diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,18970 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.0,
+  "global_step": 315900,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9984172206394435e-06,
+      "loss": 2.7157,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.996834441278886e-06,
+      "loss": 2.0977,
+      "step": 200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.995251661918329e-06,
+      "loss": 1.9549,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.993668882557772e-06,
+      "loss": 1.9347,
+      "step": 400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.992086103197215e-06,
+      "loss": 1.8296,
+      "step": 500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.990503323836657e-06,
+      "loss": 1.8133,
+      "step": 600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9889205444761e-06,
+      "loss": 1.7539,
+      "step": 700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.987337765115544e-06,
+      "loss": 1.7711,
+      "step": 800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.985754985754986e-06,
+      "loss": 1.7314,
+      "step": 900
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.984172206394429e-06,
+      "loss": 1.7503,
+      "step": 1000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.982589427033872e-06,
+      "loss": 1.722,
+      "step": 1100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.981006647673315e-06,
+      "loss": 1.719,
+      "step": 1200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9794238683127575e-06,
+      "loss": 1.7309,
+      "step": 1300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9778410889522005e-06,
+      "loss": 1.7112,
+      "step": 1400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.976258309591644e-06,
+      "loss": 1.6686,
+      "step": 1500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.974675530231086e-06,
+      "loss": 1.6855,
+      "step": 1600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.973092750870529e-06,
+      "loss": 1.6698,
+      "step": 1700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.971509971509972e-06,
+      "loss": 1.6515,
+      "step": 1800
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.969927192149415e-06,
+      "loss": 1.6495,
+      "step": 1900
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.968344412788857e-06,
+      "loss": 1.6818,
+      "step": 2000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9667616334283e-06,
+      "loss": 1.6612,
+      "step": 2100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.965178854067743e-06,
+      "loss": 1.6644,
+      "step": 2200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.963596074707186e-06,
+      "loss": 1.6534,
+      "step": 2300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.962013295346629e-06,
+      "loss": 1.6291,
+      "step": 2400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.960430515986072e-06,
+      "loss": 1.6356,
+      "step": 2500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.958847736625515e-06,
+      "loss": 1.6253,
+      "step": 2600
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.957264957264958e-06,
+      "loss": 1.6522,
+      "step": 2700
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.955682177904401e-06,
+      "loss": 1.6269,
+      "step": 2800
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.954099398543844e-06,
+      "loss": 1.5901,
+      "step": 2900
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.952516619183286e-06,
+      "loss": 1.6162,
+      "step": 3000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.950933839822729e-06,
+      "loss": 1.6118,
+      "step": 3100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.949351060462172e-06,
+      "loss": 1.6182,
+      "step": 3200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.947768281101615e-06,
+      "loss": 1.6465,
+      "step": 3300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.946185501741057e-06,
+      "loss": 1.5952,
+      "step": 3400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9446027223805e-06,
+      "loss": 1.6144,
+      "step": 3500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.943019943019943e-06,
+      "loss": 1.599,
+      "step": 3600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.941437163659386e-06,
+      "loss": 1.5459,
+      "step": 3700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.939854384298829e-06,
+      "loss": 1.6284,
+      "step": 3800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.938271604938272e-06,
+      "loss": 1.6161,
+      "step": 3900
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.936688825577715e-06,
+      "loss": 1.5846,
+      "step": 4000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.935106046217158e-06,
+      "loss": 1.6138,
+      "step": 4100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.933523266856601e-06,
+      "loss": 1.5781,
+      "step": 4200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.931940487496044e-06,
+      "loss": 1.6329,
+      "step": 4300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.930357708135486e-06,
+      "loss": 1.6064,
+      "step": 4400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.928774928774929e-06,
+      "loss": 1.5841,
+      "step": 4500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.927192149414372e-06,
+      "loss": 1.5837,
+      "step": 4600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.925609370053815e-06,
+      "loss": 1.5881,
+      "step": 4700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.924026590693257e-06,
+      "loss": 1.5882,
+      "step": 4800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9224438113327004e-06,
+      "loss": 1.5677,
+      "step": 4900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9208610319721435e-06,
+      "loss": 1.5944,
+      "step": 5000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9192782526115865e-06,
+      "loss": 1.5942,
+      "step": 5100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.917695473251029e-06,
+      "loss": 1.5794,
+      "step": 5200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.916112693890472e-06,
+      "loss": 1.5736,
+      "step": 5300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.914529914529915e-06,
+      "loss": 1.5591,
+      "step": 5400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.912947135169358e-06,
+      "loss": 1.5977,
+      "step": 5500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.911364355808801e-06,
+      "loss": 1.5528,
+      "step": 5600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.909781576448244e-06,
+      "loss": 1.5853,
+      "step": 5700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.908198797087686e-06,
+      "loss": 1.5599,
+      "step": 5800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.906616017727129e-06,
+      "loss": 1.5539,
+      "step": 5900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.905033238366572e-06,
+      "loss": 1.5947,
+      "step": 6000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.903450459006015e-06,
+      "loss": 1.5608,
+      "step": 6100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9018676796454575e-06,
+      "loss": 1.5595,
+      "step": 6200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9002849002849006e-06,
+      "loss": 1.5548,
+      "step": 6300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.898702120924344e-06,
+      "loss": 1.538,
+      "step": 6400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.897119341563787e-06,
+      "loss": 1.5576,
+      "step": 6500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.895536562203229e-06,
+      "loss": 1.5606,
+      "step": 6600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.893953782842672e-06,
+      "loss": 1.5404,
+      "step": 6700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.892371003482115e-06,
+      "loss": 1.5638,
+      "step": 6800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.890788224121558e-06,
+      "loss": 1.543,
+      "step": 6900
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.889205444761e-06,
+      "loss": 1.5597,
+      "step": 7000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.887622665400444e-06,
+      "loss": 1.5368,
+      "step": 7100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.886039886039886e-06,
+      "loss": 1.5206,
+      "step": 7200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.884457106679329e-06,
+      "loss": 1.5332,
+      "step": 7300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.882874327318772e-06,
+      "loss": 1.5274,
+      "step": 7400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8812915479582154e-06,
+      "loss": 1.5499,
+      "step": 7500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.879708768597658e-06,
+      "loss": 1.5177,
+      "step": 7600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.878125989237101e-06,
+      "loss": 1.535,
+      "step": 7700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.876543209876544e-06,
+      "loss": 1.5256,
+      "step": 7800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.874960430515987e-06,
+      "loss": 1.5351,
+      "step": 7900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.873377651155429e-06,
+      "loss": 1.5399,
+      "step": 8000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.871794871794872e-06,
+      "loss": 1.5561,
+      "step": 8100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.870212092434315e-06,
+      "loss": 1.5573,
+      "step": 8200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.868629313073758e-06,
+      "loss": 1.5409,
+      "step": 8300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8670465337132e-06,
+      "loss": 1.5533,
+      "step": 8400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.865463754352643e-06,
+      "loss": 1.5604,
+      "step": 8500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8638809749920864e-06,
+      "loss": 1.5658,
+      "step": 8600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8622981956315295e-06,
+      "loss": 1.5564,
+      "step": 8700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8607154162709725e-06,
+      "loss": 1.5404,
+      "step": 8800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.859132636910416e-06,
+      "loss": 1.5486,
+      "step": 8900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.857549857549858e-06,
+      "loss": 1.5597,
+      "step": 9000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.855967078189301e-06,
+      "loss": 1.5589,
+      "step": 9100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.854384298828744e-06,
+      "loss": 1.5272,
+      "step": 9200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.852801519468187e-06,
+      "loss": 1.5532,
+      "step": 9300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.851218740107629e-06,
+      "loss": 1.5287,
+      "step": 9400
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.849635960747072e-06,
+      "loss": 1.5108,
+      "step": 9500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.848053181386515e-06,
+      "loss": 1.528,
+      "step": 9600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.846470402025958e-06,
+      "loss": 1.5438,
+      "step": 9700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8448876226654005e-06,
+      "loss": 1.5515,
+      "step": 9800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8433048433048435e-06,
+      "loss": 1.5165,
+      "step": 9900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8417220639442866e-06,
+      "loss": 1.5248,
+      "step": 10000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.840139284583729e-06,
+      "loss": 1.5513,
+      "step": 10100
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.838556505223173e-06,
+      "loss": 1.5307,
+      "step": 10200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.836973725862616e-06,
+      "loss": 1.5391,
+      "step": 10300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.835390946502058e-06,
+      "loss": 1.5345,
+      "step": 10400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.833808167141501e-06,
+      "loss": 1.5431,
+      "step": 10500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.832225387780944e-06,
+      "loss": 1.5306,
+      "step": 10600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.830642608420387e-06,
+      "loss": 1.5027,
+      "step": 10700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.829059829059829e-06,
+      "loss": 1.5342,
+      "step": 10800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.827477049699272e-06,
+      "loss": 1.5243,
+      "step": 10900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.825894270338715e-06,
+      "loss": 1.5012,
+      "step": 11000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.824311490978158e-06,
+      "loss": 1.5478,
+      "step": 11100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.822728711617601e-06,
+      "loss": 1.5242,
+      "step": 11200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.821145932257044e-06,
+      "loss": 1.5134,
+      "step": 11300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.819563152896487e-06,
+      "loss": 1.4938,
+      "step": 11400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.817980373535929e-06,
+      "loss": 1.5079,
+      "step": 11500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.816397594175372e-06,
+      "loss": 1.5136,
+      "step": 11600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.814814814814815e-06,
+      "loss": 1.5143,
+      "step": 11700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.813232035454258e-06,
+      "loss": 1.5187,
+      "step": 11800
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.811649256093701e-06,
+      "loss": 1.5127,
+      "step": 11900
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.810066476733144e-06,
+      "loss": 1.5269,
+      "step": 12000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.808483697372587e-06,
+      "loss": 1.4965,
+      "step": 12100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.806900918012029e-06,
+      "loss": 1.531,
+      "step": 12200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8053181386514724e-06,
+      "loss": 1.526,
+      "step": 12300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8037353592909155e-06,
+      "loss": 1.5332,
+      "step": 12400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8021525799303585e-06,
+      "loss": 1.5157,
+      "step": 12500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.800569800569801e-06,
+      "loss": 1.4796,
+      "step": 12600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.798987021209244e-06,
+      "loss": 1.52,
+      "step": 12700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.797404241848687e-06,
+      "loss": 1.5035,
+      "step": 12800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.795821462488129e-06,
+      "loss": 1.5278,
+      "step": 12900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.794238683127572e-06,
+      "loss": 1.4903,
+      "step": 13000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.792655903767015e-06,
+      "loss": 1.5122,
+      "step": 13100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.791073124406458e-06,
+      "loss": 1.537,
+      "step": 13200
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.789490345045901e-06,
+      "loss": 1.5129,
+      "step": 13300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.787907565685344e-06,
+      "loss": 1.5047,
+      "step": 13400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.786324786324787e-06,
+      "loss": 1.495,
+      "step": 13500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7847420069642295e-06,
+      "loss": 1.5288,
+      "step": 13600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7831592276036726e-06,
+      "loss": 1.494,
+      "step": 13700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.781576448243116e-06,
+      "loss": 1.4887,
+      "step": 13800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.779993668882558e-06,
+      "loss": 1.4916,
+      "step": 13900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.778410889522001e-06,
+      "loss": 1.4996,
+      "step": 14000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.776828110161444e-06,
+      "loss": 1.484,
+      "step": 14100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.775245330800887e-06,
+      "loss": 1.4899,
+      "step": 14200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.773662551440329e-06,
+      "loss": 1.5139,
+      "step": 14300
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.772079772079772e-06,
+      "loss": 1.4931,
+      "step": 14400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.770496992719215e-06,
+      "loss": 1.4975,
+      "step": 14500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.768914213358658e-06,
+      "loss": 1.5215,
+      "step": 14600
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7673314339981005e-06,
+      "loss": 1.5098,
+      "step": 14700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7657486546375435e-06,
+      "loss": 1.4932,
+      "step": 14800
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7641658752769874e-06,
+      "loss": 1.4865,
+      "step": 14900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.76258309591643e-06,
+      "loss": 1.5267,
+      "step": 15000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.761000316555873e-06,
+      "loss": 1.5013,
+      "step": 15100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.759417537195316e-06,
+      "loss": 1.5071,
+      "step": 15200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.757834757834758e-06,
+      "loss": 1.4658,
+      "step": 15300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.756251978474201e-06,
+      "loss": 1.4965,
+      "step": 15400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.754669199113644e-06,
+      "loss": 1.4838,
+      "step": 15500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.753086419753087e-06,
+      "loss": 1.5135,
+      "step": 15600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.751503640392529e-06,
+      "loss": 1.4951,
+      "step": 15700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.749920861031972e-06,
+      "loss": 1.4844,
+      "step": 15800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.748338081671415e-06,
+      "loss": 1.5125,
+      "step": 15900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7467553023108584e-06,
+      "loss": 1.4884,
+      "step": 16000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.745172522950301e-06,
+      "loss": 1.4913,
+      "step": 16100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.743589743589744e-06,
+      "loss": 1.5169,
+      "step": 16200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.742006964229187e-06,
+      "loss": 1.4794,
+      "step": 16300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.74042418486863e-06,
+      "loss": 1.4748,
+      "step": 16400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.738841405508073e-06,
+      "loss": 1.4652,
+      "step": 16500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.737258626147516e-06,
+      "loss": 1.4735,
+      "step": 16600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.735675846786958e-06,
+      "loss": 1.5015,
+      "step": 16700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.734093067426401e-06,
+      "loss": 1.4786,
+      "step": 16800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.732510288065844e-06,
+      "loss": 1.5143,
+      "step": 16900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.730927508705287e-06,
+      "loss": 1.4592,
+      "step": 17000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.729344729344729e-06,
+      "loss": 1.5153,
+      "step": 17100
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7277619499841725e-06,
+      "loss": 1.4857,
+      "step": 17200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7261791706236155e-06,
+      "loss": 1.4842,
+      "step": 17300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.7245963912630586e-06,
+      "loss": 1.4627,
+      "step": 17400
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.723013611902501e-06,
+      "loss": 1.504,
+      "step": 17500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.721430832541944e-06,
+      "loss": 1.4901,
+      "step": 17600
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.719848053181387e-06,
+      "loss": 1.4816,
+      "step": 17700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.71826527382083e-06,
+      "loss": 1.5071,
+      "step": 17800
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.716682494460272e-06,
+      "loss": 1.4601,
+      "step": 17900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.715099715099716e-06,
+      "loss": 1.5007,
+      "step": 18000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.713516935739158e-06,
+      "loss": 1.4707,
+      "step": 18100
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.711934156378601e-06,
+      "loss": 1.4962,
+      "step": 18200
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.710351377018044e-06,
+      "loss": 1.4876,
+      "step": 18300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.708768597657487e-06,
+      "loss": 1.4863,
+      "step": 18400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7071858182969295e-06,
+      "loss": 1.4886,
+      "step": 18500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.705603038936373e-06,
+      "loss": 1.4909,
+      "step": 18600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.704020259575816e-06,
+      "loss": 1.4637,
+      "step": 18700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.702437480215259e-06,
+      "loss": 1.4712,
+      "step": 18800
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.700854700854701e-06,
+      "loss": 1.5079,
+      "step": 18900
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.699271921494144e-06,
+      "loss": 1.4609,
+      "step": 19000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.697689142133587e-06,
+      "loss": 1.4822,
+      "step": 19100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.69610636277303e-06,
+      "loss": 1.4737,
+      "step": 19200
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.694523583412472e-06,
+      "loss": 1.4596,
+      "step": 19300
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.692940804051915e-06,
+      "loss": 1.4944,
+      "step": 19400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.691358024691358e-06,
+      "loss": 1.4751,
+      "step": 19500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.689775245330801e-06,
+      "loss": 1.4876,
+      "step": 19600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.688192465970244e-06,
+      "loss": 1.4693,
+      "step": 19700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.6866096866096875e-06,
+      "loss": 1.4817,
+      "step": 19800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.68502690724913e-06,
+      "loss": 1.4886,
+      "step": 19900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.683444127888573e-06,
+      "loss": 1.4817,
+      "step": 20000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.681861348528016e-06,
+      "loss": 1.5018,
+      "step": 20100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.680278569167459e-06,
+      "loss": 1.4991,
+      "step": 20200
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.678695789806901e-06,
+      "loss": 1.4847,
+      "step": 20300
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.677113010446344e-06,
+      "loss": 1.4705,
+      "step": 20400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.675530231085787e-06,
+      "loss": 1.4771,
+      "step": 20500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.67394745172523e-06,
+      "loss": 1.4849,
+      "step": 20600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.672364672364672e-06,
+      "loss": 1.4823,
+      "step": 20700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.670781893004115e-06,
+      "loss": 1.4846,
+      "step": 20800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6691991136435585e-06,
+      "loss": 1.4653,
+      "step": 20900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6676163342830015e-06,
+      "loss": 1.4846,
+      "step": 21000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.666033554922444e-06,
+      "loss": 1.4902,
+      "step": 21100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.664450775561888e-06,
+      "loss": 1.4528,
+      "step": 21200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.66286799620133e-06,
+      "loss": 1.481,
+      "step": 21300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.661285216840773e-06,
+      "loss": 1.4948,
+      "step": 21400
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.659702437480216e-06,
+      "loss": 1.4664,
+      "step": 21500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.658119658119659e-06,
+      "loss": 1.4726,
+      "step": 21600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.656536878759101e-06,
+      "loss": 1.4724,
+      "step": 21700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.654954099398544e-06,
+      "loss": 1.4566,
+      "step": 21800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.653371320037987e-06,
+      "loss": 1.479,
+      "step": 21900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.65178854067743e-06,
+      "loss": 1.4506,
+      "step": 22000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6502057613168725e-06,
+      "loss": 1.4463,
+      "step": 22100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6486229819563155e-06,
+      "loss": 1.465,
+      "step": 22200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.647040202595759e-06,
+      "loss": 1.4401,
+      "step": 22300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.645457423235202e-06,
+      "loss": 1.4706,
+      "step": 22400
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.643874643874644e-06,
+      "loss": 1.4524,
+      "step": 22500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.642291864514087e-06,
+      "loss": 1.5014,
+      "step": 22600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.64070908515353e-06,
+      "loss": 1.494,
+      "step": 22700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.639126305792973e-06,
+      "loss": 1.4935,
+      "step": 22800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.637543526432416e-06,
+      "loss": 1.4658,
+      "step": 22900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.635960747071859e-06,
+      "loss": 1.4634,
+      "step": 23000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.634377967711301e-06,
+      "loss": 1.485,
+      "step": 23100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.632795188350744e-06,
+      "loss": 1.4486,
+      "step": 23200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.631212408990187e-06,
+      "loss": 1.4421,
+      "step": 23300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.62962962962963e-06,
+      "loss": 1.445,
+      "step": 23400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.628046850269073e-06,
+      "loss": 1.4629,
+      "step": 23500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.626464070908516e-06,
+      "loss": 1.4684,
+      "step": 23600
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.624881291547959e-06,
+      "loss": 1.4675,
+      "step": 23700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.623298512187401e-06,
+      "loss": 1.4695,
+      "step": 23800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.621715732826844e-06,
+      "loss": 1.4649,
+      "step": 23900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.620132953466287e-06,
+      "loss": 1.465,
+      "step": 24000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.61855017410573e-06,
+      "loss": 1.4431,
+      "step": 24100
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.616967394745172e-06,
+      "loss": 1.4568,
+      "step": 24200
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 1.4792,
+      "step": 24300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.613801836024059e-06,
+      "loss": 1.4436,
+      "step": 24400
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.612219056663501e-06,
+      "loss": 1.484,
+      "step": 24500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.6106362773029444e-06,
+      "loss": 1.4689,
+      "step": 24600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.6090534979423875e-06,
+      "loss": 1.4532,
+      "step": 24700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.6074707185818305e-06,
+      "loss": 1.4641,
+      "step": 24800
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.605887939221273e-06,
+      "loss": 1.48,
+      "step": 24900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.604305159860716e-06,
+      "loss": 1.4509,
+      "step": 25000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.602722380500159e-06,
+      "loss": 1.4455,
+      "step": 25100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.601139601139601e-06,
+      "loss": 1.483,
+      "step": 25200
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.599556821779044e-06,
+      "loss": 1.4654,
+      "step": 25300
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.597974042418487e-06,
+      "loss": 1.4881,
+      "step": 25400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.59639126305793e-06,
+      "loss": 1.4636,
+      "step": 25500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.594808483697372e-06,
+      "loss": 1.4807,
+      "step": 25600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5932257043368154e-06,
+      "loss": 1.4491,
+      "step": 25700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5916429249762585e-06,
+      "loss": 1.4742,
+      "step": 25800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5900601456157015e-06,
+      "loss": 1.4801,
+      "step": 25900
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.588477366255145e-06,
+      "loss": 1.4539,
+      "step": 26000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.586894586894588e-06,
+      "loss": 1.463,
+      "step": 26100
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.585311807534031e-06,
+      "loss": 1.4843,
+      "step": 26200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.583729028173473e-06,
+      "loss": 1.4538,
+      "step": 26300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.582146248812916e-06,
+      "loss": 1.4753,
+      "step": 26400
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.580563469452359e-06,
+      "loss": 1.437,
+      "step": 26500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.578980690091801e-06,
+      "loss": 1.4746,
+      "step": 26600
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.577397910731244e-06,
+      "loss": 1.4728,
+      "step": 26700
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.575815131370687e-06,
+      "loss": 1.4861,
+      "step": 26800
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.57423235201013e-06,
+      "loss": 1.477,
+      "step": 26900
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.5726495726495725e-06,
+      "loss": 1.4606,
+      "step": 27000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.5710667932890156e-06,
+      "loss": 1.4259,
+      "step": 27100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.569484013928459e-06,
+      "loss": 1.4523,
+      "step": 27200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.567901234567902e-06,
+      "loss": 1.473,
+      "step": 27300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.566318455207345e-06,
+      "loss": 1.4553,
+      "step": 27400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.564735675846788e-06,
+      "loss": 1.4734,
+      "step": 27500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.563152896486231e-06,
+      "loss": 1.477,
+      "step": 27600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.561570117125673e-06,
+      "loss": 1.4569,
+      "step": 27700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.559987337765116e-06,
+      "loss": 1.4534,
+      "step": 27800
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.558404558404559e-06,
+      "loss": 1.4799,
+      "step": 27900
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.556821779044001e-06,
+      "loss": 1.4623,
+      "step": 28000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.555238999683444e-06,
+      "loss": 1.4833,
+      "step": 28100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.553656220322887e-06,
+      "loss": 1.4746,
+      "step": 28200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.5520734409623304e-06,
+      "loss": 1.4497,
+      "step": 28300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.550490661601773e-06,
+      "loss": 1.4657,
+      "step": 28400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.548907882241216e-06,
+      "loss": 1.4704,
+      "step": 28500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.547325102880659e-06,
+      "loss": 1.4502,
+      "step": 28600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.545742323520102e-06,
+      "loss": 1.4662,
+      "step": 28700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.544159544159544e-06,
+      "loss": 1.4481,
+      "step": 28800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.542576764798987e-06,
+      "loss": 1.4462,
+      "step": 28900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.54099398543843e-06,
+      "loss": 1.4492,
+      "step": 29000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.539411206077873e-06,
+      "loss": 1.471,
+      "step": 29100
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.537828426717316e-06,
+      "loss": 1.4365,
+      "step": 29200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.536245647356759e-06,
+      "loss": 1.4449,
+      "step": 29300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.5346628679962014e-06,
+      "loss": 1.4707,
+      "step": 29400
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.5330800886356445e-06,
+      "loss": 1.4279,
+      "step": 29500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.5314973092750875e-06,
+      "loss": 1.4622,
+      "step": 29600
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.5299145299145306e-06,
+      "loss": 1.4691,
+      "step": 29700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.528331750553973e-06,
+      "loss": 1.436,
+      "step": 29800
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.526748971193416e-06,
+      "loss": 1.4551,
+      "step": 29900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.525166191832859e-06,
+      "loss": 1.4604,
+      "step": 30000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.523583412472302e-06,
+      "loss": 1.4489,
+      "step": 30100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.522000633111744e-06,
+      "loss": 1.4366,
+      "step": 30200
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.520417853751187e-06,
+      "loss": 1.4495,
+      "step": 30300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.51883507439063e-06,
+      "loss": 1.4462,
+      "step": 30400
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.517252295030073e-06,
+      "loss": 1.4486,
+      "step": 30500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.515669515669516e-06,
+      "loss": 1.4479,
+      "step": 30600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.514086736308959e-06,
+      "loss": 1.4437,
+      "step": 30700
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.5125039569484016e-06,
+      "loss": 1.4438,
+      "step": 30800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.510921177587845e-06,
+      "loss": 1.4517,
+      "step": 30900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.509338398227288e-06,
+      "loss": 1.4333,
+      "step": 31000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.507755618866731e-06,
+      "loss": 1.4399,
+      "step": 31100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.506172839506173e-06,
+      "loss": 1.4419,
+      "step": 31200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.504590060145616e-06,
+      "loss": 1.4447,
+      "step": 31300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.503007280785059e-06,
+      "loss": 1.4499,
+      "step": 31400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.501424501424502e-06,
+      "loss": 1.4598,
+      "step": 31500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.499841722063944e-06,
+      "loss": 1.4347,
+      "step": 31600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.498258942703387e-06,
+      "loss": 1.452,
+      "step": 31700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.49667616334283e-06,
+      "loss": 1.424,
+      "step": 31800
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.495093383982273e-06,
+      "loss": 1.4336,
+      "step": 31900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.493510604621716e-06,
+      "loss": 1.4374,
+      "step": 32000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.4919278252611595e-06,
+      "loss": 1.4201,
+      "step": 32100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.490345045900602e-06,
+      "loss": 1.438,
+      "step": 32200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.488762266540045e-06,
+      "loss": 1.4631,
+      "step": 32300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.487179487179488e-06,
+      "loss": 1.4354,
+      "step": 32400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.485596707818931e-06,
+      "loss": 1.4279,
+      "step": 32500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.484013928458373e-06,
+      "loss": 1.4515,
+      "step": 32600
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.482431149097816e-06,
+      "loss": 1.4536,
+      "step": 32700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.480848369737259e-06,
+      "loss": 1.4647,
+      "step": 32800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.479265590376702e-06,
+      "loss": 1.4503,
+      "step": 32900
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.477682811016144e-06,
+      "loss": 1.459,
+      "step": 33000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.4761000316555874e-06,
+      "loss": 1.4207,
+      "step": 33100
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4745172522950305e-06,
+      "loss": 1.4527,
+      "step": 33200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4729344729344735e-06,
+      "loss": 1.4496,
+      "step": 33300
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.471351693573916e-06,
+      "loss": 1.4407,
+      "step": 33400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.469768914213359e-06,
+      "loss": 1.4301,
+      "step": 33500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.468186134852802e-06,
+      "loss": 1.4445,
+      "step": 33600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.466603355492245e-06,
+      "loss": 1.4255,
+      "step": 33700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.465020576131688e-06,
+      "loss": 1.4477,
+      "step": 33800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.463437796771131e-06,
+      "loss": 1.4566,
+      "step": 33900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.461855017410573e-06,
+      "loss": 1.4372,
+      "step": 34000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.460272238050016e-06,
+      "loss": 1.4341,
+      "step": 34100
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.458689458689459e-06,
+      "loss": 1.4123,
+      "step": 34200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.457106679328902e-06,
+      "loss": 1.4265,
+      "step": 34300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4555238999683445e-06,
+      "loss": 1.4165,
+      "step": 34400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4539411206077876e-06,
+      "loss": 1.4402,
+      "step": 34500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.452358341247231e-06,
+      "loss": 1.4502,
+      "step": 34600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.450775561886674e-06,
+      "loss": 1.4312,
+      "step": 34700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.449192782526116e-06,
+      "loss": 1.434,
+      "step": 34800
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.447610003165559e-06,
+      "loss": 1.4188,
+      "step": 34900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.446027223805002e-06,
+      "loss": 1.4335,
+      "step": 35000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.444444444444444e-06,
+      "loss": 1.4238,
+      "step": 35100
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.442861665083888e-06,
+      "loss": 1.453,
+      "step": 35200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.441278885723331e-06,
+      "loss": 1.432,
+      "step": 35300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.439696106362773e-06,
+      "loss": 1.4386,
+      "step": 35400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.438113327002216e-06,
+      "loss": 1.4323,
+      "step": 35500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.436530547641659e-06,
+      "loss": 1.4428,
+      "step": 35600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.4349477682811024e-06,
+      "loss": 1.4313,
+      "step": 35700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.433364988920545e-06,
+      "loss": 1.4494,
+      "step": 35800
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.431782209559988e-06,
+      "loss": 1.4385,
+      "step": 35900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.430199430199431e-06,
+      "loss": 1.4008,
+      "step": 36000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.428616650838874e-06,
+      "loss": 1.4263,
