diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,25939 @@
+{
+  "best_metric": 1.145164132118225,
+  "best_model_checkpoint": "./checkpoints/bart-JES-cnn_dailymail/checkpoint-215337",
+  "epoch": 6.0,
+  "global_step": 430674,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999338246562365e-05,
+      "loss": 2.5262,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9986416639964334e-05,
+      "loss": 2.1005,
+      "step": 200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.997945081430502e-05,
+      "loss": 1.9954,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9972484988645703e-05,
+      "loss": 1.9422,
+      "step": 400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.996551916298639e-05,
+      "loss": 1.8915,
+      "step": 500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9958553337327073e-05,
+      "loss": 1.8859,
+      "step": 600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.995158751166776e-05,
+      "loss": 1.8133,
+      "step": 700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9944621686008442e-05,
+      "loss": 1.8649,
+      "step": 800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9937725518605722e-05,
+      "loss": 1.8476,
+      "step": 900
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9930759692946405e-05,
+      "loss": 1.8485,
+      "step": 1000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.992379386728709e-05,
+      "loss": 1.7922,
+      "step": 1100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9916828041627774e-05,
+      "loss": 1.8166,
+      "step": 1200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.990986221596846e-05,
+      "loss": 1.7964,
+      "step": 1300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9902896390309144e-05,
+      "loss": 1.7728,
+      "step": 1400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.989593056464983e-05,
+      "loss": 1.7414,
+      "step": 1500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9888964738990513e-05,
+      "loss": 1.7531,
+      "step": 1600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.98819989133312e-05,
+      "loss": 1.7304,
+      "step": 1700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9875033087671882e-05,
+      "loss": 1.7315,
+      "step": 1800
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.986806726201257e-05,
+      "loss": 1.7073,
+      "step": 1900
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.986110143635325e-05,
+      "loss": 1.7254,
+      "step": 2000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9854135610693938e-05,
+      "loss": 1.7187,
+      "step": 2100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.984716978503462e-05,
+      "loss": 1.7382,
+      "step": 2200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9840203959375304e-05,
+      "loss": 1.6937,
+      "step": 2300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.983323813371599e-05,
+      "loss": 1.6649,
+      "step": 2400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9826272308056673e-05,
+      "loss": 1.7018,
+      "step": 2500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.981930648239736e-05,
+      "loss": 1.6868,
+      "step": 2600
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9812340656738043e-05,
+      "loss": 1.6758,
+      "step": 2700
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.980537483107873e-05,
+      "loss": 1.7183,
+      "step": 2800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9798409005419412e-05,
+      "loss": 1.659,
+      "step": 2900
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.97914431797601e-05,
+      "loss": 1.6675,
+      "step": 3000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.978447735410078e-05,
+      "loss": 1.6956,
+      "step": 3100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9777511528441468e-05,
+      "loss": 1.6716,
+      "step": 3200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.977054570278215e-05,
+      "loss": 1.6501,
+      "step": 3300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9763579877122837e-05,
+      "loss": 1.669,
+      "step": 3400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.975661405146352e-05,
+      "loss": 1.6555,
+      "step": 3500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9749648225804207e-05,
+      "loss": 1.6304,
+      "step": 3600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.974268240014489e-05,
+      "loss": 1.7103,
+      "step": 3700
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9735716574485576e-05,
+      "loss": 1.6437,
+      "step": 3800
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.972875074882626e-05,
+      "loss": 1.6302,
+      "step": 3900
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9721784923166945e-05,
+      "loss": 1.6387,
+      "step": 4000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.971481909750763e-05,
+      "loss": 1.6215,
+      "step": 4100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9707853271848315e-05,
+      "loss": 1.6473,
+      "step": 4200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9700887446188998e-05,
+      "loss": 1.6353,
+      "step": 4300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9693921620529684e-05,
+      "loss": 1.6203,
+      "step": 4400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9686955794870364e-05,
+      "loss": 1.6379,
+      "step": 4500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.967998996921105e-05,
+      "loss": 1.6431,
+      "step": 4600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9673024143551733e-05,
+      "loss": 1.6002,
+      "step": 4700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.966605831789242e-05,
+      "loss": 1.6099,
+      "step": 4800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9659092492233106e-05,
+      "loss": 1.6515,
+      "step": 4900
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.965212666657379e-05,
+      "loss": 1.6138,
+      "step": 5000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9645160840914475e-05,
+      "loss": 1.6059,
+      "step": 5100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9638195015255158e-05,
+      "loss": 1.6292,
+      "step": 5200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9631229189595845e-05,
+      "loss": 1.6093,
+      "step": 5300
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9624263363936528e-05,
+      "loss": 1.5921,
+      "step": 5400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9617297538277214e-05,
+      "loss": 1.6198,
+      "step": 5500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9610331712617897e-05,
+      "loss": 1.5706,
+      "step": 5600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9603365886958583e-05,
+      "loss": 1.58,
+      "step": 5700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9596400061299266e-05,
+      "loss": 1.5918,
+      "step": 5800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9589434235639953e-05,
+      "loss": 1.6068,
+      "step": 5900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9582468409980636e-05,
+      "loss": 1.6168,
+      "step": 6000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9575502584321322e-05,
+      "loss": 1.6105,
+      "step": 6100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9568536758662005e-05,
+      "loss": 1.6061,
+      "step": 6200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.956157093300269e-05,
+      "loss": 1.5705,
+      "step": 6300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9554605107343374e-05,
+      "loss": 1.6033,
+      "step": 6400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.954763928168406e-05,
+      "loss": 1.5526,
+      "step": 6500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9540673456024744e-05,
+      "loss": 1.5967,
+      "step": 6600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9533707630365427e-05,
+      "loss": 1.6052,
+      "step": 6700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.952674180470611e-05,
+      "loss": 1.5469,
+      "step": 6800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9519775979046796e-05,
+      "loss": 1.5831,
+      "step": 6900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9512879811644073e-05,
+      "loss": 1.5752,
+      "step": 7000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.950591398598476e-05,
+      "loss": 1.5882,
+      "step": 7100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9498948160325442e-05,
+      "loss": 1.6126,
+      "step": 7200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9491982334666128e-05,
+      "loss": 1.5726,
+      "step": 7300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.948501650900681e-05,
+      "loss": 1.6108,
+      "step": 7400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9478050683347498e-05,
+      "loss": 1.5633,
+      "step": 7500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.947108485768818e-05,
+      "loss": 1.5745,
+      "step": 7600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9464119032028867e-05,
+      "loss": 1.5643,
+      "step": 7700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.945715320636955e-05,
+      "loss": 1.5851,
+      "step": 7800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9450187380710236e-05,
+      "loss": 1.5255,
+      "step": 7900
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.944322155505092e-05,
+      "loss": 1.5414,
+      "step": 8000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9436255729391606e-05,
+      "loss": 1.5642,
+      "step": 8100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9429289903732292e-05,
+      "loss": 1.5488,
+      "step": 8200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9422324078072975e-05,
+      "loss": 1.5419,
+      "step": 8300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.941535825241366e-05,
+      "loss": 1.5602,
+      "step": 8400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9408392426754344e-05,
+      "loss": 1.5651,
+      "step": 8500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.940142660109503e-05,
+      "loss": 1.56,
+      "step": 8600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9394460775435714e-05,
+      "loss": 1.5774,
+      "step": 8700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.93874949497764e-05,
+      "loss": 1.5753,
+      "step": 8800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9380529124117083e-05,
+      "loss": 1.5387,
+      "step": 8900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.937356329845777e-05,
+      "loss": 1.538,
+      "step": 9000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9366667131055046e-05,
+      "loss": 1.5481,
+      "step": 9100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9359701305395732e-05,
+      "loss": 1.5839,
+      "step": 9200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9352735479736412e-05,
+      "loss": 1.5186,
+      "step": 9300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.93457696540771e-05,
+      "loss": 1.5407,
+      "step": 9400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.933880382841778e-05,
+      "loss": 1.5759,
+      "step": 9500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9331838002758468e-05,
+      "loss": 1.5414,
+      "step": 9600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.932487217709915e-05,
+      "loss": 1.518,
+      "step": 9700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9317906351439837e-05,
+      "loss": 1.5332,
+      "step": 9800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.931094052578052e-05,
+      "loss": 1.5843,
+      "step": 9900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9303974700121206e-05,
+      "loss": 1.5369,
+      "step": 10000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.929700887446189e-05,
+      "loss": 1.5456,
+      "step": 10100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9290043048802576e-05,
+      "loss": 1.5237,
+      "step": 10200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.928307722314326e-05,
+      "loss": 1.5473,
+      "step": 10300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9276111397483945e-05,
+      "loss": 1.5331,
+      "step": 10400
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9269145571824628e-05,
+      "loss": 1.5466,
+      "step": 10500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9262179746165315e-05,
+      "loss": 1.5567,
+      "step": 10600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9255213920505998e-05,
+      "loss": 1.5725,
+      "step": 10700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9248248094846684e-05,
+      "loss": 1.5546,
+      "step": 10800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9241282269187367e-05,
+      "loss": 1.5172,
+      "step": 10900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9234316443528053e-05,
+      "loss": 1.511,
+      "step": 11000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9227350617868736e-05,
+      "loss": 1.4924,
+      "step": 11100
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9220384792209423e-05,
+      "loss": 1.5773,
+      "step": 11200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.92134886248067e-05,
+      "loss": 1.568,
+      "step": 11300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9206522799147385e-05,
+      "loss": 1.5465,
+      "step": 11400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.919955697348807e-05,
+      "loss": 1.5592,
+      "step": 11500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9192591147828755e-05,
+      "loss": 1.5336,
+      "step": 11600
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9185625322169438e-05,
+      "loss": 1.4984,
+      "step": 11700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.917865949651012e-05,
+      "loss": 1.5131,
+      "step": 11800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9171693670850804e-05,
+      "loss": 1.5503,
+      "step": 11900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.916472784519149e-05,
+      "loss": 1.5037,
+      "step": 12000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9157762019532173e-05,
+      "loss": 1.5283,
+      "step": 12100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.915079619387286e-05,
+      "loss": 1.5331,
+      "step": 12200
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9143830368213546e-05,
+      "loss": 1.509,
+      "step": 12300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.913686454255423e-05,
+      "loss": 1.5433,
+      "step": 12400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9129968375151505e-05,
+      "loss": 1.513,
+      "step": 12500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9123002549492192e-05,
+      "loss": 1.553,
+      "step": 12600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9116036723832875e-05,
+      "loss": 1.5211,
+      "step": 12700
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.910907089817356e-05,
+      "loss": 1.4837,
+      "step": 12800
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9102105072514244e-05,
+      "loss": 1.5275,
+      "step": 12900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.909513924685493e-05,
+      "loss": 1.4938,
+      "step": 13000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9088173421195617e-05,
+      "loss": 1.5339,
+      "step": 13100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.90812075955363e-05,
+      "loss": 1.4991,
+      "step": 13200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9074241769876986e-05,
+      "loss": 1.4775,
+      "step": 13300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.906727594421767e-05,
+      "loss": 1.4896,
+      "step": 13400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9060310118558356e-05,
+      "loss": 1.5014,
+      "step": 13500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.905334429289904e-05,
+      "loss": 1.5069,
+      "step": 13600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9046378467239725e-05,
+      "loss": 1.4634,
+      "step": 13700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9039412641580408e-05,
+      "loss": 1.4807,
+      "step": 13800
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9032446815921094e-05,
+      "loss": 1.5231,
+      "step": 13900
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9025480990261777e-05,
+      "loss": 1.482,
+      "step": 14000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9018515164602464e-05,
+      "loss": 1.5036,
+      "step": 14100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9011549338943143e-05,
+      "loss": 1.5032,
+      "step": 14200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.900458351328383e-05,
+      "loss": 1.4612,
+      "step": 14300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8997617687624513e-05,
+      "loss": 1.4982,
+      "step": 14400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.89906518619652e-05,
+      "loss": 1.4899,
+      "step": 14500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8983686036305882e-05,
+      "loss": 1.5048,
+      "step": 14600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.897672021064657e-05,
+      "loss": 1.4901,
+      "step": 14700
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.896975438498725e-05,
+      "loss": 1.4956,
+      "step": 14800
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8962788559327938e-05,
+      "loss": 1.505,
+      "step": 14900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.895582273366862e-05,
+      "loss": 1.4982,
+      "step": 15000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8948856908009307e-05,
+      "loss": 1.4777,
+      "step": 15100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.894189108234999e-05,
+      "loss": 1.4768,
+      "step": 15200
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8934925256690676e-05,
+      "loss": 1.485,
+      "step": 15300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.892795943103136e-05,
+      "loss": 1.4376,
+      "step": 15400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8920993605372046e-05,
+      "loss": 1.4533,
+      "step": 15500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8914027779712732e-05,
+      "loss": 1.4851,
+      "step": 15600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8907061954053415e-05,
+      "loss": 1.4593,
+      "step": 15700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.89000961283941e-05,
+      "loss": 1.4993,
+      "step": 15800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8893130302734785e-05,
+      "loss": 1.5023,
+      "step": 15900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.888616447707547e-05,
+      "loss": 1.4778,
+      "step": 16000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8879198651416154e-05,
+      "loss": 1.4609,
+      "step": 16100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.887223282575684e-05,
+      "loss": 1.494,
+      "step": 16200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8865267000097523e-05,
+      "loss": 1.4559,
+      "step": 16300
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8858301174438206e-05,
+      "loss": 1.503,
+      "step": 16400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.885133534877889e-05,
+      "loss": 1.5209,
+      "step": 16500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8844369523119576e-05,
+      "loss": 1.4833,
+      "step": 16600
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.883740369746026e-05,
+      "loss": 1.4805,
+      "step": 16700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8830437871800945e-05,
+      "loss": 1.4773,
+      "step": 16800
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.8823472046141628e-05,
+      "loss": 1.4863,
+      "step": 16900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.8816575878738908e-05,
+      "loss": 1.5073,
+      "step": 17000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.880961005307959e-05,
+      "loss": 1.4309,
+      "step": 17100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.8802644227420277e-05,
+      "loss": 1.486,
+      "step": 17200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.8795748060017554e-05,
+      "loss": 1.4545,
+      "step": 17300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.878878223435824e-05,
+      "loss": 1.4366,
+      "step": 17400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.8781816408698923e-05,
+      "loss": 1.4476,
+      "step": 17500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.877485058303961e-05,
+      "loss": 1.5035,
+      "step": 17600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8767884757380292e-05,
+      "loss": 1.4817,
+      "step": 17700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.876091893172098e-05,
+      "loss": 1.4922,
+      "step": 17800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8753953106061662e-05,
+      "loss": 1.4724,
+      "step": 17900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8746987280402348e-05,
+      "loss": 1.467,
+      "step": 18000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.874002145474303e-05,
+      "loss": 1.4925,
+      "step": 18100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8733055629083717e-05,
+      "loss": 1.4711,
+      "step": 18200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.87260898034244e-05,
+      "loss": 1.467,
+      "step": 18300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8719123977765087e-05,
+      "loss": 1.4444,
+      "step": 18400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.871215815210577e-05,
+      "loss": 1.462,
+      "step": 18500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8705192326446456e-05,
+      "loss": 1.4642,
+      "step": 18600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.869822650078714e-05,
+      "loss": 1.4821,
+      "step": 18700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8691260675127826e-05,
+      "loss": 1.455,
+      "step": 18800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.868429484946851e-05,
+      "loss": 1.5036,
+      "step": 18900
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.867732902380919e-05,
+      "loss": 1.4585,
+      "step": 19000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.8670363198149875e-05,
+      "loss": 1.4333,
+      "step": 19100
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.866339737249056e-05,
+      "loss": 1.4954,
+      "step": 19200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.8656431546831244e-05,
+      "loss": 1.5066,
+      "step": 19300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.864946572117193e-05,
+      "loss": 1.4496,
+      "step": 19400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.8642499895512613e-05,
+      "loss": 1.4684,
+      "step": 19500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.86355340698533e-05,
+      "loss": 1.4466,
+      "step": 19600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.8628637902450576e-05,
+      "loss": 1.4826,
+      "step": 19700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8621672076791262e-05,
+      "loss": 1.4521,
+      "step": 19800
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8614706251131945e-05,
+      "loss": 1.4588,
+      "step": 19900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8607740425472632e-05,
+      "loss": 1.4844,
+      "step": 20000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8600774599813315e-05,
+      "loss": 1.4472,
+      "step": 20100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8593808774154e-05,
+      "loss": 1.4347,
+      "step": 20200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8586842948494684e-05,
+      "loss": 1.4641,
+      "step": 20300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.857987712283537e-05,
+      "loss": 1.4338,
+      "step": 20400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8572911297176057e-05,
+      "loss": 1.4878,
+      "step": 20500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.856594547151674e-05,
+      "loss": 1.4638,
+      "step": 20600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8558979645857426e-05,
+      "loss": 1.4685,
+      "step": 20700
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.855201382019811e-05,
+      "loss": 1.455,
+      "step": 20800
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8545047994538796e-05,
+      "loss": 1.4495,
+      "step": 20900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.853808216887948e-05,
+      "loss": 1.4194,
+      "step": 21000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8531116343220165e-05,
+      "loss": 1.4316,
+      "step": 21100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8524150517560848e-05,
+      "loss": 1.4557,
+      "step": 21200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8517184691901534e-05,
+      "loss": 1.4427,
+      "step": 21300
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8510218866242217e-05,
+      "loss": 1.4752,
+      "step": 21400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.85032530405829e-05,
+      "loss": 1.4492,
+      "step": 21500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8496287214923583e-05,
+      "loss": 1.4511,
+      "step": 21600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.848932138926427e-05,
+      "loss": 1.4583,
+      "step": 21700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8482355563604953e-05,
+      "loss": 1.4913,
+      "step": 21800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.847538973794564e-05,
+      "loss": 1.4318,
+      "step": 21900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8468423912286322e-05,
+      "loss": 1.4226,
+      "step": 22000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.846145808662701e-05,
+      "loss": 1.4472,
+      "step": 22100
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.845449226096769e-05,
+      "loss": 1.4495,
+      "step": 22200
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8447526435308378e-05,
+      "loss": 1.4007,
+      "step": 22300
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.844056060964906e-05,
+      "loss": 1.4201,
+      "step": 22400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8433594783989747e-05,
+      "loss": 1.4641,
+      "step": 22500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.842662895833043e-05,
+      "loss": 1.4675,
+      "step": 22600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8419663132671117e-05,
+      "loss": 1.4365,
+      "step": 22700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.84126973070118e-05,
+      "loss": 1.4509,
+      "step": 22800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8405731481352486e-05,
+      "loss": 1.4628,
+      "step": 22900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8398765655693172e-05,
+      "loss": 1.4723,
+      "step": 23000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8391799830033855e-05,
+      "loss": 1.4234,
+      "step": 23100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.838483400437454e-05,
+      "loss": 1.4527,
+      "step": 23200
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8377868178715225e-05,
+      "loss": 1.4473,
+      "step": 23300
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.837090235305591e-05,
+      "loss": 1.4594,
+      "step": 23400
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8363936527396594e-05,
+      "loss": 1.4888,
+      "step": 23500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.835697070173728e-05,
+      "loss": 1.4387,
+      "step": 23600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.835000487607796e-05,
+      "loss": 1.4355,
+      "step": 23700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8343039050418646e-05,
+      "loss": 1.4441,
+      "step": 23800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.833607322475933e-05,
+      "loss": 1.4317,
+      "step": 23900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8329107399100016e-05,
+      "loss": 1.4386,
+      "step": 24000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.83221415734407e-05,
+      "loss": 1.4528,
+      "step": 24100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8315175747781385e-05,
+      "loss": 1.4295,
+      "step": 24200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8308209922122068e-05,
+      "loss": 1.402,
+      "step": 24300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8301244096462754e-05,
+      "loss": 1.4026,
+      "step": 24400
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8294278270803437e-05,
+      "loss": 1.4503,
+      "step": 24500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8287312445144124e-05,
+      "loss": 1.4299,
+      "step": 24600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8280346619484807e-05,
+      "loss": 1.4347,
+      "step": 24700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8273380793825493e-05,
+      "loss": 1.4493,
+      "step": 24800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8266414968166176e-05,
+      "loss": 1.4448,
+      "step": 24900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8259449142506863e-05,
+      "loss": 1.4153,
+      "step": 25000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8252483316847546e-05,
+      "loss": 1.4252,
+      "step": 25100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8245517491188232e-05,
+      "loss": 1.3862,
+      "step": 25200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.823855166552892e-05,
+      "loss": 1.4359,
+      "step": 25300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.82315858398696e-05,
+      "loss": 1.4428,
+      "step": 25400
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8224620014210288e-05,
+      "loss": 1.4598,
+      "step": 25500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.821765418855097e-05,
+      "loss": 1.4415,
+      "step": 25600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8210758021148247e-05,
+      "loss": 1.4143,
+      "step": 25700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8203792195488933e-05,
+      "loss": 1.3649,
+      "step": 25800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.819689602808621e-05,
+      "loss": 1.4065,
+      "step": 25900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8189930202426896e-05,
+      "loss": 1.4547,
+      "step": 26000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.818296437676758e-05,
+      "loss": 1.4199,
+      "step": 26100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8175998551108266e-05,
+      "loss": 1.418,
+      "step": 26200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8169032725448945e-05,
+      "loss": 1.4205,
+      "step": 26300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.816206689978963e-05,
+      "loss": 1.433,
+      "step": 26400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8155101074130315e-05,
+      "loss": 1.413,
+      "step": 26500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8148135248471e-05,
+      "loss": 1.3851,
+      "step": 26600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8141169422811684e-05,
+      "loss": 1.4563,
+      "step": 26700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.813420359715237e-05,
+      "loss": 1.4352,
+      "step": 26800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8127237771493057e-05,
+      "loss": 1.4139,
+      "step": 26900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.812027194583374e-05,
+      "loss": 1.3865,
+      "step": 27000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8113306120174426e-05,
+      "loss": 1.4146,
+      "step": 27100
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.810634029451511e-05,
+      "loss": 1.3974,
+      "step": 27200
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8099374468855795e-05,
+      "loss": 1.3841,
+      "step": 27300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.809240864319648e-05,
+      "loss": 1.4116,
+      "step": 27400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8085442817537165e-05,
+      "loss": 1.3978,
+      "step": 27500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8078476991877848e-05,
+      "loss": 1.4165,
+      "step": 27600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8071511166218534e-05,
+      "loss": 1.4303,
+      "step": 27700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8064545340559217e-05,
+      "loss": 1.4583,
+      "step": 27800
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8057579514899904e-05,
+      "loss": 1.3974,
+      "step": 27900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8050613689240587e-05,
+      "loss": 1.4024,
+      "step": 28000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8043647863581273e-05,
+      "loss": 1.4423,
+      "step": 28100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8036682037921956e-05,
+      "loss": 1.4083,
+      "step": 28200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8029716212262642e-05,
+      "loss": 1.421,
+      "step": 28300
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.8022750386603325e-05,
+      "loss": 1.3915,
+      "step": 28400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.8015784560944008e-05,
+      "loss": 1.3912,
+      "step": 28500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.8008888393541288e-05,
+      "loss": 1.3958,
+      "step": 28600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.800192256788197e-05,
+      "loss": 1.4012,
+      "step": 28700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7994956742222654e-05,
+      "loss": 1.4497,
+      "step": 28800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.798799091656334e-05,
+      "loss": 1.4186,
+      "step": 28900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7981025090904023e-05,
+      "loss": 1.4303,
+      "step": 29000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.797405926524471e-05,
+      "loss": 1.4108,
+      "step": 29100
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7967093439585393e-05,
+      "loss": 1.3992,
+      "step": 29200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.796012761392608e-05,
+      "loss": 1.4444,
+      "step": 29300
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7953161788266762e-05,
+      "loss": 1.4092,
+      "step": 29400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.794619596260745e-05,
+      "loss": 1.417,
+      "step": 29500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.793923013694813e-05,
+      "loss": 1.4212,
+      "step": 29600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7932264311288818e-05,
+      "loss": 1.4464,
+      "step": 29700
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.79252984856295e-05,
+      "loss": 1.3923,
+      "step": 29800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7918332659970187e-05,
+      "loss": 1.4111,
+      "step": 29900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.791136683431087e-05,
+      "loss": 1.4088,
+      "step": 30000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7904401008651557e-05,
+      "loss": 1.4142,
+      "step": 30100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7897435182992243e-05,
+      "loss": 1.4369,
+      "step": 30200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7890469357332926e-05,
+      "loss": 1.414,
+      "step": 30300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7883503531673612e-05,
+      "loss": 1.3888,
+      "step": 30400
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7876537706014295e-05,
+      "loss": 1.4087,
+      "step": 30500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7869641538611572e-05,
+      "loss": 1.4189,
+      "step": 30600
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7862675712952258e-05,
+      "loss": 1.3948,
+      "step": 30700
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.785570988729294e-05,
+      "loss": 1.4055,
+      "step": 30800
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7848744061633628e-05,
+      "loss": 1.3809,
+      "step": 30900
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7841778235974314e-05,
+      "loss": 1.454,
+      "step": 31000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7834812410314994e-05,
+      "loss": 1.3974,
+      "step": 31100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.782784658465568e-05,
+      "loss": 1.4263,
+      "step": 31200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7820880758996363e-05,
+      "loss": 1.3835,
+      "step": 31300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.781391493333705e-05,
+      "loss": 1.4252,
+      "step": 31400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7806949107677732e-05,
+      "loss": 1.4255,
+      "step": 31500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.779998328201842e-05,
+      "loss": 1.3858,
+      "step": 31600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.77930174563591e-05,
+      "loss": 1.3573,
+      "step": 31700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7786051630699788e-05,
+      "loss": 1.4498,
+      "step": 31800
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.777908580504047e-05,
+      "loss": 1.3861,
+      "step": 31900
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7772119979381157e-05,
+      "loss": 1.459,
+      "step": 32000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.776515415372184e-05,
+      "loss": 1.4208,
+      "step": 32100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7758188328062527e-05,
+      "loss": 1.4005,
+      "step": 32200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.775122250240321e-05,
+      "loss": 1.4153,
+      "step": 32300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7744256676743896e-05,
+      "loss": 1.3866,
+      "step": 32400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.773729085108458e-05,
+      "loss": 1.3921,
+      "step": 32500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7730325025425265e-05,
+      "loss": 1.4176,
+      "step": 32600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.772335919976595e-05,
+      "loss": 1.438,
+      "step": 32700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7716393374106635e-05,
+      "loss": 1.4075,
+      "step": 32800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7709427548447318e-05,
+      "loss": 1.4416,
+      "step": 32900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7702531381044598e-05,
+      "loss": 1.3933,
+      "step": 33000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.769556555538528e-05,
+      "loss": 1.3889,
+      "step": 33100
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7688599729725967e-05,
+      "loss": 1.4002,
+      "step": 33200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.768163390406665e-05,
+      "loss": 1.4135,
+      "step": 33300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7674668078407336e-05,
+      "loss": 1.392,
+      "step": 33400
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.766770225274802e-05,
+      "loss": 1.4133,
+      "step": 33500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7660736427088702e-05,
+      "loss": 1.397,
+      "step": 33600
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7653770601429385e-05,
+      "loss": 1.3767,
+      "step": 33700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7646804775770072e-05,
+      "loss": 1.411,
+      "step": 33800
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7639838950110755e-05,
+      "loss": 1.3778,
+      "step": 33900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.763287312445144e-05,
+      "loss": 1.4178,
+      "step": 34000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7625907298792124e-05,
+      "loss": 1.4095,
+      "step": 34100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.761894147313281e-05,
+      "loss": 1.3936,
+      "step": 34200
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7611975647473497e-05,
+      "loss": 1.3965,
+      "step": 34300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.760500982181418e-05,
+      "loss": 1.4105,
+      "step": 34400
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7598043996154866e-05,
+      "loss": 1.3629,
+      "step": 34500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7591217487008736e-05,
+      "loss": 1.3862,
+      "step": 34600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.758425166134942e-05,
+      "loss": 1.4215,
+      "step": 34700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7577285835690105e-05,
+      "loss": 1.3996,
+      "step": 34800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.757032001003079e-05,
+      "loss": 1.4127,
+      "step": 34900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7563354184371475e-05,
+      "loss": 1.3986,
+      "step": 35000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7556388358712158e-05,
+      "loss": 1.3808,
+      "step": 35100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7549422533052844e-05,
+      "loss": 1.3846,
+      "step": 35200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7542456707393527e-05,
+      "loss": 1.3943,
+      "step": 35300
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7535490881734214e-05,
+      "loss": 1.3846,
+      "step": 35400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7528525056074897e-05,
+      "loss": 1.41,
+      "step": 35500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.7521559230415583e-05,
+      "loss": 1.3983,
+      "step": 35600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.7514593404756266e-05,
+      "loss": 1.4252,
+      "step": 35700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.7507627579096952e-05,
+      "loss": 1.3834,
+      "step": 35800
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.750066175343764e-05,
+      "loss": 1.3957,
+      "step": 35900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.749369592777832e-05,
+      "loss": 1.3715,
+      "step": 36000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.7486730102119008e-05,
+      "loss": 1.3904,
+      "step": 36100
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.7479764276459688e-05,
+      "loss": 1.4317,
+      "step": 36200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.7472798450800374e-05,
+      "loss": 1.3789,
+      "step": 36300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.7465832625141057e-05,
+      "loss": 1.343,
+      "step": 36400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.7458866799481743e-05,
+      "loss": 1.3381,
+      "step": 36500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.7451900973822426e-05,
+      "loss": 1.4004,
+      "step": 36600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.7444935148163113e-05,
+      "loss": 1.4007,
+      "step": 36700
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.7437969322503796e-05,
+      "loss": 1.4021,
+      "step": 36800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.7431003496844482e-05,
+      "loss": 1.4194,
+      "step": 36900
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7424037671185165e-05,
+      "loss": 1.3586,
+      "step": 37000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.741707184552585e-05,
+      "loss": 1.3964,
+      "step": 37100
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7410106019866534e-05,
+      "loss": 1.4042,
+      "step": 37200
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.740314019420722e-05,
+      "loss": 1.4042,
+      "step": 37300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7396174368547904e-05,
+      "loss": 1.4241,
+      "step": 37400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7389278201145184e-05,
+      "loss": 1.3756,
+      "step": 37500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7382312375485867e-05,
+      "loss": 1.3848,
+      "step": 37600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.7375346549826553e-05,
+      "loss": 1.4253,
+      "step": 37700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.7368380724167236e-05,
+      "loss": 1.4135,
+      "step": 37800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.7361414898507922e-05,
+      "loss": 1.4172,
+      "step": 37900
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.7354449072848605e-05,
+      "loss": 1.416,
+      "step": 38000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.7347483247189292e-05,
+      "loss": 1.3781,
+      "step": 38100
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.7340517421529975e-05,
+      "loss": 1.3785,
+      "step": 38200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.733355159587066e-05,
+      "loss": 1.3614,
+      "step": 38300
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.7326585770211344e-05,
+      "loss": 1.3609,
+      "step": 38400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.731961994455203e-05,
+      "loss": 1.3633,
+      "step": 38500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.731265411889271e-05,
+      "loss": 1.3579,
+      "step": 38600
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.7305688293233396e-05,
+      "loss": 1.3761,
+      "step": 38700
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.729872246757408e-05,
+      "loss": 1.3834,
+      "step": 38800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.7291756641914766e-05,
+      "loss": 1.3888,
+      "step": 38900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.728479081625545e-05,
+      "loss": 1.4012,
+      "step": 39000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.7277824990596135e-05,
+      "loss": 1.4121,
+      "step": 39100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.727085916493682e-05,
+      "loss": 1.3844,
+      "step": 39200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.7263893339277505e-05,
+      "loss": 1.348,
+      "step": 39300
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.725692751361819e-05,
+      "loss": 1.4259,
+      "step": 39400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.7249961687958874e-05,
+      "loss": 1.3763,
+      "step": 39500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.724299586229956e-05,
+      "loss": 1.3797,
+      "step": 39600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.7236030036640243e-05,
+      "loss": 1.4056,
+      "step": 39700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.722906421098093e-05,
+      "loss": 1.3613,
+      "step": 39800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.7222098385321613e-05,
+      "loss": 1.387,
+      "step": 39900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.72151325596623e-05,
+      "loss": 1.3829,
+      "step": 40000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.7208166734002982e-05,
+      "loss": 1.4094,
+      "step": 40100
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.720120090834367e-05,
+      "loss": 1.378,
+      "step": 40200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.719423508268435e-05,
+      "loss": 1.4012,
+      "step": 40300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.7187269257025038e-05,
+      "loss": 1.4169,
+      "step": 40400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.718030343136572e-05,
+      "loss": 1.3587,
+      "step": 40500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.7173337605706407e-05,
+      "loss": 1.4081,
+      "step": 40600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.716637178004709e-05,
+      "loss": 1.372,
+      "step": 40700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.7159405954387773e-05,
+      "loss": 1.3797,
+      "step": 40800
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.7152440128728456e-05,
+      "loss": 1.3596,
+      "step": 40900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.7145474303069142e-05,
+      "loss": 1.3785,
+      "step": 41000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.7138508477409825e-05,
+      "loss": 1.3757,
+      "step": 41100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.7131542651750512e-05,
+      "loss": 1.3541,
+      "step": 41200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.7124576826091195e-05,
+      "loss": 1.359,
+      "step": 41300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.711761100043188e-05,
+      "loss": 1.3764,
+      "step": 41400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.7110645174772564e-05,
+      "loss": 1.397,
+      "step": 41500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.710367934911325e-05,
+      "loss": 1.378,
+      "step": 41600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.7096713523453937e-05,
+      "loss": 1.3919,
+      "step": 41700
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.708974769779462e-05,
+      "loss": 1.3885,
+      "step": 41800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.7082781872135306e-05,
+      "loss": 1.3811,
+      "step": 41900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.707581604647599e-05,
+      "loss": 1.3357,
+      "step": 42000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.7068850220816676e-05,
+      "loss": 1.378,
+      "step": 42100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.706188439515736e-05,
+      "loss": 1.3892,
+      "step": 42200
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.7054918569498045e-05,
+      "loss": 1.4016,
+      "step": 42300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.7047952743838728e-05,
+      "loss": 1.3872,
+      "step": 42400
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.7040986918179414e-05,
+      "loss": 1.3525,
+      "step": 42500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.7034021092520097e-05,
+      "loss": 1.4103,
+      "step": 42600
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.7027055266860784e-05,
+      "loss": 1.4118,
+      "step": 42700
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.7020089441201467e-05,
+      "loss": 1.3327,
+      "step": 42800
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.7013123615542153e-05,
+      "loss": 1.3728,
+      "step": 42900
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.7006157789882836e-05,
+      "loss": 1.3841,
+      "step": 43000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.699919196422352e-05,
+      "loss": 1.3852,
+      "step": 43100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.6992226138564202e-05,
+      "loss": 1.3959,
+      "step": 43200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.698526031290489e-05,
+      "loss": 1.3464,
+      "step": 43300
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.697829448724557e-05,
+      "loss": 1.3963,
+      "step": 43400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.6971328661586258e-05,
+      "loss": 1.3546,
+      "step": 43500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.696436283592694e-05,
+      "loss": 1.3473,
+      "step": 43600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.6957397010267627e-05,
+      "loss": 1.3486,
+      "step": 43700
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.695043118460831e-05,
+      "loss": 1.3969,
+      "step": 43800
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.6943465358948997e-05,
+      "loss": 1.3857,
+      "step": 43900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.6936499533289683e-05,
+      "loss": 1.3501,
+      "step": 44000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.6929533707630366e-05,
+      "loss": 1.3741,
+      "step": 44100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.6922567881971052e-05,
+      "loss": 1.3536,
+      "step": 44200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.6915602056311735e-05,
+      "loss": 1.3821,
+      "step": 44300
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.6908636230652422e-05,
+      "loss": 1.3686,
+      "step": 44400
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.6901670404993105e-05,
+      "loss": 1.3842,
+      "step": 44500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.689470457933379e-05,
+      "loss": 1.3715,
+      "step": 44600
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.6887738753674474e-05,
+      "loss": 1.3883,
+      "step": 44700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.688077292801516e-05,
+      "loss": 1.3736,
+      "step": 44800
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.6873876760612437e-05,
+      "loss": 1.4091,
+      "step": 44900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.6866910934953123e-05,
+      "loss": 1.3606,
+      "step": 45000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.6859945109293806e-05,
+      "loss": 1.3626,
+      "step": 45100
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.6852979283634493e-05,
+      "loss": 1.3744,
+      "step": 45200
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.6846013457975176e-05,
+      "loss": 1.3564,
+      "step": 45300
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.6839047632315862e-05,
+      "loss": 1.3506,
+      "step": 45400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.683208180665654e-05,
+      "loss": 1.3395,
+      "step": 45500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.6825115980997228e-05,
+      "loss": 1.3471,
+      "step": 45600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.681815015533791e-05,
+      "loss": 1.3495,
+      "step": 45700
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.6811184329678597e-05,
+      "loss": 1.3854,
+      "step": 45800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.680421850401928e-05,
+      "loss": 1.3284,
+      "step": 45900
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.6797252678359967e-05,
+      "loss": 1.3682,
+      "step": 46000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.679028685270065e-05,
+      "loss": 1.3701,
+      "step": 46100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.6783321027041336e-05,
+      "loss": 1.3407,
+      "step": 46200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.677635520138202e-05,
+      "loss": 1.39,
+      "step": 46300
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.6769389375722705e-05,
+      "loss": 1.3653,
+      "step": 46400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.676242355006339e-05,
+      "loss": 1.3297,
+      "step": 46500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.6755457724404075e-05,
+      "loss": 1.3955,
+      "step": 46600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.6748491898744758e-05,
+      "loss": 1.3503,
+      "step": 46700
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.6741526073085444e-05,
+      "loss": 1.3589,
+      "step": 46800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.6734560247426127e-05,
+      "loss": 1.3556,
+      "step": 46900
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.6727594421766814e-05,
+      "loss": 1.3429,
+      "step": 47000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6720628596107497e-05,
+      "loss": 1.3697,
+      "step": 47100
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6713662770448183e-05,
+      "loss": 1.3673,
+      "step": 47200
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.670669694478887e-05,
+      "loss": 1.392,
+      "step": 47300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6699731119129552e-05,
+      "loss": 1.3497,
+      "step": 47400
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.669283495172683e-05,
+      "loss": 1.3425,
+      "step": 47500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6685869126067515e-05,
+      "loss": 1.3948,
+      "step": 47600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6678903300408198e-05,
+      "loss": 1.3741,
+      "step": 47700
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6671937474748884e-05,
+      "loss": 1.348,
+      "step": 47800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6664971649089564e-05,
+      "loss": 1.3932,
+      "step": 47900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.665800582343025e-05,
+      "loss": 1.3822,
+      "step": 48000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6651039997770937e-05,
+      "loss": 1.3723,
+      "step": 48100
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.664407417211162e-05,
+      "loss": 1.4158,
+      "step": 48200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6637108346452306e-05,
+      "loss": 1.3831,
+      "step": 48300
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.663014252079299e-05,
+      "loss": 1.3851,
+      "step": 48400
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6623176695133676e-05,
+      "loss": 1.3529,
+      "step": 48500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.661621086947436e-05,
+      "loss": 1.4082,
+      "step": 48600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6609245043815045e-05,
+      "loss": 1.3375,
+      "step": 48700
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6602279218155728e-05,
+      "loss": 1.3496,
+      "step": 48800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6595313392496414e-05,
+      "loss": 1.3226,
+      "step": 48900
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6588347566837097e-05,
+      "loss": 1.352,
+      "step": 49000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6581381741177784e-05,
+      "loss": 1.3398,
+      "step": 49100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6574415915518467e-05,
+      "loss": 1.3801,
+      "step": 49200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6567450089859153e-05,
+      "loss": 1.3649,
+      "step": 49300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6560484264199836e-05,
+      "loss": 1.3455,
+      "step": 49400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6553518438540522e-05,
+      "loss": 1.34,
+      "step": 49500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6546552612881205e-05,
+      "loss": 1.3499,
+      "step": 49600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6539586787221892e-05,
+      "loss": 1.3164,
+      "step": 49700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6532620961562575e-05,
+      "loss": 1.3259,
+      "step": 49800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.652565513590326e-05,
+      "loss": 1.3376,
+      "step": 49900
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.6518689310243944e-05,
+      "loss": 1.3874,
+      "step": 50000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.651172348458463e-05,
+      "loss": 1.4079,
+      "step": 50100
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.650475765892531e-05,
+      "loss": 1.4041,
+      "step": 50200
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.6497791833265996e-05,
+      "loss": 1.3329,
+      "step": 50300
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.649082600760668e-05,
+      "loss": 1.3223,
+      "step": 50400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.6483860181947366e-05,
+      "loss": 1.3544,
+      "step": 50500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.6476894356288052e-05,
+      "loss": 1.3594,
+      "step": 50600
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.6469928530628735e-05,
+      "loss": 1.3821,
+      "step": 50700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.646296270496942e-05,
+      "loss": 1.3086,
+      "step": 50800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.6455996879310105e-05,
+      "loss": 1.3501,
+      "step": 50900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.644903105365079e-05,
+      "loss": 1.3729,
+      "step": 51000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.6442065227991474e-05,
+      "loss": 1.3819,
+      "step": 51100
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.643509940233216e-05,
+      "loss": 1.3432,
+      "step": 51200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.6428133576672843e-05,
+      "loss": 1.3654,
+      "step": 51300
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.642116775101353e-05,
+      "loss": 1.3374,
+      "step": 51400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.6414201925354213e-05,
+      "loss": 1.3297,
+      "step": 51500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.6407305757951492e-05,
+      "loss": 1.3748,
+      "step": 51600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.6400339932292175e-05,
+      "loss": 1.3697,
+      "step": 51700
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.6393374106632862e-05,
+      "loss": 1.3465,
+      "step": 51800
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.6386408280973545e-05,
+      "loss": 1.321,
+      "step": 51900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.637944245531423e-05,
+      "loss": 1.325,
+      "step": 52000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6372476629654914e-05,
+      "loss": 1.3387,
+      "step": 52100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.63655108039956e-05,
+      "loss": 1.3562,
+      "step": 52200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6358614636592877e-05,
+      "loss": 1.3304,
+      "step": 52300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6351648810933563e-05,
+      "loss": 1.3403,
+      "step": 52400
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.634475264353084e-05,
+      "loss": 1.3563,
+      "step": 52500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.6337786817871523e-05,
+      "loss": 1.3802,
+      "step": 52600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.633082099221221e-05,
+      "loss": 1.3703,
+      "step": 52700
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.6323855166552892e-05,
+      "loss": 1.3479,
+      "step": 52800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.6316889340893575e-05,
+      "loss": 1.3605,
+      "step": 52900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.630992351523426e-05,
+      "loss": 1.3378,
+      "step": 53000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.6302957689574945e-05,
+      "loss": 1.3712,
+      "step": 53100
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.629599186391563e-05,
+      "loss": 1.3279,
+      "step": 53200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.6289026038256314e-05,
+      "loss": 1.3342,
+      "step": 53300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.6282060212597e-05,
+      "loss": 1.3442,
+      "step": 53400
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.6275094386937683e-05,
+      "loss": 1.3437,
+      "step": 53500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.626812856127837e-05,
+      "loss": 1.3576,
+      "step": 53600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.6261162735619053e-05,
+      "loss": 1.358,
+      "step": 53700
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.625419690995974e-05,
+      "loss": 1.3469,
+      "step": 53800
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.6247231084300422e-05,
+      "loss": 1.3949,
+      "step": 53900
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.624026525864111e-05,
+      "loss": 1.3745,
+      "step": 54000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.623329943298179e-05,
+      "loss": 1.3844,
+      "step": 54100
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6226333607322478e-05,
+      "loss": 1.3565,
+      "step": 54200
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.621936778166316e-05,
+      "loss": 1.3887,
+      "step": 54300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6212401956003847e-05,
+      "loss": 1.3459,
+      "step": 54400
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.620543613034453e-05,
+      "loss": 1.3812,
+      "step": 54500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6198470304685216e-05,
+      "loss": 1.3449,
+      "step": 54600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.61915044790259e-05,
+      "loss": 1.3412,
+      "step": 54700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6184538653366586e-05,
+      "loss": 1.3461,
+      "step": 54800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.617757282770727e-05,
+      "loss": 1.3561,
