 3.124807869658777e-05,
+      "loss": 2.6748,
+      "step": 219600
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 3.1239539570311164e-05,
+      "loss": 2.7111,
+      "step": 219700
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 3.123100044403457e-05,
+      "loss": 2.7402,
+      "step": 219800
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 3.122246131775797e-05,
+      "loss": 2.6942,
+      "step": 219900
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 3.121392219148137e-05,
+      "loss": 2.7211,
+      "step": 220000
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 3.120538306520477e-05,
+      "loss": 2.649,
+      "step": 220100
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 3.119684393892817e-05,
+      "loss": 2.8113,
+      "step": 220200
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 3.118830481265157e-05,
+      "loss": 2.6578,
+      "step": 220300
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 3.1179765686374975e-05,
+      "loss": 2.8545,
+      "step": 220400
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 3.117122656009837e-05,
+      "loss": 2.9633,
+      "step": 220500
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 3.116268743382177e-05,
+      "loss": 2.8168,
+      "step": 220600
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 3.115414830754517e-05,
+      "loss": 2.6067,
+      "step": 220700
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 3.1145609181268574e-05,
+      "loss": 2.6563,
+      "step": 220800
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 3.1137070054991976e-05,
+      "loss": 2.6782,
+      "step": 220900
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 3.112853092871538e-05,
+      "loss": 2.5729,
+      "step": 221000
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 3.111999180243877e-05,
+      "loss": 2.6424,
+      "step": 221100
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 3.111145267616218e-05,
+      "loss": 2.6948,
+      "step": 221200
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 3.1102913549885575e-05,
+      "loss": 2.6781,
+      "step": 221300
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 3.109437442360898e-05,
+      "loss": 2.6164,
+      "step": 221400
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 3.108583529733238e-05,
+      "loss": 2.7599,
+      "step": 221500
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 3.107729617105578e-05,
+      "loss": 2.6885,
+      "step": 221600
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 3.106875704477918e-05,
+      "loss": 2.7714,
+      "step": 221700
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 3.106021791850258e-05,
+      "loss": 2.8554,
+      "step": 221800
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 3.105167879222598e-05,
+      "loss": 2.8403,
+      "step": 221900
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 3.1043139665949386e-05,
+      "loss": 2.7891,
+      "step": 222000
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 3.103460053967278e-05,
+      "loss": 2.7963,
+      "step": 222100
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 3.102606141339618e-05,
+      "loss": 2.8312,
+      "step": 222200
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 3.1017522287119584e-05,
+      "loss": 2.5519,
+      "step": 222300
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 3.1008983160842986e-05,
+      "loss": 2.7113,
+      "step": 222400
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 3.100044403456639e-05,
+      "loss": 2.6592,
+      "step": 222500
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 3.099190490828979e-05,
+      "loss": 2.7181,
+      "step": 222600
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 3.0983365782013184e-05,
+      "loss": 2.6204,
+      "step": 222700
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 3.097482665573659e-05,
+      "loss": 2.6717,
+      "step": 222800
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 3.0966287529459987e-05,
+      "loss": 2.8013,
+      "step": 222900
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 3.095774840318339e-05,
+      "loss": 2.7697,
+      "step": 223000
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 3.094920927690679e-05,
+      "loss": 2.6964,
+      "step": 223100
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 3.094067015063019e-05,
+      "loss": 2.7323,
+      "step": 223200
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 3.0932131024353586e-05,
+      "loss": 2.6465,
+      "step": 223300
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 3.0923591898076994e-05,
+      "loss": 2.6533,
+      "step": 223400
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 3.091505277180039e-05,
+      "loss": 2.5259,
+      "step": 223500
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 3.090651364552379e-05,
+      "loss": 2.943,
+      "step": 223600
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 3.089797451924719e-05,
+      "loss": 2.6762,
+      "step": 223700
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 3.0889435392970594e-05,
+      "loss": 2.8889,
+      "step": 223800
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 3.0880896266693995e-05,
+      "loss": 2.907,
+      "step": 223900
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 3.08723571404174e-05,
+      "loss": 2.769,
+      "step": 224000
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 3.086381801414079e-05,
+      "loss": 2.8173,
+      "step": 224100
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 3.08552788878642e-05,
+      "loss": 2.6643,
+      "step": 224200
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 3.0846739761587595e-05,
+      "loss": 2.855,
+      "step": 224300
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 3.0838200635310996e-05,
+      "loss": 2.836,
+      "step": 224400
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 3.08296615090344e-05,
+      "loss": 2.703,
+      "step": 224500
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 3.08211223827578e-05,
+      "loss": 2.8927,
+      "step": 224600
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 3.08125832564812e-05,
+      "loss": 2.6234,
+      "step": 224700
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 3.08040441302046e-05,
+      "loss": 2.715,
+      "step": 224800
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 3.0795505003928e-05,
+      "loss": 2.6856,
+      "step": 224900
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 3.0786965877651406e-05,
+      "loss": 2.6405,
+      "step": 225000
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 3.07784267513748e-05,
+      "loss": 2.6382,
+      "step": 225100
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 3.07698876250982e-05,
+      "loss": 2.6424,
+      "step": 225200
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 3.0761348498821603e-05,
+      "loss": 2.6423,
+      "step": 225300
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 3.0752809372545005e-05,
+      "loss": 2.7959,
+      "step": 225400
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 3.07442702462684e-05,
+      "loss": 2.9217,
+      "step": 225500
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 3.073573111999181e-05,
+      "loss": 2.8175,
+      "step": 225600
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 3.07271919937152e-05,
+      "loss": 2.6331,
+      "step": 225700
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 3.0718652867438604e-05,
+      "loss": 2.7062,
+      "step": 225800
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 3.0710113741162006e-05,
+      "loss": 2.6827,
+      "step": 225900
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 3.070157461488541e-05,
+      "loss": 2.6571,
+      "step": 226000
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 3.069303548860881e-05,
+      "loss": 2.5765,
+      "step": 226100
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 3.068449636233221e-05,
+      "loss": 2.7972,
+      "step": 226200
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 3.0675957236055605e-05,
+      "loss": 2.7939,
+      "step": 226300
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 3.0667418109779014e-05,
+      "loss": 2.6674,
+      "step": 226400
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 3.065887898350241e-05,
+      "loss": 2.6047,
+      "step": 226500
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 3.065033985722581e-05,
+      "loss": 2.5639,
+      "step": 226600
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 3.064180073094921e-05,
+      "loss": 2.7367,
+      "step": 226700
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 3.063326160467261e-05,
+      "loss": 2.879,
+      "step": 226800
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 3.0624722478396015e-05,
+      "loss": 2.8679,
+      "step": 226900
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 3.0616183352119416e-05,
+      "loss": 2.6516,
+      "step": 227000
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 3.060764422584281e-05,
+      "loss": 2.727,
+      "step": 227100
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 3.059910509956622e-05,
+      "loss": 2.7692,
+      "step": 227200
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 3.0590565973289614e-05,
+      "loss": 2.6338,
+      "step": 227300
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 3.0582026847013016e-05,
+      "loss": 2.6727,
+      "step": 227400
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 3.057348772073642e-05,
+      "loss": 2.5125,
+      "step": 227500
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 3.056494859445982e-05,
+      "loss": 2.8578,
+      "step": 227600
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 3.055640946818322e-05,
+      "loss": 2.792,
+      "step": 227700
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.054787034190662e-05,
+      "loss": 2.6018,
+      "step": 227800
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.0539331215630017e-05,
+      "loss": 2.7506,
+      "step": 227900
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.053079208935342e-05,
+      "loss": 2.937,
+      "step": 228000
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.052225296307682e-05,
+      "loss": 2.4773,
+      "step": 228100
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.0513713836800218e-05,
+      "loss": 2.745,
+      "step": 228200
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.050517471052362e-05,
+      "loss": 2.6591,
+      "step": 228300
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.0496635584247018e-05,
+      "loss": 2.8307,
+      "step": 228400
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.0488096457970422e-05,
+      "loss": 2.732,
+      "step": 228500
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 3.047955733169382e-05,
+      "loss": 2.703,
+      "step": 228600
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 3.0471018205417222e-05,
+      "loss": 2.7368,
+      "step": 228700
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 3.046247907914062e-05,
+      "loss": 2.7134,
+      "step": 228800
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 3.0453939952864025e-05,
+      "loss": 2.717,
+      "step": 228900
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 3.0445400826587423e-05,
+      "loss": 2.6813,
+      "step": 229000
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 3.0436861700310825e-05,
+      "loss": 2.7051,
+      "step": 229100
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 3.0428322574034223e-05,
+      "loss": 2.4863,
+      "step": 229200
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 3.0419783447757628e-05,
+      "loss": 2.8049,
+      "step": 229300
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 3.0411244321481026e-05,
+      "loss": 2.7755,
+      "step": 229400
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 3.0402705195204428e-05,
+      "loss": 2.9066,
+      "step": 229500
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 3.0394166068927826e-05,
+      "loss": 3.0177,
+      "step": 229600
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 3.038562694265123e-05,
+      "loss": 2.6426,
+      "step": 229700
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 3.037708781637463e-05,
+      "loss": 2.7463,
+      "step": 229800
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 3.036854869009803e-05,
+      "loss": 2.8526,
+      "step": 229900
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 3.036000956382143e-05,
+      "loss": 2.9398,
+      "step": 230000
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 3.0351470437544834e-05,
+      "loss": 2.5734,
+      "step": 230100
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 3.034293131126823e-05,
+      "loss": 2.7124,
+      "step": 230200
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 3.0334392184991633e-05,
+      "loss": 2.8195,
+      "step": 230300
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 3.032585305871503e-05,
+      "loss": 2.7899,
+      "step": 230400
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 3.0317313932438433e-05,
+      "loss": 2.7026,
+      "step": 230500
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 3.030877480616183e-05,
+      "loss": 2.6566,
+      "step": 230600
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 3.0300235679885236e-05,
+      "loss": 2.6287,
+      "step": 230700
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 3.0291696553608634e-05,
+      "loss": 2.8339,
+      "step": 230800
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 3.0283157427332036e-05,
+      "loss": 2.8427,
+      "step": 230900
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 3.0274618301055434e-05,
+      "loss": 2.664,
+      "step": 231000
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 3.026607917477884e-05,
+      "loss": 2.6706,
+      "step": 231100
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 3.0257540048502237e-05,
+      "loss": 2.7868,
+      "step": 231200
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 3.024900092222564e-05,
+      "loss": 2.9513,
+      "step": 231300
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 3.0240461795949037e-05,
+      "loss": 2.7365,
+      "step": 231400
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 3.0231922669672442e-05,
+      "loss": 2.5798,
+      "step": 231500
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 3.022338354339584e-05,
+      "loss": 2.4588,
+      "step": 231600
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 3.021484441711924e-05,
+      "loss": 2.84,
+      "step": 231700
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 3.020630529084264e-05,
+      "loss": 2.7311,
+      "step": 231800
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 3.0197766164566045e-05,
+      "loss": 2.6963,
+      "step": 231900
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 3.0189227038289443e-05,
+      "loss": 2.7015,
+      "step": 232000
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 3.0180687912012844e-05,
+      "loss": 2.8284,
+      "step": 232100
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 3.0172148785736242e-05,
+      "loss": 2.616,
+      "step": 232200
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 3.0163609659459647e-05,
+      "loss": 2.6611,
+      "step": 232300
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 3.0155070533183046e-05,
+      "loss": 2.8142,
+      "step": 232400
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 3.0146531406906447e-05,
+      "loss": 2.6198,
+      "step": 232500
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 3.0137992280629845e-05,
+      "loss": 2.7892,
+      "step": 232600
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 3.012945315435325e-05,
+      "loss": 2.7212,
+      "step": 232700
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 3.0120914028076645e-05,
+      "loss": 2.6871,
+      "step": 232800
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 3.011237490180005e-05,
+      "loss": 2.6104,
+      "step": 232900
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 3.0103835775523448e-05,
+      "loss": 2.5973,
+      "step": 233000
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 3.009529664924685e-05,
+      "loss": 2.7518,
+      "step": 233100
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 3.0086757522970248e-05,
+      "loss": 2.6271,
+      "step": 233200
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 3.0078218396693653e-05,
+      "loss": 2.7272,
+      "step": 233300
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 3.006967927041705e-05,
+      "loss": 2.5538,
+      "step": 233400
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 3.0061140144140452e-05,
+      "loss": 2.7496,
+      "step": 233500
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 3.005260101786385e-05,
+      "loss": 2.7422,
+      "step": 233600
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 3.0044061891587256e-05,
+      "loss": 2.6432,
+      "step": 233700
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 3.0035522765310654e-05,
+      "loss": 2.6889,
+      "step": 233800
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 3.0026983639034055e-05,
+      "loss": 2.7468,
+      "step": 233900
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 3.0018444512757453e-05,
+      "loss": 2.834,
+      "step": 234000
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.000990538648086e-05,
+      "loss": 2.7115,
+      "step": 234100
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.0001366260204257e-05,
+      "loss": 2.5906,
+      "step": 234200
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 2.9992827133927658e-05,
+      "loss": 2.6469,
+      "step": 234300
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 2.9984288007651056e-05,
+      "loss": 2.7775,
+      "step": 234400
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 2.997574888137446e-05,
+      "loss": 2.7041,
+      "step": 234500
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 2.996720975509786e-05,
+      "loss": 2.7634,
+      "step": 234600
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 2.995867062882126e-05,
+      "loss": 2.4696,
+      "step": 234700
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 2.995013150254466e-05,
+      "loss": 2.7226,
+      "step": 234800
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 2.9941592376268064e-05,
+      "loss": 2.67,
+      "step": 234900
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 2.9933053249991462e-05,
+      "loss": 2.7104,
+      "step": 235000
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 2.9924514123714864e-05,
+      "loss": 2.8479,
+      "step": 235100
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 2.9915974997438262e-05,
+      "loss": 2.7838,
+      "step": 235200
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 2.9907435871161667e-05,
+      "loss": 2.6987,
+      "step": 235300
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 2.989889674488506e-05,
+      "loss": 2.6483,
+      "step": 235400
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 2.9890357618608466e-05,
+      "loss": 2.5661,
+      "step": 235500
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 2.9881818492331865e-05,
+      "loss": 2.6686,
+      "step": 235600
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 2.987327936605527e-05,
+      "loss": 2.688,
+      "step": 235700
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 2.9864740239778664e-05,
+      "loss": 2.76,
+      "step": 235800
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 2.985620111350207e-05,
+      "loss": 2.6832,
+      "step": 235900
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 2.9847661987225467e-05,
+      "loss": 2.6808,
+      "step": 236000
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 2.983912286094887e-05,
+      "loss": 2.7677,
+      "step": 236100
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 2.9830583734672267e-05,
+      "loss": 2.6441,
+      "step": 236200
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 2.9822044608395672e-05,
+      "loss": 2.7636,
+      "step": 236300
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.981350548211907e-05,
+      "loss": 2.7828,
+      "step": 236400
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.9804966355842472e-05,
+      "loss": 2.5797,
+      "step": 236500
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.979642722956587e-05,
+      "loss": 2.7867,
+      "step": 236600
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.9787888103289275e-05,
+      "loss": 2.6842,
+      "step": 236700
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 2.9779348977012673e-05,
+      "loss": 2.7861,
+      "step": 236800
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 2.9770809850736075e-05,
+      "loss": 2.6882,
+      "step": 236900
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 2.9762270724459473e-05,
+      "loss": 2.8817,
+      "step": 237000
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 2.9753731598182878e-05,
+      "loss": 2.9218,
+      "step": 237100
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 2.9745192471906276e-05,
+      "loss": 2.7078,
+      "step": 237200
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 2.9736653345629677e-05,
+      "loss": 2.6177,
+      "step": 237300
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 2.9728114219353076e-05,
+      "loss": 2.667,
+      "step": 237400
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 2.971957509307648e-05,
+      "loss": 2.8412,
+      "step": 237500
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 2.971103596679988e-05,
+      "loss": 2.7651,
+      "step": 237600
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 2.970249684052328e-05,
+      "loss": 2.5657,
+      "step": 237700
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 2.969395771424668e-05,
+      "loss": 2.4972,
+      "step": 237800
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 2.9685418587970083e-05,
+      "loss": 2.6563,
+      "step": 237900
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 2.9676879461693478e-05,
+      "loss": 2.7086,
+      "step": 238000
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 2.9668340335416883e-05,
+      "loss": 2.4449,
+      "step": 238100
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 2.965980120914028e-05,
+      "loss": 2.5057,
+      "step": 238200
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 2.9651262082863686e-05,
+      "loss": 2.7743,
+      "step": 238300
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 2.964272295658708e-05,
+      "loss": 2.6181,
+      "step": 238400
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 2.9634183830310486e-05,
+      "loss": 2.6679,
+      "step": 238500
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 2.9625644704033884e-05,
+      "loss": 2.908,
+      "step": 238600
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 2.9617105577757286e-05,
+      "loss": 2.4055,
+      "step": 238700
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 2.9608566451480684e-05,
+      "loss": 2.7305,
+      "step": 238800
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 2.960002732520409e-05,
+      "loss": 2.7786,
+      "step": 238900
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 2.9591488198927487e-05,
+      "loss": 2.7557,
+      "step": 239000
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 2.958294907265089e-05,
+      "loss": 2.6417,
+      "step": 239100
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 2.9574409946374286e-05,
+      "loss": 2.7198,
+      "step": 239200
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 2.956587082009769e-05,
+      "loss": 2.7358,
+      "step": 239300
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 2.955733169382109e-05,
+      "loss": 2.6815,
+      "step": 239400
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 2.954879256754449e-05,
+      "loss": 2.6953,
+      "step": 239500
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 2.954025344126789e-05,
+      "loss": 2.8371,
+      "step": 239600
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 2.9531714314991294e-05,
+      "loss": 2.8867,
+      "step": 239700
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 2.9523175188714692e-05,
+      "loss": 2.7678,
+      "step": 239800
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 2.9514636062438094e-05,
+      "loss": 2.4464,
+      "step": 239900
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 2.9506096936161492e-05,
+      "loss": 2.6759,
+      "step": 240000
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 2.9497557809884897e-05,
+      "loss": 2.9032,
+      "step": 240100
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 2.9489018683608295e-05,
+      "loss": 2.7916,
+      "step": 240200
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 2.9480479557331697e-05,
+      "loss": 2.7683,
+      "step": 240300
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 2.9471940431055095e-05,
+      "loss": 2.8258,
+      "step": 240400
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 2.94634013047785e-05,
+      "loss": 2.5918,
+      "step": 240500
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 2.9454862178501898e-05,
+      "loss": 2.7946,
+      "step": 240600
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 2.94463230522253e-05,
+      "loss": 2.818,
+      "step": 240700
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 2.9437783925948698e-05,
+      "loss": 2.6471,
+      "step": 240800
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 2.9429244799672103e-05,
+      "loss": 2.6668,
+      "step": 240900
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 2.9420705673395497e-05,
+      "loss": 2.7814,
+      "step": 241000
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 2.9412166547118902e-05,
+      "loss": 2.7499,
+      "step": 241100
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 2.94036274208423e-05,
+      "loss": 2.7778,
+      "step": 241200
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 2.9395088294565702e-05,
+      "loss": 2.6787,
+      "step": 241300
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 2.93865491682891e-05,
+      "loss": 2.7438,
+      "step": 241400
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 2.9378010042012505e-05,
+      "loss": 2.7875,
+      "step": 241500
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 2.9369470915735903e-05,
+      "loss": 2.6229,
+      "step": 241600
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 2.9360931789459305e-05,
+      "loss": 2.736,
+      "step": 241700
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 2.9352392663182703e-05,
+      "loss": 2.6908,
+      "step": 241800
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 2.9343853536906108e-05,
+      "loss": 2.6863,
+      "step": 241900
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 2.9335314410629506e-05,
+      "loss": 2.6862,
+      "step": 242000
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 2.9326775284352908e-05,
+      "loss": 2.7347,
+      "step": 242100
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 2.9318236158076306e-05,
+      "loss": 2.615,
+      "step": 242200
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 2.930969703179971e-05,
+      "loss": 2.7367,
+      "step": 242300
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 2.930115790552311e-05,
+      "loss": 2.7765,
+      "step": 242400
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 2.929261877924651e-05,
+      "loss": 2.806,
+      "step": 242500
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 2.928407965296991e-05,
+      "loss": 2.7895,
+      "step": 242600
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 2.9275540526693314e-05,
+      "loss": 2.7328,
+      "step": 242700
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 2.9267001400416712e-05,
+      "loss": 2.5228,
+      "step": 242800
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 2.9258462274140113e-05,
+      "loss": 2.6651,
+      "step": 242900
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 2.924992314786351e-05,
+      "loss": 2.7362,
+      "step": 243000
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 2.9241384021586916e-05,
+      "loss": 2.8293,
+      "step": 243100
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 2.9232844895310315e-05,
+      "loss": 2.7309,
+      "step": 243200
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 2.9224305769033716e-05,
+      "loss": 2.5739,
+      "step": 243300
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 2.9215766642757114e-05,
+      "loss": 2.6963,
+      "step": 243400
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 2.920722751648052e-05,
+      "loss": 2.8438,
+      "step": 243500
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 2.9198688390203914e-05,
+      "loss": 2.5773,
+      "step": 243600
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 2.919014926392732e-05,
+      "loss": 2.6294,
+      "step": 243700
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 2.9181610137650717e-05,
+      "loss": 2.5419,
+      "step": 243800
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 2.917307101137412e-05,
+      "loss": 2.6182,
+      "step": 243900
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 2.9164531885097517e-05,
+      "loss": 2.6046,
+      "step": 244000
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 2.915599275882092e-05,
+      "loss": 2.6727,
+      "step": 244100
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 2.914745363254432e-05,
+      "loss": 2.4974,
+      "step": 244200
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 2.913891450626772e-05,
+      "loss": 2.6973,
+      "step": 244300
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 2.913037537999112e-05,
+      "loss": 2.7346,
+      "step": 244400
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 2.9121836253714524e-05,
+      "loss": 2.6768,
+      "step": 244500
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 2.9113297127437923e-05,
+      "loss": 2.6179,
+      "step": 244600
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 2.9104758001161324e-05,
+      "loss": 2.651,
+      "step": 244700
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 2.9096218874884722e-05,
+      "loss": 2.6308,
+      "step": 244800
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 2.9087679748608127e-05,
+      "loss": 2.9426,
+      "step": 244900
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 2.9079140622331525e-05,
+      "loss": 2.8143,
+      "step": 245000
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 2.9070601496054927e-05,
+      "loss": 2.5903,
+      "step": 245100
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 2.9062062369778325e-05,
+      "loss": 2.7973,
+      "step": 245200
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 2.905352324350173e-05,
+      "loss": 2.486,
+      "step": 245300
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 2.9044984117225128e-05,
+      "loss": 2.7287,
+      "step": 245400
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 2.9036444990948526e-05,
+      "loss": 2.7824,
+      "step": 245500
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 2.9027905864671928e-05,
+      "loss": 2.6783,
+      "step": 245600
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 2.9019366738395326e-05,
+      "loss": 2.7329,
+      "step": 245700
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 2.901082761211873e-05,
+      "loss": 2.5198,
+      "step": 245800
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 2.9002288485842126e-05,
+      "loss": 2.7395,
+      "step": 245900
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 2.899374935956553e-05,
+      "loss": 2.7563,
+      "step": 246000
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 2.898521023328893e-05,
+      "loss": 2.6676,
+      "step": 246100
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 2.897667110701233e-05,
+      "loss": 2.6968,
+      "step": 246200
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 2.896813198073573e-05,
+      "loss": 2.7464,
+      "step": 246300
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 2.8959592854459134e-05,
+      "loss": 2.6647,
+      "step": 246400
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 2.8951053728182532e-05,
+      "loss": 2.8056,
+      "step": 246500
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 2.8942514601905933e-05,
+      "loss": 2.7379,
+      "step": 246600
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 2.893397547562933e-05,
+      "loss": 2.6462,
+      "step": 246700
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 2.8925436349352736e-05,
+      "loss": 2.7145,
+      "step": 246800
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 2.8916897223076135e-05,
+      "loss": 2.7956,
+      "step": 246900
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 2.8908358096799536e-05,
+      "loss": 2.8627,
+      "step": 247000
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 2.8899818970522934e-05,
+      "loss": 2.736,
+      "step": 247100
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 2.889127984424634e-05,
+      "loss": 2.6867,
+      "step": 247200
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 2.8882740717969737e-05,
+      "loss": 2.7596,
+      "step": 247300
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 2.887420159169314e-05,
+      "loss": 2.6373,
+      "step": 247400
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 2.8865662465416537e-05,
+      "loss": 2.5411,
+      "step": 247500
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 2.8857123339139942e-05,
+      "loss": 2.7351,
+      "step": 247600
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 2.884858421286334e-05,
+      "loss": 2.6872,
+      "step": 247700
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 2.8840045086586742e-05,
+      "loss": 2.6734,
+      "step": 247800
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 2.883150596031014e-05,
+      "loss": 2.681,
+      "step": 247900
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 2.8822966834033545e-05,
+      "loss": 2.5425,
+      "step": 248000
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 2.8814427707756943e-05,
+      "loss": 2.7357,
+      "step": 248100
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 2.8805888581480345e-05,
+      "loss": 2.6655,
+      "step": 248200
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 2.8797349455203743e-05,
+      "loss": 2.7599,
+      "step": 248300
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 2.8788810328927148e-05,
+      "loss": 2.4836,
+      "step": 248400
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 2.8780271202650542e-05,
+      "loss": 2.4861,
+      "step": 248500
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 2.8771732076373947e-05,
+      "loss": 2.5584,
+      "step": 248600
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 2.8763192950097345e-05,
+      "loss": 2.6304,
+      "step": 248700
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 2.8754653823820747e-05,
+      "loss": 2.534,
+      "step": 248800
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 2.8746114697544145e-05,
+      "loss": 2.6754,
+      "step": 248900
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 2.873757557126755e-05,
+      "loss": 2.763,
+      "step": 249000
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 2.8729036444990948e-05,
+      "loss": 2.7081,
+      "step": 249100
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 2.872049731871435e-05,
+      "loss": 2.7015,
+      "step": 249200
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 2.8711958192437748e-05,
+      "loss": 2.5221,
+      "step": 249300
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 2.8703419066161153e-05,
+      "loss": 2.7446,
+      "step": 249400
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 2.869487993988455e-05,
+      "loss": 2.7252,
+      "step": 249500
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 2.8686340813607953e-05,
+      "loss": 2.7931,
+      "step": 249600
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 2.867780168733135e-05,
+      "loss": 2.5381,
+      "step": 249700
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 2.8669262561054756e-05,
+      "loss": 2.6714,
+      "step": 249800
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 2.8660723434778154e-05,
+      "loss": 2.5294,
+      "step": 249900
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 2.8652184308501555e-05,
+      "loss": 2.6885,
+      "step": 250000
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 2.8643645182224954e-05,
+      "loss": 2.7648,
+      "step": 250100
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 2.863510605594836e-05,
+      "loss": 2.6819,
+      "step": 250200
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 2.8626566929671757e-05,
+      "loss": 2.5423,
+      "step": 250300
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 2.8618027803395158e-05,
+      "loss": 2.5962,
+      "step": 250400
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 2.8609488677118556e-05,
+      "loss": 2.5923,
+      "step": 250500
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 2.860094955084196e-05,
+      "loss": 2.6878,
+      "step": 250600
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 2.859241042456536e-05,
+      "loss": 2.7258,
+      "step": 250700
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 2.858387129828876e-05,
+      "loss": 2.7176,
+      "step": 250800
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 2.857533217201216e-05,
+      "loss": 2.7349,
+      "step": 250900
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 2.8566793045735564e-05,
+      "loss": 2.5558,
+      "step": 251000
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 2.855825391945896e-05,
+      "loss": 2.6889,
+      "step": 251100
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 2.8549714793182364e-05,
+      "loss": 2.7017,
+      "step": 251200
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 2.8541175666905762e-05,
+      "loss": 2.7386,
+      "step": 251300
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 2.8532636540629164e-05,
+      "loss": 2.686,
+      "step": 251400
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 2.8524097414352562e-05,
+      "loss": 2.8088,
+      "step": 251500
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 2.8515558288075967e-05,
+      "loss": 2.6469,
+      "step": 251600
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 2.8507019161799365e-05,
+      "loss": 2.7537,
+      "step": 251700
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 2.8498480035522766e-05,
+      "loss": 2.7216,
+      "step": 251800
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 2.8489940909246165e-05,
+      "loss": 2.6939,
+      "step": 251900
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 2.848140178296957e-05,
+      "loss": 2.5937,
+      "step": 252000
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 2.8472862656692968e-05,
+      "loss": 2.7772,
+      "step": 252100
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 2.846432353041637e-05,
+      "loss": 2.5186,
+      "step": 252200
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 2.8455784404139767e-05,
+      "loss": 2.5647,
+      "step": 252300
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 2.8447245277863172e-05,
+      "loss": 2.7313,
+      "step": 252400
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 2.843870615158657e-05,
+      "loss": 2.8239,
+      "step": 252500
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 2.8430167025309972e-05,
+      "loss": 2.6443,
+      "step": 252600
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 2.842162789903337e-05,
+      "loss": 2.6207,
+      "step": 252700
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 2.8413088772756775e-05,
+      "loss": 2.5298,
+      "step": 252800
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 2.8404549646480173e-05,
+      "loss": 2.7086,
+      "step": 252900
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 2.8396010520203575e-05,
+      "loss": 2.7574,
+      "step": 253000
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 2.8387471393926973e-05,
+      "loss": 2.7623,
+      "step": 253100
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 2.8378932267650378e-05,
+      "loss": 2.5917,
+      "step": 253200
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 2.8370393141373776e-05,
+      "loss": 2.3612,
+      "step": 253300
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 2.8361854015097178e-05,
+      "loss": 2.4796,
+      "step": 253400
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 2.8353314888820576e-05,
+      "loss": 2.4856,
+      "step": 253500
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 2.834477576254398e-05,
+      "loss": 2.5118,
+      "step": 253600
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 2.8336236636267375e-05,
+      "loss": 2.663,
+      "step": 253700
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 2.832769750999078e-05,
+      "loss": 2.6695,
+      "step": 253800
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 2.831915838371418e-05,
+      "loss": 2.4447,
+      "step": 253900
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 2.831061925743758e-05,
+      "loss": 2.6524,
+      "step": 254000
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 2.8302080131160978e-05,
+      "loss": 2.6497,
+      "step": 254100
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 2.8293541004884383e-05,
+      "loss": 2.6326,
+      "step": 254200
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 2.828500187860778e-05,
+      "loss": 2.6074,
+      "step": 254300
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 2.8276462752331183e-05,
+      "loss": 2.4123,
+      "step": 254400
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 2.826792362605458e-05,
+      "loss": 2.6305,
+      "step": 254500
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 2.8259384499777986e-05,
+      "loss": 2.5838,
+      "step": 254600
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 2.8250845373501384e-05,
+      "loss": 2.6634,
+      "step": 254700
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 2.8242306247224786e-05,
+      "loss": 2.66,
+      "step": 254800
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 2.8233767120948184e-05,
+      "loss": 2.7131,
+      "step": 254900
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 2.822522799467159e-05,
+      "loss": 2.5279,
+      "step": 255000
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 2.8216688868394987e-05,
+      "loss": 2.6542,
+      "step": 255100
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 2.820814974211839e-05,
+      "loss": 2.4608,
+      "step": 255200
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 2.8199610615841787e-05,
+      "loss": 2.6652,
+      "step": 255300
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 2.819107148956519e-05,
+      "loss": 2.5986,
+      "step": 255400
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 2.818253236328859e-05,
+      "loss": 2.6741,
+      "step": 255500
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 2.817399323701199e-05,
+      "loss": 2.7155,
+      "step": 255600
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 2.816545411073539e-05,
+      "loss": 2.577,
+      "step": 255700
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 2.8156914984458794e-05,
+      "loss": 2.4684,
+      "step": 255800
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 2.8148375858182193e-05,
+      "loss": 2.631,
+      "step": 255900
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 2.8139836731905594e-05,
+      "loss": 2.8226,
+      "step": 256000
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 2.8131297605628992e-05,
+      "loss": 2.7944,
+      "step": 256100
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 2.8122758479352397e-05,
+      "loss": 2.5859,
+      "step": 256200
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 2.8114219353075792e-05,
+      "loss": 2.6318,
+      "step": 256300
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 2.8105680226799197e-05,
+      "loss": 2.6843,
+      "step": 256400
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 2.8097141100522595e-05,
+      "loss": 2.6963,
+      "step": 256500
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 2.8088601974245997e-05,
+      "loss": 2.6305,
+      "step": 256600
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 2.8080062847969395e-05,
+      "loss": 2.6668,
+      "step": 256700
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 2.80715237216928e-05,
+      "loss": 2.537,
+      "step": 256800
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 2.8062984595416198e-05,
+      "loss": 2.7308,
+      "step": 256900
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 2.80544454691396e-05,
+      "loss": 2.7153,
+      "step": 257000
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 2.8045906342862998e-05,
+      "loss": 2.6395,
+      "step": 257100
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 2.8037367216586403e-05,
+      "loss": 2.4912,
+      "step": 257200
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 2.80288280903098e-05,
+      "loss": 2.6566,
+      "step": 257300
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 2.8020288964033202e-05,
+      "loss": 2.5217,
+      "step": 257400
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 2.80117498377566e-05,
+      "loss": 2.7992,
+      "step": 257500
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 2.8003210711480005e-05,
+      "loss": 2.7055,
+      "step": 257600
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 2.7994671585203404e-05,
+      "loss": 2.5757,
+      "step": 257700
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 2.7986132458926805e-05,
+      "loss": 2.8099,
+      "step": 257800
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 2.7977593332650203e-05,
+      "loss": 2.5271,
+      "step": 257900
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 2.7969054206373608e-05,
+      "loss": 2.661,
+      "step": 258000
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 2.7960515080097006e-05,
+      "loss": 2.6143,
+      "step": 258100
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 2.7951975953820408e-05,
+      "loss": 2.5266,
+      "step": 258200
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 2.7943436827543806e-05,
+      "loss": 2.6032,
+      "step": 258300
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 2.793489770126721e-05,
+      "loss": 2.5501,
+      "step": 258400
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 2.792635857499061e-05,
+      "loss": 2.6541,
+      "step": 258500
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 2.791781944871401e-05,
+      "loss": 2.6406,
+      "step": 258600
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 2.790928032243741e-05,
+      "loss": 2.5852,
+      "step": 258700
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 2.7900741196160814e-05,
+      "loss": 2.6384,
+      "step": 258800
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 2.789220206988421e-05,
+      "loss": 2.5915,
+      "step": 258900
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 2.7883662943607613e-05,
+      "loss": 2.6916,
+      "step": 259000
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 2.787512381733101e-05,
+      "loss": 2.6679,
+      "step": 259100
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 2.7866584691054413e-05,
+      "loss": 2.732,
+      "step": 259200
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 2.785804556477781e-05,
+      "loss": 2.7541,
+      "step": 259300
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 2.7849506438501216e-05,
+      "loss": 2.4927,
+      "step": 259400
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 2.7840967312224614e-05,
+      "loss": 2.5912,
+      "step": 259500
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 2.7832428185948016e-05,
+      "loss": 2.5699,
+      "step": 259600
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 2.7823889059671414e-05,
+      "loss": 2.6161,
+      "step": 259700
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 2.781534993339482e-05,
+      "loss": 2.6555,
+      "step": 259800
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 2.7806810807118217e-05,
+      "loss": 2.586,
+      "step": 259900
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 2.779827168084162e-05,
+      "loss": 2.817,
+      "step": 260000
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 2.7789732554565017e-05,
+      "loss": 2.7655,
+      "step": 260100
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 2.7781193428288422e-05,
+      "loss": 2.6764,
+      "step": 260200
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 2.777265430201182e-05,
+      "loss": 2.7336,
+      "step": 260300
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 2.776411517573522e-05,
+      "loss": 2.7111,
+      "step": 260400
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 2.775557604945862e-05,
+      "loss": 2.6737,
+      "step": 260500
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 2.7747036923182025e-05,
+      "loss": 2.6634,
+      "step": 260600
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 2.7738497796905423e-05,
+      "loss": 2.6087,
+      "step": 260700
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 2.7729958670628824e-05,
+      "loss": 2.5933,
+      "step": 260800
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 2.7721419544352223e-05,
+      "loss": 2.5006,
+      "step": 260900
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 2.7712880418075627e-05,
+      "loss": 2.6066,
+      "step": 261000
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 2.7704341291799026e-05,
+      "loss": 2.7219,
+      "step": 261100
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 2.7695802165522427e-05,
+      "loss": 2.8082,
+      "step": 261200
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 2.7687263039245825e-05,
+      "loss": 2.4713,
+      "step": 261300
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 2.767872391296923e-05,
+      "loss": 2.5928,
+      "step": 261400
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 2.7670184786692625e-05,
+      "loss": 2.4251,
+      "step": 261500
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 2.766164566041603e-05,
+      "loss": 2.4727,
+      "step": 261600
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 2.7653106534139428e-05,
+      "loss": 2.6667,
+      "step": 261700
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 2.764456740786283e-05,
+      "loss": 2.6748,
+      "step": 261800
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 2.7636028281586228e-05,
+      "loss": 2.7072,
+      "step": 261900
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 2.7627489155309633e-05,
+      "loss": 2.6406,
+      "step": 262000
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 2.761895002903303e-05,
+      "loss": 2.766,
+      "step": 262100
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 2.7610410902756433e-05,
+      "loss": 2.6034,
+      "step": 262200
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 2.760187177647983e-05,
+      "loss": 2.6623,
+      "step": 262300
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 2.7593332650203236e-05,
+      "loss": 2.7657,
+      "step": 262400
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 2.7584793523926634e-05,
+      "loss": 2.6565,
+      "step": 262500
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 2.7576254397650035e-05,
+      "loss": 2.4814,
+      "step": 262600
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 2.7567715271373433e-05,
+      "loss": 2.6561,
+      "step": 262700
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 2.755917614509684e-05,
+      "loss": 2.7845,
+      "step": 262800
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 2.7550637018820237e-05,
+      "loss": 2.7197,
+      "step": 262900
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 2.7542097892543635e-05,
+      "loss": 2.4288,
+      "step": 263000
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 2.7533558766267036e-05,
+      "loss": 2.6412,
+      "step": 263100
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 2.7525019639990434e-05,
+      "loss": 2.6682,
+      "step": 263200
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 2.751648051371384e-05,
+      "loss": 2.6347,
+      "step": 263300
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 2.7507941387437238e-05,
+      "loss": 2.6485,
+      "step": 263400
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 2.749940226116064e-05,
+      "loss": 2.7453,
+      "step": 263500
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 2.7490863134884037e-05,
+      "loss": 2.5326,
+      "step": 263600
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 2.7482324008607442e-05,
+      "loss": 2.7145,
+      "step": 263700
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 2.7473784882330837e-05,
+      "loss": 2.6467,
+      "step": 263800
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 2.7465245756054242e-05,
+      "loss": 2.7114,
+      "step": 263900
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 2.745670662977764e-05,
+      "loss": 2.6683,
+      "step": 264000
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 2.744816750350104e-05,
+      "loss": 2.5731,
+      "step": 264100
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 2.743962837722444e-05,
+      "loss": 2.7838,
+      "step": 264200
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 2.7431089250947845e-05,
+      "loss": 2.6185,
+      "step": 264300
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 2.7422550124671243e-05,
+      "loss": 2.6239,
+      "step": 264400
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 2.7414010998394644e-05,
+      "loss": 2.5764,
+      "step": 264500
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 2.7405471872118043e-05,
+      "loss": 2.6734,
+      "step": 264600
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 2.7396932745841448e-05,
+      "loss": 2.7873,
+      "step": 264700
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 2.7388393619564846e-05,
+      "loss": 2.5816,
+      "step": 264800
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 2.7379854493288247e-05,
+      "loss": 2.5118,
+      "step": 264900
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 2.7371315367011645e-05,
+      "loss": 2.613,
+      "step": 265000
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 2.736277624073505e-05,
+      "loss": 2.6056,
+      "step": 265100
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 2.735423711445845e-05,
+      "loss": 2.6834,
+      "step": 265200