+      "step": 36100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.427033871478316e-06,
+      "loss": 1.4529,
+      "step": 36200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.425451092117759e-06,
+      "loss": 1.4358,
+      "step": 36300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.423868312757202e-06,
+      "loss": 1.4295,
+      "step": 36400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.422285533396644e-06,
+      "loss": 1.427,
+      "step": 36500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.420702754036087e-06,
+      "loss": 1.4718,
+      "step": 36600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.41911997467553e-06,
+      "loss": 1.4339,
+      "step": 36700
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.417537195314973e-06,
+      "loss": 1.4423,
+      "step": 36800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4159544159544165e-06,
+      "loss": 1.432,
+      "step": 36900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.4143716365938595e-06,
+      "loss": 1.4387,
+      "step": 37000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.4127888572333026e-06,
+      "loss": 1.4456,
+      "step": 37100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.411206077872745e-06,
+      "loss": 1.426,
+      "step": 37200
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.409623298512188e-06,
+      "loss": 1.4078,
+      "step": 37300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.408040519151631e-06,
+      "loss": 1.4237,
+      "step": 37400
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.406457739791074e-06,
+      "loss": 1.4276,
+      "step": 37500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.404874960430516e-06,
+      "loss": 1.4364,
+      "step": 37600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.403292181069959e-06,
+      "loss": 1.4426,
+      "step": 37700
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.401709401709402e-06,
+      "loss": 1.4288,
+      "step": 37800
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.400126622348844e-06,
+      "loss": 1.4281,
+      "step": 37900
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.3985438429882875e-06,
+      "loss": 1.4293,
+      "step": 38000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.3969610636277305e-06,
+      "loss": 1.4186,
+      "step": 38100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.3953782842671736e-06,
+      "loss": 1.4555,
+      "step": 38200
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.393795504906616e-06,
+      "loss": 1.4183,
+      "step": 38300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.39221272554606e-06,
+      "loss": 1.4672,
+      "step": 38400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.390629946185503e-06,
+      "loss": 1.448,
+      "step": 38500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.389047166824945e-06,
+      "loss": 1.4271,
+      "step": 38600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.387464387464388e-06,
+      "loss": 1.4183,
+      "step": 38700
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.385881608103831e-06,
+      "loss": 1.4348,
+      "step": 38800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.384298828743273e-06,
+      "loss": 1.432,
+      "step": 38900
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.382716049382716e-06,
+      "loss": 1.4337,
+      "step": 39000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.381133270022159e-06,
+      "loss": 1.4462,
+      "step": 39100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.379550490661602e-06,
+      "loss": 1.4178,
+      "step": 39200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.3779677113010445e-06,
+      "loss": 1.4556,
+      "step": 39300
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.376384931940488e-06,
+      "loss": 1.4308,
+      "step": 39400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.374802152579931e-06,
+      "loss": 1.4346,
+      "step": 39500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.373219373219374e-06,
+      "loss": 1.4331,
+      "step": 39600
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.371636593858816e-06,
+      "loss": 1.4558,
+      "step": 39700
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.370053814498259e-06,
+      "loss": 1.4381,
+      "step": 39800
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.368471035137703e-06,
+      "loss": 1.3987,
+      "step": 39900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.366888255777145e-06,
+      "loss": 1.4353,
+      "step": 40000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.365305476416588e-06,
+      "loss": 1.4133,
+      "step": 40100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.363722697056031e-06,
+      "loss": 1.4334,
+      "step": 40200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.362139917695473e-06,
+      "loss": 1.4134,
+      "step": 40300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.360557138334916e-06,
+      "loss": 1.4354,
+      "step": 40400
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.358974358974359e-06,
+      "loss": 1.4146,
+      "step": 40500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.3573915796138025e-06,
+      "loss": 1.4488,
+      "step": 40600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.355808800253245e-06,
+      "loss": 1.4184,
+      "step": 40700
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.354226020892688e-06,
+      "loss": 1.4268,
+      "step": 40800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.352643241532131e-06,
+      "loss": 1.4266,
+      "step": 40900
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.351060462171574e-06,
+      "loss": 1.4584,
+      "step": 41000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.349477682811016e-06,
+      "loss": 1.4223,
+      "step": 41100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.347894903450459e-06,
+      "loss": 1.4329,
+      "step": 41200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.346312124089902e-06,
+      "loss": 1.4228,
+      "step": 41300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.344729344729345e-06,
+      "loss": 1.422,
+      "step": 41400
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.343146565368788e-06,
+      "loss": 1.3957,
+      "step": 41500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.341563786008231e-06,
+      "loss": 1.4411,
+      "step": 41600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.3399810066476735e-06,
+      "loss": 1.4368,
+      "step": 41700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.3383982272871165e-06,
+      "loss": 1.417,
+      "step": 41800
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.3368154479265595e-06,
+      "loss": 1.4198,
+      "step": 41900
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.335232668566003e-06,
+      "loss": 1.4302,
+      "step": 42000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.333649889205445e-06,
+      "loss": 1.422,
+      "step": 42100
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.332067109844888e-06,
+      "loss": 1.4339,
+      "step": 42200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.330484330484331e-06,
+      "loss": 1.4336,
+      "step": 42300
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.328901551123774e-06,
+      "loss": 1.4441,
+      "step": 42400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.327318771763216e-06,
+      "loss": 1.429,
+      "step": 42500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.325735992402659e-06,
+      "loss": 1.4372,
+      "step": 42600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.324153213042102e-06,
+      "loss": 1.4223,
+      "step": 42700
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.322570433681545e-06,
+      "loss": 1.4173,
+      "step": 42800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3209876543209875e-06,
+      "loss": 1.3985,
+      "step": 42900
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3194048749604305e-06,
+      "loss": 1.4475,
+      "step": 43000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.317822095599874e-06,
+      "loss": 1.4295,
+      "step": 43100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.316239316239317e-06,
+      "loss": 1.4288,
+      "step": 43200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.31465653687876e-06,
+      "loss": 1.429,
+      "step": 43300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.313073757518203e-06,
+      "loss": 1.4169,
+      "step": 43400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.311490978157645e-06,
+      "loss": 1.4496,
+      "step": 43500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.309908198797088e-06,
+      "loss": 1.4237,
+      "step": 43600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.308325419436531e-06,
+      "loss": 1.4393,
+      "step": 43700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.306742640075974e-06,
+      "loss": 1.4146,
+      "step": 43800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.305159860715416e-06,
+      "loss": 1.3996,
+      "step": 43900
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.303577081354859e-06,
+      "loss": 1.4271,
+      "step": 44000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.301994301994302e-06,
+      "loss": 1.4479,
+      "step": 44100
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.300411522633745e-06,
+      "loss": 1.4312,
+      "step": 44200
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.298828743273188e-06,
+      "loss": 1.4182,
+      "step": 44300
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.297245963912631e-06,
+      "loss": 1.4275,
+      "step": 44400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.295663184552074e-06,
+      "loss": 1.3982,
+      "step": 44500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.294080405191517e-06,
+      "loss": 1.4644,
+      "step": 44600
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.29249762583096e-06,
+      "loss": 1.4463,
+      "step": 44700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.290914846470403e-06,
+      "loss": 1.3969,
+      "step": 44800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.289332067109845e-06,
+      "loss": 1.3956,
+      "step": 44900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.287749287749288e-06,
+      "loss": 1.4116,
+      "step": 45000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.286166508388731e-06,
+      "loss": 1.4401,
+      "step": 45100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.284583729028174e-06,
+      "loss": 1.4236,
+      "step": 45200
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.283000949667616e-06,
+      "loss": 1.4446,
+      "step": 45300
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2814181703070594e-06,
+      "loss": 1.4417,
+      "step": 45400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2798353909465025e-06,
+      "loss": 1.4123,
+      "step": 45500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2782526115859455e-06,
+      "loss": 1.3983,
+      "step": 45600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.276669832225388e-06,
+      "loss": 1.4268,
+      "step": 45700
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.275087052864831e-06,
+      "loss": 1.4113,
+      "step": 45800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.273504273504274e-06,
+      "loss": 1.4372,
+      "step": 45900
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.271921494143717e-06,
+      "loss": 1.4374,
+      "step": 46000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.270338714783159e-06,
+      "loss": 1.421,
+      "step": 46100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.268755935422603e-06,
+      "loss": 1.4311,
+      "step": 46200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.267173156062045e-06,
+      "loss": 1.4422,
+      "step": 46300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.265590376701488e-06,
+      "loss": 1.4392,
+      "step": 46400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.264007597340931e-06,
+      "loss": 1.4286,
+      "step": 46500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.262424817980374e-06,
+      "loss": 1.4098,
+      "step": 46600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2608420386198165e-06,
+      "loss": 1.439,
+      "step": 46700
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2592592592592596e-06,
+      "loss": 1.4023,
+      "step": 46800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.257676479898703e-06,
+      "loss": 1.4369,
+      "step": 46900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.256093700538146e-06,
+      "loss": 1.4135,
+      "step": 47000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.254510921177588e-06,
+      "loss": 1.4164,
+      "step": 47100
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.252928141817031e-06,
+      "loss": 1.4133,
+      "step": 47200
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.251345362456474e-06,
+      "loss": 1.4166,
+      "step": 47300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.249762583095917e-06,
+      "loss": 1.4116,
+      "step": 47400
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.248179803735359e-06,
+      "loss": 1.4534,
+      "step": 47500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.246597024374802e-06,
+      "loss": 1.4292,
+      "step": 47600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.245014245014245e-06,
+      "loss": 1.4159,
+      "step": 47700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.243431465653688e-06,
+      "loss": 1.416,
+      "step": 47800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.241848686293131e-06,
+      "loss": 1.4072,
+      "step": 47900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2402659069325745e-06,
+      "loss": 1.3975,
+      "step": 48000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.238683127572017e-06,
+      "loss": 1.4115,
+      "step": 48100
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.23710034821146e-06,
+      "loss": 1.4114,
+      "step": 48200
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.235517568850903e-06,
+      "loss": 1.4278,
+      "step": 48300
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.233934789490346e-06,
+      "loss": 1.4234,
+      "step": 48400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.232352010129788e-06,
+      "loss": 1.4219,
+      "step": 48500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.230769230769231e-06,
+      "loss": 1.4103,
+      "step": 48600
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.229186451408674e-06,
+      "loss": 1.4068,
+      "step": 48700
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.227603672048117e-06,
+      "loss": 1.4026,
+      "step": 48800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.226020892687559e-06,
+      "loss": 1.4273,
+      "step": 48900
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.224438113327002e-06,
+      "loss": 1.4096,
+      "step": 49000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2228553339664454e-06,
+      "loss": 1.4006,
+      "step": 49100
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.221272554605888e-06,
+      "loss": 1.4202,
+      "step": 49200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2196897752453315e-06,
+      "loss": 1.4173,
+      "step": 49300
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.218106995884775e-06,
+      "loss": 1.4324,
+      "step": 49400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.216524216524217e-06,
+      "loss": 1.4135,
+      "step": 49500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.21494143716366e-06,
+      "loss": 1.4551,
+      "step": 49600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.213358657803103e-06,
+      "loss": 1.4121,
+      "step": 49700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.211775878442546e-06,
+      "loss": 1.4196,
+      "step": 49800
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.210193099081988e-06,
+      "loss": 1.4014,
+      "step": 49900
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.208610319721431e-06,
+      "loss": 1.4063,
+      "step": 50000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.207027540360874e-06,
+      "loss": 1.438,
+      "step": 50100
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2054447610003164e-06,
+      "loss": 1.4008,
+      "step": 50200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2038619816397595e-06,
+      "loss": 1.4143,
+      "step": 50300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2022792022792025e-06,
+      "loss": 1.4159,
+      "step": 50400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2006964229186456e-06,
+      "loss": 1.4183,
+      "step": 50500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.199113643558088e-06,
+      "loss": 1.4222,
+      "step": 50600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.197530864197531e-06,
+      "loss": 1.4191,
+      "step": 50700
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.195948084836974e-06,
+      "loss": 1.4222,
+      "step": 50800
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.194365305476417e-06,
+      "loss": 1.4086,
+      "step": 50900
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.19278252611586e-06,
+      "loss": 1.406,
+      "step": 51000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.191199746755303e-06,
+      "loss": 1.4127,
+      "step": 51100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.189616967394746e-06,
+      "loss": 1.4103,
+      "step": 51200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.188034188034188e-06,
+      "loss": 1.4243,
+      "step": 51300
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.186451408673631e-06,
+      "loss": 1.3865,
+      "step": 51400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.184868629313074e-06,
+      "loss": 1.4028,
+      "step": 51500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1832858499525166e-06,
+      "loss": 1.4309,
+      "step": 51600
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.18170307059196e-06,
+      "loss": 1.4066,
+      "step": 51700
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.180120291231403e-06,
+      "loss": 1.3914,
+      "step": 51800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.178537511870846e-06,
+      "loss": 1.4097,
+      "step": 51900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.176954732510288e-06,
+      "loss": 1.4072,
+      "step": 52000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.175371953149731e-06,
+      "loss": 1.4314,
+      "step": 52100
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.173789173789174e-06,
+      "loss": 1.4212,
+      "step": 52200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.172206394428617e-06,
+      "loss": 1.4086,
+      "step": 52300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.17062361506806e-06,
+      "loss": 1.3854,
+      "step": 52400
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.169040835707503e-06,
+      "loss": 1.3973,
+      "step": 52500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.167458056346946e-06,
+      "loss": 1.448,
+      "step": 52600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.165875276986388e-06,
+      "loss": 1.4231,
+      "step": 52700
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.1642924976258314e-06,
+      "loss": 1.4071,
+      "step": 52800
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.1627097182652745e-06,
+      "loss": 1.4263,
+      "step": 52900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.161126938904717e-06,
+      "loss": 1.4128,
+      "step": 53000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.15954415954416e-06,
+      "loss": 1.424,
+      "step": 53100
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.157961380183603e-06,
+      "loss": 1.3951,
+      "step": 53200
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.156378600823046e-06,
+      "loss": 1.4439,
+      "step": 53300
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.154795821462488e-06,
+      "loss": 1.3815,
+      "step": 53400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.153213042101931e-06,
+      "loss": 1.4348,
+      "step": 53500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.151630262741374e-06,
+      "loss": 1.4249,
+      "step": 53600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.150047483380817e-06,
+      "loss": 1.4264,
+      "step": 53700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.148464704020259e-06,
+      "loss": 1.416,
+      "step": 53800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.1468819246597024e-06,
+      "loss": 1.4141,
+      "step": 53900
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.145299145299146e-06,
+      "loss": 1.4463,
+      "step": 54000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.1437163659385885e-06,
+      "loss": 1.4166,
+      "step": 54100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.1421335865780316e-06,
+      "loss": 1.4101,
+      "step": 54200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.140550807217475e-06,
+      "loss": 1.4364,
+      "step": 54300
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.138968027856917e-06,
+      "loss": 1.4172,
+      "step": 54400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.13738524849636e-06,
+      "loss": 1.416,
+      "step": 54500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.135802469135803e-06,
+      "loss": 1.4179,
+      "step": 54600
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.134219689775246e-06,
+      "loss": 1.3969,
+      "step": 54700
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.132636910414688e-06,
+      "loss": 1.4152,
+      "step": 54800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.131054131054131e-06,
+      "loss": 1.4426,
+      "step": 54900
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.129471351693574e-06,
+      "loss": 1.4159,
+      "step": 55000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.127888572333017e-06,
+      "loss": 1.3969,
+      "step": 55100
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1263057929724595e-06,
+      "loss": 1.4,
+      "step": 55200
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1247230136119026e-06,
+      "loss": 1.4173,
+      "step": 55300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.123140234251346e-06,
+      "loss": 1.4031,
+      "step": 55400
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.121557454890789e-06,
+      "loss": 1.417,
+      "step": 55500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.119974675530232e-06,
+      "loss": 1.4255,
+      "step": 55600
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.118391896169675e-06,
+      "loss": 1.4058,
+      "step": 55700
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.116809116809117e-06,
+      "loss": 1.4117,
+      "step": 55800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.11522633744856e-06,
+      "loss": 1.4221,
+      "step": 55900
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.113643558088003e-06,
+      "loss": 1.4102,
+      "step": 56000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.112060778727446e-06,
+      "loss": 1.4293,
+      "step": 56100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.110477999366888e-06,
+      "loss": 1.4133,
+      "step": 56200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.108895220006331e-06,
+      "loss": 1.4377,
+      "step": 56300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.107312440645774e-06,
+      "loss": 1.4066,
+      "step": 56400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.1057296612852174e-06,
+      "loss": 1.439,
+      "step": 56500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.10414688192466e-06,
+      "loss": 1.4073,
+      "step": 56600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.102564102564103e-06,
+      "loss": 1.4286,
+      "step": 56700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.100981323203546e-06,
+      "loss": 1.4036,
+      "step": 56800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.099398543842989e-06,
+      "loss": 1.4184,
+      "step": 56900
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.097815764482431e-06,
+      "loss": 1.3984,
+      "step": 57000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.096232985121875e-06,
+      "loss": 1.404,
+      "step": 57100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.094650205761317e-06,
+      "loss": 1.4018,
+      "step": 57200
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.09306742640076e-06,
+      "loss": 1.3751,
+      "step": 57300
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.091484647040203e-06,
+      "loss": 1.3923,
+      "step": 57400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.089901867679646e-06,
+      "loss": 1.4188,
+      "step": 57500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.088319088319088e-06,
+      "loss": 1.4196,
+      "step": 57600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0867363089585315e-06,
+      "loss": 1.4308,
+      "step": 57700
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0851535295979745e-06,
+      "loss": 1.399,
+      "step": 57800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0835707502374176e-06,
+      "loss": 1.3927,
+      "step": 57900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.08198797087686e-06,
+      "loss": 1.4348,
+      "step": 58000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.080405191516303e-06,
+      "loss": 1.4094,
+      "step": 58100
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.078822412155746e-06,
+      "loss": 1.3906,
+      "step": 58200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.077239632795189e-06,
+      "loss": 1.4106,
+      "step": 58300
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.075656853434631e-06,
+      "loss": 1.4167,
+      "step": 58400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.074074074074074e-06,
+      "loss": 1.4388,
+      "step": 58500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.072491294713517e-06,
+      "loss": 1.4128,
+      "step": 58600
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.07090851535296e-06,
+      "loss": 1.393,
+      "step": 58700
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.069325735992403e-06,
+      "loss": 1.397,
+      "step": 58800
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.067742956631846e-06,
+      "loss": 1.4215,
+      "step": 58900
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.0661601772712885e-06,
+      "loss": 1.4096,
+      "step": 59000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.064577397910732e-06,
+      "loss": 1.3946,
+      "step": 59100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.062994618550175e-06,
+      "loss": 1.4143,
+      "step": 59200
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.061411839189618e-06,
+      "loss": 1.4104,
+      "step": 59300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.05982905982906e-06,
+      "loss": 1.3939,
+      "step": 59400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.058246280468503e-06,
+      "loss": 1.4336,
+      "step": 59500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.056663501107946e-06,
+      "loss": 1.3844,
+      "step": 59600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.055080721747389e-06,
+      "loss": 1.4193,
+      "step": 59700
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.053497942386831e-06,
+      "loss": 1.356,
+      "step": 59800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.051915163026274e-06,
+      "loss": 1.405,
+      "step": 59900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.050332383665717e-06,
+      "loss": 1.3894,
+      "step": 60000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.0487496043051595e-06,
+      "loss": 1.421,
+      "step": 60100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.047166824944603e-06,
+      "loss": 1.4048,
+      "step": 60200
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.0455840455840465e-06,
+      "loss": 1.4176,
+      "step": 60300
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.044001266223489e-06,
+      "loss": 1.4025,
+      "step": 60400
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.042418486862932e-06,
+      "loss": 1.412,
+      "step": 60500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.040835707502375e-06,
+      "loss": 1.4001,
+      "step": 60600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.039252928141818e-06,
+      "loss": 1.4107,
+      "step": 60700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.03767014878126e-06,
+      "loss": 1.4193,
+      "step": 60800
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.036087369420703e-06,
+      "loss": 1.4219,
+      "step": 60900
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.034504590060146e-06,
+      "loss": 1.4235,
+      "step": 61000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.032921810699589e-06,
+      "loss": 1.4157,
+      "step": 61100
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.031339031339031e-06,
+      "loss": 1.3992,
+      "step": 61200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.029756251978474e-06,
+      "loss": 1.4021,
+      "step": 61300
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0281734726179175e-06,
+      "loss": 1.4146,
+      "step": 61400
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.02659069325736e-06,
+      "loss": 1.4072,
+      "step": 61500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.025007913896803e-06,
+      "loss": 1.4073,
+      "step": 61600
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.023425134536246e-06,
+      "loss": 1.3853,
+      "step": 61700
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.021842355175689e-06,
+      "loss": 1.379,
+      "step": 61800
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.020259575815132e-06,
+      "loss": 1.3842,
+      "step": 61900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.018676796454575e-06,
+      "loss": 1.4068,
+      "step": 62000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.017094017094018e-06,
+      "loss": 1.4011,
+      "step": 62100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.01551123773346e-06,
+      "loss": 1.3949,
+      "step": 62200
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.013928458372903e-06,
+      "loss": 1.4121,
+      "step": 62300
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.012345679012346e-06,
+      "loss": 1.4173,
+      "step": 62400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.010762899651789e-06,
+      "loss": 1.4121,
+      "step": 62500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0091801202912315e-06,
+      "loss": 1.4009,
+      "step": 62600
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0075973409306745e-06,
+      "loss": 1.4124,
+      "step": 62700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.006014561570118e-06,
+      "loss": 1.4065,
+      "step": 62800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.00443178220956e-06,
+      "loss": 1.3818,
+      "step": 62900
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.002849002849003e-06,
+      "loss": 1.3962,
+      "step": 63000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.001266223488446e-06,
+      "loss": 1.3835,
+      "step": 63100
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.999683444127889e-06,
+      "loss": 1.3742,
+      "step": 63200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.998100664767331e-06,
+      "loss": 1.3604,
+      "step": 63300
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.996517885406775e-06,
+      "loss": 1.3773,
+      "step": 63400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.994935106046218e-06,
+      "loss": 1.3636,
+      "step": 63500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.99335232668566e-06,
+      "loss": 1.3522,
+      "step": 63600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.991769547325103e-06,
+      "loss": 1.3363,
+      "step": 63700
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.990186767964546e-06,
+      "loss": 1.361,
+      "step": 63800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9886039886039894e-06,
+      "loss": 1.3685,
+      "step": 63900
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.987021209243432e-06,
+      "loss": 1.3688,
+      "step": 64000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.985438429882875e-06,
+      "loss": 1.384,
+      "step": 64100
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.983855650522318e-06,
+      "loss": 1.3353,
+      "step": 64200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.98227287116176e-06,
+      "loss": 1.3898,
+      "step": 64300
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.980690091801203e-06,
+      "loss": 1.3438,
+      "step": 64400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.979107312440646e-06,
+      "loss": 1.3367,
+      "step": 64500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.977524533080089e-06,
+      "loss": 1.3665,
+      "step": 64600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.975941753719531e-06,
+      "loss": 1.353,
+      "step": 64700
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.974358974358974e-06,
+      "loss": 1.3453,
+      "step": 64800
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.972776194998417e-06,
+      "loss": 1.3369,
+      "step": 64900
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.97119341563786e-06,
+      "loss": 1.348,
+      "step": 65000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.9696106362773035e-06,
+      "loss": 1.3639,
+      "step": 65100
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.9680278569167465e-06,
+      "loss": 1.3676,
+      "step": 65200
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.966445077556189e-06,
+      "loss": 1.3446,
+      "step": 65300
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.964862298195632e-06,
+      "loss": 1.3595,
+      "step": 65400
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.963279518835075e-06,
+      "loss": 1.3416,
+      "step": 65500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.961696739474518e-06,
+      "loss": 1.352,
+      "step": 65600
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.96011396011396e-06,
+      "loss": 1.3585,
+      "step": 65700
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.958531180753403e-06,
+      "loss": 1.3617,
+      "step": 65800
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.956948401392846e-06,
+      "loss": 1.3404,
+      "step": 65900
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.955365622032289e-06,
+      "loss": 1.3627,
+      "step": 66000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.953782842671731e-06,
+      "loss": 1.354,
+      "step": 66100
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9522000633111744e-06,
+      "loss": 1.3701,
+      "step": 66200
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9506172839506175e-06,
+      "loss": 1.3561,
+      "step": 66300
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9490345045900605e-06,
+      "loss": 1.3754,
+      "step": 66400
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.947451725229504e-06,
+      "loss": 1.3656,
+      "step": 66500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.945868945868947e-06,
+      "loss": 1.3586,
+      "step": 66600
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.944286166508389e-06,
+      "loss": 1.3348,
+      "step": 66700
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.942703387147832e-06,
+      "loss": 1.3549,
+      "step": 66800
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.941120607787275e-06,
+      "loss": 1.3488,
+      "step": 66900
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.939537828426718e-06,
+      "loss": 1.3692,
+      "step": 67000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.93795504906616e-06,
+      "loss": 1.3401,
+      "step": 67100
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.936372269705603e-06,
+      "loss": 1.3545,
+      "step": 67200
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.934789490345046e-06,
+      "loss": 1.347,
+      "step": 67300
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.933206710984489e-06,
+      "loss": 1.3796,
+      "step": 67400
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.9316239316239315e-06,
+      "loss": 1.3656,
+      "step": 67500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.9300411522633746e-06,
+      "loss": 1.3286,
+      "step": 67600
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.928458372902818e-06,
+      "loss": 1.3433,
+      "step": 67700
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.926875593542261e-06,
+      "loss": 1.3645,
+      "step": 67800
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.925292814181703e-06,
+      "loss": 1.3678,
+      "step": 67900
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.923710034821146e-06,
+      "loss": 1.3597,
+      "step": 68000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.922127255460589e-06,
+      "loss": 1.3543,
+      "step": 68100
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.920544476100032e-06,
+      "loss": 1.3634,
+      "step": 68200
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.918961696739475e-06,
+      "loss": 1.343,
+      "step": 68300
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.917378917378918e-06,
+      "loss": 1.3481,
+      "step": 68400
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.91579613801836e-06,
+      "loss": 1.348,
+      "step": 68500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.914213358657803e-06,
+      "loss": 1.356,
+      "step": 68600
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.912630579297246e-06,
+      "loss": 1.3819,
+      "step": 68700
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9110477999366895e-06,
+      "loss": 1.3476,
+      "step": 68800
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.909465020576132e-06,
+      "loss": 1.3359,
+      "step": 68900
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.907882241215575e-06,
+      "loss": 1.3606,
+      "step": 69000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.906299461855018e-06,
+      "loss": 1.3676,
+      "step": 69100
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.904716682494461e-06,
+      "loss": 1.3724,
+      "step": 69200
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.903133903133903e-06,
+      "loss": 1.3654,
+      "step": 69300
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.901551123773346e-06,
+      "loss": 1.3772,
+      "step": 69400
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.899968344412789e-06,
+      "loss": 1.3806,
+      "step": 69500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.898385565052232e-06,
+      "loss": 1.3633,
+      "step": 69600
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.896802785691675e-06,
+      "loss": 1.3594,
+      "step": 69700
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.895220006331118e-06,
+      "loss": 1.3504,
+      "step": 69800
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.8936372269705604e-06,
+      "loss": 1.3525,
+      "step": 69900
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.8920544476100035e-06,
+      "loss": 1.3608,
+      "step": 70000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.8904716682494465e-06,
+      "loss": 1.348,
+      "step": 70100
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.88888888888889e-06,
+      "loss": 1.3252,
+      "step": 70200
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.887306109528332e-06,
+      "loss": 1.3624,
+      "step": 70300
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.885723330167775e-06,
+      "loss": 1.348,
+      "step": 70400
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.884140550807218e-06,
+      "loss": 1.3765,
+      "step": 70500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.882557771446661e-06,
+      "loss": 1.3879,
+      "step": 70600
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.880974992086103e-06,
+      "loss": 1.3765,
+      "step": 70700
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.879392212725546e-06,
+      "loss": 1.3685,
+      "step": 70800
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.877809433364989e-06,
+      "loss": 1.37,
+      "step": 70900
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.876226654004432e-06,
+      "loss": 1.3613,
+      "step": 71000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.8746438746438745e-06,
+      "loss": 1.3731,
+      "step": 71100
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.873061095283318e-06,
+      "loss": 1.3673,
+      "step": 71200
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.8714783159227606e-06,
+      "loss": 1.3548,
+      "step": 71300
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.869895536562204e-06,
+      "loss": 1.344,
+      "step": 71400
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.868312757201647e-06,
+      "loss": 1.3536,
+      "step": 71500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.86672997784109e-06,
+      "loss": 1.3772,
+      "step": 71600
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.865147198480532e-06,
+      "loss": 1.3332,
+      "step": 71700
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.863564419119975e-06,
+      "loss": 1.3838,
+      "step": 71800
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.861981639759418e-06,
+      "loss": 1.3746,
+      "step": 71900
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.860398860398861e-06,
+      "loss": 1.3498,
+      "step": 72000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.858816081038303e-06,
+      "loss": 1.3559,
+      "step": 72100