+      "step": 54900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6170607002047955e-05,
+      "loss": 1.3813,
+      "step": 55000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6163641176388638e-05,
+      "loss": 1.374,
+      "step": 55100
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.615667535072932e-05,
+      "loss": 1.3128,
+      "step": 55200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6149709525070004e-05,
+      "loss": 1.3356,
+      "step": 55300
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.614274369941069e-05,
+      "loss": 1.3672,
+      "step": 55400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6135777873751377e-05,
+      "loss": 1.3412,
+      "step": 55500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.612881204809206e-05,
+      "loss": 1.3583,
+      "step": 55600
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6121846222432746e-05,
+      "loss": 1.3645,
+      "step": 55700
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.611488039677343e-05,
+      "loss": 1.3354,
+      "step": 55800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6107914571114116e-05,
+      "loss": 1.383,
+      "step": 55900
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.61009487454548e-05,
+      "loss": 1.2789,
+      "step": 56000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6093982919795485e-05,
+      "loss": 1.3473,
+      "step": 56100
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6087017094136168e-05,
+      "loss": 1.3763,
+      "step": 56200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6080051268476854e-05,
+      "loss": 1.3376,
+      "step": 56300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.6073085442817537e-05,
+      "loss": 1.3036,
+      "step": 56400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.6066119617158224e-05,
+      "loss": 1.3755,
+      "step": 56500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.6059153791498907e-05,
+      "loss": 1.3289,
+      "step": 56600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.6052187965839593e-05,
+      "loss": 1.357,
+      "step": 56700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.6045222140180276e-05,
+      "loss": 1.3371,
+      "step": 56800
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.6038256314520962e-05,
+      "loss": 1.2948,
+      "step": 56900
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.6031290488861645e-05,
+      "loss": 1.3417,
+      "step": 57000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.6024324663202332e-05,
+      "loss": 1.3512,
+      "step": 57100
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.6017358837543015e-05,
+      "loss": 1.3313,
+      "step": 57200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.60103930118837e-05,
+      "loss": 1.3259,
+      "step": 57300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.600342718622438e-05,
+      "loss": 1.3269,
+      "step": 57400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.5996461360565067e-05,
+      "loss": 1.3432,
+      "step": 57500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.598949553490575e-05,
+      "loss": 1.3698,
+      "step": 57600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.5982529709246437e-05,
+      "loss": 1.3546,
+      "step": 57700
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.597556388358712e-05,
+      "loss": 1.383,
+      "step": 57800
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.5968598057927806e-05,
+      "loss": 1.2983,
+      "step": 57900
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.5961632232268492e-05,
+      "loss": 1.3356,
+      "step": 58000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.595473606486577e-05,
+      "loss": 1.3308,
+      "step": 58100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.5947770239206452e-05,
+      "loss": 1.3321,
+      "step": 58200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.5940804413547138e-05,
+      "loss": 1.3815,
+      "step": 58300
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.593383858788782e-05,
+      "loss": 1.3486,
+      "step": 58400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.5926872762228507e-05,
+      "loss": 1.3451,
+      "step": 58500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.591990693656919e-05,
+      "loss": 1.3521,
+      "step": 58600
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.5912941110909877e-05,
+      "loss": 1.3315,
+      "step": 58700
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.5905975285250563e-05,
+      "loss": 1.3411,
+      "step": 58800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.5899009459591246e-05,
+      "loss": 1.2954,
+      "step": 58900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.5892043633931933e-05,
+      "loss": 1.3433,
+      "step": 59000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.5885077808272616e-05,
+      "loss": 1.3422,
+      "step": 59100
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.5878111982613302e-05,
+      "loss": 1.3384,
+      "step": 59200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.5871146156953985e-05,
+      "loss": 1.2941,
+      "step": 59300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.586418033129467e-05,
+      "loss": 1.3337,
+      "step": 59400
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.5857214505635354e-05,
+      "loss": 1.3622,
+      "step": 59500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.585024867997604e-05,
+      "loss": 1.36,
+      "step": 59600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.5843282854316724e-05,
+      "loss": 1.336,
+      "step": 59700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.583631702865741e-05,
+      "loss": 1.3074,
+      "step": 59800
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.582935120299809e-05,
+      "loss": 1.3693,
+      "step": 59900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5822385377338776e-05,
+      "loss": 1.3477,
+      "step": 60000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.581541955167946e-05,
+      "loss": 1.3297,
+      "step": 60100
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5808453726020145e-05,
+      "loss": 1.3227,
+      "step": 60200
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.580148790036083e-05,
+      "loss": 1.3525,
+      "step": 60300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5794522074701515e-05,
+      "loss": 1.374,
+      "step": 60400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5787556249042198e-05,
+      "loss": 1.3384,
+      "step": 60500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5780590423382884e-05,
+      "loss": 1.3401,
+      "step": 60600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5773624597723567e-05,
+      "loss": 1.3454,
+      "step": 60700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5766658772064253e-05,
+      "loss": 1.3388,
+      "step": 60800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5759692946404936e-05,
+      "loss": 1.3883,
+      "step": 60900
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5752727120745623e-05,
+      "loss": 1.3638,
+      "step": 61000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5745761295086306e-05,
+      "loss": 1.3263,
+      "step": 61100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5738795469426992e-05,
+      "loss": 1.3555,
+      "step": 61200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.573182964376768e-05,
+      "loss": 1.3402,
+      "step": 61300
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.572486381810836e-05,
+      "loss": 1.3395,
+      "step": 61400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5717897992449048e-05,
+      "loss": 1.3215,
+      "step": 61500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.571093216678973e-05,
+      "loss": 1.3557,
+      "step": 61600
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5703966341130417e-05,
+      "loss": 1.3658,
+      "step": 61700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.56970005154711e-05,
+      "loss": 1.3467,
+      "step": 61800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5690034689811787e-05,
+      "loss": 1.3427,
+      "step": 61900
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.568306886415247e-05,
+      "loss": 1.3249,
+      "step": 62000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5676103038493153e-05,
+      "loss": 1.3471,
+      "step": 62100
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5669206871090432e-05,
+      "loss": 1.3272,
+      "step": 62200
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5662241045431115e-05,
+      "loss": 1.307,
+      "step": 62300
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.56552752197718e-05,
+      "loss": 1.2977,
+      "step": 62400
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5648309394112485e-05,
+      "loss": 1.3312,
+      "step": 62500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5641343568453168e-05,
+      "loss": 1.3112,
+      "step": 62600
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5634377742793854e-05,
+      "loss": 1.377,
+      "step": 62700
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5627411917134537e-05,
+      "loss": 1.3737,
+      "step": 62800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.5620446091475224e-05,
+      "loss": 1.3194,
+      "step": 62900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.5613480265815907e-05,
+      "loss": 1.3558,
+      "step": 63000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.5606514440156593e-05,
+      "loss": 1.3076,
+      "step": 63100
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.5599548614497276e-05,
+      "loss": 1.3565,
+      "step": 63200
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.5592582788837962e-05,
+      "loss": 1.3515,
+      "step": 63300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.5585616963178645e-05,
+      "loss": 1.3132,
+      "step": 63400
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.557865113751933e-05,
+      "loss": 1.3345,
+      "step": 63500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5571685311860015e-05,
+      "loss": 1.3776,
+      "step": 63600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.55647194862007e-05,
+      "loss": 1.3238,
+      "step": 63700
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5557753660541384e-05,
+      "loss": 1.3039,
+      "step": 63800
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.555078783488207e-05,
+      "loss": 1.3506,
+      "step": 63900
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5543822009222753e-05,
+      "loss": 1.3469,
+      "step": 64000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.553685618356344e-05,
+      "loss": 1.3297,
+      "step": 64100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5529890357904123e-05,
+      "loss": 1.3315,
+      "step": 64200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5522994190501403e-05,
+      "loss": 1.3083,
+      "step": 64300
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5516028364842086e-05,
+      "loss": 1.3433,
+      "step": 64400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5509062539182772e-05,
+      "loss": 1.3406,
+      "step": 64500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5502096713523455e-05,
+      "loss": 1.3269,
+      "step": 64600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5495130887864138e-05,
+      "loss": 1.3314,
+      "step": 64700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.548816506220482e-05,
+      "loss": 1.3872,
+      "step": 64800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5481199236545507e-05,
+      "loss": 1.3365,
+      "step": 64900
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.547423341088619e-05,
+      "loss": 1.3184,
+      "step": 65000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.5467267585226877e-05,
+      "loss": 1.2843,
+      "step": 65100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.5460301759567563e-05,
+      "loss": 1.3652,
+      "step": 65200
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.5453335933908246e-05,
+      "loss": 1.3367,
+      "step": 65300
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.5446370108248932e-05,
+      "loss": 1.3429,
+      "step": 65400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.5439404282589615e-05,
+      "loss": 1.3395,
+      "step": 65500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.5432438456930302e-05,
+      "loss": 1.329,
+      "step": 65600
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.5425542289527578e-05,
+      "loss": 1.31,
+      "step": 65700
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.541857646386826e-05,
+      "loss": 1.3535,
+      "step": 65800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.5411610638208948e-05,
+      "loss": 1.3204,
+      "step": 65900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.540464481254963e-05,
+      "loss": 1.3465,
+      "step": 66000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.5397678986890317e-05,
+      "loss": 1.3004,
+      "step": 66100
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.5390713161231003e-05,
+      "loss": 1.3113,
+      "step": 66200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.5383747335571686e-05,
+      "loss": 1.3437,
+      "step": 66300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5376781509912373e-05,
+      "loss": 1.3098,
+      "step": 66400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5369815684253056e-05,
+      "loss": 1.3631,
+      "step": 66500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5362849858593742e-05,
+      "loss": 1.3486,
+      "step": 66600
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5355884032934425e-05,
+      "loss": 1.3174,
+      "step": 66700
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.534891820727511e-05,
+      "loss": 1.3509,
+      "step": 66800
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5341952381615794e-05,
+      "loss": 1.3222,
+      "step": 66900
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.533498655595648e-05,
+      "loss": 1.3343,
+      "step": 67000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.532802073029716e-05,
+      "loss": 1.341,
+      "step": 67100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5321054904637847e-05,
+      "loss": 1.3564,
+      "step": 67200
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.531408907897853e-05,
+      "loss": 1.3297,
+      "step": 67300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5307123253319216e-05,
+      "loss": 1.3156,
+      "step": 67400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.53001574276599e-05,
+      "loss": 1.309,
+      "step": 67500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5293191602000585e-05,
+      "loss": 1.2972,
+      "step": 67600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.528622577634127e-05,
+      "loss": 1.2935,
+      "step": 67700
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5279259950681955e-05,
+      "loss": 1.3517,
+      "step": 67800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5272294125022638e-05,
+      "loss": 1.3298,
+      "step": 67900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5265328299363324e-05,
+      "loss": 1.3111,
+      "step": 68000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5258362473704007e-05,
+      "loss": 1.2882,
+      "step": 68100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5251396648044694e-05,
+      "loss": 1.3066,
+      "step": 68200
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5244430822385377e-05,
+      "loss": 1.3078,
+      "step": 68300
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5237464996726063e-05,
+      "loss": 1.3353,
+      "step": 68400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.523049917106675e-05,
+      "loss": 1.3302,
+      "step": 68500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.5223533345407432e-05,
+      "loss": 1.3026,
+      "step": 68600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.521656751974812e-05,
+      "loss": 1.3094,
+      "step": 68700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.52096016940888e-05,
+      "loss": 1.301,
+      "step": 68800
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.5202635868429488e-05,
+      "loss": 1.3381,
+      "step": 68900
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.519567004277017e-05,
+      "loss": 1.296,
+      "step": 69000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.5188704217110857e-05,
+      "loss": 1.3327,
+      "step": 69100
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.518173839145154e-05,
+      "loss": 1.3504,
+      "step": 69200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5174772565792227e-05,
+      "loss": 1.329,
+      "step": 69300
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5167806740132906e-05,
+      "loss": 1.3304,
+      "step": 69400
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5160840914473593e-05,
+      "loss": 1.3442,
+      "step": 69500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5153875088814276e-05,
+      "loss": 1.2983,
+      "step": 69600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5146909263154962e-05,
+      "loss": 1.3069,
+      "step": 69700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5139943437495645e-05,
+      "loss": 1.3425,
+      "step": 69800
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5133116928349515e-05,
+      "loss": 1.2985,
+      "step": 69900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.51261511026902e-05,
+      "loss": 1.3073,
+      "step": 70000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5119254935287478e-05,
+      "loss": 1.3266,
+      "step": 70100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5112289109628164e-05,
+      "loss": 1.3149,
+      "step": 70200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5105323283968847e-05,
+      "loss": 1.352,
+      "step": 70300
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5098357458309534e-05,
+      "loss": 1.3348,
+      "step": 70400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5091391632650217e-05,
+      "loss": 1.3357,
+      "step": 70500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5084425806990903e-05,
+      "loss": 1.3143,
+      "step": 70600
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5077459981331586e-05,
+      "loss": 1.3632,
+      "step": 70700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5070494155672272e-05,
+      "loss": 1.3142,
+      "step": 70800
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.506352833001296e-05,
+      "loss": 1.2954,
+      "step": 70900
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.505656250435364e-05,
+      "loss": 1.3081,
+      "step": 71000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5049596678694328e-05,
+      "loss": 1.2836,
+      "step": 71100
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.504263085303501e-05,
+      "loss": 1.3663,
+      "step": 71200
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5035665027375697e-05,
+      "loss": 1.3225,
+      "step": 71300
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.502869920171638e-05,
+      "loss": 1.2865,
+      "step": 71400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5021733376057067e-05,
+      "loss": 1.3454,
+      "step": 71500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.501476755039775e-05,
+      "loss": 1.3092,
+      "step": 71600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5007801724738436e-05,
+      "loss": 1.2949,
+      "step": 71700
+    },
+    {
+      "epoch": 1.0,
+      "eval_gen_len": 20.0,
+      "eval_loss": 1.208003044128418,
+      "eval_rouge1": 11.7171,
+      "eval_rouge2": 3.3284,
+      "eval_rougeL": 11.3209,
+      "eval_rougeLsum": 11.4022,
+      "eval_runtime": 1534.4697,
+      "eval_samples_per_second": 8.712,
+      "eval_steps_per_second": 2.178,
+      "step": 71779
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.500083589907912e-05,
+      "loss": 1.2763,
+      "step": 71800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4993870073419805e-05,
+      "loss": 1.2162,
+      "step": 71900
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.498690424776049e-05,
+      "loss": 1.1719,
+      "step": 72000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.497993842210117e-05,
+      "loss": 1.2228,
+      "step": 72100
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4972972596441854e-05,
+      "loss": 1.2252,
+      "step": 72200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.496600677078254e-05,
+      "loss": 1.1962,
+      "step": 72300
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4959040945123224e-05,
+      "loss": 1.1819,
+      "step": 72400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.495207511946391e-05,
+      "loss": 1.1982,
+      "step": 72500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4945109293804593e-05,
+      "loss": 1.2265,
+      "step": 72600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.493814346814528e-05,
+      "loss": 1.1959,
+      "step": 72700
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4931177642485963e-05,
+      "loss": 1.176,
+      "step": 72800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.492421181682665e-05,
+      "loss": 1.2382,
+      "step": 72900
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4917245991167332e-05,
+      "loss": 1.2064,
+      "step": 73000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4910280165508018e-05,
+      "loss": 1.2151,
+      "step": 73100
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.49033143398487e-05,
+      "loss": 1.1908,
+      "step": 73200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4896348514189388e-05,
+      "loss": 1.2045,
+      "step": 73300
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4889382688530074e-05,
+      "loss": 1.2272,
+      "step": 73400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4882416862870757e-05,
+      "loss": 1.1965,
+      "step": 73500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.4875451037211443e-05,
+      "loss": 1.2207,
+      "step": 73600
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.4868485211552126e-05,
+      "loss": 1.1842,
+      "step": 73700
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.4861519385892813e-05,
+      "loss": 1.2278,
+      "step": 73800
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.4854553560233496e-05,
+      "loss": 1.2294,
+      "step": 73900
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.4847587734574182e-05,
+      "loss": 1.2443,
+      "step": 74000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.4840621908914865e-05,
+      "loss": 1.2025,
+      "step": 74100
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.483365608325555e-05,
+      "loss": 1.2018,
+      "step": 74200
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.4826690257596234e-05,
+      "loss": 1.202,
+      "step": 74300
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.4819724431936917e-05,
+      "loss": 1.1752,
+      "step": 74400
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.48127586062776e-05,
+      "loss": 1.2313,
+      "step": 74500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.4805792780618287e-05,
+      "loss": 1.1882,
+      "step": 74600
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.479882695495897e-05,
+      "loss": 1.2348,
+      "step": 74700
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.4791861129299656e-05,
+      "loss": 1.2377,
+      "step": 74800
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.478489530364034e-05,
+      "loss": 1.1769,
+      "step": 74900
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.4777929477981026e-05,
+      "loss": 1.2058,
+      "step": 75000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.477096365232171e-05,
+      "loss": 1.1412,
+      "step": 75100
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.4763997826662395e-05,
+      "loss": 1.2086,
+      "step": 75200
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.4757032001003078e-05,
+      "loss": 1.2231,
+      "step": 75300
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.4750066175343764e-05,
+      "loss": 1.2127,
+      "step": 75400
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.4743100349684447e-05,
+      "loss": 1.2203,
+      "step": 75500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.4736134524025134e-05,
+      "loss": 1.2357,
+      "step": 75600
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.4729168698365817e-05,
+      "loss": 1.1935,
+      "step": 75700
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.4722202872706503e-05,
+      "loss": 1.242,
+      "step": 75800
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.471523704704719e-05,
+      "loss": 1.2526,
+      "step": 75900
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.4708271221387872e-05,
+      "loss": 1.2058,
+      "step": 76000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.470137505398515e-05,
+      "loss": 1.2012,
+      "step": 76100
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.4694409228325835e-05,
+      "loss": 1.2217,
+      "step": 76200
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.4687443402666518e-05,
+      "loss": 1.2121,
+      "step": 76300
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.4680477577007205e-05,
+      "loss": 1.2092,
+      "step": 76400
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.4673511751347888e-05,
+      "loss": 1.2415,
+      "step": 76500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.4666545925688574e-05,
+      "loss": 1.219,
+      "step": 76600
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.465958010002926e-05,
+      "loss": 1.2309,
+      "step": 76700
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.465261427436994e-05,
+      "loss": 1.2061,
+      "step": 76800
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.4645648448710626e-05,
+      "loss": 1.2408,
+      "step": 76900
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.463868262305131e-05,
+      "loss": 1.1982,
+      "step": 77000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.4631716797391996e-05,
+      "loss": 1.2145,
+      "step": 77100
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.462475097173268e-05,
+      "loss": 1.2393,
+      "step": 77200
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.4617785146073365e-05,
+      "loss": 1.2282,
+      "step": 77300
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.4610819320414048e-05,
+      "loss": 1.2315,
+      "step": 77400
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.4603853494754734e-05,
+      "loss": 1.1694,
+      "step": 77500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.4596887669095417e-05,
+      "loss": 1.2348,
+      "step": 77600
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.4589921843436104e-05,
+      "loss": 1.2112,
+      "step": 77700
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.4582956017776787e-05,
+      "loss": 1.1807,
+      "step": 77800
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.4575990192117473e-05,
+      "loss": 1.197,
+      "step": 77900
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.4569024366458156e-05,
+      "loss": 1.2154,
+      "step": 78000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.4562058540798843e-05,
+      "loss": 1.2056,
+      "step": 78100
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.455516237339612e-05,
+      "loss": 1.23,
+      "step": 78200
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.4548196547736805e-05,
+      "loss": 1.2178,
+      "step": 78300
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.4541230722077488e-05,
+      "loss": 1.1859,
+      "step": 78400
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.4534264896418175e-05,
+      "loss": 1.2113,
+      "step": 78500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.4527299070758858e-05,
+      "loss": 1.1367,
+      "step": 78600
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.4520333245099544e-05,
+      "loss": 1.2035,
+      "step": 78700
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.4513367419440227e-05,
+      "loss": 1.1868,
+      "step": 78800
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.4506401593780913e-05,
+      "loss": 1.2222,
+      "step": 78900
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.4499435768121596e-05,
+      "loss": 1.1801,
+      "step": 79000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.4492469942462283e-05,
+      "loss": 1.2566,
+      "step": 79100
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.4485504116802962e-05,
+      "loss": 1.2105,
+      "step": 79200
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.447853829114365e-05,
+      "loss": 1.209,
+      "step": 79300
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.4471642123740925e-05,
+      "loss": 1.2341,
+      "step": 79400
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.446467629808161e-05,
+      "loss": 1.212,
+      "step": 79500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.4457710472422295e-05,
+      "loss": 1.2183,
+      "step": 79600
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.445074464676298e-05,
+      "loss": 1.2142,
+      "step": 79700
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.4443778821103664e-05,
+      "loss": 1.2149,
+      "step": 79800
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.443681299544435e-05,
+      "loss": 1.2276,
+      "step": 79900
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.4429847169785033e-05,
+      "loss": 1.2213,
+      "step": 80000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.442288134412572e-05,
+      "loss": 1.2164,
+      "step": 80100
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.4415915518466403e-05,
+      "loss": 1.2111,
+      "step": 80200
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.440894969280709e-05,
+      "loss": 1.2367,
+      "step": 80300
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.4401983867147772e-05,
+      "loss": 1.1896,
+      "step": 80400
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.439501804148846e-05,
+      "loss": 1.2185,
+      "step": 80500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.438805221582914e-05,
+      "loss": 1.2104,
+      "step": 80600
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.4381086390169828e-05,
+      "loss": 1.1925,
+      "step": 80700
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.4374120564510514e-05,
+      "loss": 1.2154,
+      "step": 80800
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.4367154738851197e-05,
+      "loss": 1.189,
+      "step": 80900
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.4360188913191884e-05,
+      "loss": 1.2177,
+      "step": 81000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.4353223087532566e-05,
+      "loss": 1.2071,
+      "step": 81100
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.4346257261873253e-05,
+      "loss": 1.2237,
+      "step": 81200
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.4339291436213936e-05,
+      "loss": 1.2119,
+      "step": 81300
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.4332325610554622e-05,
+      "loss": 1.2448,
+      "step": 81400
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.4325359784895305e-05,
+      "loss": 1.1896,
+      "step": 81500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.431839395923599e-05,
+      "loss": 1.1852,
+      "step": 81600
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.431142813357667e-05,
+      "loss": 1.2208,
+      "step": 81700
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.4304462307917358e-05,
+      "loss": 1.2055,
+      "step": 81800
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.429749648225804e-05,
+      "loss": 1.2003,
+      "step": 81900
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.4290530656598727e-05,
+      "loss": 1.2091,
+      "step": 82000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.428356483093941e-05,
+      "loss": 1.1914,
+      "step": 82100
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.4276599005280096e-05,
+      "loss": 1.2248,
+      "step": 82200
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.426963317962078e-05,
+      "loss": 1.2079,
+      "step": 82300
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.4262667353961466e-05,
+      "loss": 1.2115,
+      "step": 82400
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.425570152830215e-05,
+      "loss": 1.1846,
+      "step": 82500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.4248735702642835e-05,
+      "loss": 1.196,
+      "step": 82600
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.4241769876983518e-05,
+      "loss": 1.2216,
+      "step": 82700
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.4234804051324204e-05,
+      "loss": 1.1833,
+      "step": 82800
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.4227838225664887e-05,
+      "loss": 1.2104,
+      "step": 82900
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.4220872400005574e-05,
+      "loss": 1.2132,
+      "step": 83000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.4213906574346257e-05,
+      "loss": 1.2086,
+      "step": 83100
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.4206940748686943e-05,
+      "loss": 1.1941,
+      "step": 83200
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.419997492302763e-05,
+      "loss": 1.191,
+      "step": 83300
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.4193009097368313e-05,
+      "loss": 1.1874,
+      "step": 83400
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.4186043271709e-05,
+      "loss": 1.2118,
+      "step": 83500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.4179077446049682e-05,
+      "loss": 1.1804,
+      "step": 83600
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.4172111620390368e-05,
+      "loss": 1.1734,
+      "step": 83700
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.4165215452987645e-05,
+      "loss": 1.1996,
+      "step": 83800
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.4158249627328328e-05,
+      "loss": 1.156,
+      "step": 83900
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.4151283801669014e-05,
+      "loss": 1.2506,
+      "step": 84000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.4144317976009697e-05,
+      "loss": 1.2064,
+      "step": 84100
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.413735215035038e-05,
+      "loss": 1.1956,
+      "step": 84200
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.4130386324691066e-05,
+      "loss": 1.2134,
+      "step": 84300
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.412342049903175e-05,
+      "loss": 1.1889,
+      "step": 84400
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.4116454673372436e-05,
+      "loss": 1.2413,
+      "step": 84500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.410948884771312e-05,
+      "loss": 1.1893,
+      "step": 84600
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.4102523022053805e-05,
+      "loss": 1.2346,
+      "step": 84700
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.4095557196394488e-05,
+      "loss": 1.2005,
+      "step": 84800
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.4088591370735175e-05,
+      "loss": 1.2187,
+      "step": 84900
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.4081625545075858e-05,
+      "loss": 1.2475,
+      "step": 85000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.4074659719416544e-05,
+      "loss": 1.1655,
+      "step": 85100
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.4067693893757227e-05,
+      "loss": 1.2359,
+      "step": 85200
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.4060728068097913e-05,
+      "loss": 1.2474,
+      "step": 85300
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.4053762242438596e-05,
+      "loss": 1.2113,
+      "step": 85400
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.4046796416779283e-05,
+      "loss": 1.2017,
+      "step": 85500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.4039830591119966e-05,
+      "loss": 1.1962,
+      "step": 85600
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.4032864765460652e-05,
+      "loss": 1.2242,
+      "step": 85700
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.4025898939801335e-05,
+      "loss": 1.1803,
+      "step": 85800
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.4019002772398615e-05,
+      "loss": 1.2171,
+      "step": 85900
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.4012036946739298e-05,
+      "loss": 1.2158,
+      "step": 86000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.4005071121079984e-05,
+      "loss": 1.2223,
+      "step": 86100
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.3998105295420667e-05,
+      "loss": 1.234,
+      "step": 86200
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.3991139469761354e-05,
+      "loss": 1.1993,
+      "step": 86300
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.3984173644102037e-05,
+      "loss": 1.2003,
+      "step": 86400
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.397720781844272e-05,
+      "loss": 1.2102,
+      "step": 86500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.3970241992783402e-05,
+      "loss": 1.1924,
+      "step": 86600
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.396327616712409e-05,
+      "loss": 1.2074,
+      "step": 86700
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.3956310341464772e-05,
+      "loss": 1.2076,
+      "step": 86800
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.3949344515805458e-05,
+      "loss": 1.1719,
+      "step": 86900
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.394237869014614e-05,
+      "loss": 1.171,
+      "step": 87000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.3935412864486828e-05,
+      "loss": 1.2379,
+      "step": 87100
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.3928447038827514e-05,
+      "loss": 1.1785,
+      "step": 87200
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.3921481213168197e-05,
+      "loss": 1.2119,
+      "step": 87300
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.3914515387508883e-05,
+      "loss": 1.1969,
+      "step": 87400
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.3907549561849566e-05,
+      "loss": 1.1893,
+      "step": 87500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.3900583736190253e-05,
+      "loss": 1.2106,
+      "step": 87600
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.3893617910530936e-05,
+      "loss": 1.2031,
+      "step": 87700
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.3886652084871622e-05,
+      "loss": 1.2224,
+      "step": 87800
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.3879686259212305e-05,
+      "loss": 1.1619,
+      "step": 87900
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.387279009180958e-05,
+      "loss": 1.2235,
+      "step": 88000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.3865824266150268e-05,
+      "loss": 1.2187,
+      "step": 88100
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.3858858440490954e-05,
+      "loss": 1.1788,
+      "step": 88200
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.385196227308823e-05,
+      "loss": 1.1814,
+      "step": 88300
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.3844996447428914e-05,
+      "loss": 1.1988,
+      "step": 88400
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.38380306217696e-05,
+      "loss": 1.1783,
+      "step": 88500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.3831064796110283e-05,
+      "loss": 1.1934,
+      "step": 88600
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.382409897045097e-05,
+      "loss": 1.2138,
+      "step": 88700
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.3817133144791652e-05,
+      "loss": 1.2141,
+      "step": 88800
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.381016731913234e-05,
+      "loss": 1.262,
+      "step": 88900
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.3803201493473025e-05,
+      "loss": 1.2314,
+      "step": 89000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.3796235667813705e-05,
+      "loss": 1.1988,
+      "step": 89100
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.378926984215439e-05,
+      "loss": 1.2241,
+      "step": 89200
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.3782304016495074e-05,
+      "loss": 1.2096,
+      "step": 89300
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.377533819083576e-05,
+      "loss": 1.2187,
+      "step": 89400
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.3768372365176443e-05,
+      "loss": 1.1916,
+      "step": 89500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.376140653951713e-05,
+      "loss": 1.1877,
+      "step": 89600
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.3754440713857813e-05,
+      "loss": 1.2055,
+      "step": 89700
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.37474748881985e-05,
+      "loss": 1.2144,
+      "step": 89800
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.3740509062539182e-05,
+      "loss": 1.2193,
+      "step": 89900
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.373354323687987e-05,
+      "loss": 1.1965,
+      "step": 90000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.372657741122055e-05,
+      "loss": 1.2008,
+      "step": 90100
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.3719611585561238e-05,
+      "loss": 1.2098,
+      "step": 90200
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.371264575990192e-05,
+      "loss": 1.1948,
+      "step": 90300
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.3705679934242607e-05,
+      "loss": 1.1907,
+      "step": 90400
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.369871410858329e-05,
+      "loss": 1.2195,
+      "step": 90500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.3691748282923977e-05,
+      "loss": 1.195,
+      "step": 90600
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.368478245726466e-05,
+      "loss": 1.2201,
+      "step": 90700
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.3677816631605346e-05,
+      "loss": 1.1833,
+      "step": 90800
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.367085080594603e-05,
+      "loss": 1.2109,
+      "step": 90900
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.3663884980286715e-05,
+      "loss": 1.2184,
+      "step": 91000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.36569191546274e-05,
+      "loss": 1.2485,
+      "step": 91100
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.3649953328968085e-05,
+      "loss": 1.2013,
+      "step": 91200
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.3642987503308768e-05,
+      "loss": 1.2579,
+      "step": 91300
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.363602167764945e-05,
+      "loss": 1.2107,
+      "step": 91400
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.3629055851990137e-05,
+      "loss": 1.1716,
+      "step": 91500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.362209002633082e-05,
+      "loss": 1.2439,
+      "step": 91600
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.3615124200671507e-05,
+      "loss": 1.1907,
+      "step": 91700
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.360815837501219e-05,
+      "loss": 1.2171,
+      "step": 91800
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.3601192549352876e-05,
+      "loss": 1.1806,
+      "step": 91900
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.3594296381950152e-05,
+      "loss": 1.1918,
+      "step": 92000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.358733055629084e-05,
+      "loss": 1.2124,
+      "step": 92100
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.358036473063152e-05,
+      "loss": 1.2193,
+      "step": 92200
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.3573398904972208e-05,
+      "loss": 1.1929,
+      "step": 92300
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.356643307931289e-05,
+      "loss": 1.2022,
+      "step": 92400
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.3559467253653577e-05,
+      "loss": 1.2052,
+      "step": 92500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.355250142799426e-05,
+      "loss": 1.1878,
+      "step": 92600
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.3545535602334947e-05,
+      "loss": 1.1895,
+      "step": 92700
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.353856977667563e-05,
+      "loss": 1.2043,
+      "step": 92800
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.3531603951016316e-05,
+      "loss": 1.1816,
+      "step": 92900
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.3524638125357e-05,
+      "loss": 1.234,
+      "step": 93000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.3517672299697686e-05,
+      "loss": 1.2092,
+      "step": 93100
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.351070647403837e-05,
+      "loss": 1.2086,
+      "step": 93200
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.3503740648379055e-05,
+      "loss": 1.1953,
+      "step": 93300
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.3496774822719738e-05,
+      "loss": 1.2012,
+      "step": 93400
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.3489808997060424e-05,
+      "loss": 1.2169,
+      "step": 93500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.3482843171401107e-05,
+      "loss": 1.2498,
+      "step": 93600
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.3475877345741794e-05,
+      "loss": 1.2184,
+      "step": 93700
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.3468911520082473e-05,
+      "loss": 1.1944,
+      "step": 93800
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.346194569442316e-05,
+      "loss": 1.2077,
+      "step": 93900
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.3454979868763843e-05,
+      "loss": 1.1873,
+      "step": 94000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.344801404310453e-05,
+      "loss": 1.206,
+      "step": 94100
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.3441048217445212e-05,
+      "loss": 1.1979,
+      "step": 94200
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.34340823917859e-05,
+      "loss": 1.2447,
+      "step": 94300
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.342711656612658e-05,
+      "loss": 1.1973,
+      "step": 94400
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.3420150740467268e-05,
+      "loss": 1.2511,
+      "step": 94500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.3413184914807954e-05,
+      "loss": 1.2381,
+      "step": 94600
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.3406219089148637e-05,
+      "loss": 1.2367,
+      "step": 94700
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.3399253263489323e-05,
+      "loss": 1.2305,
+      "step": 94800
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.3392287437830006e-05,
+      "loss": 1.2151,
+      "step": 94900
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.3385321612170693e-05,
+      "loss": 1.2041,
+      "step": 95000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.3378355786511376e-05,
+      "loss": 1.2184,
+      "step": 95100
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.3371389960852062e-05,
+      "loss": 1.2311,
+      "step": 95200
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.3364424135192745e-05,
+      "loss": 1.186,
+      "step": 95300
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.335745830953343e-05,
+      "loss": 1.1747,
+      "step": 95400
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.3350492483874115e-05,
+      "loss": 1.2108,
+      "step": 95500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.33435266582148e-05,
+      "loss": 1.197,
+      "step": 95600
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.3336560832555484e-05,
+      "loss": 1.2411,
+      "step": 95700
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.332959500689617e-05,
+      "loss": 1.1862,
+      "step": 95800
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.3322629181236853e-05,
+      "loss": 1.2287,
+      "step": 95900
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.3315663355577536e-05,
+      "loss": 1.1929,
+      "step": 96000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.3308767188174816e-05,
+      "loss": 1.1839,
+      "step": 96100
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.33018013625155e-05,
+      "loss": 1.2291,
+      "step": 96200
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.3294835536856182e-05,
+      "loss": 1.2131,
+      "step": 96300
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.328786971119687e-05,
+      "loss": 1.2304,
+      "step": 96400
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.328090388553755e-05,
+      "loss": 1.1881,
+      "step": 96500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.3273938059878238e-05,
+      "loss": 1.2102,
+      "step": 96600
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.326697223421892e-05,
+      "loss": 1.1755,
+      "step": 96700
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.3260006408559607e-05,
+      "loss": 1.2057,
+      "step": 96800
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.325304058290029e-05,
+      "loss": 1.2049,
+      "step": 96900
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.3246074757240977e-05,
+      "loss": 1.2509,
+      "step": 97000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.323910893158166e-05,
+      "loss": 1.1966,
+      "step": 97100
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.3232143105922346e-05,
+      "loss": 1.1735,
+      "step": 97200
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.322517728026303e-05,
+      "loss": 1.1989,
+      "step": 97300
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.321828111286031e-05,
+      "loss": 1.2423,
+      "step": 97400
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.321131528720099e-05,
+      "loss": 1.2197,
+      "step": 97500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.3204349461541678e-05,
+      "loss": 1.2015,
+      "step": 97600
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.319738363588236e-05,
+      "loss": 1.2008,
+      "step": 97700
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.3190417810223047e-05,
+      "loss": 1.1893,
+      "step": 97800
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.318345198456373e-05,
+      "loss": 1.2326,
+      "step": 97900
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.3176486158904417e-05,
+      "loss": 1.2319,
+      "step": 98000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.31695203332451e-05,
+      "loss": 1.2047,
+      "step": 98100
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.3162554507585786e-05,
+      "loss": 1.21,
+      "step": 98200
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.315558868192647e-05,
+      "loss": 1.1928,
+      "step": 98300
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.3148622856267156e-05,
+      "loss": 1.2059,
+      "step": 98400
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.314165703060784e-05,
+      "loss": 1.2117,
+      "step": 98500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.313469120494852e-05,
+      "loss": 1.2217,
+      "step": 98600
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.3127725379289208e-05,
+      "loss": 1.191,
+      "step": 98700
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.312075955362989e-05,
+      "loss": 1.2245,
+      "step": 98800
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.3113793727970577e-05,
+      "loss": 1.2086,
+      "step": 98900
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.310682790231126e-05,
+      "loss": 1.1783,
+      "step": 99000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.3099862076651947e-05,
+      "loss": 1.2024,
+      "step": 99100
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.309289625099263e-05,
+      "loss": 1.1831,
+      "step": 99200
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.3085930425333316e-05,
+      "loss": 1.2045,
+      "step": 99300
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.3078964599674e-05,
+      "loss": 1.1953,
+      "step": 99400
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.3071998774014685e-05,
+      "loss": 1.1615,
+      "step": 99500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.306503294835537e-05,
+      "loss": 1.2064,
+      "step": 99600
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.3058067122696055e-05,
+      "loss": 1.1868,
+      "step": 99700
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.305117095529333e-05,
+      "loss": 1.1918,
+      "step": 99800
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.3044205129634018e-05,
+      "loss": 1.2122,
+      "step": 99900
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.30372393039747e-05,
+      "loss": 1.1998,
+      "step": 100000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.3030273478315387e-05,
+      "loss": 1.2122,
+      "step": 100100
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.302330765265607e-05,
+      "loss": 1.1924,
+      "step": 100200
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.3016341826996756e-05,
+      "loss": 1.2294,
+      "step": 100300
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.300937600133744e-05,
+      "loss": 1.198,
+      "step": 100400
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.3002410175678126e-05,
+      "loss": 1.2307,
+      "step": 100500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.299544435001881e-05,
+      "loss": 1.1966,
+      "step": 100600
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.2988478524359495e-05,
+      "loss": 1.2221,
+      "step": 100700
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.2981512698700178e-05,
+      "loss": 1.2524,
+      "step": 100800
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.2974546873040864e-05,
+      "loss": 1.2349,
+      "step": 100900
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.2967581047381544e-05,
+      "loss": 1.185,
+      "step": 101000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.296061522172223e-05,
+      "loss": 1.1917,
+      "step": 101100
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.2953649396062913e-05,
+      "loss": 1.1746,
+      "step": 101200
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.29466835704036e-05,
+      "loss": 1.1778,
+      "step": 101300
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.2939717744744283e-05,
+      "loss": 1.1927,
+      "step": 101400
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.293275191908497e-05,
+      "loss": 1.2341,
+      "step": 101500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.2925786093425652e-05,
+      "loss": 1.209,
+      "step": 101600
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.291882026776634e-05,
+      "loss": 1.2499,
+      "step": 101700
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.291185444210702e-05,
+      "loss": 1.189,
+      "step": 101800
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.2904888616447708e-05,
+      "loss": 1.1962,
+      "step": 101900
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.2897922790788394e-05,
+      "loss": 1.2561,
+      "step": 102000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.2890956965129077e-05,
+      "loss": 1.2059,
+      "step": 102100
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.2883991139469764e-05,
+      "loss": 1.1972,
+      "step": 102200
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.2877025313810447e-05,
+      "loss": 1.1873,
+      "step": 102300
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.2870059488151133e-05,
+      "loss": 1.2591,
+      "step": 102400
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.2863093662491816e-05,
+      "loss": 1.2266,
+      "step": 102500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.2856127836832502e-05,
+      "loss": 1.2194,
+      "step": 102600
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.2849162011173185e-05,
+      "loss": 1.2086,
+      "step": 102700
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.284219618551387e-05,
+      "loss": 1.2317,
+      "step": 102800
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.2835230359854555e-05,
+      "loss": 1.2056,
+      "step": 102900
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.282826453419524e-05,
+      "loss": 1.2364,
+      "step": 103000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.2821298708535924e-05,
+      "loss": 1.1771,
+      "step": 103100
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.281433288287661e-05,
+      "loss": 1.1979,
+      "step": 103200
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.280736705721729e-05,
+      "loss": 1.2382,
+      "step": 103300
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.2800401231557976e-05,
+      "loss": 1.1685,
+      "step": 103400
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.279343540589866e-05,
+      "loss": 1.2082,
+      "step": 103500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.2786469580239346e-05,
+      "loss": 1.2266,
+      "step": 103600
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.277950375458003e-05,
+      "loss": 1.2132,
+      "step": 103700
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.2772537928920715e-05,
+      "loss": 1.1876,
+      "step": 103800
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.2765572103261398e-05,
+      "loss": 1.222,
+      "step": 103900
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.2758606277602084e-05,
+      "loss": 1.2023,
+      "step": 104000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.2751640451942767e-05,
+      "loss": 1.1668,
+      "step": 104100
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.2744674626283454e-05,
+      "loss": 1.194,
+      "step": 104200
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.2737708800624137e-05,
+      "loss": 1.2058,
+      "step": 104300
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.2730742974964823e-05,
+      "loss": 1.185,
+      "step": 104400
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.27238468075621e-05,
+      "loss": 1.1681,
+      "step": 104500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.2716880981902786e-05,
+      "loss": 1.2321,
+      "step": 104600
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.270991515624347e-05,
+      "loss": 1.2104,
+      "step": 104700
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.2702949330584155e-05,
+      "loss": 1.1615,
+      "step": 104800
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.269598350492484e-05,
+      "loss": 1.2305,
+      "step": 104900
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.2689017679265525e-05,
+      "loss": 1.2212,
+      "step": 105000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.2682051853606208e-05,
+      "loss": 1.1814,
+      "step": 105100
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.2675086027946894e-05,
+      "loss": 1.2132,
+      "step": 105200
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.266812020228758e-05,
+      "loss": 1.1871,
+      "step": 105300
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.2661224034884857e-05,
+      "loss": 1.2022,
+      "step": 105400
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.265425820922554e-05,
+      "loss": 1.2129,
+      "step": 105500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.2647292383566226e-05,
+      "loss": 1.2058,
+      "step": 105600
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.264032655790691e-05,
+      "loss": 1.2106,
+      "step": 105700
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.2633360732247596e-05,
+      "loss": 1.1971,
+      "step": 105800
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.2626394906588275e-05,
+      "loss": 1.1881,
+      "step": 105900
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.261942908092896e-05,
+      "loss": 1.1922,
+      "step": 106000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.2612463255269648e-05,
+      "loss": 1.2067,
+      "step": 106100
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.260549742961033e-05,
+      "loss": 1.2249,
+      "step": 106200
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.2598531603951017e-05,
+      "loss": 1.1895,
+      "step": 106300
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.25915657782917e-05,
+      "loss": 1.1837,
+      "step": 106400
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.2584599952632387e-05,
+      "loss": 1.2043,
+      "step": 106500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.257763412697307e-05,
+      "loss": 1.2104,
+      "step": 106600
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.2570668301313756e-05,
+      "loss": 1.2394,
+      "step": 106700
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.256370247565444e-05,
+      "loss": 1.1965,
+      "step": 106800
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.2556736649995125e-05,
+      "loss": 1.2418,
+      "step": 106900
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.254977082433581e-05,
+      "loss": 1.2125,
+      "step": 107000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.2542804998676495e-05,
+      "loss": 1.2428,
+      "step": 107100
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.2535839173017178e-05,
+      "loss": 1.2245,
+      "step": 107200
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.2528873347357864e-05,
+      "loss": 1.1913,
+      "step": 107300
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.2521907521698547e-05,
+      "loss": 1.1843,
+      "step": 107400
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.2514941696039234e-05,
+      "loss": 1.2339,
+      "step": 107500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.2507975870379917e-05,
+      "loss": 1.1878,
+      "step": 107600
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.2501010044720603e-05,
+      "loss": 1.2216,
+      "step": 107700
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.2494044219061286e-05,
+      "loss": 1.1915,
+      "step": 107800
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.2487078393401972e-05,
+      "loss": 1.1568,
+      "step": 107900
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.2480112567742655e-05,
+      "loss": 1.2237,
+      "step": 108000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.2473146742083338e-05,
+      "loss": 1.1775,
+      "step": 108100
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.246618091642402e-05,
+      "loss": 1.2169,
+      "step": 108200
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.2459215090764708e-05,
+      "loss": 1.1817,
+      "step": 108300
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.2452249265105394e-05,
+      "loss": 1.2129,
+      "step": 108400
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.2445283439446077e-05,
+      "loss": 1.2335,
+      "step": 108500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.2438317613786763e-05,
+      "loss": 1.2294,
+      "step": 108600
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.2431351788127446e-05,
+      "loss": 1.1848,
+      "step": 108700
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.2424385962468133e-05,
+      "loss": 1.1672,
+      "step": 108800
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.2417420136808816e-05,
+      "loss": 1.194,
+      "step": 108900
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.2410454311149502e-05,
+      "loss": 1.2135,
+      "step": 109000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.2403488485490185e-05,
+      "loss": 1.2033,
+      "step": 109100