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 2.734569798818185e-05,
+      "loss": 2.6059,
+      "step": 265300
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 2.7337158861905248e-05,
+      "loss": 2.7789,
+      "step": 265400
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 2.7328619735628653e-05,
+      "loss": 2.679,
+      "step": 265500
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 2.732008060935205e-05,
+      "loss": 2.7703,
+      "step": 265600
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 2.7311541483075453e-05,
+      "loss": 2.6233,
+      "step": 265700
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 2.730300235679885e-05,
+      "loss": 2.7459,
+      "step": 265800
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 2.7294463230522256e-05,
+      "loss": 2.5026,
+      "step": 265900
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 2.7285924104245654e-05,
+      "loss": 2.6283,
+      "step": 266000
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 2.7277384977969056e-05,
+      "loss": 2.621,
+      "step": 266100
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 2.7268845851692454e-05,
+      "loss": 2.7844,
+      "step": 266200
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 2.726030672541586e-05,
+      "loss": 2.895,
+      "step": 266300
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 2.7251767599139253e-05,
+      "loss": 2.4861,
+      "step": 266400
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 2.724322847286266e-05,
+      "loss": 2.6531,
+      "step": 266500
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 2.7234689346586057e-05,
+      "loss": 2.6349,
+      "step": 266600
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 2.722615022030946e-05,
+      "loss": 2.6672,
+      "step": 266700
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 2.7217611094032856e-05,
+      "loss": 2.5503,
+      "step": 266800
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 2.720907196775626e-05,
+      "loss": 2.4387,
+      "step": 266900
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 2.720053284147966e-05,
+      "loss": 2.629,
+      "step": 267000
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 2.719199371520306e-05,
+      "loss": 2.7743,
+      "step": 267100
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 2.718345458892646e-05,
+      "loss": 2.6763,
+      "step": 267200
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 2.7174915462649864e-05,
+      "loss": 2.6692,
+      "step": 267300
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 2.7166376336373262e-05,
+      "loss": 2.6085,
+      "step": 267400
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 2.7157837210096664e-05,
+      "loss": 2.8022,
+      "step": 267500
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 2.7149298083820062e-05,
+      "loss": 2.4493,
+      "step": 267600
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 2.7140758957543467e-05,
+      "loss": 2.4776,
+      "step": 267700
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 2.7132219831266865e-05,
+      "loss": 2.5499,
+      "step": 267800
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 2.7123680704990267e-05,
+      "loss": 2.6288,
+      "step": 267900
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 2.7115141578713665e-05,
+      "loss": 2.6296,
+      "step": 268000
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 2.710660245243707e-05,
+      "loss": 2.6508,
+      "step": 268100
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 2.7098063326160468e-05,
+      "loss": 2.4546,
+      "step": 268200
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 2.708952419988387e-05,
+      "loss": 2.6228,
+      "step": 268300
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 2.7080985073607268e-05,
+      "loss": 2.6905,
+      "step": 268400
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 2.7072445947330672e-05,
+      "loss": 2.5638,
+      "step": 268500
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 2.706390682105407e-05,
+      "loss": 2.5877,
+      "step": 268600
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 2.7055367694777472e-05,
+      "loss": 2.7355,
+      "step": 268700
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 2.704682856850087e-05,
+      "loss": 2.5392,
+      "step": 268800
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 2.7038289442224275e-05,
+      "loss": 2.564,
+      "step": 268900
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 2.702975031594767e-05,
+      "loss": 2.7087,
+      "step": 269000
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 2.7021211189671075e-05,
+      "loss": 2.7114,
+      "step": 269100
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 2.7012672063394473e-05,
+      "loss": 2.7633,
+      "step": 269200
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 2.7004132937117878e-05,
+      "loss": 2.6081,
+      "step": 269300
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 2.6995593810841273e-05,
+      "loss": 2.5344,
+      "step": 269400
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 2.6987054684564678e-05,
+      "loss": 2.7626,
+      "step": 269500
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 2.6978515558288076e-05,
+      "loss": 2.5426,
+      "step": 269600
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 2.6969976432011477e-05,
+      "loss": 2.7273,
+      "step": 269700
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 2.6961437305734876e-05,
+      "loss": 2.5759,
+      "step": 269800
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 2.695289817945828e-05,
+      "loss": 2.5017,
+      "step": 269900
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 2.694435905318168e-05,
+      "loss": 2.592,
+      "step": 270000
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 2.693581992690508e-05,
+      "loss": 2.4799,
+      "step": 270100
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 2.692728080062848e-05,
+      "loss": 2.6806,
+      "step": 270200
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 2.6918741674351883e-05,
+      "loss": 2.6202,
+      "step": 270300
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 2.691020254807528e-05,
+      "loss": 2.5736,
+      "step": 270400
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 2.6901663421798683e-05,
+      "loss": 2.5092,
+      "step": 270500
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 2.689312429552208e-05,
+      "loss": 2.5518,
+      "step": 270600
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 2.6884585169245486e-05,
+      "loss": 2.7833,
+      "step": 270700
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 2.6876046042968884e-05,
+      "loss": 2.4738,
+      "step": 270800
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 2.6867506916692286e-05,
+      "loss": 2.6425,
+      "step": 270900
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 2.6858967790415684e-05,
+      "loss": 2.7705,
+      "step": 271000
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 2.685042866413909e-05,
+      "loss": 2.5649,
+      "step": 271100
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 2.6841889537862487e-05,
+      "loss": 2.6953,
+      "step": 271200
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 2.683335041158589e-05,
+      "loss": 2.5427,
+      "step": 271300
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 2.6824811285309287e-05,
+      "loss": 2.7958,
+      "step": 271400
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 2.6816272159032692e-05,
+      "loss": 2.6604,
+      "step": 271500
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 2.680773303275609e-05,
+      "loss": 2.6293,
+      "step": 271600
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 2.679919390647949e-05,
+      "loss": 2.6232,
+      "step": 271700
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 2.679065478020289e-05,
+      "loss": 2.6486,
+      "step": 271800
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 2.6782115653926295e-05,
+      "loss": 2.5863,
+      "step": 271900
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 2.677357652764969e-05,
+      "loss": 2.5454,
+      "step": 272000
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 2.6765037401373094e-05,
+      "loss": 2.5873,
+      "step": 272100
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 2.6756498275096492e-05,
+      "loss": 2.6361,
+      "step": 272200
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 2.6747959148819894e-05,
+      "loss": 2.7453,
+      "step": 272300
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 2.6739420022543292e-05,
+      "loss": 2.6073,
+      "step": 272400
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 2.6730880896266697e-05,
+      "loss": 2.7198,
+      "step": 272500
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 2.6722341769990095e-05,
+      "loss": 2.5774,
+      "step": 272600
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 2.6713802643713497e-05,
+      "loss": 2.6256,
+      "step": 272700
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 2.6705263517436895e-05,
+      "loss": 2.5441,
+      "step": 272800
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 2.66967243911603e-05,
+      "loss": 2.5643,
+      "step": 272900
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 2.6688185264883698e-05,
+      "loss": 2.8956,
+      "step": 273000
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 2.66796461386071e-05,
+      "loss": 2.5068,
+      "step": 273100
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 2.6671107012330498e-05,
+      "loss": 2.7411,
+      "step": 273200
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 2.6662567886053903e-05,
+      "loss": 2.6969,
+      "step": 273300
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 2.66540287597773e-05,
+      "loss": 2.4939,
+      "step": 273400
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.6645489633500702e-05,
+      "loss": 2.5886,
+      "step": 273500
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.66369505072241e-05,
+      "loss": 2.5141,
+      "step": 273600
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.6628411380947506e-05,
+      "loss": 2.6091,
+      "step": 273700
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.6619872254670904e-05,
+      "loss": 2.5974,
+      "step": 273800
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.6611333128394305e-05,
+      "loss": 2.5129,
+      "step": 273900
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.6602794002117703e-05,
+      "loss": 2.5836,
+      "step": 274000
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.659425487584111e-05,
+      "loss": 2.5638,
+      "step": 274100
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.6585715749564506e-05,
+      "loss": 2.5783,
+      "step": 274200
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.6577176623287908e-05,
+      "loss": 2.6268,
+      "step": 274300
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.6568637497011306e-05,
+      "loss": 2.6178,
+      "step": 274400
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.656009837073471e-05,
+      "loss": 2.5375,
+      "step": 274500
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.6551559244458106e-05,
+      "loss": 2.6121,
+      "step": 274600
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.654302011818151e-05,
+      "loss": 2.5868,
+      "step": 274700
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.653448099190491e-05,
+      "loss": 2.6258,
+      "step": 274800
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.652594186562831e-05,
+      "loss": 2.6168,
+      "step": 274900
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.651740273935171e-05,
+      "loss": 2.6585,
+      "step": 275000
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.6508863613075114e-05,
+      "loss": 2.7558,
+      "step": 275100
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.6500324486798512e-05,
+      "loss": 2.4509,
+      "step": 275200
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.6491785360521913e-05,
+      "loss": 2.6579,
+      "step": 275300
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.648324623424531e-05,
+      "loss": 2.5262,
+      "step": 275400
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.6474707107968716e-05,
+      "loss": 2.5844,
+      "step": 275500
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.6466167981692115e-05,
+      "loss": 2.4258,
+      "step": 275600
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.6457628855415516e-05,
+      "loss": 2.6116,
+      "step": 275700
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 2.6449089729138914e-05,
+      "loss": 2.4755,
+      "step": 275800
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 2.644055060286232e-05,
+      "loss": 2.5765,
+      "step": 275900
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 2.6432011476585717e-05,
+      "loss": 2.5403,
+      "step": 276000
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 2.642347235030912e-05,
+      "loss": 2.6769,
+      "step": 276100
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 2.6414933224032517e-05,
+      "loss": 2.4913,
+      "step": 276200
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 2.6406394097755922e-05,
+      "loss": 2.6235,
+      "step": 276300
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 2.639785497147932e-05,
+      "loss": 2.6402,
+      "step": 276400
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 2.6389315845202722e-05,
+      "loss": 2.5187,
+      "step": 276500
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 2.638077671892612e-05,
+      "loss": 2.7239,
+      "step": 276600
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 2.6372237592649525e-05,
+      "loss": 2.5616,
+      "step": 276700
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 2.6363698466372923e-05,
+      "loss": 2.5075,
+      "step": 276800
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 2.6355159340096325e-05,
+      "loss": 2.6405,
+      "step": 276900
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 2.6346620213819723e-05,
+      "loss": 2.7607,
+      "step": 277000
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 2.6338081087543128e-05,
+      "loss": 2.6004,
+      "step": 277100
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 2.6329541961266522e-05,
+      "loss": 2.3485,
+      "step": 277200
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 2.6321002834989927e-05,
+      "loss": 2.6571,
+      "step": 277300
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 2.6312463708713326e-05,
+      "loss": 2.6115,
+      "step": 277400
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 2.6303924582436727e-05,
+      "loss": 2.6166,
+      "step": 277500
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 2.6295385456160125e-05,
+      "loss": 2.6621,
+      "step": 277600
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 2.628684632988353e-05,
+      "loss": 2.652,
+      "step": 277700
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 2.627830720360693e-05,
+      "loss": 2.667,
+      "step": 277800
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 2.626976807733033e-05,
+      "loss": 2.6392,
+      "step": 277900
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 2.6261228951053728e-05,
+      "loss": 2.5462,
+      "step": 278000
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 2.6252689824777133e-05,
+      "loss": 2.514,
+      "step": 278100
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 2.624415069850053e-05,
+      "loss": 2.5877,
+      "step": 278200
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 2.6235611572223933e-05,
+      "loss": 2.6484,
+      "step": 278300
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 2.622707244594733e-05,
+      "loss": 2.772,
+      "step": 278400
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 2.6218533319670736e-05,
+      "loss": 2.6233,
+      "step": 278500
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 2.6209994193394134e-05,
+      "loss": 2.4345,
+      "step": 278600
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 2.6201455067117536e-05,
+      "loss": 2.6611,
+      "step": 278700
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 2.6192915940840934e-05,
+      "loss": 2.6447,
+      "step": 278800
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 2.618437681456434e-05,
+      "loss": 2.5533,
+      "step": 278900
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 2.6175837688287737e-05,
+      "loss": 2.5604,
+      "step": 279000
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 2.616729856201114e-05,
+      "loss": 2.7492,
+      "step": 279100
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 2.6158759435734536e-05,
+      "loss": 2.4866,
+      "step": 279200
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 2.615022030945794e-05,
+      "loss": 2.6712,
+      "step": 279300
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 2.614168118318134e-05,
+      "loss": 2.5109,
+      "step": 279400
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 2.613314205690474e-05,
+      "loss": 2.5113,
+      "step": 279500
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 2.612460293062814e-05,
+      "loss": 2.609,
+      "step": 279600
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 2.6116063804351544e-05,
+      "loss": 2.5117,
+      "step": 279700
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 2.610752467807494e-05,
+      "loss": 2.7068,
+      "step": 279800
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 2.6098985551798344e-05,
+      "loss": 2.6591,
+      "step": 279900
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 2.6090446425521742e-05,
+      "loss": 2.5596,
+      "step": 280000
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 2.6081907299245144e-05,
+      "loss": 2.7538,
+      "step": 280100
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 2.6073368172968542e-05,
+      "loss": 2.5003,
+      "step": 280200
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 2.6064829046691947e-05,
+      "loss": 2.6085,
+      "step": 280300
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 2.6056289920415345e-05,
+      "loss": 2.5711,
+      "step": 280400
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 2.6047750794138743e-05,
+      "loss": 2.6625,
+      "step": 280500
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 2.6039211667862145e-05,
+      "loss": 2.6042,
+      "step": 280600
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 2.6030672541585543e-05,
+      "loss": 2.5667,
+      "step": 280700
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 2.6022133415308948e-05,
+      "loss": 2.5048,
+      "step": 280800
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 2.6013594289032346e-05,
+      "loss": 2.5722,
+      "step": 280900
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 2.6005055162755747e-05,
+      "loss": 2.5667,
+      "step": 281000
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 2.5996516036479146e-05,
+      "loss": 2.6249,
+      "step": 281100
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 2.598797691020255e-05,
+      "loss": 2.4988,
+      "step": 281200
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 2.597943778392595e-05,
+      "loss": 2.5008,
+      "step": 281300
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 2.597089865764935e-05,
+      "loss": 2.7434,
+      "step": 281400
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 2.596235953137275e-05,
+      "loss": 2.6359,
+      "step": 281500
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 2.5953820405096153e-05,
+      "loss": 2.7356,
+      "step": 281600
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 2.594528127881955e-05,
+      "loss": 2.3953,
+      "step": 281700
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 2.5936742152542953e-05,
+      "loss": 2.4301,
+      "step": 281800
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 2.592820302626635e-05,
+      "loss": 2.5427,
+      "step": 281900
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 2.5919663899989756e-05,
+      "loss": 2.526,
+      "step": 282000
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 2.591112477371315e-05,
+      "loss": 2.6178,
+      "step": 282100
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 2.5902585647436556e-05,
+      "loss": 2.6561,
+      "step": 282200
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 2.5894046521159954e-05,
+      "loss": 2.4964,
+      "step": 282300
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 2.5885507394883356e-05,
+      "loss": 2.4569,
+      "step": 282400
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 2.5876968268606754e-05,
+      "loss": 2.5758,
+      "step": 282500
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 2.586842914233016e-05,
+      "loss": 2.5038,
+      "step": 282600
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 2.5859890016053557e-05,
+      "loss": 2.5867,
+      "step": 282700
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 2.585135088977696e-05,
+      "loss": 2.7712,
+      "step": 282800
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 2.5842811763500356e-05,
+      "loss": 2.3986,
+      "step": 282900
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 2.583427263722376e-05,
+      "loss": 2.6106,
+      "step": 283000
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 2.582573351094716e-05,
+      "loss": 2.6787,
+      "step": 283100
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 2.581719438467056e-05,
+      "loss": 2.586,
+      "step": 283200
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 2.580865525839396e-05,
+      "loss": 2.4844,
+      "step": 283300
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 2.5800116132117364e-05,
+      "loss": 2.7313,
+      "step": 283400
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 2.5791577005840762e-05,
+      "loss": 2.6168,
+      "step": 283500
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 2.5783037879564164e-05,
+      "loss": 2.4937,
+      "step": 283600
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 2.5774498753287562e-05,
+      "loss": 2.7365,
+      "step": 283700
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 2.5765959627010967e-05,
+      "loss": 2.626,
+      "step": 283800
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 2.5757420500734365e-05,
+      "loss": 2.7652,
+      "step": 283900
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 2.5748881374457767e-05,
+      "loss": 2.7086,
+      "step": 284000
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 2.5740342248181165e-05,
+      "loss": 2.5696,
+      "step": 284100
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 2.573180312190457e-05,
+      "loss": 2.6369,
+      "step": 284200
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 2.5723263995627968e-05,
+      "loss": 2.6651,
+      "step": 284300
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 2.571472486935137e-05,
+      "loss": 2.7067,
+      "step": 284400
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 2.5706185743074768e-05,
+      "loss": 2.3506,
+      "step": 284500
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 2.5697646616798173e-05,
+      "loss": 2.549,
+      "step": 284600
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 2.5689107490521567e-05,
+      "loss": 2.7959,
+      "step": 284700
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 2.5680568364244972e-05,
+      "loss": 2.7381,
+      "step": 284800
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 2.567202923796837e-05,
+      "loss": 2.6734,
+      "step": 284900
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 2.5663490111691772e-05,
+      "loss": 2.5239,
+      "step": 285000
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 2.565495098541517e-05,
+      "loss": 2.5626,
+      "step": 285100
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 2.5646411859138575e-05,
+      "loss": 2.4925,
+      "step": 285200
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 2.5637872732861973e-05,
+      "loss": 2.6307,
+      "step": 285300
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 2.5629333606585375e-05,
+      "loss": 2.4771,
+      "step": 285400
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 2.5620794480308773e-05,
+      "loss": 2.5061,
+      "step": 285500
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 2.5612255354032178e-05,
+      "loss": 2.5368,
+      "step": 285600
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 2.5603716227755576e-05,
+      "loss": 2.7254,
+      "step": 285700
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 2.5595177101478978e-05,
+      "loss": 2.6842,
+      "step": 285800
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 2.5586637975202376e-05,
+      "loss": 2.7254,
+      "step": 285900
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 2.557809884892578e-05,
+      "loss": 2.6652,
+      "step": 286000
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 2.556955972264918e-05,
+      "loss": 2.473,
+      "step": 286100
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 2.556102059637258e-05,
+      "loss": 2.6769,
+      "step": 286200
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 2.555248147009598e-05,
+      "loss": 2.6187,
+      "step": 286300
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 2.5543942343819384e-05,
+      "loss": 2.6971,
+      "step": 286400
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 2.5535403217542782e-05,
+      "loss": 2.5785,
+      "step": 286500
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 2.5526864091266183e-05,
+      "loss": 2.6179,
+      "step": 286600
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 2.551832496498958e-05,
+      "loss": 2.4922,
+      "step": 286700
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 2.5509785838712986e-05,
+      "loss": 2.623,
+      "step": 286800
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 2.5501246712436385e-05,
+      "loss": 2.6506,
+      "step": 286900
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 2.5492707586159786e-05,
+      "loss": 2.4052,
+      "step": 287000
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 2.5484168459883184e-05,
+      "loss": 2.5711,
+      "step": 287100
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 2.547562933360659e-05,
+      "loss": 2.7055,
+      "step": 287200
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 2.5467090207329984e-05,
+      "loss": 2.4519,
+      "step": 287300
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 2.545855108105339e-05,
+      "loss": 2.6229,
+      "step": 287400
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 2.5450011954776787e-05,
+      "loss": 2.6368,
+      "step": 287500
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 2.544147282850019e-05,
+      "loss": 2.7205,
+      "step": 287600
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 2.5432933702223587e-05,
+      "loss": 2.6398,
+      "step": 287700
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 2.542439457594699e-05,
+      "loss": 2.5763,
+      "step": 287800
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 2.541585544967039e-05,
+      "loss": 2.5818,
+      "step": 287900
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 2.540731632339379e-05,
+      "loss": 2.5528,
+      "step": 288000
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 2.539877719711719e-05,
+      "loss": 2.5105,
+      "step": 288100
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 2.5390238070840594e-05,
+      "loss": 2.5778,
+      "step": 288200
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 2.5381698944563993e-05,
+      "loss": 2.6856,
+      "step": 288300
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 2.5373159818287394e-05,
+      "loss": 2.6557,
+      "step": 288400
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 2.5364620692010792e-05,
+      "loss": 2.5503,
+      "step": 288500
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 2.5356081565734197e-05,
+      "loss": 2.5558,
+      "step": 288600
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 2.5347542439457595e-05,
+      "loss": 2.6848,
+      "step": 288700
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 2.5339003313180997e-05,
+      "loss": 2.5382,
+      "step": 288800
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 2.5330464186904395e-05,
+      "loss": 2.6446,
+      "step": 288900
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 2.53219250606278e-05,
+      "loss": 2.7562,
+      "step": 289000
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 2.5313385934351198e-05,
+      "loss": 2.4963,
+      "step": 289100
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 2.53048468080746e-05,
+      "loss": 2.5525,
+      "step": 289200
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 2.5296307681797998e-05,
+      "loss": 2.6787,
+      "step": 289300
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 2.5287768555521403e-05,
+      "loss": 2.6217,
+      "step": 289400
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 2.52792294292448e-05,
+      "loss": 2.3872,
+      "step": 289500
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 2.5270690302968203e-05,
+      "loss": 2.7578,
+      "step": 289600
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 2.52621511766916e-05,
+      "loss": 2.5643,
+      "step": 289700
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 2.5253612050415006e-05,
+      "loss": 2.4704,
+      "step": 289800
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 2.52450729241384e-05,
+      "loss": 2.5876,
+      "step": 289900
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 2.5236533797861805e-05,
+      "loss": 2.6602,
+      "step": 290000
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 2.5227994671585204e-05,
+      "loss": 2.4969,
+      "step": 290100
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 2.5219455545308605e-05,
+      "loss": 2.6634,
+      "step": 290200
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 2.5210916419032003e-05,
+      "loss": 2.5549,
+      "step": 290300
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 2.5202377292755408e-05,
+      "loss": 2.5112,
+      "step": 290400
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 2.5193838166478806e-05,
+      "loss": 2.4683,
+      "step": 290500
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 2.5185299040202208e-05,
+      "loss": 2.605,
+      "step": 290600
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 2.5176759913925606e-05,
+      "loss": 2.5804,
+      "step": 290700
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 2.516822078764901e-05,
+      "loss": 2.476,
+      "step": 290800
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 2.515968166137241e-05,
+      "loss": 2.672,
+      "step": 290900
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 2.515114253509581e-05,
+      "loss": 2.7042,
+      "step": 291000
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 2.514260340881921e-05,
+      "loss": 2.4737,
+      "step": 291100
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 2.5134064282542614e-05,
+      "loss": 2.6669,
+      "step": 291200
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 2.5125525156266012e-05,
+      "loss": 2.5015,
+      "step": 291300
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 2.5116986029989414e-05,
+      "loss": 2.5606,
+      "step": 291400
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 2.5108446903712812e-05,
+      "loss": 2.5226,
+      "step": 291500
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 2.5099907777436217e-05,
+      "loss": 2.6841,
+      "step": 291600
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 2.5091368651159615e-05,
+      "loss": 2.5918,
+      "step": 291700
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 2.5082829524883016e-05,
+      "loss": 2.6181,
+      "step": 291800
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 2.5074290398606415e-05,
+      "loss": 2.5589,
+      "step": 291900
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 2.506575127232982e-05,
+      "loss": 2.5423,
+      "step": 292000
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 2.5057212146053218e-05,
+      "loss": 2.7337,
+      "step": 292100
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 2.504867301977662e-05,
+      "loss": 2.6391,
+      "step": 292200
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 2.5040133893500017e-05,
+      "loss": 2.3304,
+      "step": 292300
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 2.5031594767223422e-05,
+      "loss": 2.4947,
+      "step": 292400
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 2.5023055640946817e-05,
+      "loss": 2.49,
+      "step": 292500
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 2.5014516514670222e-05,
+      "loss": 2.5896,
+      "step": 292600
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 2.500597738839362e-05,
+      "loss": 2.4626,
+      "step": 292700
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 2.499743826211702e-05,
+      "loss": 2.6996,
+      "step": 292800
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 2.498889913584042e-05,
+      "loss": 2.556,
+      "step": 292900
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 2.498036000956382e-05,
+      "loss": 2.7386,
+      "step": 293000
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 2.4971820883287223e-05,
+      "loss": 2.5582,
+      "step": 293100
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 2.4963281757010624e-05,
+      "loss": 2.5779,
+      "step": 293200
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 2.4954742630734023e-05,
+      "loss": 2.5479,
+      "step": 293300
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 2.4946203504457424e-05,
+      "loss": 2.6817,
+      "step": 293400
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 2.4937664378180826e-05,
+      "loss": 2.5313,
+      "step": 293500
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 2.4929125251904227e-05,
+      "loss": 2.7259,
+      "step": 293600
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 2.4920586125627625e-05,
+      "loss": 2.674,
+      "step": 293700
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 2.4912046999351027e-05,
+      "loss": 2.3784,
+      "step": 293800
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 2.490350787307443e-05,
+      "loss": 2.4276,
+      "step": 293900
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 2.4894968746797827e-05,
+      "loss": 2.6637,
+      "step": 294000
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 2.4886429620521228e-05,
+      "loss": 2.4605,
+      "step": 294100
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 2.487789049424463e-05,
+      "loss": 2.5052,
+      "step": 294200
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 2.486935136796803e-05,
+      "loss": 2.6614,
+      "step": 294300
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 2.486081224169143e-05,
+      "loss": 2.5257,
+      "step": 294400
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 2.485227311541483e-05,
+      "loss": 2.5516,
+      "step": 294500
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 2.4843733989138233e-05,
+      "loss": 2.7613,
+      "step": 294600
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 2.4835194862861634e-05,
+      "loss": 2.5585,
+      "step": 294700
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 2.4826655736585032e-05,
+      "loss": 2.5098,
+      "step": 294800
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 2.4818116610308434e-05,
+      "loss": 2.607,
+      "step": 294900
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 2.4809577484031835e-05,
+      "loss": 2.634,
+      "step": 295000
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 2.4801038357755234e-05,
+      "loss": 2.443,
+      "step": 295100
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 2.4792499231478635e-05,
+      "loss": 2.5236,
+      "step": 295200
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 2.4783960105202037e-05,
+      "loss": 2.6182,
+      "step": 295300
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 2.4775420978925438e-05,
+      "loss": 2.5835,
+      "step": 295400
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 2.4766881852648836e-05,
+      "loss": 2.8453,
+      "step": 295500
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 2.4758342726372238e-05,
+      "loss": 2.6775,
+      "step": 295600
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 2.474980360009564e-05,
+      "loss": 2.7386,
+      "step": 295700
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 2.474126447381904e-05,
+      "loss": 2.56,
+      "step": 295800
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 2.473272534754244e-05,
+      "loss": 2.703,
+      "step": 295900
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 2.472418622126584e-05,
+      "loss": 2.6336,
+      "step": 296000
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 2.4715647094989242e-05,
+      "loss": 2.8031,
+      "step": 296100
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 2.4707107968712644e-05,
+      "loss": 2.5844,
+      "step": 296200
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 2.4698568842436042e-05,
+      "loss": 2.4344,
+      "step": 296300
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 2.4690029716159444e-05,
+      "loss": 2.5045,
+      "step": 296400
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 2.4681490589882845e-05,
+      "loss": 2.6255,
+      "step": 296500
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 2.4672951463606243e-05,
+      "loss": 2.4738,
+      "step": 296600
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 2.4664412337329645e-05,
+      "loss": 2.8838,
+      "step": 296700
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 2.4655873211053046e-05,
+      "loss": 2.5313,
+      "step": 296800
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 2.4647334084776448e-05,
+      "loss": 2.4704,
+      "step": 296900
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 2.4638794958499846e-05,
+      "loss": 2.7296,
+      "step": 297000
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 2.4630255832223248e-05,
+      "loss": 2.4954,
+      "step": 297100
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 2.462171670594665e-05,
+      "loss": 2.4362,
+      "step": 297200
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 2.461317757967005e-05,
+      "loss": 2.6174,
+      "step": 297300
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 2.460463845339345e-05,
+      "loss": 2.7054,
+      "step": 297400
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 2.459609932711685e-05,
+      "loss": 2.6661,
+      "step": 297500
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 2.4587560200840252e-05,
+      "loss": 2.6271,
+      "step": 297600
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 2.457902107456365e-05,
+      "loss": 2.6175,
+      "step": 297700
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 2.457048194828705e-05,
+      "loss": 2.6022,
+      "step": 297800
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 2.4561942822010453e-05,
+      "loss": 2.4242,
+      "step": 297900
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 2.4553403695733855e-05,
+      "loss": 2.5952,
+      "step": 298000
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 2.4544864569457253e-05,
+      "loss": 2.5788,
+      "step": 298100
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 2.4536325443180654e-05,
+      "loss": 2.6518,
+      "step": 298200
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 2.4527786316904056e-05,
+      "loss": 2.5203,
+      "step": 298300
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 2.4519247190627458e-05,
+      "loss": 2.605,
+      "step": 298400
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 2.4510708064350856e-05,
+      "loss": 2.6588,
+      "step": 298500
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 2.4502168938074257e-05,
+      "loss": 2.5363,
+      "step": 298600
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 2.449362981179766e-05,
+      "loss": 2.4543,
+      "step": 298700
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 2.448509068552106e-05,
+      "loss": 2.7616,
+      "step": 298800
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 2.447655155924446e-05,
+      "loss": 2.6632,
+      "step": 298900
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 2.446801243296786e-05,
+      "loss": 2.5355,
+      "step": 299000
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 2.445947330669126e-05,
+      "loss": 2.8136,
+      "step": 299100
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 2.445093418041466e-05,
+      "loss": 2.7673,
+      "step": 299200
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 2.444239505413806e-05,
+      "loss": 2.6112,
+      "step": 299300
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 2.4433855927861463e-05,
+      "loss": 2.616,
+      "step": 299400
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 2.4425316801584864e-05,
+      "loss": 2.6405,
+      "step": 299500
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 2.4416777675308263e-05,
+      "loss": 2.6973,
+      "step": 299600
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 2.4408238549031664e-05,
+      "loss": 2.7004,
+      "step": 299700
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 2.4399699422755066e-05,
+      "loss": 2.5356,
+      "step": 299800
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 2.4391160296478467e-05,
+      "loss": 2.6527,
+      "step": 299900
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 2.4382621170201865e-05,
+      "loss": 2.4687,
+      "step": 300000
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 2.4374082043925267e-05,
+      "loss": 2.3935,
+      "step": 300100
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 2.436554291764867e-05,
+      "loss": 2.6342,
+      "step": 300200
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 2.435700379137207e-05,
+      "loss": 2.682,
+      "step": 300300
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.4348464665095468e-05,
+      "loss": 2.4398,
+      "step": 300400
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.433992553881887e-05,
+      "loss": 2.7451,
+      "step": 300500
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.433138641254227e-05,
+      "loss": 2.4977,
+      "step": 300600
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.432284728626567e-05,
+      "loss": 2.4256,
+      "step": 300700
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.431430815998907e-05,
+      "loss": 2.593,
+      "step": 300800
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.4305769033712473e-05,
+      "loss": 2.3915,
+      "step": 300900
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.4297229907435874e-05,
+      "loss": 2.5495,
+      "step": 301000
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.4288690781159272e-05,
+      "loss": 2.8005,
+      "step": 301100
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.4280151654882674e-05,
+      "loss": 2.5199,
+      "step": 301200
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.4271612528606075e-05,
+      "loss": 2.6356,
+      "step": 301300
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.4263073402329477e-05,
+      "loss": 2.6186,
+      "step": 301400
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.4254534276052875e-05,
+      "loss": 2.3638,
+      "step": 301500
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.4245995149776277e-05,
+      "loss": 2.4584,
+      "step": 301600
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.4237456023499678e-05,
+      "loss": 2.4968,
+      "step": 301700
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.4228916897223076e-05,
+      "loss": 2.4364,
+      "step": 301800
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.4220377770946478e-05,
+      "loss": 2.4099,
+      "step": 301900
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.421183864466988e-05,
+      "loss": 2.5179,
+      "step": 302000
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.420329951839328e-05,
+      "loss": 2.4713,
+      "step": 302100
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.419476039211668e-05,
+      "loss": 2.6405,
+      "step": 302200
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.418622126584008e-05,
+      "loss": 2.4238,
+      "step": 302300
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.4177682139563482e-05,
+      "loss": 2.4564,
+      "step": 302400
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.4169143013286884e-05,
+      "loss": 2.554,
+      "step": 302500
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.4160603887010282e-05,
+      "loss": 2.438,
+      "step": 302600
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.4152064760733683e-05,
+      "loss": 2.4269,
+      "step": 302700
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.4143525634457085e-05,
+      "loss": 2.4377,
+      "step": 302800
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.4134986508180487e-05,
+      "loss": 2.5265,
+      "step": 302900
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.4126447381903885e-05,
+      "loss": 2.639,
+      "step": 303000
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.4117908255627286e-05,
+      "loss": 2.4689,
+      "step": 303100
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.4109369129350688e-05,
+      "loss": 2.601,
+      "step": 303200
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.4100830003074086e-05,
+      "loss": 2.5779,
+      "step": 303300
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.4092290876797488e-05,
+      "loss": 2.6439,
+      "step": 303400
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.408375175052089e-05,
+      "loss": 2.6182,
+      "step": 303500
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.407521262424429e-05,
+      "loss": 2.6911,
+      "step": 303600
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.406667349796769e-05,
+      "loss": 2.453,
+      "step": 303700
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.405813437169109e-05,
+      "loss": 2.5643,
+      "step": 303800
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.4049595245414492e-05,
+      "loss": 2.6153,
+      "step": 303900
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.4041056119137893e-05,
+      "loss": 2.5002,
+      "step": 304000
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.403251699286129e-05,
+      "loss": 2.5769,
+      "step": 304100
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.4023977866584693e-05,
+      "loss": 2.4474,
+      "step": 304200
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.4015438740308095e-05,
+      "loss": 2.5044,
+      "step": 304300
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.4006899614031493e-05,
+      "loss": 2.5741,
+      "step": 304400
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.3998360487754894e-05,
+      "loss": 2.612,
+      "step": 304500
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.3989821361478296e-05,
+      "loss": 2.3832,
+      "step": 304600
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 2.3981282235201697e-05,
+      "loss": 2.5406,
+      "step": 304700
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 2.3972743108925096e-05,
+      "loss": 2.5468,
+      "step": 304800
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 2.3964203982648497e-05,
+      "loss": 2.4277,
+      "step": 304900
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 2.39556648563719e-05,
+      "loss": 2.4712,
+      "step": 305000
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 2.39471257300953e-05,
+      "loss": 2.5574,
+      "step": 305100
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 2.39385866038187e-05,
+      "loss": 2.6474,
+      "step": 305200
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 2.39300474775421e-05,
+      "loss": 2.8032,
+      "step": 305300
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 2.39215083512655e-05,
+      "loss": 2.5209,
+      "step": 305400
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 2.3912969224988903e-05,
+      "loss": 2.5471,
+      "step": 305500
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 2.39044300987123e-05,
+      "loss": 2.5731,
+      "step": 305600
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 2.3895890972435703e-05,
+      "loss": 2.5844,
+      "step": 305700
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 2.3887351846159104e-05,
+      "loss": 2.4313,
+      "step": 305800
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 2.3878812719882503e-05,
+      "loss": 2.5094,
+      "step": 305900
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 2.3870273593605904e-05,
+      "loss": 2.539,
+      "step": 306000
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 2.3861734467329306e-05,
+      "loss": 2.5007,
+      "step": 306100
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 2.3853195341052707e-05,
+      "loss": 2.4104,
+      "step": 306200
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 2.3844656214776105e-05,
+      "loss": 2.6048,
+      "step": 306300
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 2.3836117088499507e-05,
+      "loss": 2.7752,
+      "step": 306400
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 2.382757796222291e-05,
+      "loss": 2.6126,
+      "step": 306500
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 2.381903883594631e-05,
+      "loss": 2.4156,
+      "step": 306600
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 2.3810499709669708e-05,
+      "loss": 2.5983,
+      "step": 306700
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 2.3801960583393106e-05,
+      "loss": 2.4602,
+      "step": 306800
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 2.3793421457116508e-05,
+      "loss": 2.6034,
+      "step": 306900
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 2.378488233083991e-05,
+      "loss": 2.5708,
+      "step": 307000
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 2.3776343204563308e-05,
+      "loss": 2.4888,
+      "step": 307100
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 2.376780407828671e-05,
+      "loss": 2.547,
+      "step": 307200
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 2.375926495201011e-05,
+      "loss": 2.5943,
+      "step": 307300
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 2.3750725825733512e-05,
+      "loss": 2.5553,
+      "step": 307400
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 2.374218669945691e-05,
+      "loss": 2.5765,
+      "step": 307500
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 2.3733647573180312e-05,
+      "loss": 2.5925,
+      "step": 307600
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 2.3725108446903713e-05,
+      "loss": 2.4431,
+      "step": 307700
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 2.3716569320627115e-05,
+      "loss": 2.6363,
+      "step": 307800
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 2.3708030194350513e-05,
+      "loss": 2.6017,
+      "step": 307900
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 2.3699491068073915e-05,
+      "loss": 2.4631,
+      "step": 308000
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 2.3690951941797316e-05,
+      "loss": 2.4872,
+      "step": 308100
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 2.3682412815520714e-05,
+      "loss": 2.5542,
+      "step": 308200
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 2.3673873689244116e-05,
+      "loss": 2.5386,
+      "step": 308300
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 2.3665334562967517e-05,
+      "loss": 2.363,
+      "step": 308400
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 2.365679543669092e-05,
+      "loss": 2.5627,
+      "step": 308500
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 2.3648256310414317e-05,
+      "loss": 2.5397,
+      "step": 308600
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 2.363971718413772e-05,
+      "loss": 2.571,
+      "step": 308700
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 2.363117805786112e-05,
+      "loss": 2.5536,
+      "step": 308800
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 2.3622638931584522e-05,
+      "loss": 2.4963,
+      "step": 308900
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 2.361409980530792e-05,
+      "loss": 2.538,
+      "step": 309000
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 2.360556067903132e-05,
+      "loss": 2.5102,
+      "step": 309100
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 2.3597021552754723e-05,
+      "loss": 2.3795,
+      "step": 309200
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 2.358848242647812e-05,
+      "loss": 2.6052,
+      "step": 309300
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 2.3579943300201523e-05,
+      "loss": 2.6662,
+      "step": 309400
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 2.3571404173924924e-05,
+      "loss": 2.6022,
+      "step": 309500
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 2.3562865047648326e-05,
+      "loss": 2.5958,
+      "step": 309600
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 2.3554325921371724e-05,
+      "loss": 2.6017,
+      "step": 309700
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 2.3545786795095126e-05,
+      "loss": 2.5168,
+      "step": 309800
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 2.3537247668818527e-05,
+      "loss": 2.3941,
+      "step": 309900
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 2.352870854254193e-05,
+      "loss": 2.5736,
+      "step": 310000
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 2.3520169416265327e-05,
+      "loss": 2.6677,
+      "step": 310100
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 2.351163028998873e-05,
+      "loss": 2.5065,
+      "step": 310200
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 2.350309116371213e-05,
+      "loss": 2.6352,
+      "step": 310300
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 2.349455203743553e-05,
+      "loss": 2.5603,
+      "step": 310400
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 2.348601291115893e-05,
+      "loss": 2.6301,
+      "step": 310500
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 2.347747378488233e-05,
+      "loss": 2.6784,
+      "step": 310600
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 2.3468934658605733e-05,
+      "loss": 2.4355,
+      "step": 310700
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 2.346039553232913e-05,
+      "loss": 2.6839,
+      "step": 310800
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 2.3451856406052532e-05,
+      "loss": 2.665,
+      "step": 310900
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.3443317279775934e-05,