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.857233301677746e-06,
+      "loss": 1.3602,
+      "step": 72200
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.855650522317189e-06,
+      "loss": 1.3448,
+      "step": 72300
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.854067742956632e-06,
+      "loss": 1.3547,
+      "step": 72400
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.852484963596075e-06,
+      "loss": 1.3709,
+      "step": 72500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.850902184235518e-06,
+      "loss": 1.364,
+      "step": 72600
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.849319404874961e-06,
+      "loss": 1.3403,
+      "step": 72700
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.847736625514404e-06,
+      "loss": 1.3509,
+      "step": 72800
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.846153846153847e-06,
+      "loss": 1.3617,
+      "step": 72900
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.84457106679329e-06,
+      "loss": 1.3364,
+      "step": 73000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.842988287432732e-06,
+      "loss": 1.3216,
+      "step": 73100
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.841405508072175e-06,
+      "loss": 1.3694,
+      "step": 73200
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.839822728711618e-06,
+      "loss": 1.3433,
+      "step": 73300
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.838239949351061e-06,
+      "loss": 1.3569,
+      "step": 73400
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.836657169990503e-06,
+      "loss": 1.3632,
+      "step": 73500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.8350743906299464e-06,
+      "loss": 1.3487,
+      "step": 73600
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.8334916112693895e-06,
+      "loss": 1.3595,
+      "step": 73700
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.8319088319088325e-06,
+      "loss": 1.3269,
+      "step": 73800
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.830326052548275e-06,
+      "loss": 1.3452,
+      "step": 73900
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.828743273187718e-06,
+      "loss": 1.3676,
+      "step": 74000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.827160493827161e-06,
+      "loss": 1.3345,
+      "step": 74100
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.825577714466603e-06,
+      "loss": 1.3674,
+      "step": 74200
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.823994935106047e-06,
+      "loss": 1.3462,
+      "step": 74300
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.82241215574549e-06,
+      "loss": 1.3423,
+      "step": 74400
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.820829376384932e-06,
+      "loss": 1.3688,
+      "step": 74500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.819246597024375e-06,
+      "loss": 1.3577,
+      "step": 74600
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.817663817663818e-06,
+      "loss": 1.3475,
+      "step": 74700
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.816081038303261e-06,
+      "loss": 1.3605,
+      "step": 74800
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.814498258942704e-06,
+      "loss": 1.3702,
+      "step": 74900
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.8129154795821466e-06,
+      "loss": 1.3731,
+      "step": 75000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.8113327002215896e-06,
+      "loss": 1.364,
+      "step": 75100
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.8097499208610322e-06,
+      "loss": 1.3373,
+      "step": 75200
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.808167141500475e-06,
+      "loss": 1.356,
+      "step": 75300
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.806584362139918e-06,
+      "loss": 1.3346,
+      "step": 75400
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.8050015827793605e-06,
+      "loss": 1.3646,
+      "step": 75500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8034188034188036e-06,
+      "loss": 1.353,
+      "step": 75600
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.801836024058246e-06,
+      "loss": 1.3246,
+      "step": 75700
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8002532446976893e-06,
+      "loss": 1.3659,
+      "step": 75800
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.7986704653371327e-06,
+      "loss": 1.3665,
+      "step": 75900
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.7970876859765753e-06,
+      "loss": 1.3685,
+      "step": 76000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.7955049066160184e-06,
+      "loss": 1.344,
+      "step": 76100
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.793922127255461e-06,
+      "loss": 1.3183,
+      "step": 76200
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.792339347894904e-06,
+      "loss": 1.3421,
+      "step": 76300
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.7907565685343467e-06,
+      "loss": 1.3504,
+      "step": 76400
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.7891737891737893e-06,
+      "loss": 1.3559,
+      "step": 76500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.7875910098132324e-06,
+      "loss": 1.3221,
+      "step": 76600
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.786008230452675e-06,
+      "loss": 1.3545,
+      "step": 76700
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.784425451092118e-06,
+      "loss": 1.3611,
+      "step": 76800
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.7828426717315607e-06,
+      "loss": 1.3621,
+      "step": 76900
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.7812598923710037e-06,
+      "loss": 1.354,
+      "step": 77000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.7796771130104463e-06,
+      "loss": 1.3293,
+      "step": 77100
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.7780943336498894e-06,
+      "loss": 1.3496,
+      "step": 77200
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.776511554289332e-06,
+      "loss": 1.3431,
+      "step": 77300
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.774928774928775e-06,
+      "loss": 1.3338,
+      "step": 77400
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.7733459955682185e-06,
+      "loss": 1.3329,
+      "step": 77500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.771763216207661e-06,
+      "loss": 1.3387,
+      "step": 77600
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.770180436847104e-06,
+      "loss": 1.3766,
+      "step": 77700
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.768597657486547e-06,
+      "loss": 1.3163,
+      "step": 77800
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.7670148781259894e-06,
+      "loss": 1.3449,
+      "step": 77900
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.7654320987654325e-06,
+      "loss": 1.3351,
+      "step": 78000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.763849319404875e-06,
+      "loss": 1.3551,
+      "step": 78100
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.762266540044318e-06,
+      "loss": 1.3391,
+      "step": 78200
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.760683760683761e-06,
+      "loss": 1.3542,
+      "step": 78300
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.759100981323204e-06,
+      "loss": 1.332,
+      "step": 78400
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.7575182019626465e-06,
+      "loss": 1.3421,
+      "step": 78500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.7559354226020895e-06,
+      "loss": 1.3687,
+      "step": 78600
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.754352643241532e-06,
+      "loss": 1.3348,
+      "step": 78700
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.752769863880975e-06,
+      "loss": 1.3163,
+      "step": 78800
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.751187084520418e-06,
+      "loss": 1.3787,
+      "step": 78900
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.7496043051598613e-06,
+      "loss": 1.3313,
+      "step": 79000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.7480215257993043e-06,
+      "loss": 1.3514,
+      "step": 79100
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.746438746438747e-06,
+      "loss": 1.3543,
+      "step": 79200
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.7448559670781896e-06,
+      "loss": 1.3407,
+      "step": 79300
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.7432731877176326e-06,
+      "loss": 1.3521,
+      "step": 79400
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.7416904083570752e-06,
+      "loss": 1.3531,
+      "step": 79500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.7401076289965183e-06,
+      "loss": 1.3366,
+      "step": 79600
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.738524849635961e-06,
+      "loss": 1.3766,
+      "step": 79700
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.736942070275404e-06,
+      "loss": 1.3361,
+      "step": 79800
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.7353592909148466e-06,
+      "loss": 1.353,
+      "step": 79900
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7337765115542896e-06,
+      "loss": 1.3788,
+      "step": 80000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7321937321937323e-06,
+      "loss": 1.3579,
+      "step": 80100
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7306109528331753e-06,
+      "loss": 1.3516,
+      "step": 80200
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.729028173472618e-06,
+      "loss": 1.3363,
+      "step": 80300
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.727445394112061e-06,
+      "loss": 1.3517,
+      "step": 80400
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7258626147515036e-06,
+      "loss": 1.3608,
+      "step": 80500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.724279835390947e-06,
+      "loss": 1.3653,
+      "step": 80600
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7226970560303897e-06,
+      "loss": 1.3336,
+      "step": 80700
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7211142766698328e-06,
+      "loss": 1.3465,
+      "step": 80800
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7195314973092754e-06,
+      "loss": 1.3553,
+      "step": 80900
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7179487179487184e-06,
+      "loss": 1.369,
+      "step": 81000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.716365938588161e-06,
+      "loss": 1.3499,
+      "step": 81100
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.714783159227604e-06,
+      "loss": 1.3304,
+      "step": 81200
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7132003798670467e-06,
+      "loss": 1.3423,
+      "step": 81300
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7116176005064898e-06,
+      "loss": 1.3477,
+      "step": 81400
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7100348211459324e-06,
+      "loss": 1.3555,
+      "step": 81500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7084520417853754e-06,
+      "loss": 1.3383,
+      "step": 81600
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.706869262424818e-06,
+      "loss": 1.337,
+      "step": 81700
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.705286483064261e-06,
+      "loss": 1.3622,
+      "step": 81800
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7037037037037037e-06,
+      "loss": 1.3406,
+      "step": 81900
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7021209243431468e-06,
+      "loss": 1.3649,
+      "step": 82000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7005381449825894e-06,
+      "loss": 1.3329,
+      "step": 82100
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.698955365622033e-06,
+      "loss": 1.3278,
+      "step": 82200
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.6973725862614755e-06,
+      "loss": 1.3509,
+      "step": 82300
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.6957898069009186e-06,
+      "loss": 1.3532,
+      "step": 82400
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.694207027540361e-06,
+      "loss": 1.3628,
+      "step": 82500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.6926242481798042e-06,
+      "loss": 1.3405,
+      "step": 82600
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.691041468819247e-06,
+      "loss": 1.3567,
+      "step": 82700
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.68945868945869e-06,
+      "loss": 1.3371,
+      "step": 82800
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.6878759100981325e-06,
+      "loss": 1.3617,
+      "step": 82900
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.6862931307375756e-06,
+      "loss": 1.3455,
+      "step": 83000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.684710351377018e-06,
+      "loss": 1.3153,
+      "step": 83100
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6831275720164612e-06,
+      "loss": 1.3322,
+      "step": 83200
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.681544792655904e-06,
+      "loss": 1.3395,
+      "step": 83300
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.679962013295347e-06,
+      "loss": 1.3499,
+      "step": 83400
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6783792339347895e-06,
+      "loss": 1.3425,
+      "step": 83500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6767964545742326e-06,
+      "loss": 1.3486,
+      "step": 83600
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6752136752136756e-06,
+      "loss": 1.3436,
+      "step": 83700
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.6736308958531187e-06,
+      "loss": 1.3287,
+      "step": 83800
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.6720481164925613e-06,
+      "loss": 1.3592,
+      "step": 83900
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.6704653371320044e-06,
+      "loss": 1.384,
+      "step": 84000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.668882557771447e-06,
+      "loss": 1.3429,
+      "step": 84100
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.66729977841089e-06,
+      "loss": 1.3577,
+      "step": 84200
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.6657169990503327e-06,
+      "loss": 1.335,
+      "step": 84300
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6641342196897757e-06,
+      "loss": 1.351,
+      "step": 84400
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6625514403292183e-06,
+      "loss": 1.3479,
+      "step": 84500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6609686609686614e-06,
+      "loss": 1.3309,
+      "step": 84600
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.659385881608104e-06,
+      "loss": 1.3299,
+      "step": 84700
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.657803102247547e-06,
+      "loss": 1.341,
+      "step": 84800
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6562203228869897e-06,
+      "loss": 1.3577,
+      "step": 84900
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6546375435264327e-06,
+      "loss": 1.3114,
+      "step": 85000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6530547641658753e-06,
+      "loss": 1.3533,
+      "step": 85100
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.651471984805318e-06,
+      "loss": 1.3538,
+      "step": 85200
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6498892054447614e-06,
+      "loss": 1.3231,
+      "step": 85300
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6483064260842045e-06,
+      "loss": 1.3331,
+      "step": 85400
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.646723646723647e-06,
+      "loss": 1.3476,
+      "step": 85500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.64514086736309e-06,
+      "loss": 1.3395,
+      "step": 85600
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6435580880025328e-06,
+      "loss": 1.3632,
+      "step": 85700
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.641975308641976e-06,
+      "loss": 1.3583,
+      "step": 85800
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6403925292814185e-06,
+      "loss": 1.3493,
+      "step": 85900
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6388097499208615e-06,
+      "loss": 1.3395,
+      "step": 86000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.637226970560304e-06,
+      "loss": 1.3199,
+      "step": 86100
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.635644191199747e-06,
+      "loss": 1.3621,
+      "step": 86200
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.63406141183919e-06,
+      "loss": 1.3265,
+      "step": 86300
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.632478632478633e-06,
+      "loss": 1.3263,
+      "step": 86400
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6308958531180755e-06,
+      "loss": 1.3395,
+      "step": 86500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.629313073757518e-06,
+      "loss": 1.3416,
+      "step": 86600
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.627730294396961e-06,
+      "loss": 1.3286,
+      "step": 86700
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6261475150364046e-06,
+      "loss": 1.3738,
+      "step": 86800
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6245647356758472e-06,
+      "loss": 1.3188,
+      "step": 86900
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6229819563152903e-06,
+      "loss": 1.352,
+      "step": 87000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.621399176954733e-06,
+      "loss": 1.3397,
+      "step": 87100
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.619816397594176e-06,
+      "loss": 1.3484,
+      "step": 87200
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6182336182336186e-06,
+      "loss": 1.3384,
+      "step": 87300
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6166508388730616e-06,
+      "loss": 1.372,
+      "step": 87400
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6150680595125043e-06,
+      "loss": 1.3385,
+      "step": 87500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6134852801519473e-06,
+      "loss": 1.3598,
+      "step": 87600
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.61190250079139e-06,
+      "loss": 1.326,
+      "step": 87700
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6103197214308326e-06,
+      "loss": 1.368,
+      "step": 87800
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6087369420702756e-06,
+      "loss": 1.3561,
+      "step": 87900
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6071541627097182e-06,
+      "loss": 1.3387,
+      "step": 88000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6055713833491613e-06,
+      "loss": 1.3297,
+      "step": 88100
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.603988603988604e-06,
+      "loss": 1.364,
+      "step": 88200
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.602405824628047e-06,
+      "loss": 1.3651,
+      "step": 88300
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6008230452674904e-06,
+      "loss": 1.3374,
+      "step": 88400
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.599240265906933e-06,
+      "loss": 1.3361,
+      "step": 88500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.597657486546376e-06,
+      "loss": 1.3155,
+      "step": 88600
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.5960747071858187e-06,
+      "loss": 1.3294,
+      "step": 88700
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5944919278252618e-06,
+      "loss": 1.3177,
+      "step": 88800
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5929091484647044e-06,
+      "loss": 1.3472,
+      "step": 88900
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5913263691041474e-06,
+      "loss": 1.3551,
+      "step": 89000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.58974358974359e-06,
+      "loss": 1.3543,
+      "step": 89100
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5881608103830327e-06,
+      "loss": 1.338,
+      "step": 89200
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5865780310224757e-06,
+      "loss": 1.3542,
+      "step": 89300
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5849952516619184e-06,
+      "loss": 1.3447,
+      "step": 89400
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5834124723013614e-06,
+      "loss": 1.3206,
+      "step": 89500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.581829692940804e-06,
+      "loss": 1.3711,
+      "step": 89600
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.580246913580247e-06,
+      "loss": 1.3349,
+      "step": 89700
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5786641342196897e-06,
+      "loss": 1.3522,
+      "step": 89800
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5770813548591327e-06,
+      "loss": 1.3342,
+      "step": 89900
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5754985754985762e-06,
+      "loss": 1.3598,
+      "step": 90000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.573915796138019e-06,
+      "loss": 1.3272,
+      "step": 90100
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.572333016777462e-06,
+      "loss": 1.3515,
+      "step": 90200
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5707502374169045e-06,
+      "loss": 1.3289,
+      "step": 90300
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.569167458056347e-06,
+      "loss": 1.3632,
+      "step": 90400
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.56758467869579e-06,
+      "loss": 1.3324,
+      "step": 90500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.566001899335233e-06,
+      "loss": 1.3339,
+      "step": 90600
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.564419119974676e-06,
+      "loss": 1.3432,
+      "step": 90700
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5628363406141185e-06,
+      "loss": 1.3628,
+      "step": 90800
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5612535612535615e-06,
+      "loss": 1.3518,
+      "step": 90900
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.559670781893004e-06,
+      "loss": 1.3588,
+      "step": 91000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.558088002532447e-06,
+      "loss": 1.336,
+      "step": 91100
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.55650522317189e-06,
+      "loss": 1.3541,
+      "step": 91200
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.554922443811333e-06,
+      "loss": 1.3326,
+      "step": 91300
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5533396644507755e-06,
+      "loss": 1.3279,
+      "step": 91400
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.551756885090219e-06,
+      "loss": 1.3506,
+      "step": 91500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.550174105729662e-06,
+      "loss": 1.3268,
+      "step": 91600
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5485913263691046e-06,
+      "loss": 1.3444,
+      "step": 91700
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5470085470085473e-06,
+      "loss": 1.3283,
+      "step": 91800
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5454257676479903e-06,
+      "loss": 1.3466,
+      "step": 91900
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.543842988287433e-06,
+      "loss": 1.3298,
+      "step": 92000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.542260208926876e-06,
+      "loss": 1.334,
+      "step": 92100
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5406774295663186e-06,
+      "loss": 1.3271,
+      "step": 92200
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5390946502057617e-06,
+      "loss": 1.3278,
+      "step": 92300
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5375118708452043e-06,
+      "loss": 1.3547,
+      "step": 92400
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5359290914846473e-06,
+      "loss": 1.3342,
+      "step": 92500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.53434631212409e-06,
+      "loss": 1.3258,
+      "step": 92600
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.532763532763533e-06,
+      "loss": 1.3216,
+      "step": 92700
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5311807534029756e-06,
+      "loss": 1.327,
+      "step": 92800
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5295979740424187e-06,
+      "loss": 1.3605,
+      "step": 92900
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5280151946818613e-06,
+      "loss": 1.3544,
+      "step": 93000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5264324153213048e-06,
+      "loss": 1.3404,
+      "step": 93100
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5248496359607474e-06,
+      "loss": 1.3455,
+      "step": 93200
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5232668566001904e-06,
+      "loss": 1.3543,
+      "step": 93300
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.521684077239633e-06,
+      "loss": 1.3596,
+      "step": 93400
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.520101297879076e-06,
+      "loss": 1.331,
+      "step": 93500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5185185185185187e-06,
+      "loss": 1.3409,
+      "step": 93600
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.516935739157962e-06,
+      "loss": 1.3539,
+      "step": 93700
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5153529597974044e-06,
+      "loss": 1.338,
+      "step": 93800
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.5137701804368475e-06,
+      "loss": 1.3295,
+      "step": 93900
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.51218740107629e-06,
+      "loss": 1.3394,
+      "step": 94000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.510604621715733e-06,
+      "loss": 1.3233,
+      "step": 94100
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.5090218423551758e-06,
+      "loss": 1.3655,
+      "step": 94200
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.507439062994619e-06,
+      "loss": 1.3424,
+      "step": 94300
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.5058562836340614e-06,
+      "loss": 1.3325,
+      "step": 94400
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5042735042735045e-06,
+      "loss": 1.3422,
+      "step": 94500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.502690724912947e-06,
+      "loss": 1.3492,
+      "step": 94600
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5011079455523906e-06,
+      "loss": 1.3447,
+      "step": 94700
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.499525166191833e-06,
+      "loss": 1.3561,
+      "step": 94800
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.4979423868312762e-06,
+      "loss": 1.2977,
+      "step": 94900
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.496359607470719e-06,
+      "loss": 1.3085,
+      "step": 95000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.494776828110162e-06,
+      "loss": 1.3448,
+      "step": 95100
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4931940487496045e-06,
+      "loss": 1.328,
+      "step": 95200
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4916112693890476e-06,
+      "loss": 1.3417,
+      "step": 95300
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4900284900284902e-06,
+      "loss": 1.3226,
+      "step": 95400
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4884457106679333e-06,
+      "loss": 1.3326,
+      "step": 95500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.486862931307376e-06,
+      "loss": 1.3389,
+      "step": 95600
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.485280151946819e-06,
+      "loss": 1.342,
+      "step": 95700
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4836973725862616e-06,
+      "loss": 1.3514,
+      "step": 95800
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4821145932257046e-06,
+      "loss": 1.3468,
+      "step": 95900
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4805318138651472e-06,
+      "loss": 1.3344,
+      "step": 96000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4789490345045903e-06,
+      "loss": 1.3234,
+      "step": 96100
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4773662551440333e-06,
+      "loss": 1.3577,
+      "step": 96200
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4757834757834764e-06,
+      "loss": 1.324,
+      "step": 96300
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.474200696422919e-06,
+      "loss": 1.347,
+      "step": 96400
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.472617917062362e-06,
+      "loss": 1.3533,
+      "step": 96500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4710351377018047e-06,
+      "loss": 1.3493,
+      "step": 96600
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4694523583412477e-06,
+      "loss": 1.3317,
+      "step": 96700
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4678695789806903e-06,
+      "loss": 1.3557,
+      "step": 96800
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4662867996201334e-06,
+      "loss": 1.3582,
+      "step": 96900
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.464704020259576e-06,
+      "loss": 1.3431,
+      "step": 97000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.463121240899019e-06,
+      "loss": 1.372,
+      "step": 97100
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4615384615384617e-06,
+      "loss": 1.3122,
+      "step": 97200
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4599556821779047e-06,
+      "loss": 1.356,
+      "step": 97300
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4583729028173474e-06,
+      "loss": 1.3469,
+      "step": 97400
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4567901234567904e-06,
+      "loss": 1.3448,
+      "step": 97500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.455207344096233e-06,
+      "loss": 1.352,
+      "step": 97600
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4536245647356757e-06,
+      "loss": 1.3369,
+      "step": 97700
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.452041785375119e-06,
+      "loss": 1.3427,
+      "step": 97800
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.450459006014562e-06,
+      "loss": 1.3204,
+      "step": 97900
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.448876226654005e-06,
+      "loss": 1.3367,
+      "step": 98000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.447293447293448e-06,
+      "loss": 1.3189,
+      "step": 98100
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4457106679328905e-06,
+      "loss": 1.3274,
+      "step": 98200
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.4441278885723335e-06,
+      "loss": 1.3206,
+      "step": 98300
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.442545109211776e-06,
+      "loss": 1.3636,
+      "step": 98400
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.440962329851219e-06,
+      "loss": 1.3236,
+      "step": 98500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.439379550490662e-06,
+      "loss": 1.3303,
+      "step": 98600
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.437796771130105e-06,
+      "loss": 1.346,
+      "step": 98700
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.4362139917695475e-06,
+      "loss": 1.3576,
+      "step": 98800
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4346312124089905e-06,
+      "loss": 1.3425,
+      "step": 98900
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.433048433048433e-06,
+      "loss": 1.3458,
+      "step": 99000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.431465653687876e-06,
+      "loss": 1.319,
+      "step": 99100
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.429882874327319e-06,
+      "loss": 1.3236,
+      "step": 99200
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4283000949667615e-06,
+      "loss": 1.3306,
+      "step": 99300
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.426717315606205e-06,
+      "loss": 1.3295,
+      "step": 99400
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.425134536245648e-06,
+      "loss": 1.3378,
+      "step": 99500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4235517568850906e-06,
+      "loss": 1.3242,
+      "step": 99600
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4219689775245337e-06,
+      "loss": 1.3366,
+      "step": 99700
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4203861981639763e-06,
+      "loss": 1.3188,
+      "step": 99800
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4188034188034193e-06,
+      "loss": 1.3388,
+      "step": 99900
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.417220639442862e-06,
+      "loss": 1.3445,
+      "step": 100000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.415637860082305e-06,
+      "loss": 1.3382,
+      "step": 100100
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.4140550807217476e-06,
+      "loss": 1.3336,
+      "step": 100200
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.4124723013611902e-06,
+      "loss": 1.3389,
+      "step": 100300
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.4108895220006333e-06,
+      "loss": 1.3076,
+      "step": 100400
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.409306742640076e-06,
+      "loss": 1.3149,
+      "step": 100500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.407723963279519e-06,
+      "loss": 1.3258,
+      "step": 100600
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.4061411839189616e-06,
+      "loss": 1.3366,
+      "step": 100700
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.4045584045584046e-06,
+      "loss": 1.3558,
+      "step": 100800
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.402975625197848e-06,
+      "loss": 1.3258,
+      "step": 100900
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.4013928458372907e-06,
+      "loss": 1.3214,
+      "step": 101000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.3998100664767338e-06,
+      "loss": 1.3206,
+      "step": 101100
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.3982272871161764e-06,
+      "loss": 1.344,
+      "step": 101200
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.3966445077556195e-06,
+      "loss": 1.3272,
+      "step": 101300
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.395061728395062e-06,
+      "loss": 1.345,
+      "step": 101400
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.393478949034505e-06,
+      "loss": 1.3512,
+      "step": 101500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3918961696739478e-06,
+      "loss": 1.3251,
+      "step": 101600
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3903133903133904e-06,
+      "loss": 1.3308,
+      "step": 101700
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3887306109528334e-06,
+      "loss": 1.3093,
+      "step": 101800
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.387147831592276e-06,
+      "loss": 1.3604,
+      "step": 101900
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.385565052231719e-06,
+      "loss": 1.333,
+      "step": 102000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3839822728711617e-06,
+      "loss": 1.323,
+      "step": 102100
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3823994935106048e-06,
+      "loss": 1.3351,
+      "step": 102200
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3808167141500474e-06,
+      "loss": 1.3332,
+      "step": 102300
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3792339347894904e-06,
+      "loss": 1.3592,
+      "step": 102400
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.377651155428934e-06,
+      "loss": 1.3237,
+      "step": 102500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3760683760683765e-06,
+      "loss": 1.3622,
+      "step": 102600
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.3744855967078196e-06,
+      "loss": 1.3377,
+      "step": 102700
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.372902817347262e-06,
+      "loss": 1.3373,
+      "step": 102800
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.371320037986705e-06,
+      "loss": 1.3233,
+      "step": 102900
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.369737258626148e-06,
+      "loss": 1.3689,
+      "step": 103000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.3681544792655905e-06,
+      "loss": 1.3075,
+      "step": 103100
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.3665716999050336e-06,
+      "loss": 1.3618,
+      "step": 103200
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.364988920544476e-06,
+      "loss": 1.3438,
+      "step": 103300
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.3634061411839192e-06,
+      "loss": 1.3298,
+      "step": 103400
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.361823361823362e-06,
+      "loss": 1.3454,
+      "step": 103500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.360240582462805e-06,
+      "loss": 1.3319,
+      "step": 103600
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.3586578031022475e-06,
+      "loss": 1.3207,
+      "step": 103700
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.3570750237416906e-06,
+      "loss": 1.3493,
+      "step": 103800
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.355492244381133e-06,
+      "loss": 1.3293,
+      "step": 103900
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.3539094650205767e-06,
+      "loss": 1.32,
+      "step": 104000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.3523266856600197e-06,
+      "loss": 1.3179,
+      "step": 104100
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.3507439062994623e-06,
+      "loss": 1.3528,
+      "step": 104200
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.349161126938905e-06,
+      "loss": 1.3387,
+      "step": 104300
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.347578347578348e-06,
+      "loss": 1.3413,
+      "step": 104400
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.3459955682177906e-06,
+      "loss": 1.3168,
+      "step": 104500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.3444127888572337e-06,
+      "loss": 1.3271,
+      "step": 104600
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.3428300094966763e-06,
+      "loss": 1.3384,
+      "step": 104700
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.3412472301361194e-06,
+      "loss": 1.3482,
+      "step": 104800
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.339664450775562e-06,
+      "loss": 1.3239,
+      "step": 104900
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.338081671415005e-06,
+      "loss": 1.3381,
+      "step": 105000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.3364988920544477e-06,
+      "loss": 1.3379,
+      "step": 105100
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3349161126938907e-06,
+      "loss": 1.3056,
+      "step": 105200
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 1.3402,
+      "step": 105300
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3317505539727764e-06,
+      "loss": 1.3289,
+      "step": 105400
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.330167774612219e-06,
+      "loss": 1.3045,
+      "step": 105500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3285849952516625e-06,
+      "loss": 1.3117,
+      "step": 105600
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.327002215891105e-06,
+      "loss": 1.314,
+      "step": 105700
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.325419436530548e-06,
+      "loss": 1.355,
+      "step": 105800
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3238366571699908e-06,
+      "loss": 1.3459,
+      "step": 105900
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.322253877809434e-06,
+      "loss": 1.3268,
+      "step": 106000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3206710984488764e-06,
+      "loss": 1.3285,
+      "step": 106100
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3190883190883195e-06,
+      "loss": 1.3304,
+      "step": 106200
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.317505539727762e-06,
+      "loss": 1.3438,
+      "step": 106300
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.315922760367205e-06,
+      "loss": 1.3257,
+      "step": 106400
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.3143399810066478e-06,
+      "loss": 1.3216,
+      "step": 106500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.312757201646091e-06,
+      "loss": 1.3075,
+      "step": 106600
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.3111744222855335e-06,
+      "loss": 1.3277,
+      "step": 106700
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.3095916429249765e-06,
+      "loss": 1.3688,
+      "step": 106800
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.308008863564419e-06,
+      "loss": 1.3486,
+      "step": 106900
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.306426084203862e-06,
+      "loss": 1.3362,
+      "step": 107000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.304843304843305e-06,
+      "loss": 1.3165,
+      "step": 107100
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3032605254827483e-06,
+      "loss": 1.324,
+      "step": 107200
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.301677746122191e-06,
+      "loss": 1.3148,
+      "step": 107300
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.300094966761634e-06,
+      "loss": 1.3415,
+      "step": 107400
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.2985121874010766e-06,
+      "loss": 1.326,
+      "step": 107500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.2969294080405196e-06,
+      "loss": 1.3257,
+      "step": 107600
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.2953466286799622e-06,
+      "loss": 1.3203,
+      "step": 107700
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2937638493194053e-06,
+      "loss": 1.312,
+      "step": 107800
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.292181069958848e-06,
+      "loss": 1.3526,