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.239659231808746e-05,
+      "loss": 1.2111,
+      "step": 109200
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.2389626492428148e-05,
+      "loss": 1.216,
+      "step": 109300
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.2382660666768834e-05,
+      "loss": 1.2526,
+      "step": 109400
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.2375694841109517e-05,
+      "loss": 1.172,
+      "step": 109500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.2368729015450204e-05,
+      "loss": 1.1855,
+      "step": 109600
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.2361763189790887e-05,
+      "loss": 1.2075,
+      "step": 109700
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.2354797364131573e-05,
+      "loss": 1.1811,
+      "step": 109800
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.2347831538472256e-05,
+      "loss": 1.1902,
+      "step": 109900
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.2340865712812942e-05,
+      "loss": 1.2013,
+      "step": 110000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.2333899887153625e-05,
+      "loss": 1.1856,
+      "step": 110100
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.2326934061494312e-05,
+      "loss": 1.1709,
+      "step": 110200
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.2319968235834995e-05,
+      "loss": 1.1757,
+      "step": 110300
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.231300241017568e-05,
+      "loss": 1.1821,
+      "step": 110400
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.230603658451636e-05,
+      "loss": 1.2171,
+      "step": 110500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.2299070758857047e-05,
+      "loss": 1.1986,
+      "step": 110600
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.229210493319773e-05,
+      "loss": 1.206,
+      "step": 110700
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.2285139107538416e-05,
+      "loss": 1.2128,
+      "step": 110800
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.22781732818791e-05,
+      "loss": 1.1754,
+      "step": 110900
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.2271207456219786e-05,
+      "loss": 1.172,
+      "step": 111000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.226424163056047e-05,
+      "loss": 1.1641,
+      "step": 111100
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.2257275804901155e-05,
+      "loss": 1.2322,
+      "step": 111200
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.2250309979241838e-05,
+      "loss": 1.215,
+      "step": 111300
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.2243344153582525e-05,
+      "loss": 1.2181,
+      "step": 111400
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.2236378327923208e-05,
+      "loss": 1.2111,
+      "step": 111500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.2229412502263894e-05,
+      "loss": 1.2139,
+      "step": 111600
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.222244667660458e-05,
+      "loss": 1.2265,
+      "step": 111700
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.2215480850945263e-05,
+      "loss": 1.1916,
+      "step": 111800
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.220851502528595e-05,
+      "loss": 1.2285,
+      "step": 111900
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.2201549199626633e-05,
+      "loss": 1.2045,
+      "step": 112000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.219458337396732e-05,
+      "loss": 1.1994,
+      "step": 112100
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.2187617548308002e-05,
+      "loss": 1.2142,
+      "step": 112200
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.218065172264869e-05,
+      "loss": 1.1919,
+      "step": 112300
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.217368589698937e-05,
+      "loss": 1.2523,
+      "step": 112400
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2166720071330058e-05,
+      "loss": 1.2212,
+      "step": 112500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.215975424567074e-05,
+      "loss": 1.2299,
+      "step": 112600
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2152788420011427e-05,
+      "loss": 1.2114,
+      "step": 112700
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2145822594352107e-05,
+      "loss": 1.2329,
+      "step": 112800
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2138856768692793e-05,
+      "loss": 1.2369,
+      "step": 112900
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2131890943033476e-05,
+      "loss": 1.2144,
+      "step": 113000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2124925117374162e-05,
+      "loss": 1.2427,
+      "step": 113100
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2117959291714845e-05,
+      "loss": 1.2138,
+      "step": 113200
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2110993466055532e-05,
+      "loss": 1.1789,
+      "step": 113300
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2104027640396215e-05,
+      "loss": 1.2021,
+      "step": 113400
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.20970618147369e-05,
+      "loss": 1.1692,
+      "step": 113500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2090165647334178e-05,
+      "loss": 1.1913,
+      "step": 113600
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2083199821674864e-05,
+      "loss": 1.2328,
+      "step": 113700
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.2076233996015547e-05,
+      "loss": 1.2098,
+      "step": 113800
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.2069268170356233e-05,
+      "loss": 1.2152,
+      "step": 113900
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.2062302344696916e-05,
+      "loss": 1.2237,
+      "step": 114000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.2055336519037603e-05,
+      "loss": 1.1828,
+      "step": 114100
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.2048370693378286e-05,
+      "loss": 1.1819,
+      "step": 114200
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.2041474525975566e-05,
+      "loss": 1.2185,
+      "step": 114300
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.203450870031625e-05,
+      "loss": 1.1798,
+      "step": 114400
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.2027542874656935e-05,
+      "loss": 1.2069,
+      "step": 114500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.2020577048997618e-05,
+      "loss": 1.1889,
+      "step": 114600
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.2013611223338304e-05,
+      "loss": 1.2132,
+      "step": 114700
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.2006645397678987e-05,
+      "loss": 1.2078,
+      "step": 114800
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.1999679572019674e-05,
+      "loss": 1.1855,
+      "step": 114900
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.1992713746360357e-05,
+      "loss": 1.181,
+      "step": 115000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.1985747920701043e-05,
+      "loss": 1.1746,
+      "step": 115100
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.1978782095041726e-05,
+      "loss": 1.1918,
+      "step": 115200
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.1971816269382412e-05,
+      "loss": 1.1932,
+      "step": 115300
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.1964850443723092e-05,
+      "loss": 1.2226,
+      "step": 115400
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.195788461806378e-05,
+      "loss": 1.2463,
+      "step": 115500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.195091879240446e-05,
+      "loss": 1.1943,
+      "step": 115600
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.1943952966745148e-05,
+      "loss": 1.1542,
+      "step": 115700
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.1936987141085834e-05,
+      "loss": 1.2038,
+      "step": 115800
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.1930021315426517e-05,
+      "loss": 1.1774,
+      "step": 115900
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.1923055489767203e-05,
+      "loss": 1.2212,
+      "step": 116000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.1916089664107886e-05,
+      "loss": 1.1795,
+      "step": 116100
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.1909123838448573e-05,
+      "loss": 1.2041,
+      "step": 116200
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.1902158012789256e-05,
+      "loss": 1.1721,
+      "step": 116300
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.1895192187129942e-05,
+      "loss": 1.2156,
+      "step": 116400
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.1888226361470625e-05,
+      "loss": 1.1982,
+      "step": 116500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.188126053581131e-05,
+      "loss": 1.2099,
+      "step": 116600
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.1874294710151995e-05,
+      "loss": 1.1936,
+      "step": 116700
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.1867398542749274e-05,
+      "loss": 1.1853,
+      "step": 116800
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.1860432717089957e-05,
+      "loss": 1.2059,
+      "step": 116900
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.1853466891430644e-05,
+      "loss": 1.1683,
+      "step": 117000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.1846501065771327e-05,
+      "loss": 1.1649,
+      "step": 117100
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.1839535240112013e-05,
+      "loss": 1.1969,
+      "step": 117200
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.1832569414452696e-05,
+      "loss": 1.1924,
+      "step": 117300
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.1825603588793382e-05,
+      "loss": 1.2366,
+      "step": 117400
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.1818637763134065e-05,
+      "loss": 1.1937,
+      "step": 117500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.1811671937474752e-05,
+      "loss": 1.1794,
+      "step": 117600
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.1804706111815435e-05,
+      "loss": 1.2114,
+      "step": 117700
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.1797740286156118e-05,
+      "loss": 1.2059,
+      "step": 117800
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.17907744604968e-05,
+      "loss": 1.1833,
+      "step": 117900
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.1783808634837487e-05,
+      "loss": 1.1492,
+      "step": 118000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.177684280917817e-05,
+      "loss": 1.1716,
+      "step": 118100
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.1769876983518857e-05,
+      "loss": 1.156,
+      "step": 118200
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.176291115785954e-05,
+      "loss": 1.1807,
+      "step": 118300
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.1755945332200226e-05,
+      "loss": 1.1942,
+      "step": 118400
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.174897950654091e-05,
+      "loss": 1.2004,
+      "step": 118500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.1742013680881595e-05,
+      "loss": 1.227,
+      "step": 118600
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.1735047855222278e-05,
+      "loss": 1.2213,
+      "step": 118700
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1728082029562965e-05,
+      "loss": 1.2016,
+      "step": 118800
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1721116203903648e-05,
+      "loss": 1.2137,
+      "step": 118900
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1714150378244334e-05,
+      "loss": 1.1818,
+      "step": 119000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.170718455258502e-05,
+      "loss": 1.182,
+      "step": 119100
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1700218726925703e-05,
+      "loss": 1.22,
+      "step": 119200
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.169325290126639e-05,
+      "loss": 1.1563,
+      "step": 119300
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1686287075607073e-05,
+      "loss": 1.2058,
+      "step": 119400
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.167932124994776e-05,
+      "loss": 1.2657,
+      "step": 119500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1672355424288442e-05,
+      "loss": 1.182,
+      "step": 119600
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.166538959862913e-05,
+      "loss": 1.1985,
+      "step": 119700
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.165842377296981e-05,
+      "loss": 1.1724,
+      "step": 119800
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1651457947310498e-05,
+      "loss": 1.1759,
+      "step": 119900
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.164449212165118e-05,
+      "loss": 1.208,
+      "step": 120000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1637526295991864e-05,
+      "loss": 1.2373,
+      "step": 120100
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1630560470332547e-05,
+      "loss": 1.1772,
+      "step": 120200
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.1623594644673233e-05,
+      "loss": 1.1855,
+      "step": 120300
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.1616628819013916e-05,
+      "loss": 1.2035,
+      "step": 120400
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.1609662993354603e-05,
+      "loss": 1.1757,
+      "step": 120500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.1602697167695286e-05,
+      "loss": 1.2168,
+      "step": 120600
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.1595731342035972e-05,
+      "loss": 1.1868,
+      "step": 120700
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.1588765516376655e-05,
+      "loss": 1.1752,
+      "step": 120800
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.158179969071734e-05,
+      "loss": 1.1586,
+      "step": 120900
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1574903523314618e-05,
+      "loss": 1.1997,
+      "step": 121000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1567937697655304e-05,
+      "loss": 1.1929,
+      "step": 121100
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1560971871995987e-05,
+      "loss": 1.213,
+      "step": 121200
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1554006046336673e-05,
+      "loss": 1.2179,
+      "step": 121300
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1547040220677356e-05,
+      "loss": 1.1815,
+      "step": 121400
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1540074395018043e-05,
+      "loss": 1.2464,
+      "step": 121500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1533108569358726e-05,
+      "loss": 1.2206,
+      "step": 121600
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1526142743699412e-05,
+      "loss": 1.2226,
+      "step": 121700
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1519176918040095e-05,
+      "loss": 1.2398,
+      "step": 121800
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.151221109238078e-05,
+      "loss": 1.1899,
+      "step": 121900
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1505245266721465e-05,
+      "loss": 1.1707,
+      "step": 122000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.149827944106215e-05,
+      "loss": 1.206,
+      "step": 122100
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1491313615402834e-05,
+      "loss": 1.1963,
+      "step": 122200
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.148434778974352e-05,
+      "loss": 1.1995,
+      "step": 122300
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1477451622340797e-05,
+      "loss": 1.1909,
+      "step": 122400
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1470485796681483e-05,
+      "loss": 1.1914,
+      "step": 122500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1463519971022166e-05,
+      "loss": 1.2262,
+      "step": 122600
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.145655414536285e-05,
+      "loss": 1.1817,
+      "step": 122700
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1449588319703532e-05,
+      "loss": 1.1918,
+      "step": 122800
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.144262249404422e-05,
+      "loss": 1.1855,
+      "step": 122900
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.14356566683849e-05,
+      "loss": 1.202,
+      "step": 123000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1428690842725588e-05,
+      "loss": 1.2127,
+      "step": 123100
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1421725017066274e-05,
+      "loss": 1.1939,
+      "step": 123200
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1414759191406957e-05,
+      "loss": 1.1893,
+      "step": 123300
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1407793365747644e-05,
+      "loss": 1.1822,
+      "step": 123400
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1400827540088327e-05,
+      "loss": 1.203,
+      "step": 123500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1393861714429013e-05,
+      "loss": 1.2094,
+      "step": 123600
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1386895888769696e-05,
+      "loss": 1.1569,
+      "step": 123700
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1379930063110382e-05,
+      "loss": 1.1824,
+      "step": 123800
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1372964237451065e-05,
+      "loss": 1.2077,
+      "step": 123900
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.136599841179175e-05,
+      "loss": 1.241,
+      "step": 124000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1359032586132435e-05,
+      "loss": 1.2163,
+      "step": 124100
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.135206676047312e-05,
+      "loss": 1.1717,
+      "step": 124200
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1345100934813804e-05,
+      "loss": 1.1579,
+      "step": 124300
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.133813510915449e-05,
+      "loss": 1.1677,
+      "step": 124400
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1331169283495173e-05,
+      "loss": 1.17,
+      "step": 124500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.132420345783586e-05,
+      "loss": 1.2317,
+      "step": 124600
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1317237632176543e-05,
+      "loss": 1.2105,
+      "step": 124700
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.131027180651723e-05,
+      "loss": 1.2142,
+      "step": 124800
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.130330598085791e-05,
+      "loss": 1.2362,
+      "step": 124900
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1296340155198595e-05,
+      "loss": 1.2074,
+      "step": 125000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1289374329539278e-05,
+      "loss": 1.2149,
+      "step": 125100
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1282408503879965e-05,
+      "loss": 1.1955,
+      "step": 125200
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.1275442678220647e-05,
+      "loss": 1.1841,
+      "step": 125300
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.1268476852561334e-05,
+      "loss": 1.1668,
+      "step": 125400
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.1261511026902017e-05,
+      "loss": 1.2053,
+      "step": 125500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.1254545201242703e-05,
+      "loss": 1.2002,
+      "step": 125600
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.124764903383998e-05,
+      "loss": 1.1777,
+      "step": 125700
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.1240683208180666e-05,
+      "loss": 1.207,
+      "step": 125800
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.123371738252135e-05,
+      "loss": 1.1941,
+      "step": 125900
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.1226751556862035e-05,
+      "loss": 1.2201,
+      "step": 126000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.121978573120272e-05,
+      "loss": 1.1776,
+      "step": 126100
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.1212819905543405e-05,
+      "loss": 1.2108,
+      "step": 126200
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.1205854079884088e-05,
+      "loss": 1.203,
+      "step": 126300
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.1198888254224774e-05,
+      "loss": 1.1849,
+      "step": 126400
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.119192242856546e-05,
+      "loss": 1.2132,
+      "step": 126500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.1184956602906144e-05,
+      "loss": 1.1879,
+      "step": 126600
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.117799077724683e-05,
+      "loss": 1.2181,
+      "step": 126700
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.1171024951587513e-05,
+      "loss": 1.2248,
+      "step": 126800
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.11640591259282e-05,
+      "loss": 1.184,
+      "step": 126900
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.1157093300268882e-05,
+      "loss": 1.1824,
+      "step": 127000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.115012747460957e-05,
+      "loss": 1.207,
+      "step": 127100
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.114316164895025e-05,
+      "loss": 1.1663,
+      "step": 127200
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.1136195823290935e-05,
+      "loss": 1.1765,
+      "step": 127300
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.1129229997631618e-05,
+      "loss": 1.2084,
+      "step": 127400
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1122264171972304e-05,
+      "loss": 1.1856,
+      "step": 127500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1115298346312987e-05,
+      "loss": 1.2146,
+      "step": 127600
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1108332520653673e-05,
+      "loss": 1.2269,
+      "step": 127700
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.110143635325095e-05,
+      "loss": 1.1899,
+      "step": 127800
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1094470527591636e-05,
+      "loss": 1.212,
+      "step": 127900
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.108750470193232e-05,
+      "loss": 1.178,
+      "step": 128000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1080538876273006e-05,
+      "loss": 1.2044,
+      "step": 128100
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.107357305061369e-05,
+      "loss": 1.1994,
+      "step": 128200
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1066607224954375e-05,
+      "loss": 1.16,
+      "step": 128300
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1059641399295058e-05,
+      "loss": 1.2212,
+      "step": 128400
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1052675573635744e-05,
+      "loss": 1.1828,
+      "step": 128500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1045709747976427e-05,
+      "loss": 1.185,
+      "step": 128600
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1038743922317114e-05,
+      "loss": 1.2089,
+      "step": 128700
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1031778096657797e-05,
+      "loss": 1.2048,
+      "step": 128800
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.1024812270998483e-05,
+      "loss": 1.2108,
+      "step": 128900
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.1017846445339166e-05,
+      "loss": 1.1907,
+      "step": 129000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.1010880619679852e-05,
+      "loss": 1.1854,
+      "step": 129100
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.1003914794020535e-05,
+      "loss": 1.1757,
+      "step": 129200
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0996948968361222e-05,
+      "loss": 1.1574,
+      "step": 129300
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0989983142701905e-05,
+      "loss": 1.2082,
+      "step": 129400
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.098301731704259e-05,
+      "loss": 1.1877,
+      "step": 129500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0976051491383274e-05,
+      "loss": 1.1789,
+      "step": 129600
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0969085665723957e-05,
+      "loss": 1.1804,
+      "step": 129700
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0962119840064643e-05,
+      "loss": 1.212,
+      "step": 129800
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0955154014405326e-05,
+      "loss": 1.2151,
+      "step": 129900
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0948188188746013e-05,
+      "loss": 1.1798,
+      "step": 130000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0941222363086696e-05,
+      "loss": 1.1934,
+      "step": 130100
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0934256537427382e-05,
+      "loss": 1.1635,
+      "step": 130200
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0927290711768065e-05,
+      "loss": 1.2152,
+      "step": 130300
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.092032488610875e-05,
+      "loss": 1.1814,
+      "step": 130400
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0913359060449435e-05,
+      "loss": 1.2153,
+      "step": 130500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.090639323479012e-05,
+      "loss": 1.1722,
+      "step": 130600
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0899427409130804e-05,
+      "loss": 1.1957,
+      "step": 130700
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.089246158347149e-05,
+      "loss": 1.1688,
+      "step": 130800
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0885495757812173e-05,
+      "loss": 1.2204,
+      "step": 130900
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.087852993215286e-05,
+      "loss": 1.1939,
+      "step": 131000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.0871564106493543e-05,
+      "loss": 1.1904,
+      "step": 131100
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.0864667939090822e-05,
+      "loss": 1.187,
+      "step": 131200
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.0857702113431505e-05,
+      "loss": 1.1394,
+      "step": 131300
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.0850736287772192e-05,
+      "loss": 1.1678,
+      "step": 131400
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.0843770462112875e-05,
+      "loss": 1.2065,
+      "step": 131500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.083680463645356e-05,
+      "loss": 1.2385,
+      "step": 131600
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.0829838810794244e-05,
+      "loss": 1.1825,
+      "step": 131700
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.082287298513493e-05,
+      "loss": 1.2232,
+      "step": 131800
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.0815907159475614e-05,
+      "loss": 1.2067,
+      "step": 131900
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.08089413338163e-05,
+      "loss": 1.2003,
+      "step": 132000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.0801975508156983e-05,
+      "loss": 1.1924,
+      "step": 132100
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.0795009682497666e-05,
+      "loss": 1.1814,
+      "step": 132200
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.078804385683835e-05,
+      "loss": 1.2183,
+      "step": 132300
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.0781078031179035e-05,
+      "loss": 1.1836,
+      "step": 132400
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.0774112205519718e-05,
+      "loss": 1.2281,
+      "step": 132500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.0767146379860405e-05,
+      "loss": 1.2025,
+      "step": 132600
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.0760180554201088e-05,
+      "loss": 1.2013,
+      "step": 132700
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.0753214728541774e-05,
+      "loss": 1.2316,
+      "step": 132800
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.074624890288246e-05,
+      "loss": 1.2001,
+      "step": 132900
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.0739283077223143e-05,
+      "loss": 1.2178,
+      "step": 133000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.073231725156383e-05,
+      "loss": 1.1835,
+      "step": 133100
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.0725351425904513e-05,
+      "loss": 1.2126,
+      "step": 133200
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.07183856002452e-05,
+      "loss": 1.1815,
+      "step": 133300
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.0711419774585882e-05,
+      "loss": 1.1677,
+      "step": 133400
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.070445394892657e-05,
+      "loss": 1.1902,
+      "step": 133500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.069748812326725e-05,
+      "loss": 1.1912,
+      "step": 133600
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.0690522297607938e-05,
+      "loss": 1.2141,
+      "step": 133700
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.068355647194862e-05,
+      "loss": 1.229,
+      "step": 133800
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0676590646289307e-05,
+      "loss": 1.1643,
+      "step": 133900
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.066962482062999e-05,
+      "loss": 1.2217,
+      "step": 134000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0662658994970677e-05,
+      "loss": 1.164,
+      "step": 134100
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.065569316931136e-05,
+      "loss": 1.2169,
+      "step": 134200
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0648727343652046e-05,
+      "loss": 1.2057,
+      "step": 134300
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0641761517992726e-05,
+      "loss": 1.1871,
+      "step": 134400
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.063486535059001e-05,
+      "loss": 1.2043,
+      "step": 134500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.062789952493069e-05,
+      "loss": 1.2072,
+      "step": 134600
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0620933699271375e-05,
+      "loss": 1.2292,
+      "step": 134700
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0613967873612058e-05,
+      "loss": 1.2426,
+      "step": 134800
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0607002047952744e-05,
+      "loss": 1.1745,
+      "step": 134900
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0600036222293427e-05,
+      "loss": 1.2162,
+      "step": 135000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0593070396634113e-05,
+      "loss": 1.1857,
+      "step": 135100
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0586104570974796e-05,
+      "loss": 1.2112,
+      "step": 135200
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0579138745315483e-05,
+      "loss": 1.1718,
+      "step": 135300
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.0572172919656166e-05,
+      "loss": 1.189,
+      "step": 135400
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.0565207093996852e-05,
+      "loss": 1.1931,
+      "step": 135500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.0558241268337535e-05,
+      "loss": 1.1882,
+      "step": 135600
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.055127544267822e-05,
+      "loss": 1.1711,
+      "step": 135700
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.0544309617018905e-05,
+      "loss": 1.1907,
+      "step": 135800
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.053734379135959e-05,
+      "loss": 1.2113,
+      "step": 135900
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.0530377965700274e-05,
+      "loss": 1.2093,
+      "step": 136000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.052341214004096e-05,
+      "loss": 1.1927,
+      "step": 136100
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0516446314381647e-05,
+      "loss": 1.2145,
+      "step": 136200
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.050948048872233e-05,
+      "loss": 1.1858,
+      "step": 136300
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0502514663063016e-05,
+      "loss": 1.1903,
+      "step": 136400
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.04955488374037e-05,
+      "loss": 1.1671,
+      "step": 136500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0488583011744385e-05,
+      "loss": 1.1913,
+      "step": 136600
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.048161718608507e-05,
+      "loss": 1.2293,
+      "step": 136700
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.047465136042575e-05,
+      "loss": 1.1974,
+      "step": 136800
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.0467685534766434e-05,
+      "loss": 1.2024,
+      "step": 136900
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.046071970910712e-05,
+      "loss": 1.2,
+      "step": 137000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.0453753883447804e-05,
+      "loss": 1.1975,
+      "step": 137100
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.044678805778849e-05,
+      "loss": 1.2117,
+      "step": 137200
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.0439822232129173e-05,
+      "loss": 1.1786,
+      "step": 137300
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.043285640646986e-05,
+      "loss": 1.193,
+      "step": 137400
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.0425890580810542e-05,
+      "loss": 1.2061,
+      "step": 137500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.041892475515123e-05,
+      "loss": 1.1696,
+      "step": 137600
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.0411958929491912e-05,
+      "loss": 1.1869,
+      "step": 137700
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.0404993103832598e-05,
+      "loss": 1.2024,
+      "step": 137800
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.039802727817328e-05,
+      "loss": 1.1441,
+      "step": 137900
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.0391061452513968e-05,
+      "loss": 1.1947,
+      "step": 138000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.038409562685465e-05,
+      "loss": 1.1952,
+      "step": 138100
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.0377129801195337e-05,
+      "loss": 1.2063,
+      "step": 138200
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.037016397553602e-05,
+      "loss": 1.1833,
+      "step": 138300
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.0363198149876706e-05,
+      "loss": 1.199,
+      "step": 138400
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.035623232421739e-05,
+      "loss": 1.176,
+      "step": 138500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.0349266498558076e-05,
+      "loss": 1.1947,
+      "step": 138600
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.0342300672898762e-05,
+      "loss": 1.1985,
+      "step": 138700
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.0335334847239445e-05,
+      "loss": 1.19,
+      "step": 138800
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.032836902158013e-05,
+      "loss": 1.1761,
+      "step": 138900
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.0321472854177408e-05,
+      "loss": 1.1637,
+      "step": 139000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.031450702851809e-05,
+      "loss": 1.1896,
+      "step": 139100
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.0307541202858777e-05,
+      "loss": 1.1838,
+      "step": 139200
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.0300575377199457e-05,
+      "loss": 1.2248,
+      "step": 139300
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.0293609551540143e-05,
+      "loss": 1.2136,
+      "step": 139400
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.028664372588083e-05,
+      "loss": 1.1767,
+      "step": 139500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.0279677900221513e-05,
+      "loss": 1.2188,
+      "step": 139600
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.02727120745622e-05,
+      "loss": 1.1943,
+      "step": 139700
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.0265746248902882e-05,
+      "loss": 1.1964,
+      "step": 139800
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.0258780423243568e-05,
+      "loss": 1.1937,
+      "step": 139900
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.025181459758425e-05,
+      "loss": 1.1766,
+      "step": 140000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.0244848771924938e-05,
+      "loss": 1.2146,
+      "step": 140100
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.023788294626562e-05,
+      "loss": 1.2048,
+      "step": 140200
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.0230917120606307e-05,
+      "loss": 1.1679,
+      "step": 140300
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.022395129494699e-05,
+      "loss": 1.2035,
+      "step": 140400
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.0216985469287676e-05,
+      "loss": 1.1797,
+      "step": 140500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.021001964362836e-05,
+      "loss": 1.1798,
+      "step": 140600
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.0203053817969046e-05,
+      "loss": 1.1807,
+      "step": 140700
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.0196157650566322e-05,
+      "loss": 1.2107,
+      "step": 140800
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.018919182490701e-05,
+      "loss": 1.1618,
+      "step": 140900
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.018222599924769e-05,
+      "loss": 1.18,
+      "step": 141000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.0175260173588378e-05,
+      "loss": 1.257,
+      "step": 141100
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.016829434792906e-05,
+      "loss": 1.1782,
+      "step": 141200
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.0161328522269747e-05,
+      "loss": 1.2087,
+      "step": 141300
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.015436269661043e-05,
+      "loss": 1.1758,
+      "step": 141400
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.0147396870951117e-05,
+      "loss": 1.1777,
+      "step": 141500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.01404310452918e-05,
+      "loss": 1.1925,
+      "step": 141600
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.0133465219632483e-05,
+      "loss": 1.1688,
+      "step": 141700
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.0126499393973166e-05,
+      "loss": 1.1683,
+      "step": 141800
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.0119533568313852e-05,
+      "loss": 1.156,
+      "step": 141900
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.0112567742654535e-05,
+      "loss": 1.1716,
+      "step": 142000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.010560191699522e-05,
+      "loss": 1.1522,
+      "step": 142100
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.0098636091335904e-05,
+      "loss": 1.184,
+      "step": 142200
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.009167026567659e-05,
+      "loss": 1.1877,
+      "step": 142300
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.0084704440017274e-05,
+      "loss": 1.1699,
+      "step": 142400
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.007773861435796e-05,
+      "loss": 1.1616,
+      "step": 142500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.0070772788698643e-05,
+      "loss": 1.1984,
+      "step": 142600
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.006380696303933e-05,
+      "loss": 1.1819,
+      "step": 142700
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.0056841137380016e-05,
+      "loss": 1.2229,
+      "step": 142800
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.00498753117207e-05,
+      "loss": 1.1845,
+      "step": 142900
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.0042909486061385e-05,
+      "loss": 1.1822,
+      "step": 143000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.003601331865866e-05,
+      "loss": 1.2037,
+      "step": 143100
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.0029047492999345e-05,
+      "loss": 1.1949,
+      "step": 143200
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.0022151325596624e-05,
+      "loss": 1.1887,
+      "step": 143300
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.00152551581939e-05,
+      "loss": 1.1854,
+      "step": 143400
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.0008289332534587e-05,
+      "loss": 1.191,
+      "step": 143500
+    },
+    {
+      "epoch": 2.0,
+      "eval_gen_len": 20.0,
+      "eval_loss": 1.1614574193954468,
+      "eval_rouge1": 11.8484,
+      "eval_rouge2": 3.363,
+      "eval_rougeL": 11.4175,
+      "eval_rougeLsum": 11.5037,
+      "eval_runtime": 1532.2202,
+      "eval_samples_per_second": 8.725,
+      "eval_steps_per_second": 2.181,
+      "step": 143558
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.000132350687527e-05,
+      "loss": 1.1515,
+      "step": 143600
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9994357681215957e-05,
+      "loss": 1.0722,
+      "step": 143700
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.998739185555664e-05,
+      "loss": 1.0699,
+      "step": 143800
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9980426029897326e-05,
+      "loss": 1.0917,
+      "step": 143900
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.997346020423801e-05,
+      "loss": 1.0699,
+      "step": 144000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9966494378578695e-05,
+      "loss": 1.0534,
+      "step": 144100
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.995952855291938e-05,
+      "loss": 1.0903,
+      "step": 144200
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9952562727260065e-05,
+      "loss": 1.0532,
+      "step": 144300
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9945596901600748e-05,
+      "loss": 1.0783,
+      "step": 144400
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.993863107594143e-05,
+      "loss": 1.0558,
+      "step": 144500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9931665250282114e-05,
+      "loss": 1.0627,
+      "step": 144600
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.99246994246228e-05,
+      "loss": 1.1152,
+      "step": 144700
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9917733598963483e-05,
+      "loss": 1.0542,
+      "step": 144800
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.991076777330417e-05,
+      "loss": 1.0506,
+      "step": 144900
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9903801947644852e-05,
+      "loss": 1.0801,
+      "step": 145000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.989683612198554e-05,
+      "loss": 1.0663,
+      "step": 145100
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9889870296326225e-05,
+      "loss": 1.1189,
+      "step": 145200
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9882904470666908e-05,
+      "loss": 1.0802,
+      "step": 145300
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9875938645007595e-05,
+      "loss": 1.0668,
+      "step": 145400
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9868972819348278e-05,
+      "loss": 1.0909,
+      "step": 145500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9862006993688964e-05,
+      "loss": 1.0837,
+      "step": 145600
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9855041168029647e-05,
+      "loss": 1.1061,
+      "step": 145700
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9848075342370333e-05,
+      "loss": 1.0728,
+      "step": 145800
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9841109516711016e-05,
+      "loss": 1.0877,
+      "step": 145900
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9834143691051703e-05,
+      "loss": 1.0797,
+      "step": 146000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9827177865392386e-05,
+      "loss": 1.0839,
+      "step": 146100
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9820212039733072e-05,
+      "loss": 1.0552,
+      "step": 146200
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9813246214073755e-05,
+      "loss": 1.1046,
+      "step": 146300
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.980628038841444e-05,
+      "loss": 1.066,
+      "step": 146400
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9799314562755124e-05,
+      "loss": 1.0863,
+      "step": 146500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.979234873709581e-05,
+      "loss": 1.0575,
+      "step": 146600
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.978538291143649e-05,
+      "loss": 1.0753,
+      "step": 146700
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9778417085777177e-05,
+      "loss": 1.0724,
+      "step": 146800
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.977145126011786e-05,
+      "loss": 1.0711,
+      "step": 146900
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9764485434458546e-05,
+      "loss": 1.0756,
+      "step": 147000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.975751960879923e-05,
+      "loss": 1.0942,
+      "step": 147100
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9750553783139915e-05,
+      "loss": 1.0993,
+      "step": 147200
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.97435879574806e-05,
+      "loss": 1.0571,
+      "step": 147300
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9736622131821285e-05,
+      "loss": 1.0705,
+      "step": 147400
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9729656306161968e-05,
+      "loss": 1.0765,
+      "step": 147500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9722690480502654e-05,
+      "loss": 1.0937,
+      "step": 147600
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.971572465484334e-05,
+      "loss": 1.0741,
+      "step": 147700
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9708758829184024e-05,
+      "loss": 1.08,
+      "step": 147800
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.970179300352471e-05,
+      "loss": 1.0438,
+      "step": 147900
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9694827177865393e-05,
+      "loss": 1.086,
+      "step": 148000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.968786135220608e-05,
+      "loss": 1.0679,
+      "step": 148100
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9680895526546762e-05,
+      "loss": 1.0575,
+      "step": 148200
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.967392970088745e-05,
+      "loss": 1.0817,
+      "step": 148300
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.966696387522813e-05,
+      "loss": 1.0588,
+      "step": 148400
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9659998049568818e-05,
+      "loss": 1.0579,
+      "step": 148500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.96530322239095e-05,
+      "loss": 1.1229,
+      "step": 148600
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9646066398250187e-05,
+      "loss": 1.0805,
+      "step": 148700
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.963910057259087e-05,
+      "loss": 1.0663,
+      "step": 148800
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9632134746931553e-05,
+      "loss": 1.0779,
+      "step": 148900
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9625168921272236e-05,
+      "loss": 1.0883,
+      "step": 149000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9618203095612923e-05,
+      "loss": 1.0812,
+      "step": 149100
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9611237269953606e-05,
+      "loss": 1.0774,
+      "step": 149200
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9604271444294292e-05,
+      "loss": 1.0757,
+      "step": 149300
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9597305618634975e-05,
+      "loss": 1.0626,
+      "step": 149400
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.959033979297566e-05,
+      "loss": 1.0566,
+      "step": 149500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9583373967316344e-05,
+      "loss": 1.0388,
+      "step": 149600
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.957640814165703e-05,
+      "loss": 1.0847,
+      "step": 149700
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9569442315997714e-05,
+      "loss": 1.0748,
+      "step": 149800
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.95624764903384e-05,
+      "loss": 1.0418,
+      "step": 149900
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9555580322935677e-05,
+      "loss": 1.0878,
+      "step": 150000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9548614497276363e-05,
+      "loss": 1.0756,
+      "step": 150100
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9541648671617046e-05,
+      "loss": 1.11,
+      "step": 150200
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9534682845957732e-05,
+      "loss": 1.0875,
+      "step": 150300
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9527717020298415e-05,
+      "loss": 1.1018,
+      "step": 150400
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9520751194639102e-05,
+      "loss": 1.0643,
+      "step": 150500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9513855027236378e-05,
+      "loss": 1.0752,
+      "step": 150600
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9506889201577065e-05,
+      "loss": 1.0931,
+      "step": 150700
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9499923375917748e-05,
+      "loss": 1.0546,
+      "step": 150800
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9492957550258434e-05,
+      "loss": 1.0705,
+      "step": 150900
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9485991724599117e-05,
+      "loss": 1.1041,
+      "step": 151000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9479025898939803e-05,
+      "loss": 1.052,
+      "step": 151100
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9472060073280486e-05,
+      "loss": 1.0799,
+      "step": 151200
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9465094247621173e-05,
+      "loss": 1.0159,
+      "step": 151300
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9458128421961856e-05,
+      "loss": 1.0645,
+      "step": 151400
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.945116259630254e-05,
+      "loss": 1.0853,
+      "step": 151500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9444196770643225e-05,
+      "loss": 1.0862,
+      "step": 151600
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9437230944983908e-05,
+      "loss": 1.0347,
+      "step": 151700
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9430265119324594e-05,
+      "loss": 1.0947,
+      "step": 151800
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9423299293665277e-05,
+      "loss": 1.0781,
+      "step": 151900
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9416333468005964e-05,
+      "loss": 1.0664,
+      "step": 152000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9409367642346647e-05,
+      "loss": 1.0936,
+      "step": 152100
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9402401816687333e-05,
+      "loss": 1.0828,
+      "step": 152200
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9395435991028016e-05,
+      "loss": 1.0902,
+      "step": 152300
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9388470165368702e-05,
+      "loss": 1.0435,
+      "step": 152400
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9381504339709385e-05,
+      "loss": 1.07,
+      "step": 152500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9374538514050072e-05,
+      "loss": 1.0898,
+      "step": 152600
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9367572688390755e-05,
+      "loss": 1.0498,
+      "step": 152700
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.936060686273144e-05,
+      "loss": 1.0636,
+      "step": 152800
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9353641037072124e-05,
+      "loss": 1.0898,
+      "step": 152900
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.934667521141281e-05,
+      "loss": 1.0735,
+      "step": 153000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9339709385753494e-05,
+      "loss": 1.0621,
+      "step": 153100
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.933274356009418e-05,
+      "loss": 1.0774,
+      "step": 153200
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9325777734434863e-05,
+      "loss": 1.0564,
+      "step": 153300
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.931881190877555e-05,
+      "loss": 1.1064,
+      "step": 153400
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9311846083116232e-05,
+      "loss": 1.0568,
+      "step": 153500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.930488025745692e-05,
+      "loss": 1.0978,
+      "step": 153600
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.92979144317976e-05,
+      "loss": 1.0948,
+      "step": 153700
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9290948606138285e-05,
+      "loss": 1.0576,
+      "step": 153800
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9283982780478968e-05,
+      "loss": 1.0557,
+      "step": 153900
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9277016954819654e-05,
+      "loss": 1.0683,
+      "step": 154000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.927005112916034e-05,
+      "loss": 1.1081,
+      "step": 154100
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9263085303501023e-05,
+      "loss": 1.0581,
+      "step": 154200
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.925611947784171e-05,
+      "loss": 1.0605,
+      "step": 154300
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9249153652182393e-05,
+      "loss": 1.0912,
+      "step": 154400
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.924218782652308e-05,
+      "loss": 1.0877,
+      "step": 154500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9235291659120356e-05,
+      "loss": 1.0878,
+      "step": 154600
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.922832583346104e-05,
+      "loss": 1.1009,
+      "step": 154700
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9221360007801725e-05,
+      "loss": 1.0887,
+      "step": 154800
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.921439418214241e-05,
+      "loss": 1.0782,
+      "step": 154900
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9207428356483094e-05,
+      "loss": 1.0584,
+      "step": 155000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.920046253082378e-05,
+      "loss": 1.0652,
+      "step": 155100
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9193496705164464e-05,
+      "loss": 1.0993,
+      "step": 155200
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.918653087950515e-05,
+      "loss": 1.1073,
+      "step": 155300
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9179565053845833e-05,
+      "loss": 1.0592,
+      "step": 155400
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.917259922818652e-05,
+      "loss": 1.0819,
+      "step": 155500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9165633402527202e-05,
+      "loss": 1.0764,
+      "step": 155600
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.915866757686789e-05,
+      "loss": 1.103,
+      "step": 155700
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9151701751208572e-05,
+      "loss": 1.0686,
+      "step": 155800
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9144735925549258e-05,
+      "loss": 1.106,
+      "step": 155900
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.913777009988994e-05,
+      "loss": 1.0904,
+      "step": 156000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9130804274230627e-05,
+      "loss": 1.0756,
+      "step": 156100
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9123838448571307e-05,
+      "loss": 1.0832,
+      "step": 156200
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9116872622911993e-05,
+      "loss": 1.0617,
+      "step": 156300
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9109906797252676e-05,
+      "loss": 1.0715,
+      "step": 156400
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9102940971593363e-05,
+      "loss": 1.0961,
+      "step": 156500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9095975145934046e-05,
+      "loss": 1.0632,
+      "step": 156600
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9089009320274732e-05,
+      "loss": 1.0478,
+      "step": 156700
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9082043494615415e-05,
+      "loss": 1.0861,
+      "step": 156800
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.90750776689561e-05,
+      "loss": 1.0971,
+      "step": 156900
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9068181501553378e-05,
+      "loss": 1.0603,
+      "step": 157000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9061215675894064e-05,
+      "loss": 1.0885,
+      "step": 157100
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9054249850234747e-05,
+      "loss": 1.0893,
+      "step": 157200
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9047284024575434e-05,
+      "loss": 1.085,
+      "step": 157300
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9040318198916117e-05,
+      "loss": 1.0892,
+      "step": 157400
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9033352373256803e-05,
+      "loss": 1.0486,
+      "step": 157500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9026386547597486e-05,
+      "loss": 1.1074,
+      "step": 157600
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9019420721938172e-05,
+      "loss": 1.0732,
+      "step": 157700
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9012454896278855e-05,
+      "loss": 1.0925,
+      "step": 157800
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9005489070619542e-05,
+      "loss": 1.0564,
+      "step": 157900
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.8998523244960225e-05,
+      "loss": 1.0608,
+      "step": 158000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.899155741930091e-05,
+      "loss": 1.0669,
+      "step": 158100
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.8984591593641598e-05,
+      "loss": 1.066,
+      "step": 158200
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.897762576798228e-05,
+      "loss": 1.08,
+      "step": 158300
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.8970659942322967e-05,
+      "loss": 1.0728,
+      "step": 158400
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.896369411666365e-05,
+      "loss": 1.0898,
+      "step": 158500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.8956728291004333e-05,
+      "loss": 1.0327,
+      "step": 158600
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.8949762465345016e-05,
+      "loss": 1.1012,
+      "step": 158700
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.8942796639685702e-05,
+      "loss": 1.1121,
+      "step": 158800
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.8935830814026385e-05,
+      "loss": 1.0478,
+      "step": 158900
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.892886498836707e-05,
+      "loss": 1.0447,
+      "step": 159000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.8921899162707755e-05,
+      "loss": 1.0683,
+      "step": 159100
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.891493333704844e-05,
+      "loss": 1.0817,
+      "step": 159200
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.8907967511389124e-05,
+      "loss": 1.0836,
+      "step": 159300
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.8901071343986404e-05,
+      "loss": 1.1046,
+      "step": 159400
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.8894105518327087e-05,
+      "loss": 1.1115,
+      "step": 159500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.8887139692667773e-05,
+      "loss": 1.1063,
+      "step": 159600
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.8880173867008456e-05,
+      "loss": 1.1138,
+      "step": 159700
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.8873208041349143e-05,
+      "loss": 1.1093,
+      "step": 159800
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.8866242215689826e-05,
+      "loss": 1.0654,
+      "step": 159900
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.8859276390030512e-05,
+      "loss": 1.0395,
+      "step": 160000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.8852310564371195e-05,
+      "loss": 1.0777,
+      "step": 160100
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.884534473871188e-05,
+      "loss": 1.0917,
+      "step": 160200
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.8838378913052564e-05,
+      "loss": 1.096,
+      "step": 160300
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.883141308739325e-05,
+      "loss": 1.0607,
+      "step": 160400
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.8824447261733934e-05,
+      "loss": 1.0552,
+      "step": 160500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.881748143607462e-05,
+      "loss": 1.0705,
+      "step": 160600
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.8810515610415303e-05,
+      "loss": 1.0704,
+      "step": 160700
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.880354978475599e-05,
+      "loss": 1.0819,
+      "step": 160800
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.8796583959096672e-05,
+      "loss": 1.0863,
+      "step": 160900
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.878961813343736e-05,
+      "loss": 1.0816,
+      "step": 161000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.878265230777804e-05,
+      "loss": 1.0417,
+      "step": 161100
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.8775686482118725e-05,
+      "loss": 1.08,
+      "step": 161200
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.8768720656459408e-05,
+      "loss": 1.0572,
+      "step": 161300
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.8761754830800094e-05,
+      "loss": 1.0864,
+      "step": 161400
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.875478900514078e-05,
+      "loss": 1.0694,
+      "step": 161500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.8747892837738057e-05,
+      "loss": 1.0714,
+      "step": 161600
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.8740996670335333e-05,
+      "loss": 1.0979,
+      "step": 161700
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.873403084467602e-05,
+      "loss": 1.0908,
+      "step": 161800
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.8727065019016703e-05,
+      "loss": 1.0693,
+      "step": 161900
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.872009919335739e-05,
+      "loss": 1.0842,
+      "step": 162000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.8713133367698072e-05,
+      "loss": 1.1071,
+      "step": 162100
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.870616754203876e-05,
+      "loss": 1.0485,
+      "step": 162200
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.869920171637944e-05,
+      "loss": 1.0638,
+      "step": 162300
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.8692235890720128e-05,
+      "loss": 1.0964,
+      "step": 162400
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.868527006506081e-05,
+      "loss": 1.0537,
+      "step": 162500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.8678304239401497e-05,
+      "loss": 1.0453,
+      "step": 162600
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.867133841374218e-05,
+      "loss": 1.0647,
+      "step": 162700
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.8664372588082867e-05,
+      "loss": 1.1006,
+      "step": 162800
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.865740676242355e-05,
+      "loss": 1.0989,
+      "step": 162900
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.8650440936764236e-05,