+      "loss": 2.5209,
+      "step": 311000
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.3434778153499336e-05,
+      "loss": 2.4111,
+      "step": 311100
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.3426239027222734e-05,
+      "loss": 2.6543,
+      "step": 311200
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.3417699900946135e-05,
+      "loss": 2.5009,
+      "step": 311300
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.3409160774669537e-05,
+      "loss": 2.4491,
+      "step": 311400
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.340062164839294e-05,
+      "loss": 2.6839,
+      "step": 311500
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.3392082522116337e-05,
+      "loss": 2.4937,
+      "step": 311600
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.3383543395839738e-05,
+      "loss": 2.6359,
+      "step": 311700
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 2.337500426956314e-05,
+      "loss": 2.591,
+      "step": 311800
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 2.3366465143286538e-05,
+      "loss": 2.5896,
+      "step": 311900
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 2.335792601700994e-05,
+      "loss": 2.4467,
+      "step": 312000
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.334938689073334e-05,
+      "loss": 2.3812,
+      "step": 312100
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.3340847764456742e-05,
+      "loss": 2.3776,
+      "step": 312200
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.333230863818014e-05,
+      "loss": 2.6316,
+      "step": 312300
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.3323769511903542e-05,
+      "loss": 2.5576,
+      "step": 312400
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.3315230385626944e-05,
+      "loss": 2.5214,
+      "step": 312500
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.3306691259350345e-05,
+      "loss": 2.5585,
+      "step": 312600
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.3298152133073743e-05,
+      "loss": 2.5395,
+      "step": 312700
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.3289613006797145e-05,
+      "loss": 2.4142,
+      "step": 312800
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.3281073880520547e-05,
+      "loss": 2.576,
+      "step": 312900
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.3272534754243948e-05,
+      "loss": 2.5555,
+      "step": 313000
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.3263995627967346e-05,
+      "loss": 2.5797,
+      "step": 313100
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.3255456501690748e-05,
+      "loss": 2.5472,
+      "step": 313200
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.324691737541415e-05,
+      "loss": 2.4686,
+      "step": 313300
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.3238378249137547e-05,
+      "loss": 2.5522,
+      "step": 313400
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.322983912286095e-05,
+      "loss": 2.4373,
+      "step": 313500
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.322129999658435e-05,
+      "loss": 2.487,
+      "step": 313600
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.3212760870307752e-05,
+      "loss": 2.3727,
+      "step": 313700
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.320422174403115e-05,
+      "loss": 2.6973,
+      "step": 313800
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.3195682617754552e-05,
+      "loss": 2.3589,
+      "step": 313900
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.3187143491477953e-05,
+      "loss": 2.6867,
+      "step": 314000
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.3178604365201355e-05,
+      "loss": 2.6313,
+      "step": 314100
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.3170065238924753e-05,
+      "loss": 2.4842,
+      "step": 314200
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.3161526112648155e-05,
+      "loss": 2.5929,
+      "step": 314300
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.3152986986371556e-05,
+      "loss": 2.4353,
+      "step": 314400
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.3144447860094954e-05,
+      "loss": 2.5274,
+      "step": 314500
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.3135908733818356e-05,
+      "loss": 2.3902,
+      "step": 314600
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.3127369607541757e-05,
+      "loss": 2.5433,
+      "step": 314700
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.311883048126516e-05,
+      "loss": 2.7213,
+      "step": 314800
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 2.3110291354988557e-05,
+      "loss": 2.7245,
+      "step": 314900
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 2.310175222871196e-05,
+      "loss": 2.6024,
+      "step": 315000
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 2.309321310243536e-05,
+      "loss": 2.5649,
+      "step": 315100
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 2.3084673976158762e-05,
+      "loss": 2.5184,
+      "step": 315200
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 2.307613484988216e-05,
+      "loss": 2.3627,
+      "step": 315300
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 2.306759572360556e-05,
+      "loss": 2.4564,
+      "step": 315400
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 2.3059056597328963e-05,
+      "loss": 2.5968,
+      "step": 315500
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 2.3050517471052365e-05,
+      "loss": 2.5117,
+      "step": 315600
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 2.3041978344775763e-05,
+      "loss": 2.5541,
+      "step": 315700
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 2.3033439218499164e-05,
+      "loss": 2.6094,
+      "step": 315800
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 2.3024900092222566e-05,
+      "loss": 2.5938,
+      "step": 315900
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 2.3016360965945964e-05,
+      "loss": 2.5505,
+      "step": 316000
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 2.3007821839669366e-05,
+      "loss": 2.5796,
+      "step": 316100
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 2.2999282713392767e-05,
+      "loss": 2.4089,
+      "step": 316200
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 2.299074358711617e-05,
+      "loss": 2.5582,
+      "step": 316300
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 2.2982204460839567e-05,
+      "loss": 2.5246,
+      "step": 316400
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 2.297366533456297e-05,
+      "loss": 2.5751,
+      "step": 316500
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 2.296512620828637e-05,
+      "loss": 2.6564,
+      "step": 316600
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 2.295658708200977e-05,
+      "loss": 2.552,
+      "step": 316700
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 2.294804795573317e-05,
+      "loss": 2.5467,
+      "step": 316800
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 2.293950882945657e-05,
+      "loss": 2.4852,
+      "step": 316900
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 2.2930969703179973e-05,
+      "loss": 2.4988,
+      "step": 317000
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 2.2922430576903374e-05,
+      "loss": 2.505,
+      "step": 317100
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 2.2913891450626772e-05,
+      "loss": 2.4597,
+      "step": 317200
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 2.2905352324350174e-05,
+      "loss": 2.4072,
+      "step": 317300
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 2.2896813198073576e-05,
+      "loss": 2.3633,
+      "step": 317400
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 2.2888274071796974e-05,
+      "loss": 2.6036,
+      "step": 317500
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 2.2879734945520375e-05,
+      "loss": 2.5262,
+      "step": 317600
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 2.2871195819243777e-05,
+      "loss": 2.722,
+      "step": 317700
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 2.286265669296718e-05,
+      "loss": 2.6565,
+      "step": 317800
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 2.2854117566690576e-05,
+      "loss": 2.4619,
+      "step": 317900
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 2.2845578440413978e-05,
+      "loss": 2.5398,
+      "step": 318000
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 2.283703931413738e-05,
+      "loss": 2.6577,
+      "step": 318100
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 2.282850018786078e-05,
+      "loss": 2.4912,
+      "step": 318200
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 2.281996106158418e-05,
+      "loss": 2.5452,
+      "step": 318300
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 2.281142193530758e-05,
+      "loss": 2.6178,
+      "step": 318400
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 2.2802882809030982e-05,
+      "loss": 2.6458,
+      "step": 318500
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 2.279434368275438e-05,
+      "loss": 2.6709,
+      "step": 318600
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 2.2785804556477782e-05,
+      "loss": 2.5509,
+      "step": 318700
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 2.2777265430201184e-05,
+      "loss": 2.5017,
+      "step": 318800
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 2.2768726303924585e-05,
+      "loss": 2.483,
+      "step": 318900
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 2.2760187177647983e-05,
+      "loss": 2.7111,
+      "step": 319000
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 2.2751648051371385e-05,
+      "loss": 2.5306,
+      "step": 319100
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 2.2743108925094786e-05,
+      "loss": 2.6905,
+      "step": 319200
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 2.2734569798818188e-05,
+      "loss": 2.5214,
+      "step": 319300
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 2.2726030672541586e-05,
+      "loss": 2.5986,
+      "step": 319400
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 2.2717491546264988e-05,
+      "loss": 2.4584,
+      "step": 319500
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 2.270895241998839e-05,
+      "loss": 2.4888,
+      "step": 319600
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 2.270041329371179e-05,
+      "loss": 2.655,
+      "step": 319700
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 2.269187416743519e-05,
+      "loss": 2.4383,
+      "step": 319800
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 2.268333504115859e-05,
+      "loss": 2.557,
+      "step": 319900
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 2.2674795914881992e-05,
+      "loss": 2.6075,
+      "step": 320000
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 2.266625678860539e-05,
+      "loss": 2.6676,
+      "step": 320100
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 2.2657717662328792e-05,
+      "loss": 2.5704,
+      "step": 320200
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 2.2649178536052193e-05,
+      "loss": 2.5918,
+      "step": 320300
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 2.2640639409775595e-05,
+      "loss": 2.6089,
+      "step": 320400
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 2.2632100283498993e-05,
+      "loss": 2.5962,
+      "step": 320500
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 2.2623561157222395e-05,
+      "loss": 2.5116,
+      "step": 320600
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 2.2615022030945796e-05,
+      "loss": 2.6161,
+      "step": 320700
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 2.2606482904669198e-05,
+      "loss": 2.4271,
+      "step": 320800
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 2.2597943778392596e-05,
+      "loss": 2.5744,
+      "step": 320900
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 2.2589404652115997e-05,
+      "loss": 2.4381,
+      "step": 321000
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 2.25808655258394e-05,
+      "loss": 2.4659,
+      "step": 321100
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 2.2572326399562797e-05,
+      "loss": 2.566,
+      "step": 321200
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 2.25637872732862e-05,
+      "loss": 2.5267,
+      "step": 321300
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 2.25552481470096e-05,
+      "loss": 2.6958,
+      "step": 321400
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 2.2546709020733002e-05,
+      "loss": 2.4422,
+      "step": 321500
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 2.25381698944564e-05,
+      "loss": 2.3804,
+      "step": 321600
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 2.25296307681798e-05,
+      "loss": 2.5386,
+      "step": 321700
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 2.2521091641903203e-05,
+      "loss": 2.5558,
+      "step": 321800
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 2.2512552515626605e-05,
+      "loss": 2.5719,
+      "step": 321900
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 2.2504013389350003e-05,
+      "loss": 2.509,
+      "step": 322000
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 2.2495474263073404e-05,
+      "loss": 2.4556,
+      "step": 322100
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 2.2486935136796806e-05,
+      "loss": 2.6549,
+      "step": 322200
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 2.2478396010520207e-05,
+      "loss": 2.4002,
+      "step": 322300
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 2.2469856884243605e-05,
+      "loss": 2.4812,
+      "step": 322400
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 2.2461317757967007e-05,
+      "loss": 2.6121,
+      "step": 322500
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 2.245277863169041e-05,
+      "loss": 2.5983,
+      "step": 322600
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 2.2444239505413807e-05,
+      "loss": 2.4197,
+      "step": 322700
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 2.2435700379137208e-05,
+      "loss": 2.733,
+      "step": 322800
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 2.242716125286061e-05,
+      "loss": 2.6397,
+      "step": 322900
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 2.241862212658401e-05,
+      "loss": 2.3667,
+      "step": 323000
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 2.241008300030741e-05,
+      "loss": 2.5528,
+      "step": 323100
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 2.240154387403081e-05,
+      "loss": 2.515,
+      "step": 323200
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 2.2393004747754213e-05,
+      "loss": 2.5222,
+      "step": 323300
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 2.2384465621477614e-05,
+      "loss": 2.5107,
+      "step": 323400
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 2.2375926495201012e-05,
+      "loss": 2.5237,
+      "step": 323500
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 2.2367387368924414e-05,
+      "loss": 2.4484,
+      "step": 323600
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 2.2358848242647815e-05,
+      "loss": 2.5927,
+      "step": 323700
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 2.2350309116371214e-05,
+      "loss": 2.3367,
+      "step": 323800
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 2.2341769990094615e-05,
+      "loss": 2.5768,
+      "step": 323900
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 2.2333230863818017e-05,
+      "loss": 2.6143,
+      "step": 324000
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 2.2324691737541418e-05,
+      "loss": 2.4861,
+      "step": 324100
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 2.2316152611264816e-05,
+      "loss": 2.6254,
+      "step": 324200
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 2.2307613484988218e-05,
+      "loss": 2.6897,
+      "step": 324300
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 2.2299074358711616e-05,
+      "loss": 2.451,
+      "step": 324400
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 2.2290535232435018e-05,
+      "loss": 2.5221,
+      "step": 324500
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 2.228199610615842e-05,
+      "loss": 2.5239,
+      "step": 324600
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 2.2273456979881817e-05,
+      "loss": 2.5147,
+      "step": 324700
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 2.226491785360522e-05,
+      "loss": 2.5661,
+      "step": 324800
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 2.225637872732862e-05,
+      "loss": 2.5493,
+      "step": 324900
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 2.224783960105202e-05,
+      "loss": 2.731,
+      "step": 325000
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 2.223930047477542e-05,
+      "loss": 2.5801,
+      "step": 325100
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 2.2230761348498822e-05,
+      "loss": 2.3902,
+      "step": 325200
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 2.2222222222222223e-05,
+      "loss": 2.6621,
+      "step": 325300
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 2.221368309594562e-05,
+      "loss": 2.7346,
+      "step": 325400
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 2.2205143969669023e-05,
+      "loss": 2.4395,
+      "step": 325500
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 2.2196604843392425e-05,
+      "loss": 2.4414,
+      "step": 325600
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 2.2188065717115826e-05,
+      "loss": 2.4889,
+      "step": 325700
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 2.2179526590839224e-05,
+      "loss": 2.5056,
+      "step": 325800
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 2.2170987464562626e-05,
+      "loss": 2.6284,
+      "step": 325900
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 2.2162448338286027e-05,
+      "loss": 2.5589,
+      "step": 326000
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 2.2153909212009426e-05,
+      "loss": 2.5782,
+      "step": 326100
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 2.2145370085732827e-05,
+      "loss": 2.6423,
+      "step": 326200
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 2.213683095945623e-05,
+      "loss": 2.5087,
+      "step": 326300
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 2.212829183317963e-05,
+      "loss": 2.4429,
+      "step": 326400
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 2.211975270690303e-05,
+      "loss": 2.5103,
+      "step": 326500
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 2.211121358062643e-05,
+      "loss": 2.6208,
+      "step": 326600
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 2.210267445434983e-05,
+      "loss": 2.5342,
+      "step": 326700
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 2.2094135328073233e-05,
+      "loss": 2.3838,
+      "step": 326800
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 2.208559620179663e-05,
+      "loss": 2.4461,
+      "step": 326900
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 2.2077057075520033e-05,
+      "loss": 2.4148,
+      "step": 327000
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 2.2068517949243434e-05,
+      "loss": 2.5646,
+      "step": 327100
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 2.2059978822966836e-05,
+      "loss": 2.4863,
+      "step": 327200
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 2.2051439696690234e-05,
+      "loss": 2.5528,
+      "step": 327300
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 2.2042900570413635e-05,
+      "loss": 2.5117,
+      "step": 327400
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 2.2034361444137037e-05,
+      "loss": 2.5939,
+      "step": 327500
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 2.2025822317860435e-05,
+      "loss": 2.4854,
+      "step": 327600
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 2.2017283191583837e-05,
+      "loss": 2.6956,
+      "step": 327700
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 2.2008744065307238e-05,
+      "loss": 2.7891,
+      "step": 327800
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 2.200020493903064e-05,
+      "loss": 2.6001,
+      "step": 327900
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 2.1991665812754038e-05,
+      "loss": 2.5006,
+      "step": 328000
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 2.198312668647744e-05,
+      "loss": 2.5285,
+      "step": 328100
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 2.197458756020084e-05,
+      "loss": 2.3425,
+      "step": 328200
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 2.1966048433924243e-05,
+      "loss": 2.3846,
+      "step": 328300
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 2.195750930764764e-05,
+      "loss": 2.4363,
+      "step": 328400
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 2.1948970181371042e-05,
+      "loss": 2.4927,
+      "step": 328500
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 2.1940431055094444e-05,
+      "loss": 2.5043,
+      "step": 328600
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 2.1931891928817842e-05,
+      "loss": 2.4685,
+      "step": 328700
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 2.1923352802541244e-05,
+      "loss": 2.5255,
+      "step": 328800
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 2.1914813676264645e-05,
+      "loss": 2.4924,
+      "step": 328900
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 2.1906274549988047e-05,
+      "loss": 2.5157,
+      "step": 329000
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 2.1897735423711445e-05,
+      "loss": 2.6053,
+      "step": 329100
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 2.1889196297434846e-05,
+      "loss": 2.3203,
+      "step": 329200
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 2.1880657171158248e-05,
+      "loss": 2.6072,
+      "step": 329300
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 2.187211804488165e-05,
+      "loss": 2.4769,
+      "step": 329400
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 2.1863578918605048e-05,
+      "loss": 2.7022,
+      "step": 329500
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 2.185503979232845e-05,
+      "loss": 2.6139,
+      "step": 329600
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 2.184650066605185e-05,
+      "loss": 2.6004,
+      "step": 329700
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 2.1837961539775252e-05,
+      "loss": 2.5469,
+      "step": 329800
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 2.182942241349865e-05,
+      "loss": 2.6072,
+      "step": 329900
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 2.1820883287222052e-05,
+      "loss": 2.3999,
+      "step": 330000
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 2.1812344160945454e-05,
+      "loss": 2.6583,
+      "step": 330100
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 2.1803805034668852e-05,
+      "loss": 2.4339,
+      "step": 330200
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 2.1795265908392253e-05,
+      "loss": 2.5644,
+      "step": 330300
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 2.1786726782115655e-05,
+      "loss": 2.4128,
+      "step": 330400
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 2.1778187655839056e-05,
+      "loss": 2.5575,
+      "step": 330500
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 2.1769648529562455e-05,
+      "loss": 2.6073,
+      "step": 330600
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 2.1761109403285856e-05,
+      "loss": 2.5025,
+      "step": 330700
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 2.1752570277009258e-05,
+      "loss": 2.5701,
+      "step": 330800
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 2.174403115073266e-05,
+      "loss": 2.5583,
+      "step": 330900
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 2.1735492024456057e-05,
+      "loss": 2.5079,
+      "step": 331000
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 2.172695289817946e-05,
+      "loss": 2.591,
+      "step": 331100
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 2.171841377190286e-05,
+      "loss": 2.5121,
+      "step": 331200
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 2.1709874645626262e-05,
+      "loss": 2.5391,
+      "step": 331300
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 2.170133551934966e-05,
+      "loss": 2.6695,
+      "step": 331400
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 2.169279639307306e-05,
+      "loss": 2.2796,
+      "step": 331500
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 2.1684257266796463e-05,
+      "loss": 2.6824,
+      "step": 331600
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 2.167571814051986e-05,
+      "loss": 2.6698,
+      "step": 331700
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 2.1667179014243263e-05,
+      "loss": 2.5086,
+      "step": 331800
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 2.1658639887966664e-05,
+      "loss": 2.6209,
+      "step": 331900
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 2.1650100761690066e-05,
+      "loss": 2.5076,
+      "step": 332000
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 2.1641561635413464e-05,
+      "loss": 2.5979,
+      "step": 332100
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 2.1633022509136866e-05,
+      "loss": 2.4665,
+      "step": 332200
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 2.1624483382860267e-05,
+      "loss": 2.5746,
+      "step": 332300
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 2.161594425658367e-05,
+      "loss": 2.3524,
+      "step": 332400
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 2.1607405130307067e-05,
+      "loss": 2.3685,
+      "step": 332500
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 2.159886600403047e-05,
+      "loss": 2.5041,
+      "step": 332600
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 2.159032687775387e-05,
+      "loss": 2.5543,
+      "step": 332700
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 2.1581787751477268e-05,
+      "loss": 2.5392,
+      "step": 332800
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 2.157324862520067e-05,
+      "loss": 2.41,
+      "step": 332900
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 2.156470949892407e-05,
+      "loss": 2.4839,
+      "step": 333000
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 2.1556170372647473e-05,
+      "loss": 2.5286,
+      "step": 333100
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 2.154763124637087e-05,
+      "loss": 2.3548,
+      "step": 333200
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 2.1539092120094273e-05,
+      "loss": 2.6003,
+      "step": 333300
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 2.1530552993817674e-05,
+      "loss": 2.5088,
+      "step": 333400
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 2.1522013867541076e-05,
+      "loss": 2.5957,
+      "step": 333500
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 2.1513474741264474e-05,
+      "loss": 2.589,
+      "step": 333600
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 2.1504935614987875e-05,
+      "loss": 2.4592,
+      "step": 333700
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 2.1496396488711277e-05,
+      "loss": 2.5387,
+      "step": 333800
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 2.148785736243468e-05,
+      "loss": 2.4205,
+      "step": 333900
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 2.1479318236158077e-05,
+      "loss": 2.5185,
+      "step": 334000
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 2.1470779109881478e-05,
+      "loss": 2.437,
+      "step": 334100
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 2.146223998360488e-05,
+      "loss": 2.4597,
+      "step": 334200
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 2.1453700857328278e-05,
+      "loss": 2.419,
+      "step": 334300
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 2.144516173105168e-05,
+      "loss": 2.3632,
+      "step": 334400
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 2.143662260477508e-05,
+      "loss": 2.458,
+      "step": 334500
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 2.1428083478498483e-05,
+      "loss": 2.4865,
+      "step": 334600
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 2.141954435222188e-05,
+      "loss": 2.647,
+      "step": 334700
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 2.1411005225945282e-05,
+      "loss": 2.6055,
+      "step": 334800
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 2.1402466099668684e-05,
+      "loss": 2.4032,
+      "step": 334900
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 2.1393926973392085e-05,
+      "loss": 2.5299,
+      "step": 335000
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 2.1385387847115484e-05,
+      "loss": 2.4994,
+      "step": 335100
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 2.1376848720838885e-05,
+      "loss": 2.5819,
+      "step": 335200
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 2.1368309594562287e-05,
+      "loss": 2.3387,
+      "step": 335300
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 2.1359770468285685e-05,
+      "loss": 2.5313,
+      "step": 335400
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 2.1351231342009086e-05,
+      "loss": 2.5869,
+      "step": 335500
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 2.1342692215732488e-05,
+      "loss": 2.475,
+      "step": 335600
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 2.133415308945589e-05,
+      "loss": 2.5618,
+      "step": 335700
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 2.1325613963179288e-05,
+      "loss": 2.6261,
+      "step": 335800
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 2.131707483690269e-05,
+      "loss": 2.6525,
+      "step": 335900
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 2.130853571062609e-05,
+      "loss": 2.67,
+      "step": 336000
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 2.1299996584349492e-05,
+      "loss": 2.5276,
+      "step": 336100
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 2.129145745807289e-05,
+      "loss": 2.5303,
+      "step": 336200
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 2.1282918331796292e-05,
+      "loss": 2.4886,
+      "step": 336300
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 2.1274379205519693e-05,
+      "loss": 2.4955,
+      "step": 336400
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 2.1265840079243095e-05,
+      "loss": 2.5971,
+      "step": 336500
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 2.1257300952966493e-05,
+      "loss": 2.5558,
+      "step": 336600
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 2.1248761826689895e-05,
+      "loss": 2.5548,
+      "step": 336700
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 2.1240222700413296e-05,
+      "loss": 2.4898,
+      "step": 336800
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 2.1231683574136694e-05,
+      "loss": 2.5943,
+      "step": 336900
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 2.1223144447860096e-05,
+      "loss": 2.5476,
+      "step": 337000
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 2.1214605321583498e-05,
+      "loss": 2.3956,
+      "step": 337100
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 2.12060661953069e-05,
+      "loss": 2.5946,
+      "step": 337200
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 2.1197527069030297e-05,
+      "loss": 2.5176,
+      "step": 337300
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 2.11889879427537e-05,
+      "loss": 2.5441,
+      "step": 337400
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 2.11804488164771e-05,
+      "loss": 2.4069,
+      "step": 337500
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 2.1171909690200502e-05,
+      "loss": 2.4024,
+      "step": 337600
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 2.11633705639239e-05,
+      "loss": 2.5113,
+      "step": 337700
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 2.11548314376473e-05,
+      "loss": 2.5701,
+      "step": 337800
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 2.1146292311370703e-05,
+      "loss": 2.4322,
+      "step": 337900
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 2.11377531850941e-05,
+      "loss": 2.4589,
+      "step": 338000
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 2.1129214058817503e-05,
+      "loss": 2.6578,
+      "step": 338100
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 2.1120674932540904e-05,
+      "loss": 2.4323,
+      "step": 338200
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 2.1112135806264306e-05,
+      "loss": 2.4691,
+      "step": 338300
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 2.1103596679987704e-05,
+      "loss": 2.5593,
+      "step": 338400
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 2.1095057553711106e-05,
+      "loss": 2.6265,
+      "step": 338500
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 2.1086518427434507e-05,
+      "loss": 2.6088,
+      "step": 338600
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 2.107797930115791e-05,
+      "loss": 2.1967,
+      "step": 338700
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 2.1069440174881307e-05,
+      "loss": 2.5874,
+      "step": 338800
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 2.106090104860471e-05,
+      "loss": 2.3878,
+      "step": 338900
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 2.105236192232811e-05,
+      "loss": 2.4317,
+      "step": 339000
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 2.104382279605151e-05,
+      "loss": 2.4738,
+      "step": 339100
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 2.103528366977491e-05,
+      "loss": 2.5272,
+      "step": 339200
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 2.102674454349831e-05,
+      "loss": 2.5015,
+      "step": 339300
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 2.1018205417221713e-05,
+      "loss": 2.521,
+      "step": 339400
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 2.100966629094511e-05,
+      "loss": 2.5521,
+      "step": 339500
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 2.1001127164668513e-05,
+      "loss": 2.58,
+      "step": 339600
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 2.0992588038391914e-05,
+      "loss": 2.5704,
+      "step": 339700
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 2.0984048912115316e-05,
+      "loss": 2.588,
+      "step": 339800
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 2.0975509785838714e-05,
+      "loss": 2.6017,
+      "step": 339900
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 2.0966970659562115e-05,
+      "loss": 2.5664,
+      "step": 340000
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 2.0958431533285517e-05,
+      "loss": 2.493,
+      "step": 340100
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 2.094989240700892e-05,
+      "loss": 2.6741,
+      "step": 340200
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 2.0941353280732317e-05,
+      "loss": 2.2934,
+      "step": 340300
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 2.0932814154455718e-05,
+      "loss": 2.4713,
+      "step": 340400
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 2.092427502817912e-05,
+      "loss": 2.6073,
+      "step": 340500
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 2.0915735901902518e-05,
+      "loss": 2.5486,
+      "step": 340600
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 2.090719677562592e-05,
+      "loss": 2.5145,
+      "step": 340700
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 2.089865764934932e-05,
+      "loss": 2.592,
+      "step": 340800
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 2.0890118523072723e-05,
+      "loss": 2.7031,
+      "step": 340900
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 2.088157939679612e-05,
+      "loss": 2.3675,
+      "step": 341000
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 2.0873040270519522e-05,
+      "loss": 2.6031,
+      "step": 341100
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 2.0864501144242924e-05,
+      "loss": 2.4581,
+      "step": 341200
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 2.0855962017966325e-05,
+      "loss": 2.6458,
+      "step": 341300
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 2.0847422891689723e-05,
+      "loss": 2.6926,
+      "step": 341400
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 2.0838883765413125e-05,
+      "loss": 2.5385,
+      "step": 341500
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 2.0830344639136527e-05,
+      "loss": 2.5288,
+      "step": 341600
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 2.0821805512859928e-05,
+      "loss": 2.4888,
+      "step": 341700
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 2.0813266386583326e-05,
+      "loss": 2.5076,
+      "step": 341800
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 2.0804727260306724e-05,
+      "loss": 2.3994,
+      "step": 341900
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 2.0796188134030126e-05,
+      "loss": 2.6172,
+      "step": 342000
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 2.0787649007753528e-05,
+      "loss": 2.5673,
+      "step": 342100
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 2.0779109881476926e-05,
+      "loss": 2.519,
+      "step": 342200
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 2.0770570755200327e-05,
+      "loss": 2.6818,
+      "step": 342300
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 2.076203162892373e-05,
+      "loss": 2.4862,
+      "step": 342400
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 2.075349250264713e-05,
+      "loss": 2.4133,
+      "step": 342500
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 2.074495337637053e-05,
+      "loss": 2.6491,
+      "step": 342600
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 2.073641425009393e-05,
+      "loss": 2.3798,
+      "step": 342700
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 2.072787512381733e-05,
+      "loss": 2.4769,
+      "step": 342800
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 2.071933599754073e-05,
+      "loss": 2.5758,
+      "step": 342900
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 2.071079687126413e-05,
+      "loss": 2.5529,
+      "step": 343000
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 2.0702257744987533e-05,
+      "loss": 2.288,
+      "step": 343100
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 2.0693718618710934e-05,
+      "loss": 2.5995,
+      "step": 343200
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 2.0685179492434333e-05,
+      "loss": 2.4792,
+      "step": 343300
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 2.0676640366157734e-05,
+      "loss": 2.4033,
+      "step": 343400
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 2.0668101239881136e-05,
+      "loss": 2.5598,
+      "step": 343500
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 2.0659562113604537e-05,
+      "loss": 2.5351,
+      "step": 343600
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 2.0651022987327935e-05,
+      "loss": 2.3781,
+      "step": 343700
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 2.0642483861051337e-05,
+      "loss": 2.2974,
+      "step": 343800
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 2.063394473477474e-05,
+      "loss": 2.5745,
+      "step": 343900
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 2.062540560849814e-05,
+      "loss": 2.4677,
+      "step": 344000
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 2.0616866482221538e-05,
+      "loss": 2.4528,
+      "step": 344100
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 2.060832735594494e-05,
+      "loss": 2.524,
+      "step": 344200
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 2.059978822966834e-05,
+      "loss": 2.5958,
+      "step": 344300
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 2.059124910339174e-05,
+      "loss": 2.6228,
+      "step": 344400
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 2.058270997711514e-05,
+      "loss": 2.5172,
+      "step": 344500
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 2.0574170850838543e-05,
+      "loss": 2.5492,
+      "step": 344600
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 2.0565631724561944e-05,
+      "loss": 2.5259,
+      "step": 344700
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 2.0557092598285342e-05,
+      "loss": 2.494,
+      "step": 344800
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 2.0548553472008744e-05,
+      "loss": 2.601,
+      "step": 344900
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 2.0540014345732145e-05,
+      "loss": 2.4916,
+      "step": 345000
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 2.0531475219455547e-05,
+      "loss": 2.6532,
+      "step": 345100
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 2.0522936093178945e-05,
+      "loss": 2.574,
+      "step": 345200
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 2.0514396966902347e-05,
+      "loss": 2.483,
+      "step": 345300
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 2.0505857840625748e-05,
+      "loss": 2.4472,
+      "step": 345400
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 2.0497318714349146e-05,
+      "loss": 2.5283,
+      "step": 345500
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 2.0488779588072548e-05,
+      "loss": 2.4242,
+      "step": 345600
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 2.048024046179595e-05,
+      "loss": 2.311,
+      "step": 345700
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 2.047170133551935e-05,
+      "loss": 2.3901,
+      "step": 345800
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 2.046316220924275e-05,
+      "loss": 2.4731,
+      "step": 345900
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 2.045462308296615e-05,
+      "loss": 2.423,
+      "step": 346000
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 2.0446083956689552e-05,
+      "loss": 2.5762,
+      "step": 346100
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 2.0437544830412954e-05,
+      "loss": 2.5543,
+      "step": 346200
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 2.0429005704136352e-05,
+      "loss": 2.5611,
+      "step": 346300
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 2.0420466577859753e-05,
+      "loss": 2.5478,
+      "step": 346400
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 2.0411927451583155e-05,
+      "loss": 2.5788,
+      "step": 346500
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 2.0403388325306557e-05,
+      "loss": 2.6245,
+      "step": 346600
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 2.0394849199029955e-05,
+      "loss": 2.4401,
+      "step": 346700
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 2.0386310072753356e-05,
+      "loss": 2.3803,
+      "step": 346800
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 2.0377770946476758e-05,
+      "loss": 2.4374,
+      "step": 346900
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 2.0369231820200156e-05,
+      "loss": 2.7678,
+      "step": 347000
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 2.0360692693923558e-05,
+      "loss": 2.513,
+      "step": 347100
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 2.035215356764696e-05,
+      "loss": 2.646,
+      "step": 347200
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 2.034361444137036e-05,
+      "loss": 2.4675,
+      "step": 347300
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 2.033507531509376e-05,
+      "loss": 2.5015,
+      "step": 347400
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 2.032653618881716e-05,
+      "loss": 2.3507,
+      "step": 347500
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 2.0317997062540562e-05,
+      "loss": 2.4468,
+      "step": 347600
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 2.0309457936263963e-05,
+      "loss": 2.4786,
+      "step": 347700
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 2.030091880998736e-05,
+      "loss": 2.5348,
+      "step": 347800
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 2.0292379683710763e-05,
+      "loss": 2.6023,
+      "step": 347900
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 2.0283840557434165e-05,
+      "loss": 2.3332,
+      "step": 348000
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 2.0275301431157566e-05,
+      "loss": 2.6247,
+      "step": 348100
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 2.0266762304880964e-05,
+      "loss": 2.5805,
+      "step": 348200
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 2.0258223178604366e-05,
+      "loss": 2.4321,
+      "step": 348300
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 2.0249684052327767e-05,
+      "loss": 2.3058,
+      "step": 348400
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 2.0241144926051166e-05,
+      "loss": 2.2775,
+      "step": 348500
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 2.0232605799774567e-05,
+      "loss": 2.6537,
+      "step": 348600
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 2.022406667349797e-05,
+      "loss": 2.6038,
+      "step": 348700
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 2.021552754722137e-05,
+      "loss": 2.4554,
+      "step": 348800
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 2.020698842094477e-05,
+      "loss": 2.4168,
+      "step": 348900
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 2.019844929466817e-05,
+      "loss": 2.4945,
+      "step": 349000
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 2.018991016839157e-05,
+      "loss": 2.4791,
+      "step": 349100
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 2.0181371042114973e-05,
+      "loss": 2.3889,
+      "step": 349200
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 2.017283191583837e-05,
+      "loss": 2.4123,
+      "step": 349300
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 2.0164292789561773e-05,
+      "loss": 2.4595,
+      "step": 349400
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 2.0155753663285174e-05,
+      "loss": 2.3611,
+      "step": 349500
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 2.0147214537008572e-05,
+      "loss": 2.5504,
+      "step": 349600
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 2.0138675410731974e-05,
+      "loss": 2.4024,
+      "step": 349700
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 2.0130136284455376e-05,
+      "loss": 2.6099,
+      "step": 349800
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 2.0121597158178777e-05,
+      "loss": 2.5468,
+      "step": 349900
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 2.0113058031902175e-05,
+      "loss": 2.3558,
+      "step": 350000
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 2.0104518905625577e-05,
+      "loss": 2.5213,
+      "step": 350100
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 2.009597977934898e-05,
+      "loss": 2.3691,
+      "step": 350200
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 2.008744065307238e-05,
+      "loss": 2.6007,
+      "step": 350300
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 2.0078901526795778e-05,
+      "loss": 2.5096,
+      "step": 350400
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 2.007036240051918e-05,
+      "loss": 2.5878,
+      "step": 350500
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 2.006182327424258e-05,
+      "loss": 2.6019,
+      "step": 350600
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 2.0053284147965983e-05,
+      "loss": 2.5281,
+      "step": 350700
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 2.004474502168938e-05,
+      "loss": 2.3759,
+      "step": 350800
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 2.0036205895412782e-05,
+      "loss": 2.4643,
+      "step": 350900
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 2.0027666769136184e-05,
+      "loss": 2.2171,
+      "step": 351000
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 2.0019127642859582e-05,
+      "loss": 2.5775,
+      "step": 351100
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 2.0010588516582984e-05,
+      "loss": 2.36,
+      "step": 351200
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 2.0002049390306385e-05,
+      "loss": 2.3243,
+      "step": 351300
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.9993510264029787e-05,
+      "loss": 2.4882,
+      "step": 351400
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.9984971137753185e-05,
+      "loss": 2.5443,
+      "step": 351500
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.9976432011476587e-05,
+      "loss": 2.486,
+      "step": 351600
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.9967892885199988e-05,
+      "loss": 2.3474,
+      "step": 351700
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.995935375892339e-05,
+      "loss": 2.3448,
+      "step": 351800
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.9950814632646788e-05,
+      "loss": 2.4835,
+      "step": 351900
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.994227550637019e-05,
+      "loss": 2.3045,
+      "step": 352000
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.993373638009359e-05,
+      "loss": 2.5893,
+      "step": 352100
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.992519725381699e-05,
+      "loss": 2.5106,
+      "step": 352200
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.991665812754039e-05,
+      "loss": 2.4664,
+      "step": 352300
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.9908119001263792e-05,
+      "loss": 2.5472,
+      "step": 352400
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.9899579874987194e-05,
+      "loss": 2.6176,
+      "step": 352500
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.9891040748710592e-05,
+      "loss": 2.5066,
+      "step": 352600
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.9882501622433993e-05,
+      "loss": 2.4152,
+      "step": 352700
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.9873962496157395e-05,
+      "loss": 2.2691,
+      "step": 352800
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.9865423369880796e-05,
+      "loss": 2.4565,
+      "step": 352900
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.9856884243604195e-05,
+      "loss": 2.4942,
+      "step": 353000
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.9848345117327596e-05,
+      "loss": 2.5106,
+      "step": 353100
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.9839805991050998e-05,
+      "loss": 2.4466,
+      "step": 353200
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.98312668647744e-05,
+      "loss": 2.5558,
+      "step": 353300
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.9822727738497797e-05,
+      "loss": 2.7191,
+      "step": 353400
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.98141886122212e-05,
+      "loss": 2.5976,
+      "step": 353500
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.98056494859446e-05,
+      "loss": 2.425,
+      "step": 353600
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.9797110359668e-05,
+      "loss": 2.3932,
+      "step": 353700
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.97885712333914e-05,
+      "loss": 2.4325,
+      "step": 353800
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.9780032107114802e-05,
+      "loss": 2.3956,
+      "step": 353900
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.9771492980838203e-05,
+      "loss": 2.5554,
+      "step": 354000
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.97629538545616e-05,
+      "loss": 2.3288,
+      "step": 354100
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.9754414728285003e-05,
+      "loss": 2.5212,
+      "step": 354200
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 1.9745875602008405e-05,
+      "loss": 2.4385,
+      "step": 354300
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 1.9737336475731806e-05,
+      "loss": 2.5364,
+      "step": 354400
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 1.9728797349455204e-05,
+      "loss": 2.5373,
+      "step": 354500
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 1.9720258223178606e-05,
+      "loss": 2.3519,
+      "step": 354600
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 1.9711719096902007e-05,
+      "loss": 2.6551,
+      "step": 354700
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 1.9703179970625406e-05,
+      "loss": 2.4575,
+      "step": 354800
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 1.9694640844348807e-05,
+      "loss": 2.3641,
+      "step": 354900
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 1.968610171807221e-05,
+      "loss": 2.5151,
+      "step": 355000
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 1.967756259179561e-05,
+      "loss": 2.4773,
+      "step": 355100
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 1.966902346551901e-05,
+      "loss": 2.3488,
+      "step": 355200
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 1.966048433924241e-05,
+      "loss": 2.5454,
+      "step": 355300
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 1.965194521296581e-05,
+      "loss": 2.38,
+      "step": 355400
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 1.9643406086689213e-05,
+      "loss": 2.3771,
+      "step": 355500
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 1.963486696041261e-05,
+      "loss": 2.5326,
+      "step": 355600
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 1.9626327834136013e-05,
+      "loss": 2.4057,
+      "step": 355700
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 1.9617788707859414e-05,
+      "loss": 2.2275,
+      "step": 355800
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 1.9609249581582816e-05,
+      "loss": 2.4473,
+      "step": 355900
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 1.9600710455306214e-05,
+      "loss": 2.3546,
+      "step": 356000
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 1.9592171329029616e-05,
+      "loss": 2.3716,
+      "step": 356100
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 1.9583632202753017e-05,
+      "loss": 2.4092,
+      "step": 356200
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 1.9575093076476415e-05,
+      "loss": 2.4315,
+      "step": 356300
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 1.9566553950199817e-05,
+      "loss": 2.454,
+      "step": 356400
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 1.955801482392322e-05,
+      "loss": 2.5591,
+      "step": 356500
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 1.954947569764662e-05,
+      "loss": 2.4718,
+      "step": 356600
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 1.9540936571370018e-05,