+      "step": 107900
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.290598290598291e-06,
+      "loss": 1.3468,
+      "step": 108000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2890155112377336e-06,
+      "loss": 1.3067,
+      "step": 108100
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2874327318771766e-06,
+      "loss": 1.3456,
+      "step": 108200
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2858499525166193e-06,
+      "loss": 1.3207,
+      "step": 108300
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.2842671731560623e-06,
+      "loss": 1.3252,
+      "step": 108400
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.282684393795505e-06,
+      "loss": 1.3169,
+      "step": 108500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.281101614434948e-06,
+      "loss": 1.3476,
+      "step": 108600
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.279518835074391e-06,
+      "loss": 1.3189,
+      "step": 108700
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.277936055713834e-06,
+      "loss": 1.3409,
+      "step": 108800
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.2763532763532767e-06,
+      "loss": 1.3318,
+      "step": 108900
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.2747704969927197e-06,
+      "loss": 1.3365,
+      "step": 109000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.2731877176321624e-06,
+      "loss": 1.3341,
+      "step": 109100
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.2716049382716054e-06,
+      "loss": 1.324,
+      "step": 109200
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.270022158911048e-06,
+      "loss": 1.326,
+      "step": 109300
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.268439379550491e-06,
+      "loss": 1.3212,
+      "step": 109400
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.2668566001899337e-06,
+      "loss": 1.3492,
+      "step": 109500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.2652738208293768e-06,
+      "loss": 1.3452,
+      "step": 109600
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.2636910414688194e-06,
+      "loss": 1.3287,
+      "step": 109700
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.2621082621082624e-06,
+      "loss": 1.3424,
+      "step": 109800
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.260525482747705e-06,
+      "loss": 1.3312,
+      "step": 109900
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.258942703387148e-06,
+      "loss": 1.3166,
+      "step": 110000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.2573599240265907e-06,
+      "loss": 1.309,
+      "step": 110100
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.2557771446660334e-06,
+      "loss": 1.3397,
+      "step": 110200
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.254194365305477e-06,
+      "loss": 1.3423,
+      "step": 110300
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.25261158594492e-06,
+      "loss": 1.3385,
+      "step": 110400
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.2510288065843625e-06,
+      "loss": 1.3396,
+      "step": 110500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.2494460272238055e-06,
+      "loss": 1.3332,
+      "step": 110600
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.247863247863248e-06,
+      "loss": 1.3321,
+      "step": 110700
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.2462804685026912e-06,
+      "loss": 1.3402,
+      "step": 110800
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.244697689142134e-06,
+      "loss": 1.3369,
+      "step": 110900
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.243114909781577e-06,
+      "loss": 1.3383,
+      "step": 111000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.2415321304210195e-06,
+      "loss": 1.3591,
+      "step": 111100
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.2399493510604626e-06,
+      "loss": 1.3379,
+      "step": 111200
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.238366571699905e-06,
+      "loss": 1.3493,
+      "step": 111300
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.2367837923393482e-06,
+      "loss": 1.3709,
+      "step": 111400
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.235201012978791e-06,
+      "loss": 1.3163,
+      "step": 111500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.2336182336182335e-06,
+      "loss": 1.34,
+      "step": 111600
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.2320354542576765e-06,
+      "loss": 1.3182,
+      "step": 111700
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.230452674897119e-06,
+      "loss": 1.3316,
+      "step": 111800
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.2288698955365626e-06,
+      "loss": 1.3312,
+      "step": 111900
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.2272871161760057e-06,
+      "loss": 1.3393,
+      "step": 112000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.2257043368154483e-06,
+      "loss": 1.3649,
+      "step": 112100
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.2241215574548913e-06,
+      "loss": 1.3315,
+      "step": 112200
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.222538778094334e-06,
+      "loss": 1.3465,
+      "step": 112300
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.220955998733777e-06,
+      "loss": 1.3265,
+      "step": 112400
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.2193732193732196e-06,
+      "loss": 1.3435,
+      "step": 112500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.2177904400126627e-06,
+      "loss": 1.3268,
+      "step": 112600
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.2162076606521053e-06,
+      "loss": 1.345,
+      "step": 112700
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.214624881291548e-06,
+      "loss": 1.3492,
+      "step": 112800
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.213042101930991e-06,
+      "loss": 1.3159,
+      "step": 112900
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.2114593225704336e-06,
+      "loss": 1.3553,
+      "step": 113000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.2098765432098767e-06,
+      "loss": 1.3485,
+      "step": 113100
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.2082937638493193e-06,
+      "loss": 1.3291,
+      "step": 113200
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.2067109844887623e-06,
+      "loss": 1.3574,
+      "step": 113300
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.205128205128206e-06,
+      "loss": 1.3107,
+      "step": 113400
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.2035454257676484e-06,
+      "loss": 1.339,
+      "step": 113500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.2019626464070915e-06,
+      "loss": 1.3372,
+      "step": 113600
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.200379867046534e-06,
+      "loss": 1.3562,
+      "step": 113700
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.198797087685977e-06,
+      "loss": 1.3162,
+      "step": 113800
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.1972143083254198e-06,
+      "loss": 1.3386,
+      "step": 113900
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.195631528964863e-06,
+      "loss": 1.3453,
+      "step": 114000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.1940487496043054e-06,
+      "loss": 1.3481,
+      "step": 114100
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.192465970243748e-06,
+      "loss": 1.3414,
+      "step": 114200
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.190883190883191e-06,
+      "loss": 1.3339,
+      "step": 114300
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.1893004115226337e-06,
+      "loss": 1.3333,
+      "step": 114400
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.187717632162077e-06,
+      "loss": 1.3391,
+      "step": 114500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.1861348528015194e-06,
+      "loss": 1.3127,
+      "step": 114600
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.1845520734409625e-06,
+      "loss": 1.3352,
+      "step": 114700
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.182969294080405e-06,
+      "loss": 1.3104,
+      "step": 114800
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.181386514719848e-06,
+      "loss": 1.3151,
+      "step": 114900
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.1798037353592916e-06,
+      "loss": 1.3351,
+      "step": 115000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.1782209559987342e-06,
+      "loss": 1.3245,
+      "step": 115100
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.1766381766381773e-06,
+      "loss": 1.3054,
+      "step": 115200
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.17505539727762e-06,
+      "loss": 1.326,
+      "step": 115300
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.1734726179170625e-06,
+      "loss": 1.3504,
+      "step": 115400
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.1718898385565056e-06,
+      "loss": 1.3345,
+      "step": 115500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.170307059195948e-06,
+      "loss": 1.3097,
+      "step": 115600
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.1687242798353912e-06,
+      "loss": 1.3416,
+      "step": 115700
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.167141500474834e-06,
+      "loss": 1.3275,
+      "step": 115800
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.165558721114277e-06,
+      "loss": 1.318,
+      "step": 115900
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1639759417537195e-06,
+      "loss": 1.3182,
+      "step": 116000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1623931623931626e-06,
+      "loss": 1.3638,
+      "step": 116100
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1608103830326052e-06,
+      "loss": 1.3402,
+      "step": 116200
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1592276036720483e-06,
+      "loss": 1.3261,
+      "step": 116300
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.157644824311491e-06,
+      "loss": 1.3304,
+      "step": 116400
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.156062044950934e-06,
+      "loss": 1.3446,
+      "step": 116500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1544792655903774e-06,
+      "loss": 1.3422,
+      "step": 116600
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.15289648622982e-06,
+      "loss": 1.332,
+      "step": 116700
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1513137068692627e-06,
+      "loss": 1.3041,
+      "step": 116800
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1497309275087057e-06,
+      "loss": 1.3017,
+      "step": 116900
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1481481481481483e-06,
+      "loss": 1.3405,
+      "step": 117000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1465653687875914e-06,
+      "loss": 1.3474,
+      "step": 117100
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.144982589427034e-06,
+      "loss": 1.333,
+      "step": 117200
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.143399810066477e-06,
+      "loss": 1.325,
+      "step": 117300
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.1418170307059197e-06,
+      "loss": 1.3349,
+      "step": 117400
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.1402342513453627e-06,
+      "loss": 1.3099,
+      "step": 117500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.1386514719848053e-06,
+      "loss": 1.3403,
+      "step": 117600
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.1370686926242484e-06,
+      "loss": 1.3311,
+      "step": 117700
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.135485913263691e-06,
+      "loss": 1.3707,
+      "step": 117800
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.133903133903134e-06,
+      "loss": 1.3376,
+      "step": 117900
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1323203545425767e-06,
+      "loss": 1.3179,
+      "step": 118000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.13073757518202e-06,
+      "loss": 1.3404,
+      "step": 118100
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1291547958214628e-06,
+      "loss": 1.3039,
+      "step": 118200
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.127572016460906e-06,
+      "loss": 1.3341,
+      "step": 118300
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1259892371003485e-06,
+      "loss": 1.3285,
+      "step": 118400
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.1244064577397915e-06,
+      "loss": 1.3153,
+      "step": 118500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.122823678379234e-06,
+      "loss": 1.3395,
+      "step": 118600
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.121240899018677e-06,
+      "loss": 1.326,
+      "step": 118700
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.11965811965812e-06,
+      "loss": 1.3294,
+      "step": 118800
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.118075340297563e-06,
+      "loss": 1.3103,
+      "step": 118900
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.1164925609370055e-06,
+      "loss": 1.3332,
+      "step": 119000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.1149097815764485e-06,
+      "loss": 1.3203,
+      "step": 119100
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.113327002215891e-06,
+      "loss": 1.3298,
+      "step": 119200
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.111744222855334e-06,
+      "loss": 1.328,
+      "step": 119300
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.110161443494777e-06,
+      "loss": 1.332,
+      "step": 119400
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.10857866413422e-06,
+      "loss": 1.3562,
+      "step": 119500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.1069958847736625e-06,
+      "loss": 1.3013,
+      "step": 119600
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.105413105413106e-06,
+      "loss": 1.3436,
+      "step": 119700
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.1038303260525486e-06,
+      "loss": 1.3275,
+      "step": 119800
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.1022475466919916e-06,
+      "loss": 1.3135,
+      "step": 119900
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.1006647673314343e-06,
+      "loss": 1.3158,
+      "step": 120000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.0990819879708773e-06,
+      "loss": 1.3128,
+      "step": 120100
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.09749920861032e-06,
+      "loss": 1.3119,
+      "step": 120200
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.095916429249763e-06,
+      "loss": 1.3106,
+      "step": 120300
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.0943336498892056e-06,
+      "loss": 1.3477,
+      "step": 120400
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.0927508705286487e-06,
+      "loss": 1.3346,
+      "step": 120500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.0911680911680913e-06,
+      "loss": 1.3455,
+      "step": 120600
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.0895853118075343e-06,
+      "loss": 1.3575,
+      "step": 120700
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.088002532446977e-06,
+      "loss": 1.3338,
+      "step": 120800
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.08641975308642e-06,
+      "loss": 1.331,
+      "step": 120900
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.0848369737258626e-06,
+      "loss": 1.3082,
+      "step": 121000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.0832541943653057e-06,
+      "loss": 1.3121,
+      "step": 121100
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.0816714150047487e-06,
+      "loss": 1.324,
+      "step": 121200
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.0800886356441918e-06,
+      "loss": 1.327,
+      "step": 121300
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.0785058562836344e-06,
+      "loss": 1.3438,
+      "step": 121400
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 1.3315,
+      "step": 121500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.07534029756252e-06,
+      "loss": 1.3018,
+      "step": 121600
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.073757518201963e-06,
+      "loss": 1.3402,
+      "step": 121700
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.0721747388414057e-06,
+      "loss": 1.3302,
+      "step": 121800
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.0705919594808488e-06,
+      "loss": 1.3526,
+      "step": 121900
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.0690091801202914e-06,
+      "loss": 1.3117,
+      "step": 122000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.0674264007597345e-06,
+      "loss": 1.3181,
+      "step": 122100
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.065843621399177e-06,
+      "loss": 1.317,
+      "step": 122200
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.06426084203862e-06,
+      "loss": 1.3328,
+      "step": 122300
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.0626780626780627e-06,
+      "loss": 1.3485,
+      "step": 122400
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.061095283317506e-06,
+      "loss": 1.3157,
+      "step": 122500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.0595125039569484e-06,
+      "loss": 1.3191,
+      "step": 122600
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.0579297245963915e-06,
+      "loss": 1.3431,
+      "step": 122700
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.0563469452358345e-06,
+      "loss": 1.3507,
+      "step": 122800
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0547641658752776e-06,
+      "loss": 1.329,
+      "step": 122900
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.05318138651472e-06,
+      "loss": 1.3124,
+      "step": 123000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0515986071541632e-06,
+      "loss": 1.3227,
+      "step": 123100
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.050015827793606e-06,
+      "loss": 1.3298,
+      "step": 123200
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.048433048433049e-06,
+      "loss": 1.3061,
+      "step": 123300
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0468502690724915e-06,
+      "loss": 1.3505,
+      "step": 123400
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0452674897119346e-06,
+      "loss": 1.3581,
+      "step": 123500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.043684710351377e-06,
+      "loss": 1.3231,
+      "step": 123600
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0421019309908203e-06,
+      "loss": 1.3397,
+      "step": 123700
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.040519151630263e-06,
+      "loss": 1.3816,
+      "step": 123800
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.038936372269706e-06,
+      "loss": 1.346,
+      "step": 123900
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0373535929091486e-06,
+      "loss": 1.3171,
+      "step": 124000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.035770813548591e-06,
+      "loss": 1.3305,
+      "step": 124100
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0341880341880342e-06,
+      "loss": 1.3195,
+      "step": 124200
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.032605254827477e-06,
+      "loss": 1.3489,
+      "step": 124300
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0310224754669203e-06,
+      "loss": 1.3424,
+      "step": 124400
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0294396961063634e-06,
+      "loss": 1.3456,
+      "step": 124500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.027856916745806e-06,
+      "loss": 1.3357,
+      "step": 124600
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.026274137385249e-06,
+      "loss": 1.3321,
+      "step": 124700
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0246913580246917e-06,
+      "loss": 1.3198,
+      "step": 124800
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0231085786641347e-06,
+      "loss": 1.3266,
+      "step": 124900
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0215257993035773e-06,
+      "loss": 1.3294,
+      "step": 125000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0199430199430204e-06,
+      "loss": 1.3104,
+      "step": 125100
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.018360240582463e-06,
+      "loss": 1.352,
+      "step": 125200
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.016777461221906e-06,
+      "loss": 1.3145,
+      "step": 125300
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0151946818613487e-06,
+      "loss": 1.3587,
+      "step": 125400
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0136119025007913e-06,
+      "loss": 1.3504,
+      "step": 125500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0120291231402344e-06,
+      "loss": 1.3412,
+      "step": 125600
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.010446343779677e-06,
+      "loss": 1.3132,
+      "step": 125700
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.00886356441912e-06,
+      "loss": 1.3386,
+      "step": 125800
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0072807850585635e-06,
+      "loss": 1.3341,
+      "step": 125900
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.005698005698006e-06,
+      "loss": 1.3187,
+      "step": 126000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.004115226337449e-06,
+      "loss": 1.326,
+      "step": 126100
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.002532446976892e-06,
+      "loss": 1.3339,
+      "step": 126200
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.000949667616335e-06,
+      "loss": 1.3504,
+      "step": 126300
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.9993668882557775e-06,
+      "loss": 1.3061,
+      "step": 126400
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.9977841088952205e-06,
+      "loss": 1.2796,
+      "step": 126500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.996201329534663e-06,
+      "loss": 1.313,
+      "step": 126600
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.9946185501741058e-06,
+      "loss": 1.2852,
+      "step": 126700
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.993035770813549e-06,
+      "loss": 1.2886,
+      "step": 126800
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.9914529914529914e-06,
+      "loss": 1.2743,
+      "step": 126900
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.9898702120924345e-06,
+      "loss": 1.2901,
+      "step": 127000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.988287432731877e-06,
+      "loss": 1.281,
+      "step": 127100
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.98670465337132e-06,
+      "loss": 1.2892,
+      "step": 127200
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.9851218740107628e-06,
+      "loss": 1.2896,
+      "step": 127300
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.983539094650206e-06,
+      "loss": 1.2983,
+      "step": 127400
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.9819563152896493e-06,
+      "loss": 1.2872,
+      "step": 127500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.980373535929092e-06,
+      "loss": 1.2886,
+      "step": 127600
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.978790756568535e-06,
+      "loss": 1.282,
+      "step": 127700
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.9772079772079776e-06,
+      "loss": 1.2946,
+      "step": 127800
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.9756251978474206e-06,
+      "loss": 1.2729,
+      "step": 127900
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.9740424184868633e-06,
+      "loss": 1.2968,
+      "step": 128000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.972459639126306e-06,
+      "loss": 1.2882,
+      "step": 128100
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.970876859765749e-06,
+      "loss": 1.2777,
+      "step": 128200
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.9692940804051916e-06,
+      "loss": 1.3108,
+      "step": 128300
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.9677113010446346e-06,
+      "loss": 1.3069,
+      "step": 128400
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.9661285216840772e-06,
+      "loss": 1.2808,
+      "step": 128500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.9645457423235203e-06,
+      "loss": 1.2905,
+      "step": 128600
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.962962962962963e-06,
+      "loss": 1.3046,
+      "step": 128700
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.961380183602406e-06,
+      "loss": 1.2843,
+      "step": 128800
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.9597974042418486e-06,
+      "loss": 1.2923,
+      "step": 128900
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.9582146248812916e-06,
+      "loss": 1.2785,
+      "step": 129000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.956631845520735e-06,
+      "loss": 1.293,
+      "step": 129100
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.9550490661601777e-06,
+      "loss": 1.3079,
+      "step": 129200
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.9534662867996203e-06,
+      "loss": 1.2952,
+      "step": 129300
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.9518835074390634e-06,
+      "loss": 1.3053,
+      "step": 129400
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.950300728078506e-06,
+      "loss": 1.3084,
+      "step": 129500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.948717948717949e-06,
+      "loss": 1.3001,
+      "step": 129600
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.9471351693573917e-06,
+      "loss": 1.2768,
+      "step": 129700
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.9455523899968347e-06,
+      "loss": 1.2784,
+      "step": 129800
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.9439696106362774e-06,
+      "loss": 1.2952,
+      "step": 129900
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.9423868312757204e-06,
+      "loss": 1.2591,
+      "step": 130000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.940804051915163e-06,
+      "loss": 1.2895,
+      "step": 130100
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.939221272554606e-06,
+      "loss": 1.2566,
+      "step": 130200
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.9376384931940487e-06,
+      "loss": 1.2555,
+      "step": 130300
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.9360557138334918e-06,
+      "loss": 1.2803,
+      "step": 130400
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.9344729344729344e-06,
+      "loss": 1.3069,
+      "step": 130500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.932890155112378e-06,
+      "loss": 1.3106,
+      "step": 130600
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.9313073757518205e-06,
+      "loss": 1.2818,
+      "step": 130700
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.9297245963912635e-06,
+      "loss": 1.2884,
+      "step": 130800
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.928141817030706e-06,
+      "loss": 1.2982,
+      "step": 130900
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.926559037670149e-06,
+      "loss": 1.3007,
+      "step": 131000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.924976258309592e-06,
+      "loss": 1.2833,
+      "step": 131100
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.923393478949035e-06,
+      "loss": 1.3145,
+      "step": 131200
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.9218106995884775e-06,
+      "loss": 1.2746,
+      "step": 131300
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.9202279202279205e-06,
+      "loss": 1.2941,
+      "step": 131400
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.918645140867363e-06,
+      "loss": 1.2643,
+      "step": 131500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.9170623615068062e-06,
+      "loss": 1.2868,
+      "step": 131600
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.915479582146249e-06,
+      "loss": 1.2955,
+      "step": 131700
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.913896802785692e-06,
+      "loss": 1.2999,
+      "step": 131800
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.9123140234251345e-06,
+      "loss": 1.2886,
+      "step": 131900
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.9107312440645776e-06,
+      "loss": 1.2749,
+      "step": 132000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.90914846470402e-06,
+      "loss": 1.2916,
+      "step": 132100
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.9075656853434637e-06,
+      "loss": 1.2621,
+      "step": 132200
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.9059829059829063e-06,
+      "loss": 1.2701,
+      "step": 132300
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.9044001266223493e-06,
+      "loss": 1.3084,
+      "step": 132400
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.902817347261792e-06,
+      "loss": 1.3103,
+      "step": 132500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.901234567901235e-06,
+      "loss": 1.2679,
+      "step": 132600
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.8996517885406776e-06,
+      "loss": 1.2784,
+      "step": 132700
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.8980690091801207e-06,
+      "loss": 1.286,
+      "step": 132800
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.8964862298195633e-06,
+      "loss": 1.3071,
+      "step": 132900
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.8949034504590063e-06,
+      "loss": 1.281,
+      "step": 133000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.893320671098449e-06,
+      "loss": 1.2711,
+      "step": 133100
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.891737891737892e-06,
+      "loss": 1.3096,
+      "step": 133200
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.8901551123773346e-06,
+      "loss": 1.2531,
+      "step": 133300
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.8885723330167777e-06,
+      "loss": 1.2893,
+      "step": 133400
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.8869895536562203e-06,
+      "loss": 1.3043,
+      "step": 133500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.8854067742956634e-06,
+      "loss": 1.2917,
+      "step": 133600
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.883823994935106e-06,
+      "loss": 1.2718,
+      "step": 133700
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.8822412155745495e-06,
+      "loss": 1.2837,
+      "step": 133800
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.880658436213992e-06,
+      "loss": 1.2853,
+      "step": 133900
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.879075656853435e-06,
+      "loss": 1.2695,
+      "step": 134000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.8774928774928778e-06,
+      "loss": 1.2764,
+      "step": 134100
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.875910098132321e-06,
+      "loss": 1.2888,
+      "step": 134200
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.8743273187717634e-06,
+      "loss": 1.2961,
+      "step": 134300
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.8727445394112065e-06,
+      "loss": 1.2867,
+      "step": 134400
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.871161760050649e-06,
+      "loss": 1.2587,
+      "step": 134500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.869578980690092e-06,
+      "loss": 1.2934,
+      "step": 134600
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.8679962013295348e-06,
+      "loss": 1.2906,
+      "step": 134700
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.866413421968978e-06,
+      "loss": 1.3073,
+      "step": 134800
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.8648306426084204e-06,
+      "loss": 1.2593,
+      "step": 134900
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.8632478632478635e-06,
+      "loss": 1.2991,
+      "step": 135000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.861665083887306e-06,
+      "loss": 1.3139,
+      "step": 135100
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.860082304526749e-06,
+      "loss": 1.2677,
+      "step": 135200
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.858499525166192e-06,
+      "loss": 1.2836,
+      "step": 135300
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.8569167458056353e-06,
+      "loss": 1.299,
+      "step": 135400
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.855333966445078e-06,
+      "loss": 1.2876,
+      "step": 135500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.853751187084521e-06,
+      "loss": 1.2551,
+      "step": 135600
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.8521684077239636e-06,
+      "loss": 1.285,
+      "step": 135700
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.8505856283634066e-06,
+      "loss": 1.2765,
+      "step": 135800
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.8490028490028492e-06,
+      "loss": 1.2646,
+      "step": 135900
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.8474200696422923e-06,
+      "loss": 1.2749,
+      "step": 136000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.845837290281735e-06,
+      "loss": 1.3016,
+      "step": 136100
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.844254510921178e-06,
+      "loss": 1.2995,
+      "step": 136200
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.8426717315606206e-06,
+      "loss": 1.3023,
+      "step": 136300
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.8410889522000636e-06,
+      "loss": 1.2772,
+      "step": 136400
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.8395061728395062e-06,
+      "loss": 1.2784,
+      "step": 136500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.837923393478949e-06,
+      "loss": 1.3212,
+      "step": 136600
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.836340614118392e-06,
+      "loss": 1.2975,
+      "step": 136700
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.8347578347578345e-06,
+      "loss": 1.2954,
+      "step": 136800
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.833175055397278e-06,
+      "loss": 1.2626,
+      "step": 136900
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.831592276036721e-06,
+      "loss": 1.2924,
+      "step": 137000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.8300094966761637e-06,
+      "loss": 1.3133,
+      "step": 137100
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.8284267173156067e-06,
+      "loss": 1.2915,
+      "step": 137200
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.8268439379550494e-06,
+      "loss": 1.2693,
+      "step": 137300
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.8252611585944924e-06,
+      "loss": 1.2822,
+      "step": 137400
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.823678379233935e-06,
+      "loss": 1.2854,
+      "step": 137500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.822095599873378e-06,
+      "loss": 1.2886,
+      "step": 137600
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.8205128205128207e-06,
+      "loss": 1.2906,
+      "step": 137700
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.8189300411522637e-06,
+      "loss": 1.2743,
+      "step": 137800
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.8173472617917064e-06,
+      "loss": 1.2775,
+      "step": 137900
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.815764482431149e-06,
+      "loss": 1.283,
+      "step": 138000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.814181703070592e-06,
+      "loss": 1.2714,
+      "step": 138100
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.8125989237100347e-06,
+      "loss": 1.2827,
+      "step": 138200
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.8110161443494777e-06,
+      "loss": 1.2831,
+      "step": 138300
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.8094333649889203e-06,
+      "loss": 1.2901,
+      "step": 138400
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.807850585628364e-06,
+      "loss": 1.2779,
+      "step": 138500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.806267806267807e-06,
+      "loss": 1.3033,
+      "step": 138600
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.8046850269072495e-06,
+      "loss": 1.2836,
+      "step": 138700
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.8031022475466925e-06,
+      "loss": 1.2855,
+      "step": 138800
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.801519468186135e-06,
+      "loss": 1.2746,
+      "step": 138900
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.799936688825578e-06,
+      "loss": 1.2776,
+      "step": 139000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.798353909465021e-06,
+      "loss": 1.304,
+      "step": 139100
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.7967711301044635e-06,
+      "loss": 1.2999,
+      "step": 139200
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.7951883507439065e-06,
+      "loss": 1.2552,
+      "step": 139300
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.793605571383349e-06,
+      "loss": 1.2923,
+      "step": 139400
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.792022792022792e-06,
+      "loss": 1.2958,
+      "step": 139500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.790440012662235e-06,
+      "loss": 1.2625,
+      "step": 139600
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.788857233301678e-06,
+      "loss": 1.2961,
+      "step": 139700
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.7872744539411205e-06,
+      "loss": 1.2867,
+      "step": 139800
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.7856916745805635e-06,
+      "loss": 1.2944,
+      "step": 139900
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.784108895220007e-06,
+      "loss": 1.2867,
+      "step": 140000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.7825261158594496e-06,
+      "loss": 1.2793,
+      "step": 140100
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.7809433364988927e-06,
+      "loss": 1.2909,
+      "step": 140200
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.7793605571383353e-06,
+      "loss": 1.2997,
+      "step": 140300
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.7777777777777783e-06,
+      "loss": 1.2504,
+      "step": 140400
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.776194998417221e-06,
+      "loss": 1.2861,
+      "step": 140500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7746122190566636e-06,
+      "loss": 1.2919,
+      "step": 140600
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7730294396961066e-06,
+      "loss": 1.286,
+      "step": 140700
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7714466603355493e-06,
+      "loss": 1.2712,
+      "step": 140800
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7698638809749923e-06,
+      "loss": 1.2846,
+      "step": 140900
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.768281101614435e-06,
+      "loss": 1.2571,
+      "step": 141000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.766698322253878e-06,
+      "loss": 1.2722,
+      "step": 141100
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7651155428933206e-06,
+      "loss": 1.2952,
+      "step": 141200
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.7635327635327636e-06,
+      "loss": 1.2922,
+      "step": 141300
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.7619499841722063e-06,
+      "loss": 1.3144,
+      "step": 141400
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.7603672048116493e-06,
+      "loss": 1.2814,
+      "step": 141500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.758784425451093e-06,
+      "loss": 1.3146,
+      "step": 141600
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.7572016460905354e-06,
+      "loss": 1.2577,
+      "step": 141700
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.755618866729978e-06,
+      "loss": 1.2683,
+      "step": 141800
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.754036087369421e-06,
+      "loss": 1.291,
+      "step": 141900
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.7524533080088637e-06,
+      "loss": 1.2828,
+      "step": 142000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.7508705286483068e-06,
+      "loss": 1.2706,
+      "step": 142100
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.7492877492877494e-06,
+      "loss": 1.2598,
+      "step": 142200
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.7477049699271924e-06,
+      "loss": 1.3022,
+      "step": 142300
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.746122190566635e-06,
+      "loss": 1.291,
+      "step": 142400
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.744539411206078e-06,
+      "loss": 1.2797,
+      "step": 142500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.7429566318455207e-06,
+      "loss": 1.2605,
+      "step": 142600
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.7413738524849638e-06,
+      "loss": 1.3126,
+      "step": 142700
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.7397910731244064e-06,
+      "loss": 1.3085,
+      "step": 142800
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.7382082937638494e-06,
+      "loss": 1.2736,
+      "step": 142900
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.736625514403292e-06,
+      "loss": 1.2723,
+      "step": 143000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.7350427350427355e-06,
+      "loss": 1.284,
+      "step": 143100
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.733459955682178e-06,
+      "loss": 1.301,
+      "step": 143200
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.7318771763216212e-06,
+      "loss": 1.2705,
+      "step": 143300
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.730294396961064e-06,