+      "loss": 1.0766,
+      "step": 163000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.8643475111104922e-05,
+      "loss": 1.0662,
+      "step": 163100
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.8636509285445605e-05,
+      "loss": 1.1073,
+      "step": 163200
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.862954345978629e-05,
+      "loss": 1.0682,
+      "step": 163300
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.8622577634126975e-05,
+      "loss": 1.0908,
+      "step": 163400
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.861561180846766e-05,
+      "loss": 1.0464,
+      "step": 163500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.8608645982808344e-05,
+      "loss": 1.0401,
+      "step": 163600
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.8601680157149027e-05,
+      "loss": 1.1031,
+      "step": 163700
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.859471433148971e-05,
+      "loss": 1.0626,
+      "step": 163800
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.8587748505830396e-05,
+      "loss": 1.1199,
+      "step": 163900
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.858078268017108e-05,
+      "loss": 1.0946,
+      "step": 164000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.8573816854511766e-05,
+      "loss": 1.0756,
+      "step": 164100
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.856685102885245e-05,
+      "loss": 1.0578,
+      "step": 164200
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.8559885203193135e-05,
+      "loss": 1.0652,
+      "step": 164300
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.8552919377533818e-05,
+      "loss": 1.0721,
+      "step": 164400
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.8546092868387688e-05,
+      "loss": 1.0962,
+      "step": 164500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.8539127042728374e-05,
+      "loss": 1.0605,
+      "step": 164600
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.853216121706906e-05,
+      "loss": 1.0535,
+      "step": 164700
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.8525195391409744e-05,
+      "loss": 1.0563,
+      "step": 164800
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.851822956575043e-05,
+      "loss": 1.1039,
+      "step": 164900
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.8511263740091113e-05,
+      "loss": 1.0619,
+      "step": 165000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.85042979144318e-05,
+      "loss": 1.0629,
+      "step": 165100
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.8497332088772482e-05,
+      "loss": 1.0738,
+      "step": 165200
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.849036626311317e-05,
+      "loss": 1.0515,
+      "step": 165300
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.8483400437453852e-05,
+      "loss": 1.0813,
+      "step": 165400
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.8476434611794538e-05,
+      "loss": 1.0457,
+      "step": 165500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.846946878613522e-05,
+      "loss": 1.0754,
+      "step": 165600
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.8462502960475908e-05,
+      "loss": 1.1107,
+      "step": 165700
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.845553713481659e-05,
+      "loss": 1.0671,
+      "step": 165800
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.8448571309157277e-05,
+      "loss": 1.1161,
+      "step": 165900
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.844160548349796e-05,
+      "loss": 1.0534,
+      "step": 166000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.8434639657838646e-05,
+      "loss": 1.0737,
+      "step": 166100
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.842767383217933e-05,
+      "loss": 1.0896,
+      "step": 166200
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.8420708006520012e-05,
+      "loss": 1.0861,
+      "step": 166300
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.8413742180860695e-05,
+      "loss": 1.0763,
+      "step": 166400
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.840677635520138e-05,
+      "loss": 1.0291,
+      "step": 166500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.8399810529542065e-05,
+      "loss": 1.0815,
+      "step": 166600
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.839284470388275e-05,
+      "loss": 1.0771,
+      "step": 166700
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.8385878878223434e-05,
+      "loss": 1.0989,
+      "step": 166800
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.837891305256412e-05,
+      "loss": 1.0677,
+      "step": 166900
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.8371947226904803e-05,
+      "loss": 1.0529,
+      "step": 167000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.836498140124549e-05,
+      "loss": 1.0796,
+      "step": 167100
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.8358015575586176e-05,
+      "loss": 1.0991,
+      "step": 167200
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.835104974992686e-05,
+      "loss": 1.0839,
+      "step": 167300
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.8344083924267545e-05,
+      "loss": 1.1401,
+      "step": 167400
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.833711809860823e-05,
+      "loss": 1.0324,
+      "step": 167500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.8330152272948915e-05,
+      "loss": 1.0832,
+      "step": 167600
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.8323186447289598e-05,
+      "loss": 1.0761,
+      "step": 167700
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.8316220621630284e-05,
+      "loss": 1.0775,
+      "step": 167800
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.8309254795970967e-05,
+      "loss": 1.1074,
+      "step": 167900
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.8302288970311654e-05,
+      "loss": 1.0927,
+      "step": 168000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.8295323144652337e-05,
+      "loss": 1.0714,
+      "step": 168100
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.8288357318993023e-05,
+      "loss": 1.0387,
+      "step": 168200
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.8281391493333706e-05,
+      "loss": 1.096,
+      "step": 168300
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.8274425667674392e-05,
+      "loss": 1.0691,
+      "step": 168400
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.8267459842015072e-05,
+      "loss": 1.0612,
+      "step": 168500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.8260494016355758e-05,
+      "loss": 1.0746,
+      "step": 168600
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.825352819069644e-05,
+      "loss": 1.0709,
+      "step": 168700
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.8246562365037128e-05,
+      "loss": 1.0541,
+      "step": 168800
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.823959653937781e-05,
+      "loss": 1.0693,
+      "step": 168900
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.8232630713718497e-05,
+      "loss": 1.0334,
+      "step": 169000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.8225734546315773e-05,
+      "loss": 1.0684,
+      "step": 169100
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.821876872065646e-05,
+      "loss": 1.1119,
+      "step": 169200
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.8211802894997143e-05,
+      "loss": 1.0625,
+      "step": 169300
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.820483706933783e-05,
+      "loss": 1.078,
+      "step": 169400
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.8197871243678512e-05,
+      "loss": 1.0941,
+      "step": 169500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.81909054180192e-05,
+      "loss": 1.1107,
+      "step": 169600
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.818393959235988e-05,
+      "loss": 1.0978,
+      "step": 169700
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8176973766700568e-05,
+      "loss": 1.0845,
+      "step": 169800
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.817000794104125e-05,
+      "loss": 1.058,
+      "step": 169900
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8163042115381937e-05,
+      "loss": 1.0849,
+      "step": 170000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.815607628972262e-05,
+      "loss": 1.1009,
+      "step": 170100
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8149110464063307e-05,
+      "loss": 1.1185,
+      "step": 170200
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.814214463840399e-05,
+      "loss": 1.0886,
+      "step": 170300
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8135178812744676e-05,
+      "loss": 1.0958,
+      "step": 170400
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.8128212987085362e-05,
+      "loss": 1.0732,
+      "step": 170500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.8121247161426045e-05,
+      "loss": 1.0757,
+      "step": 170600
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.8114281335766732e-05,
+      "loss": 1.0901,
+      "step": 170700
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.8107315510107415e-05,
+      "loss": 1.1048,
+      "step": 170800
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.8100349684448098e-05,
+      "loss": 1.0625,
+      "step": 170900
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.809338385878878e-05,
+      "loss": 1.0933,
+      "step": 171000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.8086418033129467e-05,
+      "loss": 1.0984,
+      "step": 171100
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.807945220747015e-05,
+      "loss": 1.0805,
+      "step": 171200
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.8072486381810836e-05,
+      "loss": 1.0746,
+      "step": 171300
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.806552055615152e-05,
+      "loss": 1.0634,
+      "step": 171400
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.8058554730492206e-05,
+      "loss": 1.074,
+      "step": 171500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.805158890483289e-05,
+      "loss": 1.0932,
+      "step": 171600
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.8044623079173575e-05,
+      "loss": 1.0733,
+      "step": 171700
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.8037657253514258e-05,
+      "loss": 1.0979,
+      "step": 171800
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.8030691427854945e-05,
+      "loss": 1.1366,
+      "step": 171900
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.8023725602195628e-05,
+      "loss": 1.0854,
+      "step": 172000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.8016759776536314e-05,
+      "loss": 1.0839,
+      "step": 172100
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.8009793950876997e-05,
+      "loss": 1.0771,
+      "step": 172200
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.8002828125217683e-05,
+      "loss": 1.0912,
+      "step": 172300
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.7995862299558366e-05,
+      "loss": 1.0832,
+      "step": 172400
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.7988896473899053e-05,
+      "loss": 1.0872,
+      "step": 172500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.7981930648239736e-05,
+      "loss": 1.1261,
+      "step": 172600
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.7974964822580422e-05,
+      "loss": 1.0622,
+      "step": 172700
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.7967998996921105e-05,
+      "loss": 1.0816,
+      "step": 172800
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.796103317126179e-05,
+      "loss": 1.0919,
+      "step": 172900
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.7954067345602478e-05,
+      "loss": 1.0751,
+      "step": 173000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.794710151994316e-05,
+      "loss": 1.0811,
+      "step": 173100
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.7940135694283844e-05,
+      "loss": 1.1039,
+      "step": 173200
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.7933169868624527e-05,
+      "loss": 1.079,
+      "step": 173300
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.7926204042965213e-05,
+      "loss": 1.074,
+      "step": 173400
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.7919238217305896e-05,
+      "loss": 1.1371,
+      "step": 173500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.7912272391646583e-05,
+      "loss": 1.0723,
+      "step": 173600
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.7905306565987266e-05,
+      "loss": 1.1076,
+      "step": 173700
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.7898340740327952e-05,
+      "loss": 1.0711,
+      "step": 173800
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.7891374914668635e-05,
+      "loss": 1.0799,
+      "step": 173900
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.788440908900932e-05,
+      "loss": 1.0692,
+      "step": 174000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7877443263350004e-05,
+      "loss": 1.0965,
+      "step": 174100
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7870547095947284e-05,
+      "loss": 1.061,
+      "step": 174200
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7863581270287967e-05,
+      "loss": 1.0913,
+      "step": 174300
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7856615444628653e-05,
+      "loss": 1.0876,
+      "step": 174400
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7849649618969336e-05,
+      "loss": 1.0911,
+      "step": 174500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7842683793310023e-05,
+      "loss": 1.0707,
+      "step": 174600
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7835717967650706e-05,
+      "loss": 1.0763,
+      "step": 174700
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7828752141991392e-05,
+      "loss": 1.1131,
+      "step": 174800
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7821786316332075e-05,
+      "loss": 1.0556,
+      "step": 174900
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.781482049067276e-05,
+      "loss": 1.0817,
+      "step": 175000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7807854665013445e-05,
+      "loss": 1.052,
+      "step": 175100
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.780088883935413e-05,
+      "loss": 1.0742,
+      "step": 175200
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7793923013694814e-05,
+      "loss": 1.1105,
+      "step": 175300
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.77869571880355e-05,
+      "loss": 1.0677,
+      "step": 175400
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7779991362376183e-05,
+      "loss": 1.0662,
+      "step": 175500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7773025536716866e-05,
+      "loss": 1.0562,
+      "step": 175600
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.776605971105755e-05,
+      "loss": 1.0825,
+      "step": 175700
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7759093885398236e-05,
+      "loss": 1.0602,
+      "step": 175800
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.775212805973892e-05,
+      "loss": 1.0643,
+      "step": 175900
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7745162234079605e-05,
+      "loss": 1.099,
+      "step": 176000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.773819640842029e-05,
+      "loss": 1.0983,
+      "step": 176100
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7731230582760974e-05,
+      "loss": 1.0967,
+      "step": 176200
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.772426475710166e-05,
+      "loss": 1.0932,
+      "step": 176300
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7717298931442344e-05,
+      "loss": 1.1169,
+      "step": 176400
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.771033310578303e-05,
+      "loss": 1.0628,
+      "step": 176500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7703367280123713e-05,
+      "loss": 1.0565,
+      "step": 176600
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.76964014544644e-05,
+      "loss": 1.0957,
+      "step": 176700
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7689435628805082e-05,
+      "loss": 1.102,
+      "step": 176800
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.768246980314577e-05,
+      "loss": 1.0591,
+      "step": 176900
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.7675503977486452e-05,
+      "loss": 1.0839,
+      "step": 177000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.7668538151827138e-05,
+      "loss": 1.0878,
+      "step": 177100
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.766157232616782e-05,
+      "loss": 1.126,
+      "step": 177200
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.7654606500508508e-05,
+      "loss": 1.0994,
+      "step": 177300
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.764764067484919e-05,
+      "loss": 1.0516,
+      "step": 177400
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.7640674849189877e-05,
+      "loss": 1.0712,
+      "step": 177500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.763370902353056e-05,
+      "loss": 1.0929,
+      "step": 177600
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.7626743197871246e-05,
+      "loss": 1.0808,
+      "step": 177700
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.7619777372211926e-05,
+      "loss": 1.0607,
+      "step": 177800
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.7612811546552612e-05,
+      "loss": 1.1114,
+      "step": 177900
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.7605845720893295e-05,
+      "loss": 1.0941,
+      "step": 178000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.759887989523398e-05,
+      "loss": 1.0774,
+      "step": 178100
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.7591914069574665e-05,
+      "loss": 1.0861,
+      "step": 178200
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.758494824391535e-05,
+      "loss": 1.0952,
+      "step": 178300
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.7577982418256034e-05,
+      "loss": 1.081,
+      "step": 178400
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.7571086250853314e-05,
+      "loss": 1.0808,
+      "step": 178500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.7564120425193997e-05,
+      "loss": 1.069,
+      "step": 178600
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.7557154599534683e-05,
+      "loss": 1.0716,
+      "step": 178700
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.7550188773875366e-05,
+      "loss": 1.0677,
+      "step": 178800
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.7543222948216053e-05,
+      "loss": 1.0636,
+      "step": 178900
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.753632678081333e-05,
+      "loss": 1.074,
+      "step": 179000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.7529360955154015e-05,
+      "loss": 1.0737,
+      "step": 179100
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.75223951294947e-05,
+      "loss": 1.0935,
+      "step": 179200
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.7515429303835385e-05,
+      "loss": 1.0565,
+      "step": 179300
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.7508463478176068e-05,
+      "loss": 1.0837,
+      "step": 179400
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.7501497652516754e-05,
+      "loss": 1.0872,
+      "step": 179500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.7494531826857437e-05,
+      "loss": 1.0564,
+      "step": 179600
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.7487566001198123e-05,
+      "loss": 1.0401,
+      "step": 179700
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.7480600175538806e-05,
+      "loss": 1.0782,
+      "step": 179800
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.7473634349879493e-05,
+      "loss": 1.0679,
+      "step": 179900
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.7466668524220176e-05,
+      "loss": 1.07,
+      "step": 180000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.7459702698560862e-05,
+      "loss": 1.0541,
+      "step": 180100
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.745273687290155e-05,
+      "loss": 1.0779,
+      "step": 180200
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.744577104724223e-05,
+      "loss": 1.0653,
+      "step": 180300
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.7438805221582915e-05,
+      "loss": 1.0546,
+      "step": 180400
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.7431839395923598e-05,
+      "loss": 1.1075,
+      "step": 180500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.7424873570264284e-05,
+      "loss": 1.0602,
+      "step": 180600
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.7417907744604967e-05,
+      "loss": 1.0899,
+      "step": 180700
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.7410941918945653e-05,
+      "loss": 1.0677,
+      "step": 180800
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.7403976093286336e-05,
+      "loss": 1.1195,
+      "step": 180900
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.7397010267627023e-05,
+      "loss": 1.0993,
+      "step": 181000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.7390044441967706e-05,
+      "loss": 1.0639,
+      "step": 181100
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.7383078616308392e-05,
+      "loss": 1.0561,
+      "step": 181200
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.7376112790649075e-05,
+      "loss": 1.0863,
+      "step": 181300
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.736914696498976e-05,
+      "loss": 1.1163,
+      "step": 181400
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.7362181139330444e-05,
+      "loss": 1.1151,
+      "step": 181500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.735521531367113e-05,
+      "loss": 1.093,
+      "step": 181600
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.7348249488011814e-05,
+      "loss": 1.0777,
+      "step": 181700
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.73412836623525e-05,
+      "loss": 1.0636,
+      "step": 181800
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.7334317836693183e-05,
+      "loss": 1.0995,
+      "step": 181900
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.732735201103387e-05,
+      "loss": 1.1191,
+      "step": 182000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.7320386185374552e-05,
+      "loss": 1.0695,
+      "step": 182100
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.731342035971524e-05,
+      "loss": 1.078,
+      "step": 182200
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.7306524192312515e-05,
+      "loss": 1.0821,
+      "step": 182300
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.72995583666532e-05,
+      "loss": 1.103,
+      "step": 182400
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.7292592540993885e-05,
+      "loss": 1.0857,
+      "step": 182500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.728562671533457e-05,
+      "loss": 1.0897,
+      "step": 182600
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.7278660889675254e-05,
+      "loss": 1.082,
+      "step": 182700
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.727169506401594e-05,
+      "loss": 1.0776,
+      "step": 182800
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.726472923835662e-05,
+      "loss": 1.0197,
+      "step": 182900
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.7257763412697306e-05,
+      "loss": 1.0843,
+      "step": 183000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.725079758703799e-05,
+      "loss": 1.0958,
+      "step": 183100
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.7243831761378676e-05,
+      "loss": 1.1242,
+      "step": 183200
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.723686593571936e-05,
+      "loss": 1.0766,
+      "step": 183300
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.7229900110060045e-05,
+      "loss": 1.0642,
+      "step": 183400
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.722293428440073e-05,
+      "loss": 1.0896,
+      "step": 183500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.7215968458741414e-05,
+      "loss": 1.0786,
+      "step": 183600
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.72090026330821e-05,
+      "loss": 1.1146,
+      "step": 183700
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.7202036807422784e-05,
+      "loss": 1.1031,
+      "step": 183800
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.719507098176347e-05,
+      "loss": 1.0745,
+      "step": 183900
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.7188105156104153e-05,
+      "loss": 1.0727,
+      "step": 184000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.718113933044484e-05,
+      "loss": 1.0804,
+      "step": 184100
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.7174173504785523e-05,
+      "loss": 1.0792,
+      "step": 184200
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.716720767912621e-05,
+      "loss": 1.0806,
+      "step": 184300
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.7160241853466892e-05,
+      "loss": 1.0611,
+      "step": 184400
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.7153276027807578e-05,
+      "loss": 1.0855,
+      "step": 184500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.7146379860404855e-05,
+      "loss": 1.0496,
+      "step": 184600
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.713941403474554e-05,
+      "loss": 1.0807,
+      "step": 184700
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.7132448209086224e-05,
+      "loss": 1.0464,
+      "step": 184800
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.712548238342691e-05,
+      "loss": 1.0566,
+      "step": 184900
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.7118516557767593e-05,
+      "loss": 1.07,
+      "step": 185000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.711155073210828e-05,
+      "loss": 1.0464,
+      "step": 185100
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.7104584906448963e-05,
+      "loss": 1.0723,
+      "step": 185200
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.7097619080789646e-05,
+      "loss": 1.0604,
+      "step": 185300
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.709065325513033e-05,
+      "loss": 1.0921,
+      "step": 185400
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.7083687429471015e-05,
+      "loss": 1.0819,
+      "step": 185500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.7076721603811698e-05,
+      "loss": 1.0824,
+      "step": 185600
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.7069755778152385e-05,
+      "loss": 1.0815,
+      "step": 185700
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.7062789952493068e-05,
+      "loss": 1.0748,
+      "step": 185800
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.7055824126833754e-05,
+      "loss": 1.0746,
+      "step": 185900
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.7048858301174437e-05,
+      "loss": 1.1041,
+      "step": 186000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.7041892475515123e-05,
+      "loss": 1.056,
+      "step": 186100
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.7034926649855806e-05,
+      "loss": 1.0595,
+      "step": 186200
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.7027960824196493e-05,
+      "loss": 1.1017,
+      "step": 186300
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.7020994998537176e-05,
+      "loss": 1.0718,
+      "step": 186400
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.7014029172877862e-05,
+      "loss": 1.0939,
+      "step": 186500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.7007063347218545e-05,
+      "loss": 1.0789,
+      "step": 186600
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.700009752155923e-05,
+      "loss": 1.0848,
+      "step": 186700
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.6993201354156508e-05,
+      "loss": 1.1197,
+      "step": 186800
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.6986235528497194e-05,
+      "loss": 1.0867,
+      "step": 186900
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.6979269702837877e-05,
+      "loss": 1.1139,
+      "step": 187000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.6972303877178564e-05,
+      "loss": 1.1051,
+      "step": 187100
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.6965338051519247e-05,
+      "loss": 1.1226,
+      "step": 187200
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.6958372225859933e-05,
+      "loss": 1.0646,
+      "step": 187300
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.6951406400200616e-05,
+      "loss": 1.079,
+      "step": 187400
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.6944440574541302e-05,
+      "loss": 1.1046,
+      "step": 187500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.693747474888199e-05,
+      "loss": 1.0883,
+      "step": 187600
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.6930508923222668e-05,
+      "loss": 1.0878,
+      "step": 187700
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6923543097563355e-05,
+      "loss": 1.045,
+      "step": 187800
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6916577271904038e-05,
+      "loss": 1.1132,
+      "step": 187900
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6909611446244724e-05,
+      "loss": 1.0981,
+      "step": 188000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6902645620585407e-05,
+      "loss": 1.0895,
+      "step": 188100
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6895679794926093e-05,
+      "loss": 1.0495,
+      "step": 188200
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6888713969266776e-05,
+      "loss": 1.0456,
+      "step": 188300
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6881748143607463e-05,
+      "loss": 1.1049,
+      "step": 188400
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.6874782317948146e-05,
+      "loss": 1.0673,
+      "step": 188500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.6867816492288832e-05,
+      "loss": 1.0426,
+      "step": 188600
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.6860850666629515e-05,
+      "loss": 1.1219,
+      "step": 188700
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.68538848409702e-05,
+      "loss": 1.1085,
+      "step": 188800
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.6846919015310884e-05,
+      "loss": 1.1014,
+      "step": 188900
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.683995318965157e-05,
+      "loss": 1.1089,
+      "step": 189000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.6832987363992254e-05,
+      "loss": 1.0614,
+      "step": 189100
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.682602153833294e-05,
+      "loss": 1.0551,
+      "step": 189200
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.6819055712673623e-05,
+      "loss": 1.0755,
+      "step": 189300
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.681208988701431e-05,
+      "loss": 1.062,
+      "step": 189400
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.6805124061354993e-05,
+      "loss": 1.0667,
+      "step": 189500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.679815823569568e-05,
+      "loss": 1.0606,
+      "step": 189600
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.6791192410036362e-05,
+      "loss": 1.0815,
+      "step": 189700
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.6784226584377048e-05,
+      "loss": 1.0908,
+      "step": 189800
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6777260758717728e-05,
+      "loss": 1.0862,
+      "step": 189900
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6770294933058414e-05,
+      "loss": 1.0912,
+      "step": 190000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.676339876565569e-05,
+      "loss": 1.1108,
+      "step": 190100
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6756432939996377e-05,
+      "loss": 1.0989,
+      "step": 190200
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.674946711433706e-05,
+      "loss": 1.0668,
+      "step": 190300
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6742501288677746e-05,
+      "loss": 1.1185,
+      "step": 190400
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.673553546301843e-05,
+      "loss": 1.0967,
+      "step": 190500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.6728569637359116e-05,
+      "loss": 1.0713,
+      "step": 190600
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.67216038116998e-05,
+      "loss": 1.072,
+      "step": 190700
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.6714637986040485e-05,
+      "loss": 1.0556,
+      "step": 190800
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.670767216038117e-05,
+      "loss": 1.1068,
+      "step": 190900
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.6700706334721855e-05,
+      "loss": 1.0761,
+      "step": 191000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.669374050906254e-05,
+      "loss": 1.08,
+      "step": 191100
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.6686774683403224e-05,
+      "loss": 1.0995,
+      "step": 191200
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.667980885774391e-05,
+      "loss": 1.1142,
+      "step": 191300
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.6672843032084593e-05,
+      "loss": 1.0918,
+      "step": 191400
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.666587720642528e-05,
+      "loss": 1.0756,
+      "step": 191500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.6658911380765963e-05,
+      "loss": 1.1105,
+      "step": 191600
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.665194555510665e-05,
+      "loss": 1.0649,
+      "step": 191700
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.6644979729447332e-05,
+      "loss": 1.0424,
+      "step": 191800
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.663801390378802e-05,
+      "loss": 1.0836,
+      "step": 191900
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.66310480781287e-05,
+      "loss": 1.0363,
+      "step": 192000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.6624082252469388e-05,
+      "loss": 1.085,
+      "step": 192100
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.661711642681007e-05,
+      "loss": 1.1017,
+      "step": 192200
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.6610150601150757e-05,
+      "loss": 1.0552,
+      "step": 192300
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.6603184775491437e-05,
+      "loss": 1.0586,
+      "step": 192400
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.6596218949832123e-05,
+      "loss": 1.0569,
+      "step": 192500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.6589253124172806e-05,
+      "loss": 1.0751,
+      "step": 192600
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.6582287298513492e-05,
+      "loss": 1.0993,
+      "step": 192700
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.6575321472854175e-05,
+      "loss": 1.0761,
+      "step": 192800
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.6568355647194862e-05,
+      "loss": 1.0457,
+      "step": 192900
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.6561389821535545e-05,
+      "loss": 1.0316,
+      "step": 193000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.655442399587623e-05,
+      "loss": 1.0627,
+      "step": 193100
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.6547458170216914e-05,
+      "loss": 1.0716,
+      "step": 193200
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.65404923445576e-05,
+      "loss": 1.1109,
+      "step": 193300
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.6533526518898287e-05,
+      "loss": 1.1173,
+      "step": 193400
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.652656069323897e-05,
+      "loss": 1.0714,
+      "step": 193500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.6519594867579656e-05,
+      "loss": 1.0919,
+      "step": 193600
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.651262904192034e-05,
+      "loss": 1.0783,
+      "step": 193700
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.6505663216261026e-05,
+      "loss": 1.1043,
+      "step": 193800
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.649869739060171e-05,
+      "loss": 1.0848,
+      "step": 193900
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.6491731564942395e-05,
+      "loss": 1.1154,
+      "step": 194000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.6484765739283078e-05,
+      "loss": 1.077,
+      "step": 194100
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.6477799913623764e-05,
+      "loss": 1.126,
+      "step": 194200
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.647090374622104e-05,
+      "loss": 1.1002,
+      "step": 194300
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.6463937920561727e-05,
+      "loss": 1.075,
+      "step": 194400
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.645697209490241e-05,
+      "loss": 1.0668,
+      "step": 194500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.6450075927499687e-05,
+      "loss": 1.0743,
+      "step": 194600
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.6443110101840373e-05,
+      "loss": 1.0968,
+      "step": 194700
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.6436144276181056e-05,
+      "loss": 1.0653,
+      "step": 194800
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.6429178450521742e-05,
+      "loss": 1.0899,
+      "step": 194900
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.6422212624862425e-05,
+      "loss": 1.0995,
+      "step": 195000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.641524679920311e-05,
+      "loss": 1.0823,
+      "step": 195100
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.6408280973543795e-05,
+      "loss": 1.0645,
+      "step": 195200
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.6401315147884478e-05,
+      "loss": 1.0751,
+      "step": 195300
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.6394349322225164e-05,
+      "loss": 1.0748,
+      "step": 195400
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.6387383496565847e-05,
+      "loss": 1.1133,
+      "step": 195500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.6380417670906533e-05,
+      "loss": 1.086,
+      "step": 195600
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.6373451845247216e-05,
+      "loss": 1.0626,
+      "step": 195700
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.6366486019587903e-05,
+      "loss": 1.0763,
+      "step": 195800
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.6359520193928586e-05,
+      "loss": 1.0454,
+      "step": 195900
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.6352554368269272e-05,
+      "loss": 1.1118,
+      "step": 196000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.6345588542609955e-05,
+      "loss": 1.0515,
+      "step": 196100
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.633862271695064e-05,
+      "loss": 1.0801,
+      "step": 196200
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.6331656891291325e-05,
+      "loss": 1.0742,
+      "step": 196300
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.632469106563201e-05,
+      "loss": 1.0821,
+      "step": 196400
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.6317725239972694e-05,
+      "loss": 1.1027,
+      "step": 196500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.631075941431338e-05,
+      "loss": 1.0542,
+      "step": 196600
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.6303793588654063e-05,
+      "loss": 1.092,
+      "step": 196700
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.629682776299475e-05,
+      "loss": 1.1127,
+      "step": 196800
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.6289861937335433e-05,
+      "loss": 1.0915,
+      "step": 196900
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.628289611167612e-05,
+      "loss": 1.0801,
+      "step": 197000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6275930286016802e-05,
+      "loss": 1.0764,
+      "step": 197100
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6268964460357485e-05,
+      "loss": 1.073,
+      "step": 197200
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.626199863469817e-05,
+      "loss": 1.1099,
+      "step": 197300
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6255032809038854e-05,
+      "loss": 1.087,
+      "step": 197400
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.624806698337954e-05,
+      "loss": 1.0516,
+      "step": 197500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6241101157720224e-05,
+      "loss": 1.0765,
+      "step": 197600
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.62342049903175e-05,
+      "loss": 1.0956,
+      "step": 197700
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6227239164658187e-05,
+      "loss": 1.0769,
+      "step": 197800
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.622027333899887e-05,
+      "loss": 1.0753,
+      "step": 197900
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6213307513339556e-05,
+      "loss": 1.0535,
+      "step": 198000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6206341687680242e-05,
+      "loss": 1.0612,
+      "step": 198100
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6199375862020925e-05,
+      "loss": 1.0586,
+      "step": 198200
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.619241003636161e-05,
+      "loss": 1.0834,
+      "step": 198300
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6185444210702295e-05,
+      "loss": 1.1046,
+      "step": 198400
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.617847838504298e-05,
+      "loss": 1.0751,
+      "step": 198500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.6171512559383664e-05,
+      "loss": 1.0657,
+      "step": 198600
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.616454673372435e-05,
+      "loss": 1.0797,
+      "step": 198700
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.6157580908065033e-05,
+      "loss": 1.0562,
+      "step": 198800
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.615061508240572e-05,
+      "loss": 1.0602,
+      "step": 198900
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.6143649256746403e-05,
+      "loss": 1.0542,
+      "step": 199000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.6136753089343683e-05,
+      "loss": 1.0838,
+      "step": 199100
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6129787263684366e-05,
+      "loss": 1.0975,
+      "step": 199200
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6122821438025052e-05,
+      "loss": 1.0557,
+      "step": 199300
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6115855612365735e-05,
+      "loss": 1.0981,
+      "step": 199400
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.610888978670642e-05,
+      "loss": 1.1035,
+      "step": 199500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6101923961047104e-05,
+      "loss": 1.0754,
+      "step": 199600
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.609495813538779e-05,
+      "loss": 1.0956,
+      "step": 199700
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.608799230972847e-05,
+      "loss": 1.0734,
+      "step": 199800
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6081026484069157e-05,
+      "loss": 1.0515,
+      "step": 199900
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.607406065840984e-05,
+      "loss": 1.0866,
+      "step": 200000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6067094832750526e-05,
+      "loss": 1.0826,
+      "step": 200100
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.606012900709121e-05,
+      "loss": 1.0871,
+      "step": 200200
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6053163181431895e-05,
+      "loss": 1.0809,
+      "step": 200300
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.604619735577258e-05,
+      "loss": 1.1262,
+      "step": 200400
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6039231530113265e-05,
+      "loss": 1.1024,
+      "step": 200500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6032265704453948e-05,
+      "loss": 1.0786,
+      "step": 200600
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6025299878794634e-05,
+      "loss": 1.0486,
+      "step": 200700
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6018334053135317e-05,
+      "loss": 1.0435,
+      "step": 200800
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6011368227476003e-05,
+      "loss": 1.1039,
+      "step": 200900
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6004402401816686e-05,
+      "loss": 1.0653,
+      "step": 201000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.5997436576157373e-05,
+      "loss": 1.0518,
+      "step": 201100
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.5990470750498056e-05,
+      "loss": 1.104,
+      "step": 201200
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.5983504924838742e-05,
+      "loss": 1.0759,
+      "step": 201300
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.597653909917943e-05,
+      "loss": 1.0754,
+      "step": 201400
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.5969642931776705e-05,
+      "loss": 1.0443,
+      "step": 201500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.5962677106117388e-05,
+      "loss": 1.0767,
+      "step": 201600
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.5955711280458074e-05,
+      "loss": 1.0629,
+      "step": 201700
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.5948745454798757e-05,
+      "loss": 1.0433,
+      "step": 201800
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.5941779629139444e-05,
+      "loss": 1.1352,
+      "step": 201900
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.5934813803480127e-05,
+      "loss": 1.0735,
+      "step": 202000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.5927847977820813e-05,
+      "loss": 1.0932,
+      "step": 202100
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.5920882152161496e-05,
+      "loss": 1.0893,
+      "step": 202200
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.591391632650218e-05,
+      "loss": 1.0984,
+      "step": 202300
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.5906950500842865e-05,
+      "loss": 1.0618,
+      "step": 202400
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.589998467518355e-05,
+      "loss": 1.0906,
+      "step": 202500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.5893018849524235e-05,
+      "loss": 1.078,
+      "step": 202600
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.5886053023864918e-05,
+      "loss": 1.1111,
+      "step": 202700
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5879087198205604e-05,
+      "loss": 1.0833,
+      "step": 202800
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5872121372546287e-05,
+      "loss": 1.0899,
+      "step": 202900
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5865155546886974e-05,
+      "loss": 1.0476,
+      "step": 203000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5858189721227657e-05,
+      "loss": 1.0827,
+      "step": 203100
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5851223895568343e-05,
+      "loss": 1.0933,
+      "step": 203200
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5844258069909026e-05,
+      "loss": 1.136,
+      "step": 203300
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5837292244249712e-05,
+      "loss": 1.1106,
+      "step": 203400
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5830326418590395e-05,
+      "loss": 1.0957,
+      "step": 203500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.582336059293108e-05,
+      "loss": 1.1121,
+      "step": 203600
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5816394767271765e-05,
+      "loss": 1.0751,
+      "step": 203700
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.580942894161245e-05,
+      "loss": 1.0797,
+      "step": 203800
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5802463115953134e-05,
+      "loss": 1.0954,
+      "step": 203900
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.579549729029382e-05,
+      "loss": 1.0925,
+      "step": 204000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5788531464634503e-05,
+      "loss": 1.0731,
+      "step": 204100
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.578156563897519e-05,
+      "loss": 1.0628,
+      "step": 204200
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5774599813315873e-05,
+      "loss": 1.095,
+      "step": 204300
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.576763398765656e-05,
+      "loss": 1.074,
+      "step": 204400
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.576066816199724e-05,
+      "loss": 1.0821,
+      "step": 204500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5753702336337925e-05,
+      "loss": 1.122,
+      "step": 204600
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.574673651067861e-05,
+      "loss": 1.0234,
+      "step": 204700
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5739770685019294e-05,
+      "loss": 1.0927,
+      "step": 204800
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.573280485935998e-05,
+      "loss": 1.0797,
+      "step": 204900
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5725839033700664e-05,
+      "loss": 1.0855,
+      "step": 205000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.571887320804135e-05,
+      "loss": 1.0915,
+      "step": 205100
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5711907382382033e-05,
+      "loss": 1.0885,
+      "step": 205200
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.570494155672272e-05,
+      "loss": 1.0944,
+      "step": 205300
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5697975731063403e-05,
+      "loss": 1.1138,
+      "step": 205400
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.569100990540409e-05,
+      "loss": 1.0337,
+      "step": 205500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5684044079744772e-05,
+      "loss": 1.0726,
+      "step": 205600
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.567707825408546e-05,
+      "loss": 1.0861,
+      "step": 205700
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.567011242842614e-05,
+      "loss": 1.0717,
+      "step": 205800
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5663146602766828e-05,
+      "loss": 1.1152,
+      "step": 205900
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.565618077710751e-05,
+      "loss": 1.0978,
+      "step": 206000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5649214951448197e-05,
+      "loss": 1.0851,
+      "step": 206100
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.564224912578888e-05,
+      "loss": 1.0665,
+      "step": 206200
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5635283300129566e-05,
+      "loss": 1.0686,
+      "step": 206300
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.562831747447025e-05,
+      "loss": 1.0869,
+      "step": 206400
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5621351648810936e-05,
+      "loss": 1.1102,
+      "step": 206500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.561438582315162e-05,
+      "loss": 1.0685,
+      "step": 206600
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5607419997492302e-05,
+      "loss": 1.0487,
+      "step": 206700
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5600454171832985e-05,
+      "loss": 1.0682,
+      "step": 206800
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.559348834617367e-05,
+      "loss": 1.0777,
+      "step": 206900
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5586522520514354e-05,
+      "loss": 1.0471,
+      "step": 207000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.557955669485504e-05,
+      "loss": 1.0578,
+      "step": 207100
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5572590869195727e-05,
+      "loss": 1.0603,
+      "step": 207200
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.556562504353641e-05,
+      "loss": 1.0783,
+      "step": 207300
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5558659217877096e-05,
+      "loss": 1.1009,
+      "step": 207400
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.555169339221778e-05,
+      "loss": 1.0468,
+      "step": 207500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5544797224815056e-05,
+      "loss": 1.0694,
+      "step": 207600
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5537831399155742e-05,
+      "loss": 1.0634,
+      "step": 207700
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5530865573496425e-05,
+      "loss": 1.0974,
+      "step": 207800
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.552389974783711e-05,
+      "loss": 1.1073,
+      "step": 207900
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5516933922177798e-05,
+      "loss": 1.1259,
+      "step": 208000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.550996809651848e-05,
+      "loss": 1.0786,
+      "step": 208100
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5503002270859167e-05,
+      "loss": 1.0566,
+      "step": 208200
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.549603644519985e-05,
+      "loss": 1.1088,
+      "step": 208300
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5489070619540537e-05,
+      "loss": 1.0866,
+      "step": 208400
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.548210479388122e-05,
+      "loss": 1.1046,
+      "step": 208500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5475138968221906e-05,
+      "loss": 1.0708,
+      "step": 208600
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.546817314256259e-05,
+      "loss": 1.0877,
+      "step": 208700
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.546127697515987e-05,
+      "loss": 1.0964,
+      "step": 208800
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.545431114950055e-05,
+      "loss": 1.0895,
+      "step": 208900
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5447345323841238e-05,
+      "loss": 1.0904,
+      "step": 209000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.544037949818192e-05,
+      "loss": 1.1009,
+      "step": 209100
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5433413672522607e-05,
+      "loss": 1.0585,
+      "step": 209200
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5426447846863287e-05,
+      "loss": 1.0974,
+      "step": 209300
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5419482021203973e-05,
+      "loss": 1.1168,
+      "step": 209400
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5412516195544656e-05,
+      "loss": 1.0974,
+      "step": 209500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5405550369885343e-05,
+      "loss": 1.0855,
+      "step": 209600
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5398584544226026e-05,
+      "loss": 1.0696,
+      "step": 209700
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5391618718566712e-05,
+      "loss": 1.0956,
+      "step": 209800
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5384652892907395e-05,
+      "loss": 1.0661,
+      "step": 209900
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.537768706724808e-05,
+      "loss": 1.0785,
+      "step": 210000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.5370721241588764e-05,
+      "loss": 1.0751,
+      "step": 210100
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.536375541592945e-05,
+      "loss": 1.1149,
+      "step": 210200
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.5356789590270134e-05,
+      "loss": 1.1071,
+      "step": 210300
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.534982376461082e-05,
+      "loss": 1.0892,
+      "step": 210400
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.5342857938951503e-05,
+      "loss": 1.0664,
+      "step": 210500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.533589211329219e-05,
+      "loss": 1.0815,
+      "step": 210600
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.5328926287632873e-05,
+      "loss": 1.07,
+      "step": 210700
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.532196046197356e-05,
+      "loss": 1.0735,
+      "step": 210800
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.5314994636314242e-05,
+      "loss": 1.0877,
+      "step": 210900
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.530802881065493e-05,
+      "loss": 1.0944,
+      "step": 211000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.530106298499561e-05,
+      "loss": 1.0892,
+      "step": 211100
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.5294097159336298e-05,
+      "loss": 1.0595,
+      "step": 211200
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.5287131333676984e-05,
+      "loss": 1.0714,
+      "step": 211300
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5280165508017667e-05,
+      "loss": 1.1189,
+      "step": 211400
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5273199682358353e-05,
+      "loss": 1.0914,
+      "step": 211500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5266233856699033e-05,
+      "loss": 1.0677,
+      "step": 211600
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.525926803103972e-05,
+      "loss": 1.0775,
+      "step": 211700
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5252302205380402e-05,
+      "loss": 1.065,
+      "step": 211800
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.524540603797768e-05,
+      "loss": 1.0778,
+      "step": 211900
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5238440212318365e-05,
+      "loss": 1.0582,
+      "step": 212000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.523147438665905e-05,
+      "loss": 1.0656,
+      "step": 212100
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.5224508560999735e-05,
+      "loss": 1.0801,
+      "step": 212200
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.521754273534042e-05,
+      "loss": 1.0756,
+      "step": 212300
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.5210576909681104e-05,
+      "loss": 1.0706,
+      "step": 212400
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.5203611084021789e-05,
+      "loss": 1.0565,
+      "step": 212500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.5196645258362473e-05,
+      "loss": 1.1021,
+      "step": 212600
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.5189749090959751e-05,
+      "loss": 1.0911,
+      "step": 212700
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.5182783265300436e-05,
+      "loss": 1.0772,
+      "step": 212800
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.517581743964112e-05,
+      "loss": 1.0479,
+      "step": 212900
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.5168851613981805e-05,
+      "loss": 1.0724,
+      "step": 213000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.516188578832249e-05,
+      "loss": 1.0897,
+      "step": 213100
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.5154919962663175e-05,
+      "loss": 1.0473,
+      "step": 213200
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.514795413700386e-05,
+      "loss": 1.0869,
+      "step": 213300
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.5140988311344544e-05,
+      "loss": 1.0665,
+      "step": 213400
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.5134022485685229e-05,
+      "loss": 1.076,
+      "step": 213500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5127056660025914e-05,
+      "loss": 1.107,
+      "step": 213600
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5120090834366598e-05,
+      "loss": 1.0896,
+      "step": 213700
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5113125008707283e-05,
+      "loss": 1.0652,
+      "step": 213800
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5106159183047968e-05,
+      "loss": 1.0795,
+      "step": 213900
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5099193357388652e-05,
+      "loss": 1.0711,
+      "step": 214000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5092227531729339e-05,
+      "loss": 1.0763,
+      "step": 214100
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.508526170607002e-05,
+      "loss": 1.0809,
+      "step": 214200
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5078295880410705e-05,
+      "loss": 1.0848,
+      "step": 214300
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.507133005475139e-05,
+      "loss": 1.1072,
+      "step": 214400
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5064364229092074e-05,
+      "loss": 1.0776,
+      "step": 214500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5057398403432759e-05,
+      "loss": 1.0591,
+      "step": 214600
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5050432577773443e-05,
+      "loss": 1.0677,
+      "step": 214700
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5043466752114128e-05,
+      "loss": 1.0943,
+      "step": 214800
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5036500926454813e-05,
+      "loss": 1.1069,
+      "step": 214900
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.5029535100795497e-05,
+      "loss": 1.0775,
+      "step": 215000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.5022569275136182e-05,
+      "loss": 1.0876,
+      "step": 215100
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.5015603449476867e-05,
+      "loss": 1.108,
+      "step": 215200
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.5008637623817552e-05,
+      "loss": 1.0907,
+      "step": 215300
+    },
+    {
+      "epoch": 3.0,
+      "eval_gen_len": 20.0,
+      "eval_loss": 1.145164132118225,
+      "eval_rouge1": 12.6221,
+      "eval_rouge2": 3.773,
+      "eval_rougeL": 12.1226,
+      "eval_rougeLsum": 12.2359,
+      "eval_runtime": 1533.982,
+      "eval_samples_per_second": 8.715,
+      "eval_steps_per_second": 2.179,
+      "step": 215337
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.5001671798158236e-05,
+      "loss": 0.9923,
+      "step": 215400
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4994705972498921e-05,
+      "loss": 0.9906,
+      "step": 215500
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4987740146839606e-05,
+      "loss": 0.9636,
+      "step": 215600
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.498077432118029e-05,
+      "loss": 0.9861,
+      "step": 215700
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4973808495520975e-05,
+      "loss": 0.9771,
+      "step": 215800
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4966842669861658e-05,
+      "loss": 0.9615,
+      "step": 215900
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4959876844202343e-05,
+      "loss": 0.9824,
+      "step": 216000
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4952911018543027e-05,
+      "loss": 0.9312,
+      "step": 216100
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4945945192883712e-05,
+      "loss": 0.9745,
+      "step": 216200
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4938979367224397e-05,
+      "loss": 0.9686,
+      "step": 216300
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.4932013541565081e-05,
+      "loss": 0.9777,
+      "step": 216400
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4925047715905768e-05,
+      "loss": 0.9322,
+      "step": 216500
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4918081890246452e-05,
+      "loss": 0.9771,
+      "step": 216600
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4911116064587137e-05,
+      "loss": 0.9627,
+      "step": 216700
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4904150238927822e-05,
+      "loss": 0.9897,