+      "loss": 2.2909,
+      "step": 356700
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 1.953239744509342e-05,
+      "loss": 2.6231,
+      "step": 356800
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 1.952385831881682e-05,
+      "loss": 2.3351,
+      "step": 356900
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 1.9515319192540223e-05,
+      "loss": 2.5459,
+      "step": 357000
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 1.950678006626362e-05,
+      "loss": 2.5014,
+      "step": 357100
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 1.9498240939987022e-05,
+      "loss": 2.5505,
+      "step": 357200
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 1.9489701813710424e-05,
+      "loss": 2.3888,
+      "step": 357300
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 1.9481162687433822e-05,
+      "loss": 2.5272,
+      "step": 357400
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 1.9472623561157224e-05,
+      "loss": 2.4352,
+      "step": 357500
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 1.9464084434880625e-05,
+      "loss": 2.4541,
+      "step": 357600
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 1.9455545308604027e-05,
+      "loss": 2.2915,
+      "step": 357700
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 1.9447006182327425e-05,
+      "loss": 2.5464,
+      "step": 357800
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 1.9438467056050826e-05,
+      "loss": 2.3428,
+      "step": 357900
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 1.9429927929774228e-05,
+      "loss": 2.3803,
+      "step": 358000
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 1.942138880349763e-05,
+      "loss": 2.4407,
+      "step": 358100
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 1.9412849677221028e-05,
+      "loss": 2.4178,
+      "step": 358200
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 1.940431055094443e-05,
+      "loss": 2.5146,
+      "step": 358300
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 1.939577142466783e-05,
+      "loss": 2.4201,
+      "step": 358400
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 1.9387232298391232e-05,
+      "loss": 2.5639,
+      "step": 358500
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 1.937869317211463e-05,
+      "loss": 2.4283,
+      "step": 358600
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 1.9370154045838032e-05,
+      "loss": 2.4783,
+      "step": 358700
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 1.9361614919561434e-05,
+      "loss": 2.5163,
+      "step": 358800
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 1.9353075793284832e-05,
+      "loss": 2.3794,
+      "step": 358900
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 1.9344536667008233e-05,
+      "loss": 2.6358,
+      "step": 359000
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 1.9335997540731635e-05,
+      "loss": 2.5169,
+      "step": 359100
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 1.9327458414455036e-05,
+      "loss": 2.3942,
+      "step": 359200
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 1.9318919288178435e-05,
+      "loss": 2.5266,
+      "step": 359300
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 1.9310380161901833e-05,
+      "loss": 2.4013,
+      "step": 359400
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 1.9301841035625234e-05,
+      "loss": 2.4278,
+      "step": 359500
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 1.9293301909348636e-05,
+      "loss": 2.3956,
+      "step": 359600
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 1.9284762783072034e-05,
+      "loss": 2.4784,
+      "step": 359700
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 1.9276223656795436e-05,
+      "loss": 2.4385,
+      "step": 359800
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 1.9267684530518837e-05,
+      "loss": 2.4665,
+      "step": 359900
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 1.925914540424224e-05,
+      "loss": 2.369,
+      "step": 360000
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 1.9250606277965637e-05,
+      "loss": 2.4647,
+      "step": 360100
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 1.924206715168904e-05,
+      "loss": 2.4181,
+      "step": 360200
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 1.923352802541244e-05,
+      "loss": 2.3927,
+      "step": 360300
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 1.922498889913584e-05,
+      "loss": 2.4444,
+      "step": 360400
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 1.921644977285924e-05,
+      "loss": 2.6982,
+      "step": 360500
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 1.920791064658264e-05,
+      "loss": 2.6013,
+      "step": 360600
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 1.9199371520306043e-05,
+      "loss": 2.4856,
+      "step": 360700
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 1.9190832394029444e-05,
+      "loss": 2.5226,
+      "step": 360800
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 1.9182293267752842e-05,
+      "loss": 2.4428,
+      "step": 360900
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 1.9173754141476244e-05,
+      "loss": 2.5025,
+      "step": 361000
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 1.9165215015199646e-05,
+      "loss": 2.4364,
+      "step": 361100
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 1.9156675888923044e-05,
+      "loss": 2.3382,
+      "step": 361200
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 1.9148136762646445e-05,
+      "loss": 2.3355,
+      "step": 361300
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 1.9139597636369847e-05,
+      "loss": 2.3755,
+      "step": 361400
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 1.913105851009325e-05,
+      "loss": 2.5798,
+      "step": 361500
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 1.9122519383816646e-05,
+      "loss": 2.5586,
+      "step": 361600
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 1.9113980257540048e-05,
+      "loss": 2.5779,
+      "step": 361700
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 1.910544113126345e-05,
+      "loss": 2.4358,
+      "step": 361800
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 1.909690200498685e-05,
+      "loss": 2.4411,
+      "step": 361900
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 1.908836287871025e-05,
+      "loss": 2.556,
+      "step": 362000
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 1.907982375243365e-05,
+      "loss": 2.4819,
+      "step": 362100
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 1.9071284626157052e-05,
+      "loss": 2.3784,
+      "step": 362200
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 1.906274549988045e-05,
+      "loss": 2.2813,
+      "step": 362300
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 1.9054206373603852e-05,
+      "loss": 2.3302,
+      "step": 362400
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 1.9045667247327254e-05,
+      "loss": 2.5037,
+      "step": 362500
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 1.9037128121050655e-05,
+      "loss": 2.5283,
+      "step": 362600
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 1.9028588994774053e-05,
+      "loss": 2.3458,
+      "step": 362700
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 1.9020049868497455e-05,
+      "loss": 2.4362,
+      "step": 362800
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 1.9011510742220856e-05,
+      "loss": 2.5364,
+      "step": 362900
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 1.9002971615944258e-05,
+      "loss": 2.3834,
+      "step": 363000
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 1.8994432489667656e-05,
+      "loss": 2.4197,
+      "step": 363100
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 1.8985893363391058e-05,
+      "loss": 2.4561,
+      "step": 363200
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 1.897735423711446e-05,
+      "loss": 2.4097,
+      "step": 363300
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 1.896881511083786e-05,
+      "loss": 2.4191,
+      "step": 363400
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 1.896027598456126e-05,
+      "loss": 2.3627,
+      "step": 363500
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 1.895173685828466e-05,
+      "loss": 2.5534,
+      "step": 363600
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 1.8943197732008062e-05,
+      "loss": 2.3401,
+      "step": 363700
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 1.893465860573146e-05,
+      "loss": 2.5124,
+      "step": 363800
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 1.8926119479454862e-05,
+      "loss": 2.5377,
+      "step": 363900
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 1.8917580353178263e-05,
+      "loss": 2.4968,
+      "step": 364000
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 1.8909041226901665e-05,
+      "loss": 2.5767,
+      "step": 364100
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 1.8900502100625063e-05,
+      "loss": 2.46,
+      "step": 364200
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 1.8891962974348465e-05,
+      "loss": 2.2387,
+      "step": 364300
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 1.8883423848071866e-05,
+      "loss": 2.4068,
+      "step": 364400
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.8874884721795268e-05,
+      "loss": 2.5661,
+      "step": 364500
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.8866345595518666e-05,
+      "loss": 2.5201,
+      "step": 364600
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.8857806469242067e-05,
+      "loss": 2.4884,
+      "step": 364700
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.884926734296547e-05,
+      "loss": 2.6423,
+      "step": 364800
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.884072821668887e-05,
+      "loss": 2.5374,
+      "step": 364900
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.883218909041227e-05,
+      "loss": 2.3762,
+      "step": 365000
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.882364996413567e-05,
+      "loss": 2.4098,
+      "step": 365100
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.8815110837859072e-05,
+      "loss": 2.5898,
+      "step": 365200
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.880657171158247e-05,
+      "loss": 2.3937,
+      "step": 365300
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.879803258530587e-05,
+      "loss": 2.5132,
+      "step": 365400
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.8789493459029273e-05,
+      "loss": 2.3442,
+      "step": 365500
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.8780954332752675e-05,
+      "loss": 2.4829,
+      "step": 365600
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.8772415206476073e-05,
+      "loss": 2.5117,
+      "step": 365700
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.8763876080199474e-05,
+      "loss": 2.4341,
+      "step": 365800
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.8755336953922876e-05,
+      "loss": 2.66,
+      "step": 365900
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.8746797827646277e-05,
+      "loss": 2.3417,
+      "step": 366000
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.8738258701369675e-05,
+      "loss": 2.4579,
+      "step": 366100
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.8729719575093077e-05,
+      "loss": 2.5,
+      "step": 366200
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.872118044881648e-05,
+      "loss": 2.3192,
+      "step": 366300
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.8712641322539877e-05,
+      "loss": 2.5531,
+      "step": 366400
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.8704102196263278e-05,
+      "loss": 2.5078,
+      "step": 366500
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.869556306998668e-05,
+      "loss": 2.3509,
+      "step": 366600
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.868702394371008e-05,
+      "loss": 2.4027,
+      "step": 366700
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.867848481743348e-05,
+      "loss": 2.2721,
+      "step": 366800
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.866994569115688e-05,
+      "loss": 2.3922,
+      "step": 366900
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.8661406564880283e-05,
+      "loss": 2.4485,
+      "step": 367000
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.8652867438603684e-05,
+      "loss": 2.5269,
+      "step": 367100
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.8644328312327082e-05,
+      "loss": 2.4858,
+      "step": 367200
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.8635789186050484e-05,
+      "loss": 2.5382,
+      "step": 367300
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.8627250059773885e-05,
+      "loss": 2.3414,
+      "step": 367400
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.8618710933497287e-05,
+      "loss": 2.4237,
+      "step": 367500
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.8610171807220685e-05,
+      "loss": 2.6329,
+      "step": 367600
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.8601632680944087e-05,
+      "loss": 2.3772,
+      "step": 367700
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.8593093554667488e-05,
+      "loss": 2.3128,
+      "step": 367800
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.8584554428390886e-05,
+      "loss": 2.4733,
+      "step": 367900
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.8576015302114288e-05,
+      "loss": 2.4864,
+      "step": 368000
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.856747617583769e-05,
+      "loss": 2.6485,
+      "step": 368100
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.855893704956109e-05,
+      "loss": 2.4724,
+      "step": 368200
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.855039792328449e-05,
+      "loss": 2.4789,
+      "step": 368300
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.854185879700789e-05,
+      "loss": 2.5142,
+      "step": 368400
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.8533319670731292e-05,
+      "loss": 2.5463,
+      "step": 368500
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.8524780544454694e-05,
+      "loss": 2.3217,
+      "step": 368600
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.8516241418178092e-05,
+      "loss": 2.384,
+      "step": 368700
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.8507702291901494e-05,
+      "loss": 2.4266,
+      "step": 368800
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.8499163165624895e-05,
+      "loss": 2.4248,
+      "step": 368900
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.8490624039348293e-05,
+      "loss": 2.5011,
+      "step": 369000
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.8482084913071695e-05,
+      "loss": 2.3596,
+      "step": 369100
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.8473545786795096e-05,
+      "loss": 2.5138,
+      "step": 369200
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.8465006660518498e-05,
+      "loss": 2.5326,
+      "step": 369300
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.8456467534241896e-05,
+      "loss": 2.7676,
+      "step": 369400
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.8447928407965298e-05,
+      "loss": 2.5415,
+      "step": 369500
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.84393892816887e-05,
+      "loss": 2.4893,
+      "step": 369600
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.84308501554121e-05,
+      "loss": 2.4449,
+      "step": 369700
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.84223110291355e-05,
+      "loss": 2.438,
+      "step": 369800
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.84137719028589e-05,
+      "loss": 2.5547,
+      "step": 369900
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.8405232776582302e-05,
+      "loss": 2.5122,
+      "step": 370000
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.8396693650305704e-05,
+      "loss": 2.403,
+      "step": 370100
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.83881545240291e-05,
+      "loss": 2.5822,
+      "step": 370200
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.8379615397752503e-05,
+      "loss": 2.2563,
+      "step": 370300
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.8371076271475905e-05,
+      "loss": 2.4158,
+      "step": 370400
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.8362537145199303e-05,
+      "loss": 2.4103,
+      "step": 370500
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.8353998018922704e-05,
+      "loss": 2.3934,
+      "step": 370600
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.8345458892646106e-05,
+      "loss": 2.6031,
+      "step": 370700
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.8336919766369508e-05,
+      "loss": 2.5486,
+      "step": 370800
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.8328380640092906e-05,
+      "loss": 2.3811,
+      "step": 370900
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.8319841513816307e-05,
+      "loss": 2.5578,
+      "step": 371000
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.831130238753971e-05,
+      "loss": 2.3826,
+      "step": 371100
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.830276326126311e-05,
+      "loss": 2.5775,
+      "step": 371200
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.829422413498651e-05,
+      "loss": 2.6319,
+      "step": 371300
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.828568500870991e-05,
+      "loss": 2.2904,
+      "step": 371400
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.827714588243331e-05,
+      "loss": 2.4549,
+      "step": 371500
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.826860675615671e-05,
+      "loss": 2.612,
+      "step": 371600
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.826006762988011e-05,
+      "loss": 2.444,
+      "step": 371700
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.8251528503603513e-05,
+      "loss": 2.4744,
+      "step": 371800
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.8242989377326914e-05,
+      "loss": 2.5093,
+      "step": 371900
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.8234450251050313e-05,
+      "loss": 2.3673,
+      "step": 372000
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.8225911124773714e-05,
+      "loss": 2.3785,
+      "step": 372100
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.8217371998497116e-05,
+      "loss": 2.5666,
+      "step": 372200
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.8208832872220517e-05,
+      "loss": 2.251,
+      "step": 372300
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.8200293745943915e-05,
+      "loss": 2.5145,
+      "step": 372400
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.8191754619667317e-05,
+      "loss": 2.4607,
+      "step": 372500
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.818321549339072e-05,
+      "loss": 2.5412,
+      "step": 372600
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.817467636711412e-05,
+      "loss": 2.4565,
+      "step": 372700
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.8166137240837518e-05,
+      "loss": 2.4761,
+      "step": 372800
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.815759811456092e-05,
+      "loss": 2.4169,
+      "step": 372900
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.814905898828432e-05,
+      "loss": 2.298,
+      "step": 373000
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.814051986200772e-05,
+      "loss": 2.4365,
+      "step": 373100
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.813198073573112e-05,
+      "loss": 2.5362,
+      "step": 373200
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.8123441609454523e-05,
+      "loss": 2.4336,
+      "step": 373300
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.8114902483177924e-05,
+      "loss": 2.2315,
+      "step": 373400
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.8106363356901322e-05,
+      "loss": 2.304,
+      "step": 373500
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.8097824230624724e-05,
+      "loss": 2.444,
+      "step": 373600
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.8089285104348125e-05,
+      "loss": 2.4737,
+      "step": 373700
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.8080745978071527e-05,
+      "loss": 2.4807,
+      "step": 373800
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.8072206851794925e-05,
+      "loss": 2.4411,
+      "step": 373900
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.8063667725518327e-05,
+      "loss": 2.4845,
+      "step": 374000
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.8055128599241728e-05,
+      "loss": 2.5516,
+      "step": 374100
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.8046589472965126e-05,
+      "loss": 2.4418,
+      "step": 374200
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.8038050346688528e-05,
+      "loss": 2.4746,
+      "step": 374300
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.802951122041193e-05,
+      "loss": 2.2424,
+      "step": 374400
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.802097209413533e-05,
+      "loss": 2.537,
+      "step": 374500
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.801243296785873e-05,
+      "loss": 2.4697,
+      "step": 374600
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.800389384158213e-05,
+      "loss": 2.4984,
+      "step": 374700
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.7995354715305532e-05,
+      "loss": 2.6161,
+      "step": 374800
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.7986815589028934e-05,
+      "loss": 2.4842,
+      "step": 374900
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.7978276462752332e-05,
+      "loss": 2.2586,
+      "step": 375000
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.7969737336475734e-05,
+      "loss": 2.5688,
+      "step": 375100
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.7961198210199135e-05,
+      "loss": 2.3439,
+      "step": 375200
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.7952659083922537e-05,
+      "loss": 2.4299,
+      "step": 375300
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.7944119957645935e-05,
+      "loss": 2.6774,
+      "step": 375400
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.7935580831369336e-05,
+      "loss": 2.4918,
+      "step": 375500
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.7927041705092738e-05,
+      "loss": 2.3825,
+      "step": 375600
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.7918502578816136e-05,
+      "loss": 2.4744,
+      "step": 375700
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.7909963452539538e-05,
+      "loss": 2.4258,
+      "step": 375800
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.790142432626294e-05,
+      "loss": 2.5512,
+      "step": 375900
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.789288519998634e-05,
+      "loss": 2.3063,
+      "step": 376000
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.788434607370974e-05,
+      "loss": 2.4228,
+      "step": 376100
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.787580694743314e-05,
+      "loss": 2.409,
+      "step": 376200
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.7867267821156542e-05,
+      "loss": 2.4866,
+      "step": 376300
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.7858728694879943e-05,
+      "loss": 2.4487,
+      "step": 376400
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.785018956860334e-05,
+      "loss": 2.3571,
+      "step": 376500
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.7841650442326743e-05,
+      "loss": 2.6049,
+      "step": 376600
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.7833111316050145e-05,
+      "loss": 2.4646,
+      "step": 376700
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.7824572189773546e-05,
+      "loss": 2.3863,
+      "step": 376800
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.7816033063496944e-05,
+      "loss": 2.4522,
+      "step": 376900
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.7807493937220343e-05,
+      "loss": 2.4233,
+      "step": 377000
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.7798954810943744e-05,
+      "loss": 2.5585,
+      "step": 377100
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.7790415684667146e-05,
+      "loss": 2.6039,
+      "step": 377200
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.7781876558390544e-05,
+      "loss": 2.502,
+      "step": 377300
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.7773337432113945e-05,
+      "loss": 2.5336,
+      "step": 377400
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.7764798305837347e-05,
+      "loss": 2.4525,
+      "step": 377500
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.775625917956075e-05,
+      "loss": 2.5984,
+      "step": 377600
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.7747720053284147e-05,
+      "loss": 2.4445,
+      "step": 377700
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.7739180927007548e-05,
+      "loss": 2.2989,
+      "step": 377800
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.773064180073095e-05,
+      "loss": 2.4097,
+      "step": 377900
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.7722102674454348e-05,
+      "loss": 2.4614,
+      "step": 378000
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.771356354817775e-05,
+      "loss": 2.4558,
+      "step": 378100
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.770502442190115e-05,
+      "loss": 2.4049,
+      "step": 378200
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.7696485295624553e-05,
+      "loss": 2.5365,
+      "step": 378300
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.768794616934795e-05,
+      "loss": 2.5574,
+      "step": 378400
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.7679407043071352e-05,
+      "loss": 2.5316,
+      "step": 378500
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.7670867916794754e-05,
+      "loss": 2.5792,
+      "step": 378600
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.7662328790518155e-05,
+      "loss": 2.372,
+      "step": 378700
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.7653789664241554e-05,
+      "loss": 2.5235,
+      "step": 378800
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.7645250537964955e-05,
+      "loss": 2.5767,
+      "step": 378900
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.7636711411688357e-05,
+      "loss": 2.4736,
+      "step": 379000
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.7628172285411755e-05,
+      "loss": 2.4114,
+      "step": 379100
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.7619633159135156e-05,
+      "loss": 2.3932,
+      "step": 379200
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.7611094032858558e-05,
+      "loss": 2.5267,
+      "step": 379300
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.760255490658196e-05,
+      "loss": 2.5374,
+      "step": 379400
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.7594015780305358e-05,
+      "loss": 2.4665,
+      "step": 379500
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.758547665402876e-05,
+      "loss": 2.491,
+      "step": 379600
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.757693752775216e-05,
+      "loss": 2.5024,
+      "step": 379700
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.7568398401475562e-05,
+      "loss": 2.5219,
+      "step": 379800
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.755985927519896e-05,
+      "loss": 2.4444,
+      "step": 379900
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.7551320148922362e-05,
+      "loss": 2.4486,
+      "step": 380000
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.7542781022645763e-05,
+      "loss": 2.4555,
+      "step": 380100
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.7534241896369165e-05,
+      "loss": 2.4537,
+      "step": 380200
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.7525702770092563e-05,
+      "loss": 2.4673,
+      "step": 380300
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.7517163643815965e-05,
+      "loss": 2.5531,
+      "step": 380400
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.7508624517539366e-05,
+      "loss": 2.5371,
+      "step": 380500
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.7500085391262764e-05,
+      "loss": 2.692,
+      "step": 380600
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.7491546264986166e-05,
+      "loss": 2.2664,
+      "step": 380700
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 1.7483007138709568e-05,
+      "loss": 2.4715,
+      "step": 380800
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 1.747446801243297e-05,
+      "loss": 2.4072,
+      "step": 380900
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 1.7465928886156367e-05,
+      "loss": 2.3953,
+      "step": 381000
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 1.745738975987977e-05,
+      "loss": 2.3845,
+      "step": 381100
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.744885063360317e-05,
+      "loss": 2.4376,
+      "step": 381200
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.7440311507326572e-05,
+      "loss": 2.5648,
+      "step": 381300
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.743177238104997e-05,
+      "loss": 2.4225,
+      "step": 381400
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.742323325477337e-05,
+      "loss": 2.2958,
+      "step": 381500
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.7414694128496773e-05,
+      "loss": 2.5642,
+      "step": 381600
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.7406155002220175e-05,
+      "loss": 2.4446,
+      "step": 381700
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.7397615875943573e-05,
+      "loss": 2.4259,
+      "step": 381800
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.7389076749666974e-05,
+      "loss": 2.4643,
+      "step": 381900
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.7380537623390376e-05,
+      "loss": 2.5778,
+      "step": 382000
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.7371998497113774e-05,
+      "loss": 2.3056,
+      "step": 382100
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.7363459370837176e-05,
+      "loss": 2.4528,
+      "step": 382200
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.7354920244560577e-05,
+      "loss": 2.5293,
+      "step": 382300
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.734638111828398e-05,
+      "loss": 2.552,
+      "step": 382400
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.7337841992007377e-05,
+      "loss": 2.4343,
+      "step": 382500
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.732930286573078e-05,
+      "loss": 2.4091,
+      "step": 382600
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.732076373945418e-05,
+      "loss": 2.5018,
+      "step": 382700
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 1.731222461317758e-05,
+      "loss": 2.4653,
+      "step": 382800
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 1.730368548690098e-05,
+      "loss": 2.4213,
+      "step": 382900
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 1.729514636062438e-05,
+      "loss": 2.5149,
+      "step": 383000
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 1.7286607234347783e-05,
+      "loss": 2.3389,
+      "step": 383100
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.727806810807118e-05,
+      "loss": 2.2932,
+      "step": 383200
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.7269528981794583e-05,
+      "loss": 2.425,
+      "step": 383300
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.7260989855517984e-05,
+      "loss": 2.5606,
+      "step": 383400
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.7252450729241386e-05,
+      "loss": 2.2893,
+      "step": 383500
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.7243911602964784e-05,
+      "loss": 2.6422,
+      "step": 383600
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.7235372476688185e-05,
+      "loss": 2.4667,
+      "step": 383700
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.7226833350411587e-05,
+      "loss": 2.4505,
+      "step": 383800
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.721829422413499e-05,
+      "loss": 2.4373,
+      "step": 383900
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 1.7209755097858387e-05,
+      "loss": 2.5523,
+      "step": 384000
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 1.7201215971581788e-05,
+      "loss": 2.5825,
+      "step": 384100
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 1.719267684530519e-05,
+      "loss": 2.2537,
+      "step": 384200
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 1.718413771902859e-05,
+      "loss": 2.592,
+      "step": 384300
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.717559859275199e-05,
+      "loss": 2.391,
+      "step": 384400
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.716705946647539e-05,
+      "loss": 2.5364,
+      "step": 384500
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.7158520340198792e-05,
+      "loss": 2.3426,
+      "step": 384600
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 1.714998121392219e-05,
+      "loss": 2.6583,
+      "step": 384700
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 1.7141442087645592e-05,
+      "loss": 2.4238,
+      "step": 384800
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 1.7132902961368994e-05,
+      "loss": 2.2801,
+      "step": 384900
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 1.7124363835092395e-05,
+      "loss": 2.495,
+      "step": 385000
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 1.7115824708815793e-05,
+      "loss": 2.4237,
+      "step": 385100
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 1.7107285582539195e-05,
+      "loss": 2.4236,
+      "step": 385200
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 1.7098746456262597e-05,
+      "loss": 2.2743,
+      "step": 385300
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 1.7090207329985998e-05,
+      "loss": 2.4369,
+      "step": 385400
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 1.7081668203709396e-05,
+      "loss": 2.568,
+      "step": 385500
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 1.7073129077432798e-05,
+      "loss": 2.3912,
+      "step": 385600
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 1.70645899511562e-05,
+      "loss": 2.4135,
+      "step": 385700
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 1.7056050824879598e-05,
+      "loss": 2.3469,
+      "step": 385800
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 1.7047511698603e-05,
+      "loss": 2.5563,
+      "step": 385900
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 1.70389725723264e-05,
+      "loss": 2.456,
+      "step": 386000
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 1.7030433446049802e-05,
+      "loss": 2.4716,
+      "step": 386100
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 1.70218943197732e-05,
+      "loss": 2.4082,
+      "step": 386200
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 1.7013355193496602e-05,
+      "loss": 2.5351,
+      "step": 386300
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 1.7004816067220003e-05,
+      "loss": 2.4178,
+      "step": 386400
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 1.6996276940943405e-05,
+      "loss": 2.474,
+      "step": 386500
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 1.6987737814666803e-05,
+      "loss": 2.4404,
+      "step": 386600
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 1.6979198688390205e-05,
+      "loss": 2.4461,
+      "step": 386700
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 1.6970659562113606e-05,
+      "loss": 2.5338,
+      "step": 386800
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 1.6962120435837008e-05,
+      "loss": 2.4256,
+      "step": 386900
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 1.6953581309560406e-05,
+      "loss": 2.5794,
+      "step": 387000
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 1.6945042183283807e-05,
+      "loss": 2.3708,
+      "step": 387100
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 1.693650305700721e-05,
+      "loss": 2.5113,
+      "step": 387200
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 1.6927963930730607e-05,
+      "loss": 2.3322,
+      "step": 387300
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 1.691942480445401e-05,
+      "loss": 2.4048,
+      "step": 387400
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 1.691088567817741e-05,
+      "loss": 2.4437,
+      "step": 387500
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 1.6902346551900812e-05,
+      "loss": 2.4859,
+      "step": 387600
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 1.689380742562421e-05,
+      "loss": 2.516,
+      "step": 387700
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 1.688526829934761e-05,
+      "loss": 2.5643,
+      "step": 387800
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.6876729173071013e-05,
+      "loss": 2.4569,
+      "step": 387900
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.6868190046794415e-05,
+      "loss": 2.5162,
+      "step": 388000
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.6859650920517813e-05,
+      "loss": 2.3696,
+      "step": 388100
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.6851111794241214e-05,
+      "loss": 2.4488,
+      "step": 388200
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.6842572667964616e-05,
+      "loss": 2.5093,
+      "step": 388300
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.6834033541688014e-05,
+      "loss": 2.4386,
+      "step": 388400
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.6825494415411416e-05,
+      "loss": 2.3979,
+      "step": 388500
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.6816955289134817e-05,
+      "loss": 2.6886,
+      "step": 388600
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.680841616285822e-05,
+      "loss": 2.5377,
+      "step": 388700
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.6799877036581617e-05,
+      "loss": 2.5474,
+      "step": 388800
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.679133791030502e-05,
+      "loss": 2.562,
+      "step": 388900
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.678279878402842e-05,
+      "loss": 2.4386,
+      "step": 389000
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.677425965775182e-05,
+      "loss": 2.6628,
+      "step": 389100
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.676572053147522e-05,
+      "loss": 2.381,
+      "step": 389200
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.675718140519862e-05,
+      "loss": 2.5905,
+      "step": 389300
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 1.6748642278922023e-05,
+      "loss": 2.2932,
+      "step": 389400
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 1.6740103152645424e-05,
+      "loss": 2.3794,
+      "step": 389500
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 1.6731564026368822e-05,
+      "loss": 2.4643,
+      "step": 389600
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 1.6723024900092224e-05,
+      "loss": 2.3464,
+      "step": 389700
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 1.6714485773815626e-05,
+      "loss": 2.4667,
+      "step": 389800
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 1.6705946647539024e-05,
+      "loss": 2.4188,
+      "step": 389900
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 1.6697407521262425e-05,
+      "loss": 2.273,
+      "step": 390000
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 1.6688868394985827e-05,
+      "loss": 2.517,
+      "step": 390100
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.668032926870923e-05,
+      "loss": 2.405,
+      "step": 390200
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.6671790142432627e-05,
+      "loss": 2.4721,
+      "step": 390300
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.6663251016156028e-05,
+      "loss": 2.3936,
+      "step": 390400
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.665471188987943e-05,
+      "loss": 2.3981,
+      "step": 390500
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.664617276360283e-05,
+      "loss": 2.5434,
+      "step": 390600
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.663763363732623e-05,
+      "loss": 2.4544,
+      "step": 390700
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.662909451104963e-05,
+      "loss": 2.3425,
+      "step": 390800
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.6620555384773032e-05,
+      "loss": 2.4189,
+      "step": 390900
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.661201625849643e-05,
+      "loss": 2.4276,
+      "step": 391000
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.6603477132219832e-05,
+      "loss": 2.3498,
+      "step": 391100
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.6594938005943234e-05,
+      "loss": 2.3557,
+      "step": 391200
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.6586398879666635e-05,
+      "loss": 2.4556,
+      "step": 391300
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.6577859753390033e-05,
+      "loss": 2.477,
+      "step": 391400
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.6569320627113435e-05,
+      "loss": 2.3401,
+      "step": 391500
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.6560781500836836e-05,
+      "loss": 2.5818,
+      "step": 391600
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.6552242374560238e-05,
+      "loss": 2.3227,
+      "step": 391700
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.6543703248283636e-05,
+      "loss": 2.5602,
+      "step": 391800
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.6535164122007038e-05,
+      "loss": 2.6084,
+      "step": 391900
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.652662499573044e-05,
+      "loss": 2.7121,
+      "step": 392000
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.651808586945384e-05,
+      "loss": 2.3617,
+      "step": 392100
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.650954674317724e-05,
+      "loss": 2.3958,
+      "step": 392200
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.650100761690064e-05,
+      "loss": 2.3304,
+      "step": 392300
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.6492468490624042e-05,
+      "loss": 2.2532,
+      "step": 392400
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.648392936434744e-05,
+      "loss": 2.3591,
+      "step": 392500
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.6475390238070842e-05,
+      "loss": 2.3205,
+      "step": 392600
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.6466851111794243e-05,
+      "loss": 2.504,
+      "step": 392700
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.6458311985517645e-05,
+      "loss": 2.4209,
+      "step": 392800
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 1.6449772859241043e-05,
+      "loss": 2.4739,
+      "step": 392900
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 1.6441233732964445e-05,
+      "loss": 2.3814,
+      "step": 393000
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 1.6432694606687846e-05,
+      "loss": 2.5431,
+      "step": 393100
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 1.6424155480411248e-05,
+      "loss": 2.2433,
+      "step": 393200
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 1.6415616354134646e-05,
+      "loss": 2.3893,
+      "step": 393300
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 1.6407077227858047e-05,
+      "loss": 2.3591,
+      "step": 393400
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 1.639853810158145e-05,
+      "loss": 2.3984,
+      "step": 393500
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 1.638999897530485e-05,
+      "loss": 2.3859,
+      "step": 393600
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 1.638145984902825e-05,
+      "loss": 2.4652,
+      "step": 393700
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 1.637292072275165e-05,
+      "loss": 2.5129,
+      "step": 393800
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 1.6364381596475052e-05,
+      "loss": 2.414,
+      "step": 393900
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 1.635584247019845e-05,
+      "loss": 2.4516,
+      "step": 394000
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 1.634730334392185e-05,
+      "loss": 2.4073,
+      "step": 394100
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 1.6338764217645253e-05,
+      "loss": 2.39,
+      "step": 394200
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 1.6330225091368655e-05,
+      "loss": 2.3152,
+      "step": 394300
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 1.6321685965092053e-05,
+      "loss": 2.5527,
+      "step": 394400
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 1.631314683881545e-05,
+      "loss": 2.3535,
+      "step": 394500
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 1.6304607712538852e-05,
+      "loss": 2.4134,
+      "step": 394600
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 1.6296068586262254e-05,
+      "loss": 2.4877,
+      "step": 394700
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 1.6287529459985652e-05,
+      "loss": 2.449,
+      "step": 394800
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 1.6278990333709054e-05,
+      "loss": 2.4072,
+      "step": 394900
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 1.6270451207432455e-05,
+      "loss": 2.4447,
+      "step": 395000
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 1.6261912081155857e-05,
+      "loss": 2.3707,
+      "step": 395100
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 1.6253372954879255e-05,
+      "loss": 2.3866,
+      "step": 395200
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 1.6244833828602657e-05,
+      "loss": 2.4695,
+      "step": 395300
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 1.6236294702326058e-05,
+      "loss": 2.3628,
+      "step": 395400
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 1.622775557604946e-05,
+      "loss": 2.4668,
+      "step": 395500
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 1.6219216449772858e-05,
+      "loss": 2.4534,
+      "step": 395600
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 1.621067732349626e-05,
+      "loss": 2.3748,
+      "step": 395700
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 1.620213819721966e-05,
+      "loss": 2.3559,
+      "step": 395800
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 1.6193599070943062e-05,
+      "loss": 2.4506,
+      "step": 395900
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 1.618505994466646e-05,
+      "loss": 2.1887,
+      "step": 396000
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 1.6176520818389862e-05,
+      "loss": 2.3987,
+      "step": 396100
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 1.6167981692113264e-05,
+      "loss": 2.2123,
+      "step": 396200
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 1.6159442565836662e-05,
+      "loss": 2.4118,
+      "step": 396300
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 1.6150903439560063e-05,
+      "loss": 2.4012,
+      "step": 396400
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 1.6142364313283465e-05,
+      "loss": 2.4437,
+      "step": 396500
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 1.6133825187006866e-05,
+      "loss": 2.4596,
+      "step": 396600
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 1.6125286060730265e-05,
+      "loss": 2.272,
+      "step": 396700
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 1.6116746934453666e-05,
+      "loss": 2.4186,
+      "step": 396800
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 1.6108207808177068e-05,
+      "loss": 2.3864,
+      "step": 396900
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 1.609966868190047e-05,
+      "loss": 2.5098,
+      "step": 397000
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 1.6091129555623867e-05,
+      "loss": 2.2682,
+      "step": 397100
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 1.608259042934727e-05,
+      "loss": 2.4155,
+      "step": 397200
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 1.607405130307067e-05,
+      "loss": 2.5484,
+      "step": 397300
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 1.606551217679407e-05,
+      "loss": 2.3424,
+      "step": 397400
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 1.605697305051747e-05,
+      "loss": 2.439,
+      "step": 397500
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 1.6048433924240872e-05,
+      "loss": 2.3458,
+      "step": 397600
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 1.6039894797964273e-05,
+      "loss": 2.4439,
+      "step": 397700
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 1.603135567168767e-05,
+      "loss": 2.3554,
+      "step": 397800
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 1.6022816545411073e-05,
+      "loss": 2.4306,
+      "step": 397900
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 1.6014277419134475e-05,
+      "loss": 2.4042,
+      "step": 398000
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 1.6005738292857876e-05,
+      "loss": 2.5338,
+      "step": 398100
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 1.5997199166581274e-05,
+      "loss": 2.452,
+      "step": 398200
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 1.5988660040304676e-05,
+      "loss": 2.3639,
+      "step": 398300
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 1.5980120914028077e-05,
+      "loss": 2.3826,
+      "step": 398400
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 1.597158178775148e-05,
+      "loss": 2.3524,
+      "step": 398500
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 1.5963042661474877e-05,
+      "loss": 2.5094,
+      "step": 398600
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 1.595450353519828e-05,
+      "loss": 2.2688,
+      "step": 398700
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 1.594596440892168e-05,
+      "loss": 2.3228,
+      "step": 398800
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 1.593742528264508e-05,
+      "loss": 2.366,
+      "step": 398900
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 1.592888615636848e-05,
+      "loss": 2.4432,
+      "step": 399000
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 1.592034703009188e-05,
+      "loss": 2.5214,
+      "step": 399100
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 1.5911807903815283e-05,
+      "loss": 2.4662,
+      "step": 399200
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 1.590326877753868e-05,
+      "loss": 2.3677,
+      "step": 399300
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 1.5894729651262083e-05,
+      "loss": 2.3994,
+      "step": 399400
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 1.5886190524985484e-05,
+      "loss": 2.4339,
+      "step": 399500
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 1.5877651398708886e-05,
+      "loss": 2.465,
+      "step": 399600
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 1.5869112272432284e-05,
+      "loss": 2.5653,
+      "step": 399700
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 1.5860573146155686e-05,
+      "loss": 2.4402,
+      "step": 399800
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 1.5852034019879087e-05,
+      "loss": 2.4643,
+      "step": 399900
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 1.5843494893602485e-05,
+      "loss": 2.4224,
+      "step": 400000
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 1.5834955767325887e-05,
+      "loss": 2.3672,
+      "step": 400100
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 1.582641664104929e-05,
+      "loss": 2.3876,
+      "step": 400200
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 1.581787751477269e-05,
+      "loss": 2.2916,
+      "step": 400300
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 1.5809338388496088e-05,
+      "loss": 2.4448,
+      "step": 400400
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 1.580079926221949e-05,
+      "loss": 2.4192,
+      "step": 400500
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 1.579226013594289e-05,
+      "loss": 2.4131,
+      "step": 400600
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 1.5783721009666293e-05,
+      "loss": 2.4761,
+      "step": 400700
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 1.577518188338969e-05,
+      "loss": 2.4109,
+      "step": 400800
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 1.5766642757113092e-05,
+      "loss": 2.3231,
+      "step": 400900
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 1.5758103630836494e-05,
+      "loss": 2.3753,
+      "step": 401000
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 1.5749564504559895e-05,
+      "loss": 2.3626,
+      "step": 401100
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 1.5741025378283294e-05,
+      "loss": 2.4065,
+      "step": 401200
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 1.5732486252006695e-05,
+      "loss": 2.3983,
+      "step": 401300
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 1.5723947125730097e-05,
+      "loss": 2.5158,
+      "step": 401400
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 1.5715407999453495e-05,
+      "loss": 2.4137,
+      "step": 401500
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 1.5706868873176896e-05,
+      "loss": 2.5449,
+      "step": 401600
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 1.5698329746900298e-05,
+      "loss": 2.1586,
+      "step": 401700
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 1.56897906206237e-05,
+      "loss": 2.2384,
+      "step": 401800
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 1.5681251494347098e-05,
+      "loss": 2.4827,
+      "step": 401900
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 1.56727123680705e-05,
+      "loss": 2.6007,
+      "step": 402000
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 1.56641732417939e-05,
+      "loss": 2.5747,
+      "step": 402100
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 1.5655634115517302e-05,
+      "loss": 2.3761,
+      "step": 402200
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 1.56470949892407e-05,
+      "loss": 2.3155,
+      "step": 402300
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 1.5638555862964102e-05,
+      "loss": 2.4255,
+      "step": 402400
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 1.5630016736687504e-05,
+      "loss": 2.5638,
+      "step": 402500
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 1.5621477610410902e-05,
+      "loss": 2.4141,
+      "step": 402600
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 1.5612938484134303e-05,