+      "loss": 1.2826,
+      "step": 143400
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.728711617600507e-06,
+      "loss": 1.2732,
+      "step": 143500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.7271288382399495e-06,
+      "loss": 1.298,
+      "step": 143600
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.7255460588793926e-06,
+      "loss": 1.285,
+      "step": 143700
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.723963279518835e-06,
+      "loss": 1.2824,
+      "step": 143800
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.7223805001582782e-06,
+      "loss": 1.2705,
+      "step": 143900
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.720797720797721e-06,
+      "loss": 1.2845,
+      "step": 144000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.719214941437164e-06,
+      "loss": 1.2869,
+      "step": 144100
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.7176321620766065e-06,
+      "loss": 1.2995,
+      "step": 144200
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.7160493827160496e-06,
+      "loss": 1.3236,
+      "step": 144300
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.714466603355492e-06,
+      "loss": 1.2612,
+      "step": 144400
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7128838239949353e-06,
+      "loss": 1.3095,
+      "step": 144500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.711301044634378e-06,
+      "loss": 1.2815,
+      "step": 144600
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7097182652738213e-06,
+      "loss": 1.2872,
+      "step": 144700
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.708135485913264e-06,
+      "loss": 1.2812,
+      "step": 144800
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.706552706552707e-06,
+      "loss": 1.3119,
+      "step": 144900
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7049699271921496e-06,
+      "loss": 1.2929,
+      "step": 145000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7033871478315927e-06,
+      "loss": 1.284,
+      "step": 145100
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7018043684710353e-06,
+      "loss": 1.2944,
+      "step": 145200
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7002215891104784e-06,
+      "loss": 1.2904,
+      "step": 145300
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.698638809749921e-06,
+      "loss": 1.2871,
+      "step": 145400
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.697056030389364e-06,
+      "loss": 1.2858,
+      "step": 145500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6954732510288067e-06,
+      "loss": 1.2898,
+      "step": 145600
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6938904716682497e-06,
+      "loss": 1.2727,
+      "step": 145700
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6923076923076923e-06,
+      "loss": 1.2853,
+      "step": 145800
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6907249129471354e-06,
+      "loss": 1.2831,
+      "step": 145900
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.689142133586578e-06,
+      "loss": 1.2912,
+      "step": 146000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.687559354226021e-06,
+      "loss": 1.2902,
+      "step": 146100
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6859765748654637e-06,
+      "loss": 1.2999,
+      "step": 146200
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.684393795504907e-06,
+      "loss": 1.2977,
+      "step": 146300
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6828110161443498e-06,
+      "loss": 1.2701,
+      "step": 146400
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.681228236783793e-06,
+      "loss": 1.2808,
+      "step": 146500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6796454574232354e-06,
+      "loss": 1.291,
+      "step": 146600
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6780626780626785e-06,
+      "loss": 1.3003,
+      "step": 146700
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.676479898702121e-06,
+      "loss": 1.2885,
+      "step": 146800
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.674897119341564e-06,
+      "loss": 1.2894,
+      "step": 146900
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.673314339981007e-06,
+      "loss": 1.2799,
+      "step": 147000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.67173156062045e-06,
+      "loss": 1.2916,
+      "step": 147100
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.6701487812598925e-06,
+      "loss": 1.2769,
+      "step": 147200
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.6685660018993355e-06,
+      "loss": 1.259,
+      "step": 147300
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.666983222538778e-06,
+      "loss": 1.2908,
+      "step": 147400
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.665400443178221e-06,
+      "loss": 1.3013,
+      "step": 147500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.663817663817664e-06,
+      "loss": 1.2974,
+      "step": 147600
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.662234884457107e-06,
+      "loss": 1.281,
+      "step": 147700
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.66065210509655e-06,
+      "loss": 1.2776,
+      "step": 147800
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.659069325735993e-06,
+      "loss": 1.3079,
+      "step": 147900
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.6574865463754356e-06,
+      "loss": 1.2748,
+      "step": 148000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.6559037670148786e-06,
+      "loss": 1.2847,
+      "step": 148100
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.6543209876543212e-06,
+      "loss": 1.2881,
+      "step": 148200
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.6527382082937643e-06,
+      "loss": 1.2602,
+      "step": 148300
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.651155428933207e-06,
+      "loss": 1.2774,
+      "step": 148400
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.64957264957265e-06,
+      "loss": 1.2866,
+      "step": 148500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.6479898702120926e-06,
+      "loss": 1.2806,
+      "step": 148600
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.6464070908515356e-06,
+      "loss": 1.3143,
+      "step": 148700
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.6448243114909783e-06,
+      "loss": 1.2822,
+      "step": 148800
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.6432415321304213e-06,
+      "loss": 1.2634,
+      "step": 148900
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.641658752769864e-06,
+      "loss": 1.2712,
+      "step": 149000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.6400759734093066e-06,
+      "loss": 1.2949,
+      "step": 149100
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.6384931940487496e-06,
+      "loss": 1.2724,
+      "step": 149200
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.6369104146881922e-06,
+      "loss": 1.2675,
+      "step": 149300
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.6353276353276357e-06,
+      "loss": 1.2976,
+      "step": 149400
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.6337448559670788e-06,
+      "loss": 1.2862,
+      "step": 149500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.6321620766065214e-06,
+      "loss": 1.2641,
+      "step": 149600
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.6305792972459644e-06,
+      "loss": 1.2874,
+      "step": 149700
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.628996517885407e-06,
+      "loss": 1.2915,
+      "step": 149800
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.62741373852485e-06,
+      "loss": 1.2946,
+      "step": 149900
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.6258309591642927e-06,
+      "loss": 1.2815,
+      "step": 150000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.6242481798037358e-06,
+      "loss": 1.2842,
+      "step": 150100
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.6226654004431784e-06,
+      "loss": 1.2673,
+      "step": 150200
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.6210826210826214e-06,
+      "loss": 1.2714,
+      "step": 150300
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.619499841722064e-06,
+      "loss": 1.2804,
+      "step": 150400
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.6179170623615067e-06,
+      "loss": 1.2692,
+      "step": 150500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.6163342830009497e-06,
+      "loss": 1.255,
+      "step": 150600
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6147515036403924e-06,
+      "loss": 1.2839,
+      "step": 150700
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6131687242798354e-06,
+      "loss": 1.282,
+      "step": 150800
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.611585944919278e-06,
+      "loss": 1.2551,
+      "step": 150900
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6100031655587215e-06,
+      "loss": 1.279,
+      "step": 151000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6084203861981646e-06,
+      "loss": 1.2863,
+      "step": 151100
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.606837606837607e-06,
+      "loss": 1.2644,
+      "step": 151200
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6052548274770502e-06,
+      "loss": 1.2841,
+      "step": 151300
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.603672048116493e-06,
+      "loss": 1.2954,
+      "step": 151400
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.602089268755936e-06,
+      "loss": 1.2965,
+      "step": 151500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.6005064893953785e-06,
+      "loss": 1.2769,
+      "step": 151600
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.598923710034821e-06,
+      "loss": 1.3023,
+      "step": 151700
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.597340930674264e-06,
+      "loss": 1.3077,
+      "step": 151800
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.595758151313707e-06,
+      "loss": 1.2736,
+      "step": 151900
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.59417537195315e-06,
+      "loss": 1.3131,
+      "step": 152000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.5925925925925925e-06,
+      "loss": 1.3021,
+      "step": 152100
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.5910098132320355e-06,
+      "loss": 1.2775,
+      "step": 152200
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.589427033871478e-06,
+      "loss": 1.2894,
+      "step": 152300
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.587844254510921e-06,
+      "loss": 1.3002,
+      "step": 152400
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.5862614751503647e-06,
+      "loss": 1.2762,
+      "step": 152500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5846786957898073e-06,
+      "loss": 1.283,
+      "step": 152600
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5830959164292504e-06,
+      "loss": 1.2851,
+      "step": 152700
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.581513137068693e-06,
+      "loss": 1.289,
+      "step": 152800
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.579930357708136e-06,
+      "loss": 1.2804,
+      "step": 152900
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5783475783475787e-06,
+      "loss": 1.2968,
+      "step": 153000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5767647989870213e-06,
+      "loss": 1.2773,
+      "step": 153100
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5751820196264643e-06,
+      "loss": 1.2509,
+      "step": 153200
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.573599240265907e-06,
+      "loss": 1.2857,
+      "step": 153300
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.57201646090535e-06,
+      "loss": 1.2871,
+      "step": 153400
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.5704336815447926e-06,
+      "loss": 1.2932,
+      "step": 153500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.5688509021842357e-06,
+      "loss": 1.2669,
+      "step": 153600
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.5672681228236783e-06,
+      "loss": 1.2839,
+      "step": 153700
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.5656853434631213e-06,
+      "loss": 1.2592,
+      "step": 153800
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.564102564102564e-06,
+      "loss": 1.268,
+      "step": 153900
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.562519784742007e-06,
+      "loss": 1.2906,
+      "step": 154000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.5609370053814505e-06,
+      "loss": 1.2903,
+      "step": 154100
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.559354226020893e-06,
+      "loss": 1.2691,
+      "step": 154200
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.5577714466603357e-06,
+      "loss": 1.3002,
+      "step": 154300
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.5561886672997788e-06,
+      "loss": 1.281,
+      "step": 154400
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.5546058879392214e-06,
+      "loss": 1.2702,
+      "step": 154500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.5530231085786645e-06,
+      "loss": 1.2892,
+      "step": 154600
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.551440329218107e-06,
+      "loss": 1.2741,
+      "step": 154700
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.54985754985755e-06,
+      "loss": 1.2741,
+      "step": 154800
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.5482747704969927e-06,
+      "loss": 1.3219,
+      "step": 154900
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.546691991136436e-06,
+      "loss": 1.2679,
+      "step": 155000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.5451092117758784e-06,
+      "loss": 1.2688,
+      "step": 155100
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.5435264324153215e-06,
+      "loss": 1.2968,
+      "step": 155200
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.541943653054764e-06,
+      "loss": 1.3159,
+      "step": 155300
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.540360873694207e-06,
+      "loss": 1.2848,
+      "step": 155400
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.5387780943336498e-06,
+      "loss": 1.2605,
+      "step": 155500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.537195314973093e-06,
+      "loss": 1.276,
+      "step": 155600
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.535612535612536e-06,
+      "loss": 1.2905,
+      "step": 155700
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.534029756251979e-06,
+      "loss": 1.2514,
+      "step": 155800
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.5324469768914215e-06,
+      "loss": 1.2799,
+      "step": 155900
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.5308641975308646e-06,
+      "loss": 1.2998,
+      "step": 156000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.529281418170307e-06,
+      "loss": 1.291,
+      "step": 156100
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.5276986388097503e-06,
+      "loss": 1.2712,
+      "step": 156200
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.526115859449193e-06,
+      "loss": 1.2876,
+      "step": 156300
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.524533080088636e-06,
+      "loss": 1.2815,
+      "step": 156400
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.5229503007280786e-06,
+      "loss": 1.2951,
+      "step": 156500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.5213675213675216e-06,
+      "loss": 1.2857,
+      "step": 156600
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.5197847420069642e-06,
+      "loss": 1.2703,
+      "step": 156700
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.5182019626464073e-06,
+      "loss": 1.2849,
+      "step": 156800
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.51661918328585e-06,
+      "loss": 1.3097,
+      "step": 156900
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.515036403925293e-06,
+      "loss": 1.2862,
+      "step": 157000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.5134536245647356e-06,
+      "loss": 1.2717,
+      "step": 157100
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.511870845204179e-06,
+      "loss": 1.2889,
+      "step": 157200
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.5102880658436217e-06,
+      "loss": 1.3,
+      "step": 157300
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.5087052864830647e-06,
+      "loss": 1.2936,
+      "step": 157400
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.5071225071225073e-06,
+      "loss": 1.2755,
+      "step": 157500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.5055397277619504e-06,
+      "loss": 1.2885,
+      "step": 157600
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.503956948401393e-06,
+      "loss": 1.2881,
+      "step": 157700
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.502374169040836e-06,
+      "loss": 1.3128,
+      "step": 157800
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.5007913896802787e-06,
+      "loss": 1.2987,
+      "step": 157900
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.4992086103197217e-06,
+      "loss": 1.2818,
+      "step": 158000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.4976258309591644e-06,
+      "loss": 1.2771,
+      "step": 158100
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.4960430515986074e-06,
+      "loss": 1.2731,
+      "step": 158200
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.49446027223805e-06,
+      "loss": 1.2701,
+      "step": 158300
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.492877492877493e-06,
+      "loss": 1.2838,
+      "step": 158400
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.491294713516936e-06,
+      "loss": 1.2786,
+      "step": 158500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.4897119341563787e-06,
+      "loss": 1.2603,
+      "step": 158600
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.488129154795822e-06,
+      "loss": 1.292,
+      "step": 158700
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.4865463754352644e-06,
+      "loss": 1.2703,
+      "step": 158800
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.4849635960747075e-06,
+      "loss": 1.2658,
+      "step": 158900
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.48338081671415e-06,
+      "loss": 1.2922,
+      "step": 159000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.481798037353593e-06,
+      "loss": 1.2781,
+      "step": 159100
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.480215257993036e-06,
+      "loss": 1.2777,
+      "step": 159200
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.478632478632479e-06,
+      "loss": 1.2759,
+      "step": 159300
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.477049699271922e-06,
+      "loss": 1.287,
+      "step": 159400
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.4754669199113645e-06,
+      "loss": 1.2505,
+      "step": 159500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4738841405508075e-06,
+      "loss": 1.2884,
+      "step": 159600
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.47230136119025e-06,
+      "loss": 1.2841,
+      "step": 159700
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.470718581829693e-06,
+      "loss": 1.2788,
+      "step": 159800
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.469135802469136e-06,
+      "loss": 1.2529,
+      "step": 159900
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.467553023108579e-06,
+      "loss": 1.2602,
+      "step": 160000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.465970243748022e-06,
+      "loss": 1.2755,
+      "step": 160100
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4643874643874645e-06,
+      "loss": 1.2771,
+      "step": 160200
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4628046850269076e-06,
+      "loss": 1.2608,
+      "step": 160300
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4612219056663502e-06,
+      "loss": 1.2642,
+      "step": 160400
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4596391263057933e-06,
+      "loss": 1.2574,
+      "step": 160500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.458056346945236e-06,
+      "loss": 1.2792,
+      "step": 160600
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.456473567584679e-06,
+      "loss": 1.2582,
+      "step": 160700
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.454890788224122e-06,
+      "loss": 1.2884,
+      "step": 160800
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.4533080088635646e-06,
+      "loss": 1.2669,
+      "step": 160900
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.4517252295030077e-06,
+      "loss": 1.2882,
+      "step": 161000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.4501424501424503e-06,
+      "loss": 1.2643,
+      "step": 161100
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.4485596707818933e-06,
+      "loss": 1.2855,
+      "step": 161200
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.446976891421336e-06,
+      "loss": 1.2679,
+      "step": 161300
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.445394112060779e-06,
+      "loss": 1.2799,
+      "step": 161400
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.443811332700222e-06,
+      "loss": 1.2646,
+      "step": 161500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.4422285533396647e-06,
+      "loss": 1.2751,
+      "step": 161600
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.4406457739791077e-06,
+      "loss": 1.2822,
+      "step": 161700
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.4390629946185503e-06,
+      "loss": 1.2619,
+      "step": 161800
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.4374802152579934e-06,
+      "loss": 1.2626,
+      "step": 161900
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.435897435897436e-06,
+      "loss": 1.2936,
+      "step": 162000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.434314656536879e-06,
+      "loss": 1.248,
+      "step": 162100
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4327318771763217e-06,
+      "loss": 1.2828,
+      "step": 162200
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4311490978157647e-06,
+      "loss": 1.2596,
+      "step": 162300
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.429566318455208e-06,
+      "loss": 1.2892,
+      "step": 162400
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4279835390946504e-06,
+      "loss": 1.2711,
+      "step": 162500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4264007597340935e-06,
+      "loss": 1.2868,
+      "step": 162600
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.424817980373536e-06,
+      "loss": 1.2772,
+      "step": 162700
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.423235201012979e-06,
+      "loss": 1.2728,
+      "step": 162800
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4216524216524218e-06,
+      "loss": 1.2838,
+      "step": 162900
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4200696422918644e-06,
+      "loss": 1.2813,
+      "step": 163000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.418486862931308e-06,
+      "loss": 1.2664,
+      "step": 163100
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4169040835707505e-06,
+      "loss": 1.2921,
+      "step": 163200
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4153213042101935e-06,
+      "loss": 1.3039,
+      "step": 163300
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.413738524849636e-06,
+      "loss": 1.2669,
+      "step": 163400
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.412155745489079e-06,
+      "loss": 1.2897,
+      "step": 163500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.410572966128522e-06,
+      "loss": 1.2796,
+      "step": 163600
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4089901867679644e-06,
+      "loss": 1.2657,
+      "step": 163700
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4074074074074075e-06,
+      "loss": 1.2913,
+      "step": 163800
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4058246280468505e-06,
+      "loss": 1.2962,
+      "step": 163900
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4042418486862936e-06,
+      "loss": 1.2736,
+      "step": 164000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4026590693257362e-06,
+      "loss": 1.2807,
+      "step": 164100
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4010762899651793e-06,
+      "loss": 1.2704,
+      "step": 164200
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.399493510604622e-06,
+      "loss": 1.2899,
+      "step": 164300
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.3979107312440645e-06,
+      "loss": 1.247,
+      "step": 164400
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.3963279518835076e-06,
+      "loss": 1.2702,
+      "step": 164500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.3947451725229506e-06,
+      "loss": 1.3203,
+      "step": 164600
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.3931623931623937e-06,
+      "loss": 1.3029,
+      "step": 164700
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.3915796138018363e-06,
+      "loss": 1.3014,
+      "step": 164800
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.389996834441279e-06,
+      "loss": 1.2693,
+      "step": 164900
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.388414055080722e-06,
+      "loss": 1.2998,
+      "step": 165000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.3868312757201646e-06,
+      "loss": 1.2903,
+      "step": 165100
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.3852484963596076e-06,
+      "loss": 1.2914,
+      "step": 165200
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.3836657169990502e-06,
+      "loss": 1.2729,
+      "step": 165300
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.3820829376384937e-06,
+      "loss": 1.2835,
+      "step": 165400
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.3805001582779363e-06,
+      "loss": 1.2828,
+      "step": 165500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.378917378917379e-06,
+      "loss": 1.2891,
+      "step": 165600
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.377334599556822e-06,
+      "loss": 1.2837,
+      "step": 165700
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.3757518201962646e-06,
+      "loss": 1.3014,
+      "step": 165800
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.3741690408357077e-06,
+      "loss": 1.271,
+      "step": 165900
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.3725862614751503e-06,
+      "loss": 1.275,
+      "step": 166000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.3710034821145934e-06,
+      "loss": 1.281,
+      "step": 166100
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.3694207027540364e-06,
+      "loss": 1.2778,
+      "step": 166200
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.367837923393479e-06,
+      "loss": 1.2923,
+      "step": 166300
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.366255144032922e-06,
+      "loss": 1.2918,
+      "step": 166400
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3646723646723647e-06,
+      "loss": 1.2531,
+      "step": 166500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3630895853118078e-06,
+      "loss": 1.3017,
+      "step": 166600
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3615068059512504e-06,
+      "loss": 1.3031,
+      "step": 166700
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3599240265906934e-06,
+      "loss": 1.2741,
+      "step": 166800
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.358341247230136e-06,
+      "loss": 1.2919,
+      "step": 166900
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.356758467869579e-06,
+      "loss": 1.2754,
+      "step": 167000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.355175688509022e-06,
+      "loss": 1.3103,
+      "step": 167100
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3535929091484648e-06,
+      "loss": 1.3057,
+      "step": 167200
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.352010129787908e-06,
+      "loss": 1.3086,
+      "step": 167300
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3504273504273504e-06,
+      "loss": 1.2568,
+      "step": 167400
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3488445710667935e-06,
+      "loss": 1.267,
+      "step": 167500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.347261791706236e-06,
+      "loss": 1.2607,
+      "step": 167600
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.345679012345679e-06,
+      "loss": 1.2655,
+      "step": 167700
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.344096232985122e-06,
+      "loss": 1.2726,
+      "step": 167800
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.342513453624565e-06,
+      "loss": 1.2957,
+      "step": 167900
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.340930674264008e-06,
+      "loss": 1.303,
+      "step": 168000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3393478949034505e-06,
+      "loss": 1.2565,
+      "step": 168100
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3377651155428936e-06,
+      "loss": 1.2627,
+      "step": 168200
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.336182336182336e-06,
+      "loss": 1.2835,
+      "step": 168300
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3345995568217792e-06,
+      "loss": 1.2903,
+      "step": 168400
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.333016777461222e-06,
+      "loss": 1.2856,
+      "step": 168500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.331433998100665e-06,
+      "loss": 1.2892,
+      "step": 168600
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.329851218740108e-06,
+      "loss": 1.2951,
+      "step": 168700
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3282684393795506e-06,
+      "loss": 1.2871,
+      "step": 168800
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3266856600189936e-06,
+      "loss": 1.2634,
+      "step": 168900
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3251028806584362e-06,
+      "loss": 1.2854,
+      "step": 169000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3235201012978793e-06,
+      "loss": 1.2907,
+      "step": 169100
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.321937321937322e-06,
+      "loss": 1.2936,
+      "step": 169200
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.320354542576765e-06,
+      "loss": 1.286,
+      "step": 169300
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.318771763216208e-06,
+      "loss": 1.2911,
+      "step": 169400
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3171889838556506e-06,
+      "loss": 1.3014,
+      "step": 169500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3156062044950937e-06,
+      "loss": 1.294,
+      "step": 169600
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3140234251345363e-06,
+      "loss": 1.281,
+      "step": 169700
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3124406457739794e-06,
+      "loss": 1.3081,
+      "step": 169800
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.310857866413422e-06,
+      "loss": 1.2751,
+      "step": 169900
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.309275087052865e-06,
+      "loss": 1.2948,
+      "step": 170000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.307692307692308e-06,
+      "loss": 1.2792,
+      "step": 170100
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3061095283317507e-06,
+      "loss": 1.2642,
+      "step": 170200
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3045267489711937e-06,
+      "loss": 1.2786,
+      "step": 170300
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3029439696106364e-06,
+      "loss": 1.3118,
+      "step": 170400
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3013611902500794e-06,
+      "loss": 1.2878,
+      "step": 170500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.299778410889522e-06,
+      "loss": 1.3062,
+      "step": 170600
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.298195631528965e-06,
+      "loss": 1.2905,
+      "step": 170700
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2966128521684077e-06,
+      "loss": 1.2815,
+      "step": 170800
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2950300728078508e-06,
+      "loss": 1.2741,
+      "step": 170900
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.293447293447294e-06,
+      "loss": 1.2942,
+      "step": 171000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2918645140867364e-06,
+      "loss": 1.2476,
+      "step": 171100
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2902817347261795e-06,
+      "loss": 1.2634,
+      "step": 171200
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.288698955365622e-06,
+      "loss": 1.2898,
+      "step": 171300
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.287116176005065e-06,
+      "loss": 1.2751,
+      "step": 171400
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2855333966445078e-06,
+      "loss": 1.3013,
+      "step": 171500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.283950617283951e-06,
+      "loss": 1.2886,
+      "step": 171600
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.282367837923394e-06,
+      "loss": 1.2535,
+      "step": 171700
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2807850585628365e-06,
+      "loss": 1.2751,
+      "step": 171800
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2792022792022796e-06,
+      "loss": 1.2938,
+      "step": 171900
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.277619499841722e-06,
+      "loss": 1.2777,
+      "step": 172000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2760367204811652e-06,
+      "loss": 1.2935,
+      "step": 172100
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.274453941120608e-06,
+      "loss": 1.2723,
+      "step": 172200
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.272871161760051e-06,
+      "loss": 1.2664,
+      "step": 172300
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2712883823994935e-06,
+      "loss": 1.2758,
+      "step": 172400
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2697056030389366e-06,
+      "loss": 1.2746,
+      "step": 172500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2681228236783796e-06,
+      "loss": 1.2814,
+      "step": 172600
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2665400443178222e-06,
+      "loss": 1.3054,
+      "step": 172700
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2649572649572653e-06,
+      "loss": 1.249,
+      "step": 172800
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.263374485596708e-06,
+      "loss": 1.2662,
+      "step": 172900
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.261791706236151e-06,
+      "loss": 1.2934,
+      "step": 173000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2602089268755936e-06,
+      "loss": 1.2933,
+      "step": 173100
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2586261475150366e-06,
+      "loss": 1.2583,
+      "step": 173200
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2570433681544797e-06,
+      "loss": 1.2778,
+      "step": 173300
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2554605887939223e-06,
+      "loss": 1.2888,
+      "step": 173400
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2538778094333654e-06,
+      "loss": 1.2716,
+      "step": 173500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.252295030072808e-06,
+      "loss": 1.2647,
+      "step": 173600
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.250712250712251e-06,
+      "loss": 1.2459,
+      "step": 173700
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2491294713516936e-06,
+      "loss": 1.259,
+      "step": 173800
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2475466919911367e-06,
+      "loss": 1.2816,
+      "step": 173900
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2459639126305797e-06,
+      "loss": 1.2973,
+      "step": 174000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2443811332700224e-06,
+      "loss": 1.263,
+      "step": 174100
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2427983539094654e-06,
+      "loss": 1.2899,
+      "step": 174200
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.241215574548908e-06,
+      "loss": 1.2728,
+      "step": 174300
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.239632795188351e-06,
+      "loss": 1.2855,
+      "step": 174400
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2380500158277937e-06,
+      "loss": 1.2836,
+      "step": 174500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2364672364672368e-06,
+      "loss": 1.2571,
+      "step": 174600
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.2348844571066794e-06,
+      "loss": 1.2874,
+      "step": 174700
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.2333016777461224e-06,
+      "loss": 1.2911,
+      "step": 174800
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.2317188983855655e-06,
+      "loss": 1.2679,
+      "step": 174900
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.230136119025008e-06,
+      "loss": 1.2628,
+      "step": 175000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.228553339664451e-06,
+      "loss": 1.2869,
+      "step": 175100
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.2269705603038938e-06,
+      "loss": 1.2891,
+      "step": 175200
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.225387780943337e-06,
+      "loss": 1.2748,
+      "step": 175300
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2238050015827795e-06,
+      "loss": 1.2615,
+      "step": 175400
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.222222222222222e-06,
+      "loss": 1.2676,
+      "step": 175500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2206394428616655e-06,
+      "loss": 1.2829,
+      "step": 175600
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.219056663501108e-06,
+      "loss": 1.2769,
+      "step": 175700
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2174738841405512e-06,
+      "loss": 1.2651,
+      "step": 175800
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.215891104779994e-06,
+      "loss": 1.288,
+      "step": 175900
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.214308325419437e-06,
+      "loss": 1.2966,
+      "step": 176000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.2127255460588795e-06,
+      "loss": 1.2975,
+      "step": 176100
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.211142766698322e-06,
+      "loss": 1.2709,
+      "step": 176200
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.209559987337765e-06,
+      "loss": 1.2717,
+      "step": 176300
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.2079772079772082e-06,
+      "loss": 1.2737,
+      "step": 176400
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.2063944286166513e-06,
+      "loss": 1.2705,
+      "step": 176500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.204811649256094e-06,
+      "loss": 1.2532,
+      "step": 176600
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.203228869895537e-06,
+      "loss": 1.2836,
+      "step": 176700
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2016460905349796e-06,
+      "loss": 1.2802,
+      "step": 176800
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.200063311174422e-06,
+      "loss": 1.2907,
+      "step": 176900
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1984805318138653e-06,
+      "loss": 1.2926,
+      "step": 177000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.196897752453308e-06,
+      "loss": 1.2737,
+      "step": 177100
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1953149730927513e-06,
+      "loss": 1.3116,
+      "step": 177200
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.193732193732194e-06,
+      "loss": 1.283,
+      "step": 177300
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1921494143716366e-06,
+      "loss": 1.2873,
+      "step": 177400
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1905666350110796e-06,
+      "loss": 1.271,
+      "step": 177500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1889838556505223e-06,
+      "loss": 1.293,
+      "step": 177600
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1874010762899653e-06,
+      "loss": 1.2981,
+      "step": 177700
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.185818296929408e-06,
+      "loss": 1.3009,
+      "step": 177800
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1842355175688514e-06,
+      "loss": 1.2716,
+      "step": 177900
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.182652738208294e-06,
+      "loss": 1.2789,
+      "step": 178000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1810699588477367e-06,
+      "loss": 1.2717,
+      "step": 178100
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1794871794871797e-06,
+      "loss": 1.2844,
+      "step": 178200
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1779044001266223e-06,
+      "loss": 1.3079,
+      "step": 178300
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1763216207660654e-06,
+      "loss": 1.2719,
+      "step": 178400
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.174738841405508e-06,
+      "loss": 1.3024,
+      "step": 178500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.173156062044951e-06,
+      "loss": 1.2851,
+      "step": 178600
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.171573282684394e-06,
+      "loss": 1.2829,
+      "step": 178700
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.1699905033238367e-06,
+      "loss": 1.2971,
+      "step": 178800
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.1684077239632798e-06,