+      "step": 216800
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4897184413268506e-05,
+      "loss": 0.9642,
+      "step": 216900
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.489021858760919e-05,
+      "loss": 0.9726,
+      "step": 217000
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.4883252761949874e-05,
+      "loss": 0.9797,
+      "step": 217100
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4876286936290559e-05,
+      "loss": 0.991,
+      "step": 217200
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4869321110631243e-05,
+      "loss": 0.9746,
+      "step": 217300
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4862355284971928e-05,
+      "loss": 0.9629,
+      "step": 217400
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4855389459312613e-05,
+      "loss": 0.967,
+      "step": 217500
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4848423633653298e-05,
+      "loss": 0.985,
+      "step": 217600
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4841457807993982e-05,
+      "loss": 0.9809,
+      "step": 217700
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.4834491982334667e-05,
+      "loss": 0.9763,
+      "step": 217800
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.4827526156675352e-05,
+      "loss": 0.9669,
+      "step": 217900
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.4820560331016036e-05,
+      "loss": 0.9642,
+      "step": 218000
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.481359450535672e-05,
+      "loss": 0.9634,
+      "step": 218100
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.4806628679697404e-05,
+      "loss": 0.9793,
+      "step": 218200
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.4799662854038089e-05,
+      "loss": 0.9267,
+      "step": 218300
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.4792697028378773e-05,
+      "loss": 0.9831,
+      "step": 218400
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.4785731202719458e-05,
+      "loss": 0.9944,
+      "step": 218500
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.4778765377060143e-05,
+      "loss": 0.9952,
+      "step": 218600
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.4771799551400827e-05,
+      "loss": 0.9637,
+      "step": 218700
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.4764903383998105e-05,
+      "loss": 0.9783,
+      "step": 218800
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.475793755833879e-05,
+      "loss": 0.9377,
+      "step": 218900
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.4750971732679475e-05,
+      "loss": 0.9509,
+      "step": 219000
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.474400590702016e-05,
+      "loss": 0.99,
+      "step": 219100
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.4737040081360844e-05,
+      "loss": 0.9931,
+      "step": 219200
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.4730074255701529e-05,
+      "loss": 1.02,
+      "step": 219300
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.4723108430042212e-05,
+      "loss": 0.9756,
+      "step": 219400
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.4716142604382897e-05,
+      "loss": 1.015,
+      "step": 219500
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.4709176778723581e-05,
+      "loss": 0.9802,
+      "step": 219600
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.4702210953064266e-05,
+      "loss": 1.0137,
+      "step": 219700
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.4695314785661544e-05,
+      "loss": 0.9492,
+      "step": 219800
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.4688348960002229e-05,
+      "loss": 0.9642,
+      "step": 219900
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.4681383134342913e-05,
+      "loss": 0.9745,
+      "step": 220000
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.4674417308683598e-05,
+      "loss": 0.9685,
+      "step": 220100
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.4667451483024283e-05,
+      "loss": 0.9723,
+      "step": 220200
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.4660485657364967e-05,
+      "loss": 1.0045,
+      "step": 220300
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.4653519831705652e-05,
+      "loss": 0.9695,
+      "step": 220400
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.4646554006046337e-05,
+      "loss": 0.9483,
+      "step": 220500
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.4639588180387023e-05,
+      "loss": 0.9441,
+      "step": 220600
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.4632622354727706e-05,
+      "loss": 0.9568,
+      "step": 220700
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.4625656529068391e-05,
+      "loss": 0.9447,
+      "step": 220800
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.4618690703409076e-05,
+      "loss": 0.9856,
+      "step": 220900
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.461172487774976e-05,
+      "loss": 0.9654,
+      "step": 221000
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.4604759052090445e-05,
+      "loss": 0.9757,
+      "step": 221100
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.459779322643113e-05,
+      "loss": 0.9999,
+      "step": 221200
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.4590827400771814e-05,
+      "loss": 0.9977,
+      "step": 221300
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.4583861575112499e-05,
+      "loss": 0.9694,
+      "step": 221400
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.4576895749453184e-05,
+      "loss": 0.9623,
+      "step": 221500
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.4569929923793868e-05,
+      "loss": 0.9558,
+      "step": 221600
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.4562964098134553e-05,
+      "loss": 0.9392,
+      "step": 221700
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.4555998272475236e-05,
+      "loss": 0.9699,
+      "step": 221800
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.454903244681592e-05,
+      "loss": 0.9772,
+      "step": 221900
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.4542066621156605e-05,
+      "loss": 0.9531,
+      "step": 222000
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.453510079549729e-05,
+      "loss": 0.9337,
+      "step": 222100
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.4528134969837975e-05,
+      "loss": 1.0065,
+      "step": 222200
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.452116914417866e-05,
+      "loss": 0.9943,
+      "step": 222300
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.4514203318519344e-05,
+      "loss": 0.9692,
+      "step": 222400
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.4507237492860029e-05,
+      "loss": 0.9519,
+      "step": 222500
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.4500271667200713e-05,
+      "loss": 0.9975,
+      "step": 222600
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.4493305841541398e-05,
+      "loss": 0.9832,
+      "step": 222700
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.4486340015882083e-05,
+      "loss": 0.956,
+      "step": 222800
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.4479374190222766e-05,
+      "loss": 0.9643,
+      "step": 222900
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.4472408364563452e-05,
+      "loss": 0.9638,
+      "step": 223000
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.4465442538904137e-05,
+      "loss": 0.976,
+      "step": 223100
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.4458476713244822e-05,
+      "loss": 1.0135,
+      "step": 223200
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.4451510887585506e-05,
+      "loss": 0.9846,
+      "step": 223300
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.4444545061926191e-05,
+      "loss": 1.0115,
+      "step": 223400
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.4437579236266876e-05,
+      "loss": 0.9863,
+      "step": 223500
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.443061341060756e-05,
+      "loss": 0.9686,
+      "step": 223600
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.4423647584948245e-05,
+      "loss": 0.9709,
+      "step": 223700
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.441668175928893e-05,
+      "loss": 0.9712,
+      "step": 223800
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.4409785591886208e-05,
+      "loss": 0.9408,
+      "step": 223900
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.4402819766226892e-05,
+      "loss": 0.9603,
+      "step": 224000
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.4395853940567577e-05,
+      "loss": 0.9748,
+      "step": 224100
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.438888811490826e-05,
+      "loss": 0.9562,
+      "step": 224200
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.4381922289248945e-05,
+      "loss": 1.0136,
+      "step": 224300
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.437495646358963e-05,
+      "loss": 0.9822,
+      "step": 224400
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.4367990637930314e-05,
+      "loss": 1.0055,
+      "step": 224500
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.4361024812270999e-05,
+      "loss": 0.9875,
+      "step": 224600
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.4354058986611684e-05,
+      "loss": 1.0064,
+      "step": 224700
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.4347162819208962e-05,
+      "loss": 0.9762,
+      "step": 224800
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.4340196993549646e-05,
+      "loss": 0.9445,
+      "step": 224900
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.4333231167890331e-05,
+      "loss": 0.9514,
+      "step": 225000
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.4326265342231016e-05,
+      "loss": 0.9584,
+      "step": 225100
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.43192995165717e-05,
+      "loss": 1.0148,
+      "step": 225200
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.4312333690912385e-05,
+      "loss": 0.9802,
+      "step": 225300
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.430536786525307e-05,
+      "loss": 0.98,
+      "step": 225400
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.4298402039593753e-05,
+      "loss": 0.9218,
+      "step": 225500
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.4291436213934437e-05,
+      "loss": 0.9785,
+      "step": 225600
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.4284470388275122e-05,
+      "loss": 0.9941,
+      "step": 225700
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.4277504562615807e-05,
+      "loss": 0.9878,
+      "step": 225800
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.4270538736956492e-05,
+      "loss": 0.9692,
+      "step": 225900
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.4263572911297176e-05,
+      "loss": 0.9958,
+      "step": 226000
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.4256607085637861e-05,
+      "loss": 0.9872,
+      "step": 226100
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.4249641259978546e-05,
+      "loss": 0.9813,
+      "step": 226200
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.424267543431923e-05,
+      "loss": 0.9601,
+      "step": 226300
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.4235709608659915e-05,
+      "loss": 0.9642,
+      "step": 226400
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.42287437830006e-05,
+      "loss": 0.9719,
+      "step": 226500
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.4221777957341284e-05,
+      "loss": 0.9609,
+      "step": 226600
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.4214812131681967e-05,
+      "loss": 0.9422,
+      "step": 226700
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.4207846306022652e-05,
+      "loss": 0.9826,
+      "step": 226800
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.4200880480363337e-05,
+      "loss": 0.9508,
+      "step": 226900
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.4193914654704021e-05,
+      "loss": 0.9779,
+      "step": 227000
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.4186948829044708e-05,
+      "loss": 0.9608,
+      "step": 227100
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.4179983003385392e-05,
+      "loss": 0.9876,
+      "step": 227200
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.4173017177726077e-05,
+      "loss": 0.9501,
+      "step": 227300
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.4166051352066762e-05,
+      "loss": 0.9924,
+      "step": 227400
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.4159085526407446e-05,
+      "loss": 0.9669,
+      "step": 227500
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.4152119700748131e-05,
+      "loss": 0.9911,
+      "step": 227600
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.4145153875088816e-05,
+      "loss": 0.9584,
+      "step": 227700
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.4138188049429499e-05,
+      "loss": 0.9577,
+      "step": 227800
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.4131222223770183e-05,
+      "loss": 0.9707,
+      "step": 227900
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.4124326056367462e-05,
+      "loss": 0.9537,
+      "step": 228000
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.4117360230708146e-05,
+      "loss": 0.9705,
+      "step": 228100
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.4110394405048831e-05,
+      "loss": 0.9787,
+      "step": 228200
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.4103428579389516e-05,
+      "loss": 0.9366,
+      "step": 228300
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.40964627537302e-05,
+      "loss": 0.9885,
+      "step": 228400
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.4089496928070885e-05,
+      "loss": 1.0184,
+      "step": 228500
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.408253110241157e-05,
+      "loss": 1.0104,
+      "step": 228600
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.4075565276752254e-05,
+      "loss": 0.9694,
+      "step": 228700
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.4068599451092939e-05,
+      "loss": 0.9776,
+      "step": 228800
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.4061633625433624e-05,
+      "loss": 0.9612,
+      "step": 228900
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.4054667799774308e-05,
+      "loss": 0.9772,
+      "step": 229000
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.4047701974114991e-05,
+      "loss": 0.9404,
+      "step": 229100
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.4040736148455676e-05,
+      "loss": 0.927,
+      "step": 229200
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.403377032279636e-05,
+      "loss": 0.9735,
+      "step": 229300
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.4026804497137045e-05,
+      "loss": 0.9728,
+      "step": 229400
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.401983867147773e-05,
+      "loss": 0.9853,
+      "step": 229500
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.4012872845818415e-05,
+      "loss": 0.9633,
+      "step": 229600
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.40059070201591e-05,
+      "loss": 0.9723,
+      "step": 229700
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3998941194499784e-05,
+      "loss": 0.9575,
+      "step": 229800
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3991975368840469e-05,
+      "loss": 0.9777,
+      "step": 229900
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3985009543181154e-05,
+      "loss": 0.9754,
+      "step": 230000
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3978043717521838e-05,
+      "loss": 0.9709,
+      "step": 230100
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3971077891862521e-05,
+      "loss": 0.9731,
+      "step": 230200
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3964112066203206e-05,
+      "loss": 0.9468,
+      "step": 230300
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3957146240543892e-05,
+      "loss": 1.0028,
+      "step": 230400
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3950180414884577e-05,
+      "loss": 0.9617,
+      "step": 230500
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3943214589225262e-05,
+      "loss": 0.9848,
+      "step": 230600
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3936248763565946e-05,
+      "loss": 0.9833,
+      "step": 230700
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3929282937906631e-05,
+      "loss": 0.9816,
+      "step": 230800
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3922317112247316e-05,
+      "loss": 0.9549,
+      "step": 230900
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3915351286588e-05,
+      "loss": 0.9583,
+      "step": 231000
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3908385460928685e-05,
+      "loss": 0.99,
+      "step": 231100
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.390141963526937e-05,
+      "loss": 0.9897,
+      "step": 231200
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3894453809610053e-05,
+      "loss": 0.9987,
+      "step": 231300
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3887487983950737e-05,
+      "loss": 0.9997,
+      "step": 231400
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3880522158291422e-05,
+      "loss": 0.9936,
+      "step": 231500
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3873556332632107e-05,
+      "loss": 0.9856,
+      "step": 231600
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3866660165229385e-05,
+      "loss": 0.9799,
+      "step": 231700
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.385969433957007e-05,
+      "loss": 0.9946,
+      "step": 231800
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3852728513910754e-05,
+      "loss": 0.9955,
+      "step": 231900
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3845762688251439e-05,
+      "loss": 0.9836,
+      "step": 232000
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3838796862592124e-05,
+      "loss": 0.9599,
+      "step": 232100
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3831831036932808e-05,
+      "loss": 0.9956,
+      "step": 232200
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3824865211273493e-05,
+      "loss": 1.0097,
+      "step": 232300
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3817899385614178e-05,
+      "loss": 1.0054,
+      "step": 232400
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3810933559954862e-05,
+      "loss": 0.9995,
+      "step": 232500
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3803967734295545e-05,
+      "loss": 0.9805,
+      "step": 232600
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.379700190863623e-05,
+      "loss": 0.9738,
+      "step": 232700
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3790036082976915e-05,
+      "loss": 0.969,
+      "step": 232800
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.37830702573176e-05,
+      "loss": 0.9727,
+      "step": 232900
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3776104431658284e-05,
+      "loss": 0.9536,
+      "step": 233000
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3769138605998969e-05,
+      "loss": 0.9871,
+      "step": 233100
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3762172780339654e-05,
+      "loss": 1.0029,
+      "step": 233200
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3755206954680338e-05,
+      "loss": 0.9848,
+      "step": 233300
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3748241129021023e-05,
+      "loss": 1.0025,
+      "step": 233400
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3741275303361708e-05,
+      "loss": 0.9891,
+      "step": 233500
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3734309477702394e-05,
+      "loss": 0.997,
+      "step": 233600
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3727343652043079e-05,
+      "loss": 0.9582,
+      "step": 233700
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3720377826383762e-05,
+      "loss": 0.9687,
+      "step": 233800
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3713412000724446e-05,
+      "loss": 0.9688,
+      "step": 233900
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3706446175065131e-05,
+      "loss": 0.9618,
+      "step": 234000
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3699550007662407e-05,
+      "loss": 0.9736,
+      "step": 234100
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3692584182003092e-05,
+      "loss": 0.983,
+      "step": 234200
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3685618356343777e-05,
+      "loss": 0.963,
+      "step": 234300
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3678652530684463e-05,
+      "loss": 1.0001,
+      "step": 234400
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3671686705025148e-05,
+      "loss": 0.9748,
+      "step": 234500
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3664720879365833e-05,
+      "loss": 0.9721,
+      "step": 234600
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3657755053706517e-05,
+      "loss": 0.9612,
+      "step": 234700
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3650789228047202e-05,
+      "loss": 0.993,
+      "step": 234800
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3643823402387887e-05,
+      "loss": 0.9825,
+      "step": 234900
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3636857576728571e-05,
+      "loss": 1.0081,
+      "step": 235000
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3629891751069254e-05,
+      "loss": 0.9997,
+      "step": 235100
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3622925925409939e-05,
+      "loss": 0.9768,
+      "step": 235200
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3615960099750624e-05,
+      "loss": 1.0161,
+      "step": 235300
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3608994274091308e-05,
+      "loss": 1.0124,
+      "step": 235400
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3602028448431993e-05,
+      "loss": 0.9675,
+      "step": 235500
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3595062622772678e-05,
+      "loss": 0.9708,
+      "step": 235600
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3588096797113362e-05,
+      "loss": 0.9902,
+      "step": 235700
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3581130971454047e-05,
+      "loss": 0.983,
+      "step": 235800
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3574165145794732e-05,
+      "loss": 0.9454,
+      "step": 235900
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3567199320135416e-05,
+      "loss": 0.9914,
+      "step": 236000
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3560233494476101e-05,
+      "loss": 1.0168,
+      "step": 236100
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3553267668816784e-05,
+      "loss": 0.9914,
+      "step": 236200
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3546301843157469e-05,
+      "loss": 0.9612,
+      "step": 236300
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3539336017498153e-05,
+      "loss": 0.974,
+      "step": 236400
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3532370191838838e-05,
+      "loss": 0.9705,
+      "step": 236500
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3525404366179523e-05,
+      "loss": 0.9984,
+      "step": 236600
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3518438540520207e-05,
+      "loss": 0.9975,
+      "step": 236700
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3511472714860892e-05,
+      "loss": 0.9893,
+      "step": 236800
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3504506889201579e-05,
+      "loss": 0.9882,
+      "step": 236900
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3497541063542263e-05,
+      "loss": 1.0054,
+      "step": 237000
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3490575237882948e-05,
+      "loss": 0.9796,
+      "step": 237100
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3483609412223633e-05,
+      "loss": 0.9589,
+      "step": 237200
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3476713244820909e-05,
+      "loss": 1.0009,
+      "step": 237300
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3469747419161594e-05,
+      "loss": 0.9728,
+      "step": 237400
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3462781593502277e-05,
+      "loss": 0.9735,
+      "step": 237500
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3455815767842961e-05,
+      "loss": 0.9765,
+      "step": 237600
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3448849942183648e-05,
+      "loss": 0.9593,
+      "step": 237700
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3441884116524332e-05,
+      "loss": 1.0069,
+      "step": 237800
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3434918290865017e-05,
+      "loss": 0.9707,
+      "step": 237900
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3427952465205702e-05,
+      "loss": 0.9854,
+      "step": 238000
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3420986639546386e-05,
+      "loss": 0.9702,
+      "step": 238100
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3414020813887071e-05,
+      "loss": 1.0152,
+      "step": 238200
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3407054988227756e-05,
+      "loss": 1.0205,
+      "step": 238300
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.340008916256844e-05,
+      "loss": 0.9861,
+      "step": 238400
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3393123336909125e-05,
+      "loss": 1.0026,
+      "step": 238500
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3386157511249808e-05,
+      "loss": 1.0146,
+      "step": 238600
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3379191685590493e-05,
+      "loss": 0.9556,
+      "step": 238700
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3372225859931178e-05,
+      "loss": 0.9612,
+      "step": 238800
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3365260034271862e-05,
+      "loss": 0.9914,
+      "step": 238900
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3358294208612547e-05,
+      "loss": 0.9871,
+      "step": 239000
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3351328382953232e-05,
+      "loss": 0.995,
+      "step": 239100
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3344362557293916e-05,
+      "loss": 0.9541,
+      "step": 239200
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3337396731634601e-05,
+      "loss": 0.9721,
+      "step": 239300
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3330430905975286e-05,
+      "loss": 1.0121,
+      "step": 239400
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.332346508031597e-05,
+      "loss": 0.9509,
+      "step": 239500
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3316568912913248e-05,
+      "loss": 0.9673,
+      "step": 239600
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3309603087253933e-05,
+      "loss": 0.9598,
+      "step": 239700
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3302637261594618e-05,
+      "loss": 0.982,
+      "step": 239800
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.32956714359353e-05,
+      "loss": 0.9916,
+      "step": 239900
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3288705610275986e-05,
+      "loss": 0.9989,
+      "step": 240000
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.328173978461667e-05,
+      "loss": 0.9761,
+      "step": 240100
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3274773958957355e-05,
+      "loss": 0.9695,
+      "step": 240200
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.326780813329804e-05,
+      "loss": 0.9946,
+      "step": 240300
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3260842307638724e-05,
+      "loss": 0.9662,
+      "step": 240400
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3253876481979409e-05,
+      "loss": 1.0043,
+      "step": 240500
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3246910656320094e-05,
+      "loss": 0.9889,
+      "step": 240600
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3239944830660778e-05,
+      "loss": 0.9748,
+      "step": 240700
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3232979005001463e-05,
+      "loss": 0.9632,
+      "step": 240800
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3226013179342148e-05,
+      "loss": 1.026,
+      "step": 240900
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3219047353682832e-05,
+      "loss": 0.9639,
+      "step": 241000
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3212081528023517e-05,
+      "loss": 0.973,
+      "step": 241100
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3205115702364202e-05,
+      "loss": 0.9861,
+      "step": 241200
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3198149876704886e-05,
+      "loss": 0.9995,
+      "step": 241300
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3191184051045571e-05,
+      "loss": 0.9796,
+      "step": 241400
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3184218225386256e-05,
+      "loss": 0.9778,
+      "step": 241500
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.317725239972694e-05,
+      "loss": 0.9788,
+      "step": 241600
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.3170286574067625e-05,
+      "loss": 1.0043,
+      "step": 241700
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.316332074840831e-05,
+      "loss": 0.9934,
+      "step": 241800
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.3156354922748994e-05,
+      "loss": 0.976,
+      "step": 241900
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.3149458755346273e-05,
+      "loss": 0.9997,
+      "step": 242000
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.3142492929686957e-05,
+      "loss": 0.9845,
+      "step": 242100
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.3135527104027642e-05,
+      "loss": 0.9935,
+      "step": 242200
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.3128561278368325e-05,
+      "loss": 0.9589,
+      "step": 242300
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.312159545270901e-05,
+      "loss": 0.9656,
+      "step": 242400
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.3114629627049694e-05,
+      "loss": 0.9979,
+      "step": 242500
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.3107663801390379e-05,
+      "loss": 0.9738,
+      "step": 242600
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.3100697975731064e-05,
+      "loss": 1.0074,
+      "step": 242700
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.3093732150071748e-05,
+      "loss": 1.0119,
+      "step": 242800
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.3086766324412433e-05,
+      "loss": 0.9412,
+      "step": 242900
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3079800498753118e-05,
+      "loss": 1.0002,
+      "step": 243000
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3072834673093802e-05,
+      "loss": 0.9525,
+      "step": 243100
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3065868847434487e-05,
+      "loss": 0.959,
+      "step": 243200
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3058903021775172e-05,
+      "loss": 1.0097,
+      "step": 243300
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3051937196115855e-05,
+      "loss": 0.9807,
+      "step": 243400
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.304497137045654e-05,
+      "loss": 0.9537,
+      "step": 243500
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.3038005544797224e-05,
+      "loss": 0.9929,
+      "step": 243600
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3031039719137909e-05,
+      "loss": 0.9862,
+      "step": 243700
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3024143551735187e-05,
+      "loss": 0.9913,
+      "step": 243800
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3017177726075872e-05,
+      "loss": 0.9867,
+      "step": 243900
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3010211900416556e-05,
+      "loss": 0.9704,
+      "step": 244000
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.3003246074757241e-05,
+      "loss": 0.9614,
+      "step": 244100
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2996280249097926e-05,
+      "loss": 0.9972,
+      "step": 244200
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.298931442343861e-05,
+      "loss": 0.9898,
+      "step": 244300
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2982348597779295e-05,
+      "loss": 0.9753,
+      "step": 244400
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.297538277211998e-05,
+      "loss": 0.9839,
+      "step": 244500
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2968416946460664e-05,
+      "loss": 0.9777,
+      "step": 244600
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2961451120801347e-05,
+      "loss": 0.9914,
+      "step": 244700
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2954485295142032e-05,
+      "loss": 0.9803,
+      "step": 244800
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2947519469482717e-05,
+      "loss": 0.9592,
+      "step": 244900
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2940553643823403e-05,
+      "loss": 0.9881,
+      "step": 245000
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2933587818164088e-05,
+      "loss": 0.9817,
+      "step": 245100
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2926621992504773e-05,
+      "loss": 0.9738,
+      "step": 245200
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2919656166845457e-05,
+      "loss": 0.9989,
+      "step": 245300
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2912690341186142e-05,
+      "loss": 0.9551,
+      "step": 245400
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2905724515526827e-05,
+      "loss": 0.9904,
+      "step": 245500
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2898758689867511e-05,
+      "loss": 0.9919,
+      "step": 245600
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2891792864208196e-05,
+      "loss": 0.9967,
+      "step": 245700
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.288482703854888e-05,
+      "loss": 1.0002,
+      "step": 245800
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.2877861212889564e-05,
+      "loss": 0.9913,
+      "step": 245900
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.2870895387230248e-05,
+      "loss": 0.9849,
+      "step": 246000
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.2863929561570933e-05,
+      "loss": 0.9998,
+      "step": 246100
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.2856963735911618e-05,
+      "loss": 1.0058,
+      "step": 246200
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.2849997910252302e-05,
+      "loss": 0.9815,
+      "step": 246300
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.2843032084592987e-05,
+      "loss": 0.9764,
+      "step": 246400
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.2836066258933672e-05,
+      "loss": 1.0131,
+      "step": 246500
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.2829100433274356e-05,
+      "loss": 0.9728,
+      "step": 246600
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.2822134607615041e-05,
+      "loss": 0.9944,
+      "step": 246700
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.2815168781955726e-05,
+      "loss": 0.9332,
+      "step": 246800
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.280820295629641e-05,
+      "loss": 0.961,
+      "step": 246900
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.2801237130637093e-05,
+      "loss": 1.0046,
+      "step": 247000
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.2794271304977778e-05,
+      "loss": 0.975,
+      "step": 247100
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.2787305479318463e-05,
+      "loss": 0.9783,
+      "step": 247200
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.2780339653659147e-05,
+      "loss": 0.9852,
+      "step": 247300
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.2773373827999832e-05,
+      "loss": 1.0075,
+      "step": 247400
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.2766408002340519e-05,
+      "loss": 0.9754,
+      "step": 247500
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.2759442176681203e-05,
+      "loss": 0.9657,
+      "step": 247600
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.2752476351021888e-05,
+      "loss": 1.0082,
+      "step": 247700
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.2745510525362573e-05,
+      "loss": 0.9673,
+      "step": 247800
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.2738544699703257e-05,
+      "loss": 0.9751,
+      "step": 247900
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.2731578874043942e-05,
+      "loss": 1.0117,
+      "step": 248000
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.2724613048384625e-05,
+      "loss": 1.0074,
+      "step": 248100
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.2717716880981903e-05,
+      "loss": 1.0053,
+      "step": 248200
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.2710751055322588e-05,
+      "loss": 1.0105,
+      "step": 248300
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.2703785229663272e-05,
+      "loss": 0.9761,
+      "step": 248400
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.2696819404003957e-05,
+      "loss": 0.9702,
+      "step": 248500
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.2689853578344642e-05,
+      "loss": 1.0102,
+      "step": 248600
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.2682887752685327e-05,
+      "loss": 0.9613,
+      "step": 248700
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.2675921927026011e-05,
+      "loss": 0.9878,
+      "step": 248800
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.2668956101366696e-05,
+      "loss": 0.9962,
+      "step": 248900
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.266199027570738e-05,
+      "loss": 0.9722,
+      "step": 249000
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.2655024450048065e-05,
+      "loss": 0.9736,
+      "step": 249100
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.264805862438875e-05,
+      "loss": 0.9761,
+      "step": 249200
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.2641092798729435e-05,
+      "loss": 0.9865,
+      "step": 249300
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.2634126973070118e-05,
+      "loss": 1.0022,
+      "step": 249400
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.2627161147410802e-05,
+      "loss": 0.9916,
+      "step": 249500
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.2620195321751487e-05,
+      "loss": 0.9595,
+      "step": 249600
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.2613229496092172e-05,
+      "loss": 0.9772,
+      "step": 249700
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.2606263670432856e-05,
+      "loss": 0.99,
+      "step": 249800
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.2599297844773541e-05,
+      "loss": 0.9679,
+      "step": 249900
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.2592332019114226e-05,
+      "loss": 1.0168,
+      "step": 250000
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.258536619345491e-05,
+      "loss": 0.9696,
+      "step": 250100
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.2578400367795595e-05,
+      "loss": 0.9751,
+      "step": 250200
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.257143454213628e-05,
+      "loss": 0.9922,
+      "step": 250300
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.2564468716476964e-05,
+      "loss": 0.9825,
+      "step": 250400
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.2557572549074243e-05,
+      "loss": 0.9756,
+      "step": 250500
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.2550606723414927e-05,
+      "loss": 0.9796,
+      "step": 250600
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.254364089775561e-05,
+      "loss": 0.9671,
+      "step": 250700
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.2536675072096295e-05,
+      "loss": 0.9738,
+      "step": 250800
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.252970924643698e-05,
+      "loss": 1.0037,
+      "step": 250900
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.2522743420777664e-05,
+      "loss": 0.9856,
+      "step": 251000
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.2515777595118349e-05,
+      "loss": 0.9816,
+      "step": 251100
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.2508811769459034e-05,
+      "loss": 0.982,
+      "step": 251200
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.2501845943799718e-05,
+      "loss": 0.9655,
+      "step": 251300
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.2494880118140403e-05,
+      "loss": 0.9474,
+      "step": 251400
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.2487914292481088e-05,
+      "loss": 1.0095,
+      "step": 251500
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.2480948466821774e-05,
+      "loss": 0.9443,
+      "step": 251600
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.2473982641162459e-05,
+      "loss": 1.0083,
+      "step": 251700
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.2467016815503142e-05,
+      "loss": 0.9412,
+      "step": 251800
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.2460050989843826e-05,
+      "loss": 0.9921,
+      "step": 251900
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.2453085164184511e-05,
+      "loss": 0.9875,
+      "step": 252000
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.2446119338525196e-05,
+      "loss": 0.9961,
+      "step": 252100
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.243915351286588e-05,
+      "loss": 0.9483,
+      "step": 252200
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.2432187687206565e-05,
+      "loss": 0.9692,
+      "step": 252300
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.242522186154725e-05,
+      "loss": 0.9758,
+      "step": 252400
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.2418256035887935e-05,
+      "loss": 0.9745,
+      "step": 252500
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.241129021022862e-05,
+      "loss": 0.9939,
+      "step": 252600
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.2404324384569304e-05,
+      "loss": 1.0025,
+      "step": 252700
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.2397358558909989e-05,
+      "loss": 0.9738,
+      "step": 252800
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.2390392733250673e-05,
+      "loss": 0.957,
+      "step": 252900
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.2383426907591356e-05,
+      "loss": 0.9619,
+      "step": 253000
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.2376461081932041e-05,
+      "loss": 0.9671,
+      "step": 253100
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.2369564914529319e-05,
+      "loss": 0.9897,
+      "step": 253200
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.2362599088870004e-05,
+      "loss": 0.9783,
+      "step": 253300
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.2355633263210688e-05,
+      "loss": 0.9969,
+      "step": 253400
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.2348667437551373e-05,
+      "loss": 0.9811,
+      "step": 253500
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.2341701611892058e-05,
+      "loss": 0.9867,
+      "step": 253600
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.2334805444489336e-05,
+      "loss": 1.0169,
+      "step": 253700
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.232783961883002e-05,
+      "loss": 0.9899,
+      "step": 253800
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.2320873793170705e-05,
+      "loss": 1.0126,
+      "step": 253900
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.231390796751139e-05,
+      "loss": 0.9635,
+      "step": 254000
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.2306942141852075e-05,
+      "loss": 1.0019,
+      "step": 254100
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.229997631619276e-05,
+      "loss": 0.9804,
+      "step": 254200
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.2293010490533444e-05,
+      "loss": 0.9913,
+      "step": 254300
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.2286044664874127e-05,
+      "loss": 0.9763,
+      "step": 254400
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.2279078839214812e-05,
+      "loss": 0.9634,
+      "step": 254500
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.2272113013555496e-05,
+      "loss": 0.9622,
+      "step": 254600
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.2265147187896181e-05,
+      "loss": 0.9788,
+      "step": 254700
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.2258181362236866e-05,
+      "loss": 1.0044,
+      "step": 254800
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.225121553657755e-05,
+      "loss": 0.9843,
+      "step": 254900
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.2244249710918235e-05,
+      "loss": 0.9691,
+      "step": 255000
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.223728388525892e-05,
+      "loss": 0.9985,
+      "step": 255100
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2230318059599604e-05,
+      "loss": 0.9796,
+      "step": 255200
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2223352233940289e-05,
+      "loss": 0.9725,
+      "step": 255300
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2216386408280974e-05,
+      "loss": 0.9806,
+      "step": 255400
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2209420582621659e-05,
+      "loss": 0.9748,
+      "step": 255500
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2202454756962343e-05,
+      "loss": 0.9934,
+      "step": 255600
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2195488931303028e-05,
+      "loss": 0.989,
+      "step": 255700
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.2188523105643713e-05,
+      "loss": 1.0401,
+      "step": 255800
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2181557279984397e-05,
+      "loss": 1.0054,
+      "step": 255900
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2174591454325082e-05,
+      "loss": 0.9631,
+      "step": 256000
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2167625628665767e-05,
+      "loss": 0.9756,
+      "step": 256100
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2160729461263043e-05,
+      "loss": 0.9753,
+      "step": 256200
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2153763635603728e-05,
+      "loss": 0.9604,
+      "step": 256300
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2146797809944412e-05,
+      "loss": 0.9536,
+      "step": 256400
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2139831984285099e-05,
+      "loss": 0.9592,
+      "step": 256500
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.2132866158625783e-05,
+      "loss": 0.9903,
+      "step": 256600
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2125900332966468e-05,
+      "loss": 0.9939,
+      "step": 256700
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2118934507307153e-05,
+      "loss": 0.9895,
+      "step": 256800
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2111968681647836e-05,
+      "loss": 1.0161,
+      "step": 256900
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.210500285598852e-05,
+      "loss": 0.9908,
+      "step": 257000
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2098037030329205e-05,
+      "loss": 0.968,
+      "step": 257100
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.209107120466989e-05,
+      "loss": 0.9951,
+      "step": 257200
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.2084105379010575e-05,
+      "loss": 1.0156,
+      "step": 257300
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.207713955335126e-05,
+      "loss": 0.9902,
+      "step": 257400
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2070173727691944e-05,
+      "loss": 0.9568,
+      "step": 257500
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2063207902032629e-05,
+      "loss": 0.9984,
+      "step": 257600
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2056242076373313e-05,
+      "loss": 0.9871,
+      "step": 257700
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2049276250713998e-05,
+      "loss": 0.9664,
+      "step": 257800
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2042310425054683e-05,
+      "loss": 0.9597,
+      "step": 257900
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.2035344599395366e-05,
+      "loss": 1.025,
+      "step": 258000
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.202837877373605e-05,
+      "loss": 0.9871,
+      "step": 258100
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2021412948076735e-05,
+      "loss": 1.0194,
+      "step": 258200
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.201444712241742e-05,
+      "loss": 1.0025,
+      "step": 258300
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2007481296758104e-05,
+      "loss": 1.0267,
+      "step": 258400
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.2000515471098789e-05,
+      "loss": 0.9666,
+      "step": 258500
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.1993549645439474e-05,
+      "loss": 0.9595,
+      "step": 258600
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.1986583819780158e-05,
+      "loss": 0.9739,
+      "step": 258700
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.1979617994120843e-05,
+      "loss": 0.9822,
+      "step": 258800
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.197265216846153e-05,
+      "loss": 0.975,
+      "step": 258900
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.1965686342802214e-05,
+      "loss": 0.9816,
+      "step": 259000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.1958720517142897e-05,
+      "loss": 0.9882,
+      "step": 259100
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.1951754691483582e-05,
+      "loss": 0.955,
+      "step": 259200
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.1944788865824267e-05,
+      "loss": 1.0038,
+      "step": 259300
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.1937823040164951e-05,
+      "loss": 0.9749,
+      "step": 259400
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.1930857214505636e-05,
+      "loss": 1.0001,
+      "step": 259500
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.192389138884632e-05,
+      "loss": 0.9507,
+      "step": 259600
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.1916925563187005e-05,
+      "loss": 0.9973,
+      "step": 259700
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.190995973752769e-05,
+      "loss": 0.989,
+      "step": 259800
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.1902993911868375e-05,
+      "loss": 1.0236,
+      "step": 259900
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.189602808620906e-05,
+      "loss": 0.9796,
+      "step": 260000
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.1889062260549744e-05,
+      "loss": 1.0098,
+      "step": 260100
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.1882096434890427e-05,
+      "loss": 0.986,
+      "step": 260200
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.1875130609231112e-05,
+      "loss": 0.9521,
+      "step": 260300
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.1868164783571796e-05,
+      "loss": 0.9656,
+      "step": 260400
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.1861198957912481e-05,
+      "loss": 0.9844,
+      "step": 260500
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.1854233132253166e-05,
+      "loss": 0.9965,
+      "step": 260600
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.184726730659385e-05,
+      "loss": 0.9732,
+      "step": 260700
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.1840371139191129e-05,
+      "loss": 0.9969,
+      "step": 260800
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.1833405313531813e-05,
+      "loss": 0.9837,
+      "step": 260900
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.1826439487872498e-05,
+      "loss": 1.0085,
+      "step": 261000
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.1819473662213183e-05,
+      "loss": 0.9855,
+      "step": 261100
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.1812507836553867e-05,
+      "loss": 0.9981,
+      "step": 261200
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.1805542010894552e-05,
+      "loss": 1.0086,
+      "step": 261300
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.1798576185235237e-05,
+      "loss": 0.9939,
+      "step": 261400
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.179161035957592e-05,
+      "loss": 1.0008,
+      "step": 261500
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.1784644533916604e-05,
+      "loss": 0.9769,
+      "step": 261600
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.1777678708257289e-05,
+      "loss": 1.0016,
+      "step": 261700
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.1770712882597974e-05,
+      "loss": 1.0011,
+      "step": 261800
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.1763747056938658e-05,
+      "loss": 0.9821,
+      "step": 261900
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.1756781231279343e-05,
+      "loss": 0.9878,
+      "step": 262000
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.1749815405620028e-05,
+      "loss": 1.0042,
+      "step": 262100
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.1742849579960714e-05,
+      "loss": 0.9959,
+      "step": 262200
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.1735883754301399e-05,
+      "loss": 0.9901,
+      "step": 262300
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.1728917928642083e-05,
+      "loss": 0.9912,
+      "step": 262400
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.1721952102982768e-05,
+      "loss": 0.9897,
+      "step": 262500
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.1714986277323451e-05,
+      "loss": 0.949,
+      "step": 262600
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.1708020451664136e-05,
+      "loss": 0.9607,
+      "step": 262700
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.170105462600482e-05,
+      "loss": 0.9794,
+      "step": 262800
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.1694088800345505e-05,
+      "loss": 0.9869,
+      "step": 262900
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.168712297468619e-05,
+      "loss": 0.9877,
+      "step": 263000
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.1680157149026875e-05,
+      "loss": 0.9649,
+      "step": 263100
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.167319132336756e-05,
+      "loss": 0.9846,
+      "step": 263200
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.1666225497708244e-05,
+      "loss": 0.994,
+      "step": 263300
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.1659259672048929e-05,
+      "loss": 0.95,
+      "step": 263400
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.1652293846389613e-05,
+      "loss": 0.9849,
+      "step": 263500
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.1645328020730298e-05,
+      "loss": 0.9534,
+      "step": 263600
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.1638362195070983e-05,
+      "loss": 0.97,
+      "step": 263700
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.1631396369411666e-05,
+      "loss": 1.004,
+      "step": 263800
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.162443054375235e-05,
+      "loss": 0.9785,
+      "step": 263900
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.1617534376349628e-05,
+      "loss": 0.9533,
+      "step": 264000
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.1610568550690313e-05,
+      "loss": 0.9831,
+      "step": 264100
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.1603602725030998e-05,
+      "loss": 1.0029,
+      "step": 264200
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.1596706557628276e-05,
+      "loss": 0.9957,
+      "step": 264300
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.158974073196896e-05,
+      "loss": 1.0092,
+      "step": 264400
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.1582774906309645e-05,
+      "loss": 0.9792,
+      "step": 264500
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.157580908065033e-05,
+      "loss": 0.9532,
+      "step": 264600
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.1568843254991015e-05,
+      "loss": 0.959,
+      "step": 264700
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.15618774293317e-05,
+      "loss": 0.9557,
+      "step": 264800
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.1554911603672384e-05,
+      "loss": 0.9499,
+      "step": 264900
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.1547945778013069e-05,
+      "loss": 0.9418,
+      "step": 265000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.1540979952353753e-05,
+      "loss": 1.0181,
+      "step": 265100
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.1534014126694436e-05,
+      "loss": 0.9944,
+      "step": 265200
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.1527048301035121e-05,
+      "loss": 0.9985,
+      "step": 265300
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.1520082475375806e-05,
+      "loss": 0.9748,
+      "step": 265400
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.151311664971649e-05,
+      "loss": 0.9544,
+      "step": 265500
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.1506150824057175e-05,
+      "loss": 0.9731,
+      "step": 265600
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.149918499839786e-05,
+      "loss": 0.9681,
+      "step": 265700
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.1492219172738544e-05,
+      "loss": 0.9926,
+      "step": 265800
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.148525334707923e-05,
+      "loss": 0.9781,
+      "step": 265900
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.1478287521419914e-05,
+      "loss": 0.9916,
+      "step": 266000
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.1471321695760599e-05,
+      "loss": 0.9642,
+      "step": 266100
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.1464355870101283e-05,
+      "loss": 0.9716,
+      "step": 266200
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.145739004444197e-05,
+      "loss": 0.9785,
+      "step": 266300
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.1450424218782653e-05,
+      "loss": 0.9607,
+      "step": 266400
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.1443458393123337e-05,
+      "loss": 0.972,
+      "step": 266500
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.1436492567464022e-05,
+      "loss": 0.9862,
+      "step": 266600
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.1429526741804707e-05,
+      "loss": 0.9673,
+      "step": 266700
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.1422560916145391e-05,
+      "loss": 0.9721,
+      "step": 266800
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.1415595090486076e-05,
+      "loss": 0.9615,
+      "step": 266900
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.140862926482676e-05,
+      "loss": 0.9706,
+      "step": 267000
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.1401663439167445e-05,
+      "loss": 1.0159,
+      "step": 267100
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.139469761350813e-05,
+      "loss": 0.997,
+      "step": 267200
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.1387731787848815e-05,
+      "loss": 0.97,
+      "step": 267300
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.13807659621895e-05,
+      "loss": 1.0179,
+      "step": 267400
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.1373800136530182e-05,
+      "loss": 0.9738,
+      "step": 267500
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.1366834310870867e-05,
+      "loss": 0.9676,
+      "step": 267600
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.1359868485211552e-05,
+      "loss": 0.9892,
+      "step": 267700
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.1352902659552236e-05,
+      "loss": 1.0037,
+      "step": 267800
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.1345936833892921e-05,
+      "loss": 1.0042,
+      "step": 267900
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.1338971008233606e-05,
+      "loss": 0.9883,
+      "step": 268000
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.133200518257429e-05,
+      "loss": 1.0058,
+      "step": 268100
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.1325039356914975e-05,
+      "loss": 0.9622,
+      "step": 268200
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.131807353125566e-05,
+      "loss": 1.002,
+      "step": 268300
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.1311107705596345e-05,
+      "loss": 0.9946,
+      "step": 268400
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.130414187993703e-05,
+      "loss": 1.0038,
+      "step": 268500
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.1297176054277712e-05,
+      "loss": 0.9864,
+      "step": 268600
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.1290210228618399e-05,
+      "loss": 0.9908,
+      "step": 268700
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.1283244402959083e-05,
+      "loss": 1.0071,
+      "step": 268800
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.1276278577299768e-05,
+      "loss": 0.9852,
+      "step": 268900
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.1269382409897044e-05,
+      "loss": 0.9787,
+      "step": 269000
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.1262416584237729e-05,
+      "loss": 0.9853,
+      "step": 269100
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.1255450758578414e-05,
+      "loss": 0.9849,
+      "step": 269200
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.1248484932919098e-05,
+      "loss": 0.9848,
+      "step": 269300
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.1241519107259783e-05,
+      "loss": 0.967,
+      "step": 269400
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.123455328160047e-05,
+      "loss": 0.9804,
+      "step": 269500
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.1227587455941154e-05,
+      "loss": 0.9966,
+      "step": 269600
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.1220621630281839e-05,
+      "loss": 0.9848,
+      "step": 269700
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.1213655804622524e-05,
+      "loss": 0.9583,
+      "step": 269800
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.1206689978963207e-05,
+      "loss": 0.9921,
+      "step": 269900
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.1199724153303891e-05,
+      "loss": 0.9721,
+      "step": 270000
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.1192758327644576e-05,
+      "loss": 0.9834,
+      "step": 270100
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.118579250198526e-05,
+      "loss": 0.9992,
+      "step": 270200
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.1178826676325945e-05,
+      "loss": 0.9495,
+      "step": 270300
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.117186085066663e-05,
+      "loss": 0.9699,
+      "step": 270400
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.1164895025007315e-05,
+      "loss": 0.9755,
+      "step": 270500
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.1157929199348e-05,
+      "loss": 0.9905,
+      "step": 270600
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.1150963373688684e-05,
+      "loss": 0.9889,
+      "step": 270700
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.1143997548029369e-05,