+      "loss": 2.506,
+      "step": 402700
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 1.5604399357857705e-05,
+      "loss": 2.3318,
+      "step": 402800
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 1.5595860231581106e-05,
+      "loss": 2.5507,
+      "step": 402900
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 1.5587321105304505e-05,
+      "loss": 2.4233,
+      "step": 403000
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 1.5578781979027906e-05,
+      "loss": 2.5108,
+      "step": 403100
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 1.5570242852751308e-05,
+      "loss": 2.3689,
+      "step": 403200
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 1.556170372647471e-05,
+      "loss": 2.2936,
+      "step": 403300
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 1.5553164600198107e-05,
+      "loss": 2.3457,
+      "step": 403400
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 1.554462547392151e-05,
+      "loss": 2.336,
+      "step": 403500
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 1.553608634764491e-05,
+      "loss": 2.5982,
+      "step": 403600
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 1.5527547221368312e-05,
+      "loss": 2.3622,
+      "step": 403700
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 1.551900809509171e-05,
+      "loss": 2.5142,
+      "step": 403800
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 1.5510468968815112e-05,
+      "loss": 2.3936,
+      "step": 403900
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 1.5501929842538513e-05,
+      "loss": 2.5448,
+      "step": 404000
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 1.549339071626191e-05,
+      "loss": 2.3943,
+      "step": 404100
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 1.5484851589985313e-05,
+      "loss": 2.5677,
+      "step": 404200
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 1.5476312463708715e-05,
+      "loss": 2.4401,
+      "step": 404300
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 1.5467773337432116e-05,
+      "loss": 2.34,
+      "step": 404400
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 1.5459234211155514e-05,
+      "loss": 2.3057,
+      "step": 404500
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 1.5450695084878916e-05,
+      "loss": 2.4371,
+      "step": 404600
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 1.5442155958602317e-05,
+      "loss": 2.2235,
+      "step": 404700
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 1.543361683232572e-05,
+      "loss": 2.3583,
+      "step": 404800
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 1.5425077706049117e-05,
+      "loss": 2.3671,
+      "step": 404900
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 1.541653857977252e-05,
+      "loss": 2.4501,
+      "step": 405000
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 1.540799945349592e-05,
+      "loss": 2.4837,
+      "step": 405100
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 1.539946032721932e-05,
+      "loss": 2.4959,
+      "step": 405200
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 1.539092120094272e-05,
+      "loss": 2.5588,
+      "step": 405300
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 1.538238207466612e-05,
+      "loss": 2.388,
+      "step": 405400
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 1.5373842948389523e-05,
+      "loss": 2.3401,
+      "step": 405500
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 1.536530382211292e-05,
+      "loss": 2.513,
+      "step": 405600
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 1.5356764695836323e-05,
+      "loss": 2.3475,
+      "step": 405700
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 1.5348225569559724e-05,
+      "loss": 2.3806,
+      "step": 405800
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 1.5339686443283126e-05,
+      "loss": 2.4245,
+      "step": 405900
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 1.5331147317006524e-05,
+      "loss": 2.3078,
+      "step": 406000
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 1.5322608190729925e-05,
+      "loss": 2.578,
+      "step": 406100
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 1.5314069064453327e-05,
+      "loss": 2.3105,
+      "step": 406200
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 1.530552993817673e-05,
+      "loss": 2.4754,
+      "step": 406300
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 1.5296990811900127e-05,
+      "loss": 2.4158,
+      "step": 406400
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 1.5288451685623528e-05,
+      "loss": 2.5898,
+      "step": 406500
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 1.527991255934693e-05,
+      "loss": 2.578,
+      "step": 406600
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 1.5271373433070328e-05,
+      "loss": 2.5228,
+      "step": 406700
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 1.526283430679373e-05,
+      "loss": 2.3272,
+      "step": 406800
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 1.5254295180517131e-05,
+      "loss": 2.4769,
+      "step": 406900
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 1.5245756054240531e-05,
+      "loss": 2.5348,
+      "step": 407000
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 1.5237216927963932e-05,
+      "loss": 2.4132,
+      "step": 407100
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 1.5228677801687332e-05,
+      "loss": 2.3936,
+      "step": 407200
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 1.5220138675410734e-05,
+      "loss": 2.4092,
+      "step": 407300
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 1.5211599549134134e-05,
+      "loss": 2.3868,
+      "step": 407400
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 1.5203060422857535e-05,
+      "loss": 2.3178,
+      "step": 407500
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 1.5194521296580935e-05,
+      "loss": 2.5125,
+      "step": 407600
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 1.5185982170304337e-05,
+      "loss": 2.7254,
+      "step": 407700
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 1.5177443044027737e-05,
+      "loss": 2.4515,
+      "step": 407800
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 1.5168903917751136e-05,
+      "loss": 2.4938,
+      "step": 407900
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 1.5160364791474538e-05,
+      "loss": 2.396,
+      "step": 408000
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 1.5151825665197938e-05,
+      "loss": 2.3851,
+      "step": 408100
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 1.514328653892134e-05,
+      "loss": 2.5115,
+      "step": 408200
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 1.513474741264474e-05,
+      "loss": 2.4618,
+      "step": 408300
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 1.512620828636814e-05,
+      "loss": 2.343,
+      "step": 408400
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 1.511766916009154e-05,
+      "loss": 2.4015,
+      "step": 408500
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 1.5109130033814942e-05,
+      "loss": 2.3877,
+      "step": 408600
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 1.5100590907538342e-05,
+      "loss": 2.5644,
+      "step": 408700
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 1.5092051781261744e-05,
+      "loss": 2.4597,
+      "step": 408800
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 1.5083512654985143e-05,
+      "loss": 2.5105,
+      "step": 408900
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 1.5074973528708545e-05,
+      "loss": 2.4664,
+      "step": 409000
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 1.5066434402431945e-05,
+      "loss": 2.5024,
+      "step": 409100
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 1.5057895276155345e-05,
+      "loss": 2.3704,
+      "step": 409200
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 1.5049356149878746e-05,
+      "loss": 2.5143,
+      "step": 409300
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 1.5040817023602146e-05,
+      "loss": 2.3204,
+      "step": 409400
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 1.5032277897325548e-05,
+      "loss": 2.4267,
+      "step": 409500
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 1.5023738771048947e-05,
+      "loss": 2.4022,
+      "step": 409600
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 1.5015199644772349e-05,
+      "loss": 2.4902,
+      "step": 409700
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 1.5006660518495749e-05,
+      "loss": 2.4062,
+      "step": 409800
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 1.499812139221915e-05,
+      "loss": 2.3848,
+      "step": 409900
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 1.498958226594255e-05,
+      "loss": 2.3646,
+      "step": 410000
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 1.4981043139665952e-05,
+      "loss": 2.4942,
+      "step": 410100
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 1.4972504013389352e-05,
+      "loss": 2.5237,
+      "step": 410200
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 1.4963964887112753e-05,
+      "loss": 2.4084,
+      "step": 410300
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 1.4955425760836153e-05,
+      "loss": 2.4429,
+      "step": 410400
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 1.4946886634559553e-05,
+      "loss": 2.3814,
+      "step": 410500
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 1.4938347508282954e-05,
+      "loss": 2.6651,
+      "step": 410600
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 1.4929808382006354e-05,
+      "loss": 2.3788,
+      "step": 410700
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 1.4921269255729756e-05,
+      "loss": 2.3782,
+      "step": 410800
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 1.4912730129453156e-05,
+      "loss": 2.444,
+      "step": 410900
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 1.4904191003176557e-05,
+      "loss": 2.4786,
+      "step": 411000
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 1.4895651876899957e-05,
+      "loss": 2.408,
+      "step": 411100
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 1.4887112750623359e-05,
+      "loss": 2.348,
+      "step": 411200
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 1.4878573624346759e-05,
+      "loss": 2.4413,
+      "step": 411300
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 1.487003449807016e-05,
+      "loss": 2.2968,
+      "step": 411400
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 1.486149537179356e-05,
+      "loss": 2.4226,
+      "step": 411500
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 1.4852956245516961e-05,
+      "loss": 2.4854,
+      "step": 411600
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 1.4844417119240361e-05,
+      "loss": 2.3968,
+      "step": 411700
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 1.4835877992963761e-05,
+      "loss": 2.4726,
+      "step": 411800
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 1.4827338866687163e-05,
+      "loss": 2.4465,
+      "step": 411900
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 1.4818799740410561e-05,
+      "loss": 2.4776,
+      "step": 412000
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 1.481026061413396e-05,
+      "loss": 2.3093,
+      "step": 412100
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 1.4801721487857362e-05,
+      "loss": 2.4531,
+      "step": 412200
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 1.4793182361580762e-05,
+      "loss": 2.462,
+      "step": 412300
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 1.4784643235304164e-05,
+      "loss": 2.4598,
+      "step": 412400
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 1.4776104109027564e-05,
+      "loss": 2.3987,
+      "step": 412500
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 1.4767564982750965e-05,
+      "loss": 2.5386,
+      "step": 412600
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 1.4759025856474365e-05,
+      "loss": 2.4256,
+      "step": 412700
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 1.4750486730197765e-05,
+      "loss": 2.4934,
+      "step": 412800
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 1.4741947603921166e-05,
+      "loss": 2.4629,
+      "step": 412900
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 1.4733408477644566e-05,
+      "loss": 2.6451,
+      "step": 413000
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 1.4724869351367968e-05,
+      "loss": 2.2613,
+      "step": 413100
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 1.4716330225091368e-05,
+      "loss": 2.4202,
+      "step": 413200
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 1.470779109881477e-05,
+      "loss": 2.5201,
+      "step": 413300
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 1.4699251972538169e-05,
+      "loss": 2.5148,
+      "step": 413400
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 1.469071284626157e-05,
+      "loss": 2.4361,
+      "step": 413500
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 1.468217371998497e-05,
+      "loss": 2.2988,
+      "step": 413600
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 1.4673634593708372e-05,
+      "loss": 2.3311,
+      "step": 413700
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 1.4665095467431772e-05,
+      "loss": 2.5351,
+      "step": 413800
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 1.4656556341155173e-05,
+      "loss": 2.4125,
+      "step": 413900
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 1.4648017214878573e-05,
+      "loss": 2.3415,
+      "step": 414000
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 1.4639478088601973e-05,
+      "loss": 2.4118,
+      "step": 414100
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 1.4630938962325375e-05,
+      "loss": 2.3756,
+      "step": 414200
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 1.4622399836048774e-05,
+      "loss": 2.5816,
+      "step": 414300
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 1.4613860709772176e-05,
+      "loss": 2.4063,
+      "step": 414400
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 1.4605321583495576e-05,
+      "loss": 2.5107,
+      "step": 414500
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 1.4596782457218977e-05,
+      "loss": 2.3821,
+      "step": 414600
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 1.4588243330942377e-05,
+      "loss": 2.4562,
+      "step": 414700
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 1.4579704204665779e-05,
+      "loss": 2.3442,
+      "step": 414800
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 1.4571165078389179e-05,
+      "loss": 2.4321,
+      "step": 414900
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 1.456262595211258e-05,
+      "loss": 2.2946,
+      "step": 415000
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 1.455408682583598e-05,
+      "loss": 2.2737,
+      "step": 415100
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 1.4545547699559382e-05,
+      "loss": 2.4658,
+      "step": 415200
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 1.4537008573282782e-05,
+      "loss": 2.4046,
+      "step": 415300
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 1.4528469447006181e-05,
+      "loss": 2.443,
+      "step": 415400
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 1.4519930320729583e-05,
+      "loss": 2.4225,
+      "step": 415500
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 1.4511391194452983e-05,
+      "loss": 2.454,
+      "step": 415600
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 1.4502852068176384e-05,
+      "loss": 2.5816,
+      "step": 415700
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 1.4494312941899784e-05,
+      "loss": 2.3252,
+      "step": 415800
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 1.4485773815623186e-05,
+      "loss": 2.192,
+      "step": 415900
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 1.4477234689346586e-05,
+      "loss": 2.4979,
+      "step": 416000
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 1.4468695563069987e-05,
+      "loss": 2.4697,
+      "step": 416100
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 1.4460156436793387e-05,
+      "loss": 2.4549,
+      "step": 416200
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 1.4451617310516789e-05,
+      "loss": 2.353,
+      "step": 416300
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 1.4443078184240188e-05,
+      "loss": 2.3207,
+      "step": 416400
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 1.443453905796359e-05,
+      "loss": 2.4548,
+      "step": 416500
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 1.442599993168699e-05,
+      "loss": 2.54,
+      "step": 416600
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 1.441746080541039e-05,
+      "loss": 2.5123,
+      "step": 416700
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 1.4408921679133791e-05,
+      "loss": 2.3531,
+      "step": 416800
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 1.4400382552857191e-05,
+      "loss": 2.5471,
+      "step": 416900
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 1.4391843426580593e-05,
+      "loss": 2.3131,
+      "step": 417000
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 1.4383304300303992e-05,
+      "loss": 2.3687,
+      "step": 417100
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 1.4374765174027394e-05,
+      "loss": 2.5053,
+      "step": 417200
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 1.4366226047750794e-05,
+      "loss": 2.3076,
+      "step": 417300
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 1.4357686921474195e-05,
+      "loss": 2.5531,
+      "step": 417400
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 1.4349147795197595e-05,
+      "loss": 2.5181,
+      "step": 417500
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 1.4340608668920997e-05,
+      "loss": 2.4228,
+      "step": 417600
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 1.4332069542644397e-05,
+      "loss": 2.529,
+      "step": 417700
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 1.4323530416367798e-05,
+      "loss": 2.3639,
+      "step": 417800
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 1.4314991290091198e-05,
+      "loss": 2.301,
+      "step": 417900
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 1.43064521638146e-05,
+      "loss": 2.5247,
+      "step": 418000
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 1.4297913037538e-05,
+      "loss": 2.4488,
+      "step": 418100
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 1.42893739112614e-05,
+      "loss": 2.5862,
+      "step": 418200
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 1.42808347849848e-05,
+      "loss": 2.1738,
+      "step": 418300
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 1.42722956587082e-05,
+      "loss": 2.4319,
+      "step": 418400
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 1.4263756532431602e-05,
+      "loss": 2.3571,
+      "step": 418500
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 1.4255217406155002e-05,
+      "loss": 2.3887,
+      "step": 418600
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 1.4246678279878404e-05,
+      "loss": 2.5333,
+      "step": 418700
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 1.4238139153601804e-05,
+      "loss": 2.3024,
+      "step": 418800
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 1.4229600027325205e-05,
+      "loss": 2.4367,
+      "step": 418900
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 1.4221060901048605e-05,
+      "loss": 2.4616,
+      "step": 419000
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 1.4212521774772006e-05,
+      "loss": 2.5386,
+      "step": 419100
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 1.4203982648495406e-05,
+      "loss": 2.4597,
+      "step": 419200
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 1.4195443522218808e-05,
+      "loss": 2.3248,
+      "step": 419300
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 1.4186904395942208e-05,
+      "loss": 2.5811,
+      "step": 419400
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 1.4178365269665608e-05,
+      "loss": 2.3857,
+      "step": 419500
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 1.4169826143389009e-05,
+      "loss": 2.2709,
+      "step": 419600
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 1.4161287017112409e-05,
+      "loss": 2.4978,
+      "step": 419700
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 1.415274789083581e-05,
+      "loss": 2.4434,
+      "step": 419800
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 1.414420876455921e-05,
+      "loss": 2.4206,
+      "step": 419900
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 1.4135669638282612e-05,
+      "loss": 2.3937,
+      "step": 420000
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 1.4127130512006012e-05,
+      "loss": 2.3275,
+      "step": 420100
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 1.4118591385729413e-05,
+      "loss": 2.3028,
+      "step": 420200
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 1.4110052259452813e-05,
+      "loss": 2.458,
+      "step": 420300
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 1.4101513133176215e-05,
+      "loss": 2.5485,
+      "step": 420400
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 1.4092974006899615e-05,
+      "loss": 2.4615,
+      "step": 420500
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 1.4084434880623016e-05,
+      "loss": 2.4002,
+      "step": 420600
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 1.4075895754346416e-05,
+      "loss": 2.3733,
+      "step": 420700
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 1.4067356628069816e-05,
+      "loss": 2.4245,
+      "step": 420800
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 1.4058817501793217e-05,
+      "loss": 2.3932,
+      "step": 420900
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 1.4050278375516617e-05,
+      "loss": 2.5509,
+      "step": 421000
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 1.4041739249240019e-05,
+      "loss": 2.3252,
+      "step": 421100
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 1.4033200122963419e-05,
+      "loss": 2.4302,
+      "step": 421200
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 1.402466099668682e-05,
+      "loss": 2.4215,
+      "step": 421300
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 1.401612187041022e-05,
+      "loss": 2.314,
+      "step": 421400
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 1.4007582744133622e-05,
+      "loss": 2.3784,
+      "step": 421500
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 1.3999043617857021e-05,
+      "loss": 2.4432,
+      "step": 421600
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 1.3990504491580423e-05,
+      "loss": 2.2759,
+      "step": 421700
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 1.3981965365303823e-05,
+      "loss": 2.2875,
+      "step": 421800
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 1.3973426239027224e-05,
+      "loss": 2.3403,
+      "step": 421900
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 1.3964887112750624e-05,
+      "loss": 2.4533,
+      "step": 422000
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 1.3956347986474024e-05,
+      "loss": 2.3412,
+      "step": 422100
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 1.3947808860197426e-05,
+      "loss": 2.5796,
+      "step": 422200
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 1.3939269733920826e-05,
+      "loss": 2.275,
+      "step": 422300
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 1.3930730607644227e-05,
+      "loss": 2.3748,
+      "step": 422400
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 1.3922191481367627e-05,
+      "loss": 2.5889,
+      "step": 422500
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 1.3913652355091028e-05,
+      "loss": 2.5048,
+      "step": 422600
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 1.3905113228814428e-05,
+      "loss": 2.4468,
+      "step": 422700
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 1.389657410253783e-05,
+      "loss": 2.229,
+      "step": 422800
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 1.388803497626123e-05,
+      "loss": 2.4576,
+      "step": 422900
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 1.3879495849984631e-05,
+      "loss": 2.4826,
+      "step": 423000
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 1.3870956723708031e-05,
+      "loss": 2.436,
+      "step": 423100
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 1.3862417597431433e-05,
+      "loss": 2.3486,
+      "step": 423200
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 1.3853878471154833e-05,
+      "loss": 2.4808,
+      "step": 423300
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 1.3845339344878232e-05,
+      "loss": 2.5013,
+      "step": 423400
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 1.3836800218601634e-05,
+      "loss": 2.4706,
+      "step": 423500
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 1.3828261092325034e-05,
+      "loss": 2.3887,
+      "step": 423600
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 1.3819721966048435e-05,
+      "loss": 2.3089,
+      "step": 423700
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 1.3811182839771835e-05,
+      "loss": 2.3603,
+      "step": 423800
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 1.3802643713495237e-05,
+      "loss": 2.2136,
+      "step": 423900
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 1.3794104587218637e-05,
+      "loss": 2.3481,
+      "step": 424000
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 1.3785565460942038e-05,
+      "loss": 2.2901,
+      "step": 424100
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 1.3777026334665438e-05,
+      "loss": 2.3176,
+      "step": 424200
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 1.376848720838884e-05,
+      "loss": 2.4147,
+      "step": 424300
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 1.375994808211224e-05,
+      "loss": 2.5023,
+      "step": 424400
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 1.3751408955835641e-05,
+      "loss": 2.468,
+      "step": 424500
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 1.374286982955904e-05,
+      "loss": 2.3227,
+      "step": 424600
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 1.373433070328244e-05,
+      "loss": 2.3984,
+      "step": 424700
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 1.3725791577005842e-05,
+      "loss": 2.393,
+      "step": 424800
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 1.3717252450729242e-05,
+      "loss": 2.4668,
+      "step": 424900
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 1.3708713324452644e-05,
+      "loss": 2.5237,
+      "step": 425000
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 1.3700174198176043e-05,
+      "loss": 2.3748,
+      "step": 425100
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 1.3691635071899445e-05,
+      "loss": 2.3424,
+      "step": 425200
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 1.3683095945622845e-05,
+      "loss": 2.4823,
+      "step": 425300
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 1.3674556819346246e-05,
+      "loss": 2.4264,
+      "step": 425400
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 1.3666017693069646e-05,
+      "loss": 2.459,
+      "step": 425500
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 1.3657478566793048e-05,
+      "loss": 2.4273,
+      "step": 425600
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 1.3648939440516448e-05,
+      "loss": 2.3875,
+      "step": 425700
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 1.364040031423985e-05,
+      "loss": 2.39,
+      "step": 425800
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 1.3631861187963249e-05,
+      "loss": 2.5833,
+      "step": 425900
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 1.3623322061686649e-05,
+      "loss": 2.5368,
+      "step": 426000
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 1.361478293541005e-05,
+      "loss": 2.2617,
+      "step": 426100
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 1.360624380913345e-05,
+      "loss": 2.4038,
+      "step": 426200
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 1.3597704682856852e-05,
+      "loss": 2.3052,
+      "step": 426300
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 1.3589165556580252e-05,
+      "loss": 2.3489,
+      "step": 426400
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 1.3580626430303653e-05,
+      "loss": 2.4012,
+      "step": 426500
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 1.3572087304027053e-05,
+      "loss": 2.3726,
+      "step": 426600
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 1.3563548177750455e-05,
+      "loss": 2.3671,
+      "step": 426700
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 1.3555009051473855e-05,
+      "loss": 2.3331,
+      "step": 426800
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 1.3546469925197256e-05,
+      "loss": 2.2823,
+      "step": 426900
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 1.3537930798920656e-05,
+      "loss": 2.4966,
+      "step": 427000
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 1.3529391672644057e-05,
+      "loss": 2.233,
+      "step": 427100
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 1.3520852546367457e-05,
+      "loss": 2.5499,
+      "step": 427200
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 1.3512313420090857e-05,
+      "loss": 2.4112,
+      "step": 427300
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 1.3503774293814259e-05,
+      "loss": 2.2821,
+      "step": 427400
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 1.3495235167537659e-05,
+      "loss": 2.3467,
+      "step": 427500
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 1.348669604126106e-05,
+      "loss": 2.436,
+      "step": 427600
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 1.347815691498446e-05,
+      "loss": 2.3273,
+      "step": 427700
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 1.3469617788707862e-05,
+      "loss": 2.3415,
+      "step": 427800
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 1.3461078662431261e-05,
+      "loss": 2.4805,
+      "step": 427900
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 1.3452539536154663e-05,
+      "loss": 2.1796,
+      "step": 428000
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 1.3444000409878063e-05,
+      "loss": 2.2917,
+      "step": 428100
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 1.3435461283601464e-05,
+      "loss": 2.3663,
+      "step": 428200
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 1.3426922157324864e-05,
+      "loss": 2.3382,
+      "step": 428300
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 1.3418383031048266e-05,
+      "loss": 2.5369,
+      "step": 428400
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 1.3409843904771666e-05,
+      "loss": 2.4268,
+      "step": 428500
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 1.3401304778495065e-05,
+      "loss": 2.6242,
+      "step": 428600
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 1.3392765652218467e-05,
+      "loss": 2.2714,
+      "step": 428700
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 1.3384226525941867e-05,
+      "loss": 2.5422,
+      "step": 428800
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 1.3375687399665268e-05,
+      "loss": 2.4616,
+      "step": 428900
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 1.3367148273388668e-05,
+      "loss": 2.47,
+      "step": 429000
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 1.335860914711207e-05,
+      "loss": 2.4516,
+      "step": 429100
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 1.335007002083547e-05,
+      "loss": 2.26,
+      "step": 429200
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.3341530894558871e-05,
+      "loss": 2.2387,
+      "step": 429300
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.3332991768282271e-05,
+      "loss": 2.3875,
+      "step": 429400
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.3324452642005673e-05,
+      "loss": 2.441,
+      "step": 429500
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.3315913515729069e-05,
+      "loss": 2.4267,
+      "step": 429600
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.330737438945247e-05,
+      "loss": 2.5106,
+      "step": 429700
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.329883526317587e-05,
+      "loss": 2.2311,
+      "step": 429800
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.3290296136899272e-05,
+      "loss": 2.3798,
+      "step": 429900
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3281757010622672e-05,
+      "loss": 2.313,
+      "step": 430000
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3273217884346073e-05,
+      "loss": 2.2904,
+      "step": 430100
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3264678758069473e-05,
+      "loss": 2.461,
+      "step": 430200
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3256139631792875e-05,
+      "loss": 2.422,
+      "step": 430300
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.3247600505516275e-05,
+      "loss": 2.5934,
+      "step": 430400
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.3239061379239676e-05,
+      "loss": 2.4207,
+      "step": 430500
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.3230522252963076e-05,
+      "loss": 2.3234,
+      "step": 430600
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.3221983126686478e-05,
+      "loss": 2.2679,
+      "step": 430700
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.3213444000409877e-05,
+      "loss": 2.5224,
+      "step": 430800
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.3204904874133277e-05,
+      "loss": 2.6412,
+      "step": 430900
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.3196365747856679e-05,
+      "loss": 2.3197,
+      "step": 431000
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.3187826621580079e-05,
+      "loss": 2.3307,
+      "step": 431100
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.317928749530348e-05,
+      "loss": 2.3558,
+      "step": 431200
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.317074836902688e-05,
+      "loss": 2.4468,
+      "step": 431300
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.3162209242750282e-05,
+      "loss": 2.2056,
+      "step": 431400
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.3153670116473682e-05,
+      "loss": 2.4161,
+      "step": 431500
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.3145130990197083e-05,
+      "loss": 2.3587,
+      "step": 431600
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.3136591863920483e-05,
+      "loss": 2.488,
+      "step": 431700
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.3128052737643884e-05,
+      "loss": 2.3696,
+      "step": 431800
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.3119513611367284e-05,
+      "loss": 2.4558,
+      "step": 431900
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 1.3110974485090686e-05,
+      "loss": 2.3515,
+      "step": 432000
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 1.3102435358814086e-05,
+      "loss": 2.1881,
+      "step": 432100
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 1.3093896232537486e-05,
+      "loss": 2.3457,
+      "step": 432200
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 1.3085357106260887e-05,
+      "loss": 2.3209,
+      "step": 432300
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 1.3076817979984287e-05,
+      "loss": 2.3805,
+      "step": 432400
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 1.3068278853707689e-05,
+      "loss": 2.4787,
+      "step": 432500
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 1.3059739727431088e-05,
+      "loss": 2.4628,
+      "step": 432600
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 1.305120060115449e-05,
+      "loss": 2.4036,
+      "step": 432700
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 1.304266147487789e-05,
+      "loss": 2.2851,
+      "step": 432800
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 1.3034122348601291e-05,
+      "loss": 2.3833,
+      "step": 432900
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 1.3025583222324691e-05,
+      "loss": 2.2168,
+      "step": 433000
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 1.3017044096048093e-05,
+      "loss": 2.4524,
+      "step": 433100
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 1.3008504969771493e-05,
+      "loss": 2.3643,
+      "step": 433200
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 1.2999965843494894e-05,
+      "loss": 2.335,
+      "step": 433300
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 1.2991426717218294e-05,
+      "loss": 2.457,
+      "step": 433400
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 1.2982887590941694e-05,
+      "loss": 2.2603,
+      "step": 433500
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 1.2974348464665095e-05,
+      "loss": 2.3553,
+      "step": 433600
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 1.2965809338388495e-05,
+      "loss": 2.4858,
+      "step": 433700
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 1.2957270212111897e-05,
+      "loss": 2.5419,
+      "step": 433800
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 1.2948731085835297e-05,
+      "loss": 2.3164,
+      "step": 433900
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 1.2940191959558698e-05,
+      "loss": 2.333,
+      "step": 434000
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 1.2931652833282098e-05,
+      "loss": 2.2533,
+      "step": 434100
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 1.29231137070055e-05,
+      "loss": 2.5749,
+      "step": 434200
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 1.29145745807289e-05,
+      "loss": 2.4057,
+      "step": 434300
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 1.2906035454452301e-05,
+      "loss": 2.3012,
+      "step": 434400
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 1.2897496328175701e-05,
+      "loss": 2.4791,
+      "step": 434500
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 1.2888957201899102e-05,
+      "loss": 2.4412,
+      "step": 434600
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 1.2880418075622502e-05,
+      "loss": 2.3738,
+      "step": 434700
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 1.2871878949345904e-05,
+      "loss": 2.3775,
+      "step": 434800
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 1.2863339823069304e-05,
+      "loss": 2.391,
+      "step": 434900
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 1.2854800696792704e-05,
+      "loss": 2.437,
+      "step": 435000
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 1.2846261570516105e-05,
+      "loss": 2.4915,
+      "step": 435100
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 1.2837722444239505e-05,
+      "loss": 2.5653,
+      "step": 435200
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 1.2829183317962906e-05,
+      "loss": 2.3941,
+      "step": 435300
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 1.2820644191686306e-05,
+      "loss": 2.6094,
+      "step": 435400
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 1.2812105065409708e-05,
+      "loss": 2.3746,
+      "step": 435500
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 1.2803565939133108e-05,
+      "loss": 2.5318,
+      "step": 435600
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 1.279502681285651e-05,
+      "loss": 2.3462,
+      "step": 435700
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 1.2786487686579909e-05,
+      "loss": 2.3414,
+      "step": 435800
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 1.277794856030331e-05,
+      "loss": 2.4234,
+      "step": 435900
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 1.276940943402671e-05,
+      "loss": 2.3881,
+      "step": 436000
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 1.2760870307750112e-05,
+      "loss": 2.4492,
+      "step": 436100
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 1.2752331181473512e-05,
+      "loss": 2.3043,
+      "step": 436200
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 1.2743792055196912e-05,
+      "loss": 2.2461,
+      "step": 436300
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 1.2735252928920313e-05,
+      "loss": 2.4854,
+      "step": 436400
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 1.2726713802643713e-05,
+      "loss": 2.3799,
+      "step": 436500
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 1.2718174676367115e-05,
+      "loss": 2.4182,
+      "step": 436600
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 1.2709635550090515e-05,
+      "loss": 2.3611,
+      "step": 436700
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 1.2701096423813916e-05,
+      "loss": 2.3318,
+      "step": 436800
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 1.2692557297537316e-05,
+      "loss": 2.3417,
+      "step": 436900
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 1.2684018171260718e-05,
+      "loss": 2.5033,
+      "step": 437000
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 1.2675479044984117e-05,
+      "loss": 2.371,
+      "step": 437100
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 1.2666939918707519e-05,
+      "loss": 2.4119,
+      "step": 437200
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 1.2658400792430919e-05,
+      "loss": 2.3906,
+      "step": 437300
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 1.264986166615432e-05,
+      "loss": 2.3757,
+      "step": 437400
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 1.264132253987772e-05,
+      "loss": 2.3541,
+      "step": 437500
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 1.263278341360112e-05,
+      "loss": 2.3389,
+      "step": 437600
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 1.2624244287324522e-05,
+      "loss": 2.4163,
+      "step": 437700
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 1.2615705161047921e-05,
+      "loss": 2.2918,
+      "step": 437800
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 1.2607166034771323e-05,
+      "loss": 2.4072,
+      "step": 437900
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 1.2598626908494723e-05,
+      "loss": 2.2518,
+      "step": 438000
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 1.2590087782218124e-05,
+      "loss": 2.4738,
+      "step": 438100
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 1.2581548655941524e-05,
+      "loss": 2.2788,
+      "step": 438200
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 1.2573009529664926e-05,
+      "loss": 2.3523,
+      "step": 438300
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 1.2564470403388326e-05,
+      "loss": 2.5445,
+      "step": 438400
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 1.2555931277111727e-05,
+      "loss": 2.391,
+      "step": 438500
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 1.2547392150835127e-05,
+      "loss": 2.4425,
+      "step": 438600
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 1.2538853024558529e-05,
+      "loss": 2.4095,
+      "step": 438700
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 1.2530313898281928e-05,
+      "loss": 2.5371,
+      "step": 438800
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 1.2521774772005328e-05,
+      "loss": 2.3896,
+      "step": 438900
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 1.251323564572873e-05,
+      "loss": 2.4364,
+      "step": 439000
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 1.250469651945213e-05,
+      "loss": 2.4755,
+      "step": 439100
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 1.2496157393175531e-05,
+      "loss": 2.2854,
+      "step": 439200
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 1.2487618266898931e-05,
+      "loss": 2.5707,
+      "step": 439300
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 1.2479079140622333e-05,
+      "loss": 2.3527,
+      "step": 439400
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 1.2470540014345733e-05,
+      "loss": 2.4184,
+      "step": 439500
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 1.2462000888069134e-05,
+      "loss": 2.2633,
+      "step": 439600
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 1.2453461761792534e-05,
+      "loss": 2.2966,
+      "step": 439700
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 1.2444922635515936e-05,
+      "loss": 2.3354,
+      "step": 439800
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 1.2436383509239335e-05,
+      "loss": 2.2734,
+      "step": 439900
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 1.2427844382962737e-05,
+      "loss": 2.2729,
+      "step": 440000
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 1.2419305256686137e-05,
+      "loss": 2.4983,
+      "step": 440100
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 1.2410766130409537e-05,
+      "loss": 2.3772,
+      "step": 440200
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 1.2402227004132938e-05,
+      "loss": 2.4128,
+      "step": 440300
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 1.2393687877856338e-05,
+      "loss": 2.3746,
+      "step": 440400
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 1.238514875157974e-05,
+      "loss": 2.4524,
+      "step": 440500
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 1.237660962530314e-05,
+      "loss": 2.41,
+      "step": 440600
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 1.2368070499026541e-05,
+      "loss": 2.3369,
+      "step": 440700
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 1.235953137274994e-05,
+      "loss": 2.3432,
+      "step": 440800
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 1.2350992246473342e-05,
+      "loss": 2.4829,
+      "step": 440900
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 1.2342453120196742e-05,
+      "loss": 2.4246,
+      "step": 441000
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 1.2333913993920144e-05,
+      "loss": 2.5314,
+      "step": 441100
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 1.2325374867643544e-05,
+      "loss": 2.3748,
+      "step": 441200
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 1.2316835741366945e-05,
+      "loss": 2.4952,
+      "step": 441300
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 1.2308296615090345e-05,
+      "loss": 2.3488,
+      "step": 441400
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 1.2299757488813745e-05,
+      "loss": 2.5501,
+      "step": 441500
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 1.2291218362537146e-05,
+      "loss": 2.3405,
+      "step": 441600
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 1.2282679236260546e-05,
+      "loss": 2.5885,
+      "step": 441700
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 1.2274140109983948e-05,
+      "loss": 2.3693,
+      "step": 441800
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 1.2265600983707348e-05,
+      "loss": 2.4954,
+      "step": 441900
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 1.225706185743075e-05,
+      "loss": 2.4335,
+      "step": 442000
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 1.2248522731154149e-05,
+      "loss": 2.3323,
+      "step": 442100
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 1.223998360487755e-05,
+      "loss": 2.4281,
+      "step": 442200
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 1.223144447860095e-05,
+      "loss": 2.4198,
+      "step": 442300
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 1.2222905352324352e-05,
+      "loss": 2.515,
+      "step": 442400
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 1.2214366226047752e-05,
+      "loss": 2.2983,
+      "step": 442500
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 1.2205827099771153e-05,
+      "loss": 2.401,
+      "step": 442600
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 1.2197287973494552e-05,
+      "loss": 2.3863,
+      "step": 442700
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 1.2188748847217953e-05,
+      "loss": 2.3331,
+      "step": 442800
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 1.2180209720941353e-05,
+      "loss": 2.4432,
+      "step": 442900
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 1.2171670594664755e-05,
+      "loss": 2.4303,
+      "step": 443000
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 1.2163131468388154e-05,
+      "loss": 2.3619,
+      "step": 443100
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 1.2154592342111556e-05,
+      "loss": 2.18,
+      "step": 443200
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 1.2146053215834956e-05,
+      "loss": 2.3913,
+      "step": 443300
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 1.2137514089558356e-05,
+      "loss": 2.4747,
+      "step": 443400
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 1.2128974963281757e-05,
+      "loss": 2.3488,
+      "step": 443500
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 1.2120435837005157e-05,
+      "loss": 2.4871,
+      "step": 443600
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 1.2111896710728559e-05,
+      "loss": 2.3957,
+      "step": 443700
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 1.2103357584451958e-05,
+      "loss": 2.5369,
+      "step": 443800
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 1.209481845817536e-05,
+      "loss": 2.4831,
+      "step": 443900
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 1.208627933189876e-05,
+      "loss": 2.3505,
+      "step": 444000
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 1.2077740205622161e-05,
+      "loss": 2.4901,
+      "step": 444100
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 1.2069201079345561e-05,
+      "loss": 2.3502,
+      "step": 444200
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 1.2060661953068963e-05,
+      "loss": 2.2579,
+      "step": 444300
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 1.2052122826792363e-05,
+      "loss": 2.3426,
+      "step": 444400
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 1.2043583700515764e-05,
+      "loss": 2.2187,
+      "step": 444500
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 1.2035044574239164e-05,
+      "loss": 2.2769,
+      "step": 444600
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 1.2026505447962564e-05,
+      "loss": 2.4459,
+      "step": 444700
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 1.2017966321685965e-05,
+      "loss": 2.3774,
+      "step": 444800
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 1.2009427195409365e-05,
+      "loss": 2.4176,
+      "step": 444900
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 1.2000888069132767e-05,
+      "loss": 2.3299,
+      "step": 445000
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 1.1992348942856167e-05,
+      "loss": 2.2312,
+      "step": 445100
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 1.1983809816579568e-05,
+      "loss": 2.3771,
+      "step": 445200
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 1.1975270690302968e-05,
+      "loss": 2.3815,
+      "step": 445300
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 1.196673156402637e-05,
+      "loss": 2.3026,
+      "step": 445400
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 1.195819243774977e-05,
+      "loss": 2.486,
+      "step": 445500
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 1.1949653311473171e-05,
+      "loss": 2.4579,
+      "step": 445600
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 1.1941114185196571e-05,
+      "loss": 2.298,
+      "step": 445700
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 1.1932575058919972e-05,
+      "loss": 2.2979,
+      "step": 445800
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 1.1924035932643372e-05,
+      "loss": 2.5232,
+      "step": 445900
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 1.1915496806366772e-05,
+      "loss": 2.289,
+      "step": 446000
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 1.1906957680090174e-05,
+      "loss": 2.2656,
+      "step": 446100
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 1.1898418553813574e-05,
+      "loss": 2.3435,
+      "step": 446200
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 1.1889879427536975e-05,
+      "loss": 2.3753,
+      "step": 446300
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 1.1881340301260375e-05,
+      "loss": 2.4717,
+      "step": 446400
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 1.1872801174983777e-05,
+      "loss": 2.4375,
+      "step": 446500
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 1.1864262048707176e-05,
+      "loss": 2.3796,
+      "step": 446600
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 1.1855722922430578e-05,
+      "loss": 2.3647,
+      "step": 446700
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 1.1847183796153978e-05,
+      "loss": 2.5385,
+      "step": 446800
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 1.183864466987738e-05,
+      "loss": 2.3397,
+      "step": 446900
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 1.183010554360078e-05,
+      "loss": 2.3079,
+      "step": 447000
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 1.182156641732418e-05,
+      "loss": 2.4263,
+      "step": 447100
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 1.181302729104758e-05,
+      "loss": 2.3297,
+      "step": 447200
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 1.180448816477098e-05,
+      "loss": 2.2843,
+      "step": 447300
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 1.1795949038494382e-05,
+      "loss": 2.3753,
+      "step": 447400
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 1.1787409912217782e-05,
+      "loss": 2.448,
+      "step": 447500
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 1.1778870785941183e-05,
+      "loss": 2.2821,
+      "step": 447600
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 1.1770331659664583e-05,
+      "loss": 2.4746,
+      "step": 447700
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 1.1761792533387985e-05,
+      "loss": 2.4732,
+      "step": 447800
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 1.1753253407111385e-05,
+      "loss": 2.3822,
+      "step": 447900
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 1.1744714280834786e-05,
+      "loss": 2.3055,
+      "step": 448000
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 1.1736175154558186e-05,
+      "loss": 2.2832,
+      "step": 448100
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 1.1727636028281588e-05,
+      "loss": 2.4172,
+      "step": 448200
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 1.1719096902004987e-05,