+      "loss": 1.2774,
+      "step": 178900
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.1668249446027224e-06,
+      "loss": 1.2784,
+      "step": 179000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.1652421652421654e-06,
+      "loss": 1.2699,
+      "step": 179100
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.163659385881608e-06,
+      "loss": 1.2804,
+      "step": 179200
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.162076606521051e-06,
+      "loss": 1.2879,
+      "step": 179300
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.1604938271604937e-06,
+      "loss": 1.2742,
+      "step": 179400
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.158911047799937e-06,
+      "loss": 1.2722,
+      "step": 179500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.15732826843938e-06,
+      "loss": 1.2702,
+      "step": 179600
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.1557454890788225e-06,
+      "loss": 1.2969,
+      "step": 179700
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.1541627097182655e-06,
+      "loss": 1.2828,
+      "step": 179800
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.152579930357708e-06,
+      "loss": 1.2896,
+      "step": 179900
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.150997150997151e-06,
+      "loss": 1.2637,
+      "step": 180000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.149414371636594e-06,
+      "loss": 1.2725,
+      "step": 180100
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.147831592276037e-06,
+      "loss": 1.2688,
+      "step": 180200
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.14624881291548e-06,
+      "loss": 1.3062,
+      "step": 180300
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1446660335549225e-06,
+      "loss": 1.2713,
+      "step": 180400
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1430832541943656e-06,
+      "loss": 1.3065,
+      "step": 180500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.141500474833808e-06,
+      "loss": 1.2709,
+      "step": 180600
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1399176954732512e-06,
+      "loss": 1.2857,
+      "step": 180700
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.138334916112694e-06,
+      "loss": 1.2632,
+      "step": 180800
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.136752136752137e-06,
+      "loss": 1.2468,
+      "step": 180900
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1351693573915795e-06,
+      "loss": 1.2784,
+      "step": 181000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1335865780310226e-06,
+      "loss": 1.2511,
+      "step": 181100
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1320037986704656e-06,
+      "loss": 1.2788,
+      "step": 181200
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1304210193099083e-06,
+      "loss": 1.2766,
+      "step": 181300
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1288382399493513e-06,
+      "loss": 1.2914,
+      "step": 181400
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.127255460588794e-06,
+      "loss": 1.2572,
+      "step": 181500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.125672681228237e-06,
+      "loss": 1.2663,
+      "step": 181600
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1240899018676796e-06,
+      "loss": 1.273,
+      "step": 181700
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1225071225071227e-06,
+      "loss": 1.3021,
+      "step": 181800
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1209243431465657e-06,
+      "loss": 1.2695,
+      "step": 181900
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1193415637860083e-06,
+      "loss": 1.2785,
+      "step": 182000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1177587844254514e-06,
+      "loss": 1.308,
+      "step": 182100
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.116176005064894e-06,
+      "loss": 1.275,
+      "step": 182200
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.114593225704337e-06,
+      "loss": 1.2769,
+      "step": 182300
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.1130104463437797e-06,
+      "loss": 1.2948,
+      "step": 182400
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.1114276669832227e-06,
+      "loss": 1.2922,
+      "step": 182500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.1098448876226658e-06,
+      "loss": 1.2751,
+      "step": 182600
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.1082621082621084e-06,
+      "loss": 1.2982,
+      "step": 182700
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.1066793289015514e-06,
+      "loss": 1.2891,
+      "step": 182800
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.105096549540994e-06,
+      "loss": 1.2642,
+      "step": 182900
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.103513770180437e-06,
+      "loss": 1.282,
+      "step": 183000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.1019309908198797e-06,
+      "loss": 1.2484,
+      "step": 183100
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.1003482114593228e-06,
+      "loss": 1.2937,
+      "step": 183200
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.0987654320987654e-06,
+      "loss": 1.2722,
+      "step": 183300
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.0971826527382085e-06,
+      "loss": 1.2755,
+      "step": 183400
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.0955998733776515e-06,
+      "loss": 1.2612,
+      "step": 183500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.094017094017094e-06,
+      "loss": 1.2788,
+      "step": 183600
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.092434314656537e-06,
+      "loss": 1.302,
+      "step": 183700
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.09085153529598e-06,
+      "loss": 1.2644,
+      "step": 183800
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.089268755935423e-06,
+      "loss": 1.3188,
+      "step": 183900
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.0876859765748655e-06,
+      "loss": 1.2803,
+      "step": 184000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.0861031972143085e-06,
+      "loss": 1.2829,
+      "step": 184100
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.0845204178537516e-06,
+      "loss": 1.2981,
+      "step": 184200
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.082937638493194e-06,
+      "loss": 1.264,
+      "step": 184300
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.0813548591326372e-06,
+      "loss": 1.2735,
+      "step": 184400
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.07977207977208e-06,
+      "loss": 1.2693,
+      "step": 184500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.078189300411523e-06,
+      "loss": 1.2836,
+      "step": 184600
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.0766065210509655e-06,
+      "loss": 1.2621,
+      "step": 184700
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.0750237416904086e-06,
+      "loss": 1.2567,
+      "step": 184800
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.0734409623298512e-06,
+      "loss": 1.269,
+      "step": 184900
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.0718581829692943e-06,
+      "loss": 1.2755,
+      "step": 185000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.0702754036087373e-06,
+      "loss": 1.2778,
+      "step": 185100
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.06869262424818e-06,
+      "loss": 1.2769,
+      "step": 185200
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.067109844887623e-06,
+      "loss": 1.2784,
+      "step": 185300
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.0655270655270656e-06,
+      "loss": 1.2756,
+      "step": 185400
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0639442861665087e-06,
+      "loss": 1.2588,
+      "step": 185500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0623615068059513e-06,
+      "loss": 1.2923,
+      "step": 185600
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0607787274453943e-06,
+      "loss": 1.2884,
+      "step": 185700
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0591959480848374e-06,
+      "loss": 1.2911,
+      "step": 185800
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.05761316872428e-06,
+      "loss": 1.2752,
+      "step": 185900
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.056030389363723e-06,
+      "loss": 1.2973,
+      "step": 186000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.0544476100031657e-06,
+      "loss": 1.2618,
+      "step": 186100
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.0528648306426087e-06,
+      "loss": 1.2849,
+      "step": 186200
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.0512820512820513e-06,
+      "loss": 1.2698,
+      "step": 186300
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.0496992719214944e-06,
+      "loss": 1.2903,
+      "step": 186400
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.0481164925609374e-06,
+      "loss": 1.2819,
+      "step": 186500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.04653371320038e-06,
+      "loss": 1.2618,
+      "step": 186600
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.044950933839823e-06,
+      "loss": 1.2659,
+      "step": 186700
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.0433681544792657e-06,
+      "loss": 1.2797,
+      "step": 186800
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.0417853751187088e-06,
+      "loss": 1.2685,
+      "step": 186900
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.0402025957581514e-06,
+      "loss": 1.2847,
+      "step": 187000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.0386198163975945e-06,
+      "loss": 1.2835,
+      "step": 187100
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.037037037037037e-06,
+      "loss": 1.2828,
+      "step": 187200
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.03545425767648e-06,
+      "loss": 1.2635,
+      "step": 187300
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.033871478315923e-06,
+      "loss": 1.2737,
+      "step": 187400
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.032288698955366e-06,
+      "loss": 1.2667,
+      "step": 187500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.030705919594809e-06,
+      "loss": 1.2842,
+      "step": 187600
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0291231402342515e-06,
+      "loss": 1.2708,
+      "step": 187700
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0275403608736945e-06,
+      "loss": 1.2875,
+      "step": 187800
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.025957581513137e-06,
+      "loss": 1.2784,
+      "step": 187900
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0243748021525798e-06,
+      "loss": 1.2915,
+      "step": 188000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0227920227920232e-06,
+      "loss": 1.2535,
+      "step": 188100
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.021209243431466e-06,
+      "loss": 1.2692,
+      "step": 188200
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.019626464070909e-06,
+      "loss": 1.2783,
+      "step": 188300
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0180436847103515e-06,
+      "loss": 1.254,
+      "step": 188400
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0164609053497946e-06,
+      "loss": 1.2628,
+      "step": 188500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.014878125989237e-06,
+      "loss": 1.2939,
+      "step": 188600
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.01329534662868e-06,
+      "loss": 1.2817,
+      "step": 188700
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.011712567268123e-06,
+      "loss": 1.276,
+      "step": 188800
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.010129787907566e-06,
+      "loss": 1.2744,
+      "step": 188900
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.008547008547009e-06,
+      "loss": 1.2634,
+      "step": 189000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.0069642291864516e-06,
+      "loss": 1.2708,
+      "step": 189100
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.0053814498258946e-06,
+      "loss": 1.2825,
+      "step": 189200
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.0037986704653373e-06,
+      "loss": 1.3043,
+      "step": 189300
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.00221589110478e-06,
+      "loss": 1.2951,
+      "step": 189400
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.000633111744223e-06,
+      "loss": 1.2802,
+      "step": 189500
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9990503323836656e-06,
+      "loss": 1.2612,
+      "step": 189600
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.997467553023109e-06,
+      "loss": 1.2405,
+      "step": 189700
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9958847736625517e-06,
+      "loss": 1.2401,
+      "step": 189800
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9943019943019947e-06,
+      "loss": 1.2274,
+      "step": 189900
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9927192149414373e-06,
+      "loss": 1.2452,
+      "step": 190000
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.99113643558088e-06,
+      "loss": 1.2378,
+      "step": 190100
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.989553656220323e-06,
+      "loss": 1.2384,
+      "step": 190200
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9879708768597656e-06,
+      "loss": 1.2451,
+      "step": 190300
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9863880974992087e-06,
+      "loss": 1.2583,
+      "step": 190400
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9848053181386517e-06,
+      "loss": 1.2231,
+      "step": 190500
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9832225387780944e-06,
+      "loss": 1.2353,
+      "step": 190600
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9816397594175374e-06,
+      "loss": 1.2643,
+      "step": 190700
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.98005698005698e-06,
+      "loss": 1.2729,
+      "step": 190800
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.978474200696423e-06,
+      "loss": 1.2783,
+      "step": 190900
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9768914213358657e-06,
+      "loss": 1.2253,
+      "step": 191000
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9753086419753087e-06,
+      "loss": 1.2331,
+      "step": 191100
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.973725862614752e-06,
+      "loss": 1.2509,
+      "step": 191200
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9721430832541944e-06,
+      "loss": 1.2361,
+      "step": 191300
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9705603038936375e-06,
+      "loss": 1.253,
+      "step": 191400
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.96897752453308e-06,
+      "loss": 1.2475,
+      "step": 191500
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.967394745172523e-06,
+      "loss": 1.2385,
+      "step": 191600
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9658119658119658e-06,
+      "loss": 1.2559,
+      "step": 191700
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.964229186451409e-06,
+      "loss": 1.2609,
+      "step": 191800
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9626464070908514e-06,
+      "loss": 1.2372,
+      "step": 191900
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9610636277302945e-06,
+      "loss": 1.2321,
+      "step": 192000
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9594808483697375e-06,
+      "loss": 1.2499,
+      "step": 192100
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.95789806900918e-06,
+      "loss": 1.2412,
+      "step": 192200
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.956315289648623e-06,
+      "loss": 1.2322,
+      "step": 192300
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.954732510288066e-06,
+      "loss": 1.2493,
+      "step": 192400
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.953149730927509e-06,
+      "loss": 1.2598,
+      "step": 192500
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.9515669515669515e-06,
+      "loss": 1.2199,
+      "step": 192600
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.9499841722063945e-06,
+      "loss": 1.2525,
+      "step": 192700
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.9484013928458376e-06,
+      "loss": 1.2735,
+      "step": 192800
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.9468186134852802e-06,
+      "loss": 1.2406,
+      "step": 192900
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.9452358341247233e-06,
+      "loss": 1.236,
+      "step": 193000
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.943653054764166e-06,
+      "loss": 1.2599,
+      "step": 193100
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.942070275403609e-06,
+      "loss": 1.2612,
+      "step": 193200
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.9404874960430516e-06,
+      "loss": 1.2271,
+      "step": 193300
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.9389047166824946e-06,
+      "loss": 1.2178,
+      "step": 193400
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.9373219373219372e-06,
+      "loss": 1.2459,
+      "step": 193500
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.9357391579613803e-06,
+      "loss": 1.2591,
+      "step": 193600
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.9341563786008233e-06,
+      "loss": 1.2471,
+      "step": 193700
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.932573599240266e-06,
+      "loss": 1.2676,
+      "step": 193800
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.930990819879709e-06,
+      "loss": 1.249,
+      "step": 193900
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.9294080405191516e-06,
+      "loss": 1.2489,
+      "step": 194000
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.9278252611585947e-06,
+      "loss": 1.2576,
+      "step": 194100
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.9262424817980373e-06,
+      "loss": 1.2459,
+      "step": 194200
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.9246597024374803e-06,
+      "loss": 1.229,
+      "step": 194300
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.9230769230769234e-06,
+      "loss": 1.2454,
+      "step": 194400
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.921494143716366e-06,
+      "loss": 1.2555,
+      "step": 194500
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.919911364355809e-06,
+      "loss": 1.24,
+      "step": 194600
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.9183285849952517e-06,
+      "loss": 1.2455,
+      "step": 194700
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.9167458056346947e-06,
+      "loss": 1.2728,
+      "step": 194800
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.9151630262741374e-06,
+      "loss": 1.2666,
+      "step": 194900
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.9135802469135804e-06,
+      "loss": 1.238,
+      "step": 195000
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.9119974675530235e-06,
+      "loss": 1.2514,
+      "step": 195100
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.910414688192466e-06,
+      "loss": 1.2493,
+      "step": 195200
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.908831908831909e-06,
+      "loss": 1.2498,
+      "step": 195300
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.907249129471352e-06,
+      "loss": 1.2379,
+      "step": 195400
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.9056663501107948e-06,
+      "loss": 1.2071,
+      "step": 195500
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.9040835707502374e-06,
+      "loss": 1.2708,
+      "step": 195600
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.9025007913896803e-06,
+      "loss": 1.2384,
+      "step": 195700
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.900918012029123e-06,
+      "loss": 1.2763,
+      "step": 195800
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.8993352326685664e-06,
+      "loss": 1.2591,
+      "step": 195900
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.8977524533080092e-06,
+      "loss": 1.2338,
+      "step": 196000
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.896169673947452e-06,
+      "loss": 1.255,
+      "step": 196100
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.8945868945868947e-06,
+      "loss": 1.2154,
+      "step": 196200
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.8930041152263375e-06,
+      "loss": 1.2323,
+      "step": 196300
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.8914213358657803e-06,
+      "loss": 1.2435,
+      "step": 196400
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.8898385565052232e-06,
+      "loss": 1.2459,
+      "step": 196500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.888255777144666e-06,
+      "loss": 1.2382,
+      "step": 196600
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.8866729977841093e-06,
+      "loss": 1.2192,
+      "step": 196700
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.885090218423552e-06,
+      "loss": 1.232,
+      "step": 196800
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8835074390629947e-06,
+      "loss": 1.2427,
+      "step": 196900
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8819246597024376e-06,
+      "loss": 1.2299,
+      "step": 197000
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8803418803418804e-06,
+      "loss": 1.2627,
+      "step": 197100
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8787591009813232e-06,
+      "loss": 1.2491,
+      "step": 197200
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.877176321620766e-06,
+      "loss": 1.2408,
+      "step": 197300
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.875593542260209e-06,
+      "loss": 1.2532,
+      "step": 197400
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8740107628996522e-06,
+      "loss": 1.2322,
+      "step": 197500
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8724279835390948e-06,
+      "loss": 1.2301,
+      "step": 197600
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8708452041785376e-06,
+      "loss": 1.2351,
+      "step": 197700
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8692624248179805e-06,
+      "loss": 1.2381,
+      "step": 197800
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8676796454574233e-06,
+      "loss": 1.2292,
+      "step": 197900
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8660968660968661e-06,
+      "loss": 1.2353,
+      "step": 198000
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.864514086736309e-06,
+      "loss": 1.2478,
+      "step": 198100
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8629313073757518e-06,
+      "loss": 1.2674,
+      "step": 198200
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8613485280151949e-06,
+      "loss": 1.2407,
+      "step": 198300
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8597657486546377e-06,
+      "loss": 1.252,
+      "step": 198400
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8581829692940805e-06,
+      "loss": 1.2296,
+      "step": 198500
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8566001899335234e-06,
+      "loss": 1.2212,
+      "step": 198600
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8550174105729662e-06,
+      "loss": 1.2593,
+      "step": 198700
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.853434631212409e-06,
+      "loss": 1.2589,
+      "step": 198800
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.8518518518518519e-06,
+      "loss": 1.2658,
+      "step": 198900
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.8502690724912947e-06,
+      "loss": 1.2562,
+      "step": 199000
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.8486862931307378e-06,
+      "loss": 1.2194,
+      "step": 199100
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.8471035137701806e-06,
+      "loss": 1.2305,
+      "step": 199200
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.8455207344096234e-06,
+      "loss": 1.2331,
+      "step": 199300
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.8439379550490663e-06,
+      "loss": 1.2488,
+      "step": 199400
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.842355175688509e-06,
+      "loss": 1.2838,
+      "step": 199500
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.840772396327952e-06,
+      "loss": 1.2588,
+      "step": 199600
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.8391896169673948e-06,
+      "loss": 1.2524,
+      "step": 199700
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.8376068376068378e-06,
+      "loss": 1.2301,
+      "step": 199800
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.8360240582462807e-06,
+      "loss": 1.2341,
+      "step": 199900
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8344412788857235e-06,
+      "loss": 1.2495,
+      "step": 200000
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8328584995251663e-06,
+      "loss": 1.249,
+      "step": 200100
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8312757201646092e-06,
+      "loss": 1.2413,
+      "step": 200200
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.829692940804052e-06,
+      "loss": 1.251,
+      "step": 200300
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8281101614434948e-06,
+      "loss": 1.2444,
+      "step": 200400
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8265273820829377e-06,
+      "loss": 1.2483,
+      "step": 200500
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8249446027223807e-06,
+      "loss": 1.2452,
+      "step": 200600
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8233618233618236e-06,
+      "loss": 1.2351,
+      "step": 200700
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8217790440012664e-06,
+      "loss": 1.2388,
+      "step": 200800
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8201962646407092e-06,
+      "loss": 1.2543,
+      "step": 200900
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.818613485280152e-06,
+      "loss": 1.2439,
+      "step": 201000
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.817030705919595e-06,
+      "loss": 1.2244,
+      "step": 201100
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8154479265590377e-06,
+      "loss": 1.2469,
+      "step": 201200
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8138651471984806e-06,
+      "loss": 1.2573,
+      "step": 201300
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8122823678379236e-06,
+      "loss": 1.2158,
+      "step": 201400
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8106995884773665e-06,
+      "loss": 1.2588,
+      "step": 201500
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8091168091168093e-06,
+      "loss": 1.2726,
+      "step": 201600
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8075340297562521e-06,
+      "loss": 1.2704,
+      "step": 201700
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.805951250395695e-06,
+      "loss": 1.2376,
+      "step": 201800
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.8043684710351378e-06,
+      "loss": 1.2476,
+      "step": 201900
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.8027856916745806e-06,
+      "loss": 1.25,
+      "step": 202000
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.8012029123140235e-06,
+      "loss": 1.2597,
+      "step": 202100
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7996201329534665e-06,
+      "loss": 1.2478,
+      "step": 202200
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7980373535929094e-06,
+      "loss": 1.2715,
+      "step": 202300
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7964545742323522e-06,
+      "loss": 1.2387,
+      "step": 202400
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.794871794871795e-06,
+      "loss": 1.2514,
+      "step": 202500
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7932890155112379e-06,
+      "loss": 1.2611,
+      "step": 202600
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7917062361506807e-06,
+      "loss": 1.2358,
+      "step": 202700
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7901234567901235e-06,
+      "loss": 1.2601,
+      "step": 202800
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7885406774295664e-06,
+      "loss": 1.2259,
+      "step": 202900
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7869578980690094e-06,
+      "loss": 1.2487,
+      "step": 203000
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7853751187084523e-06,
+      "loss": 1.2563,
+      "step": 203100
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.783792339347895e-06,
+      "loss": 1.267,
+      "step": 203200
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.782209559987338e-06,
+      "loss": 1.2506,
+      "step": 203300
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7806267806267808e-06,
+      "loss": 1.2634,
+      "step": 203400
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7790440012662236e-06,
+      "loss": 1.2611,
+      "step": 203500
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7774612219056664e-06,
+      "loss": 1.2441,
+      "step": 203600
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7758784425451095e-06,
+      "loss": 1.2578,
+      "step": 203700
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7742956631845523e-06,
+      "loss": 1.2381,
+      "step": 203800
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7727128838239952e-06,
+      "loss": 1.2298,
+      "step": 203900
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.771130104463438e-06,
+      "loss": 1.2178,
+      "step": 204000
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7695473251028808e-06,
+      "loss": 1.2671,
+      "step": 204100
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7679645457423237e-06,
+      "loss": 1.2419,
+      "step": 204200
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7663817663817665e-06,
+      "loss": 1.2411,
+      "step": 204300
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7647989870212093e-06,
+      "loss": 1.2656,
+      "step": 204400
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7632162076606524e-06,
+      "loss": 1.256,
+      "step": 204500
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7616334283000952e-06,
+      "loss": 1.249,
+      "step": 204600
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.760050648939538e-06,
+      "loss": 1.2494,
+      "step": 204700
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.758467869578981e-06,
+      "loss": 1.2468,
+      "step": 204800
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7568850902184237e-06,
+      "loss": 1.2431,
+      "step": 204900
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7553023108578666e-06,
+      "loss": 1.2088,
+      "step": 205000
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7537195314973094e-06,
+      "loss": 1.2292,
+      "step": 205100
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7521367521367522e-06,
+      "loss": 1.2328,
+      "step": 205200
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7505539727761953e-06,
+      "loss": 1.273,
+      "step": 205300
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7489711934156381e-06,
+      "loss": 1.2562,
+      "step": 205400
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.747388414055081e-06,
+      "loss": 1.2129,
+      "step": 205500
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7458056346945238e-06,
+      "loss": 1.241,
+      "step": 205600
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7442228553339666e-06,
+      "loss": 1.2482,
+      "step": 205700
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7426400759734095e-06,
+      "loss": 1.2848,
+      "step": 205800
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7410572966128523e-06,
+      "loss": 1.2477,
+      "step": 205900
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7394745172522951e-06,
+      "loss": 1.2461,
+      "step": 206000
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7378917378917382e-06,
+      "loss": 1.2796,
+      "step": 206100
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.736308958531181e-06,
+      "loss": 1.2306,
+      "step": 206200
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7347261791706239e-06,
+      "loss": 1.2568,
+      "step": 206300
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7331433998100667e-06,
+      "loss": 1.227,
+      "step": 206400
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7315606204495095e-06,
+      "loss": 1.2486,
+      "step": 206500
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7299778410889524e-06,
+      "loss": 1.2436,
+      "step": 206600
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7283950617283952e-06,
+      "loss": 1.2449,
+      "step": 206700
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7268122823678378e-06,
+      "loss": 1.2456,
+      "step": 206800
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.725229503007281e-06,
+      "loss": 1.2389,
+      "step": 206900
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.723646723646724e-06,
+      "loss": 1.2394,
+      "step": 207000
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7220639442861668e-06,
+      "loss": 1.2534,
+      "step": 207100
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7204811649256096e-06,
+      "loss": 1.2481,
+      "step": 207200
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7188983855650524e-06,
+      "loss": 1.2398,
+      "step": 207300
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7173156062044953e-06,
+      "loss": 1.2393,
+      "step": 207400
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.715732826843938e-06,
+      "loss": 1.2248,
+      "step": 207500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7141500474833807e-06,
+      "loss": 1.2493,
+      "step": 207600
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.712567268122824e-06,
+      "loss": 1.2608,
+      "step": 207700
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7109844887622668e-06,
+      "loss": 1.2488,
+      "step": 207800
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7094017094017097e-06,
+      "loss": 1.252,
+      "step": 207900
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7078189300411525e-06,
+      "loss": 1.247,
+      "step": 208000
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7062361506805951e-06,
+      "loss": 1.2525,
+      "step": 208100
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.704653371320038e-06,
+      "loss": 1.2471,
+      "step": 208200
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7030705919594808e-06,
+      "loss": 1.2298,
+      "step": 208300
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.701487812598924e-06,
+      "loss": 1.2743,
+      "step": 208400
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.6999050332383669e-06,
+      "loss": 1.2406,
+      "step": 208500
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.6983222538778097e-06,
+      "loss": 1.2562,
+      "step": 208600
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.6967394745172526e-06,
+      "loss": 1.2634,
+      "step": 208700
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.6951566951566952e-06,
+      "loss": 1.2591,
+      "step": 208800
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.693573915796138e-06,
+      "loss": 1.252,
+      "step": 208900
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.6919911364355809e-06,
+      "loss": 1.2353,
+      "step": 209000
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.6904083570750237e-06,
+      "loss": 1.2397,
+      "step": 209100
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.688825577714467e-06,
+      "loss": 1.2503,
+      "step": 209200
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.6872427983539098e-06,
+      "loss": 1.2477,
+      "step": 209300
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.6856600189933524e-06,
+      "loss": 1.2638,
+      "step": 209400
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.6840772396327953e-06,
+      "loss": 1.2328,
+      "step": 209500
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.682494460272238e-06,
+      "loss": 1.2335,
+      "step": 209600
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.680911680911681e-06,
+      "loss": 1.2445,
+      "step": 209700
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.6793289015511238e-06,
+      "loss": 1.2558,
+      "step": 209800
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.6777461221905666e-06,
+      "loss": 1.2411,
+      "step": 209900
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.6761633428300099e-06,
+      "loss": 1.2376,
+      "step": 210000
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.6745805634694525e-06,
+      "loss": 1.2362,
+      "step": 210100
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.6729977841088953e-06,
+      "loss": 1.2222,
+      "step": 210200
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.6714150047483382e-06,
+      "loss": 1.2419,
+      "step": 210300
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.669832225387781e-06,
+      "loss": 1.2606,
+      "step": 210400
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.6682494460272238e-06,
+      "loss": 1.26,
+      "step": 210500
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 1.249,
+      "step": 210600
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.6650838873061095e-06,
+      "loss": 1.2807,
+      "step": 210700
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6635011079455525e-06,
+      "loss": 1.2249,
+      "step": 210800
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6619183285849954e-06,
+      "loss": 1.2672,
+      "step": 210900
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6603355492244382e-06,
+      "loss": 1.2518,
+      "step": 211000
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.658752769863881e-06,
+      "loss": 1.2495,
+      "step": 211100
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6571699905033239e-06,
+      "loss": 1.2569,
+      "step": 211200
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6555872111427667e-06,
+      "loss": 1.2431,
+      "step": 211300
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6540044317822096e-06,
+      "loss": 1.2373,
+      "step": 211400
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6524216524216524e-06,
+      "loss": 1.2348,
+      "step": 211500
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6508388730610954e-06,
+      "loss": 1.2432,
+      "step": 211600
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6492560937005383e-06,
+      "loss": 1.2426,
+      "step": 211700
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6476733143399811e-06,
+      "loss": 1.2697,
+      "step": 211800
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.646090534979424e-06,
+      "loss": 1.2502,
+      "step": 211900
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.6445077556188668e-06,
+      "loss": 1.253,
+      "step": 212000
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.6429249762583096e-06,
+      "loss": 1.2616,
+      "step": 212100
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.6413421968977525e-06,
+      "loss": 1.2869,
+      "step": 212200
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.6397594175371955e-06,
+      "loss": 1.2456,
+      "step": 212300
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.6381766381766383e-06,
+      "loss": 1.2771,
+      "step": 212400
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.6365938588160812e-06,
+      "loss": 1.2429,
+      "step": 212500
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.635011079455524e-06,
+      "loss": 1.2574,
+      "step": 212600
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.6334283000949669e-06,
+      "loss": 1.2408,
+      "step": 212700
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.6318455207344097e-06,
+      "loss": 1.2654,
+      "step": 212800
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.6302627413738525e-06,
+      "loss": 1.2428,
+      "step": 212900
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.6286799620132954e-06,
+      "loss": 1.2224,
+      "step": 213000
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.6270971826527384e-06,
+      "loss": 1.2795,
+      "step": 213100
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.6255144032921812e-06,
+      "loss": 1.2448,
+      "step": 213200
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.623931623931624e-06,
+      "loss": 1.2453,
+      "step": 213300
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.622348844571067e-06,
+      "loss": 1.2457,
+      "step": 213400
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.6207660652105098e-06,
+      "loss": 1.2379,
+      "step": 213500
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.6191832858499526e-06,
+      "loss": 1.2555,
+      "step": 213600
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.6176005064893954e-06,
+      "loss": 1.2506,
+      "step": 213700
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.6160177271288383e-06,
+      "loss": 1.2602,
+      "step": 213800
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6144349477682813e-06,
+      "loss": 1.2719,
+      "step": 213900
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6128521684077241e-06,
+      "loss": 1.245,
+      "step": 214000
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.611269389047167e-06,
+      "loss": 1.2508,
+      "step": 214100
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6096866096866098e-06,
+      "loss": 1.2412,
+      "step": 214200
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6081038303260527e-06,
+      "loss": 1.2291,
+      "step": 214300
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6065210509654955e-06,