+      "loss": 0.9678,
+      "step": 270800
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.1137031722370053e-05,
+      "loss": 0.9644,
+      "step": 270900
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.1130065896710736e-05,
+      "loss": 0.9962,
+      "step": 271000
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.1123100071051421e-05,
+      "loss": 1.0022,
+      "step": 271100
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.1116134245392106e-05,
+      "loss": 0.9839,
+      "step": 271200
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.110916841973279e-05,
+      "loss": 0.9932,
+      "step": 271300
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.1102202594073475e-05,
+      "loss": 0.9739,
+      "step": 271400
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.109523676841416e-05,
+      "loss": 1.0009,
+      "step": 271500
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.1088270942754844e-05,
+      "loss": 0.9738,
+      "step": 271600
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.1081305117095529e-05,
+      "loss": 0.9487,
+      "step": 271700
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.1074339291436214e-05,
+      "loss": 0.9752,
+      "step": 271800
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.1067373465776899e-05,
+      "loss": 0.9955,
+      "step": 271900
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.1060407640117585e-05,
+      "loss": 0.9662,
+      "step": 272000
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.105344181445827e-05,
+      "loss": 0.9841,
+      "step": 272100
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.1046475988798953e-05,
+      "loss": 0.9586,
+      "step": 272200
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.1039510163139637e-05,
+      "loss": 0.9878,
+      "step": 272300
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.1032544337480322e-05,
+      "loss": 0.9602,
+      "step": 272400
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.1025578511821007e-05,
+      "loss": 0.9938,
+      "step": 272500
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.1018612686161691e-05,
+      "loss": 0.9777,
+      "step": 272600
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.1011646860502376e-05,
+      "loss": 0.9831,
+      "step": 272700
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.100468103484306e-05,
+      "loss": 0.9644,
+      "step": 272800
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.0997715209183745e-05,
+      "loss": 0.9664,
+      "step": 272900
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.099074938352443e-05,
+      "loss": 0.9777,
+      "step": 273000
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.0983783557865115e-05,
+      "loss": 0.9728,
+      "step": 273100
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.09768177322058e-05,
+      "loss": 0.9791,
+      "step": 273200
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.0969921564803078e-05,
+      "loss": 0.9805,
+      "step": 273300
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.0962955739143762e-05,
+      "loss": 0.9774,
+      "step": 273400
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.0955989913484445e-05,
+      "loss": 0.9813,
+      "step": 273500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.094902408782513e-05,
+      "loss": 0.9358,
+      "step": 273600
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.0942058262165815e-05,
+      "loss": 1.0004,
+      "step": 273700
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.09350924365065e-05,
+      "loss": 0.97,
+      "step": 273800
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.0928126610847184e-05,
+      "loss": 0.9578,
+      "step": 273900
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.0921160785187869e-05,
+      "loss": 0.994,
+      "step": 274000
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.0914194959528553e-05,
+      "loss": 0.9845,
+      "step": 274100
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.0907229133869238e-05,
+      "loss": 0.9934,
+      "step": 274200
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.0900263308209923e-05,
+      "loss": 0.9909,
+      "step": 274300
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.0893297482550607e-05,
+      "loss": 0.9895,
+      "step": 274400
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.0886331656891292e-05,
+      "loss": 0.9869,
+      "step": 274500
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.0879365831231975e-05,
+      "loss": 0.9759,
+      "step": 274600
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.087240000557266e-05,
+      "loss": 1.013,
+      "step": 274700
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.0865434179913344e-05,
+      "loss": 0.9553,
+      "step": 274800
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.0858468354254029e-05,
+      "loss": 0.9715,
+      "step": 274900
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.0851502528594714e-05,
+      "loss": 0.9699,
+      "step": 275000
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.0844536702935398e-05,
+      "loss": 1.0056,
+      "step": 275100
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.0837640535532677e-05,
+      "loss": 1.0038,
+      "step": 275200
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.0830674709873361e-05,
+      "loss": 0.9863,
+      "step": 275300
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.0823708884214046e-05,
+      "loss": 0.9697,
+      "step": 275400
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.081674305855473e-05,
+      "loss": 0.9998,
+      "step": 275500
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.0809777232895415e-05,
+      "loss": 1.0124,
+      "step": 275600
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.08028114072361e-05,
+      "loss": 0.979,
+      "step": 275700
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.0795845581576785e-05,
+      "loss": 1.0042,
+      "step": 275800
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.0788879755917468e-05,
+      "loss": 0.9549,
+      "step": 275900
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.0781913930258154e-05,
+      "loss": 0.9615,
+      "step": 276000
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.0774948104598839e-05,
+      "loss": 0.9758,
+      "step": 276100
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.0767982278939523e-05,
+      "loss": 0.9454,
+      "step": 276200
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.0761016453280208e-05,
+      "loss": 0.9954,
+      "step": 276300
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.0754050627620893e-05,
+      "loss": 0.9784,
+      "step": 276400
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.0747084801961577e-05,
+      "loss": 0.9846,
+      "step": 276500
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.0740118976302262e-05,
+      "loss": 0.9897,
+      "step": 276600
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.0733153150642947e-05,
+      "loss": 1.0244,
+      "step": 276700
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.0726187324983631e-05,
+      "loss": 0.9911,
+      "step": 276800
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.0719221499324316e-05,
+      "loss": 1.0317,
+      "step": 276900
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.0712255673665e-05,
+      "loss": 0.9978,
+      "step": 277000
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.0705289848005684e-05,
+      "loss": 1.0002,
+      "step": 277100
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.0698324022346369e-05,
+      "loss": 0.9814,
+      "step": 277200
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.0691358196687053e-05,
+      "loss": 0.9785,
+      "step": 277300
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.0684392371027738e-05,
+      "loss": 0.9724,
+      "step": 277400
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.0677426545368423e-05,
+      "loss": 0.9989,
+      "step": 277500
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.0670460719709107e-05,
+      "loss": 0.961,
+      "step": 277600
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.0663494894049792e-05,
+      "loss": 0.96,
+      "step": 277700
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.0656529068390477e-05,
+      "loss": 0.9835,
+      "step": 277800
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.0649563242731161e-05,
+      "loss": 0.9815,
+      "step": 277900
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.0642597417071846e-05,
+      "loss": 0.9932,
+      "step": 278000
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.0635631591412529e-05,
+      "loss": 0.9926,
+      "step": 278100
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.0628665765753214e-05,
+      "loss": 0.9705,
+      "step": 278200
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.0621699940093898e-05,
+      "loss": 1.0001,
+      "step": 278300
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.0614734114434583e-05,
+      "loss": 0.9889,
+      "step": 278400
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.060776828877527e-05,
+      "loss": 0.9962,
+      "step": 278500
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.0600802463115954e-05,
+      "loss": 0.9582,
+      "step": 278600
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.0593836637456639e-05,
+      "loss": 0.9944,
+      "step": 278700
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.0586870811797323e-05,
+      "loss": 0.9763,
+      "step": 278800
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.0579904986138008e-05,
+      "loss": 0.9497,
+      "step": 278900
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.0572939160478693e-05,
+      "loss": 0.9902,
+      "step": 279000
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.0565973334819378e-05,
+      "loss": 0.9809,
+      "step": 279100
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.0559077167416654e-05,
+      "loss": 0.9747,
+      "step": 279200
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.055211134175734e-05,
+      "loss": 0.9805,
+      "step": 279300
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.0545145516098023e-05,
+      "loss": 1.0024,
+      "step": 279400
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.0538179690438708e-05,
+      "loss": 0.9853,
+      "step": 279500
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.0531213864779393e-05,
+      "loss": 0.9734,
+      "step": 279600
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.0524317697376669e-05,
+      "loss": 0.9694,
+      "step": 279700
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.0517351871717354e-05,
+      "loss": 1.0061,
+      "step": 279800
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.0510386046058038e-05,
+      "loss": 0.9687,
+      "step": 279900
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.0503420220398723e-05,
+      "loss": 0.9794,
+      "step": 280000
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.049645439473941e-05,
+      "loss": 0.9878,
+      "step": 280100
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.0489488569080094e-05,
+      "loss": 0.9465,
+      "step": 280200
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0482522743420779e-05,
+      "loss": 1.0336,
+      "step": 280300
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0475556917761464e-05,
+      "loss": 0.9868,
+      "step": 280400
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0468591092102148e-05,
+      "loss": 1.0007,
+      "step": 280500
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0461625266442833e-05,
+      "loss": 0.9596,
+      "step": 280600
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0454659440783516e-05,
+      "loss": 1.0274,
+      "step": 280700
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.04476936151242e-05,
+      "loss": 0.9902,
+      "step": 280800
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0440727789464885e-05,
+      "loss": 0.9704,
+      "step": 280900
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.043376196380557e-05,
+      "loss": 0.9813,
+      "step": 281000
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0426796138146255e-05,
+      "loss": 1.0052,
+      "step": 281100
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.041983031248694e-05,
+      "loss": 0.9405,
+      "step": 281200
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0412864486827624e-05,
+      "loss": 0.9918,
+      "step": 281300
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0405898661168309e-05,
+      "loss": 0.9681,
+      "step": 281400
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0398932835508993e-05,
+      "loss": 0.9744,
+      "step": 281500
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0391967009849678e-05,
+      "loss": 0.974,
+      "step": 281600
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0385001184190363e-05,
+      "loss": 1.0004,
+      "step": 281700
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0378035358531046e-05,
+      "loss": 0.9524,
+      "step": 281800
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.037106953287173e-05,
+      "loss": 0.9796,
+      "step": 281900
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0364103707212415e-05,
+      "loss": 1.0044,
+      "step": 282000
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.03571378815531e-05,
+      "loss": 0.9573,
+      "step": 282100
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0350172055893784e-05,
+      "loss": 1.0033,
+      "step": 282200
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.034320623023447e-05,
+      "loss": 1.0084,
+      "step": 282300
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0336240404575154e-05,
+      "loss": 0.9516,
+      "step": 282400
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0329274578915839e-05,
+      "loss": 0.9972,
+      "step": 282500
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0322308753256525e-05,
+      "loss": 0.9877,
+      "step": 282600
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.031534292759721e-05,
+      "loss": 1.0119,
+      "step": 282700
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0308377101937894e-05,
+      "loss": 0.9725,
+      "step": 282800
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0301411276278579e-05,
+      "loss": 0.9483,
+      "step": 282900
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0294445450619262e-05,
+      "loss": 1.0138,
+      "step": 283000
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0287479624959947e-05,
+      "loss": 0.9888,
+      "step": 283100
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.0280513799300631e-05,
+      "loss": 0.9672,
+      "step": 283200
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.0273547973641316e-05,
+      "loss": 0.9859,
+      "step": 283300
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.0266582147982e-05,
+      "loss": 0.9606,
+      "step": 283400
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.0259616322322685e-05,
+      "loss": 1.0003,
+      "step": 283500
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.025265049666337e-05,
+      "loss": 0.9798,
+      "step": 283600
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.0245684671004055e-05,
+      "loss": 0.9869,
+      "step": 283700
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.023871884534474e-05,
+      "loss": 1.0049,
+      "step": 283800
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0231753019685424e-05,
+      "loss": 0.9581,
+      "step": 283900
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0224787194026109e-05,
+      "loss": 1.0021,
+      "step": 284000
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0217821368366792e-05,
+      "loss": 0.9775,
+      "step": 284100
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0210855542707476e-05,
+      "loss": 0.9659,
+      "step": 284200
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0203889717048161e-05,
+      "loss": 0.9469,
+      "step": 284300
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.019699354964544e-05,
+      "loss": 0.9738,
+      "step": 284400
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0190027723986124e-05,
+      "loss": 0.9421,
+      "step": 284500
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0183061898326809e-05,
+      "loss": 0.989,
+      "step": 284600
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0176096072667493e-05,
+      "loss": 0.9879,
+      "step": 284700
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0169199905264771e-05,
+      "loss": 0.9911,
+      "step": 284800
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0162234079605456e-05,
+      "loss": 0.9613,
+      "step": 284900
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.015526825394614e-05,
+      "loss": 0.9824,
+      "step": 285000
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0148302428286825e-05,
+      "loss": 0.9766,
+      "step": 285100
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.014133660262751e-05,
+      "loss": 0.9803,
+      "step": 285200
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0134370776968195e-05,
+      "loss": 1.0073,
+      "step": 285300
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.012740495130888e-05,
+      "loss": 0.9942,
+      "step": 285400
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0120439125649564e-05,
+      "loss": 0.9659,
+      "step": 285500
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0113473299990247e-05,
+      "loss": 0.9612,
+      "step": 285600
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0106507474330932e-05,
+      "loss": 0.969,
+      "step": 285700
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0099541648671617e-05,
+      "loss": 0.9585,
+      "step": 285800
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0092575823012301e-05,
+      "loss": 0.9946,
+      "step": 285900
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0085609997352986e-05,
+      "loss": 0.9986,
+      "step": 286000
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.007864417169367e-05,
+      "loss": 1.0097,
+      "step": 286100
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0071678346034355e-05,
+      "loss": 1.0117,
+      "step": 286200
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.006471252037504e-05,
+      "loss": 0.973,
+      "step": 286300
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0057746694715725e-05,
+      "loss": 0.9853,
+      "step": 286400
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.005078086905641e-05,
+      "loss": 0.9752,
+      "step": 286500
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0043815043397094e-05,
+      "loss": 0.9869,
+      "step": 286600
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0036849217737779e-05,
+      "loss": 0.9808,
+      "step": 286700
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.0029883392078463e-05,
+      "loss": 0.9659,
+      "step": 286800
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.0022917566419148e-05,
+      "loss": 0.9788,
+      "step": 286900
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.0015951740759833e-05,
+      "loss": 0.9915,
+      "step": 287000
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.0008985915100517e-05,
+      "loss": 0.9798,
+      "step": 287100
+    },
+    {
+      "epoch": 4.0,
+      "eval_gen_len": 20.0,
+      "eval_loss": 1.1669589281082153,
+      "eval_rouge1": 12.4306,
+      "eval_rouge2": 3.7329,
+      "eval_rougeL": 11.9497,
+      "eval_rougeLsum": 12.0617,
+      "eval_runtime": 1509.6015,
+      "eval_samples_per_second": 8.855,
+      "eval_steps_per_second": 2.214,
+      "step": 287116
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.0002020089441202e-05,
+      "loss": 0.8798,
+      "step": 287200
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 9.995054263781887e-06,
+      "loss": 0.887,
+      "step": 287300
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 9.988088438122572e-06,
+      "loss": 0.9167,
+      "step": 287400
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 9.981122612463256e-06,
+      "loss": 0.8906,
+      "step": 287500
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 9.974156786803941e-06,
+      "loss": 0.8845,
+      "step": 287600
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 9.967190961144626e-06,
+      "loss": 0.8998,
+      "step": 287700
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 9.960225135485309e-06,
+      "loss": 0.9037,
+      "step": 287800
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 9.953259309825993e-06,
+      "loss": 0.86,
+      "step": 287900
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 9.946293484166678e-06,
+      "loss": 0.9148,
+      "step": 288000
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 9.939327658507363e-06,
+      "loss": 0.9055,
+      "step": 288100
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 9.932361832848047e-06,
+      "loss": 0.8924,
+      "step": 288200
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 9.925396007188732e-06,
+      "loss": 0.8662,
+      "step": 288300
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 9.918430181529417e-06,
+      "loss": 0.8848,
+      "step": 288400
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 9.911464355870101e-06,
+      "loss": 0.8896,
+      "step": 288500
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 9.90456818846738e-06,
+      "loss": 0.8681,
+      "step": 288600
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 9.897602362808064e-06,
+      "loss": 0.9143,
+      "step": 288700
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 9.890636537148749e-06,
+      "loss": 0.8555,
+      "step": 288800
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 9.883670711489434e-06,
+      "loss": 0.898,
+      "step": 288900
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.876704885830118e-06,
+      "loss": 0.8996,
+      "step": 289000
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.869739060170801e-06,
+      "loss": 0.8892,
+      "step": 289100
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.862773234511486e-06,
+      "loss": 0.8774,
+      "step": 289200
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.85580740885217e-06,
+      "loss": 0.8788,
+      "step": 289300
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.848841583192855e-06,
+      "loss": 0.8799,
+      "step": 289400
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.84187575753354e-06,
+      "loss": 0.888,
+      "step": 289500
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.834909931874225e-06,
+      "loss": 0.8484,
+      "step": 289600
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 9.82794410621491e-06,
+      "loss": 0.9076,
+      "step": 289700
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 9.820978280555594e-06,
+      "loss": 0.879,
+      "step": 289800
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 9.81401245489628e-06,
+      "loss": 0.8713,
+      "step": 289900
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 9.807046629236965e-06,
+      "loss": 0.8748,
+      "step": 290000
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 9.80008080357765e-06,
+      "loss": 0.8888,
+      "step": 290100
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 9.793114977918333e-06,
+      "loss": 0.8755,
+      "step": 290200
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 9.786149152259017e-06,
+      "loss": 0.8682,
+      "step": 290300
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 9.779183326599702e-06,
+      "loss": 0.8733,
+      "step": 290400
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 9.772217500940387e-06,
+      "loss": 0.8791,
+      "step": 290500
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 9.765251675281071e-06,
+      "loss": 0.8938,
+      "step": 290600
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 9.758285849621756e-06,
+      "loss": 0.9049,
+      "step": 290700
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 9.75132002396244e-06,
+      "loss": 0.8843,
+      "step": 290800
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 9.744354198303125e-06,
+      "loss": 0.9157,
+      "step": 290900
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 9.73738837264381e-06,
+      "loss": 0.9201,
+      "step": 291000
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 9.730422546984495e-06,
+      "loss": 0.8714,
+      "step": 291100
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 9.72345672132518e-06,
+      "loss": 0.8757,
+      "step": 291200
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 9.716490895665864e-06,
+      "loss": 0.8866,
+      "step": 291300
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 9.709525070006547e-06,
+      "loss": 0.9118,
+      "step": 291400
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 9.702559244347232e-06,
+      "loss": 0.9034,
+      "step": 291500
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 9.695593418687917e-06,
+      "loss": 0.8982,
+      "step": 291600
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 9.688627593028601e-06,
+      "loss": 0.9282,
+      "step": 291700
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 9.681661767369286e-06,
+      "loss": 0.8808,
+      "step": 291800
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 9.67469594170997e-06,
+      "loss": 0.8979,
+      "step": 291900
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 9.667799774307249e-06,
+      "loss": 0.9066,
+      "step": 292000
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 9.660833948647933e-06,
+      "loss": 0.9289,
+      "step": 292100
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 9.653868122988618e-06,
+      "loss": 0.9009,
+      "step": 292200
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 9.646902297329303e-06,
+      "loss": 0.9069,
+      "step": 292300
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 9.639936471669987e-06,
+      "loss": 0.9099,
+      "step": 292400
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 9.632970646010672e-06,
+      "loss": 0.8847,
+      "step": 292500
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 9.626004820351357e-06,
+      "loss": 0.8912,
+      "step": 292600
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 9.61903899469204e-06,
+      "loss": 0.891,
+      "step": 292700
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 9.612073169032725e-06,
+      "loss": 0.9028,
+      "step": 292800
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 9.60510734337341e-06,
+      "loss": 0.9037,
+      "step": 292900
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 9.598141517714094e-06,
+      "loss": 0.9124,
+      "step": 293000
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 9.591175692054779e-06,
+      "loss": 0.8872,
+      "step": 293100
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 9.584209866395465e-06,
+      "loss": 0.8969,
+      "step": 293200
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 9.57724404073615e-06,
+      "loss": 0.9182,
+      "step": 293300
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 9.570278215076834e-06,
+      "loss": 0.9193,
+      "step": 293400
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 9.563312389417519e-06,
+      "loss": 0.8957,
+      "step": 293500
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 9.556346563758204e-06,
+      "loss": 0.8839,
+      "step": 293600
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 9.549380738098888e-06,
+      "loss": 0.903,
+      "step": 293700
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 9.542414912439571e-06,
+      "loss": 0.8874,
+      "step": 293800
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 9.535449086780256e-06,
+      "loss": 0.9027,
+      "step": 293900
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 9.52848326112094e-06,
+      "loss": 0.8889,
+      "step": 294000
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 9.521587093718219e-06,
+      "loss": 0.8773,
+      "step": 294100
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 9.514621268058904e-06,
+      "loss": 0.91,
+      "step": 294200
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 9.507655442399588e-06,
+      "loss": 0.8912,
+      "step": 294300
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 9.500689616740273e-06,
+      "loss": 0.8588,
+      "step": 294400
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 9.493723791080958e-06,
+      "loss": 0.8836,
+      "step": 294500
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 9.486757965421642e-06,
+      "loss": 0.8813,
+      "step": 294600
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 9.479792139762327e-06,
+      "loss": 0.8857,
+      "step": 294700
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 9.472826314103012e-06,
+      "loss": 0.9061,
+      "step": 294800
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 9.465860488443696e-06,
+      "loss": 0.9027,
+      "step": 294900
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 9.458964321040974e-06,
+      "loss": 0.9203,
+      "step": 295000
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 9.451998495381659e-06,
+      "loss": 0.8871,
+      "step": 295100
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 9.445032669722344e-06,
+      "loss": 0.9012,
+      "step": 295200
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 9.438066844063027e-06,
+      "loss": 0.9211,
+      "step": 295300
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 9.431101018403711e-06,
+      "loss": 0.8783,
+      "step": 295400
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 9.424135192744396e-06,
+      "loss": 0.8633,
+      "step": 295500
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 9.41716936708508e-06,
+      "loss": 0.9149,
+      "step": 295600
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 9.410203541425766e-06,
+      "loss": 0.8882,
+      "step": 295700
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 9.40323771576645e-06,
+      "loss": 0.8843,
+      "step": 295800
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 9.396271890107135e-06,
+      "loss": 0.8888,
+      "step": 295900
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 9.38930606444782e-06,
+      "loss": 0.8865,
+      "step": 296000
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 9.382340238788504e-06,
+      "loss": 0.8992,
+      "step": 296100
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 9.375374413129189e-06,
+      "loss": 0.8829,
+      "step": 296200
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 9.368408587469874e-06,
+      "loss": 0.9391,
+      "step": 296300
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 9.361442761810557e-06,
+      "loss": 0.8892,
+      "step": 296400
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 9.354476936151241e-06,
+      "loss": 0.8723,
+      "step": 296500
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 9.347511110491926e-06,
+      "loss": 0.905,
+      "step": 296600
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 9.34054528483261e-06,
+      "loss": 0.8861,
+      "step": 296700
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 9.333579459173295e-06,
+      "loss": 0.875,
+      "step": 296800
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 9.32661363351398e-06,
+      "loss": 0.8967,
+      "step": 296900
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 9.319647807854665e-06,
+      "loss": 0.9027,
+      "step": 297000
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 9.31268198219535e-06,
+      "loss": 0.885,
+      "step": 297100
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 9.305716156536036e-06,
+      "loss": 0.9072,
+      "step": 297200
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 9.29875033087672e-06,
+      "loss": 0.9043,
+      "step": 297300
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 9.291784505217405e-06,
+      "loss": 0.864,
+      "step": 297400
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 9.284818679558088e-06,
+      "loss": 0.8637,
+      "step": 297500
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 9.277852853898773e-06,
+      "loss": 0.887,
+      "step": 297600
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 9.270887028239457e-06,
+      "loss": 0.9015,
+      "step": 297700
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 9.263921202580142e-06,
+      "loss": 0.9068,
+      "step": 297800
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 9.256955376920827e-06,
+      "loss": 0.8656,
+      "step": 297900
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 9.249989551261512e-06,
+      "loss": 0.9118,
+      "step": 298000
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 9.243023725602196e-06,
+      "loss": 0.8942,
+      "step": 298100
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 9.236127558199474e-06,
+      "loss": 0.8867,
+      "step": 298200
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 9.229161732540159e-06,
+      "loss": 0.8914,
+      "step": 298300
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 9.222195906880844e-06,
+      "loss": 0.8917,
+      "step": 298400
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 9.215230081221528e-06,
+      "loss": 0.8779,
+      "step": 298500
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 9.208264255562213e-06,
+      "loss": 0.9147,
+      "step": 298600
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 9.201298429902898e-06,
+      "loss": 0.882,
+      "step": 298700
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 9.19433260424358e-06,
+      "loss": 0.8678,
+      "step": 298800
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 9.187366778584265e-06,
+      "loss": 0.8821,
+      "step": 298900
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 9.18040095292495e-06,
+      "loss": 0.8871,
+      "step": 299000
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 9.173435127265635e-06,
+      "loss": 0.9118,
+      "step": 299100
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 9.16646930160632e-06,
+      "loss": 0.9057,
+      "step": 299200
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 9.159503475947004e-06,
+      "loss": 0.8713,
+      "step": 299300
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 9.152537650287689e-06,
+      "loss": 0.8827,
+      "step": 299400
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 9.145571824628374e-06,
+      "loss": 0.8992,
+      "step": 299500
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 9.138605998969058e-06,
+      "loss": 0.9391,
+      "step": 299600
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 9.131640173309743e-06,
+      "loss": 0.9138,
+      "step": 299700
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 9.124674347650428e-06,
+      "loss": 0.8712,
+      "step": 299800
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 9.11770852199111e-06,
+      "loss": 0.9157,
+      "step": 299900
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 9.110742696331795e-06,
+      "loss": 0.8838,
+      "step": 300000
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 9.10377687067248e-06,
+      "loss": 0.9131,
+      "step": 300100
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 9.096811045013165e-06,
+      "loss": 0.9078,
+      "step": 300200
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 9.08984521935385e-06,
+      "loss": 0.9053,
+      "step": 300300
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 9.082879393694534e-06,
+      "loss": 0.9065,
+      "step": 300400
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 9.07591356803522e-06,
+      "loss": 0.9008,
+      "step": 300500
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 9.068947742375905e-06,
+      "loss": 0.8981,
+      "step": 300600
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 9.06198191671659e-06,
+      "loss": 0.8974,
+      "step": 300700
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 9.055016091057274e-06,
+      "loss": 0.9151,
+      "step": 300800
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 9.048050265397959e-06,
+      "loss": 0.9115,
+      "step": 300900
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 9.041084439738642e-06,
+      "loss": 0.9199,
+      "step": 301000
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 9.034118614079327e-06,
+      "loss": 0.8923,
+      "step": 301100
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 9.027152788420011e-06,
+      "loss": 0.8718,
+      "step": 301200
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 9.020186962760696e-06,
+      "loss": 0.8895,
+      "step": 301300
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 9.01322113710138e-06,
+      "loss": 0.9087,
+      "step": 301400
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 9.006255311442065e-06,
+      "loss": 0.8972,
+      "step": 301500
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 8.99928948578275e-06,
+      "loss": 0.9067,
+      "step": 301600
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 8.992323660123435e-06,
+      "loss": 0.9087,
+      "step": 301700
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 8.98535783446412e-06,
+      "loss": 0.9044,
+      "step": 301800
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 8.978392008804804e-06,
+      "loss": 0.8673,
+      "step": 301900
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 8.971426183145489e-06,
+      "loss": 0.8897,
+      "step": 302000
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 8.964460357486174e-06,
+      "loss": 0.8733,
+      "step": 302100
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 8.957494531826857e-06,
+      "loss": 0.8798,
+      "step": 302200
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 8.950598364424135e-06,
+      "loss": 0.8735,
+      "step": 302300
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 8.94363253876482e-06,
+      "loss": 0.8819,
+      "step": 302400
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 8.936666713105504e-06,
+      "loss": 0.9019,
+      "step": 302500
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 8.929700887446189e-06,
+      "loss": 0.8547,
+      "step": 302600
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 8.922735061786873e-06,
+      "loss": 0.8871,
+      "step": 302700
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 8.915769236127558e-06,
+      "loss": 0.8906,
+      "step": 302800
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 8.908803410468243e-06,
+      "loss": 0.9111,
+      "step": 302900
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 8.901837584808927e-06,
+      "loss": 0.8933,
+      "step": 303000
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 8.894871759149612e-06,
+      "loss": 0.8749,
+      "step": 303100
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 8.887905933490297e-06,
+      "loss": 0.9143,
+      "step": 303200
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 8.880940107830982e-06,
+      "loss": 0.8993,
+      "step": 303300
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 8.873974282171666e-06,
+      "loss": 0.9112,
+      "step": 303400
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 8.86700845651235e-06,
+      "loss": 0.8785,
+      "step": 303500
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 8.860042630853034e-06,
+      "loss": 0.8886,
+      "step": 303600
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 8.85307680519372e-06,
+      "loss": 0.938,
+      "step": 303700
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 8.846110979534405e-06,
+      "loss": 0.854,
+      "step": 303800
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 8.83914515387509e-06,
+      "loss": 0.926,
+      "step": 303900
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 8.832179328215774e-06,
+      "loss": 0.9124,
+      "step": 304000
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 8.825213502556459e-06,
+      "loss": 0.8968,
+      "step": 304100
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 8.818247676897144e-06,
+      "loss": 0.9058,
+      "step": 304200
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 8.811281851237828e-06,
+      "loss": 0.908,
+      "step": 304300
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 8.804385683835105e-06,
+      "loss": 0.9382,
+      "step": 304400
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 8.79741985817579e-06,
+      "loss": 0.8834,
+      "step": 304500
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 8.790454032516476e-06,
+      "loss": 0.8964,
+      "step": 304600
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 8.78348820685716e-06,
+      "loss": 0.9125,
+      "step": 304700
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 8.776522381197844e-06,
+      "loss": 0.8903,
+      "step": 304800
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 8.769556555538528e-06,
+      "loss": 0.8758,
+      "step": 304900
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 8.762590729879213e-06,
+      "loss": 0.8857,
+      "step": 305000
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 8.755624904219898e-06,
+      "loss": 0.9212,
+      "step": 305100
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 8.748728736817174e-06,
+      "loss": 0.8922,
+      "step": 305200
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 8.741762911157859e-06,
+      "loss": 0.9003,
+      "step": 305300
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 8.734797085498545e-06,
+      "loss": 0.9132,
+      "step": 305400
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 8.72783125983923e-06,
+      "loss": 0.8758,
+      "step": 305500
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 8.720865434179914e-06,
+      "loss": 0.8875,
+      "step": 305600
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 8.713899608520599e-06,
+      "loss": 0.9164,
+      "step": 305700
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 8.706933782861284e-06,
+      "loss": 0.8996,
+      "step": 305800
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 8.699967957201968e-06,
+      "loss": 0.9045,
+      "step": 305900
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 8.693002131542653e-06,
+      "loss": 0.8863,
+      "step": 306000
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 8.686036305883336e-06,
+      "loss": 0.8951,
+      "step": 306100
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 8.67907048022402e-06,
+      "loss": 0.9016,
+      "step": 306200
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 8.672104654564706e-06,
+      "loss": 0.9075,
+      "step": 306300
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 8.66513882890539e-06,
+      "loss": 0.8794,
+      "step": 306400
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 8.658173003246075e-06,
+      "loss": 0.9076,
+      "step": 306500
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 8.65120717758676e-06,
+      "loss": 0.8812,
+      "step": 306600
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 8.644241351927444e-06,
+      "loss": 0.8843,
+      "step": 306700
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 8.637275526268129e-06,
+      "loss": 0.8714,
+      "step": 306800
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 8.630309700608814e-06,
+      "loss": 0.879,
+      "step": 306900
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 8.623343874949498e-06,
+      "loss": 0.8801,
+      "step": 307000
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 8.616378049290183e-06,
+      "loss": 0.8613,
+      "step": 307100
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 8.609412223630866e-06,
+      "loss": 0.887,
+      "step": 307200
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 8.60244639797155e-06,
+      "loss": 0.8832,
+      "step": 307300
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 8.595480572312235e-06,
+      "loss": 0.9144,
+      "step": 307400
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 8.58851474665292e-06,
+      "loss": 0.8777,
+      "step": 307500
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 8.581548920993605e-06,
+      "loss": 0.8765,
+      "step": 307600
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 8.57458309533429e-06,
+      "loss": 0.9076,
+      "step": 307700
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 8.567617269674976e-06,
+      "loss": 0.8719,
+      "step": 307800
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 8.56065144401566e-06,
+      "loss": 0.8705,
+      "step": 307900
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 8.553685618356345e-06,
+      "loss": 0.9109,
+      "step": 308000
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 8.54671979269703e-06,
+      "loss": 0.9082,
+      "step": 308100
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 8.539753967037715e-06,
+      "loss": 0.9211,
+      "step": 308200
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 8.532788141378398e-06,
+      "loss": 0.8623,
+      "step": 308300
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 8.525822315719082e-06,
+      "loss": 0.9063,
+      "step": 308400
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 8.518856490059767e-06,
+      "loss": 0.91,
+      "step": 308500
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 8.511890664400452e-06,
+      "loss": 0.9026,
+      "step": 308600
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 8.504924838741136e-06,
+      "loss": 0.9274,
+      "step": 308700
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 8.498028671338414e-06,
+      "loss": 0.8974,
+      "step": 308800
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 8.491062845679099e-06,
+      "loss": 0.914,
+      "step": 308900
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 8.484097020019784e-06,
+      "loss": 0.9032,
+      "step": 309000
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 8.477131194360468e-06,
+      "loss": 0.895,
+      "step": 309100
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 8.470165368701153e-06,
+      "loss": 0.9005,
+      "step": 309200
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 8.463199543041838e-06,
+      "loss": 0.9147,
+      "step": 309300
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 8.456233717382522e-06,
+      "loss": 0.913,
+      "step": 309400
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 8.449267891723207e-06,
+      "loss": 0.9146,
+      "step": 309500
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 8.44230206606389e-06,
+      "loss": 0.9002,
+      "step": 309600
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 8.435336240404575e-06,
+      "loss": 0.9239,
+      "step": 309700
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 8.42837041474526e-06,
+      "loss": 0.8964,
+      "step": 309800
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 8.421404589085944e-06,
+      "loss": 0.9108,
+      "step": 309900
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 8.414438763426629e-06,
+      "loss": 0.8997,
+      "step": 310000
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 8.407472937767314e-06,
+      "loss": 0.8904,
+      "step": 310100
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 8.400507112107998e-06,
+      "loss": 0.8925,
+      "step": 310200
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 8.393541286448683e-06,
+      "loss": 0.8845,
+      "step": 310300
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 8.386575460789368e-06,
+      "loss": 0.859,
+      "step": 310400
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 8.379609635130052e-06,
+      "loss": 0.882,
+      "step": 310500
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 8.372643809470737e-06,
+      "loss": 0.9013,
+      "step": 310600
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 8.36567798381142e-06,
+      "loss": 0.8792,
+      "step": 310700
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 8.358712158152105e-06,
+      "loss": 0.8754,
+      "step": 310800
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 8.35174633249279e-06,
+      "loss": 0.8944,
+      "step": 310900
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 8.344780506833474e-06,
+      "loss": 0.9278,
+      "step": 311000
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 8.33781468117416e-06,
+      "loss": 0.9196,
+      "step": 311100
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.330848855514845e-06,
+      "loss": 0.9093,
+      "step": 311200
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.32388302985553e-06,
+      "loss": 0.8672,
+      "step": 311300
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.316917204196214e-06,
+      "loss": 0.889,
+      "step": 311400
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.309951378536899e-06,
+      "loss": 0.9111,
+      "step": 311500
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.302985552877584e-06,
+      "loss": 0.9003,
+      "step": 311600
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.296019727218268e-06,
+      "loss": 0.9248,
+      "step": 311700
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.289053901558953e-06,
+      "loss": 0.9375,
+      "step": 311800
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.282088075899636e-06,
+      "loss": 0.8977,
+      "step": 311900
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.27512225024032e-06,
+      "loss": 0.9032,
+      "step": 312000
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.268156424581006e-06,
+      "loss": 0.8761,
+      "step": 312100
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.26119059892169e-06,
+      "loss": 0.9121,
+      "step": 312200
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.254224773262375e-06,
+      "loss": 0.9079,
+      "step": 312300
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.24725894760306e-06,
+      "loss": 0.8987,
+      "step": 312400
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.240293121943744e-06,
+      "loss": 0.8772,
+      "step": 312500
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.233327296284429e-06,
+      "loss": 0.8698,
+      "step": 312600
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.226361470625114e-06,
+      "loss": 0.8869,
+      "step": 312700
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.219395644965798e-06,
+      "loss": 0.8906,
+      "step": 312800
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.212429819306483e-06,
+      "loss": 0.9012,
+      "step": 312900
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.205463993647166e-06,
+      "loss": 0.9043,
+      "step": 313000
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.19849816798785e-06,
+      "loss": 0.8731,
+      "step": 313100
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.191532342328535e-06,
+      "loss": 0.8759,
+      "step": 313200
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.18456651666922e-06,
+      "loss": 0.8886,
+      "step": 313300
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.177600691009905e-06,
+      "loss": 0.9356,
+      "step": 313400
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.17063486535059e-06,
+      "loss": 0.8879,
+      "step": 313500
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.163669039691276e-06,
+      "loss": 0.8747,
+      "step": 313600
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.15670321403196e-06,
+      "loss": 0.8869,
+      "step": 313700
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.149737388372645e-06,
+      "loss": 0.8997,
+      "step": 313800
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.14277156271333e-06,
+      "loss": 0.8978,
+      "step": 313900
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.135805737054014e-06,
+      "loss": 0.8883,
+      "step": 314000
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.128909569651291e-06,
+      "loss": 0.8676,
+      "step": 314100
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.121943743991976e-06,
+      "loss": 0.9107,
+      "step": 314200
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.11497791833266e-06,
+      "loss": 0.8825,
+      "step": 314300
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.108012092673345e-06,
+      "loss": 0.9159,
+      "step": 314400
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.10104626701403e-06,
+      "loss": 0.8902,
+      "step": 314500
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.094080441354714e-06,
+      "loss": 0.913,
+      "step": 314600
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.087114615695399e-06,
+      "loss": 0.8953,
+      "step": 314700
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 8.080148790036084e-06,
+      "loss": 0.8857,
+      "step": 314800
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 8.073182964376768e-06,
+      "loss": 0.8929,
+      "step": 314900
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 8.066217138717453e-06,
+      "loss": 0.9126,
+      "step": 315000
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 8.059251313058138e-06,
+      "loss": 0.8801,
+      "step": 315100
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 8.052285487398822e-06,
+      "loss": 0.8607,
+      "step": 315200
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 8.045319661739507e-06,
+      "loss": 0.934,
+      "step": 315300
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 8.03835383608019e-06,
+      "loss": 0.9185,
+      "step": 315400
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 8.031388010420875e-06,
+      "loss": 0.8689,
+      "step": 315500
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 8.02442218476156e-06,
+      "loss": 0.8656,
+      "step": 315600
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 8.017456359102244e-06,
+      "loss": 0.9096,
+      "step": 315700
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 8.010490533442929e-06,
+      "loss": 0.8914,
+      "step": 315800
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 8.003524707783614e-06,
+      "loss": 0.9163,
+      "step": 315900
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.996558882124298e-06,
+      "loss": 0.9187,
+      "step": 316000
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.989593056464983e-06,
+      "loss": 0.8844,
+      "step": 316100
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.982696889062261e-06,
+      "loss": 0.8941,
+      "step": 316200
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.975731063402946e-06,
+      "loss": 0.9091,
+      "step": 316300
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.96876523774363e-06,
+      "loss": 0.9058,
+      "step": 316400
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.961799412084315e-06,
+      "loss": 0.902,
+      "step": 316500
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.954833586425e-06,
+      "loss": 0.9069,
+      "step": 316600
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.947867760765683e-06,
+      "loss": 0.9132,
+      "step": 316700
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.940901935106367e-06,
+      "loss": 0.9092,
+      "step": 316800
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.933936109447052e-06,
+      "loss": 0.8787,
+      "step": 316900
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.926970283787737e-06,
+      "loss": 0.8912,
+      "step": 317000
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.920004458128421e-06,
+      "loss": 0.8981,
+      "step": 317100
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.913038632469106e-06,
+      "loss": 0.8828,
+      "step": 317200
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.90607280680979e-06,
+      "loss": 0.8916,
+      "step": 317300
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.899106981150476e-06,
+      "loss": 0.8966,
+      "step": 317400
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.89214115549116e-06,
+      "loss": 0.8944,
+      "step": 317500
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.885175329831847e-06,
+      "loss": 0.9162,
+      "step": 317600
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 7.878209504172531e-06,
+      "loss": 0.8874,
+      "step": 317700
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 7.871243678513214e-06,
+      "loss": 0.8966,
+      "step": 317800
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 7.864277852853899e-06,
+      "loss": 0.9229,
+      "step": 317900
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 7.857312027194584e-06,
+      "loss": 0.9151,
+      "step": 318000
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 7.850346201535268e-06,
+      "loss": 0.9226,
+      "step": 318100
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 7.843380375875953e-06,
+      "loss": 0.9051,
+      "step": 318200
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 7.836414550216638e-06,
+      "loss": 0.9119,
+      "step": 318300
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 7.829448724557322e-06,
+      "loss": 0.8696,
+      "step": 318400
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 7.822482898898007e-06,
+      "loss": 0.877,
+      "step": 318500
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 7.815517073238692e-06,
+      "loss": 0.9048,
+      "step": 318600
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 7.808551247579376e-06,
+      "loss": 0.8981,
+      "step": 318700
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 7.801585421920061e-06,
+      "loss": 0.8949,
+      "step": 318800
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 7.794619596260744e-06,
+      "loss": 0.8869,
+      "step": 318900
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 7.787653770601429e-06,
+      "loss": 0.9065,
+      "step": 319000
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 7.780687944942113e-06,
+      "loss": 0.8938,
+      "step": 319100
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 7.773722119282798e-06,
+      "loss": 0.9054,
+      "step": 319200
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 7.766756293623483e-06,
+      "loss": 0.8662,
+      "step": 319300
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 7.759790467964167e-06,
+      "loss": 0.8957,
+      "step": 319400
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 7.752824642304852e-06,
+      "loss": 0.8956,
+      "step": 319500
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 7.745858816645537e-06,
+      "loss": 0.8782,
+      "step": 319600
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 7.738892990986222e-06,
+      "loss": 0.8943,
+      "step": 319700
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 7.731927165326906e-06,
+      "loss": 0.8972,
+      "step": 319800
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 7.724961339667591e-06,
+      "loss": 0.9111,
+      "step": 319900
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 7.717995514008276e-06,
+      "loss": 0.908,
+      "step": 320000
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 7.71102968834896e-06,
+      "loss": 0.9101,
+      "step": 320100
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 7.704133520946237e-06,
+      "loss": 0.8888,
+      "step": 320200
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 7.697167695286921e-06,
+      "loss": 0.9402,
+      "step": 320300
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 7.690201869627606e-06,
+      "loss": 0.9185,
+      "step": 320400
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 7.68323604396829e-06,
+      "loss": 0.9445,
+      "step": 320500
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 7.676339876565569e-06,
+      "loss": 0.862,
+      "step": 320600
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 7.669374050906254e-06,
+      "loss": 0.8918,
+      "step": 320700
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 7.662408225246938e-06,
+      "loss": 0.888,
+      "step": 320800
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 7.655442399587623e-06,
+      "loss": 0.8846,
+      "step": 320900
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 7.648476573928308e-06,
+      "loss": 0.8908,
+      "step": 321000
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 7.641510748268992e-06,
+      "loss": 0.8922,
+      "step": 321100
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 7.634544922609677e-06,
+      "loss": 0.8926,
+      "step": 321200
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.6275790969503625e-06,
+      "loss": 0.8947,
+      "step": 321300
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.620613271291047e-06,
+      "loss": 0.892,
+      "step": 321400
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.61364744563173e-06,
+      "loss": 0.8965,
+      "step": 321500
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.606681619972415e-06,
+      "loss": 0.9022,
+      "step": 321600
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.5997157943130996e-06,
+      "loss": 0.8852,
+      "step": 321700
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.592749968653784e-06,
+      "loss": 0.9176,
+      "step": 321800
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.585784142994469e-06,
+      "loss": 0.8962,
+      "step": 321900
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.578818317335154e-06,
+      "loss": 0.9282,
+      "step": 322000
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.571852491675838e-06,
+      "loss": 0.9027,
+      "step": 322100
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.564886666016523e-06,
+      "loss": 0.9116,
+      "step": 322200
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.5579208403572085e-06,
+      "loss": 0.8951,
+      "step": 322300
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.550955014697893e-06,
+      "loss": 0.9081,
+      "step": 322400
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.543989189038578e-06,
+      "loss": 0.9026,
+      "step": 322500
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.5370233633792625e-06,
+      "loss": 0.8898,
+      "step": 322600
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.53012719597654e-06,
+      "loss": 0.8865,
+      "step": 322700
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.523161370317223e-06,
+      "loss": 0.9049,
+      "step": 322800
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.5161955446579075e-06,
+      "loss": 0.9006,
+      "step": 322900
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.509229718998592e-06,
+      "loss": 0.8915,
+      "step": 323000
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.502263893339278e-06,
+      "loss": 0.8986,
+      "step": 323100
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.495298067679962e-06,
+      "loss": 0.8966,
+      "step": 323200
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.488332242020647e-06,
+      "loss": 0.8789,
+      "step": 323300
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.481366416361332e-06,
+      "loss": 0.9292,
+      "step": 323400
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.4744005907020164e-06,
+      "loss": 0.9006,
+      "step": 323500
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.467434765042701e-06,
+      "loss": 0.9103,
+      "step": 323600
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.460468939383385e-06,
+      "loss": 0.902,
+      "step": 323700
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.45350311372407e-06,
+      "loss": 0.904,
+      "step": 323800
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.446537288064754e-06,
+      "loss": 0.897,
+      "step": 323900
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.439571462405439e-06,
+      "loss": 0.9162,
+      "step": 324000
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 7.432605636746124e-06,
+      "loss": 0.9034,
+      "step": 324100
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 7.4256398110868075e-06,
+      "loss": 0.8902,