+      "loss": 2.5879,
+      "step": 448300
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 1.1710557775728389e-05,
+      "loss": 2.3739,
+      "step": 448400
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 1.1702018649451789e-05,
+      "loss": 2.412,
+      "step": 448500
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 1.1693479523175189e-05,
+      "loss": 2.2493,
+      "step": 448600
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 1.168494039689859e-05,
+      "loss": 2.5437,
+      "step": 448700
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 1.167640127062199e-05,
+      "loss": 2.4373,
+      "step": 448800
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 1.1667862144345392e-05,
+      "loss": 2.3772,
+      "step": 448900
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 1.1659323018068792e-05,
+      "loss": 2.491,
+      "step": 449000
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 1.1650783891792193e-05,
+      "loss": 2.3499,
+      "step": 449100
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 1.1642244765515593e-05,
+      "loss": 2.4852,
+      "step": 449200
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 1.1633705639238994e-05,
+      "loss": 2.2343,
+      "step": 449300
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 1.1625166512962394e-05,
+      "loss": 2.2328,
+      "step": 449400
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 1.1616627386685796e-05,
+      "loss": 2.293,
+      "step": 449500
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 1.1608088260409196e-05,
+      "loss": 2.2728,
+      "step": 449600
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 1.1599549134132597e-05,
+      "loss": 2.3773,
+      "step": 449700
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 1.1591010007855997e-05,
+      "loss": 2.5609,
+      "step": 449800
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 1.1582470881579397e-05,
+      "loss": 2.3769,
+      "step": 449900
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 1.1573931755302799e-05,
+      "loss": 2.4467,
+      "step": 450000
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 1.1565392629026198e-05,
+      "loss": 2.4738,
+      "step": 450100
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 1.15568535027496e-05,
+      "loss": 2.3608,
+      "step": 450200
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 1.1548314376473e-05,
+      "loss": 2.307,
+      "step": 450300
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 1.1539775250196401e-05,
+      "loss": 2.386,
+      "step": 450400
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 1.1531236123919801e-05,
+      "loss": 2.5637,
+      "step": 450500
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 1.1522696997643203e-05,
+      "loss": 2.3824,
+      "step": 450600
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 1.1514157871366603e-05,
+      "loss": 2.4299,
+      "step": 450700
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 1.1505618745090004e-05,
+      "loss": 2.359,
+      "step": 450800
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 1.1497079618813404e-05,
+      "loss": 2.4056,
+      "step": 450900
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 1.1488540492536806e-05,
+      "loss": 2.4729,
+      "step": 451000
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 1.1480001366260205e-05,
+      "loss": 2.4801,
+      "step": 451100
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 1.1471462239983605e-05,
+      "loss": 2.4627,
+      "step": 451200
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 1.1462923113707007e-05,
+      "loss": 2.2652,
+      "step": 451300
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 1.1454383987430407e-05,
+      "loss": 2.3068,
+      "step": 451400
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 1.1445844861153807e-05,
+      "loss": 2.4512,
+      "step": 451500
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 1.1437305734877208e-05,
+      "loss": 2.382,
+      "step": 451600
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 1.1428766608600608e-05,
+      "loss": 2.2647,
+      "step": 451700
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 1.1420227482324008e-05,
+      "loss": 2.3449,
+      "step": 451800
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 1.141168835604741e-05,
+      "loss": 2.3118,
+      "step": 451900
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 1.140314922977081e-05,
+      "loss": 2.4311,
+      "step": 452000
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 1.139461010349421e-05,
+      "loss": 2.2573,
+      "step": 452100
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 1.138607097721761e-05,
+      "loss": 2.2957,
+      "step": 452200
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 1.1377531850941012e-05,
+      "loss": 2.4046,
+      "step": 452300
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 1.1368992724664412e-05,
+      "loss": 2.3726,
+      "step": 452400
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 1.1360453598387814e-05,
+      "loss": 2.2297,
+      "step": 452500
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 1.1351914472111213e-05,
+      "loss": 2.4903,
+      "step": 452600
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 1.1343375345834615e-05,
+      "loss": 2.2608,
+      "step": 452700
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 1.1334836219558015e-05,
+      "loss": 2.2555,
+      "step": 452800
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 1.1326297093281416e-05,
+      "loss": 2.4198,
+      "step": 452900
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 1.1317757967004816e-05,
+      "loss": 2.2123,
+      "step": 453000
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 1.1309218840728216e-05,
+      "loss": 2.4742,
+      "step": 453100
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 1.1300679714451618e-05,
+      "loss": 2.413,
+      "step": 453200
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 1.1292140588175017e-05,
+      "loss": 2.5419,
+      "step": 453300
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 1.1283601461898419e-05,
+      "loss": 2.3603,
+      "step": 453400
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 1.1275062335621819e-05,
+      "loss": 2.5009,
+      "step": 453500
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 1.126652320934522e-05,
+      "loss": 2.3219,
+      "step": 453600
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 1.125798408306862e-05,
+      "loss": 2.2979,
+      "step": 453700
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 1.1249444956792022e-05,
+      "loss": 2.3667,
+      "step": 453800
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 1.1240905830515422e-05,
+      "loss": 2.4106,
+      "step": 453900
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 1.1232366704238823e-05,
+      "loss": 2.4102,
+      "step": 454000
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 1.1223827577962223e-05,
+      "loss": 2.5527,
+      "step": 454100
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 1.1215288451685625e-05,
+      "loss": 2.2172,
+      "step": 454200
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 1.1206749325409024e-05,
+      "loss": 2.5036,
+      "step": 454300
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 1.1198210199132424e-05,
+      "loss": 2.4123,
+      "step": 454400
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 1.1189671072855826e-05,
+      "loss": 2.5316,
+      "step": 454500
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 1.1181131946579226e-05,
+      "loss": 2.126,
+      "step": 454600
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 1.1172592820302627e-05,
+      "loss": 2.4036,
+      "step": 454700
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 1.1164053694026027e-05,
+      "loss": 2.3227,
+      "step": 454800
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 1.1155514567749429e-05,
+      "loss": 2.5125,
+      "step": 454900
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 1.1146975441472829e-05,
+      "loss": 2.3822,
+      "step": 455000
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 1.113843631519623e-05,
+      "loss": 2.2891,
+      "step": 455100
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 1.112989718891963e-05,
+      "loss": 2.3755,
+      "step": 455200
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 1.1121358062643031e-05,
+      "loss": 2.3895,
+      "step": 455300
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 1.1112818936366431e-05,
+      "loss": 2.4506,
+      "step": 455400
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 1.1104279810089833e-05,
+      "loss": 2.3326,
+      "step": 455500
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 1.1095740683813233e-05,
+      "loss": 2.2889,
+      "step": 455600
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 1.1087201557536633e-05,
+      "loss": 2.3132,
+      "step": 455700
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 1.1078662431260034e-05,
+      "loss": 2.5263,
+      "step": 455800
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 1.1070123304983434e-05,
+      "loss": 2.2508,
+      "step": 455900
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 1.1061584178706836e-05,
+      "loss": 2.2724,
+      "step": 456000
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 1.1053045052430235e-05,
+      "loss": 2.3963,
+      "step": 456100
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 1.1044505926153637e-05,
+      "loss": 2.354,
+      "step": 456200
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 1.1035966799877037e-05,
+      "loss": 2.3623,
+      "step": 456300
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 1.1027427673600438e-05,
+      "loss": 2.3601,
+      "step": 456400
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 1.1018888547323838e-05,
+      "loss": 2.3038,
+      "step": 456500
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 1.101034942104724e-05,
+      "loss": 2.3352,
+      "step": 456600
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 1.100181029477064e-05,
+      "loss": 2.4089,
+      "step": 456700
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 1.0993271168494041e-05,
+      "loss": 2.3569,
+      "step": 456800
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 1.0984732042217441e-05,
+      "loss": 2.3535,
+      "step": 456900
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 1.0976192915940841e-05,
+      "loss": 2.3747,
+      "step": 457000
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 1.0967653789664242e-05,
+      "loss": 2.4065,
+      "step": 457100
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 1.0959114663387642e-05,
+      "loss": 2.524,
+      "step": 457200
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 1.0950575537111044e-05,
+      "loss": 2.3989,
+      "step": 457300
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 1.0942036410834444e-05,
+      "loss": 2.3519,
+      "step": 457400
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 1.0933497284557845e-05,
+      "loss": 2.4087,
+      "step": 457500
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 1.0924958158281245e-05,
+      "loss": 2.3264,
+      "step": 457600
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 1.0916419032004647e-05,
+      "loss": 2.223,
+      "step": 457700
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 1.0907879905728046e-05,
+      "loss": 2.3941,
+      "step": 457800
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 1.0899340779451448e-05,
+      "loss": 2.2319,
+      "step": 457900
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 1.0890801653174848e-05,
+      "loss": 2.2449,
+      "step": 458000
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 1.088226252689825e-05,
+      "loss": 2.316,
+      "step": 458100
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 1.087372340062165e-05,
+      "loss": 2.4809,
+      "step": 458200
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 1.0865184274345049e-05,
+      "loss": 2.2141,
+      "step": 458300
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 1.085664514806845e-05,
+      "loss": 2.2298,
+      "step": 458400
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 1.084810602179185e-05,
+      "loss": 2.1899,
+      "step": 458500
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 1.0839566895515252e-05,
+      "loss": 2.347,
+      "step": 458600
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 1.0831027769238652e-05,
+      "loss": 2.489,
+      "step": 458700
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 1.0822488642962053e-05,
+      "loss": 2.2976,
+      "step": 458800
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 1.0813949516685453e-05,
+      "loss": 2.3948,
+      "step": 458900
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 1.0805410390408855e-05,
+      "loss": 2.3881,
+      "step": 459000
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 1.0796871264132255e-05,
+      "loss": 2.2191,
+      "step": 459100
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 1.0788332137855656e-05,
+      "loss": 2.3434,
+      "step": 459200
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 1.0779793011579056e-05,
+      "loss": 2.4501,
+      "step": 459300
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 1.0771253885302458e-05,
+      "loss": 2.3837,
+      "step": 459400
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 1.0762714759025858e-05,
+      "loss": 2.3998,
+      "step": 459500
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 1.0754175632749257e-05,
+      "loss": 2.3488,
+      "step": 459600
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 1.0745636506472659e-05,
+      "loss": 2.3771,
+      "step": 459700
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 1.0737097380196059e-05,
+      "loss": 2.4188,
+      "step": 459800
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 1.072855825391946e-05,
+      "loss": 2.4259,
+      "step": 459900
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 1.072001912764286e-05,
+      "loss": 2.4023,
+      "step": 460000
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 1.0711480001366262e-05,
+      "loss": 2.4128,
+      "step": 460100
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 1.070294087508966e-05,
+      "loss": 2.3,
+      "step": 460200
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 1.0694401748813061e-05,
+      "loss": 2.2961,
+      "step": 460300
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 1.0685862622536461e-05,
+      "loss": 2.2691,
+      "step": 460400
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 1.0677323496259863e-05,
+      "loss": 2.3439,
+      "step": 460500
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 1.0668784369983263e-05,
+      "loss": 2.2766,
+      "step": 460600
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 1.0660245243706664e-05,
+      "loss": 2.2719,
+      "step": 460700
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 1.0651706117430064e-05,
+      "loss": 2.4551,
+      "step": 460800
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 1.0643166991153466e-05,
+      "loss": 2.3891,
+      "step": 460900
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 1.0634627864876866e-05,
+      "loss": 2.186,
+      "step": 461000
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 1.0626088738600267e-05,
+      "loss": 2.4296,
+      "step": 461100
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 1.0617549612323667e-05,
+      "loss": 2.646,
+      "step": 461200
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 1.0609010486047068e-05,
+      "loss": 2.2722,
+      "step": 461300
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 1.0600471359770468e-05,
+      "loss": 2.3699,
+      "step": 461400
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 1.0591932233493868e-05,
+      "loss": 2.5427,
+      "step": 461500
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 1.058339310721727e-05,
+      "loss": 2.4104,
+      "step": 461600
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 1.057485398094067e-05,
+      "loss": 2.2132,
+      "step": 461700
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 1.0566314854664071e-05,
+      "loss": 2.2677,
+      "step": 461800
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 1.0557775728387471e-05,
+      "loss": 2.5672,
+      "step": 461900
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 1.0549236602110873e-05,
+      "loss": 2.3104,
+      "step": 462000
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 1.0540697475834272e-05,
+      "loss": 2.3313,
+      "step": 462100
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 1.0532158349557674e-05,
+      "loss": 2.2646,
+      "step": 462200
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 1.0523619223281074e-05,
+      "loss": 2.3769,
+      "step": 462300
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 1.0515080097004475e-05,
+      "loss": 2.4205,
+      "step": 462400
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 1.0506540970727875e-05,
+      "loss": 2.4462,
+      "step": 462500
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 1.0498001844451277e-05,
+      "loss": 2.3421,
+      "step": 462600
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 1.0489462718174677e-05,
+      "loss": 2.3728,
+      "step": 462700
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 1.0480923591898076e-05,
+      "loss": 2.2476,
+      "step": 462800
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 1.0472384465621478e-05,
+      "loss": 2.2497,
+      "step": 462900
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 1.0463845339344878e-05,
+      "loss": 2.5004,
+      "step": 463000
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 1.045530621306828e-05,
+      "loss": 2.323,
+      "step": 463100
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 1.044676708679168e-05,
+      "loss": 2.2561,
+      "step": 463200
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 1.043822796051508e-05,
+      "loss": 2.4128,
+      "step": 463300
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 1.042968883423848e-05,
+      "loss": 2.1714,
+      "step": 463400
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 1.0421149707961882e-05,
+      "loss": 2.475,
+      "step": 463500
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 1.0412610581685282e-05,
+      "loss": 2.2916,
+      "step": 463600
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 1.0404071455408684e-05,
+      "loss": 2.1948,
+      "step": 463700
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 1.0395532329132083e-05,
+      "loss": 2.4248,
+      "step": 463800
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 1.0386993202855485e-05,
+      "loss": 2.5744,
+      "step": 463900
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 1.0378454076578885e-05,
+      "loss": 2.2751,
+      "step": 464000
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 1.0369914950302285e-05,
+      "loss": 2.5041,
+      "step": 464100
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 1.0361375824025686e-05,
+      "loss": 2.4111,
+      "step": 464200
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 1.0352836697749086e-05,
+      "loss": 2.3358,
+      "step": 464300
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 1.0344297571472488e-05,
+      "loss": 2.3797,
+      "step": 464400
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 1.0335758445195888e-05,
+      "loss": 2.3143,
+      "step": 464500
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 1.0327219318919289e-05,
+      "loss": 2.4891,
+      "step": 464600
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 1.0318680192642689e-05,
+      "loss": 2.315,
+      "step": 464700
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 1.031014106636609e-05,
+      "loss": 2.3748,
+      "step": 464800
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 1.030160194008949e-05,
+      "loss": 2.3514,
+      "step": 464900
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 1.0293062813812892e-05,
+      "loss": 2.3933,
+      "step": 465000
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 1.0284523687536292e-05,
+      "loss": 2.5657,
+      "step": 465100
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 1.0275984561259693e-05,
+      "loss": 2.3748,
+      "step": 465200
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 1.0267445434983093e-05,
+      "loss": 2.3294,
+      "step": 465300
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 1.0258906308706493e-05,
+      "loss": 2.3038,
+      "step": 465400
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 1.0250367182429895e-05,
+      "loss": 2.3801,
+      "step": 465500
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 1.0241828056153294e-05,
+      "loss": 2.2168,
+      "step": 465600
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 1.0233288929876696e-05,
+      "loss": 2.2804,
+      "step": 465700
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 1.0224749803600096e-05,
+      "loss": 2.3886,
+      "step": 465800
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 1.0216210677323497e-05,
+      "loss": 2.5101,
+      "step": 465900
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 1.0207671551046897e-05,
+      "loss": 2.477,
+      "step": 466000
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 1.0199132424770299e-05,
+      "loss": 2.2089,
+      "step": 466100
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 1.0190593298493699e-05,
+      "loss": 2.4535,
+      "step": 466200
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 1.01820541722171e-05,
+      "loss": 2.2373,
+      "step": 466300
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 1.01735150459405e-05,
+      "loss": 2.3935,
+      "step": 466400
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 1.0164975919663902e-05,
+      "loss": 2.3327,
+      "step": 466500
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 1.0156436793387301e-05,
+      "loss": 2.3853,
+      "step": 466600
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 1.0147897667110701e-05,
+      "loss": 2.2725,
+      "step": 466700
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 1.0139358540834103e-05,
+      "loss": 2.3891,
+      "step": 466800
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 1.0130819414557503e-05,
+      "loss": 2.3847,
+      "step": 466900
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 1.0122280288280904e-05,
+      "loss": 2.6729,
+      "step": 467000
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 1.0113741162004304e-05,
+      "loss": 2.4303,
+      "step": 467100
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 1.0105202035727706e-05,
+      "loss": 2.6356,
+      "step": 467200
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 1.0096662909451105e-05,
+      "loss": 2.3694,
+      "step": 467300
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 1.0088123783174507e-05,
+      "loss": 2.2721,
+      "step": 467400
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 1.0079584656897907e-05,
+      "loss": 2.4132,
+      "step": 467500
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 1.0071045530621308e-05,
+      "loss": 2.3676,
+      "step": 467600
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 1.0062506404344708e-05,
+      "loss": 2.2973,
+      "step": 467700
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 1.005396727806811e-05,
+      "loss": 2.2985,
+      "step": 467800
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 1.004542815179151e-05,
+      "loss": 2.3777,
+      "step": 467900
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 1.003688902551491e-05,
+      "loss": 2.3674,
+      "step": 468000
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 1.0028349899238311e-05,
+      "loss": 2.3113,
+      "step": 468100
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 1.0019810772961711e-05,
+      "loss": 2.3669,
+      "step": 468200
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 1.0011271646685112e-05,
+      "loss": 2.363,
+      "step": 468300
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 1.0002732520408512e-05,
+      "loss": 2.3587,
+      "step": 468400
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 9.994193394131914e-06,
+      "loss": 2.2949,
+      "step": 468500
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 9.985654267855314e-06,
+      "loss": 2.3748,
+      "step": 468600
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 9.977115141578715e-06,
+      "loss": 2.3995,
+      "step": 468700
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 9.968576015302115e-06,
+      "loss": 2.3465,
+      "step": 468800
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 9.960036889025517e-06,
+      "loss": 2.2719,
+      "step": 468900
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 9.951497762748915e-06,
+      "loss": 2.1681,
+      "step": 469000
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 9.942958636472316e-06,
+      "loss": 2.2478,
+      "step": 469100
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 9.934419510195716e-06,
+      "loss": 2.3767,
+      "step": 469200
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 9.925880383919118e-06,
+      "loss": 2.3495,
+      "step": 469300
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 9.917341257642518e-06,
+      "loss": 2.4604,
+      "step": 469400
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 9.90880213136592e-06,
+      "loss": 2.281,
+      "step": 469500
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 9.900263005089319e-06,
+      "loss": 2.2163,
+      "step": 469600
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 9.89172387881272e-06,
+      "loss": 2.2095,
+      "step": 469700
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 9.88318475253612e-06,
+      "loss": 2.3108,
+      "step": 469800
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 9.87464562625952e-06,
+      "loss": 2.2822,
+      "step": 469900
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 9.866106499982922e-06,
+      "loss": 2.4625,
+      "step": 470000
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 9.857567373706322e-06,
+      "loss": 2.291,
+      "step": 470100
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 9.849028247429723e-06,
+      "loss": 2.3437,
+      "step": 470200
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 9.840489121153123e-06,
+      "loss": 2.3757,
+      "step": 470300
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 9.831949994876525e-06,
+      "loss": 2.4156,
+      "step": 470400
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 9.823410868599925e-06,
+      "loss": 2.3461,
+      "step": 470500
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 9.814871742323326e-06,
+      "loss": 2.303,
+      "step": 470600
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 9.806332616046726e-06,
+      "loss": 2.3965,
+      "step": 470700
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 9.797793489770127e-06,
+      "loss": 2.4787,
+      "step": 470800
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 9.789254363493527e-06,
+      "loss": 2.2388,
+      "step": 470900
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 9.780715237216929e-06,
+      "loss": 2.1807,
+      "step": 471000
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 9.772176110940329e-06,
+      "loss": 2.3898,
+      "step": 471100
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 9.763636984663729e-06,
+      "loss": 2.3237,
+      "step": 471200
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 9.75509785838713e-06,
+      "loss": 2.3549,
+      "step": 471300
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 9.74655873211053e-06,
+      "loss": 2.4313,
+      "step": 471400
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 9.738019605833932e-06,
+      "loss": 2.3672,
+      "step": 471500
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 9.729480479557331e-06,
+      "loss": 2.4768,
+      "step": 471600
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 9.720941353280733e-06,
+      "loss": 2.4814,
+      "step": 471700
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 9.712402227004133e-06,
+      "loss": 2.2807,
+      "step": 471800
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 9.703863100727534e-06,
+      "loss": 2.4814,
+      "step": 471900
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 9.695323974450934e-06,
+      "loss": 2.1597,
+      "step": 472000
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 9.686784848174336e-06,
+      "loss": 2.5317,
+      "step": 472100
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 9.678245721897736e-06,
+      "loss": 2.3068,
+      "step": 472200
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 9.669706595621137e-06,
+      "loss": 2.5735,
+      "step": 472300
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 9.661167469344537e-06,
+      "loss": 2.3386,
+      "step": 472400
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 9.652628343067937e-06,
+      "loss": 2.1372,
+      "step": 472500
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 9.644089216791338e-06,
+      "loss": 2.3445,
+      "step": 472600
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 9.635550090514738e-06,
+      "loss": 2.3872,
+      "step": 472700
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 9.62701096423814e-06,
+      "loss": 2.3145,
+      "step": 472800
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 9.61847183796154e-06,
+      "loss": 2.3415,
+      "step": 472900
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 9.609932711684941e-06,
+      "loss": 2.4201,
+      "step": 473000
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 9.601393585408341e-06,
+      "loss": 2.4163,
+      "step": 473100
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 9.592854459131743e-06,
+      "loss": 2.2282,
+      "step": 473200
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 9.584315332855142e-06,
+      "loss": 2.3337,
+      "step": 473300
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 9.575776206578544e-06,
+      "loss": 2.5322,
+      "step": 473400
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 9.567237080301944e-06,
+      "loss": 2.2854,
+      "step": 473500
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 9.558697954025345e-06,
+      "loss": 2.3463,
+      "step": 473600
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 9.550158827748745e-06,
+      "loss": 2.1978,
+      "step": 473700
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 9.541619701472145e-06,
+      "loss": 2.4242,
+      "step": 473800
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 9.533080575195547e-06,
+      "loss": 2.3271,
+      "step": 473900
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 9.524541448918947e-06,
+      "loss": 2.3342,
+      "step": 474000
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 9.516002322642348e-06,
+      "loss": 2.4026,
+      "step": 474100
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 9.507463196365748e-06,
+      "loss": 2.404,
+      "step": 474200
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 9.49892407008915e-06,
+      "loss": 2.3344,
+      "step": 474300
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 9.49038494381255e-06,
+      "loss": 2.4367,
+      "step": 474400
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 9.481845817535951e-06,
+      "loss": 2.3966,
+      "step": 474500
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 9.47330669125935e-06,
+      "loss": 2.3476,
+      "step": 474600
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 9.464767564982752e-06,
+      "loss": 2.2906,
+      "step": 474700
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 9.456228438706152e-06,
+      "loss": 2.5284,
+      "step": 474800
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 9.447689312429554e-06,
+      "loss": 2.3994,
+      "step": 474900
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 9.439150186152954e-06,
+      "loss": 2.2996,
+      "step": 475000
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 9.430611059876353e-06,
+      "loss": 2.3269,
+      "step": 475100
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 9.422071933599755e-06,
+      "loss": 2.3843,
+      "step": 475200
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 9.413532807323155e-06,
+      "loss": 2.4582,
+      "step": 475300
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 9.404993681046556e-06,
+      "loss": 2.473,
+      "step": 475400
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 9.396454554769956e-06,
+      "loss": 2.2896,
+      "step": 475500
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 9.387915428493358e-06,
+      "loss": 2.3914,
+      "step": 475600
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 9.379376302216758e-06,
+      "loss": 2.3402,
+      "step": 475700
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 9.370837175940159e-06,
+      "loss": 2.4208,
+      "step": 475800
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 9.362298049663559e-06,
+      "loss": 2.4028,
+      "step": 475900
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 9.35375892338696e-06,
+      "loss": 2.3018,
+      "step": 476000
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 9.34521979711036e-06,
+      "loss": 2.3314,
+      "step": 476100
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 9.336680670833762e-06,
+      "loss": 2.2372,
+      "step": 476200
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 9.328141544557162e-06,
+      "loss": 2.267,
+      "step": 476300
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 9.319602418280562e-06,
+      "loss": 2.2984,
+      "step": 476400
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 9.311063292003963e-06,
+      "loss": 2.2714,
+      "step": 476500
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 9.302524165727363e-06,
+      "loss": 2.4087,
+      "step": 476600
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 9.293985039450765e-06,
+      "loss": 2.3647,
+      "step": 476700
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 9.285445913174164e-06,
+      "loss": 2.3222,
+      "step": 476800
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 9.276906786897566e-06,
+      "loss": 2.3455,
+      "step": 476900
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 9.268367660620966e-06,
+      "loss": 2.2878,
+      "step": 477000
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 9.259828534344367e-06,
+      "loss": 2.2694,
+      "step": 477100
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 9.251289408067767e-06,
+      "loss": 2.3443,
+      "step": 477200
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 9.242750281791169e-06,
+      "loss": 2.2092,
+      "step": 477300
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 9.234211155514569e-06,
+      "loss": 2.336,
+      "step": 477400
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 9.22567202923797e-06,
+      "loss": 2.4295,
+      "step": 477500
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 9.21713290296137e-06,
+      "loss": 2.323,
+      "step": 477600
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 9.20859377668477e-06,
+      "loss": 2.4561,
+      "step": 477700
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 9.20005465040817e-06,
+      "loss": 2.2738,
+      "step": 477800
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 9.191515524131571e-06,
+      "loss": 2.2431,
+      "step": 477900
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 9.182976397854971e-06,
+      "loss": 2.2159,
+      "step": 478000
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 9.174437271578373e-06,
+      "loss": 2.4646,
+      "step": 478100
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 9.165898145301773e-06,
+      "loss": 2.6006,
+      "step": 478200
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 9.157359019025172e-06,
+      "loss": 2.4189,
+      "step": 478300
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 9.148819892748574e-06,
+      "loss": 2.4249,
+      "step": 478400
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 9.140280766471974e-06,
+      "loss": 2.441,
+      "step": 478500
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 9.131741640195375e-06,
+      "loss": 2.3329,
+      "step": 478600
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 9.123202513918775e-06,
+      "loss": 2.3692,
+      "step": 478700
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 9.114663387642177e-06,
+      "loss": 2.2183,
+      "step": 478800
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 9.106124261365577e-06,
+      "loss": 2.2556,
+      "step": 478900
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 9.097585135088978e-06,
+      "loss": 2.2597,
+      "step": 479000
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 9.089046008812378e-06,
+      "loss": 2.2906,
+      "step": 479100
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 9.08050688253578e-06,
+      "loss": 2.4126,
+      "step": 479200
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 9.07196775625918e-06,
+      "loss": 2.4707,
+      "step": 479300
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 9.063428629982581e-06,
+      "loss": 2.2973,
+      "step": 479400
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 9.05488950370598e-06,
+      "loss": 2.3448,
+      "step": 479500
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 9.04635037742938e-06,
+      "loss": 2.2431,
+      "step": 479600
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 9.037811251152782e-06,
+      "loss": 2.1868,
+      "step": 479700
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 9.029272124876182e-06,
+      "loss": 2.4682,
+      "step": 479800
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 9.020732998599584e-06,
+      "loss": 2.2371,
+      "step": 479900
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 9.012193872322983e-06,
+      "loss": 2.3678,
+      "step": 480000
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 9.003654746046385e-06,
+      "loss": 2.2952,
+      "step": 480100
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 8.995115619769785e-06,
+      "loss": 2.3105,
+      "step": 480200
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 8.986576493493186e-06,
+      "loss": 2.2523,
+      "step": 480300
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 8.978037367216586e-06,
+      "loss": 2.2952,
+      "step": 480400
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 8.969498240939988e-06,
+      "loss": 2.1216,
+      "step": 480500
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 8.960959114663388e-06,
+      "loss": 2.2869,
+      "step": 480600
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 8.95241998838679e-06,
+      "loss": 2.2056,
+      "step": 480700
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 8.943880862110189e-06,
+      "loss": 2.4429,
+      "step": 480800
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 8.935341735833589e-06,
+      "loss": 2.4611,
+      "step": 480900
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 8.92680260955699e-06,
+      "loss": 2.3674,
+      "step": 481000
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 8.91826348328039e-06,
+      "loss": 2.2883,
+      "step": 481100
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 8.909724357003792e-06,
+      "loss": 2.3646,
+      "step": 481200
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 8.901185230727192e-06,
+      "loss": 2.1688,
+      "step": 481300
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 8.892646104450593e-06,
+      "loss": 2.4568,
+      "step": 481400
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 8.884106978173993e-06,
+      "loss": 2.3121,
+      "step": 481500
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 8.875567851897395e-06,
+      "loss": 2.2561,
+      "step": 481600
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 8.867028725620795e-06,
+      "loss": 2.441,
+      "step": 481700
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 8.858489599344196e-06,
+      "loss": 2.279,
+      "step": 481800
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 8.849950473067596e-06,
+      "loss": 2.2011,
+      "step": 481900
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 8.841411346790998e-06,
+      "loss": 2.4275,
+      "step": 482000
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 8.832872220514397e-06,
+      "loss": 2.4149,
+      "step": 482100
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 8.824333094237797e-06,
+      "loss": 2.3443,
+      "step": 482200
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 8.815793967961199e-06,
+      "loss": 2.4427,
+      "step": 482300
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 8.807254841684599e-06,
+      "loss": 2.4005,
+      "step": 482400
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 8.798715715408e-06,
+      "loss": 2.2363,
+      "step": 482500
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 8.7901765891314e-06,
+      "loss": 2.3844,
+      "step": 482600
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 8.781637462854802e-06,
+      "loss": 2.2652,
+      "step": 482700
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 8.773098336578201e-06,
+      "loss": 2.4342,
+      "step": 482800
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 8.764559210301603e-06,
+      "loss": 2.313,
+      "step": 482900
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 8.756020084025003e-06,
+      "loss": 2.3698,
+      "step": 483000
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 8.747480957748404e-06,
+      "loss": 2.2924,
+      "step": 483100
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 8.738941831471804e-06,
+      "loss": 2.3884,
+      "step": 483200
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 8.730402705195206e-06,
+      "loss": 2.3257,
+      "step": 483300
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 8.721863578918606e-06,
+      "loss": 2.315,
+      "step": 483400
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 8.713324452642005e-06,
+      "loss": 2.4,
+      "step": 483500
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 8.704785326365407e-06,
+      "loss": 2.4332,
+      "step": 483600
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 8.696246200088807e-06,
+      "loss": 2.4837,
+      "step": 483700
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 8.687707073812208e-06,
+      "loss": 2.2764,
+      "step": 483800
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 8.679167947535608e-06,
+      "loss": 2.3559,
+      "step": 483900
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 8.67062882125901e-06,
+      "loss": 2.4048,
+      "step": 484000
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 8.66208969498241e-06,
+      "loss": 2.2568,
+      "step": 484100
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 8.653550568705811e-06,
+      "loss": 2.1394,
+      "step": 484200
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 8.645011442429211e-06,
+      "loss": 2.3126,
+      "step": 484300
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 8.636472316152613e-06,
+      "loss": 2.365,
+      "step": 484400
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 8.627933189876013e-06,
+      "loss": 2.3888,
+      "step": 484500
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 8.619394063599414e-06,
+      "loss": 2.309,
+      "step": 484600
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 8.610854937322814e-06,
+      "loss": 2.3908,
+      "step": 484700
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 8.602315811046214e-06,
+      "loss": 2.2101,
+      "step": 484800
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 8.593776684769615e-06,
+      "loss": 2.2009,
+      "step": 484900
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 8.585237558493015e-06,
+      "loss": 2.4389,
+      "step": 485000
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 8.576698432216417e-06,
+      "loss": 2.4014,
+      "step": 485100
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 8.568159305939817e-06,
+      "loss": 2.2933,
+      "step": 485200
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 8.559620179663218e-06,
+      "loss": 2.2193,
+      "step": 485300
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 8.551081053386618e-06,
+      "loss": 2.2519,
+      "step": 485400
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 8.54254192711002e-06,
+      "loss": 2.4025,
+      "step": 485500
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 8.53400280083342e-06,
+      "loss": 2.4623,
+      "step": 485600
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 8.525463674556821e-06,
+      "loss": 2.3038,
+      "step": 485700
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 8.51692454828022e-06,
+      "loss": 2.3208,
+      "step": 485800
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 8.508385422003622e-06,
+      "loss": 2.3442,
+      "step": 485900
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 8.499846295727022e-06,
+      "loss": 2.2255,
+      "step": 486000
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 8.491307169450422e-06,
+      "loss": 2.4155,
+      "step": 486100
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 8.482768043173824e-06,
+      "loss": 2.3096,
+      "step": 486200
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 8.474228916897223e-06,
+      "loss": 2.3084,
+      "step": 486300
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 8.465689790620625e-06,
+      "loss": 2.3511,
+      "step": 486400
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 8.457150664344025e-06,
+      "loss": 2.2618,
+      "step": 486500
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 8.448611538067425e-06,
+      "loss": 2.4265,
+      "step": 486600
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 8.440072411790825e-06,
+      "loss": 2.4689,
+      "step": 486700
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 8.431533285514226e-06,
+      "loss": 2.4635,
+      "step": 486800
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 8.422994159237626e-06,
+      "loss": 2.3413,
+      "step": 486900
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 8.414455032961027e-06,
+      "loss": 2.4025,
+      "step": 487000
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 8.405915906684427e-06,
+      "loss": 2.3439,
+      "step": 487100
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 8.397376780407829e-06,
+      "loss": 2.1309,
+      "step": 487200
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 8.388837654131229e-06,
+      "loss": 2.2746,
+      "step": 487300
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 8.38029852785463e-06,
+      "loss": 2.2547,
+      "step": 487400
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 8.37175940157803e-06,
+      "loss": 2.2953,
+      "step": 487500
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 8.363220275301432e-06,
+      "loss": 2.2432,
+      "step": 487600
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 8.354681149024832e-06,
+      "loss": 2.3293,
+      "step": 487700
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 8.346142022748233e-06,
+      "loss": 2.4646,
+      "step": 487800
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 8.337602896471633e-06,
+      "loss": 2.3454,
+      "step": 487900
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 8.329063770195033e-06,
+      "loss": 2.3126,
+      "step": 488000
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 8.320524643918434e-06,
+      "loss": 2.3632,
+      "step": 488100
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 8.311985517641834e-06,
+      "loss": 2.27,
+      "step": 488200
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 8.303446391365236e-06,
+      "loss": 2.4339,
+      "step": 488300
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 8.294907265088636e-06,
+      "loss": 2.3237,
+      "step": 488400
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 8.286368138812037e-06,
+      "loss": 2.4811,
+      "step": 488500
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 8.277829012535437e-06,
+      "loss": 2.3788,
+      "step": 488600
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 8.269289886258839e-06,
+      "loss": 2.3031,
+      "step": 488700
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 8.260750759982238e-06,
+      "loss": 2.4946,
+      "step": 488800
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 8.25221163370564e-06,
+      "loss": 2.386,
+      "step": 488900
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 8.24367250742904e-06,
+      "loss": 2.3604,
+      "step": 489000
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 8.235133381152441e-06,
+      "loss": 2.3326,
+      "step": 489100
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 8.226594254875841e-06,
+      "loss": 2.3458,
+      "step": 489200
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 8.218055128599241e-06,
+      "loss": 2.4537,
+      "step": 489300
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 8.209516002322643e-06,
+      "loss": 2.445,
+      "step": 489400
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 8.200976876046042e-06,
+      "loss": 2.3172,
+      "step": 489500
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 8.192437749769444e-06,
+      "loss": 2.5434,
+      "step": 489600
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 8.183898623492844e-06,
+      "loss": 2.2746,
+      "step": 489700
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 8.175359497216245e-06,
+      "loss": 2.5382,
+      "step": 489800
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 8.166820370939645e-06,
+      "loss": 2.4194,
+      "step": 489900
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 8.158281244663047e-06,
+      "loss": 2.4621,
+      "step": 490000
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 8.149742118386447e-06,
+      "loss": 2.1997,
+      "step": 490100
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 8.141202992109848e-06,
+      "loss": 2.3675,
+      "step": 490200
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 8.132663865833248e-06,
+      "loss": 2.3056,
+      "step": 490300
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 8.12412473955665e-06,
+      "loss": 2.359,
+      "step": 490400
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 8.11558561328005e-06,
+      "loss": 2.2198,
+      "step": 490500
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 8.10704648700345e-06,
+      "loss": 2.2349,
+      "step": 490600
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 8.098507360726851e-06,
+      "loss": 2.4303,
+      "step": 490700
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 8.08996823445025e-06,
+      "loss": 2.4193,
+      "step": 490800
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 8.081429108173652e-06,
+      "loss": 2.2192,
+      "step": 490900
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 8.072889981897052e-06,
+      "loss": 2.3109,
+      "step": 491000
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 8.064350855620454e-06,
+      "loss": 2.3217,
+      "step": 491100
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 8.055811729343854e-06,
+      "loss": 2.2739,
+      "step": 491200
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 8.047272603067255e-06,
+      "loss": 2.4991,
+      "step": 491300
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 8.038733476790655e-06,
+      "loss": 2.234,
+      "step": 491400
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 8.030194350514057e-06,
+      "loss": 2.27,
+      "step": 491500
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 8.021655224237456e-06,
+      "loss": 2.4502,
+      "step": 491600
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 8.013116097960858e-06,
+      "loss": 2.3828,
+      "step": 491700
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 8.004576971684258e-06,
+      "loss": 2.3753,
+      "step": 491800
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 7.996037845407658e-06,
+      "loss": 2.5328,
+      "step": 491900
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 7.98749871913106e-06,
+      "loss": 2.0865,
+      "step": 492000
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 7.978959592854459e-06,
+      "loss": 2.3216,
+      "step": 492100
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 7.97042046657786e-06,
+      "loss": 2.1444,
+      "step": 492200
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 7.96188134030126e-06,
+      "loss": 2.1694,
+      "step": 492300
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 7.953342214024662e-06,
+      "loss": 2.258,
+      "step": 492400
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 7.944803087748062e-06,
+      "loss": 2.4035,
+      "step": 492500
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 7.936263961471463e-06,
+      "loss": 2.3153,
+      "step": 492600
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 7.927724835194863e-06,
+      "loss": 2.4465,
+      "step": 492700
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 7.919185708918265e-06,
+      "loss": 2.341,
+      "step": 492800
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 7.910646582641665e-06,
+      "loss": 2.3799,
+      "step": 492900
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 7.902107456365066e-06,
+      "loss": 2.2845,
+      "step": 493000
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 7.893568330088466e-06,
+      "loss": 2.0654,
+      "step": 493100
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 7.885029203811866e-06,
+      "loss": 2.4565,
+      "step": 493200
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 7.876490077535267e-06,
+      "loss": 2.3472,
+      "step": 493300
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 7.867950951258667e-06,
+      "loss": 2.3162,
+      "step": 493400
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 7.859411824982069e-06,
+      "loss": 2.3647,
+      "step": 493500
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 7.850872698705469e-06,
+      "loss": 2.4958,
+      "step": 493600
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 7.84233357242887e-06,
+      "loss": 2.2391,
+      "step": 493700
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 7.83379444615227e-06,
+      "loss": 2.2725,
+      "step": 493800
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 7.825255319875672e-06,
+      "loss": 2.3836,
+      "step": 493900
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 7.816716193599071e-06,
+      "loss": 2.3073,
+      "step": 494000