+      "loss": 1.2507,
+      "step": 214400
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6049382716049383e-06,
+      "loss": 1.2273,
+      "step": 214500
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6033554922443812e-06,
+      "loss": 1.2531,
+      "step": 214600
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6017727128838242e-06,
+      "loss": 1.2221,
+      "step": 214700
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.600189933523267e-06,
+      "loss": 1.2632,
+      "step": 214800
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.5986071541627099e-06,
+      "loss": 1.2397,
+      "step": 214900
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.5970243748021527e-06,
+      "loss": 1.236,
+      "step": 215000
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.5954415954415956e-06,
+      "loss": 1.2437,
+      "step": 215100
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.5938588160810384e-06,
+      "loss": 1.2488,
+      "step": 215200
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.5922760367204812e-06,
+      "loss": 1.229,
+      "step": 215300
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.590693257359924e-06,
+      "loss": 1.2075,
+      "step": 215400
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.5891104779993671e-06,
+      "loss": 1.2419,
+      "step": 215500
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.58752769863881e-06,
+      "loss": 1.2519,
+      "step": 215600
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.5859449192782528e-06,
+      "loss": 1.2538,
+      "step": 215700
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.5843621399176956e-06,
+      "loss": 1.242,
+      "step": 215800
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.5827793605571385e-06,
+      "loss": 1.2382,
+      "step": 215900
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.5811965811965813e-06,
+      "loss": 1.2555,
+      "step": 216000
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.5796138018360241e-06,
+      "loss": 1.2095,
+      "step": 216100
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.578031022475467e-06,
+      "loss": 1.2067,
+      "step": 216200
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.57644824311491e-06,
+      "loss": 1.2299,
+      "step": 216300
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.5748654637543529e-06,
+      "loss": 1.2532,
+      "step": 216400
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.5732826843937957e-06,
+      "loss": 1.2626,
+      "step": 216500
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.5716999050332385e-06,
+      "loss": 1.2278,
+      "step": 216600
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.5701171256726814e-06,
+      "loss": 1.2322,
+      "step": 216700
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.5685343463121242e-06,
+      "loss": 1.2374,
+      "step": 216800
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.566951566951567e-06,
+      "loss": 1.2623,
+      "step": 216900
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.56536878759101e-06,
+      "loss": 1.2337,
+      "step": 217000
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.563786008230453e-06,
+      "loss": 1.2624,
+      "step": 217100
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.5622032288698958e-06,
+      "loss": 1.2655,
+      "step": 217200
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.5606204495093386e-06,
+      "loss": 1.2449,
+      "step": 217300
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.5590376701487814e-06,
+      "loss": 1.249,
+      "step": 217400
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.5574548907882243e-06,
+      "loss": 1.2301,
+      "step": 217500
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.555872111427667e-06,
+      "loss": 1.2445,
+      "step": 217600
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.55428933206711e-06,
+      "loss": 1.2463,
+      "step": 217700
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.552706552706553e-06,
+      "loss": 1.2428,
+      "step": 217800
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.5511237733459958e-06,
+      "loss": 1.2259,
+      "step": 217900
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.5495409939854387e-06,
+      "loss": 1.2691,
+      "step": 218000
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.5479582146248815e-06,
+      "loss": 1.2534,
+      "step": 218100
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.5463754352643243e-06,
+      "loss": 1.2381,
+      "step": 218200
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.5447926559037672e-06,
+      "loss": 1.2581,
+      "step": 218300
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.54320987654321e-06,
+      "loss": 1.2314,
+      "step": 218400
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.5416270971826528e-06,
+      "loss": 1.2477,
+      "step": 218500
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.5400443178220959e-06,
+      "loss": 1.2518,
+      "step": 218600
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.5384615384615387e-06,
+      "loss": 1.2412,
+      "step": 218700
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.5368787591009816e-06,
+      "loss": 1.2748,
+      "step": 218800
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.5352959797404244e-06,
+      "loss": 1.2315,
+      "step": 218900
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.5337132003798672e-06,
+      "loss": 1.2391,
+      "step": 219000
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.53213042101931e-06,
+      "loss": 1.2561,
+      "step": 219100
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.530547641658753e-06,
+      "loss": 1.2338,
+      "step": 219200
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.5289648622981957e-06,
+      "loss": 1.2484,
+      "step": 219300
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.5273820829376388e-06,
+      "loss": 1.242,
+      "step": 219400
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.5257993035770816e-06,
+      "loss": 1.2733,
+      "step": 219500
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.5242165242165245e-06,
+      "loss": 1.2338,
+      "step": 219600
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.5226337448559673e-06,
+      "loss": 1.2568,
+      "step": 219700
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.5210509654954101e-06,
+      "loss": 1.2168,
+      "step": 219800
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.519468186134853e-06,
+      "loss": 1.2227,
+      "step": 219900
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.5178854067742956e-06,
+      "loss": 1.2381,
+      "step": 220000
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.5163026274137384e-06,
+      "loss": 1.236,
+      "step": 220100
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.5147198480531817e-06,
+      "loss": 1.2648,
+      "step": 220200
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.5131370686926245e-06,
+      "loss": 1.2629,
+      "step": 220300
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.5115542893320674e-06,
+      "loss": 1.2506,
+      "step": 220400
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.5099715099715102e-06,
+      "loss": 1.2663,
+      "step": 220500
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.508388730610953e-06,
+      "loss": 1.2484,
+      "step": 220600
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.5068059512503957e-06,
+      "loss": 1.2824,
+      "step": 220700
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.5052231718898385e-06,
+      "loss": 1.2388,
+      "step": 220800
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.5036403925292817e-06,
+      "loss": 1.2586,
+      "step": 220900
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.5020576131687246e-06,
+      "loss": 1.2517,
+      "step": 221000
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.5004748338081674e-06,
+      "loss": 1.2667,
+      "step": 221100
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.4988920544476103e-06,
+      "loss": 1.2656,
+      "step": 221200
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.4973092750870529e-06,
+      "loss": 1.2394,
+      "step": 221300
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.4957264957264957e-06,
+      "loss": 1.215,
+      "step": 221400
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.4941437163659386e-06,
+      "loss": 1.2729,
+      "step": 221500
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.4925609370053814e-06,
+      "loss": 1.2605,
+      "step": 221600
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.4909781576448246e-06,
+      "loss": 1.214,
+      "step": 221700
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.4893953782842675e-06,
+      "loss": 1.2753,
+      "step": 221800
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.4878125989237103e-06,
+      "loss": 1.2588,
+      "step": 221900
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.486229819563153e-06,
+      "loss": 1.2522,
+      "step": 222000
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.4846470402025958e-06,
+      "loss": 1.2276,
+      "step": 222100
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.4830642608420386e-06,
+      "loss": 1.2501,
+      "step": 222200
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.4814814814814815e-06,
+      "loss": 1.2317,
+      "step": 222300
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.4798987021209243e-06,
+      "loss": 1.2656,
+      "step": 222400
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.4783159227603675e-06,
+      "loss": 1.2541,
+      "step": 222500
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.4767331433998102e-06,
+      "loss": 1.259,
+      "step": 222600
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.475150364039253e-06,
+      "loss": 1.2336,
+      "step": 222700
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.4735675846786958e-06,
+      "loss": 1.2568,
+      "step": 222800
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.4719848053181387e-06,
+      "loss": 1.237,
+      "step": 222900
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.4704020259575815e-06,
+      "loss": 1.2408,
+      "step": 223000
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.4688192465970244e-06,
+      "loss": 1.2355,
+      "step": 223100
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.4672364672364672e-06,
+      "loss": 1.2569,
+      "step": 223200
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.4656536878759102e-06,
+      "loss": 1.2255,
+      "step": 223300
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.464070908515353e-06,
+      "loss": 1.2289,
+      "step": 223400
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.462488129154796e-06,
+      "loss": 1.2258,
+      "step": 223500
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.4609053497942387e-06,
+      "loss": 1.2568,
+      "step": 223600
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.4593225704336816e-06,
+      "loss": 1.2682,
+      "step": 223700
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.4577397910731244e-06,
+      "loss": 1.222,
+      "step": 223800
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.4561570117125673e-06,
+      "loss": 1.2279,
+      "step": 223900
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.45457423235201e-06,
+      "loss": 1.2364,
+      "step": 224000
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4529914529914531e-06,
+      "loss": 1.236,
+      "step": 224100
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.451408673630896e-06,
+      "loss": 1.2355,
+      "step": 224200
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4498258942703388e-06,
+      "loss": 1.2598,
+      "step": 224300
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4482431149097816e-06,
+      "loss": 1.2585,
+      "step": 224400
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4466603355492245e-06,
+      "loss": 1.2652,
+      "step": 224500
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4450775561886673e-06,
+      "loss": 1.2346,
+      "step": 224600
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.4434947768281102e-06,
+      "loss": 1.2436,
+      "step": 224700
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.441911997467553e-06,
+      "loss": 1.2388,
+      "step": 224800
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.440329218106996e-06,
+      "loss": 1.2255,
+      "step": 224900
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.4387464387464389e-06,
+      "loss": 1.2149,
+      "step": 225000
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.4371636593858817e-06,
+      "loss": 1.2252,
+      "step": 225100
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.4355808800253245e-06,
+      "loss": 1.2472,
+      "step": 225200
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4339981006647674e-06,
+      "loss": 1.2453,
+      "step": 225300
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4324153213042102e-06,
+      "loss": 1.2376,
+      "step": 225400
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.430832541943653e-06,
+      "loss": 1.2351,
+      "step": 225500
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.429249762583096e-06,
+      "loss": 1.2365,
+      "step": 225600
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.427666983222539e-06,
+      "loss": 1.2242,
+      "step": 225700
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4260842038619818e-06,
+      "loss": 1.2405,
+      "step": 225800
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4245014245014246e-06,
+      "loss": 1.238,
+      "step": 225900
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4229186451408674e-06,
+      "loss": 1.2643,
+      "step": 226000
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4213358657803103e-06,
+      "loss": 1.2401,
+      "step": 226100
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4197530864197531e-06,
+      "loss": 1.2699,
+      "step": 226200
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.418170307059196e-06,
+      "loss": 1.2191,
+      "step": 226300
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.416587527698639e-06,
+      "loss": 1.2451,
+      "step": 226400
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4150047483380818e-06,
+      "loss": 1.244,
+      "step": 226500
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4134219689775247e-06,
+      "loss": 1.2351,
+      "step": 226600
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4118391896169675e-06,
+      "loss": 1.2529,
+      "step": 226700
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4102564102564104e-06,
+      "loss": 1.2416,
+      "step": 226800
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4086736308958532e-06,
+      "loss": 1.2517,
+      "step": 226900
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.407090851535296e-06,
+      "loss": 1.2529,
+      "step": 227000
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4055080721747389e-06,
+      "loss": 1.2695,
+      "step": 227100
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.403925292814182e-06,
+      "loss": 1.2373,
+      "step": 227200
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.4023425134536247e-06,
+      "loss": 1.2243,
+      "step": 227300
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.4007597340930676e-06,
+      "loss": 1.2498,
+      "step": 227400
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.3991769547325104e-06,
+      "loss": 1.2622,
+      "step": 227500
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.3975941753719533e-06,
+      "loss": 1.2382,
+      "step": 227600
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.396011396011396e-06,
+      "loss": 1.2502,
+      "step": 227700
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.394428616650839e-06,
+      "loss": 1.2512,
+      "step": 227800
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.3928458372902818e-06,
+      "loss": 1.2618,
+      "step": 227900
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.3912630579297248e-06,
+      "loss": 1.2564,
+      "step": 228000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.3896802785691676e-06,
+      "loss": 1.2455,
+      "step": 228100
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.3880974992086105e-06,
+      "loss": 1.2435,
+      "step": 228200
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.3865147198480533e-06,
+      "loss": 1.246,
+      "step": 228300
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.3849319404874962e-06,
+      "loss": 1.2369,
+      "step": 228400
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.383349161126939e-06,
+      "loss": 1.2676,
+      "step": 228500
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.3817663817663818e-06,
+      "loss": 1.2402,
+      "step": 228600
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.3801836024058247e-06,
+      "loss": 1.2247,
+      "step": 228700
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.3786008230452677e-06,
+      "loss": 1.2288,
+      "step": 228800
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.3770180436847105e-06,
+      "loss": 1.2513,
+      "step": 228900
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.3754352643241534e-06,
+      "loss": 1.2511,
+      "step": 229000
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.3738524849635962e-06,
+      "loss": 1.2334,
+      "step": 229100
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.372269705603039e-06,
+      "loss": 1.2434,
+      "step": 229200
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.3706869262424819e-06,
+      "loss": 1.2542,
+      "step": 229300
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.3691041468819247e-06,
+      "loss": 1.2356,
+      "step": 229400
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.3675213675213678e-06,
+      "loss": 1.2485,
+      "step": 229500
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.3659385881608106e-06,
+      "loss": 1.2517,
+      "step": 229600
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.3643558088002534e-06,
+      "loss": 1.2377,
+      "step": 229700
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.3627730294396963e-06,
+      "loss": 1.2531,
+      "step": 229800
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.3611902500791391e-06,
+      "loss": 1.2534,
+      "step": 229900
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.359607470718582e-06,
+      "loss": 1.2547,
+      "step": 230000
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.3580246913580248e-06,
+      "loss": 1.2592,
+      "step": 230100
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.3564419119974676e-06,
+      "loss": 1.2643,
+      "step": 230200
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.3548591326369107e-06,
+      "loss": 1.2549,
+      "step": 230300
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.3532763532763535e-06,
+      "loss": 1.2592,
+      "step": 230400
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.3516935739157963e-06,
+      "loss": 1.2509,
+      "step": 230500
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.3501107945552392e-06,
+      "loss": 1.2405,
+      "step": 230600
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.348528015194682e-06,
+      "loss": 1.25,
+      "step": 230700
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.3469452358341249e-06,
+      "loss": 1.2289,
+      "step": 230800
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.3453624564735677e-06,
+      "loss": 1.2611,
+      "step": 230900
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.3437796771130105e-06,
+      "loss": 1.2404,
+      "step": 231000
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.3421968977524536e-06,
+      "loss": 1.23,
+      "step": 231100
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.3406141183918964e-06,
+      "loss": 1.2433,
+      "step": 231200
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.3390313390313392e-06,
+      "loss": 1.2484,
+      "step": 231300
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.337448559670782e-06,
+      "loss": 1.2499,
+      "step": 231400
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.335865780310225e-06,
+      "loss": 1.218,
+      "step": 231500
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.3342830009496678e-06,
+      "loss": 1.2401,
+      "step": 231600
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.3327002215891106e-06,
+      "loss": 1.2423,
+      "step": 231700
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.3311174422285534e-06,
+      "loss": 1.2505,
+      "step": 231800
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.3295346628679965e-06,
+      "loss": 1.2788,
+      "step": 231900
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.3279518835074393e-06,
+      "loss": 1.2591,
+      "step": 232000
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.3263691041468821e-06,
+      "loss": 1.2424,
+      "step": 232100
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.324786324786325e-06,
+      "loss": 1.2317,
+      "step": 232200
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.3232035454257678e-06,
+      "loss": 1.2419,
+      "step": 232300
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.3216207660652107e-06,
+      "loss": 1.2291,
+      "step": 232400
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.3200379867046533e-06,
+      "loss": 1.2314,
+      "step": 232500
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.3184552073440961e-06,
+      "loss": 1.2604,
+      "step": 232600
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.3168724279835394e-06,
+      "loss": 1.2483,
+      "step": 232700
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.3152896486229822e-06,
+      "loss": 1.2415,
+      "step": 232800
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.313706869262425e-06,
+      "loss": 1.2391,
+      "step": 232900
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.3121240899018679e-06,
+      "loss": 1.2217,
+      "step": 233000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.3105413105413107e-06,
+      "loss": 1.2702,
+      "step": 233100
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.3089585311807533e-06,
+      "loss": 1.2514,
+      "step": 233200
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.3073757518201962e-06,
+      "loss": 1.27,
+      "step": 233300
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.305792972459639e-06,
+      "loss": 1.2208,
+      "step": 233400
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.3042101930990823e-06,
+      "loss": 1.2622,
+      "step": 233500
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.3026274137385251e-06,
+      "loss": 1.2381,
+      "step": 233600
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.301044634377968e-06,
+      "loss": 1.2485,
+      "step": 233700
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.2994618550174106e-06,
+      "loss": 1.2427,
+      "step": 233800
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.2978790756568534e-06,
+      "loss": 1.2365,
+      "step": 233900
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.2962962962962962e-06,
+      "loss": 1.2406,
+      "step": 234000
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.294713516935739e-06,
+      "loss": 1.2346,
+      "step": 234100
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2931307375751823e-06,
+      "loss": 1.236,
+      "step": 234200
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2915479582146252e-06,
+      "loss": 1.2465,
+      "step": 234300
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.289965178854068e-06,
+      "loss": 1.2357,
+      "step": 234400
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2883823994935106e-06,
+      "loss": 1.2427,
+      "step": 234500
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2867996201329535e-06,
+      "loss": 1.2411,
+      "step": 234600
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2852168407723963e-06,
+      "loss": 1.2346,
+      "step": 234700
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2836340614118391e-06,
+      "loss": 1.2327,
+      "step": 234800
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.282051282051282e-06,
+      "loss": 1.2583,
+      "step": 234900
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2804685026907252e-06,
+      "loss": 1.2683,
+      "step": 235000
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2788857233301679e-06,
+      "loss": 1.2552,
+      "step": 235100
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2773029439696107e-06,
+      "loss": 1.2634,
+      "step": 235200
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2757201646090535e-06,
+      "loss": 1.2703,
+      "step": 235300
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2741373852484964e-06,
+      "loss": 1.2357,
+      "step": 235400
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2725546058879392e-06,
+      "loss": 1.2311,
+      "step": 235500
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.270971826527382e-06,
+      "loss": 1.2546,
+      "step": 235600
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2693890471668249e-06,
+      "loss": 1.2428,
+      "step": 235700
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.267806267806268e-06,
+      "loss": 1.2363,
+      "step": 235800
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2662234884457108e-06,
+      "loss": 1.2555,
+      "step": 235900
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2646407090851536e-06,
+      "loss": 1.2612,
+      "step": 236000
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2630579297245964e-06,
+      "loss": 1.2647,
+      "step": 236100
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2614751503640393e-06,
+      "loss": 1.244,
+      "step": 236200
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2598923710034821e-06,
+      "loss": 1.247,
+      "step": 236300
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.258309591642925e-06,
+      "loss": 1.2188,
+      "step": 236400
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2567268122823678e-06,
+      "loss": 1.2389,
+      "step": 236500
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2551440329218108e-06,
+      "loss": 1.2465,
+      "step": 236600
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2535612535612537e-06,
+      "loss": 1.2485,
+      "step": 236700
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2519784742006965e-06,
+      "loss": 1.2558,
+      "step": 236800
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2503956948401393e-06,
+      "loss": 1.2336,
+      "step": 236900
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2488129154795822e-06,
+      "loss": 1.2336,
+      "step": 237000
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.247230136119025e-06,
+      "loss": 1.2232,
+      "step": 237100
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.245647356758468e-06,
+      "loss": 1.2447,
+      "step": 237200
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.244064577397911e-06,
+      "loss": 1.2451,
+      "step": 237300
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2424817980373537e-06,
+      "loss": 1.2262,
+      "step": 237400
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2408990186767966e-06,
+      "loss": 1.2374,
+      "step": 237500
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2393162393162394e-06,
+      "loss": 1.257,
+      "step": 237600
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2377334599556822e-06,
+      "loss": 1.2302,
+      "step": 237700
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.236150680595125e-06,
+      "loss": 1.2363,
+      "step": 237800
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.234567901234568e-06,
+      "loss": 1.2497,
+      "step": 237900
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.232985121874011e-06,
+      "loss": 1.2276,
+      "step": 238000
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2314023425134538e-06,
+      "loss": 1.2419,
+      "step": 238100
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2298195631528966e-06,
+      "loss": 1.2191,
+      "step": 238200
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2282367837923395e-06,
+      "loss": 1.2259,
+      "step": 238300
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2266540044317823e-06,
+      "loss": 1.2626,
+      "step": 238400
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2250712250712251e-06,
+      "loss": 1.2465,
+      "step": 238500
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.223488445710668e-06,
+      "loss": 1.233,
+      "step": 238600
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.221905666350111e-06,
+      "loss": 1.2341,
+      "step": 238700
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2203228869895539e-06,
+      "loss": 1.2505,
+      "step": 238800
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2187401076289967e-06,
+      "loss": 1.2452,
+      "step": 238900
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2171573282684395e-06,
+      "loss": 1.2341,
+      "step": 239000
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2155745489078824e-06,
+      "loss": 1.2519,
+      "step": 239100
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2139917695473252e-06,
+      "loss": 1.2568,
+      "step": 239200
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.212408990186768e-06,
+      "loss": 1.2417,
+      "step": 239300
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2108262108262109e-06,
+      "loss": 1.2238,
+      "step": 239400
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.209243431465654e-06,
+      "loss": 1.2603,
+      "step": 239500
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2076606521050968e-06,
+      "loss": 1.2635,
+      "step": 239600
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2060778727445396e-06,
+      "loss": 1.2414,
+      "step": 239700
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2044950933839822e-06,
+      "loss": 1.2324,
+      "step": 239800
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2029123140234253e-06,
+      "loss": 1.2282,
+      "step": 239900
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2013295346628681e-06,
+      "loss": 1.2431,
+      "step": 240000
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.199746755302311e-06,
+      "loss": 1.2492,
+      "step": 240100
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.1981639759417538e-06,
+      "loss": 1.2618,
+      "step": 240200
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.1965811965811968e-06,
+      "loss": 1.2537,
+      "step": 240300
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1949984172206395e-06,
+      "loss": 1.2318,
+      "step": 240400
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1934156378600823e-06,
+      "loss": 1.2362,
+      "step": 240500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1918328584995251e-06,
+      "loss": 1.2557,
+      "step": 240600
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1902500791389682e-06,
+      "loss": 1.2382,
+      "step": 240700
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.188667299778411e-06,
+      "loss": 1.2282,
+      "step": 240800
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1870845204178538e-06,
+      "loss": 1.2551,
+      "step": 240900
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1855017410572967e-06,
+      "loss": 1.2666,
+      "step": 241000
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.1839189616967395e-06,
+      "loss": 1.2554,
+      "step": 241100
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.1823361823361824e-06,
+      "loss": 1.249,
+      "step": 241200
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.1807534029756252e-06,
+      "loss": 1.2368,
+      "step": 241300
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.179170623615068e-06,
+      "loss": 1.2152,
+      "step": 241400
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.177587844254511e-06,
+      "loss": 1.2264,
+      "step": 241500
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.176005064893954e-06,
+      "loss": 1.2387,
+      "step": 241600
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.1744222855333967e-06,
+      "loss": 1.2525,
+      "step": 241700
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.1728395061728396e-06,
+      "loss": 1.2551,
+      "step": 241800
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.1712567268122824e-06,
+      "loss": 1.2517,
+      "step": 241900
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.1696739474517253e-06,
+      "loss": 1.2395,
+      "step": 242000
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.168091168091168e-06,
+      "loss": 1.2725,
+      "step": 242100
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.166508388730611e-06,
+      "loss": 1.2384,
+      "step": 242200
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.164925609370054e-06,
+      "loss": 1.2415,
+      "step": 242300
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.1633428300094968e-06,
+      "loss": 1.2425,
+      "step": 242400
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.1617600506489396e-06,
+      "loss": 1.2423,
+      "step": 242500
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.1601772712883825e-06,
+      "loss": 1.2445,
+      "step": 242600
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.1585944919278253e-06,
+      "loss": 1.2468,
+      "step": 242700
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.1570117125672682e-06,
+      "loss": 1.2683,
+      "step": 242800
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.155428933206711e-06,
+      "loss": 1.2497,
+      "step": 242900
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.153846153846154e-06,
+      "loss": 1.2405,
+      "step": 243000
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.1522633744855969e-06,
+      "loss": 1.2389,
+      "step": 243100
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.1506805951250397e-06,
+      "loss": 1.2181,
+      "step": 243200
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.1490978157644825e-06,
+      "loss": 1.244,
+      "step": 243300
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.1475150364039254e-06,
+      "loss": 1.2576,
+      "step": 243400
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.1459322570433682e-06,
+      "loss": 1.2137,
+      "step": 243500
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.144349477682811e-06,
+      "loss": 1.2546,
+      "step": 243600
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.1427666983222539e-06,
+      "loss": 1.2682,
+      "step": 243700
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.141183918961697e-06,
+      "loss": 1.2553,
+      "step": 243800
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.1396011396011398e-06,
+      "loss": 1.2623,
+      "step": 243900
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.1380183602405826e-06,
+      "loss": 1.2418,
+      "step": 244000
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.1364355808800254e-06,
+      "loss": 1.2509,
+      "step": 244100
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.1348528015194683e-06,
+      "loss": 1.2589,
+      "step": 244200
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.1332700221589111e-06,
+      "loss": 1.2667,
+      "step": 244300
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.131687242798354e-06,
+      "loss": 1.2304,
+      "step": 244400
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.1301044634377968e-06,
+      "loss": 1.2306,
+      "step": 244500
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.1285216840772398e-06,
+      "loss": 1.2414,
+      "step": 244600
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.1269389047166827e-06,
+      "loss": 1.2494,
+      "step": 244700
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.1253561253561255e-06,
+      "loss": 1.233,
+      "step": 244800
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.1237733459955683e-06,
+      "loss": 1.2552,
+      "step": 244900
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.1221905666350112e-06,
+      "loss": 1.237,
+      "step": 245000
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.120607787274454e-06,
+      "loss": 1.2529,
+      "step": 245100
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.1190250079138969e-06,
+      "loss": 1.236,
+      "step": 245200
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.1174422285533397e-06,
+      "loss": 1.2658,
+      "step": 245300
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.1158594491927827e-06,
+      "loss": 1.2609,
+      "step": 245400
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.1142766698322256e-06,
+      "loss": 1.2423,
+      "step": 245500
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.1126938904716684e-06,
+      "loss": 1.2388,
+      "step": 245600
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.111111111111111e-06,
+      "loss": 1.2253,
+      "step": 245700
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.109528331750554e-06,
+      "loss": 1.2525,
+      "step": 245800
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.107945552389997e-06,
+      "loss": 1.2705,
+      "step": 245900
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.1063627730294398e-06,
+      "loss": 1.2519,
+      "step": 246000
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.1047799936688826e-06,
+      "loss": 1.2501,
+      "step": 246100
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.1031972143083256e-06,
+      "loss": 1.2612,
+      "step": 246200
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.1016144349477685e-06,
+      "loss": 1.237,
+      "step": 246300
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.100031655587211e-06,
+      "loss": 1.2562,
+      "step": 246400
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.098448876226654e-06,
+      "loss": 1.2435,
+      "step": 246500
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.096866096866097e-06,
+      "loss": 1.2513,
+      "step": 246600
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.0952833175055398e-06,
+      "loss": 1.2429,
+      "step": 246700
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0937005381449827e-06,
+      "loss": 1.2486,
+      "step": 246800
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0921177587844257e-06,
+      "loss": 1.2547,
+      "step": 246900
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0905349794238683e-06,
+      "loss": 1.2369,
+      "step": 247000
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0889522000633112e-06,
+      "loss": 1.242,
+      "step": 247100
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.087369420702754e-06,
+      "loss": 1.2345,
+      "step": 247200
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.085786641342197e-06,
+      "loss": 1.2694,
+      "step": 247300
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0842038619816399e-06,
+      "loss": 1.2281,
+      "step": 247400
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0826210826210827e-06,
+      "loss": 1.225,
+      "step": 247500
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0810383032605256e-06,
+      "loss": 1.2525,
+      "step": 247600
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0794555238999684e-06,
+      "loss": 1.2331,
+      "step": 247700
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0778727445394112e-06,
+      "loss": 1.2281,
+      "step": 247800
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.076289965178854e-06,
+      "loss": 1.2252,
+      "step": 247900
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.074707185818297e-06,
+      "loss": 1.254,
+      "step": 248000
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.07312440645774e-06,
+      "loss": 1.2318,
+      "step": 248100
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0715416270971828e-06,
+      "loss": 1.2201,
+      "step": 248200
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0699588477366256e-06,
+      "loss": 1.2346,
+      "step": 248300
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0683760683760685e-06,
+      "loss": 1.2332,
+      "step": 248400
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0667932890155113e-06,
+      "loss": 1.2522,
+      "step": 248500
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0652105096549541e-06,
+      "loss": 1.2432,
+      "step": 248600
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.063627730294397e-06,
+      "loss": 1.243,
+      "step": 248700
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0620449509338398e-06,
+      "loss": 1.2388,
+      "step": 248800
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0604621715732829e-06,
+      "loss": 1.2575,
+      "step": 248900
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0588793922127257e-06,
+      "loss": 1.2382,
+      "step": 249000
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0572966128521685e-06,
+      "loss": 1.2576,
+      "step": 249100
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0557138334916114e-06,
+      "loss": 1.2447,
+      "step": 249200
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.0541310541310542e-06,
+      "loss": 1.2306,
+      "step": 249300
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.052548274770497e-06,
+      "loss": 1.2472,
+      "step": 249400
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.0509654954099399e-06,
+      "loss": 1.264,
+      "step": 249500
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.0493827160493827e-06,
+      "loss": 1.2432,
+      "step": 249600
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.0477999366888258e-06,
+      "loss": 1.2438,
+      "step": 249700