+      "step": 324200
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 7.418673985427493e-06,
+      "loss": 0.9144,
+      "step": 324300
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 7.411708159768178e-06,
+      "loss": 0.8973,
+      "step": 324400
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 7.4047423341088624e-06,
+      "loss": 0.8659,
+      "step": 324500
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 7.397776508449547e-06,
+      "loss": 0.9041,
+      "step": 324600
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 7.390810682790232e-06,
+      "loss": 0.9243,
+      "step": 324700
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 7.383844857130916e-06,
+      "loss": 0.9291,
+      "step": 324800
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 7.376948689728194e-06,
+      "loss": 0.9036,
+      "step": 324900
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 7.369982864068878e-06,
+      "loss": 0.9243,
+      "step": 325000
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 7.363017038409562e-06,
+      "loss": 0.8956,
+      "step": 325100
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 7.356051212750248e-06,
+      "loss": 0.8965,
+      "step": 325200
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 7.3490853870909325e-06,
+      "loss": 0.8957,
+      "step": 325300
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 7.342119561431617e-06,
+      "loss": 0.9157,
+      "step": 325400
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 7.335153735772301e-06,
+      "loss": 0.9129,
+      "step": 325500
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 7.328187910112986e-06,
+      "loss": 0.862,
+      "step": 325600
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 7.32122208445367e-06,
+      "loss": 0.9181,
+      "step": 325700
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 7.314256258794355e-06,
+      "loss": 0.9186,
+      "step": 325800
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 7.30729043313504e-06,
+      "loss": 0.9071,
+      "step": 325900
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 7.3003246074757244e-06,
+      "loss": 0.9064,
+      "step": 326000
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 7.293358781816408e-06,
+      "loss": 0.8919,
+      "step": 326100
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 7.286392956157093e-06,
+      "loss": 0.8977,
+      "step": 326200
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 7.279427130497778e-06,
+      "loss": 0.8911,
+      "step": 326300
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 7.272461304838463e-06,
+      "loss": 0.9122,
+      "step": 326400
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 7.265495479179148e-06,
+      "loss": 0.8897,
+      "step": 326500
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 7.2585296535198325e-06,
+      "loss": 0.8913,
+      "step": 326600
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 7.251563827860516e-06,
+      "loss": 0.9271,
+      "step": 326700
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 7.244598002201201e-06,
+      "loss": 0.8734,
+      "step": 326800
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 7.237632176541886e-06,
+      "loss": 0.9086,
+      "step": 326900
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 7.230736009139163e-06,
+      "loss": 0.8985,
+      "step": 327000
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 7.223770183479848e-06,
+      "loss": 0.8944,
+      "step": 327100
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 7.216804357820533e-06,
+      "loss": 0.8879,
+      "step": 327200
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 7.209838532161218e-06,
+      "loss": 0.8855,
+      "step": 327300
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 7.202872706501902e-06,
+      "loss": 0.9174,
+      "step": 327400
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 7.195906880842586e-06,
+      "loss": 0.9044,
+      "step": 327500
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 7.188941055183271e-06,
+      "loss": 0.9211,
+      "step": 327600
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 7.181975229523956e-06,
+      "loss": 0.8787,
+      "step": 327700
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 7.1750094038646405e-06,
+      "loss": 0.9017,
+      "step": 327800
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 7.168043578205325e-06,
+      "loss": 0.8862,
+      "step": 327900
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 7.161077752546009e-06,
+      "loss": 0.8903,
+      "step": 328000
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 7.154111926886694e-06,
+      "loss": 0.8577,
+      "step": 328100
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 7.147146101227378e-06,
+      "loss": 0.889,
+      "step": 328200
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 7.140180275568063e-06,
+      "loss": 0.8597,
+      "step": 328300
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 7.1332144499087486e-06,
+      "loss": 0.8797,
+      "step": 328400
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 7.126248624249432e-06,
+      "loss": 0.8777,
+      "step": 328500
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 7.119282798590117e-06,
+      "loss": 0.8944,
+      "step": 328600
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 7.112316972930802e-06,
+      "loss": 0.9213,
+      "step": 328700
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 7.1053511472714865e-06,
+      "loss": 0.9316,
+      "step": 328800
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 7.098385321612171e-06,
+      "loss": 0.9004,
+      "step": 328900
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 7.091419495952856e-06,
+      "loss": 0.9317,
+      "step": 329000
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 7.08445367029354e-06,
+      "loss": 0.8835,
+      "step": 329100
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 7.077487844634224e-06,
+      "loss": 0.9031,
+      "step": 329200
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 7.070522018974909e-06,
+      "loss": 0.8982,
+      "step": 329300
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 7.063556193315594e-06,
+      "loss": 0.9028,
+      "step": 329400
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 7.056590367656278e-06,
+      "loss": 0.8781,
+      "step": 329500
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 7.049624541996963e-06,
+      "loss": 0.9082,
+      "step": 329600
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 7.042658716337648e-06,
+      "loss": 0.9128,
+      "step": 329700
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 7.0356928906783324e-06,
+      "loss": 0.9191,
+      "step": 329800
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 7.028727065019017e-06,
+      "loss": 0.8662,
+      "step": 329900
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 7.021761239359702e-06,
+      "loss": 0.9145,
+      "step": 330000
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 7.0147954137003865e-06,
+      "loss": 0.9019,
+      "step": 330100
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 7.00782958804107e-06,
+      "loss": 0.9101,
+      "step": 330200
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 7.000863762381755e-06,
+      "loss": 0.9263,
+      "step": 330300
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 6.99389793672244e-06,
+      "loss": 0.9352,
+      "step": 330400
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 6.986932111063124e-06,
+      "loss": 0.9041,
+      "step": 330500
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 6.979966285403809e-06,
+      "loss": 0.9118,
+      "step": 330600
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 6.973000459744494e-06,
+      "loss": 0.8861,
+      "step": 330700
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 6.966034634085178e-06,
+      "loss": 0.9053,
+      "step": 330800
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 6.959068808425863e-06,
+      "loss": 0.8984,
+      "step": 330900
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 6.952102982766548e-06,
+      "loss": 0.8753,
+      "step": 331000
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 6.945206815363825e-06,
+      "loss": 0.8981,
+      "step": 331100
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 6.93824098970451e-06,
+      "loss": 0.9,
+      "step": 331200
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 6.9312751640451944e-06,
+      "loss": 0.887,
+      "step": 331300
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 6.924309338385879e-06,
+      "loss": 0.9234,
+      "step": 331400
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 6.917343512726563e-06,
+      "loss": 0.9033,
+      "step": 331500
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 6.910377687067248e-06,
+      "loss": 0.8558,
+      "step": 331600
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 6.903411861407933e-06,
+      "loss": 0.8931,
+      "step": 331700
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 6.896446035748618e-06,
+      "loss": 0.9008,
+      "step": 331800
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 6.8894802100893025e-06,
+      "loss": 0.8831,
+      "step": 331900
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 6.882514384429987e-06,
+      "loss": 0.8829,
+      "step": 332000
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 6.875548558770671e-06,
+      "loss": 0.9051,
+      "step": 332100
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 6.868582733111356e-06,
+      "loss": 0.905,
+      "step": 332200
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 6.86161690745204e-06,
+      "loss": 0.9167,
+      "step": 332300
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 6.854651081792725e-06,
+      "loss": 0.9018,
+      "step": 332400
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 6.84768525613341e-06,
+      "loss": 0.8984,
+      "step": 332500
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 6.840719430474094e-06,
+      "loss": 0.9024,
+      "step": 332600
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.833753604814778e-06,
+      "loss": 0.89,
+      "step": 332700
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.826787779155463e-06,
+      "loss": 0.9009,
+      "step": 332800
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.8198219534961485e-06,
+      "loss": 0.9188,
+      "step": 332900
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.812856127836833e-06,
+      "loss": 0.9017,
+      "step": 333000
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.8059599604341105e-06,
+      "loss": 0.9005,
+      "step": 333100
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.798994134774795e-06,
+      "loss": 0.9046,
+      "step": 333200
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.79202830911548e-06,
+      "loss": 0.9067,
+      "step": 333300
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.785062483456164e-06,
+      "loss": 0.9117,
+      "step": 333400
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.778096657796848e-06,
+      "loss": 0.9139,
+      "step": 333500
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.771130832137533e-06,
+      "loss": 0.9028,
+      "step": 333600
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.764165006478219e-06,
+      "loss": 0.8987,
+      "step": 333700
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.757268839075496e-06,
+      "loss": 0.8911,
+      "step": 333800
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.7503030134161805e-06,
+      "loss": 0.8636,
+      "step": 333900
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.743337187756865e-06,
+      "loss": 0.902,
+      "step": 334000
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.736371362097549e-06,
+      "loss": 0.8968,
+      "step": 334100
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.729405536438234e-06,
+      "loss": 0.8989,
+      "step": 334200
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.7224397107789184e-06,
+      "loss": 0.887,
+      "step": 334300
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.715473885119603e-06,
+      "loss": 0.886,
+      "step": 334400
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.708508059460289e-06,
+      "loss": 0.9102,
+      "step": 334500
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.701542233800973e-06,
+      "loss": 0.895,
+      "step": 334600
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.694576408141657e-06,
+      "loss": 0.8741,
+      "step": 334700
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.687610582482342e-06,
+      "loss": 0.9149,
+      "step": 334800
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 6.6806447568230265e-06,
+      "loss": 0.8864,
+      "step": 334900
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 6.673678931163711e-06,
+      "loss": 0.895,
+      "step": 335000
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 6.666713105504396e-06,
+      "loss": 0.902,
+      "step": 335100
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 6.65974727984508e-06,
+      "loss": 0.8989,
+      "step": 335200
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 6.652781454185764e-06,
+      "loss": 0.9056,
+      "step": 335300
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 6.645815628526449e-06,
+      "loss": 0.8864,
+      "step": 335400
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 6.638849802867134e-06,
+      "loss": 0.9214,
+      "step": 335500
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 6.6318839772078185e-06,
+      "loss": 0.8958,
+      "step": 335600
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 6.624918151548503e-06,
+      "loss": 0.9103,
+      "step": 335700
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 6.617952325889188e-06,
+      "loss": 0.8671,
+      "step": 335800
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 6.6109865002298725e-06,
+      "loss": 0.9166,
+      "step": 335900
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 6.604020674570557e-06,
+      "loss": 0.8945,
+      "step": 336000
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 6.597054848911242e-06,
+      "loss": 0.923,
+      "step": 336100
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 6.5900890232519266e-06,
+      "loss": 0.9177,
+      "step": 336200
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 6.583123197592611e-06,
+      "loss": 0.905,
+      "step": 336300
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 6.576157371933295e-06,
+      "loss": 0.9267,
+      "step": 336400
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 6.56919154627398e-06,
+      "loss": 0.8953,
+      "step": 336500
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 6.5622257206146644e-06,
+      "loss": 0.8962,
+      "step": 336600
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 6.555259894955349e-06,
+      "loss": 0.9365,
+      "step": 336700
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 6.548294069296034e-06,
+      "loss": 0.8649,
+      "step": 336800
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 6.541328243636718e-06,
+      "loss": 0.8944,
+      "step": 336900
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 6.534362417977403e-06,
+      "loss": 0.9301,
+      "step": 337000
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 6.527396592318088e-06,
+      "loss": 0.8917,
+      "step": 337100
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 6.520500424915365e-06,
+      "loss": 0.8989,
+      "step": 337200
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 6.51353459925605e-06,
+      "loss": 0.8881,
+      "step": 337300
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 6.5065687735967345e-06,
+      "loss": 0.9066,
+      "step": 337400
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 6.499602947937419e-06,
+      "loss": 0.8719,
+      "step": 337500
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 6.492637122278104e-06,
+      "loss": 0.9109,
+      "step": 337600
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 6.485671296618788e-06,
+      "loss": 0.8819,
+      "step": 337700
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 6.478705470959473e-06,
+      "loss": 0.9111,
+      "step": 337800
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 6.471739645300158e-06,
+      "loss": 0.9033,
+      "step": 337900
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 6.464773819640843e-06,
+      "loss": 0.9152,
+      "step": 338000
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 6.457807993981527e-06,
+      "loss": 0.8903,
+      "step": 338100
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 6.450842168322211e-06,
+      "loss": 0.8815,
+      "step": 338200
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 6.443876342662896e-06,
+      "loss": 0.9236,
+      "step": 338300
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 6.4369105170035805e-06,
+      "loss": 0.899,
+      "step": 338400
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 6.429944691344265e-06,
+      "loss": 0.8957,
+      "step": 338500
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 6.42297886568495e-06,
+      "loss": 0.8671,
+      "step": 338600
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 6.4160130400256345e-06,
+      "loss": 0.9103,
+      "step": 338700
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 6.409047214366318e-06,
+      "loss": 0.8624,
+      "step": 338800
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 6.402081388707003e-06,
+      "loss": 0.8942,
+      "step": 338900
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 6.395115563047689e-06,
+      "loss": 0.9036,
+      "step": 339000
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 6.388149737388373e-06,
+      "loss": 0.8815,
+      "step": 339100
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 6.381183911729058e-06,
+      "loss": 0.9228,
+      "step": 339200
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 6.374218086069742e-06,
+      "loss": 0.8905,
+      "step": 339300
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 6.3672522604104265e-06,
+      "loss": 0.8905,
+      "step": 339400
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 6.360356093007704e-06,
+      "loss": 0.871,
+      "step": 339500
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 6.3533902673483884e-06,
+      "loss": 0.8743,
+      "step": 339600
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 6.346424441689073e-06,
+      "loss": 0.8953,
+      "step": 339700
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 6.339458616029759e-06,
+      "loss": 0.9128,
+      "step": 339800
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 6.332492790370443e-06,
+      "loss": 0.9017,
+      "step": 339900
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 6.325526964711128e-06,
+      "loss": 0.8974,
+      "step": 340000
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 6.318561139051812e-06,
+      "loss": 0.9075,
+      "step": 340100
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 6.3115953133924965e-06,
+      "loss": 0.9063,
+      "step": 340200
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 6.304629487733181e-06,
+      "loss": 0.9167,
+      "step": 340300
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 6.297663662073866e-06,
+      "loss": 0.873,
+      "step": 340400
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 6.290697836414551e-06,
+      "loss": 0.8829,
+      "step": 340500
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.283732010755234e-06,
+      "loss": 0.8914,
+      "step": 340600
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.276766185095919e-06,
+      "loss": 0.9371,
+      "step": 340700
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.269800359436604e-06,
+      "loss": 0.8828,
+      "step": 340800
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.2628345337772885e-06,
+      "loss": 0.895,
+      "step": 340900
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.255868708117973e-06,
+      "loss": 0.8914,
+      "step": 341000
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.248972540715251e-06,
+      "loss": 0.9313,
+      "step": 341100
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.242006715055936e-06,
+      "loss": 0.879,
+      "step": 341200
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.235040889396621e-06,
+      "loss": 0.8908,
+      "step": 341300
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 6.2280750637373045e-06,
+      "loss": 0.9114,
+      "step": 341400
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 6.221109238077989e-06,
+      "loss": 0.9086,
+      "step": 341500
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 6.214143412418674e-06,
+      "loss": 0.9252,
+      "step": 341600
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 6.2071775867593585e-06,
+      "loss": 0.9261,
+      "step": 341700
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 6.200211761100043e-06,
+      "loss": 0.8984,
+      "step": 341800
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 6.193245935440728e-06,
+      "loss": 0.8705,
+      "step": 341900
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 6.186280109781413e-06,
+      "loss": 0.8985,
+      "step": 342000
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 6.179314284122097e-06,
+      "loss": 0.8767,
+      "step": 342100
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 6.172348458462782e-06,
+      "loss": 0.8924,
+      "step": 342200
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 6.165382632803467e-06,
+      "loss": 0.9175,
+      "step": 342300
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 6.158416807144151e-06,
+      "loss": 0.9246,
+      "step": 342400
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 6.151450981484835e-06,
+      "loss": 0.8804,
+      "step": 342500
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 6.14448515582552e-06,
+      "loss": 0.8869,
+      "step": 342600
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 6.1375193301662045e-06,
+      "loss": 0.919,
+      "step": 342700
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 6.130553504506889e-06,
+      "loss": 0.9046,
+      "step": 342800
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 6.123587678847574e-06,
+      "loss": 0.8937,
+      "step": 342900
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 6.1166218531882586e-06,
+      "loss": 0.8938,
+      "step": 343000
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 6.109656027528943e-06,
+      "loss": 0.903,
+      "step": 343100
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 6.102690201869628e-06,
+      "loss": 0.8774,
+      "step": 343200
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 6.095724376210313e-06,
+      "loss": 0.8893,
+      "step": 343300
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 6.088758550550997e-06,
+      "loss": 0.8893,
+      "step": 343400
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 6.081792724891682e-06,
+      "loss": 0.8624,
+      "step": 343500
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 6.074826899232366e-06,
+      "loss": 0.8782,
+      "step": 343600
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 6.0678610735730505e-06,
+      "loss": 0.8608,
+      "step": 343700
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 6.060895247913735e-06,
+      "loss": 0.8896,
+      "step": 343800
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 6.05392942225442e-06,
+      "loss": 0.9133,
+      "step": 343900
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 6.0469635965951045e-06,
+      "loss": 0.9189,
+      "step": 344000
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 6.039997770935789e-06,
+      "loss": 0.908,
+      "step": 344100
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 6.033031945276473e-06,
+      "loss": 0.902,
+      "step": 344200
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 6.026066119617159e-06,
+      "loss": 0.8951,
+      "step": 344300
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 6.019100293957843e-06,
+      "loss": 0.9014,
+      "step": 344400
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 6.012134468298528e-06,
+      "loss": 0.8941,
+      "step": 344500
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 6.005168642639213e-06,
+      "loss": 0.9028,
+      "step": 344600
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 5.9982028169798965e-06,
+      "loss": 0.8792,
+      "step": 344700
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 5.991236991320581e-06,
+      "loss": 0.9255,
+      "step": 344800
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 5.984271165661266e-06,
+      "loss": 0.8769,
+      "step": 344900
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 5.9773053400019505e-06,
+      "loss": 0.9103,
+      "step": 345000
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 5.970339514342635e-06,
+      "loss": 0.924,
+      "step": 345100
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 5.96337368868332e-06,
+      "loss": 0.8969,
+      "step": 345200
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 5.956407863024004e-06,
+      "loss": 0.9029,
+      "step": 345300
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 5.949442037364688e-06,
+      "loss": 0.9153,
+      "step": 345400
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 5.9425458699619665e-06,
+      "loss": 0.8868,
+      "step": 345500
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 5.935580044302651e-06,
+      "loss": 0.9184,
+      "step": 345600
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 5.928614218643336e-06,
+      "loss": 0.8876,
+      "step": 345700
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 5.921648392984021e-06,
+      "loss": 0.9123,
+      "step": 345800
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 5.914682567324705e-06,
+      "loss": 0.8983,
+      "step": 345900
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 5.90771674166539e-06,
+      "loss": 0.9041,
+      "step": 346000
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 5.900750916006074e-06,
+      "loss": 0.9088,
+      "step": 346100
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 5.8937850903467585e-06,
+      "loss": 0.8975,
+      "step": 346200
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 5.886819264687443e-06,
+      "loss": 0.8626,
+      "step": 346300
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.879853439028129e-06,
+      "loss": 0.9134,
+      "step": 346400
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.872887613368813e-06,
+      "loss": 0.9149,
+      "step": 346500
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.865991445966091e-06,
+      "loss": 0.9046,
+      "step": 346600
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.859025620306775e-06,
+      "loss": 0.8771,
+      "step": 346700
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.852059794647459e-06,
+      "loss": 0.9007,
+      "step": 346800
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.845093968988144e-06,
+      "loss": 0.8749,
+      "step": 346900
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.8381281433288285e-06,
+      "loss": 0.924,
+      "step": 347000
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.831162317669513e-06,
+      "loss": 0.9206,
+      "step": 347100
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.824266150266791e-06,
+      "loss": 0.8967,
+      "step": 347200
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.817300324607476e-06,
+      "loss": 0.8749,
+      "step": 347300
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.810334498948161e-06,
+      "loss": 0.9193,
+      "step": 347400
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.8033686732888446e-06,
+      "loss": 0.8796,
+      "step": 347500
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.796402847629529e-06,
+      "loss": 0.9018,
+      "step": 347600
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.789437021970214e-06,
+      "loss": 0.8721,
+      "step": 347700
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.782471196310899e-06,
+      "loss": 0.9011,
+      "step": 347800
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.775505370651583e-06,
+      "loss": 0.9137,
+      "step": 347900
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.768539544992269e-06,
+      "loss": 0.8995,
+      "step": 348000
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.761573719332953e-06,
+      "loss": 0.9014,
+      "step": 348100
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.754607893673637e-06,
+      "loss": 0.9109,
+      "step": 348200
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.747642068014322e-06,
+      "loss": 0.8916,
+      "step": 348300
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.740676242355007e-06,
+      "loss": 0.8899,
+      "step": 348400
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.733710416695691e-06,
+      "loss": 0.8821,
+      "step": 348500
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.726744591036376e-06,
+      "loss": 0.8959,
+      "step": 348600
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.71977876537706e-06,
+      "loss": 0.8975,
+      "step": 348700
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.712812939717745e-06,
+      "loss": 0.8898,
+      "step": 348800
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.705847114058429e-06,
+      "loss": 0.8829,
+      "step": 348900
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.698881288399114e-06,
+      "loss": 0.8892,
+      "step": 349000
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.691915462739799e-06,
+      "loss": 0.8734,
+      "step": 349100
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.684949637080483e-06,
+      "loss": 0.91,
+      "step": 349200
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.677983811421168e-06,
+      "loss": 0.8991,
+      "step": 349300
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.671017985761853e-06,
+      "loss": 0.9066,
+      "step": 349400
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.664052160102537e-06,
+      "loss": 0.8815,
+      "step": 349500
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.657086334443222e-06,
+      "loss": 0.9071,
+      "step": 349600
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.650120508783907e-06,
+      "loss": 0.902,
+      "step": 349700
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.6431546831245906e-06,
+      "loss": 0.9186,
+      "step": 349800
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.636188857465275e-06,
+      "loss": 0.895,
+      "step": 349900
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 5.62922303180596e-06,
+      "loss": 0.8841,
+      "step": 350000
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 5.622257206146645e-06,
+      "loss": 0.8849,
+      "step": 350100
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 5.615291380487329e-06,
+      "loss": 0.8815,
+      "step": 350200
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 5.608325554828013e-06,
+      "loss": 0.8902,
+      "step": 350300
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 5.601359729168699e-06,
+      "loss": 0.9048,
+      "step": 350400
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 5.594393903509383e-06,
+      "loss": 0.9038,
+      "step": 350500
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 5.587428077850068e-06,
+      "loss": 0.8853,
+      "step": 350600
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 5.580462252190753e-06,
+      "loss": 0.9107,
+      "step": 350700
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 5.573496426531437e-06,
+      "loss": 0.9116,
+      "step": 350800
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 5.566530600872121e-06,
+      "loss": 0.8804,
+      "step": 350900
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 5.559564775212806e-06,
+      "loss": 0.8989,
+      "step": 351000
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 5.552598949553491e-06,
+      "loss": 0.8993,
+      "step": 351100
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 5.545633123894175e-06,
+      "loss": 0.8658,
+      "step": 351200
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 5.53866729823486e-06,
+      "loss": 0.8993,
+      "step": 351300
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.531701472575545e-06,
+      "loss": 0.9247,
+      "step": 351400
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.5247356469162285e-06,
+      "loss": 0.8967,
+      "step": 351500
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.517769821256913e-06,
+      "loss": 0.9107,
+      "step": 351600
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.510803995597599e-06,
+      "loss": 0.8948,
+      "step": 351700
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.503838169938283e-06,
+      "loss": 0.8823,
+      "step": 351800
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.496872344278968e-06,
+      "loss": 0.8879,
+      "step": 351900
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.489906518619652e-06,
+      "loss": 0.9149,
+      "step": 352000
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 5.482940692960337e-06,
+      "loss": 0.8996,
+      "step": 352100
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 5.475974867301021e-06,
+      "loss": 0.9015,
+      "step": 352200
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 5.469009041641706e-06,
+      "loss": 0.9153,
+      "step": 352300
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 5.462043215982391e-06,
+      "loss": 0.883,
+      "step": 352400
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 5.455077390323075e-06,
+      "loss": 0.8792,
+      "step": 352500
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 5.448111564663759e-06,
+      "loss": 0.9128,
+      "step": 352600
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 5.441145739004444e-06,
+      "loss": 0.8933,
+      "step": 352700
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 5.4341799133451285e-06,
+      "loss": 0.8921,
+      "step": 352800
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 5.427214087685814e-06,
+      "loss": 0.8886,
+      "step": 352900
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 5.420248262026499e-06,
+      "loss": 0.8993,
+      "step": 353000
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 5.4132824363671826e-06,
+      "loss": 0.9102,
+      "step": 353100
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 5.406316610707867e-06,
+      "loss": 0.904,
+      "step": 353200
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 5.399350785048552e-06,
+      "loss": 0.8972,
+      "step": 353300
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 5.392384959389237e-06,
+      "loss": 0.9171,
+      "step": 353400
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 5.385419133729921e-06,
+      "loss": 0.9246,
+      "step": 353500
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 5.378453308070606e-06,
+      "loss": 0.9323,
+      "step": 353600
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 5.37148748241129e-06,
+      "loss": 0.8837,
+      "step": 353700
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 5.364591315008569e-06,
+      "loss": 0.8957,
+      "step": 353800
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 5.357625489349253e-06,
+      "loss": 0.875,
+      "step": 353900
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 5.350659663689937e-06,
+      "loss": 0.8903,
+      "step": 354000
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 5.343693838030622e-06,
+      "loss": 0.8809,
+      "step": 354100
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 5.336728012371307e-06,
+      "loss": 0.8936,
+      "step": 354200
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 5.329831844968584e-06,
+      "loss": 0.8777,
+      "step": 354300
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 5.322866019309269e-06,
+      "loss": 0.8892,
+      "step": 354400
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 5.315900193649954e-06,
+      "loss": 0.8496,
+      "step": 354500
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 5.308934367990638e-06,
+      "loss": 0.8915,
+      "step": 354600
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 5.301968542331323e-06,
+      "loss": 0.9091,
+      "step": 354700
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 5.295002716672007e-06,
+      "loss": 0.8996,
+      "step": 354800
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 5.288036891012692e-06,
+      "loss": 0.8934,
+      "step": 354900
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.281071065353377e-06,
+      "loss": 0.914,
+      "step": 355000
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.2741052396940614e-06,
+      "loss": 0.8762,
+      "step": 355100
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.267139414034745e-06,
+      "loss": 0.9032,
+      "step": 355200
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.26017358837543e-06,
+      "loss": 0.8971,
+      "step": 355300
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.253207762716115e-06,
+      "loss": 0.9071,
+      "step": 355400
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.246241937056799e-06,
+      "loss": 0.9246,
+      "step": 355500
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.239276111397484e-06,
+      "loss": 0.9098,
+      "step": 355600
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.232310285738169e-06,
+      "loss": 0.8697,
+      "step": 355700
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.225344460078853e-06,
+      "loss": 0.914,
+      "step": 355800
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.218378634419538e-06,
+      "loss": 0.8898,
+      "step": 355900
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.211412808760223e-06,
+      "loss": 0.9309,
+      "step": 356000
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.204446983100907e-06,
+      "loss": 0.906,
+      "step": 356100
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.197481157441592e-06,
+      "loss": 0.8984,
+      "step": 356200
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.190515331782276e-06,
+      "loss": 0.9126,
+      "step": 356300
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.183549506122961e-06,
+      "loss": 0.894,
+      "step": 356400
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.176583680463645e-06,
+      "loss": 0.8923,
+      "step": 356500
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.16961785480433e-06,
+      "loss": 0.8962,
+      "step": 356600
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.162652029145015e-06,
+      "loss": 0.8862,
+      "step": 356700
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.155686203485699e-06,
+      "loss": 0.8902,
+      "step": 356800
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.148720377826383e-06,
+      "loss": 0.9039,
+      "step": 356900
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.141754552167069e-06,
+      "loss": 0.8818,
+      "step": 357000
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.134788726507753e-06,
+      "loss": 0.8793,
+      "step": 357100
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.127822900848438e-06,
+      "loss": 0.9053,
+      "step": 357200
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.120857075189123e-06,
+      "loss": 0.8951,
+      "step": 357300
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.113891249529807e-06,
+      "loss": 0.8964,
+      "step": 357400
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.106925423870491e-06,
+      "loss": 0.8916,
+      "step": 357500
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.099959598211176e-06,
+      "loss": 0.8703,
+      "step": 357600
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.092993772551861e-06,
+      "loss": 0.9257,
+      "step": 357700
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.086027946892545e-06,
+      "loss": 0.8877,
+      "step": 357800
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.07906212123323e-06,
+      "loss": 0.8723,
+      "step": 357900
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.072096295573914e-06,
+      "loss": 0.8969,
+      "step": 358000
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.0651304699145985e-06,
+      "loss": 0.8809,
+      "step": 358100
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.058164644255284e-06,
+      "loss": 0.916,
+      "step": 358200
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.051198818595969e-06,
+      "loss": 0.9012,
+      "step": 358300
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.0442329929366534e-06,
+      "loss": 0.9146,
+      "step": 358400
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.037267167277337e-06,
+      "loss": 0.8625,
+      "step": 358500
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 5.030301341618022e-06,
+      "loss": 0.9046,
+      "step": 358600
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 5.023335515958707e-06,
+      "loss": 0.8989,
+      "step": 358700
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 5.016369690299391e-06,
+      "loss": 0.9112,
+      "step": 358800
+    },
+    {
+      "epoch": 5.0,
+      "eval_gen_len": 20.0,
+      "eval_loss": 1.1667309999465942,
+      "eval_rouge1": 12.5404,
+      "eval_rouge2": 3.7842,
+      "eval_rougeL": 12.0541,
+      "eval_rougeLsum": 12.1643,
+      "eval_runtime": 1508.2531,
+      "eval_samples_per_second": 8.863,
+      "eval_steps_per_second": 2.216,
+      "step": 358895
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 5.009403864640076e-06,
+      "loss": 0.9271,
+      "step": 358900
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 5.002438038980761e-06,
+      "loss": 0.8275,
+      "step": 359000
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.9954722133214445e-06,
+      "loss": 0.8209,
+      "step": 359100
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.9885760459187235e-06,
+      "loss": 0.8225,
+      "step": 359200
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.981610220259407e-06,
+      "loss": 0.8078,
+      "step": 359300
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.974644394600092e-06,
+      "loss": 0.8394,
+      "step": 359400
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.967678568940777e-06,
+      "loss": 0.8552,
+      "step": 359500
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.960712743281461e-06,
+      "loss": 0.8287,
+      "step": 359600
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.953746917622146e-06,
+      "loss": 0.7978,
+      "step": 359700
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.94678109196283e-06,
+      "loss": 0.8254,
+      "step": 359800
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.939815266303515e-06,
+      "loss": 0.805,
+      "step": 359900
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.932849440644199e-06,
+      "loss": 0.8448,
+      "step": 360000
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.925883614984884e-06,
+      "loss": 0.8456,
+      "step": 360100
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.9189177893255695e-06,
+      "loss": 0.8382,
+      "step": 360200
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.911951963666254e-06,
+      "loss": 0.8349,
+      "step": 360300
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.904986138006938e-06,
+      "loss": 0.8105,
+      "step": 360400
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.898020312347623e-06,
+      "loss": 0.8326,
+      "step": 360500
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.891054486688307e-06,
+      "loss": 0.8429,
+      "step": 360600
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.884088661028992e-06,
+      "loss": 0.8391,
+      "step": 360700
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.877122835369677e-06,
+      "loss": 0.8196,
+      "step": 360800
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.870226667966954e-06,
+      "loss": 0.8342,
+      "step": 360900
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.8632608423076395e-06,
+      "loss": 0.8288,
+      "step": 361000
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.856295016648323e-06,
+      "loss": 0.8454,
+      "step": 361100
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.849329190989008e-06,
+      "loss": 0.8364,
+      "step": 361200
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.842363365329693e-06,
+      "loss": 0.8249,
+      "step": 361300
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.8353975396703774e-06,
+      "loss": 0.7996,
+      "step": 361400
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.828431714011062e-06,
+      "loss": 0.8294,
+      "step": 361500
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.821465888351747e-06,
+      "loss": 0.816,
+      "step": 361600
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.814500062692431e-06,
+      "loss": 0.8391,
+      "step": 361700
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.807534237033115e-06,
+      "loss": 0.8331,
+      "step": 361800
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.8005684113738e-06,
+      "loss": 0.8248,
+      "step": 361900
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.793602585714485e-06,
+      "loss": 0.8309,
+      "step": 362000
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.786636760055169e-06,
+      "loss": 0.8254,
+      "step": 362100
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.779670934395854e-06,
+      "loss": 0.8449,
+      "step": 362200
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.772705108736539e-06,
+      "loss": 0.8179,
+      "step": 362300
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.765739283077223e-06,
+      "loss": 0.8216,
+      "step": 362400
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.758773457417908e-06,
+      "loss": 0.8372,
+      "step": 362500
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.751807631758593e-06,
+      "loss": 0.8232,
+      "step": 362600
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.7448418060992775e-06,
+      "loss": 0.8381,
+      "step": 362700
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.737875980439961e-06,
+      "loss": 0.8593,
+      "step": 362800
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.730910154780646e-06,
+      "loss": 0.8482,
+      "step": 362900
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.723944329121331e-06,
+      "loss": 0.8209,
+      "step": 363000
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.716978503462015e-06,
+      "loss": 0.8023,
+      "step": 363100
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.7100126778027e-06,
+      "loss": 0.8405,
+      "step": 363200
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.703046852143385e-06,
+      "loss": 0.8098,
+      "step": 363300
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.6960810264840685e-06,
+      "loss": 0.8205,
+      "step": 363400
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.689115200824754e-06,
+      "loss": 0.8516,
+      "step": 363500
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.682149375165439e-06,
+      "loss": 0.8504,
+      "step": 363600
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.6751835495061234e-06,
+      "loss": 0.8465,
+      "step": 363700
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.668217723846808e-06,
+      "loss": 0.8444,
+      "step": 363800
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.661251898187492e-06,
+      "loss": 0.8536,
+      "step": 363900
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.654286072528177e-06,
+      "loss": 0.7937,
+      "step": 364000
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.647320246868861e-06,
+      "loss": 0.8394,
+      "step": 364100
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.640354421209546e-06,
+      "loss": 0.8356,
+      "step": 364200
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.633388595550231e-06,
+      "loss": 0.8217,
+      "step": 364300
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.626422769890915e-06,
+      "loss": 0.8573,
+      "step": 364400
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.619456944231599e-06,
+      "loss": 0.8214,
+      "step": 364500
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.612491118572284e-06,
+      "loss": 0.8246,
+      "step": 364600
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.6055252929129694e-06,
+      "loss": 0.8251,
+      "step": 364700
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.598559467253654e-06,
+      "loss": 0.84,
+      "step": 364800
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.591593641594339e-06,
+      "loss": 0.8238,
+      "step": 364900
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.5846278159350235e-06,
+      "loss": 0.8194,
+      "step": 365000
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.577661990275707e-06,
+      "loss": 0.8537,
+      "step": 365100
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.570696164616392e-06,
+      "loss": 0.8201,
+      "step": 365200
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.563730338957077e-06,
+      "loss": 0.8323,
+      "step": 365300
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.556764513297761e-06,
+      "loss": 0.8201,
+      "step": 365400
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.549798687638446e-06,
+      "loss": 0.8394,
+      "step": 365500
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.54283286197913e-06,
+      "loss": 0.8441,
+      "step": 365600
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.5358670363198146e-06,
+      "loss": 0.8349,
+      "step": 365700
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.528901210660499e-06,
+      "loss": 0.8408,
+      "step": 365800
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.521935385001185e-06,
+      "loss": 0.8681,
+      "step": 365900
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.5149695593418695e-06,
+      "loss": 0.8023,
+      "step": 366000
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.508003733682554e-06,
+      "loss": 0.8192,
+      "step": 366100
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.501037908023238e-06,
+      "loss": 0.842,
+      "step": 366200
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.494072082363923e-06,
+      "loss": 0.8262,
+      "step": 366300
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.487106256704607e-06,
+      "loss": 0.8162,
+      "step": 366400
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.480140431045292e-06,
+      "loss": 0.8301,
+      "step": 366500
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.473174605385977e-06,
+      "loss": 0.8343,
+      "step": 366600
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.466208779726661e-06,
+      "loss": 0.8363,
+      "step": 366700
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.4593126123239395e-06,
+      "loss": 0.8299,
+      "step": 366800
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.452346786664623e-06,
+      "loss": 0.8213,
+      "step": 366900
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.445380961005308e-06,
+      "loss": 0.8165,
+      "step": 367000
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.438415135345993e-06,
+      "loss": 0.8285,
+      "step": 367100
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.431449309686677e-06,
+      "loss": 0.8347,
+      "step": 367200
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.424483484027362e-06,
+      "loss": 0.8282,
+      "step": 367300
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.417517658368047e-06,
+      "loss": 0.8539,
+      "step": 367400
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.410551832708731e-06,
+      "loss": 0.8361,
+      "step": 367500
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.403586007049415e-06,
+      "loss": 0.8209,
+      "step": 367600
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.3966201813901e-06,
+      "loss": 0.8303,
+      "step": 367700
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.389654355730785e-06,
+      "loss": 0.8435,
+      "step": 367800
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.382688530071469e-06,
+      "loss": 0.8139,
+      "step": 367900
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.375722704412155e-06,
+      "loss": 0.8544,
+      "step": 368000
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.368756878752839e-06,
+      "loss": 0.8244,
+      "step": 368100
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.361791053093523e-06,
+      "loss": 0.8323,
+      "step": 368200
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.354825227434208e-06,
+      "loss": 0.8592,
+      "step": 368300
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.347859401774893e-06,
+      "loss": 0.8287,
+      "step": 368400
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.3408935761155774e-06,
+      "loss": 0.8361,
+      "step": 368500
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.333927750456261e-06,
+      "loss": 0.823,
+      "step": 368600
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.326961924796946e-06,
+      "loss": 0.8323,
+      "step": 368700
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.319996099137631e-06,
+      "loss": 0.8575,
+      "step": 368800
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.313030273478315e-06,
+      "loss": 0.8333,
+      "step": 368900
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.306064447819e-06,
+      "loss": 0.8535,
+      "step": 369000
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.299098622159685e-06,
+      "loss": 0.8152,
+      "step": 369100
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.292132796500369e-06,
+      "loss": 0.8207,
+      "step": 369200
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.285166970841054e-06,
+      "loss": 0.8351,
+      "step": 369300
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.278201145181739e-06,
+      "loss": 0.8194,
+      "step": 369400
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.2712353195224234e-06,
+      "loss": 0.8532,
+      "step": 369500
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.264339152119701e-06,
+      "loss": 0.8306,
+      "step": 369600
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.257373326460385e-06,
+      "loss": 0.8295,
+      "step": 369700
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.25040750080107e-06,
+      "loss": 0.8504,
+      "step": 369800
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.243441675141754e-06,
+      "loss": 0.8283,
+      "step": 369900
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.2364758494824394e-06,
+      "loss": 0.8485,
+      "step": 370000
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.229510023823124e-06,
+      "loss": 0.81,
+      "step": 370100
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.222544198163809e-06,
+      "loss": 0.8437,
+      "step": 370200
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.2155783725044935e-06,
+      "loss": 0.8365,
+      "step": 370300
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.208612546845178e-06,
+      "loss": 0.8538,
+      "step": 370400
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.201646721185862e-06,
+      "loss": 0.8265,
+      "step": 370500
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.194680895526547e-06,
+      "loss": 0.8176,
+      "step": 370600
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.187715069867231e-06,
+      "loss": 0.8338,
+      "step": 370700
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.180749244207916e-06,
+      "loss": 0.8237,
+      "step": 370800
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.173783418548601e-06,
+      "loss": 0.8213,
+      "step": 370900
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.166817592889285e-06,
+      "loss": 0.8419,
+      "step": 371000
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.159851767229969e-06,
+      "loss": 0.8052,
+      "step": 371100
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.152885941570655e-06,
+      "loss": 0.7705,
+      "step": 371200
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.1459201159113395e-06,
+      "loss": 0.8226,
+      "step": 371300
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.138954290252024e-06,
+      "loss": 0.8284,
+      "step": 371400
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.131988464592709e-06,
+      "loss": 0.836,
+      "step": 371500
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.125022638933393e-06,
+      "loss": 0.8067,
+      "step": 371600
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.118056813274077e-06,
+      "loss": 0.8206,
+      "step": 371700
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.111090987614762e-06,
+      "loss": 0.8185,
+      "step": 371800
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.104125161955447e-06,
+      "loss": 0.8344,
+      "step": 371900
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.097159336296131e-06,
+      "loss": 0.8392,
+      "step": 372000
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.090193510636816e-06,
+      "loss": 0.8224,
+      "step": 372100
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.0832276849775e-06,
+      "loss": 0.8247,
+      "step": 372200
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.076261859318185e-06,
+      "loss": 0.8541,
+      "step": 372300
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.069296033658869e-06,
+      "loss": 0.8669,
+      "step": 372400
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.062330207999555e-06,
+      "loss": 0.8228,
+      "step": 372500
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.0553643823402395e-06,
+      "loss": 0.8359,
+      "step": 372600
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.048398556680923e-06,
+      "loss": 0.8489,
+      "step": 372700
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.041432731021608e-06,
+      "loss": 0.8537,
+      "step": 372800
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.034536563618885e-06,
+      "loss": 0.8625,
+      "step": 372900
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.02757073795957e-06,
+      "loss": 0.8186,
+      "step": 373000
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.020604912300255e-06,
+      "loss": 0.8274,
+      "step": 373100
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.013708744897533e-06,
+      "loss": 0.8246,
+      "step": 373200
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.0067429192382175e-06,
+      "loss": 0.8173,
+      "step": 373300
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 3.999777093578902e-06,
+      "loss": 0.8404,
+      "step": 373400
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 3.992811267919587e-06,
+      "loss": 0.8254,
+      "step": 373500
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 3.985845442260271e-06,
+      "loss": 0.8032,
+      "step": 373600
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.978879616600955e-06,
+      "loss": 0.81,
+      "step": 373700
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.97191379094164e-06,
+      "loss": 0.8396,
+      "step": 373800
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.964947965282325e-06,
+      "loss": 0.8314,
+      "step": 373900
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.957982139623009e-06,
+      "loss": 0.8444,
+      "step": 374000
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.951016313963695e-06,
+      "loss": 0.8479,
+      "step": 374100
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.944050488304379e-06,
+      "loss": 0.8289,
+      "step": 374200
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.9370846626450635e-06,
+      "loss": 0.8545,
+      "step": 374300
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.930118836985748e-06,
+      "loss": 0.8515,
+      "step": 374400
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.923153011326433e-06,
+      "loss": 0.8247,
+      "step": 374500
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.9161871856671175e-06,
+      "loss": 0.8546,
+      "step": 374600
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.909221360007802e-06,
+      "loss": 0.8536,
+      "step": 374700
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.902255534348486e-06,
+      "loss": 0.8446,
+      "step": 374800
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.895289708689171e-06,
+      "loss": 0.8714,
+      "step": 374900
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.888323883029855e-06,
+      "loss": 0.8566,
+      "step": 375000
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.88135805737054e-06,
+      "loss": 0.8568,
+      "step": 375100
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.874392231711225e-06,
+      "loss": 0.8445,
+      "step": 375200
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.8674264060519094e-06,
+      "loss": 0.8545,
+      "step": 375300
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.860460580392594e-06,
+      "loss": 0.8354,
+      "step": 375400
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.853494754733279e-06,
+      "loss": 0.8702,
+      "step": 375500
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.8465289290739635e-06,
+      "loss": 0.83,
+      "step": 375600
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.839563103414648e-06,
+      "loss": 0.8258,
+      "step": 375700
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 3.832597277755333e-06,
+      "loss": 0.8523,
+      "step": 375800
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 3.825631452096017e-06,
+      "loss": 0.8535,
+      "step": 375900
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 3.818665626436701e-06,
+      "loss": 0.8125,
+      "step": 376000
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 3.811699800777386e-06,
+      "loss": 0.8218,
+      "step": 376100
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 3.804733975118071e-06,
+      "loss": 0.8384,
+      "step": 376200
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 3.797768149458756e-06,
+      "loss": 0.8437,
+      "step": 376300
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 3.7908023237994397e-06,
+      "loss": 0.8364,
+      "step": 376400
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.7838364981401244e-06,
+      "loss": 0.8538,
+      "step": 376500
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.776870672480809e-06,
+      "loss": 0.8263,
+      "step": 376600
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.7699048468214937e-06,
+      "loss": 0.8514,
+      "step": 376700
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.7629390211621784e-06,
+      "loss": 0.8167,
+      "step": 376800
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.7559731955028635e-06,
+      "loss": 0.8557,
+      "step": 376900
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.749007369843548e-06,
+      "loss": 0.8237,
+      "step": 377000
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.742041544184232e-06,
+      "loss": 0.8251,
+      "step": 377100
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.7350757185249167e-06,
+      "loss": 0.8271,
+      "step": 377200
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.7281098928656014e-06,
+      "loss": 0.8176,
+      "step": 377300
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.7211440672062857e-06,
+      "loss": 0.8048,
+      "step": 377400
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.7141782415469708e-06,
+      "loss": 0.8213,
+      "step": 377500
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.7072124158876555e-06,
+      "loss": 0.8204,
+      "step": 377600
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.7002465902283397e-06,