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 7.808177067322473e-06,
+      "loss": 2.152,
+      "step": 494100
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 7.799637941045873e-06,
+      "loss": 2.2982,
+      "step": 494200
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 7.791098814769274e-06,
+      "loss": 2.4182,
+      "step": 494300
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 7.782559688492674e-06,
+      "loss": 2.1968,
+      "step": 494400
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 7.774020562216074e-06,
+      "loss": 2.4608,
+      "step": 494500
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 7.765481435939476e-06,
+      "loss": 2.3586,
+      "step": 494600
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 7.756942309662876e-06,
+      "loss": 2.397,
+      "step": 494700
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 7.748403183386277e-06,
+      "loss": 2.3283,
+      "step": 494800
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 7.739864057109677e-06,
+      "loss": 2.4132,
+      "step": 494900
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 7.731324930833079e-06,
+      "loss": 2.4419,
+      "step": 495000
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 7.722785804556478e-06,
+      "loss": 2.1422,
+      "step": 495100
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 7.71424667827988e-06,
+      "loss": 2.289,
+      "step": 495200
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 7.705707552003278e-06,
+      "loss": 2.4988,
+      "step": 495300
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 7.69716842572668e-06,
+      "loss": 2.2937,
+      "step": 495400
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 7.68862929945008e-06,
+      "loss": 2.4985,
+      "step": 495500
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 7.680090173173481e-06,
+      "loss": 2.3567,
+      "step": 495600
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 7.671551046896881e-06,
+      "loss": 2.1894,
+      "step": 495700
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 7.663011920620282e-06,
+      "loss": 2.232,
+      "step": 495800
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 7.654472794343682e-06,
+      "loss": 2.4518,
+      "step": 495900
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 7.645933668067084e-06,
+      "loss": 2.3342,
+      "step": 496000
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 7.637394541790484e-06,
+      "loss": 2.3339,
+      "step": 496100
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 7.628855415513884e-06,
+      "loss": 2.4171,
+      "step": 496200
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 7.620316289237285e-06,
+      "loss": 2.3569,
+      "step": 496300
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 7.611777162960686e-06,
+      "loss": 2.2601,
+      "step": 496400
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 7.6032380366840865e-06,
+      "loss": 2.3,
+      "step": 496500
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 7.594698910407487e-06,
+      "loss": 2.1885,
+      "step": 496600
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 7.586159784130888e-06,
+      "loss": 2.4208,
+      "step": 496700
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 7.577620657854289e-06,
+      "loss": 2.2028,
+      "step": 496800
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 7.569081531577689e-06,
+      "loss": 2.1802,
+      "step": 496900
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 7.560542405301089e-06,
+      "loss": 2.4717,
+      "step": 497000
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 7.55200327902449e-06,
+      "loss": 2.5148,
+      "step": 497100
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 7.5434641527478905e-06,
+      "loss": 2.3434,
+      "step": 497200
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 7.534925026471291e-06,
+      "loss": 2.2749,
+      "step": 497300
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 7.526385900194692e-06,
+      "loss": 2.4476,
+      "step": 497400
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 7.517846773918093e-06,
+      "loss": 2.4586,
+      "step": 497500
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 7.509307647641493e-06,
+      "loss": 2.3242,
+      "step": 497600
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 7.500768521364894e-06,
+      "loss": 2.2898,
+      "step": 497700
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 7.492229395088295e-06,
+      "loss": 2.1807,
+      "step": 497800
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 7.4836902688116955e-06,
+      "loss": 2.3854,
+      "step": 497900
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 7.475151142535096e-06,
+      "loss": 2.3335,
+      "step": 498000
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 7.466612016258497e-06,
+      "loss": 2.4008,
+      "step": 498100
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 7.4580728899818976e-06,
+      "loss": 2.3006,
+      "step": 498200
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 7.449533763705297e-06,
+      "loss": 2.4518,
+      "step": 498300
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 7.440994637428698e-06,
+      "loss": 2.2865,
+      "step": 498400
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 7.432455511152099e-06,
+      "loss": 2.4045,
+      "step": 498500
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 7.4239163848754995e-06,
+      "loss": 2.2694,
+      "step": 498600
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 7.4153772585989e-06,
+      "loss": 2.3668,
+      "step": 498700
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 7.406838132322301e-06,
+      "loss": 2.3019,
+      "step": 498800
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 7.398299006045702e-06,
+      "loss": 2.3908,
+      "step": 498900
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 7.389759879769102e-06,
+      "loss": 2.3854,
+      "step": 499000
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 7.381220753492503e-06,
+      "loss": 2.4618,
+      "step": 499100
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 7.372681627215904e-06,
+      "loss": 2.4981,
+      "step": 499200
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 7.364142500939304e-06,
+      "loss": 2.2234,
+      "step": 499300
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 7.355603374662705e-06,
+      "loss": 2.2864,
+      "step": 499400
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 7.347064248386106e-06,
+      "loss": 2.3078,
+      "step": 499500
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 7.338525122109506e-06,
+      "loss": 2.311,
+      "step": 499600
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 7.329985995832906e-06,
+      "loss": 2.3657,
+      "step": 499700
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 7.321446869556307e-06,
+      "loss": 2.4746,
+      "step": 499800
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 7.312907743279708e-06,
+      "loss": 2.2827,
+      "step": 499900
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 7.3043686170031085e-06,
+      "loss": 2.4597,
+      "step": 500000
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 7.295829490726509e-06,
+      "loss": 2.1647,
+      "step": 500100
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 7.28729036444991e-06,
+      "loss": 2.3248,
+      "step": 500200
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 7.278751238173311e-06,
+      "loss": 2.2732,
+      "step": 500300
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 7.270212111896711e-06,
+      "loss": 2.4343,
+      "step": 500400
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 7.261672985620112e-06,
+      "loss": 2.388,
+      "step": 500500
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 7.253133859343513e-06,
+      "loss": 2.2686,
+      "step": 500600
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 7.244594733066913e-06,
+      "loss": 2.2298,
+      "step": 500700
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 7.236055606790314e-06,
+      "loss": 2.3278,
+      "step": 500800
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 7.227516480513715e-06,
+      "loss": 2.3579,
+      "step": 500900
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 7.218977354237115e-06,
+      "loss": 2.3337,
+      "step": 501000
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 7.210438227960515e-06,
+      "loss": 2.3084,
+      "step": 501100
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 7.201899101683916e-06,
+      "loss": 2.3216,
+      "step": 501200
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 7.193359975407317e-06,
+      "loss": 2.381,
+      "step": 501300
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 7.1848208491307175e-06,
+      "loss": 2.2307,
+      "step": 501400
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 7.176281722854118e-06,
+      "loss": 2.3823,
+      "step": 501500
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 7.167742596577519e-06,
+      "loss": 2.4136,
+      "step": 501600
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 7.1592034703009196e-06,
+      "loss": 2.4216,
+      "step": 501700
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 7.15066434402432e-06,
+      "loss": 2.3522,
+      "step": 501800
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 7.142125217747721e-06,
+      "loss": 2.1623,
+      "step": 501900
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 7.133586091471122e-06,
+      "loss": 2.313,
+      "step": 502000
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 7.125046965194522e-06,
+      "loss": 2.2511,
+      "step": 502100
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 7.116507838917923e-06,
+      "loss": 2.284,
+      "step": 502200
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 7.107968712641323e-06,
+      "loss": 2.3836,
+      "step": 502300
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 7.099429586364724e-06,
+      "loss": 2.1526,
+      "step": 502400
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 7.090890460088124e-06,
+      "loss": 2.3357,
+      "step": 502500
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 7.082351333811525e-06,
+      "loss": 2.2915,
+      "step": 502600
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 7.073812207534926e-06,
+      "loss": 2.4807,
+      "step": 502700
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 7.065273081258326e-06,
+      "loss": 2.4016,
+      "step": 502800
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 7.056733954981727e-06,
+      "loss": 2.3555,
+      "step": 502900
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 7.048194828705128e-06,
+      "loss": 2.3149,
+      "step": 503000
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 7.0396557024285285e-06,
+      "loss": 2.3492,
+      "step": 503100
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 7.031116576151929e-06,
+      "loss": 2.3255,
+      "step": 503200
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 7.02257744987533e-06,
+      "loss": 2.3781,
+      "step": 503300
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 7.014038323598731e-06,
+      "loss": 2.3089,
+      "step": 503400
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 7.005499197322131e-06,
+      "loss": 2.3618,
+      "step": 503500
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 6.996960071045531e-06,
+      "loss": 2.2661,
+      "step": 503600
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 6.988420944768932e-06,
+      "loss": 2.4096,
+      "step": 503700
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 6.979881818492333e-06,
+      "loss": 2.2042,
+      "step": 503800
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 6.971342692215733e-06,
+      "loss": 2.3418,
+      "step": 503900
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 6.962803565939134e-06,
+      "loss": 2.2529,
+      "step": 504000
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 6.954264439662533e-06,
+      "loss": 2.3809,
+      "step": 504100
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 6.945725313385934e-06,
+      "loss": 2.3171,
+      "step": 504200
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 6.937186187109334e-06,
+      "loss": 2.3075,
+      "step": 504300
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 6.928647060832735e-06,
+      "loss": 2.5179,
+      "step": 504400
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 6.920107934556136e-06,
+      "loss": 2.3905,
+      "step": 504500
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 6.9115688082795365e-06,
+      "loss": 2.2362,
+      "step": 504600
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 6.903029682002937e-06,
+      "loss": 2.4593,
+      "step": 504700
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 6.894490555726338e-06,
+      "loss": 2.3502,
+      "step": 504800
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 6.885951429449739e-06,
+      "loss": 2.4884,
+      "step": 504900
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 6.877412303173139e-06,
+      "loss": 2.3433,
+      "step": 505000
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 6.86887317689654e-06,
+      "loss": 2.3418,
+      "step": 505100
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 6.860334050619941e-06,
+      "loss": 2.2611,
+      "step": 505200
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 6.851794924343341e-06,
+      "loss": 2.4576,
+      "step": 505300
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 6.843255798066741e-06,
+      "loss": 2.354,
+      "step": 505400
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 6.834716671790142e-06,
+      "loss": 2.2216,
+      "step": 505500
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 6.826177545513543e-06,
+      "loss": 2.3089,
+      "step": 505600
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 6.817638419236943e-06,
+      "loss": 2.405,
+      "step": 505700
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 6.809099292960344e-06,
+      "loss": 2.2704,
+      "step": 505800
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 6.800560166683745e-06,
+      "loss": 2.3242,
+      "step": 505900
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 6.7920210404071455e-06,
+      "loss": 2.2489,
+      "step": 506000
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 6.783481914130546e-06,
+      "loss": 2.2966,
+      "step": 506100
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 6.774942787853947e-06,
+      "loss": 2.3083,
+      "step": 506200
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 6.7664036615773476e-06,
+      "loss": 2.4113,
+      "step": 506300
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 6.757864535300748e-06,
+      "loss": 2.3892,
+      "step": 506400
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 6.749325409024149e-06,
+      "loss": 2.2771,
+      "step": 506500
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 6.74078628274755e-06,
+      "loss": 2.3133,
+      "step": 506600
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 6.7322471564709495e-06,
+      "loss": 2.3161,
+      "step": 506700
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 6.72370803019435e-06,
+      "loss": 2.2451,
+      "step": 506800
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 6.715168903917751e-06,
+      "loss": 2.3036,
+      "step": 506900
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 6.706629777641152e-06,
+      "loss": 2.3547,
+      "step": 507000
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 6.698090651364552e-06,
+      "loss": 2.3271,
+      "step": 507100
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 6.689551525087953e-06,
+      "loss": 2.2541,
+      "step": 507200
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 6.681012398811354e-06,
+      "loss": 2.2262,
+      "step": 507300
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 6.6724732725347544e-06,
+      "loss": 2.2099,
+      "step": 507400
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 6.663934146258155e-06,
+      "loss": 2.285,
+      "step": 507500
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 6.655395019981556e-06,
+      "loss": 2.2368,
+      "step": 507600
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 6.6468558937049565e-06,
+      "loss": 2.2398,
+      "step": 507700
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 6.638316767428357e-06,
+      "loss": 2.3463,
+      "step": 507800
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 6.629777641151758e-06,
+      "loss": 2.2428,
+      "step": 507900
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 6.621238514875159e-06,
+      "loss": 2.5164,
+      "step": 508000
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.6126993885985585e-06,
+      "loss": 2.2446,
+      "step": 508100
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.604160262321959e-06,
+      "loss": 2.1762,
+      "step": 508200
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.59562113604536e-06,
+      "loss": 2.3407,
+      "step": 508300
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.587082009768761e-06,
+      "loss": 2.3899,
+      "step": 508400
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 6.578542883492161e-06,
+      "loss": 2.2486,
+      "step": 508500
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 6.570003757215562e-06,
+      "loss": 2.2667,
+      "step": 508600
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 6.561464630938963e-06,
+      "loss": 2.3153,
+      "step": 508700
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 6.552925504662363e-06,
+      "loss": 2.2504,
+      "step": 508800
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 6.544386378385764e-06,
+      "loss": 2.3833,
+      "step": 508900
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 6.535847252109165e-06,
+      "loss": 2.1683,
+      "step": 509000
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 6.5273081258325655e-06,
+      "loss": 2.2484,
+      "step": 509100
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 6.518768999555966e-06,
+      "loss": 2.4141,
+      "step": 509200
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 6.510229873279367e-06,
+      "loss": 2.0949,
+      "step": 509300
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 6.501690747002767e-06,
+      "loss": 2.3086,
+      "step": 509400
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 6.4931516207261675e-06,
+      "loss": 2.0418,
+      "step": 509500
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 6.484612494449568e-06,
+      "loss": 2.2146,
+      "step": 509600
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 6.476073368172969e-06,
+      "loss": 2.2324,
+      "step": 509700
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 6.4675342418963696e-06,
+      "loss": 2.4706,
+      "step": 509800
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 6.45899511561977e-06,
+      "loss": 2.114,
+      "step": 509900
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 6.450455989343171e-06,
+      "loss": 2.4077,
+      "step": 510000
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 6.441916863066572e-06,
+      "loss": 2.4391,
+      "step": 510100
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 6.433377736789972e-06,
+      "loss": 2.2424,
+      "step": 510200
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 6.424838610513373e-06,
+      "loss": 2.2792,
+      "step": 510300
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 6.416299484236774e-06,
+      "loss": 2.3093,
+      "step": 510400
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 6.4077603579601745e-06,
+      "loss": 2.4905,
+      "step": 510500
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 6.399221231683575e-06,
+      "loss": 2.4747,
+      "step": 510600
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 6.390682105406975e-06,
+      "loss": 2.35,
+      "step": 510700
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 6.382142979130376e-06,
+      "loss": 2.2641,
+      "step": 510800
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 6.3736038528537764e-06,
+      "loss": 2.1705,
+      "step": 510900
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 6.365064726577177e-06,
+      "loss": 2.203,
+      "step": 511000
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 6.356525600300578e-06,
+      "loss": 2.3118,
+      "step": 511100
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 6.3479864740239785e-06,
+      "loss": 2.3299,
+      "step": 511200
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 6.339447347747379e-06,
+      "loss": 2.3658,
+      "step": 511300
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 6.33090822147078e-06,
+      "loss": 2.4442,
+      "step": 511400
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 6.322369095194181e-06,
+      "loss": 2.4042,
+      "step": 511500
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 6.313829968917581e-06,
+      "loss": 2.4569,
+      "step": 511600
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 6.305290842640982e-06,
+      "loss": 2.3307,
+      "step": 511700
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 6.296751716364383e-06,
+      "loss": 2.1856,
+      "step": 511800
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 6.2882125900877835e-06,
+      "loss": 2.218,
+      "step": 511900
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 6.279673463811183e-06,
+      "loss": 2.173,
+      "step": 512000
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 6.271134337534584e-06,
+      "loss": 2.3561,
+      "step": 512100
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 6.262595211257985e-06,
+      "loss": 2.3229,
+      "step": 512200
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 6.254056084981385e-06,
+      "loss": 2.2665,
+      "step": 512300
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 6.245516958704785e-06,
+      "loss": 2.2048,
+      "step": 512400
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 6.236977832428186e-06,
+      "loss": 2.1869,
+      "step": 512500
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 6.228438706151587e-06,
+      "loss": 2.2518,
+      "step": 512600
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 6.219899579874987e-06,
+      "loss": 2.3848,
+      "step": 512700
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 6.211360453598388e-06,
+      "loss": 2.2685,
+      "step": 512800
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 6.202821327321789e-06,
+      "loss": 2.2747,
+      "step": 512900
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 6.1942822010451895e-06,
+      "loss": 2.2388,
+      "step": 513000
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 6.18574307476859e-06,
+      "loss": 2.2014,
+      "step": 513100
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 6.177203948491991e-06,
+      "loss": 2.2756,
+      "step": 513200
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 6.1686648222153916e-06,
+      "loss": 2.4186,
+      "step": 513300
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 6.160125695938792e-06,
+      "loss": 2.4534,
+      "step": 513400
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 6.151586569662193e-06,
+      "loss": 2.3178,
+      "step": 513500
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 6.143047443385593e-06,
+      "loss": 2.1683,
+      "step": 513600
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 6.1345083171089935e-06,
+      "loss": 2.4042,
+      "step": 513700
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 6.125969190832394e-06,
+      "loss": 2.3706,
+      "step": 513800
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 6.117430064555795e-06,
+      "loss": 2.3816,
+      "step": 513900
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 6.108890938279196e-06,
+      "loss": 2.3901,
+      "step": 514000
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 6.100351812002596e-06,
+      "loss": 2.4142,
+      "step": 514100
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 6.091812685725997e-06,
+      "loss": 2.3527,
+      "step": 514200
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 6.083273559449398e-06,
+      "loss": 2.2065,
+      "step": 514300
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 6.0747344331727984e-06,
+      "loss": 2.2061,
+      "step": 514400
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 6.066195306896199e-06,
+      "loss": 2.392,
+      "step": 514500
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 6.0576561806196e-06,
+      "loss": 2.3672,
+      "step": 514600
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 6.0491170543430005e-06,
+      "loss": 2.4435,
+      "step": 514700
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 6.040577928066401e-06,
+      "loss": 2.3277,
+      "step": 514800
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 6.032038801789801e-06,
+      "loss": 2.2059,
+      "step": 514900
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 6.023499675513202e-06,
+      "loss": 2.3794,
+      "step": 515000
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 6.014960549236602e-06,
+      "loss": 2.3631,
+      "step": 515100
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 6.006421422960002e-06,
+      "loss": 2.2747,
+      "step": 515200
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 5.997882296683403e-06,
+      "loss": 2.2276,
+      "step": 515300
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 5.989343170406804e-06,
+      "loss": 2.3502,
+      "step": 515400
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 5.9808040441302045e-06,
+      "loss": 2.2641,
+      "step": 515500
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 5.972264917853605e-06,
+      "loss": 2.4032,
+      "step": 515600
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 5.963725791577006e-06,
+      "loss": 2.2596,
+      "step": 515700
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 5.9551866653004066e-06,
+      "loss": 2.2879,
+      "step": 515800
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 5.946647539023807e-06,
+      "loss": 2.2841,
+      "step": 515900
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 5.938108412747208e-06,
+      "loss": 2.23,
+      "step": 516000
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 5.929569286470609e-06,
+      "loss": 2.4033,
+      "step": 516100
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 5.921030160194009e-06,
+      "loss": 2.1437,
+      "step": 516200
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 5.91249103391741e-06,
+      "loss": 2.2891,
+      "step": 516300
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 5.903951907640811e-06,
+      "loss": 2.1633,
+      "step": 516400
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 5.895412781364211e-06,
+      "loss": 2.2754,
+      "step": 516500
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 5.886873655087611e-06,
+      "loss": 2.3523,
+      "step": 516600
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 5.878334528811012e-06,
+      "loss": 2.2518,
+      "step": 516700
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 5.869795402534413e-06,
+      "loss": 2.3984,
+      "step": 516800
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 5.8612562762578134e-06,
+      "loss": 2.3942,
+      "step": 516900
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 5.852717149981214e-06,
+      "loss": 2.3366,
+      "step": 517000
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 5.844178023704615e-06,
+      "loss": 2.2217,
+      "step": 517100
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 5.8356388974280155e-06,
+      "loss": 2.3396,
+      "step": 517200
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 5.827099771151416e-06,
+      "loss": 2.2669,
+      "step": 517300
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 5.818560644874817e-06,
+      "loss": 2.3906,
+      "step": 517400
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 5.810021518598218e-06,
+      "loss": 2.3159,
+      "step": 517500
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 5.801482392321618e-06,
+      "loss": 2.5124,
+      "step": 517600
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 5.792943266045019e-06,
+      "loss": 2.2749,
+      "step": 517700
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 5.784404139768419e-06,
+      "loss": 2.417,
+      "step": 517800
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 5.77586501349182e-06,
+      "loss": 2.4673,
+      "step": 517900
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 5.76732588721522e-06,
+      "loss": 2.343,
+      "step": 518000
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 5.758786760938621e-06,
+      "loss": 2.3252,
+      "step": 518100
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 5.750247634662022e-06,
+      "loss": 2.5362,
+      "step": 518200
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 5.741708508385422e-06,
+      "loss": 2.3327,
+      "step": 518300
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 5.733169382108823e-06,
+      "loss": 2.347,
+      "step": 518400
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 5.724630255832224e-06,
+      "loss": 2.3024,
+      "step": 518500
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 5.7160911295556245e-06,
+      "loss": 2.278,
+      "step": 518600
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 5.707552003279025e-06,
+      "loss": 2.3998,
+      "step": 518700
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 5.699012877002426e-06,
+      "loss": 2.2746,
+      "step": 518800
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 5.690473750725827e-06,
+      "loss": 2.2974,
+      "step": 518900
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 5.681934624449227e-06,
+      "loss": 2.3175,
+      "step": 519000
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 5.673395498172627e-06,
+      "loss": 2.3615,
+      "step": 519100
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 5.664856371896028e-06,
+      "loss": 2.2808,
+      "step": 519200
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 5.6563172456194286e-06,
+      "loss": 2.4497,
+      "step": 519300
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 5.647778119342828e-06,
+      "loss": 2.4673,
+      "step": 519400
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 5.639238993066229e-06,
+      "loss": 2.1913,
+      "step": 519500
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 5.63069986678963e-06,
+      "loss": 2.1462,
+      "step": 519600
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 5.6221607405130305e-06,
+      "loss": 2.3437,
+      "step": 519700
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 5.613621614236431e-06,
+      "loss": 2.2628,
+      "step": 519800
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 5.605082487959832e-06,
+      "loss": 2.1694,
+      "step": 519900
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 5.596543361683233e-06,
+      "loss": 2.3273,
+      "step": 520000
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 5.588004235406633e-06,
+      "loss": 2.2511,
+      "step": 520100
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 5.579465109130034e-06,
+      "loss": 2.3144,
+      "step": 520200
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 5.570925982853435e-06,
+      "loss": 2.3039,
+      "step": 520300
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 5.5623868565768354e-06,
+      "loss": 2.286,
+      "step": 520400
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 5.553847730300236e-06,
+      "loss": 2.2385,
+      "step": 520500
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 5.545308604023637e-06,
+      "loss": 2.127,
+      "step": 520600
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 5.536769477747037e-06,
+      "loss": 2.3676,
+      "step": 520700
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 5.528230351470437e-06,
+      "loss": 2.3496,
+      "step": 520800
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 5.519691225193838e-06,
+      "loss": 2.3599,
+      "step": 520900
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 5.511152098917239e-06,
+      "loss": 2.1873,
+      "step": 521000
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 5.5026129726406395e-06,
+      "loss": 2.292,
+      "step": 521100
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 5.49407384636404e-06,
+      "loss": 2.2371,
+      "step": 521200
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 5.485534720087441e-06,
+      "loss": 2.3498,
+      "step": 521300
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 5.476995593810842e-06,
+      "loss": 2.2792,
+      "step": 521400
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 5.468456467534242e-06,
+      "loss": 2.2241,
+      "step": 521500
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 5.459917341257643e-06,
+      "loss": 2.3171,
+      "step": 521600
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 5.451378214981044e-06,
+      "loss": 2.3056,
+      "step": 521700
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 5.442839088704444e-06,
+      "loss": 2.1791,
+      "step": 521800
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 5.434299962427845e-06,
+      "loss": 2.2324,
+      "step": 521900
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 5.425760836151245e-06,
+      "loss": 2.2704,
+      "step": 522000
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 5.417221709874646e-06,
+      "loss": 2.3341,
+      "step": 522100
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 5.408682583598046e-06,
+      "loss": 2.1766,
+      "step": 522200
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 5.400143457321447e-06,
+      "loss": 2.1573,
+      "step": 522300
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 5.391604331044848e-06,
+      "loss": 2.2716,
+      "step": 522400
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 5.3830652047682485e-06,
+      "loss": 2.2186,
+      "step": 522500
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 5.374526078491649e-06,
+      "loss": 2.4815,
+      "step": 522600
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 5.36598695221505e-06,
+      "loss": 2.5104,
+      "step": 522700
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 5.3574478259384506e-06,
+      "loss": 2.3255,
+      "step": 522800
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 5.348908699661851e-06,
+      "loss": 2.3178,
+      "step": 522900
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 5.340369573385252e-06,
+      "loss": 2.2776,
+      "step": 523000
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 5.331830447108653e-06,
+      "loss": 2.1851,
+      "step": 523100
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 5.323291320832053e-06,
+      "loss": 2.4296,
+      "step": 523200
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 5.314752194555453e-06,
+      "loss": 2.2819,
+      "step": 523300
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 5.306213068278854e-06,
+      "loss": 2.322,
+      "step": 523400
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 5.297673942002255e-06,
+      "loss": 2.1871,
+      "step": 523500
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 5.289134815725655e-06,
+      "loss": 2.382,
+      "step": 523600
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 5.280595689449056e-06,
+      "loss": 2.51,
+      "step": 523700
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 5.272056563172456e-06,
+      "loss": 2.5047,
+      "step": 523800
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 5.263517436895857e-06,
+      "loss": 2.1496,
+      "step": 523900
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 5.254978310619257e-06,
+      "loss": 2.3528,
+      "step": 524000
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 5.246439184342658e-06,
+      "loss": 2.23,
+      "step": 524100
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 5.237900058066059e-06,
+      "loss": 2.2,
+      "step": 524200
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 5.229360931789459e-06,
+      "loss": 2.4509,
+      "step": 524300
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 5.22082180551286e-06,
+      "loss": 2.218,
+      "step": 524400
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 5.212282679236261e-06,
+      "loss": 2.1552,
+      "step": 524500
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 5.2037435529596615e-06,
+      "loss": 2.4082,
+      "step": 524600
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 5.195204426683062e-06,
+      "loss": 2.1625,
+      "step": 524700
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 5.186665300406463e-06,
+      "loss": 2.3216,
+      "step": 524800
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 5.178126174129863e-06,
+      "loss": 2.2809,
+      "step": 524900
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 5.1695870478532634e-06,
+      "loss": 2.185,
+      "step": 525000
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 5.161047921576664e-06,
+      "loss": 2.2444,
+      "step": 525100
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 5.152508795300065e-06,
+      "loss": 2.3672,
+      "step": 525200
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 5.1439696690234656e-06,
+      "loss": 2.2422,
+      "step": 525300
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 5.135430542746866e-06,
+      "loss": 2.1887,
+      "step": 525400
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 5.126891416470267e-06,
+      "loss": 2.2537,
+      "step": 525500
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 5.118352290193668e-06,
+      "loss": 2.4022,
+      "step": 525600
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 5.109813163917068e-06,
+      "loss": 2.3139,
+      "step": 525700
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 5.101274037640469e-06,
+      "loss": 2.2153,
+      "step": 525800
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 5.09273491136387e-06,
+      "loss": 2.3421,
+      "step": 525900
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 5.0841957850872705e-06,
+      "loss": 2.3735,
+      "step": 526000
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 5.075656658810671e-06,
+      "loss": 2.3653,
+      "step": 526100
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 5.067117532534071e-06,
+      "loss": 2.3226,
+      "step": 526200
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 5.058578406257472e-06,
+      "loss": 2.2424,
+      "step": 526300
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 5.050039279980872e-06,
+      "loss": 2.319,
+      "step": 526400
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 5.041500153704273e-06,
+      "loss": 2.1001,
+      "step": 526500
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 5.032961027427674e-06,
+      "loss": 2.3242,
+      "step": 526600
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 5.0244219011510745e-06,
+      "loss": 2.1981,
+      "step": 526700
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 5.015882774874475e-06,
+      "loss": 2.3837,
+      "step": 526800
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 5.007343648597876e-06,
+      "loss": 2.2706,
+      "step": 526900
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.998804522321277e-06,
+      "loss": 2.3966,
+      "step": 527000
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.990265396044677e-06,
+      "loss": 2.1872,
+      "step": 527100
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.981726269768078e-06,
+      "loss": 2.3085,
+      "step": 527200
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.973187143491479e-06,
+      "loss": 2.1895,
+      "step": 527300
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.9646480172148794e-06,
+      "loss": 2.2384,
+      "step": 527400
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.956108890938279e-06,
+      "loss": 2.3372,
+      "step": 527500
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.94756976466168e-06,
+      "loss": 2.2382,
+      "step": 527600
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.939030638385081e-06,
+      "loss": 2.4529,
+      "step": 527700
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.930491512108481e-06,
+      "loss": 2.2532,
+      "step": 527800
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.921952385831882e-06,
+      "loss": 2.386,
+      "step": 527900
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.913413259555283e-06,
+      "loss": 2.3378,
+      "step": 528000
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.9048741332786835e-06,
+      "loss": 2.4536,
+      "step": 528100
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.896335007002083e-06,
+      "loss": 2.3369,
+      "step": 528200
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.887795880725484e-06,
+      "loss": 2.4021,
+      "step": 528300
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.879256754448885e-06,
+      "loss": 2.283,
+      "step": 528400
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.8707176281722854e-06,
+      "loss": 2.2235,
+      "step": 528500
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.862178501895686e-06,
+      "loss": 2.2042,
+      "step": 528600
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.853639375619087e-06,
+      "loss": 2.1249,
+      "step": 528700
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.8451002493424876e-06,
+      "loss": 2.469,
+      "step": 528800
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.836561123065888e-06,
+      "loss": 2.3044,
+      "step": 528900
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.828021996789289e-06,
+      "loss": 2.4205,
+      "step": 529000
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.819482870512689e-06,
+      "loss": 2.3678,
+      "step": 529100
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.8109437442360895e-06,
+      "loss": 2.2873,
+      "step": 529200
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.80240461795949e-06,
+      "loss": 2.2397,
+      "step": 529300
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.793865491682891e-06,
+      "loss": 2.2666,
+      "step": 529400
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.785326365406292e-06,
+      "loss": 2.1234,
+      "step": 529500
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.776787239129692e-06,
+      "loss": 2.297,
+      "step": 529600
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.768248112853093e-06,
+      "loss": 2.2197,
+      "step": 529700
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.759708986576494e-06,
+      "loss": 2.3912,
+      "step": 529800
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.751169860299894e-06,
+      "loss": 2.2514,
+      "step": 529900
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.742630734023295e-06,
+      "loss": 2.2146,
+      "step": 530000
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.734091607746696e-06,
+      "loss": 2.2581,
+      "step": 530100
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.7255524814700965e-06,
+      "loss": 2.2877,
+      "step": 530200
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.717013355193497e-06,
+      "loss": 2.277,
+      "step": 530300
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.708474228916897e-06,
+      "loss": 2.1507,
+      "step": 530400
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.699935102640298e-06,
+      "loss": 2.3655,
+      "step": 530500
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.6913959763636985e-06,
+      "loss": 2.3029,
+      "step": 530600
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.682856850087099e-06,
+      "loss": 2.3902,
+      "step": 530700
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.6743177238105e-06,
+      "loss": 2.4023,
+      "step": 530800
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.665778597533901e-06,
+      "loss": 2.3543,
+      "step": 530900
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.657239471257301e-06,
+      "loss": 2.4241,
+      "step": 531000
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.648700344980702e-06,
+      "loss": 2.4799,
+      "step": 531100
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.640161218704103e-06,
+      "loss": 2.2627,
+      "step": 531200
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.631622092427503e-06,
+      "loss": 2.2909,
+      "step": 531300
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.623082966150904e-06,
+      "loss": 2.3981,
+      "step": 531400
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.614543839874305e-06,
+      "loss": 2.2313,
+      "step": 531500
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.6060047135977055e-06,
+      "loss": 2.2898,
+      "step": 531600
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.597465587321105e-06,
+      "loss": 2.0953,
+      "step": 531700
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.588926461044506e-06,
+      "loss": 2.3292,
+      "step": 531800
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.580387334767907e-06,
+      "loss": 2.2417,
+      "step": 531900
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.5718482084913074e-06,
+      "loss": 2.2797,
+      "step": 532000
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.563309082214708e-06,
+      "loss": 2.4571,
+      "step": 532100
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.554769955938109e-06,
+      "loss": 2.3237,
+      "step": 532200
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.5462308296615096e-06,
+      "loss": 2.2959,
+      "step": 532300
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.53769170338491e-06,
+      "loss": 2.474,
+      "step": 532400
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.529152577108311e-06,
+      "loss": 2.2593,
+      "step": 532500
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.520613450831711e-06,
+      "loss": 2.4216,
+      "step": 532600
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.5120743245551115e-06,
+      "loss": 2.4381,
+      "step": 532700
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.503535198278512e-06,
+      "loss": 2.3255,
+      "step": 532800
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.494996072001913e-06,
+      "loss": 2.2518,
+      "step": 532900
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.486456945725314e-06,
+      "loss": 2.1913,
+      "step": 533000
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.477917819448714e-06,
+      "loss": 2.2473,
+      "step": 533100
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.469378693172115e-06,
+      "loss": 2.4085,
+      "step": 533200
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.460839566895515e-06,
+      "loss": 2.2575,
+      "step": 533300
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.4523004406189156e-06,
+      "loss": 2.4073,
+      "step": 533400
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.443761314342316e-06,
+      "loss": 2.356,
+      "step": 533500
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.435222188065717e-06,
+      "loss": 2.306,
+      "step": 533600
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.426683061789118e-06,
+      "loss": 2.1347,
+      "step": 533700
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.418143935512518e-06,
+      "loss": 2.2145,
+      "step": 533800
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.409604809235919e-06,
+      "loss": 2.2796,
+      "step": 533900
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.40106568295932e-06,
+      "loss": 2.3241,
+      "step": 534000
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.3925265566827205e-06,
+      "loss": 2.2354,
+      "step": 534100
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.383987430406121e-06,
+      "loss": 2.1824,
+      "step": 534200
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.375448304129522e-06,
+      "loss": 2.35,
+      "step": 534300
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.366909177852923e-06,
+      "loss": 2.3045,
+      "step": 534400
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.358370051576323e-06,
+      "loss": 2.3268,
+      "step": 534500
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.349830925299723e-06,
+      "loss": 2.2335,
+      "step": 534600
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.341291799023124e-06,
+      "loss": 2.2382,
+      "step": 534700
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.3327526727465245e-06,
+      "loss": 2.3745,
+      "step": 534800
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.324213546469925e-06,
+      "loss": 2.4012,
+      "step": 534900
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.315674420193326e-06,
+      "loss": 2.3895,
+      "step": 535000
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.307135293916727e-06,
+      "loss": 2.276,
+      "step": 535100
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.298596167640127e-06,
+      "loss": 2.3365,
+      "step": 535200
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.290057041363528e-06,
+      "loss": 2.1478,
+      "step": 535300
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.281517915086929e-06,
+      "loss": 2.2136,
+      "step": 535400
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.2729787888103294e-06,
+      "loss": 2.1603,
+      "step": 535500
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.26443966253373e-06,
+      "loss": 2.2522,
+      "step": 535600
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.255900536257131e-06,
+      "loss": 2.3462,
+      "step": 535700
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.2473614099805316e-06,
+      "loss": 2.4739,
+      "step": 535800
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.238822283703931e-06,
+      "loss": 2.2846,
+      "step": 535900
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.230283157427332e-06,
+      "loss": 2.3267,
+      "step": 536000
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.221744031150733e-06,
+      "loss": 2.2946,
+      "step": 536100
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.2132049048741335e-06,
+      "loss": 2.1301,
+      "step": 536200
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.204665778597534e-06,
+      "loss": 2.4257,
+      "step": 536300
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.196126652320935e-06,
+      "loss": 2.2667,
+      "step": 536400
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.187587526044336e-06,
+      "loss": 2.2935,
+      "step": 536500
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.179048399767736e-06,
+      "loss": 2.4029,
+      "step": 536600
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.170509273491137e-06,
+      "loss": 2.204,
+      "step": 536700
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.161970147214538e-06,
+      "loss": 2.4656,
+      "step": 536800
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.1534310209379376e-06,
+      "loss": 2.3937,
+      "step": 536900
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.144891894661338e-06,
+      "loss": 2.2908,
+      "step": 537000
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.136352768384739e-06,
+      "loss": 2.2052,
+      "step": 537100
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.12781364210814e-06,
+      "loss": 2.3669,
+      "step": 537200
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.11927451583154e-06,
+      "loss": 2.2515,
+      "step": 537300
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.110735389554941e-06,
+      "loss": 2.2041,
+      "step": 537400
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.102196263278341e-06,
+      "loss": 2.2503,
+      "step": 537500
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.093657137001742e-06,
+      "loss": 2.3934,
+      "step": 537600
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.085118010725142e-06,
+      "loss": 2.3733,
+      "step": 537700
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.076578884448543e-06,
+      "loss": 2.3535,
+      "step": 537800
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.068039758171944e-06,
+      "loss": 2.1226,
+      "step": 537900
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.0595006318953444e-06,
+      "loss": 2.3963,
+      "step": 538000
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.050961505618745e-06,
+      "loss": 2.3109,
+      "step": 538100
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.042422379342146e-06,
+      "loss": 2.2519,
+      "step": 538200
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.0338832530655465e-06,
+      "loss": 2.2264,
+      "step": 538300
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.025344126788947e-06,
+      "loss": 2.3696,
+      "step": 538400
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.016805000512348e-06,
+      "loss": 2.3853,
+      "step": 538500
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.008265874235749e-06,
+      "loss": 2.243,
+      "step": 538600
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 3.999726747959149e-06,
+      "loss": 2.1949,
+      "step": 538700
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 3.991187621682549e-06,
+      "loss": 2.4244,
+      "step": 538800
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 3.98264849540595e-06,
+      "loss": 2.3751,
+      "step": 538900
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 3.974109369129351e-06,
+      "loss": 2.3418,
+      "step": 539000
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 3.965570242852751e-06,
+      "loss": 2.3028,
+      "step": 539100
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 3.957031116576152e-06,
+      "loss": 2.3154,
+      "step": 539200
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 3.948491990299553e-06,
+      "loss": 2.2674,
+      "step": 539300
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 3.939952864022953e-06,
+      "loss": 2.3125,
+      "step": 539400
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 3.931413737746354e-06,
+      "loss": 2.4033,
+      "step": 539500
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 3.922874611469755e-06,
+      "loss": 2.2926,
+      "step": 539600
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 3.9143354851931555e-06,
+      "loss": 2.4927,
+      "step": 539700
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 3.905796358916556e-06,
+      "loss": 2.3426,
+      "step": 539800
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 3.897257232639957e-06,
+      "loss": 2.3957,
+      "step": 539900
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 3.888718106363358e-06,
+      "loss": 2.3603,
+      "step": 540000
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 3.8801789800867575e-06,