+      "loss": 0.8593,
+      "step": 377700
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.6932807645690244e-06,
+      "loss": 0.8453,
+      "step": 377800
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.686314938909709e-06,
+      "loss": 0.852,
+      "step": 377900
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.6793491132503934e-06,
+      "loss": 0.8347,
+      "step": 378000
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.6723832875910785e-06,
+      "loss": 0.8226,
+      "step": 378100
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.665417461931763e-06,
+      "loss": 0.8266,
+      "step": 378200
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.6584516362724474e-06,
+      "loss": 0.8069,
+      "step": 378300
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.651555468869725e-06,
+      "loss": 0.8233,
+      "step": 378400
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.64458964321041e-06,
+      "loss": 0.8328,
+      "step": 378500
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.6376238175510945e-06,
+      "loss": 0.8203,
+      "step": 378600
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 3.6306579918917787e-06,
+      "loss": 0.8252,
+      "step": 378700
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 3.6236921662324634e-06,
+      "loss": 0.831,
+      "step": 378800
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 3.6167263405731485e-06,
+      "loss": 0.8212,
+      "step": 378900
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 3.6097605149138328e-06,
+      "loss": 0.8431,
+      "step": 379000
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 3.6027946892545175e-06,
+      "loss": 0.81,
+      "step": 379100
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 3.595828863595202e-06,
+      "loss": 0.8325,
+      "step": 379200
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 3.5888630379358864e-06,
+      "loss": 0.8471,
+      "step": 379300
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3.581897212276571e-06,
+      "loss": 0.8566,
+      "step": 379400
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3.574931386617256e-06,
+      "loss": 0.8169,
+      "step": 379500
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3.5679655609579405e-06,
+      "loss": 0.8231,
+      "step": 379600
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3.560999735298625e-06,
+      "loss": 0.8377,
+      "step": 379700
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3.5540339096393094e-06,
+      "loss": 0.8441,
+      "step": 379800
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3.547068083979994e-06,
+      "loss": 0.834,
+      "step": 379900
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3.5401022583206788e-06,
+      "loss": 0.8248,
+      "step": 380000
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.5331364326613634e-06,
+      "loss": 0.8165,
+      "step": 380100
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.526170607002048e-06,
+      "loss": 0.8424,
+      "step": 380200
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.519204781342733e-06,
+      "loss": 0.8317,
+      "step": 380300
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.512238955683417e-06,
+      "loss": 0.8206,
+      "step": 380400
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.505342788280695e-06,
+      "loss": 0.8332,
+      "step": 380500
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.4983769626213795e-06,
+      "loss": 0.8391,
+      "step": 380600
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.491411136962064e-06,
+      "loss": 0.8209,
+      "step": 380700
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.484445311302749e-06,
+      "loss": 0.8131,
+      "step": 380800
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.4774794856434335e-06,
+      "loss": 0.8374,
+      "step": 380900
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.470513659984118e-06,
+      "loss": 0.8362,
+      "step": 381000
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.4635478343248025e-06,
+      "loss": 0.8236,
+      "step": 381100
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.456582008665487e-06,
+      "loss": 0.8184,
+      "step": 381200
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.449616183006172e-06,
+      "loss": 0.8451,
+      "step": 381300
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.442650357346856e-06,
+      "loss": 0.821,
+      "step": 381400
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.435684531687541e-06,
+      "loss": 0.8402,
+      "step": 381500
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.428718706028226e-06,
+      "loss": 0.8643,
+      "step": 381600
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.42175288036891e-06,
+      "loss": 0.8172,
+      "step": 381700
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.414787054709595e-06,
+      "loss": 0.8328,
+      "step": 381800
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.4078212290502795e-06,
+      "loss": 0.8191,
+      "step": 381900
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.4008554033909638e-06,
+      "loss": 0.8503,
+      "step": 382000
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.393889577731649e-06,
+      "loss": 0.8623,
+      "step": 382100
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.3869237520723335e-06,
+      "loss": 0.8358,
+      "step": 382200
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.379957926413018e-06,
+      "loss": 0.8478,
+      "step": 382300
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.3729921007537025e-06,
+      "loss": 0.825,
+      "step": 382400
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.3660262750943867e-06,
+      "loss": 0.8273,
+      "step": 382500
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.3590604494350714e-06,
+      "loss": 0.8264,
+      "step": 382600
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.352094623775756e-06,
+      "loss": 0.8397,
+      "step": 382700
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.345128798116441e-06,
+      "loss": 0.8351,
+      "step": 382800
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.3381629724571255e-06,
+      "loss": 0.8359,
+      "step": 382900
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.33119714679781e-06,
+      "loss": 0.8243,
+      "step": 383000
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.3242313211384944e-06,
+      "loss": 0.8176,
+      "step": 383100
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.317265495479179e-06,
+      "loss": 0.8401,
+      "step": 383200
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.310299669819864e-06,
+      "loss": 0.825,
+      "step": 383300
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.3033338441605485e-06,
+      "loss": 0.8245,
+      "step": 383400
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.296368018501233e-06,
+      "loss": 0.8646,
+      "step": 383500
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.289402192841918e-06,
+      "loss": 0.8621,
+      "step": 383600
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.282436367182602e-06,
+      "loss": 0.83,
+      "step": 383700
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.2754705415232868e-06,
+      "loss": 0.8519,
+      "step": 383800
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.268504715863971e-06,
+      "loss": 0.8233,
+      "step": 383900
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.261538890204656e-06,
+      "loss": 0.8262,
+      "step": 384000
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.254573064545341e-06,
+      "loss": 0.8491,
+      "step": 384100
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.2476768971426185e-06,
+      "loss": 0.8227,
+      "step": 384200
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.240780729739896e-06,
+      "loss": 0.8404,
+      "step": 384300
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.233814904080581e-06,
+      "loss": 0.8342,
+      "step": 384400
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.2268490784212656e-06,
+      "loss": 0.8264,
+      "step": 384500
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.21988325276195e-06,
+      "loss": 0.8257,
+      "step": 384600
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.2129174271026345e-06,
+      "loss": 0.8442,
+      "step": 384700
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.2059516014433192e-06,
+      "loss": 0.8405,
+      "step": 384800
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.1989857757840035e-06,
+      "loss": 0.8456,
+      "step": 384900
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.1920199501246886e-06,
+      "loss": 0.8348,
+      "step": 385000
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.1850541244653733e-06,
+      "loss": 0.8129,
+      "step": 385100
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.1780882988060575e-06,
+      "loss": 0.847,
+      "step": 385200
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.1711224731467422e-06,
+      "loss": 0.8226,
+      "step": 385300
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.1641566474874265e-06,
+      "loss": 0.8348,
+      "step": 385400
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.157190821828111e-06,
+      "loss": 0.821,
+      "step": 385500
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.1502249961687963e-06,
+      "loss": 0.8228,
+      "step": 385600
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.1432591705094805e-06,
+      "loss": 0.8163,
+      "step": 385700
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.136293344850165e-06,
+      "loss": 0.8451,
+      "step": 385800
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.12932751919085e-06,
+      "loss": 0.8191,
+      "step": 385900
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.122361693531534e-06,
+      "loss": 0.8147,
+      "step": 386000
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.115395867872219e-06,
+      "loss": 0.8197,
+      "step": 386100
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.108430042212904e-06,
+      "loss": 0.8634,
+      "step": 386200
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.101464216553588e-06,
+      "loss": 0.8023,
+      "step": 386300
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.094498390894273e-06,
+      "loss": 0.8237,
+      "step": 386400
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.0875325652349576e-06,
+      "loss": 0.8171,
+      "step": 386500
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3.080566739575642e-06,
+      "loss": 0.798,
+      "step": 386600
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3.0736009139163265e-06,
+      "loss": 0.8277,
+      "step": 386700
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3.066635088257011e-06,
+      "loss": 0.8288,
+      "step": 386800
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3.059669262597696e-06,
+      "loss": 0.8021,
+      "step": 386900
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3.0527034369383806e-06,
+      "loss": 0.8283,
+      "step": 387000
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3.045737611279065e-06,
+      "loss": 0.8425,
+      "step": 387100
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3.0387717856197495e-06,
+      "loss": 0.818,
+      "step": 387200
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 3.031805959960434e-06,
+      "loss": 0.8403,
+      "step": 387300
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 3.024840134301119e-06,
+      "loss": 0.8149,
+      "step": 387400
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 3.0178743086418036e-06,
+      "loss": 0.8069,
+      "step": 387500
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 3.0109084829824882e-06,
+      "loss": 0.8554,
+      "step": 387600
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 3.0039426573231725e-06,
+      "loss": 0.8302,
+      "step": 387700
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 2.996976831663857e-06,
+      "loss": 0.8193,
+      "step": 387800
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 2.990011006004542e-06,
+      "loss": 0.8342,
+      "step": 387900
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 2.983045180345226e-06,
+      "loss": 0.8531,
+      "step": 388000
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 2.9760793546859112e-06,
+      "loss": 0.8356,
+      "step": 388100
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 2.9691135290265955e-06,
+      "loss": 0.8063,
+      "step": 388200
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 2.96214770336728e-06,
+      "loss": 0.8303,
+      "step": 388300
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 2.955181877707965e-06,
+      "loss": 0.8202,
+      "step": 388400
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 2.948216052048649e-06,
+      "loss": 0.8353,
+      "step": 388500
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 2.941250226389334e-06,
+      "loss": 0.838,
+      "step": 388600
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 2.934284400730019e-06,
+      "loss": 0.8324,
+      "step": 388700
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 2.927318575070703e-06,
+      "loss": 0.8324,
+      "step": 388800
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 2.920352749411388e-06,
+      "loss": 0.8674,
+      "step": 388900
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 2.9133869237520725e-06,
+      "loss": 0.8197,
+      "step": 389000
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 2.906421098092757e-06,
+      "loss": 0.8349,
+      "step": 389100
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 2.8994552724334415e-06,
+      "loss": 0.8212,
+      "step": 389200
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 2.8924894467741266e-06,
+      "loss": 0.8134,
+      "step": 389300
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 2.885523621114811e-06,
+      "loss": 0.8445,
+      "step": 389400
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 2.8785577954554955e-06,
+      "loss": 0.8518,
+      "step": 389500
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 2.8716616280527732e-06,
+      "loss": 0.842,
+      "step": 389600
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 2.864695802393458e-06,
+      "loss": 0.819,
+      "step": 389700
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 2.857729976734142e-06,
+      "loss": 0.8206,
+      "step": 389800
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 2.850764151074827e-06,
+      "loss": 0.8133,
+      "step": 389900
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 2.8437983254155115e-06,
+      "loss": 0.8705,
+      "step": 390000
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 2.8368324997561962e-06,
+      "loss": 0.7895,
+      "step": 390100
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 2.829866674096881e-06,
+      "loss": 0.8369,
+      "step": 390200
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 2.8229008484375656e-06,
+      "loss": 0.8482,
+      "step": 390300
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 2.81593502277825e-06,
+      "loss": 0.839,
+      "step": 390400
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 2.8089691971189345e-06,
+      "loss": 0.8442,
+      "step": 390500
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 2.802003371459619e-06,
+      "loss": 0.8403,
+      "step": 390600
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 2.795037545800304e-06,
+      "loss": 0.8377,
+      "step": 390700
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 2.7880717201409886e-06,
+      "loss": 0.7882,
+      "step": 390800
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 2.781105894481673e-06,
+      "loss": 0.8355,
+      "step": 390900
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 2.7741400688223575e-06,
+      "loss": 0.8392,
+      "step": 391000
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 2.767174243163042e-06,
+      "loss": 0.8438,
+      "step": 391100
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 2.7602084175037265e-06,
+      "loss": 0.8294,
+      "step": 391200
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 2.7532425918444116e-06,
+      "loss": 0.8563,
+      "step": 391300
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 2.7462767661850963e-06,
+      "loss": 0.7958,
+      "step": 391400
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 2.7393109405257805e-06,
+      "loss": 0.8421,
+      "step": 391500
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 2.732345114866465e-06,
+      "loss": 0.8167,
+      "step": 391600
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 2.72537928920715e-06,
+      "loss": 0.815,
+      "step": 391700
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 2.718413463547834e-06,
+      "loss": 0.8459,
+      "step": 391800
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 2.7114476378885192e-06,
+      "loss": 0.83,
+      "step": 391900
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 2.704481812229204e-06,
+      "loss": 0.8363,
+      "step": 392000
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 2.697515986569888e-06,
+      "loss": 0.8393,
+      "step": 392100
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 2.690550160910573e-06,
+      "loss": 0.8446,
+      "step": 392200
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 2.6836539935078506e-06,
+      "loss": 0.8503,
+      "step": 392300
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 2.6766881678485353e-06,
+      "loss": 0.858,
+      "step": 392400
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 2.6697223421892195e-06,
+      "loss": 0.8208,
+      "step": 392500
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 2.662756516529904e-06,
+      "loss": 0.8277,
+      "step": 392600
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 2.6557906908705893e-06,
+      "loss": 0.8217,
+      "step": 392700
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 2.6488248652112736e-06,
+      "loss": 0.7853,
+      "step": 392800
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 2.6418590395519582e-06,
+      "loss": 0.8538,
+      "step": 392900
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 2.634893213892643e-06,
+      "loss": 0.8455,
+      "step": 393000
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 2.627927388233327e-06,
+      "loss": 0.8299,
+      "step": 393100
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 2.620961562574012e-06,
+      "loss": 0.8385,
+      "step": 393200
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 2.6139957369146966e-06,
+      "loss": 0.8374,
+      "step": 393300
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 2.6070299112553812e-06,
+      "loss": 0.818,
+      "step": 393400
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 2.600064085596066e-06,
+      "loss": 0.8294,
+      "step": 393500
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 2.59309825993675e-06,
+      "loss": 0.8212,
+      "step": 393600
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 2.586132434277435e-06,
+      "loss": 0.8319,
+      "step": 393700
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 2.5791666086181196e-06,
+      "loss": 0.8353,
+      "step": 393800
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 2.572200782958804e-06,
+      "loss": 0.8453,
+      "step": 393900
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 2.565234957299489e-06,
+      "loss": 0.8301,
+      "step": 394000
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 2.5582691316401736e-06,
+      "loss": 0.8026,
+      "step": 394100
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 2.551303305980858e-06,
+      "loss": 0.8317,
+      "step": 394200
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 2.5443374803215425e-06,
+      "loss": 0.8567,
+      "step": 394300
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 2.5373716546622272e-06,
+      "loss": 0.842,
+      "step": 394400
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 2.5304058290029115e-06,
+      "loss": 0.8337,
+      "step": 394500
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 2.5234400033435966e-06,
+      "loss": 0.8284,
+      "step": 394600
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 2.5164741776842813e-06,
+      "loss": 0.8391,
+      "step": 394700
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 2.5095083520249655e-06,
+      "loss": 0.8143,
+      "step": 394800
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 2.5025425263656502e-06,
+      "loss": 0.8268,
+      "step": 394900
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 2.4955767007063345e-06,
+      "loss": 0.8075,
+      "step": 395000
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 2.488610875047019e-06,
+      "loss": 0.8218,
+      "step": 395100
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 2.4816450493877043e-06,
+      "loss": 0.8452,
+      "step": 395200
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 2.4746792237283885e-06,
+      "loss": 0.8349,
+      "step": 395300
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 2.467713398069073e-06,
+      "loss": 0.8594,
+      "step": 395400
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 2.460747572409758e-06,
+      "loss": 0.8357,
+      "step": 395500
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 2.453781746750442e-06,
+      "loss": 0.8482,
+      "step": 395600
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 2.446815921091127e-06,
+      "loss": 0.8132,
+      "step": 395700
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 2.439850095431812e-06,
+      "loss": 0.8626,
+      "step": 395800
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 2.432884269772496e-06,
+      "loss": 0.8244,
+      "step": 395900
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 2.425918444113181e-06,
+      "loss": 0.8427,
+      "step": 396000
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 2.4189526184538656e-06,
+      "loss": 0.8401,
+      "step": 396100
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 2.4120564510511433e-06,
+      "loss": 0.8222,
+      "step": 396200
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 2.4050906253918275e-06,
+      "loss": 0.845,
+      "step": 396300
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 2.3981247997325122e-06,
+      "loss": 0.818,
+      "step": 396400
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 2.391158974073197e-06,
+      "loss": 0.8338,
+      "step": 396500
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 2.3841931484138816e-06,
+      "loss": 0.8357,
+      "step": 396600
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 2.3772273227545663e-06,
+      "loss": 0.8255,
+      "step": 396700
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 2.370261497095251e-06,
+      "loss": 0.8069,
+      "step": 396800
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 2.363295671435935e-06,
+      "loss": 0.8002,
+      "step": 396900
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 2.35632984577662e-06,
+      "loss": 0.8583,
+      "step": 397000
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 2.3493640201173046e-06,
+      "loss": 0.8307,
+      "step": 397100
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 2.3423981944579893e-06,
+      "loss": 0.8362,
+      "step": 397200
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 2.335432368798674e-06,
+      "loss": 0.8156,
+      "step": 397300
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 2.3284665431393586e-06,
+      "loss": 0.8103,
+      "step": 397400
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 2.321500717480043e-06,
+      "loss": 0.8506,
+      "step": 397500
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 2.3145348918207276e-06,
+      "loss": 0.8415,
+      "step": 397600
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 2.307569066161412e-06,
+      "loss": 0.8264,
+      "step": 397700
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 2.300603240502097e-06,
+      "loss": 0.8058,
+      "step": 397800
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 2.2936374148427816e-06,
+      "loss": 0.8303,
+      "step": 397900
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 2.286671589183466e-06,
+      "loss": 0.8256,
+      "step": 398000
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.2797057635241506e-06,
+      "loss": 0.8057,
+      "step": 398100
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.2727399378648352e-06,
+      "loss": 0.8375,
+      "step": 398200
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.2657741122055195e-06,
+      "loss": 0.8279,
+      "step": 398300
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.258808286546204e-06,
+      "loss": 0.8181,
+      "step": 398400
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.2518424608868893e-06,
+      "loss": 0.8052,
+      "step": 398500
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.2448766352275736e-06,
+      "loss": 0.8453,
+      "step": 398600
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.2379108095682582e-06,
+      "loss": 0.8587,
+      "step": 398700
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.230944983908943e-06,
+      "loss": 0.8429,
+      "step": 398800
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.223979158249627e-06,
+      "loss": 0.8207,
+      "step": 398900
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.217082990846905e-06,
+      "loss": 0.8515,
+      "step": 399000
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.2101171651875896e-06,
+      "loss": 0.8403,
+      "step": 399100
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.2031513395282742e-06,
+      "loss": 0.8369,
+      "step": 399200
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.196185513868959e-06,
+      "loss": 0.841,
+      "step": 399300
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.1892196882096436e-06,
+      "loss": 0.8074,
+      "step": 399400
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.1822538625503283e-06,
+      "loss": 0.7971,
+      "step": 399500
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.1752880368910126e-06,
+      "loss": 0.8242,
+      "step": 399600
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.1683222112316972e-06,
+      "loss": 0.8233,
+      "step": 399700
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.161356385572382e-06,
+      "loss": 0.8398,
+      "step": 399800
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.1543905599130666e-06,
+      "loss": 0.8336,
+      "step": 399900
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.1474247342537513e-06,
+      "loss": 0.8367,
+      "step": 400000
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.140458908594436e-06,
+      "loss": 0.8471,
+      "step": 400100
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.1334930829351202e-06,
+      "loss": 0.8388,
+      "step": 400200
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.126527257275805e-06,
+      "loss": 0.8394,
+      "step": 400300
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.119561431616489e-06,
+      "loss": 0.8561,
+      "step": 400400
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.1125956059571743e-06,
+      "loss": 0.8279,
+      "step": 400500
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.105629780297859e-06,
+      "loss": 0.818,
+      "step": 400600
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.0986639546385432e-06,
+      "loss": 0.8218,
+      "step": 400700
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.091698128979228e-06,
+      "loss": 0.831,
+      "step": 400800
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.0847323033199126e-06,
+      "loss": 0.8279,
+      "step": 400900
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.077766477660597e-06,
+      "loss": 0.8396,
+      "step": 401000
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.070800652001282e-06,
+      "loss": 0.8516,
+      "step": 401100
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.0638348263419666e-06,
+      "loss": 0.8415,
+      "step": 401200
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.056869000682651e-06,
+      "loss": 0.8537,
+      "step": 401300
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.0499031750233356e-06,
+      "loss": 0.8165,
+      "step": 401400
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.0429373493640203e-06,
+      "loss": 0.8216,
+      "step": 401500
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.0359715237047045e-06,
+      "loss": 0.8369,
+      "step": 401600
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.0290056980453896e-06,
+      "loss": 0.8306,
+      "step": 401700
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.022039872386074e-06,
+      "loss": 0.8369,
+      "step": 401800
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.0150740467267586e-06,
+      "loss": 0.8312,
+      "step": 401900
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.0081082210674433e-06,
+      "loss": 0.8204,
+      "step": 402000
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.0011423954081275e-06,
+      "loss": 0.8472,
+      "step": 402100
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.994176569748812e-06,
+      "loss": 0.8314,
+      "step": 402200
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.9872107440894973e-06,
+      "loss": 0.8275,
+      "step": 402300
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.9802449184301816e-06,
+      "loss": 0.8527,
+      "step": 402400
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.9733487510274597e-06,
+      "loss": 0.8428,
+      "step": 402500
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.966382925368144e-06,
+      "loss": 0.8339,
+      "step": 402600
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.9594170997088286e-06,
+      "loss": 0.8574,
+      "step": 402700
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.9524512740495133e-06,
+      "loss": 0.8299,
+      "step": 402800
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.9454854483901976e-06,
+      "loss": 0.8344,
+      "step": 402900
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.9385196227308823e-06,
+      "loss": 0.7946,
+      "step": 403000
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.931553797071567e-06,
+      "loss": 0.806,
+      "step": 403100
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.9245879714122516e-06,
+      "loss": 0.8634,
+      "step": 403200
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.9176221457529363e-06,
+      "loss": 0.8452,
+      "step": 403300
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.9106563200936206e-06,
+      "loss": 0.8458,
+      "step": 403400
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.9036904944343053e-06,
+      "loss": 0.8422,
+      "step": 403500
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.8967246687749901e-06,
+      "loss": 0.8609,
+      "step": 403600
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.8897588431156744e-06,
+      "loss": 0.8105,
+      "step": 403700
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.882793017456359e-06,
+      "loss": 0.8159,
+      "step": 403800
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.875827191797044e-06,
+      "loss": 0.8604,
+      "step": 403900
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.8688613661377285e-06,
+      "loss": 0.8309,
+      "step": 404000
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.861895540478413e-06,
+      "loss": 0.8026,
+      "step": 404100
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.8549297148190976e-06,
+      "loss": 0.8271,
+      "step": 404200
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.847963889159782e-06,
+      "loss": 0.837,
+      "step": 404300
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.8409980635004668e-06,
+      "loss": 0.824,
+      "step": 404400
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.8340322378411514e-06,
+      "loss": 0.8557,
+      "step": 404500
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.827066412181836e-06,
+      "loss": 0.8109,
+      "step": 404600
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.8201005865225206e-06,
+      "loss": 0.8213,
+      "step": 404700
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.8131347608632053e-06,
+      "loss": 0.8316,
+      "step": 404800
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.8061689352038898e-06,
+      "loss": 0.8072,
+      "step": 404900
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.7992031095445742e-06,
+      "loss": 0.8013,
+      "step": 405000
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.7923069421418521e-06,
+      "loss": 0.8226,
+      "step": 405100
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.7853411164825366e-06,
+      "loss": 0.8322,
+      "step": 405200
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.7783752908232215e-06,
+      "loss": 0.8296,
+      "step": 405300
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.771409465163906e-06,
+      "loss": 0.8525,
+      "step": 405400
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.7644436395045905e-06,
+      "loss": 0.8294,
+      "step": 405500
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.7574778138452751e-06,
+      "loss": 0.8088,
+      "step": 405600
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.7505119881859598e-06,
+      "loss": 0.8109,
+      "step": 405700
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.7435461625266443e-06,
+      "loss": 0.836,
+      "step": 405800
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.7365803368673288e-06,
+      "loss": 0.8353,
+      "step": 405900
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.7296145112080137e-06,
+      "loss": 0.8345,
+      "step": 406000
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.7226486855486981e-06,
+      "loss": 0.8032,
+      "step": 406100
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.7156828598893826e-06,
+      "loss": 0.8448,
+      "step": 406200
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.7087170342300673e-06,
+      "loss": 0.8559,
+      "step": 406300
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.701751208570752e-06,
+      "loss": 0.8094,
+      "step": 406400
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.6947853829114364e-06,
+      "loss": 0.8585,
+      "step": 406500
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.6878195572521211e-06,
+      "loss": 0.8252,
+      "step": 406600
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.6808537315928058e-06,
+      "loss": 0.7963,
+      "step": 406700
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.6738879059334903e-06,
+      "loss": 0.8484,
+      "step": 406800
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.666922080274175e-06,
+      "loss": 0.8171,
+      "step": 406900
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.6599562546148594e-06,
+      "loss": 0.8206,
+      "step": 407000
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.6529904289555441e-06,
+      "loss": 0.8143,
+      "step": 407100
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.6460942615528218e-06,
+      "loss": 0.8433,
+      "step": 407200
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.6391284358935065e-06,
+      "loss": 0.8295,
+      "step": 407300
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.6321626102341912e-06,
+      "loss": 0.8231,
+      "step": 407400
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.6251967845748757e-06,
+      "loss": 0.8251,
+      "step": 407500
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.6182309589155601e-06,
+      "loss": 0.8477,
+      "step": 407600
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.611265133256245e-06,
+      "loss": 0.8329,
+      "step": 407700
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.6042993075969295e-06,
+      "loss": 0.8176,
+      "step": 407800
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.597333481937614e-06,
+      "loss": 0.8167,
+      "step": 407900
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.5903676562782989e-06,
+      "loss": 0.8161,
+      "step": 408000
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.5834018306189833e-06,
+      "loss": 0.805,
+      "step": 408100
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.5764360049596678e-06,
+      "loss": 0.8254,
+      "step": 408200
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.5694701793003525e-06,
+      "loss": 0.847,
+      "step": 408300
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.5625043536410372e-06,
+      "loss": 0.7958,
+      "step": 408400
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.5555385279817216e-06,
+      "loss": 0.8541,
+      "step": 408500
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.5485727023224063e-06,
+      "loss": 0.8399,
+      "step": 408600
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.541606876663091e-06,
+      "loss": 0.8413,
+      "step": 408700
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.5346410510037755e-06,
+      "loss": 0.8071,
+      "step": 408800
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.5276752253444602e-06,
+      "loss": 0.8483,
+      "step": 408900
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.5207093996851446e-06,
+      "loss": 0.827,
+      "step": 409000
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.5137435740258293e-06,
+      "loss": 0.8573,
+      "step": 409100
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.506777748366514e-06,
+      "loss": 0.8162,
+      "step": 409200
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.4998119227071985e-06,
+      "loss": 0.8183,
+      "step": 409300
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.4928460970478832e-06,
+      "loss": 0.847,
+      "step": 409400
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.4858802713885678e-06,
+      "loss": 0.8509,
+      "step": 409500
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.4789144457292523e-06,
+      "loss": 0.8433,
+      "step": 409600
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.4719486200699368e-06,
+      "loss": 0.8343,
+      "step": 409700
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.4649827944106217e-06,
+      "loss": 0.8438,
+      "step": 409800
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.4580169687513061e-06,
+      "loss": 0.8591,
+      "step": 409900
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.4510511430919906e-06,
+      "loss": 0.8236,
+      "step": 410000
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.4440853174326755e-06,
+      "loss": 0.8381,
+      "step": 410100
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.43711949177336e-06,
+      "loss": 0.8261,
+      "step": 410200
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.4301536661140445e-06,
+      "loss": 0.8465,
+      "step": 410300
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.4231878404547291e-06,
+      "loss": 0.8374,
+      "step": 410400
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.4162220147954138e-06,
+      "loss": 0.8262,
+      "step": 410500
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.4092561891360983e-06,
+      "loss": 0.8437,
+      "step": 410600
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.402290363476783e-06,
+      "loss": 0.801,
+      "step": 410700
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.3953245378174677e-06,
+      "loss": 0.8187,
+      "step": 410800
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.3884283704147454e-06,
+      "loss": 0.827,
+      "step": 410900
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.3814625447554298e-06,
+      "loss": 0.816,
+      "step": 411000
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.3744967190961145e-06,
+      "loss": 0.8482,
+      "step": 411100
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.367530893436799e-06,
+      "loss": 0.8296,
+      "step": 411200
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.3605650677774837e-06,
+      "loss": 0.8425,
+      "step": 411300
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.3535992421181683e-06,
+      "loss": 0.8138,
+      "step": 411400
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.3466334164588528e-06,
+      "loss": 0.8248,
+      "step": 411500
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.3396675907995375e-06,
+      "loss": 0.8417,
+      "step": 411600
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.332701765140222e-06,
+      "loss": 0.845,
+      "step": 411700
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.3257359394809067e-06,
+      "loss": 0.8463,
+      "step": 411800
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.3187701138215913e-06,
+      "loss": 0.8482,
+      "step": 411900
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.3118042881622758e-06,
+      "loss": 0.8133,
+      "step": 412000
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.3048384625029605e-06,
+      "loss": 0.8457,
+      "step": 412100
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.2978726368436452e-06,
+      "loss": 0.7992,
+      "step": 412200
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.2909068111843297e-06,
+      "loss": 0.8254,
+      "step": 412300
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.2839409855250141e-06,
+      "loss": 0.8398,
+      "step": 412400
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.276975159865699e-06,
+      "loss": 0.8129,
+      "step": 412500
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.2700093342063835e-06,
+      "loss": 0.8211,
+      "step": 412600
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.263043508547068e-06,
+      "loss": 0.8203,
+      "step": 412700
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.2560776828877529e-06,
+      "loss": 0.8585,
+      "step": 412800
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.2491815154850303e-06,
+      "loss": 0.8181,
+      "step": 412900
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.242215689825715e-06,
+      "loss": 0.8559,
+      "step": 413000
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.2352498641663997e-06,
+      "loss": 0.8108,
+      "step": 413100
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.2282840385070842e-06,
+      "loss": 0.8364,
+      "step": 413200
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.2213182128477689e-06,
+      "loss": 0.8198,
+      "step": 413300
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.2143523871884535e-06,
+      "loss": 0.8501,
+      "step": 413400
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.207386561529138e-06,
+      "loss": 0.8246,
+      "step": 413500
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.2004207358698227e-06,
+      "loss": 0.79,
+      "step": 413600
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.1934549102105074e-06,
+      "loss": 0.8225,
+      "step": 413700
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.1864890845511919e-06,
+      "loss": 0.8366,
+      "step": 413800
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.1795232588918765e-06,
+      "loss": 0.8177,
+      "step": 413900
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.172557433232561e-06,
+      "loss": 0.8278,
+      "step": 414000
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.1655916075732457e-06,
+      "loss": 0.8216,
+      "step": 414100
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.1586257819139304e-06,
+      "loss": 0.7972,
+      "step": 414200
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.1516599562546149e-06,
+      "loss": 0.8423,
+      "step": 414300
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.1446941305952995e-06,
+      "loss": 0.8451,
+      "step": 414400
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.1377283049359842e-06,
+      "loss": 0.8252,
+      "step": 414500
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.1307624792766687e-06,
+      "loss": 0.8351,
+      "step": 414600
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.1237966536173532e-06,
+      "loss": 0.854,
+      "step": 414700
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.116830827958038e-06,
+      "loss": 0.8025,
+      "step": 414800
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.1098650022987225e-06,
+      "loss": 0.8122,
+      "step": 414900
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.102899176639407e-06,
+      "loss": 0.8385,
+      "step": 415000
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.096003009236685e-06,
+      "loss": 0.8497,
+      "step": 415100
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.0890371835773694e-06,
+      "loss": 0.8524,
+      "step": 415200
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.082071357918054e-06,
+      "loss": 0.8413,
+      "step": 415300
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.0751055322587387e-06,
+      "loss": 0.8408,
+      "step": 415400
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.0681397065994232e-06,
+      "loss": 0.8346,
+      "step": 415500
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.061173880940108e-06,
+      "loss": 0.8443,
+      "step": 415600
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.0542080552807926e-06,
+      "loss": 0.8039,
+      "step": 415700
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.047242229621477e-06,
+      "loss": 0.8309,
+      "step": 415800
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.0402764039621617e-06,
+      "loss": 0.8151,
+      "step": 415900
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.0333105783028462e-06,
+      "loss": 0.8235,
+      "step": 416000
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.026344752643531e-06,
+      "loss": 0.8586,
+      "step": 416100
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.0193789269842156e-06,
+      "loss": 0.8187,
+      "step": 416200
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.0124131013249e-06,
+      "loss": 0.8205,
+      "step": 416300
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.0054472756655847e-06,
+      "loss": 0.8414,
+      "step": 416400
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 9.984814500062692e-07,
+      "loss": 0.8385,
+      "step": 416500
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 9.915156243469539e-07,
+      "loss": 0.8484,
+      "step": 416600
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 9.845497986876384e-07,
+      "loss": 0.8357,
+      "step": 416700
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 9.77583973028323e-07,
+      "loss": 0.8065,
+      "step": 416800
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 9.706181473690077e-07,
+      "loss": 0.8406,
+      "step": 416900
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 9.636523217096922e-07,
+      "loss": 0.8033,
+      "step": 417000
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 9.566864960503769e-07,
+      "loss": 0.7802,
+      "step": 417100
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 9.497206703910616e-07,
+      "loss": 0.8161,
+      "step": 417200
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 9.42754844731746e-07,
+      "loss": 0.8294,
+      "step": 417300
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 9.357890190724307e-07,
+      "loss": 0.8428,
+      "step": 417400
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 9.288231934131153e-07,
+      "loss": 0.8358,
+      "step": 417500
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 9.218573677537999e-07,
+      "loss": 0.8485,
+      "step": 417600
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 9.148915420944845e-07,
+      "loss": 0.8188,
+      "step": 417700
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 9.07925716435169e-07,
+      "loss": 0.8022,
+      "step": 417800
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 9.009598907758537e-07,
+      "loss": 0.8418,
+      "step": 417900
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 8.939940651165383e-07,
+      "loss": 0.8339,
+      "step": 418000
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 8.870282394572229e-07,
+      "loss": 0.8511,
+      "step": 418100
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 8.800624137979074e-07,
+      "loss": 0.8283,
+      "step": 418200
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 8.730965881385921e-07,
+      "loss": 0.7943,
+      "step": 418300
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 8.661307624792766e-07,
+      "loss": 0.8262,
+      "step": 418400
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 8.591649368199613e-07,
+      "loss": 0.8273,
+      "step": 418500
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 8.52268769417239e-07,
+      "loss": 0.8275,
+      "step": 418600
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 8.453029437579237e-07,
+      "loss": 0.8275,
+      "step": 418700
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 8.383371180986082e-07,
+      "loss": 0.857,
+      "step": 418800
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 8.313712924392928e-07,
+      "loss": 0.8198,
+      "step": 418900
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 8.244054667799775e-07,
+      "loss": 0.8198,
+      "step": 419000
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 8.174396411206621e-07,
+      "loss": 0.8533,
+      "step": 419100
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 8.104738154613467e-07,
+      "loss": 0.8134,
+      "step": 419200
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 8.035079898020312e-07,
+      "loss": 0.837,
+      "step": 419300
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 7.965421641427159e-07,
+      "loss": 0.8556,
+      "step": 419400
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 7.895763384834004e-07,
+      "loss": 0.8463,
+      "step": 419500
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 7.826105128240851e-07,
+      "loss": 0.856,
+      "step": 419600
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 7.756446871647698e-07,
+      "loss": 0.8344,
+      "step": 419700
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 7.686788615054542e-07,
+      "loss": 0.8052,
+      "step": 419800
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 7.617130358461389e-07,
+      "loss": 0.8433,
+      "step": 419900
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 7.547472101868235e-07,
+      "loss": 0.8475,
+      "step": 420000
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 7.477813845275081e-07,
+      "loss": 0.8058,
+      "step": 420100
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 7.408155588681926e-07,
+      "loss": 0.8244,
+      "step": 420200
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 7.338497332088772e-07,
+      "loss": 0.8162,
+      "step": 420300
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 7.268839075495619e-07,
+      "loss": 0.8354,
+      "step": 420400
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 7.199180818902465e-07,
+      "loss": 0.8327,
+      "step": 420500
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 7.129522562309311e-07,
+      "loss": 0.8383,
+      "step": 420600
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 7.059864305716156e-07,
+      "loss": 0.8278,
+      "step": 420700
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 6.990206049123003e-07,
+      "loss": 0.8357,
+      "step": 420800
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 6.920547792529848e-07,
+      "loss": 0.8503,
+      "step": 420900
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 6.850889535936695e-07,
+      "loss": 0.8103,
+      "step": 421000
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 6.781231279343542e-07,
+      "loss": 0.8206,
+      "step": 421100
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 6.711573022750386e-07,
+      "loss": 0.8256,
+      "step": 421200
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 6.641914766157233e-07,
+      "loss": 0.8429,
+      "step": 421300
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 6.572256509564079e-07,
+      "loss": 0.8381,
+      "step": 421400
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 6.502598252970925e-07,
+      "loss": 0.8227,
+      "step": 421500
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 6.43293999637777e-07,
+      "loss": 0.8342,
+      "step": 421600
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 6.363281739784617e-07,
+      "loss": 0.8107,
+      "step": 421700
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 6.293623483191463e-07,
+      "loss": 0.8456,
+      "step": 421800
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 6.223965226598309e-07,
+      "loss": 0.8257,
+      "step": 421900
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 6.154306970005156e-07,
+      "loss": 0.8303,
+      "step": 422000
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 6.084648713412e-07,
+      "loss": 0.8303,
+      "step": 422100
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 6.014990456818847e-07,
+      "loss": 0.8193,
+      "step": 422200
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 5.946028782791624e-07,
+      "loss": 0.8381,
+      "step": 422300
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 5.876370526198471e-07,
+      "loss": 0.8263,
+      "step": 422400
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 5.806712269605317e-07,
+      "loss": 0.8478,
+      "step": 422500
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 5.737054013012163e-07,
+      "loss": 0.8227,
+      "step": 422600
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 5.667395756419008e-07,
+      "loss": 0.8371,
+      "step": 422700
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 5.597737499825854e-07,
+      "loss": 0.8067,
+      "step": 422800
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 5.5280792432327e-07,
+      "loss": 0.8135,
+      "step": 422900
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 5.458420986639547e-07,
+      "loss": 0.8414,
+      "step": 423000
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 5.388762730046393e-07,
+      "loss": 0.8406,
+      "step": 423100
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 5.319104473453238e-07,
+      "loss": 0.8134,
+      "step": 423200
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 5.249446216860085e-07,
+      "loss": 0.8446,
+      "step": 423300
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 5.17978796026693e-07,
+      "loss": 0.8345,
+      "step": 423400
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 5.110129703673777e-07,
+      "loss": 0.8047,
+      "step": 423500
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 5.040471447080622e-07,
+      "loss": 0.8492,
+      "step": 423600
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.970813190487468e-07,
+      "loss": 0.8023,
+      "step": 423700
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.901154933894315e-07,
+      "loss": 0.8161,
+      "step": 423800
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.831496677301161e-07,
+      "loss": 0.8424,
+      "step": 423900
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.761838420708007e-07,
+      "loss": 0.836,
+      "step": 424000
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.6921801641148524e-07,
+      "loss": 0.8129,
+      "step": 424100
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.6225219075216987e-07,
+      "loss": 0.8521,
+      "step": 424200
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.552863650928545e-07,
+      "loss": 0.8312,
+      "step": 424300
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.483205394335391e-07,
+      "loss": 0.8064,
+      "step": 424400
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.4135471377422365e-07,
+      "loss": 0.8389,
+      "step": 424500
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.3438888811490823e-07,
+      "loss": 0.8164,
+      "step": 424600
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.2742306245559286e-07,
+      "loss": 0.8236,
+      "step": 424700
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.204572367962775e-07,
+      "loss": 0.8056,
+      "step": 424800
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.1349141113696207e-07,
+      "loss": 0.8357,
+      "step": 424900
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.065255854776467e-07,
+      "loss": 0.8209,
+      "step": 425000
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 3.995597598183313e-07,
+      "loss": 0.8222,
+      "step": 425100
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 3.9259393415901585e-07,
+      "loss": 0.8028,
+      "step": 425200
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 3.856281084997005e-07,
+      "loss": 0.8125,
+      "step": 425300
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 3.7873194109697824e-07,
+      "loss": 0.8536,
+      "step": 425400
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 3.7176611543766287e-07,
+      "loss": 0.8474,
+      "step": 425500
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 3.6480028977834744e-07,
+      "loss": 0.8179,
+      "step": 425600
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 3.57834464119032e-07,
+      "loss": 0.8267,
+      "step": 425700
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 3.5086863845971665e-07,
+      "loss": 0.8247,
+      "step": 425800
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 3.4390281280040123e-07,
+      "loss": 0.8374,
+      "step": 425900
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 3.3693698714108586e-07,
+      "loss": 0.7893,
+      "step": 426000
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 3.2997116148177044e-07,
+      "loss": 0.8325,
+      "step": 426100
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 3.2300533582245507e-07,
+      "loss": 0.8137,
+      "step": 426200
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 3.1603951016313964e-07,
+      "loss": 0.8398,
+      "step": 426300
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 3.090736845038242e-07,
+      "loss": 0.8326,
+      "step": 426400
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 3.0210785884450885e-07,
+      "loss": 0.8341,
+      "step": 426500
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 2.9514203318519343e-07,
+      "loss": 0.8363,
+      "step": 426600
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 2.8817620752587806e-07,
+      "loss": 0.825,
+      "step": 426700
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 2.812103818665627e-07,
+      "loss": 0.8365,
+      "step": 426800
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 2.7424455620724727e-07,
+      "loss": 0.8254,
+      "step": 426900
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 2.6727873054793184e-07,
+      "loss": 0.8007,
+      "step": 427000
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 2.603129048886165e-07,
+      "loss": 0.8189,
+      "step": 427100
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 2.5334707922930105e-07,
+      "loss": 0.8243,
+      "step": 427200
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 2.4638125356998563e-07,
+      "loss": 0.8188,
+      "step": 427300
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 2.3941542791067026e-07,
+      "loss": 0.8351,
+      "step": 427400
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 2.32519260507948e-07,
+      "loss": 0.8359,
+      "step": 427500
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 2.2555343484863262e-07,
+      "loss": 0.8314,
+      "step": 427600
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 2.1858760918931722e-07,
+      "loss": 0.802,
+      "step": 427700
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 2.1162178353000182e-07,
+      "loss": 0.8333,
+      "step": 427800
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 2.0465595787068643e-07,
+      "loss": 0.8329,
+      "step": 427900
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.97690132211371e-07,
+      "loss": 0.805,
+      "step": 428000
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.907243065520556e-07,
+      "loss": 0.8559,
+      "step": 428100
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.8375848089274024e-07,
+      "loss": 0.836,
+      "step": 428200
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.7679265523342482e-07,
+      "loss": 0.8504,
+      "step": 428300
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.6982682957410942e-07,
+      "loss": 0.8513,
+      "step": 428400
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.6286100391479402e-07,
+      "loss": 0.8228,
+      "step": 428500
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.5589517825547863e-07,
+      "loss": 0.8337,
+      "step": 428600
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.4892935259616323e-07,
+      "loss": 0.8333,
+      "step": 428700
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.419635269368478e-07,
+      "loss": 0.816,
+      "step": 428800
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.3499770127753244e-07,
+      "loss": 0.8424,
+      "step": 428900
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.2803187561821702e-07,
+      "loss": 0.8313,
+      "step": 429000
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.2106604995890162e-07,
+      "loss": 0.8185,
+      "step": 429100
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.1410022429958622e-07,
+      "loss": 0.811,
+      "step": 429200
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.0713439864027084e-07,
+      "loss": 0.8382,
+      "step": 429300
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.0016857298095543e-07,
+      "loss": 0.8313,
+      "step": 429400
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 9.320274732164004e-08,
+      "loss": 0.8064,
+      "step": 429500
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 8.623692166232464e-08,
+      "loss": 0.8072,
+      "step": 429600
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 7.927109600300923e-08,
+      "loss": 0.8101,
+      "step": 429700
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 7.230527034369385e-08,
+      "loss": 0.8358,
+      "step": 429800
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 6.533944468437844e-08,
+      "loss": 0.8318,
+      "step": 429900
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 5.837361902506304e-08,
+      "loss": 0.8202,
+      "step": 430000
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 5.1407793365747645e-08,
+      "loss": 0.8055,
+      "step": 430100
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.444196770643224e-08,
+      "loss": 0.8096,
+      "step": 430200
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 3.7476142047116846e-08,
+      "loss": 0.8211,
+      "step": 430300
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.051031638780144e-08,
+      "loss": 0.8008,
+      "step": 430400
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 2.3544490728486047e-08,
+      "loss": 0.8318,
+      "step": 430500
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.657866506917065e-08,
+      "loss": 0.8358,
+      "step": 430600
+    },
+    {
+      "epoch": 6.0,
+      "eval_gen_len": 20.0,
+      "eval_loss": 1.199650764465332,
+      "eval_rouge1": 12.5153,
+      "eval_rouge2": 3.778,
+      "eval_rougeL": 12.0382,
+      "eval_rougeLsum": 12.1332,
+      "eval_runtime": 1506.4096,
+      "eval_samples_per_second": 8.874,
+      "eval_steps_per_second": 2.219,
+      "step": 430674
+    },
+    {
+      "epoch": 6.0,
+      "step": 430674,
+      "total_flos": 4.0669778872979374e+18,
+      "train_loss": 1.0701684290170863,
+      "train_runtime": 175298.0292,
+      "train_samples_per_second": 9.827,
+      "train_steps_per_second": 2.457
+    }
+  ],
+  "max_steps": 430674,
+  "num_train_epochs": 6,
+  "total_flos": 4.0669778872979374e+18,
+  "trial_name": null,
+  "trial_params": null
+}