+      "loss": 2.2371,
+      "step": 540100
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 3.871639853810158e-06,
+      "loss": 2.393,
+      "step": 540200
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 3.863100727533559e-06,
+      "loss": 2.1292,
+      "step": 540300
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 3.8545616012569596e-06,
+      "loss": 2.3997,
+      "step": 540400
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 3.84602247498036e-06,
+      "loss": 2.4442,
+      "step": 540500
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 3.837483348703761e-06,
+      "loss": 2.1941,
+      "step": 540600
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 3.828944222427162e-06,
+      "loss": 2.3835,
+      "step": 540700
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 3.820405096150562e-06,
+      "loss": 2.225,
+      "step": 540800
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 3.811865969873963e-06,
+      "loss": 2.2939,
+      "step": 540900
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 3.8033268435973638e-06,
+      "loss": 2.1806,
+      "step": 541000
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 3.794787717320764e-06,
+      "loss": 2.4158,
+      "step": 541100
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 3.7862485910441648e-06,
+      "loss": 2.3077,
+      "step": 541200
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 3.7777094647675646e-06,
+      "loss": 2.2717,
+      "step": 541300
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 3.7691703384909653e-06,
+      "loss": 2.4561,
+      "step": 541400
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 3.760631212214366e-06,
+      "loss": 2.2191,
+      "step": 541500
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 3.7520920859377667e-06,
+      "loss": 2.2821,
+      "step": 541600
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 3.7435529596611674e-06,
+      "loss": 2.1932,
+      "step": 541700
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 3.735013833384568e-06,
+      "loss": 2.1749,
+      "step": 541800
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 3.726474707107969e-06,
+      "loss": 2.3083,
+      "step": 541900
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 3.717935580831369e-06,
+      "loss": 2.2537,
+      "step": 542000
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 3.70939645455477e-06,
+      "loss": 2.2607,
+      "step": 542100
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 3.7008573282781705e-06,
+      "loss": 2.3312,
+      "step": 542200
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 3.692318202001571e-06,
+      "loss": 2.3231,
+      "step": 542300
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 3.683779075724972e-06,
+      "loss": 2.2841,
+      "step": 542400
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 3.6752399494483726e-06,
+      "loss": 2.1106,
+      "step": 542500
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 3.6667008231717733e-06,
+      "loss": 2.5235,
+      "step": 542600
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 3.6581616968951736e-06,
+      "loss": 2.3049,
+      "step": 542700
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 3.6496225706185743e-06,
+      "loss": 2.1696,
+      "step": 542800
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 3.641083444341975e-06,
+      "loss": 2.3529,
+      "step": 542900
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 3.6325443180653757e-06,
+      "loss": 2.5494,
+      "step": 543000
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 3.6240051917887764e-06,
+      "loss": 2.4923,
+      "step": 543100
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 3.615466065512177e-06,
+      "loss": 2.2472,
+      "step": 543200
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 3.6069269392355774e-06,
+      "loss": 2.1455,
+      "step": 543300
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 3.598387812958978e-06,
+      "loss": 2.2442,
+      "step": 543400
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 3.5898486866823788e-06,
+      "loss": 2.2608,
+      "step": 543500
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 3.5813095604057795e-06,
+      "loss": 2.4307,
+      "step": 543600
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 3.57277043412918e-06,
+      "loss": 2.5147,
+      "step": 543700
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 3.564231307852581e-06,
+      "loss": 2.2385,
+      "step": 543800
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 3.5556921815759816e-06,
+      "loss": 2.1395,
+      "step": 543900
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 3.547153055299382e-06,
+      "loss": 2.3087,
+      "step": 544000
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 3.5386139290227826e-06,
+      "loss": 2.2105,
+      "step": 544100
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 3.5300748027461833e-06,
+      "loss": 2.3762,
+      "step": 544200
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 3.521535676469584e-06,
+      "loss": 2.297,
+      "step": 544300
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 3.5129965501929847e-06,
+      "loss": 2.3126,
+      "step": 544400
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 3.5044574239163854e-06,
+      "loss": 2.1932,
+      "step": 544500
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 3.4959182976397856e-06,
+      "loss": 2.3053,
+      "step": 544600
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 3.4873791713631863e-06,
+      "loss": 2.4063,
+      "step": 544700
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 3.478840045086587e-06,
+      "loss": 2.2226,
+      "step": 544800
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 3.4703009188099877e-06,
+      "loss": 2.3015,
+      "step": 544900
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 3.4617617925333884e-06,
+      "loss": 2.0615,
+      "step": 545000
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 3.453222666256789e-06,
+      "loss": 2.3217,
+      "step": 545100
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 3.44468353998019e-06,
+      "loss": 2.2258,
+      "step": 545200
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 3.43614441370359e-06,
+      "loss": 2.5063,
+      "step": 545300
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 3.427605287426991e-06,
+      "loss": 2.1059,
+      "step": 545400
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 3.4190661611503915e-06,
+      "loss": 2.294,
+      "step": 545500
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 3.4105270348737922e-06,
+      "loss": 2.1908,
+      "step": 545600
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 3.401987908597192e-06,
+      "loss": 2.4956,
+      "step": 545700
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 3.3934487823205928e-06,
+      "loss": 2.4304,
+      "step": 545800
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 3.3849096560439935e-06,
+      "loss": 2.228,
+      "step": 545900
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 3.376370529767394e-06,
+      "loss": 2.1937,
+      "step": 546000
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 3.367831403490795e-06,
+      "loss": 2.1773,
+      "step": 546100
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 3.359292277214195e-06,
+      "loss": 2.2041,
+      "step": 546200
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 3.350753150937596e-06,
+      "loss": 2.359,
+      "step": 546300
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 3.3422140246609966e-06,
+      "loss": 2.4075,
+      "step": 546400
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 3.3336748983843973e-06,
+      "loss": 2.3136,
+      "step": 546500
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 3.325135772107798e-06,
+      "loss": 2.3028,
+      "step": 546600
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.3165966458311987e-06,
+      "loss": 2.3709,
+      "step": 546700
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.3080575195545994e-06,
+      "loss": 2.084,
+      "step": 546800
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.2995183932779996e-06,
+      "loss": 2.4734,
+      "step": 546900
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.2909792670014003e-06,
+      "loss": 2.1511,
+      "step": 547000
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.282440140724801e-06,
+      "loss": 2.2915,
+      "step": 547100
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.2739010144482017e-06,
+      "loss": 2.2365,
+      "step": 547200
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.2653618881716024e-06,
+      "loss": 2.3072,
+      "step": 547300
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.256822761895003e-06,
+      "loss": 2.2433,
+      "step": 547400
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.2482836356184034e-06,
+      "loss": 2.3882,
+      "step": 547500
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.239744509341804e-06,
+      "loss": 2.1129,
+      "step": 547600
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.231205383065205e-06,
+      "loss": 2.3175,
+      "step": 547700
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.2226662567886055e-06,
+      "loss": 2.402,
+      "step": 547800
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.2141271305120062e-06,
+      "loss": 2.2457,
+      "step": 547900
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.205588004235407e-06,
+      "loss": 2.285,
+      "step": 548000
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.1970488779588076e-06,
+      "loss": 2.2699,
+      "step": 548100
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.188509751682208e-06,
+      "loss": 2.1153,
+      "step": 548200
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.1799706254056086e-06,
+      "loss": 2.2285,
+      "step": 548300
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.1714314991290093e-06,
+      "loss": 2.2391,
+      "step": 548400
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.16289237285241e-06,
+      "loss": 2.4201,
+      "step": 548500
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.1543532465758107e-06,
+      "loss": 2.2739,
+      "step": 548600
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.1458141202992114e-06,
+      "loss": 2.3189,
+      "step": 548700
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.137274994022612e-06,
+      "loss": 2.39,
+      "step": 548800
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.1287358677460124e-06,
+      "loss": 2.2188,
+      "step": 548900
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.1201967414694127e-06,
+      "loss": 2.2681,
+      "step": 549000
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.1116576151928134e-06,
+      "loss": 2.262,
+      "step": 549100
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.103118488916214e-06,
+      "loss": 2.3291,
+      "step": 549200
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.0945793626396148e-06,
+      "loss": 2.1329,
+      "step": 549300
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.0860402363630155e-06,
+      "loss": 2.0695,
+      "step": 549400
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.077501110086416e-06,
+      "loss": 2.5006,
+      "step": 549500
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.0689619838098165e-06,
+      "loss": 2.2184,
+      "step": 549600
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.060422857533217e-06,
+      "loss": 2.088,
+      "step": 549700
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.051883731256618e-06,
+      "loss": 2.2616,
+      "step": 549800
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.0433446049800186e-06,
+      "loss": 2.2769,
+      "step": 549900
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.0348054787034193e-06,
+      "loss": 2.3596,
+      "step": 550000
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.02626635242682e-06,
+      "loss": 2.2853,
+      "step": 550100
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.0177272261502207e-06,
+      "loss": 2.2473,
+      "step": 550200
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3.009188099873621e-06,
+      "loss": 2.3455,
+      "step": 550300
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3.0006489735970216e-06,
+      "loss": 2.3338,
+      "step": 550400
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 2.9921098473204223e-06,
+      "loss": 2.2896,
+      "step": 550500
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 2.983570721043823e-06,
+      "loss": 2.2142,
+      "step": 550600
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 2.9750315947672237e-06,
+      "loss": 2.3097,
+      "step": 550700
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 2.9664924684906244e-06,
+      "loss": 2.2682,
+      "step": 550800
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 2.957953342214025e-06,
+      "loss": 2.301,
+      "step": 550900
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 2.9494142159374254e-06,
+      "loss": 2.5299,
+      "step": 551000
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 2.940875089660826e-06,
+      "loss": 2.3803,
+      "step": 551100
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 2.9323359633842264e-06,
+      "loss": 2.1536,
+      "step": 551200
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 2.923796837107627e-06,
+      "loss": 2.3938,
+      "step": 551300
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 2.915257710831028e-06,
+      "loss": 2.2973,
+      "step": 551400
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 2.9067185845544285e-06,
+      "loss": 2.2337,
+      "step": 551500
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 2.898179458277829e-06,
+      "loss": 2.2885,
+      "step": 551600
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 2.8896403320012295e-06,
+      "loss": 2.3434,
+      "step": 551700
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 2.88110120572463e-06,
+      "loss": 2.3131,
+      "step": 551800
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 2.872562079448031e-06,
+      "loss": 2.3793,
+      "step": 551900
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 2.8640229531714316e-06,
+      "loss": 2.0242,
+      "step": 552000
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 2.8554838268948323e-06,
+      "loss": 2.3354,
+      "step": 552100
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 2.846944700618233e-06,
+      "loss": 2.2313,
+      "step": 552200
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 2.8384055743416337e-06,
+      "loss": 2.346,
+      "step": 552300
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 2.829866448065034e-06,
+      "loss": 2.3449,
+      "step": 552400
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 2.8213273217884347e-06,
+      "loss": 2.2808,
+      "step": 552500
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 2.8127881955118354e-06,
+      "loss": 2.205,
+      "step": 552600
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 2.804249069235236e-06,
+      "loss": 2.3683,
+      "step": 552700
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 2.7957099429586368e-06,
+      "loss": 2.3848,
+      "step": 552800
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 2.7871708166820375e-06,
+      "loss": 2.3793,
+      "step": 552900
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 2.778631690405438e-06,
+      "loss": 2.4302,
+      "step": 553000
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 2.7700925641288385e-06,
+      "loss": 2.5048,
+      "step": 553100
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 2.761553437852239e-06,
+      "loss": 2.3093,
+      "step": 553200
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 2.75301431157564e-06,
+      "loss": 2.1302,
+      "step": 553300
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 2.74447518529904e-06,
+      "loss": 2.5015,
+      "step": 553400
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 2.735936059022441e-06,
+      "loss": 2.4615,
+      "step": 553500
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 2.7273969327458415e-06,
+      "loss": 2.5099,
+      "step": 553600
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 2.7188578064692422e-06,
+      "loss": 2.2814,
+      "step": 553700
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 2.7103186801926425e-06,
+      "loss": 2.1708,
+      "step": 553800
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 2.7017795539160432e-06,
+      "loss": 2.3461,
+      "step": 553900
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 2.693240427639444e-06,
+      "loss": 2.1178,
+      "step": 554000
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 2.6847013013628446e-06,
+      "loss": 2.2476,
+      "step": 554100
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 2.6761621750862453e-06,
+      "loss": 2.1886,
+      "step": 554200
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 2.667623048809646e-06,
+      "loss": 2.235,
+      "step": 554300
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 2.6590839225330467e-06,
+      "loss": 2.3423,
+      "step": 554400
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 2.650544796256447e-06,
+      "loss": 2.3436,
+      "step": 554500
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 2.6420056699798477e-06,
+      "loss": 2.2064,
+      "step": 554600
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 2.6334665437032484e-06,
+      "loss": 2.4161,
+      "step": 554700
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 2.624927417426649e-06,
+      "loss": 2.241,
+      "step": 554800
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 2.61638829115005e-06,
+      "loss": 2.3325,
+      "step": 554900
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 2.6078491648734505e-06,
+      "loss": 2.223,
+      "step": 555000
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 2.599310038596851e-06,
+      "loss": 2.2619,
+      "step": 555100
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 2.5907709123202515e-06,
+      "loss": 2.3514,
+      "step": 555200
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 2.582231786043652e-06,
+      "loss": 2.2996,
+      "step": 555300
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 2.573692659767053e-06,
+      "loss": 2.3588,
+      "step": 555400
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 2.5651535334904536e-06,
+      "loss": 2.3086,
+      "step": 555500
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 2.556614407213854e-06,
+      "loss": 2.1997,
+      "step": 555600
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 2.5480752809372546e-06,
+      "loss": 2.1863,
+      "step": 555700
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 2.5395361546606553e-06,
+      "loss": 2.171,
+      "step": 555800
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 2.5309970283840555e-06,
+      "loss": 2.2636,
+      "step": 555900
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 2.5224579021074563e-06,
+      "loss": 2.3338,
+      "step": 556000
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 2.513918775830857e-06,
+      "loss": 2.3938,
+      "step": 556100
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 2.5053796495542577e-06,
+      "loss": 2.4101,
+      "step": 556200
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 2.4968405232776584e-06,
+      "loss": 2.2689,
+      "step": 556300
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 2.488301397001059e-06,
+      "loss": 2.2431,
+      "step": 556400
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 2.4797622707244598e-06,
+      "loss": 2.3691,
+      "step": 556500
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 2.47122314444786e-06,
+      "loss": 2.3202,
+      "step": 556600
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 2.4626840181712607e-06,
+      "loss": 2.094,
+      "step": 556700
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 2.4541448918946614e-06,
+      "loss": 2.4063,
+      "step": 556800
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 2.445605765618062e-06,
+      "loss": 2.2024,
+      "step": 556900
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 2.437066639341463e-06,
+      "loss": 2.1287,
+      "step": 557000
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 2.4285275130648635e-06,
+      "loss": 2.1329,
+      "step": 557100
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 2.4199883867882642e-06,
+      "loss": 2.3669,
+      "step": 557200
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 2.4114492605116645e-06,
+      "loss": 2.3657,
+      "step": 557300
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 2.4029101342350652e-06,
+      "loss": 2.4,
+      "step": 557400
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 2.394371007958466e-06,
+      "loss": 2.3162,
+      "step": 557500
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 2.3858318816818666e-06,
+      "loss": 2.3129,
+      "step": 557600
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 2.3772927554052673e-06,
+      "loss": 2.1451,
+      "step": 557700
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 2.3687536291286676e-06,
+      "loss": 2.4304,
+      "step": 557800
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 2.3602145028520683e-06,
+      "loss": 2.2704,
+      "step": 557900
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 2.3516753765754686e-06,
+      "loss": 2.3049,
+      "step": 558000
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 2.3431362502988693e-06,
+      "loss": 2.2524,
+      "step": 558100
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 2.33459712402227e-06,
+      "loss": 2.3015,
+      "step": 558200
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 2.3260579977456707e-06,
+      "loss": 2.278,
+      "step": 558300
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 2.3175188714690714e-06,
+      "loss": 2.3302,
+      "step": 558400
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 2.308979745192472e-06,
+      "loss": 2.2837,
+      "step": 558500
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 2.3004406189158728e-06,
+      "loss": 2.3509,
+      "step": 558600
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 2.291901492639273e-06,
+      "loss": 2.3553,
+      "step": 558700
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 2.2833623663626738e-06,
+      "loss": 2.1597,
+      "step": 558800
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 2.2748232400860745e-06,
+      "loss": 2.3153,
+      "step": 558900
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 2.266284113809475e-06,
+      "loss": 2.2201,
+      "step": 559000
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 2.257744987532876e-06,
+      "loss": 2.3287,
+      "step": 559100
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 2.2492058612562766e-06,
+      "loss": 2.2204,
+      "step": 559200
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 2.2406667349796773e-06,
+      "loss": 2.2555,
+      "step": 559300
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 2.2321276087030775e-06,
+      "loss": 2.3604,
+      "step": 559400
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 2.2235884824264783e-06,
+      "loss": 2.3152,
+      "step": 559500
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 2.215049356149879e-06,
+      "loss": 2.2703,
+      "step": 559600
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 2.2065102298732797e-06,
+      "loss": 2.4655,
+      "step": 559700
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 2.1979711035966804e-06,
+      "loss": 2.1722,
+      "step": 559800
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 2.189431977320081e-06,
+      "loss": 2.2994,
+      "step": 559900
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 2.1808928510434813e-06,
+      "loss": 2.2292,
+      "step": 560000
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 2.1723537247668816e-06,
+      "loss": 2.5873,
+      "step": 560100
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 2.1638145984902823e-06,
+      "loss": 2.2973,
+      "step": 560200
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 2.155275472213683e-06,
+      "loss": 2.4311,
+      "step": 560300
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 2.1467363459370837e-06,
+      "loss": 2.1605,
+      "step": 560400
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 2.1381972196604844e-06,
+      "loss": 2.2169,
+      "step": 560500
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 2.129658093383885e-06,
+      "loss": 2.3904,
+      "step": 560600
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 2.121118967107286e-06,
+      "loss": 2.2097,
+      "step": 560700
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 2.112579840830686e-06,
+      "loss": 2.2065,
+      "step": 560800
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 2.104040714554087e-06,
+      "loss": 2.2926,
+      "step": 560900
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 2.0955015882774875e-06,
+      "loss": 2.3529,
+      "step": 561000
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 2.086962462000888e-06,
+      "loss": 2.2113,
+      "step": 561100
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 2.078423335724289e-06,
+      "loss": 2.1736,
+      "step": 561200
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 2.0698842094476896e-06,
+      "loss": 2.1903,
+      "step": 561300
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 2.0613450831710903e-06,
+      "loss": 2.3195,
+      "step": 561400
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 2.0528059568944906e-06,
+      "loss": 2.3093,
+      "step": 561500
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 2.0442668306178913e-06,
+      "loss": 2.3438,
+      "step": 561600
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 2.035727704341292e-06,
+      "loss": 2.2253,
+      "step": 561700
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 2.0271885780646927e-06,
+      "loss": 2.2596,
+      "step": 561800
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 2.0186494517880934e-06,
+      "loss": 2.3994,
+      "step": 561900
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 2.010110325511494e-06,
+      "loss": 2.2312,
+      "step": 562000
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 2.0015711992348944e-06,
+      "loss": 2.4136,
+      "step": 562100
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 1.9930320729582946e-06,
+      "loss": 2.305,
+      "step": 562200
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 1.9844929466816953e-06,
+      "loss": 2.311,
+      "step": 562300
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 1.975953820405096e-06,
+      "loss": 2.3172,
+      "step": 562400
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 1.9674146941284967e-06,
+      "loss": 2.253,
+      "step": 562500
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 1.9588755678518974e-06,
+      "loss": 2.1828,
+      "step": 562600
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 1.950336441575298e-06,
+      "loss": 2.3561,
+      "step": 562700
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 1.941797315298699e-06,
+      "loss": 2.3354,
+      "step": 562800
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 1.933258189022099e-06,
+      "loss": 2.3418,
+      "step": 562900
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 1.9247190627455e-06,
+      "loss": 2.1337,
+      "step": 563000
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 1.9161799364689005e-06,
+      "loss": 2.2881,
+      "step": 563100
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 1.9076408101923012e-06,
+      "loss": 2.1455,
+      "step": 563200
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 1.899101683915702e-06,
+      "loss": 2.2943,
+      "step": 563300
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 1.8905625576391024e-06,
+      "loss": 2.4468,
+      "step": 563400
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 1.8820234313625031e-06,
+      "loss": 2.2302,
+      "step": 563500
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 1.8734843050859038e-06,
+      "loss": 2.2683,
+      "step": 563600
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 1.8649451788093043e-06,
+      "loss": 2.3384,
+      "step": 563700
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 1.856406052532705e-06,
+      "loss": 2.4106,
+      "step": 563800
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 1.8478669262561057e-06,
+      "loss": 2.3334,
+      "step": 563900
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 1.8393277999795064e-06,
+      "loss": 2.3531,
+      "step": 564000
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 1.830788673702907e-06,
+      "loss": 2.3331,
+      "step": 564100
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 1.8222495474263076e-06,
+      "loss": 2.176,
+      "step": 564200
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 1.8137104211497079e-06,
+      "loss": 2.3081,
+      "step": 564300
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 1.8051712948731086e-06,
+      "loss": 2.3887,
+      "step": 564400
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 1.796632168596509e-06,
+      "loss": 2.3987,
+      "step": 564500
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 1.7880930423199098e-06,
+      "loss": 2.3948,
+      "step": 564600
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 1.7795539160433105e-06,
+      "loss": 2.2204,
+      "step": 564700
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 1.771014789766711e-06,
+      "loss": 2.2972,
+      "step": 564800
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 1.7624756634901117e-06,
+      "loss": 2.3433,
+      "step": 564900
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 1.7539365372135124e-06,
+      "loss": 2.2312,
+      "step": 565000
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 1.745397410936913e-06,
+      "loss": 2.2213,
+      "step": 565100
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 1.7368582846603136e-06,
+      "loss": 2.3164,
+      "step": 565200
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 1.7283191583837143e-06,
+      "loss": 2.2808,
+      "step": 565300
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 1.719780032107115e-06,
+      "loss": 2.0961,
+      "step": 565400
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 1.7112409058305155e-06,
+      "loss": 2.3284,
+      "step": 565500
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 1.7027017795539162e-06,
+      "loss": 2.2878,
+      "step": 565600
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 1.6941626532773169e-06,
+      "loss": 2.22,
+      "step": 565700
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 1.6856235270007173e-06,
+      "loss": 2.2838,
+      "step": 565800
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 1.677084400724118e-06,
+      "loss": 2.1683,
+      "step": 565900
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 1.6685452744475187e-06,
+      "loss": 2.1824,
+      "step": 566000
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 1.6600061481709194e-06,
+      "loss": 2.2465,
+      "step": 566100
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 1.65146702189432e-06,
+      "loss": 2.2038,
+      "step": 566200
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 1.6429278956177206e-06,
+      "loss": 2.2572,
+      "step": 566300
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 1.6343887693411213e-06,
+      "loss": 2.2173,
+      "step": 566400
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 1.6258496430645216e-06,
+      "loss": 2.3154,
+      "step": 566500
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 1.617310516787922e-06,
+      "loss": 2.3011,
+      "step": 566600
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 1.6087713905113228e-06,
+      "loss": 2.2621,
+      "step": 566700
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 1.6002322642347235e-06,
+      "loss": 2.3119,
+      "step": 566800
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 1.591693137958124e-06,
+      "loss": 2.331,
+      "step": 566900
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 1.5831540116815247e-06,
+      "loss": 2.4661,
+      "step": 567000
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 1.5746148854049254e-06,
+      "loss": 2.3069,
+      "step": 567100
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 1.566075759128326e-06,
+      "loss": 2.2827,
+      "step": 567200
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 1.5575366328517266e-06,
+      "loss": 2.3809,
+      "step": 567300
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 1.5489975065751273e-06,
+      "loss": 2.2665,
+      "step": 567400
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 1.540458380298528e-06,
+      "loss": 2.2672,
+      "step": 567500
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 1.5319192540219285e-06,
+      "loss": 2.3254,
+      "step": 567600
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 1.5233801277453292e-06,
+      "loss": 2.1762,
+      "step": 567700
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 1.5148410014687299e-06,
+      "loss": 2.1415,
+      "step": 567800
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 1.5063018751921304e-06,
+      "loss": 2.2914,
+      "step": 567900
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 1.497762748915531e-06,
+      "loss": 2.2524,
+      "step": 568000
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 1.4892236226389318e-06,
+      "loss": 2.2628,
+      "step": 568100
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 1.4806844963623323e-06,
+      "loss": 2.3373,
+      "step": 568200
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 1.4721453700857328e-06,
+      "loss": 2.3604,
+      "step": 568300
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 1.4636062438091335e-06,
+      "loss": 2.2078,
+      "step": 568400
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 1.4550671175325342e-06,
+      "loss": 2.3737,
+      "step": 568500
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 1.4465279912559346e-06,
+      "loss": 2.1506,
+      "step": 568600
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 1.4379888649793353e-06,
+      "loss": 2.4117,
+      "step": 568700
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 1.429449738702736e-06,
+      "loss": 2.2271,
+      "step": 568800
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 1.4209106124261367e-06,
+      "loss": 2.3585,
+      "step": 568900
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 1.4123714861495372e-06,
+      "loss": 2.4083,
+      "step": 569000
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 1.403832359872938e-06,
+      "loss": 2.2435,
+      "step": 569100
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 1.3952932335963386e-06,
+      "loss": 2.1267,
+      "step": 569200
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 1.3867541073197391e-06,
+      "loss": 2.2659,
+      "step": 569300
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 1.3782149810431396e-06,
+      "loss": 2.234,
+      "step": 569400
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 1.3696758547665403e-06,
+      "loss": 2.3186,
+      "step": 569500
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 1.361136728489941e-06,
+      "loss": 2.3054,
+      "step": 569600
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 1.3525976022133415e-06,
+      "loss": 2.243,
+      "step": 569700
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 1.3440584759367422e-06,
+      "loss": 2.1987,
+      "step": 569800
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 1.335519349660143e-06,
+      "loss": 2.4739,
+      "step": 569900
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 1.3269802233835434e-06,
+      "loss": 2.156,
+      "step": 570000
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 1.318441097106944e-06,
+      "loss": 2.2558,
+      "step": 570100
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 1.3099019708303448e-06,
+      "loss": 2.3416,
+      "step": 570200
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 1.3013628445537455e-06,
+      "loss": 2.3596,
+      "step": 570300
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 1.2928237182771458e-06,
+      "loss": 2.3077,
+      "step": 570400
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 1.2842845920005465e-06,
+      "loss": 2.377,
+      "step": 570500
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 1.2757454657239472e-06,
+      "loss": 2.2575,
+      "step": 570600
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 1.2672063394473479e-06,
+      "loss": 2.3954,
+      "step": 570700
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 1.2586672131707484e-06,
+      "loss": 2.1197,
+      "step": 570800
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 1.250128086894149e-06,
+      "loss": 2.3559,
+      "step": 570900
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 1.2415889606175498e-06,
+      "loss": 2.1791,
+      "step": 571000
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 1.2330498343409503e-06,
+      "loss": 2.2896,
+      "step": 571100
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 1.224510708064351e-06,
+      "loss": 2.3246,
+      "step": 571200
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 1.2159715817877517e-06,
+      "loss": 2.3038,
+      "step": 571300
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 1.2074324555111522e-06,
+      "loss": 2.4088,
+      "step": 571400
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 1.1988933292345527e-06,
+      "loss": 2.2095,
+      "step": 571500
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 1.1903542029579534e-06,
+      "loss": 2.407,
+      "step": 571600
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 1.181815076681354e-06,
+      "loss": 2.1209,
+      "step": 571700
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 1.1732759504047545e-06,
+      "loss": 2.1136,
+      "step": 571800
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 1.1647368241281552e-06,
+      "loss": 2.3447,
+      "step": 571900
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 1.156197697851556e-06,
+      "loss": 2.2854,
+      "step": 572000
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 1.1476585715749564e-06,
+      "loss": 2.2554,
+      "step": 572100
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 1.1391194452983571e-06,
+      "loss": 2.3975,
+      "step": 572200
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 1.1305803190217578e-06,
+      "loss": 2.1929,
+      "step": 572300
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 1.1220411927451585e-06,
+      "loss": 2.3007,
+      "step": 572400
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 1.113502066468559e-06,
+      "loss": 2.433,
+      "step": 572500
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 1.1049629401919595e-06,
+      "loss": 2.1499,
+      "step": 572600
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 1.0964238139153602e-06,
+      "loss": 2.1606,
+      "step": 572700
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 1.087884687638761e-06,
+      "loss": 2.2014,
+      "step": 572800
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 1.0793455613621614e-06,
+      "loss": 2.229,
+      "step": 572900
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 1.0708064350855621e-06,
+      "loss": 2.1355,
+      "step": 573000
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 1.0622673088089628e-06,
+      "loss": 2.1702,
+      "step": 573100
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 1.0537281825323633e-06,
+      "loss": 2.3781,
+      "step": 573200
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 1.045189056255764e-06,
+      "loss": 2.2928,
+      "step": 573300
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 1.0366499299791647e-06,
+      "loss": 2.2605,
+      "step": 573400
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 1.0281108037025652e-06,
+      "loss": 2.3233,
+      "step": 573500
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 1.0195716774259657e-06,
+      "loss": 2.2551,
+      "step": 573600
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 1.0110325511493664e-06,
+      "loss": 2.4301,
+      "step": 573700
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 1.002493424872767e-06,
+      "loss": 2.2655,
+      "step": 573800
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 9.939542985961676e-07,
+      "loss": 2.3177,
+      "step": 573900
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 9.854151723195683e-07,
+      "loss": 2.2581,
+      "step": 574000
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 9.76876046042969e-07,
+      "loss": 2.2795,
+      "step": 574100
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 9.683369197663695e-07,
+      "loss": 2.4648,
+      "step": 574200
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 9.597977934897702e-07,
+      "loss": 2.2556,
+      "step": 574300
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 9.512586672131709e-07,
+      "loss": 2.2184,
+      "step": 574400
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 9.427195409365715e-07,
+      "loss": 2.2618,
+      "step": 574500
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 9.341804146599721e-07,
+      "loss": 2.3441,
+      "step": 574600
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 9.256412883833725e-07,
+      "loss": 2.1492,
+      "step": 574700
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 9.171021621067732e-07,
+      "loss": 2.3674,
+      "step": 574800
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 9.085630358301738e-07,
+      "loss": 2.1999,
+      "step": 574900
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 9.000239095535744e-07,
+      "loss": 2.2784,
+      "step": 575000
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 8.914847832769751e-07,
+      "loss": 2.2225,
+      "step": 575100
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 8.829456570003757e-07,
+      "loss": 2.2886,
+      "step": 575200
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 8.744065307237764e-07,
+      "loss": 2.3574,
+      "step": 575300
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 8.65867404447177e-07,
+      "loss": 2.2308,
+      "step": 575400
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 8.573282781705776e-07,
+      "loss": 2.2534,
+      "step": 575500
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 8.487891518939783e-07,
+      "loss": 2.2353,
+      "step": 575600
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 8.402500256173789e-07,
+      "loss": 2.2454,
+      "step": 575700
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 8.317108993407794e-07,
+      "loss": 2.4992,
+      "step": 575800
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 8.2317177306418e-07,
+      "loss": 2.4162,
+      "step": 575900
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 8.146326467875807e-07,
+      "loss": 2.2603,
+      "step": 576000
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 8.060935205109813e-07,
+      "loss": 2.4212,
+      "step": 576100
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 7.97554394234382e-07,
+      "loss": 2.1244,
+      "step": 576200
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 7.890152679577826e-07,
+      "loss": 2.3865,
+      "step": 576300
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 7.804761416811832e-07,
+      "loss": 2.2334,
+      "step": 576400
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 7.719370154045839e-07,
+      "loss": 2.3344,
+      "step": 576500
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 7.633978891279845e-07,
+      "loss": 2.2259,
+      "step": 576600
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 7.548587628513851e-07,
+      "loss": 2.4459,
+      "step": 576700
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 7.463196365747857e-07,
+      "loss": 2.3647,
+      "step": 576800
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 7.377805102981863e-07,
+      "loss": 2.4897,
+      "step": 576900
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 7.29241384021587e-07,
+      "loss": 2.3553,
+      "step": 577000
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 7.207022577449876e-07,
+      "loss": 2.2556,
+      "step": 577100
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 7.121631314683882e-07,
+      "loss": 2.3734,
+      "step": 577200
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 7.036240051917888e-07,
+      "loss": 2.414,
+      "step": 577300
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 6.950848789151895e-07,
+      "loss": 2.2522,
+      "step": 577400
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 6.865457526385901e-07,
+      "loss": 2.1318,
+      "step": 577500
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 6.780066263619907e-07,
+      "loss": 2.1168,
+      "step": 577600
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 6.694675000853914e-07,
+      "loss": 2.3635,
+      "step": 577700
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 6.609283738087918e-07,
+      "loss": 2.3013,
+      "step": 577800
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 6.523892475321925e-07,
+      "loss": 2.3134,
+      "step": 577900
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 6.438501212555931e-07,
+      "loss": 2.3957,
+      "step": 578000
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 6.353109949789938e-07,
+      "loss": 2.1935,
+      "step": 578100
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 6.267718687023944e-07,
+      "loss": 2.2694,
+      "step": 578200
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 6.18232742425795e-07,
+      "loss": 2.3684,
+      "step": 578300
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 6.096936161491956e-07,
+      "loss": 2.2628,
+      "step": 578400
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 6.011544898725962e-07,
+      "loss": 2.2747,
+      "step": 578500
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 5.926153635959969e-07,
+      "loss": 2.2606,
+      "step": 578600
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 5.840762373193975e-07,
+      "loss": 2.3139,
+      "step": 578700
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 5.755371110427981e-07,
+      "loss": 2.329,
+      "step": 578800
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 5.669979847661987e-07,
+      "loss": 2.2978,
+      "step": 578900
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 5.584588584895993e-07,
+      "loss": 2.1899,
+      "step": 579000
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 5.49919732213e-07,
+      "loss": 2.2918,
+      "step": 579100
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 5.413806059364006e-07,
+      "loss": 2.1719,
+      "step": 579200
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 5.328414796598013e-07,
+      "loss": 2.4,
+      "step": 579300
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 5.243023533832018e-07,
+      "loss": 2.2283,
+      "step": 579400
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 5.157632271066025e-07,
+      "loss": 2.4378,
+      "step": 579500
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 5.072241008300031e-07,
+      "loss": 2.2668,
+      "step": 579600
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 4.986849745534037e-07,
+      "loss": 2.1497,
+      "step": 579700
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 4.901458482768044e-07,
+      "loss": 2.2747,
+      "step": 579800
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 4.816067220002049e-07,
+      "loss": 2.2221,
+      "step": 579900
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 4.730675957236056e-07,
+      "loss": 2.3611,
+      "step": 580000
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 4.645284694470062e-07,
+      "loss": 2.2331,
+      "step": 580100
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 4.559893431704068e-07,
+      "loss": 2.3034,
+      "step": 580200
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 4.4745021689380747e-07,
+      "loss": 2.2968,
+      "step": 580300
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 4.389110906172081e-07,
+      "loss": 2.3666,
+      "step": 580400
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 4.3037196434060866e-07,
+      "loss": 2.3805,
+      "step": 580500
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 4.218328380640093e-07,
+      "loss": 2.3279,
+      "step": 580600
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 4.132937117874099e-07,
+      "loss": 2.1707,
+      "step": 580700
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 4.0475458551081055e-07,
+      "loss": 2.0922,
+      "step": 580800
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 3.962154592342112e-07,
+      "loss": 2.3051,
+      "step": 580900
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 3.876763329576118e-07,
+      "loss": 2.1741,
+      "step": 581000
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 3.7913720668101245e-07,
+      "loss": 2.3168,
+      "step": 581100
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 3.7059808040441304e-07,
+      "loss": 2.2626,
+      "step": 581200
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 3.620589541278137e-07,
+      "loss": 2.435,
+      "step": 581300
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 3.535198278512143e-07,
+      "loss": 2.3655,
+      "step": 581400
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 3.449807015746149e-07,
+      "loss": 2.2915,
+      "step": 581500
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 3.3644157529801553e-07,
+      "loss": 2.2458,
+      "step": 581600
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 3.279024490214161e-07,
+      "loss": 2.23,
+      "step": 581700
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 3.1936332274481677e-07,
+      "loss": 2.3178,
+      "step": 581800
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 3.108241964682174e-07,
+      "loss": 2.3426,
+      "step": 581900
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 3.02285070191618e-07,
+      "loss": 2.1817,
+      "step": 582000
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 2.937459439150186e-07,
+      "loss": 2.1318,
+      "step": 582100
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 2.8520681763841926e-07,
+      "loss": 2.3357,
+      "step": 582200
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 2.7666769136181985e-07,
+      "loss": 2.4255,
+      "step": 582300
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 2.681285650852205e-07,
+      "loss": 2.3859,
+      "step": 582400
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 2.595894388086211e-07,
+      "loss": 2.2031,
+      "step": 582500
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 2.5105031253202175e-07,
+      "loss": 2.296,
+      "step": 582600
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 2.4251118625542234e-07,
+      "loss": 2.1092,
+      "step": 582700
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 2.3397205997882296e-07,
+      "loss": 2.2336,
+      "step": 582800
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 2.2543293370222359e-07,
+      "loss": 2.3101,
+      "step": 582900
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 2.1689380742562423e-07,
+      "loss": 2.2666,
+      "step": 583000
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 2.0835468114902483e-07,
+      "loss": 2.31,
+      "step": 583100
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 1.9981555487242548e-07,
+      "loss": 2.2143,
+      "step": 583200
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 1.9127642859582607e-07,
+      "loss": 2.3068,
+      "step": 583300
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 1.827373023192267e-07,
+      "loss": 2.2878,
+      "step": 583400
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 1.7419817604262734e-07,
+      "loss": 2.131,
+      "step": 583500
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 1.6565904976602794e-07,
+      "loss": 2.1412,
+      "step": 583600
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 1.5711992348942856e-07,
+      "loss": 2.3785,
+      "step": 583700
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 1.485807972128292e-07,
+      "loss": 2.3242,
+      "step": 583800
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 1.4004167093622983e-07,
+      "loss": 2.2279,
+      "step": 583900
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 1.3150254465963042e-07,
+      "loss": 2.3326,
+      "step": 584000
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 1.2296341838303105e-07,
+      "loss": 2.2253,
+      "step": 584100
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 1.1442429210643168e-07,
+      "loss": 2.3627,
+      "step": 584200
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 1.058851658298323e-07,
+      "loss": 2.1505,
+      "step": 584300
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 9.734603955323291e-08,
+      "loss": 2.4305,
+      "step": 584400
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 8.880691327663355e-08,
+      "loss": 2.312,
+      "step": 584500
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 8.026778700003416e-08,
+      "loss": 2.1861,
+      "step": 584600
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 7.172866072343478e-08,
+      "loss": 2.3235,
+      "step": 584700
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 6.31895344468354e-08,
+      "loss": 2.2686,
+      "step": 584800
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 5.465040817023603e-08,
+      "loss": 2.2017,
+      "step": 584900
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 4.611128189363664e-08,
+      "loss": 2.1246,
+      "step": 585000
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 3.7572155617037264e-08,
+      "loss": 2.2377,
+      "step": 585100
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 2.9033029340437886e-08,
+      "loss": 2.396,
+      "step": 585200
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 2.0493903063838508e-08,
+      "loss": 2.3171,
+      "step": 585300
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 1.195477678723913e-08,
+      "loss": 2.271,
+      "step": 585400
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 3.4156505106397517e-09,
+      "loss": 2.2955,
+      "step": 585500
+    },
+    {
+      "epoch": 15.0,
+      "step": 585540,
+      "total_flos": 4.952930904057427e+16,
+      "train_runtime": 597.6283,
+      "train_samples_per_second": 979.773
+    }
+  ],
+  "max_steps": 585540,
+  "num_train_epochs": 15,
+  "total_flos": 4.952930904057427e+16,
+  "trial_name": null,
+  "trial_params": null
+}