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.0462171573282686e-06,
+      "loss": 1.2565,
+      "step": 249800
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0446343779677114e-06,
+      "loss": 1.2437,
+      "step": 249900
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0430515986071543e-06,
+      "loss": 1.2349,
+      "step": 250000
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.041468819246597e-06,
+      "loss": 1.2452,
+      "step": 250100
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.03988603988604e-06,
+      "loss": 1.2405,
+      "step": 250200
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0383032605254828e-06,
+      "loss": 1.2347,
+      "step": 250300
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0367204811649256e-06,
+      "loss": 1.2407,
+      "step": 250400
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0351377018043687e-06,
+      "loss": 1.2514,
+      "step": 250500
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0335549224438115e-06,
+      "loss": 1.2345,
+      "step": 250600
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0319721430832543e-06,
+      "loss": 1.2235,
+      "step": 250700
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0303893637226972e-06,
+      "loss": 1.211,
+      "step": 250800
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.02880658436214e-06,
+      "loss": 1.271,
+      "step": 250900
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0272238050015828e-06,
+      "loss": 1.2656,
+      "step": 251000
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0256410256410257e-06,
+      "loss": 1.2491,
+      "step": 251100
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0240582462804687e-06,
+      "loss": 1.2323,
+      "step": 251200
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0224754669199116e-06,
+      "loss": 1.2322,
+      "step": 251300
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0208926875593544e-06,
+      "loss": 1.2369,
+      "step": 251400
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0193099081987972e-06,
+      "loss": 1.2164,
+      "step": 251500
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.01772712883824e-06,
+      "loss": 1.2385,
+      "step": 251600
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.016144349477683e-06,
+      "loss": 1.2391,
+      "step": 251700
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0145615701171257e-06,
+      "loss": 1.2339,
+      "step": 251800
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0129787907565686e-06,
+      "loss": 1.2648,
+      "step": 251900
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0113960113960116e-06,
+      "loss": 1.2392,
+      "step": 252000
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0098132320354545e-06,
+      "loss": 1.2161,
+      "step": 252100
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0082304526748973e-06,
+      "loss": 1.2221,
+      "step": 252200
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.00664767331434e-06,
+      "loss": 1.2626,
+      "step": 252300
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.005064893953783e-06,
+      "loss": 1.2323,
+      "step": 252400
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.0034821145932258e-06,
+      "loss": 1.2225,
+      "step": 252500
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.0018993352326686e-06,
+      "loss": 1.2237,
+      "step": 252600
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.0003165558721115e-06,
+      "loss": 1.2393,
+      "step": 252700
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 9.987337765115545e-07,
+      "loss": 1.2331,
+      "step": 252800
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 9.971509971509974e-07,
+      "loss": 1.2363,
+      "step": 252900
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 9.9556821779044e-07,
+      "loss": 1.2385,
+      "step": 253000
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 9.939854384298828e-07,
+      "loss": 1.2217,
+      "step": 253100
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 9.924026590693259e-07,
+      "loss": 1.2262,
+      "step": 253200
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 9.908198797087687e-07,
+      "loss": 1.2354,
+      "step": 253300
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 9.892371003482115e-07,
+      "loss": 1.2145,
+      "step": 253400
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 9.876543209876544e-07,
+      "loss": 1.2181,
+      "step": 253500
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 9.860715416270972e-07,
+      "loss": 1.226,
+      "step": 253600
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 9.8448876226654e-07,
+      "loss": 1.2222,
+      "step": 253700
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 9.829059829059829e-07,
+      "loss": 1.2253,
+      "step": 253800
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 9.813232035454257e-07,
+      "loss": 1.2306,
+      "step": 253900
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 9.797404241848688e-07,
+      "loss": 1.2386,
+      "step": 254000
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 9.781576448243116e-07,
+      "loss": 1.2346,
+      "step": 254100
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 9.765748654637544e-07,
+      "loss": 1.2149,
+      "step": 254200
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.749920861031973e-07,
+      "loss": 1.2564,
+      "step": 254300
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.734093067426401e-07,
+      "loss": 1.2195,
+      "step": 254400
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.71826527382083e-07,
+      "loss": 1.2174,
+      "step": 254500
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.702437480215258e-07,
+      "loss": 1.2197,
+      "step": 254600
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.686609686609686e-07,
+      "loss": 1.2259,
+      "step": 254700
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.670781893004117e-07,
+      "loss": 1.2089,
+      "step": 254800
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.654954099398545e-07,
+      "loss": 1.2201,
+      "step": 254900
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 9.639126305792973e-07,
+      "loss": 1.2364,
+      "step": 255000
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 9.623298512187402e-07,
+      "loss": 1.2486,
+      "step": 255100
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 9.60747071858183e-07,
+      "loss": 1.2484,
+      "step": 255200
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 9.591642924976258e-07,
+      "loss": 1.2237,
+      "step": 255300
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 9.575815131370687e-07,
+      "loss": 1.197,
+      "step": 255400
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 9.559987337765117e-07,
+      "loss": 1.2274,
+      "step": 255500
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 9.544159544159546e-07,
+      "loss": 1.2307,
+      "step": 255600
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 9.528331750553974e-07,
+      "loss": 1.2244,
+      "step": 255700
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 9.512503956948401e-07,
+      "loss": 1.2181,
+      "step": 255800
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 9.496676163342832e-07,
+      "loss": 1.206,
+      "step": 255900
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 9.48084836973726e-07,
+      "loss": 1.2437,
+      "step": 256000
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 9.465020576131687e-07,
+      "loss": 1.2216,
+      "step": 256100
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 9.449192782526116e-07,
+      "loss": 1.2245,
+      "step": 256200
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 9.433364988920546e-07,
+      "loss": 1.223,
+      "step": 256300
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 9.417537195314974e-07,
+      "loss": 1.2319,
+      "step": 256400
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 9.401709401709402e-07,
+      "loss": 1.2453,
+      "step": 256500
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 9.38588160810383e-07,
+      "loss": 1.234,
+      "step": 256600
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 9.370053814498261e-07,
+      "loss": 1.2314,
+      "step": 256700
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 9.354226020892688e-07,
+      "loss": 1.2013,
+      "step": 256800
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 9.338398227287116e-07,
+      "loss": 1.2229,
+      "step": 256900
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 9.322570433681545e-07,
+      "loss": 1.2013,
+      "step": 257000
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 9.306742640075974e-07,
+      "loss": 1.2169,
+      "step": 257100
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 9.290914846470403e-07,
+      "loss": 1.2046,
+      "step": 257200
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 9.275087052864831e-07,
+      "loss": 1.2601,
+      "step": 257300
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 9.259259259259259e-07,
+      "loss": 1.2527,
+      "step": 257400
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 9.243431465653689e-07,
+      "loss": 1.2336,
+      "step": 257500
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 9.227603672048117e-07,
+      "loss": 1.2232,
+      "step": 257600
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 9.211775878442545e-07,
+      "loss": 1.2148,
+      "step": 257700
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 9.195948084836974e-07,
+      "loss": 1.217,
+      "step": 257800
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 9.180120291231403e-07,
+      "loss": 1.2315,
+      "step": 257900
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 9.164292497625832e-07,
+      "loss": 1.2368,
+      "step": 258000
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 9.14846470402026e-07,
+      "loss": 1.2126,
+      "step": 258100
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 9.132636910414688e-07,
+      "loss": 1.2545,
+      "step": 258200
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 9.116809116809118e-07,
+      "loss": 1.2237,
+      "step": 258300
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 9.100981323203546e-07,
+      "loss": 1.2131,
+      "step": 258400
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 9.085153529597975e-07,
+      "loss": 1.2287,
+      "step": 258500
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 9.069325735992403e-07,
+      "loss": 1.2276,
+      "step": 258600
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 9.053497942386832e-07,
+      "loss": 1.1935,
+      "step": 258700
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 9.037670148781261e-07,
+      "loss": 1.205,
+      "step": 258800
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 9.021842355175689e-07,
+      "loss": 1.2195,
+      "step": 258900
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 9.006014561570117e-07,
+      "loss": 1.2067,
+      "step": 259000
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 8.990186767964547e-07,
+      "loss": 1.2016,
+      "step": 259100
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 8.974358974358975e-07,
+      "loss": 1.2289,
+      "step": 259200
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 8.958531180753404e-07,
+      "loss": 1.2059,
+      "step": 259300
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 8.942703387147832e-07,
+      "loss": 1.2441,
+      "step": 259400
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 8.926875593542261e-07,
+      "loss": 1.2385,
+      "step": 259500
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 8.91104779993669e-07,
+      "loss": 1.2383,
+      "step": 259600
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 8.895220006331118e-07,
+      "loss": 1.2277,
+      "step": 259700
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 8.879392212725547e-07,
+      "loss": 1.2149,
+      "step": 259800
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 8.863564419119976e-07,
+      "loss": 1.2047,
+      "step": 259900
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 8.847736625514404e-07,
+      "loss": 1.2147,
+      "step": 260000
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 8.831908831908833e-07,
+      "loss": 1.2101,
+      "step": 260100
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 8.816081038303262e-07,
+      "loss": 1.2134,
+      "step": 260200
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 8.80025324469769e-07,
+      "loss": 1.2361,
+      "step": 260300
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 8.784425451092119e-07,
+      "loss": 1.2083,
+      "step": 260400
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 8.768597657486547e-07,
+      "loss": 1.2284,
+      "step": 260500
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 8.752769863880976e-07,
+      "loss": 1.2245,
+      "step": 260600
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 8.736942070275405e-07,
+      "loss": 1.2216,
+      "step": 260700
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 8.721114276669833e-07,
+      "loss": 1.2279,
+      "step": 260800
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 8.705286483064262e-07,
+      "loss": 1.2534,
+      "step": 260900
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 8.689458689458691e-07,
+      "loss": 1.2281,
+      "step": 261000
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 8.673630895853119e-07,
+      "loss": 1.2376,
+      "step": 261100
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 8.657803102247548e-07,
+      "loss": 1.2057,
+      "step": 261200
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 8.641975308641976e-07,
+      "loss": 1.2238,
+      "step": 261300
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 8.626147515036405e-07,
+      "loss": 1.2365,
+      "step": 261400
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 8.610319721430834e-07,
+      "loss": 1.2265,
+      "step": 261500
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 8.594491927825262e-07,
+      "loss": 1.2328,
+      "step": 261600
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 8.57866413421969e-07,
+      "loss": 1.2375,
+      "step": 261700
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 8.56283634061412e-07,
+      "loss": 1.2156,
+      "step": 261800
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 8.547008547008548e-07,
+      "loss": 1.2192,
+      "step": 261900
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 8.531180753402976e-07,
+      "loss": 1.2027,
+      "step": 262000
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 8.515352959797404e-07,
+      "loss": 1.21,
+      "step": 262100
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 8.499525166191834e-07,
+      "loss": 1.2121,
+      "step": 262200
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 8.483697372586263e-07,
+      "loss": 1.2086,
+      "step": 262300
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 8.46786957898069e-07,
+      "loss": 1.2432,
+      "step": 262400
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 8.452041785375118e-07,
+      "loss": 1.2317,
+      "step": 262500
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 8.436213991769549e-07,
+      "loss": 1.2257,
+      "step": 262600
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 8.420386198163976e-07,
+      "loss": 1.2124,
+      "step": 262700
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 8.404558404558405e-07,
+      "loss": 1.2134,
+      "step": 262800
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 8.388730610952833e-07,
+      "loss": 1.1924,
+      "step": 262900
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 8.372902817347262e-07,
+      "loss": 1.2177,
+      "step": 263000
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 8.357075023741691e-07,
+      "loss": 1.2286,
+      "step": 263100
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 8.341247230136119e-07,
+      "loss": 1.2167,
+      "step": 263200
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 8.325419436530547e-07,
+      "loss": 1.2271,
+      "step": 263300
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 8.309591642924977e-07,
+      "loss": 1.2244,
+      "step": 263400
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 8.293763849319405e-07,
+      "loss": 1.2319,
+      "step": 263500
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 8.277936055713834e-07,
+      "loss": 1.2276,
+      "step": 263600
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 8.262108262108262e-07,
+      "loss": 1.2367,
+      "step": 263700
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 8.246280468502691e-07,
+      "loss": 1.2384,
+      "step": 263800
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 8.23045267489712e-07,
+      "loss": 1.2504,
+      "step": 263900
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 8.214624881291548e-07,
+      "loss": 1.2502,
+      "step": 264000
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 8.198797087685978e-07,
+      "loss": 1.24,
+      "step": 264100
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 8.182969294080406e-07,
+      "loss": 1.2296,
+      "step": 264200
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 8.167141500474834e-07,
+      "loss": 1.2239,
+      "step": 264300
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 8.151313706869263e-07,
+      "loss": 1.2178,
+      "step": 264400
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 8.135485913263692e-07,
+      "loss": 1.2132,
+      "step": 264500
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 8.11965811965812e-07,
+      "loss": 1.2107,
+      "step": 264600
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 8.103830326052549e-07,
+      "loss": 1.214,
+      "step": 264700
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 8.088002532446977e-07,
+      "loss": 1.2367,
+      "step": 264800
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 8.072174738841407e-07,
+      "loss": 1.2442,
+      "step": 264900
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 8.056346945235835e-07,
+      "loss": 1.2177,
+      "step": 265000
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 8.040519151630263e-07,
+      "loss": 1.1992,
+      "step": 265100
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 8.024691358024692e-07,
+      "loss": 1.2128,
+      "step": 265200
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 8.008863564419121e-07,
+      "loss": 1.2227,
+      "step": 265300
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 7.993035770813549e-07,
+      "loss": 1.2192,
+      "step": 265400
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 7.977207977207978e-07,
+      "loss": 1.2104,
+      "step": 265500
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 7.961380183602406e-07,
+      "loss": 1.2301,
+      "step": 265600
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 7.945552389996836e-07,
+      "loss": 1.2364,
+      "step": 265700
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 7.929724596391264e-07,
+      "loss": 1.2211,
+      "step": 265800
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 7.913896802785692e-07,
+      "loss": 1.2089,
+      "step": 265900
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 7.898069009180121e-07,
+      "loss": 1.217,
+      "step": 266000
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 7.88224121557455e-07,
+      "loss": 1.2247,
+      "step": 266100
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 7.866413421968978e-07,
+      "loss": 1.209,
+      "step": 266200
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 7.850585628363407e-07,
+      "loss": 1.2159,
+      "step": 266300
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 7.834757834757835e-07,
+      "loss": 1.2051,
+      "step": 266400
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 7.818930041152265e-07,
+      "loss": 1.2303,
+      "step": 266500
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 7.803102247546693e-07,
+      "loss": 1.2138,
+      "step": 266600
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 7.787274453941121e-07,
+      "loss": 1.2188,
+      "step": 266700
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 7.77144666033555e-07,
+      "loss": 1.2283,
+      "step": 266800
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 7.755618866729979e-07,
+      "loss": 1.1906,
+      "step": 266900
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 7.739791073124407e-07,
+      "loss": 1.2488,
+      "step": 267000
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 7.723963279518836e-07,
+      "loss": 1.2176,
+      "step": 267100
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 7.708135485913264e-07,
+      "loss": 1.215,
+      "step": 267200
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 7.692307692307694e-07,
+      "loss": 1.2364,
+      "step": 267300
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 7.676479898702122e-07,
+      "loss": 1.2603,
+      "step": 267400
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 7.66065210509655e-07,
+      "loss": 1.2257,
+      "step": 267500
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 7.644824311490979e-07,
+      "loss": 1.2262,
+      "step": 267600
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 7.628996517885408e-07,
+      "loss": 1.2265,
+      "step": 267700
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 7.613168724279836e-07,
+      "loss": 1.2112,
+      "step": 267800
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 7.597340930674265e-07,
+      "loss": 1.2195,
+      "step": 267900
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 7.581513137068692e-07,
+      "loss": 1.1993,
+      "step": 268000
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 7.565685343463123e-07,
+      "loss": 1.2303,
+      "step": 268100
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 7.549857549857551e-07,
+      "loss": 1.2501,
+      "step": 268200
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 7.534029756251978e-07,
+      "loss": 1.2242,
+      "step": 268300
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 7.518201962646409e-07,
+      "loss": 1.2056,
+      "step": 268400
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 7.502374169040837e-07,
+      "loss": 1.2103,
+      "step": 268500
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 7.486546375435264e-07,
+      "loss": 1.2102,
+      "step": 268600
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 7.470718581829693e-07,
+      "loss": 1.2354,
+      "step": 268700
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 7.454890788224123e-07,
+      "loss": 1.2311,
+      "step": 268800
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 7.439062994618552e-07,
+      "loss": 1.2315,
+      "step": 268900
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 7.423235201012979e-07,
+      "loss": 1.2165,
+      "step": 269000
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 7.407407407407407e-07,
+      "loss": 1.23,
+      "step": 269100
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 7.391579613801838e-07,
+      "loss": 1.2151,
+      "step": 269200
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 7.375751820196265e-07,
+      "loss": 1.2227,
+      "step": 269300
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 7.359924026590693e-07,
+      "loss": 1.1945,
+      "step": 269400
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 7.344096232985122e-07,
+      "loss": 1.2185,
+      "step": 269500
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 7.328268439379551e-07,
+      "loss": 1.2332,
+      "step": 269600
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 7.31244064577398e-07,
+      "loss": 1.2355,
+      "step": 269700
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 7.296612852168408e-07,
+      "loss": 1.222,
+      "step": 269800
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 7.280785058562836e-07,
+      "loss": 1.2238,
+      "step": 269900
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 7.264957264957266e-07,
+      "loss": 1.2278,
+      "step": 270000
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 7.249129471351694e-07,
+      "loss": 1.223,
+      "step": 270100
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 7.233301677746122e-07,
+      "loss": 1.236,
+      "step": 270200
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 7.217473884140551e-07,
+      "loss": 1.237,
+      "step": 270300
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 7.20164609053498e-07,
+      "loss": 1.21,
+      "step": 270400
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 7.185818296929409e-07,
+      "loss": 1.2176,
+      "step": 270500
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 7.169990503323837e-07,
+      "loss": 1.2053,
+      "step": 270600
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 7.154162709718265e-07,
+      "loss": 1.2249,
+      "step": 270700
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 7.138334916112695e-07,
+      "loss": 1.2079,
+      "step": 270800
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 7.122507122507123e-07,
+      "loss": 1.2091,
+      "step": 270900
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 7.106679328901551e-07,
+      "loss": 1.2176,
+      "step": 271000
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 7.09085153529598e-07,
+      "loss": 1.2224,
+      "step": 271100
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 7.075023741690409e-07,
+      "loss": 1.1919,
+      "step": 271200
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 7.059195948084838e-07,
+      "loss": 1.2166,
+      "step": 271300
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 7.043368154479266e-07,
+      "loss": 1.2271,
+      "step": 271400
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 7.027540360873694e-07,
+      "loss": 1.2394,
+      "step": 271500
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 7.011712567268124e-07,
+      "loss": 1.2107,
+      "step": 271600
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 6.995884773662552e-07,
+      "loss": 1.2444,
+      "step": 271700
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 6.98005698005698e-07,
+      "loss": 1.2411,
+      "step": 271800
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 6.964229186451409e-07,
+      "loss": 1.2274,
+      "step": 271900
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 6.948401392845838e-07,
+      "loss": 1.199,
+      "step": 272000
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 6.932573599240267e-07,
+      "loss": 1.231,
+      "step": 272100
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 6.916745805634695e-07,
+      "loss": 1.2462,
+      "step": 272200
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 6.900918012029123e-07,
+      "loss": 1.235,
+      "step": 272300
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 6.885090218423553e-07,
+      "loss": 1.2136,
+      "step": 272400
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 6.869262424817981e-07,
+      "loss": 1.2406,
+      "step": 272500
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 6.853434631212409e-07,
+      "loss": 1.2045,
+      "step": 272600
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 6.837606837606839e-07,
+      "loss": 1.2363,
+      "step": 272700
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 6.821779044001267e-07,
+      "loss": 1.2156,
+      "step": 272800
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 6.805951250395696e-07,
+      "loss": 1.2183,
+      "step": 272900
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 6.790123456790124e-07,
+      "loss": 1.2254,
+      "step": 273000
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 6.774295663184553e-07,
+      "loss": 1.2367,
+      "step": 273100
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 6.758467869578982e-07,
+      "loss": 1.2146,
+      "step": 273200
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 6.74264007597341e-07,
+      "loss": 1.2576,
+      "step": 273300
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 6.726812282367838e-07,
+      "loss": 1.2112,
+      "step": 273400
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 6.710984488762268e-07,
+      "loss": 1.2366,
+      "step": 273500
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 6.695156695156696e-07,
+      "loss": 1.2143,
+      "step": 273600
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 6.679328901551125e-07,
+      "loss": 1.2253,
+      "step": 273700
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 6.663501107945553e-07,
+      "loss": 1.2102,
+      "step": 273800
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 6.647673314339982e-07,
+      "loss": 1.2178,
+      "step": 273900
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 6.631845520734411e-07,
+      "loss": 1.2304,
+      "step": 274000
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 6.616017727128839e-07,
+      "loss": 1.2131,
+      "step": 274100
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 6.600189933523266e-07,
+      "loss": 1.2112,
+      "step": 274200
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 6.584362139917697e-07,
+      "loss": 1.234,
+      "step": 274300
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 6.568534346312125e-07,
+      "loss": 1.2111,
+      "step": 274400
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 6.552706552706554e-07,
+      "loss": 1.2304,
+      "step": 274500
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 6.536878759100981e-07,
+      "loss": 1.2073,
+      "step": 274600
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 6.521050965495411e-07,
+      "loss": 1.2147,
+      "step": 274700
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 6.50522317188984e-07,
+      "loss": 1.2278,
+      "step": 274800
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 6.489395378284267e-07,
+      "loss": 1.255,
+      "step": 274900
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 6.473567584678695e-07,
+      "loss": 1.2166,
+      "step": 275000
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 6.457739791073126e-07,
+      "loss": 1.2096,
+      "step": 275100
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 6.441911997467553e-07,
+      "loss": 1.2371,
+      "step": 275200
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 6.426084203861982e-07,
+      "loss": 1.2566,
+      "step": 275300
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 6.41025641025641e-07,
+      "loss": 1.2389,
+      "step": 275400
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 6.394428616650839e-07,
+      "loss": 1.2247,
+      "step": 275500
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 6.378600823045268e-07,
+      "loss": 1.2499,
+      "step": 275600
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 6.362773029439696e-07,
+      "loss": 1.2511,
+      "step": 275700
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 6.346945235834124e-07,
+      "loss": 1.2391,
+      "step": 275800
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 6.331117442228554e-07,
+      "loss": 1.203,
+      "step": 275900
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 6.315289648622982e-07,
+      "loss": 1.2202,
+      "step": 276000
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 6.299461855017411e-07,
+      "loss": 1.2067,
+      "step": 276100
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 6.283634061411839e-07,
+      "loss": 1.2445,
+      "step": 276200
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 6.267806267806268e-07,
+      "loss": 1.2124,
+      "step": 276300
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 6.251978474200697e-07,
+      "loss": 1.2279,
+      "step": 276400
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 6.236150680595125e-07,
+      "loss": 1.2147,
+      "step": 276500
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 6.220322886989554e-07,
+      "loss": 1.2426,
+      "step": 276600
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 6.204495093383983e-07,
+      "loss": 1.2176,
+      "step": 276700
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 6.188667299778411e-07,
+      "loss": 1.2209,
+      "step": 276800
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 6.17283950617284e-07,
+      "loss": 1.2265,
+      "step": 276900
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 6.157011712567269e-07,
+      "loss": 1.2294,
+      "step": 277000
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 6.141183918961697e-07,
+      "loss": 1.2359,
+      "step": 277100
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 6.125356125356126e-07,
+      "loss": 1.2245,
+      "step": 277200
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 6.109528331750555e-07,
+      "loss": 1.2216,
+      "step": 277300
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 6.093700538144983e-07,
+      "loss": 1.2325,
+      "step": 277400
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 6.077872744539412e-07,
+      "loss": 1.2228,
+      "step": 277500
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 6.06204495093384e-07,
+      "loss": 1.2395,
+      "step": 277600
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 6.04621715732827e-07,
+      "loss": 1.2272,
+      "step": 277700
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 6.030389363722698e-07,
+      "loss": 1.2402,
+      "step": 277800
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 6.014561570117126e-07,
+      "loss": 1.2283,
+      "step": 277900
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 5.998733776511555e-07,
+      "loss": 1.2129,
+      "step": 278000
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 5.982905982905984e-07,
+      "loss": 1.2297,
+      "step": 278100
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 5.967078189300411e-07,
+      "loss": 1.2234,
+      "step": 278200
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 5.951250395694841e-07,
+      "loss": 1.2344,
+      "step": 278300
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 5.935422602089269e-07,
+      "loss": 1.2161,
+      "step": 278400
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 5.919594808483698e-07,
+      "loss": 1.2153,
+      "step": 278500
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 5.903767014878126e-07,
+      "loss": 1.2233,
+      "step": 278600
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 5.887939221272555e-07,
+      "loss": 1.2114,
+      "step": 278700
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 5.872111427666984e-07,
+      "loss": 1.2281,
+      "step": 278800
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 5.856283634061412e-07,
+      "loss": 1.241,
+      "step": 278900
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 5.84045584045584e-07,
+      "loss": 1.2111,
+      "step": 279000
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 5.82462804685027e-07,
+      "loss": 1.211,
+      "step": 279100
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 5.808800253244698e-07,
+      "loss": 1.2186,
+      "step": 279200
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 5.792972459639127e-07,
+      "loss": 1.2508,
+      "step": 279300
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 5.777144666033555e-07,
+      "loss": 1.198,
+      "step": 279400
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 5.761316872427984e-07,
+      "loss": 1.2177,
+      "step": 279500
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 5.745489078822413e-07,
+      "loss": 1.2377,
+      "step": 279600
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 5.729661285216841e-07,
+      "loss": 1.2285,
+      "step": 279700
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 5.713833491611269e-07,
+      "loss": 1.2058,
+      "step": 279800
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 5.698005698005699e-07,
+      "loss": 1.199,
+      "step": 279900
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 5.682177904400127e-07,
+      "loss": 1.2427,
+      "step": 280000
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 5.666350110794556e-07,
+      "loss": 1.2292,
+      "step": 280100
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 5.650522317188984e-07,
+      "loss": 1.2054,
+      "step": 280200
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 5.634694523583413e-07,
+      "loss": 1.24,
+      "step": 280300
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 5.618866729977842e-07,
+      "loss": 1.2224,
+      "step": 280400
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 5.60303893637227e-07,
+      "loss": 1.2031,
+      "step": 280500
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 5.587211142766698e-07,
+      "loss": 1.2405,
+      "step": 280600
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 5.571383349161128e-07,
+      "loss": 1.2526,
+      "step": 280700
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 5.555555555555555e-07,
+      "loss": 1.2041,
+      "step": 280800
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 5.539727761949985e-07,
+      "loss": 1.2327,
+      "step": 280900
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 5.523899968344413e-07,
+      "loss": 1.2322,
+      "step": 281000
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 5.508072174738842e-07,
+      "loss": 1.2342,
+      "step": 281100
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 5.49224438113327e-07,
+      "loss": 1.2051,
+      "step": 281200
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 5.476416587527699e-07,
+      "loss": 1.242,
+      "step": 281300
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 5.460588793922129e-07,
+      "loss": 1.1998,
+      "step": 281400
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.444761000316556e-07,
+      "loss": 1.2141,
+      "step": 281500
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.428933206710985e-07,
+      "loss": 1.2291,
+      "step": 281600
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.413105413105414e-07,
+      "loss": 1.2211,
+      "step": 281700
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.397277619499842e-07,
+      "loss": 1.2303,
+      "step": 281800
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.38144982589427e-07,
+      "loss": 1.2191,
+      "step": 281900
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.3656220322887e-07,
+      "loss": 1.2361,
+      "step": 282000
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.349794238683128e-07,
+      "loss": 1.2239,
+      "step": 282100
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.333966445077556e-07,
+      "loss": 1.2313,
+      "step": 282200
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.318138651471985e-07,
+      "loss": 1.1873,
+      "step": 282300
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.302310857866414e-07,
+      "loss": 1.2486,
+      "step": 282400
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.286483064260843e-07,
+      "loss": 1.1866,
+      "step": 282500
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.270655270655271e-07,
+      "loss": 1.2402,
+      "step": 282600
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.254827477049699e-07,
+      "loss": 1.2061,
+      "step": 282700
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.238999683444129e-07,
+      "loss": 1.2431,
+      "step": 282800
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.223171889838557e-07,
+      "loss": 1.2132,
+      "step": 282900
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.207344096232985e-07,
+      "loss": 1.205,
+      "step": 283000
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.191516302627414e-07,
+      "loss": 1.2221,
+      "step": 283100
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.175688509021843e-07,
+      "loss": 1.2291,
+      "step": 283200
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.159860715416272e-07,
+      "loss": 1.2316,
+      "step": 283300
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.1440329218107e-07,
+      "loss": 1.2116,
+      "step": 283400
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.128205128205128e-07,
+      "loss": 1.2375,
+      "step": 283500
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.112377334599558e-07,
+      "loss": 1.2154,
+      "step": 283600
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.096549540993986e-07,
+      "loss": 1.2367,
+      "step": 283700
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.080721747388414e-07,
+      "loss": 1.2299,
+      "step": 283800
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.064893953782843e-07,
+      "loss": 1.2425,
+      "step": 283900
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.049066160177272e-07,
+      "loss": 1.2154,
+      "step": 284000
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.0332383665717e-07,
+      "loss": 1.2427,
+      "step": 284100
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.017410572966129e-07,
+      "loss": 1.2093,
+      "step": 284200
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.001582779360557e-07,
+      "loss": 1.2424,
+      "step": 284300
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.985754985754987e-07,
+      "loss": 1.2285,
+      "step": 284400
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.969927192149414e-07,
+      "loss": 1.2309,
+      "step": 284500
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.954099398543844e-07,
+      "loss": 1.2446,
+      "step": 284600
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.938271604938272e-07,
+      "loss": 1.2018,
+      "step": 284700
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.9224438113327e-07,
+      "loss": 1.2095,
+      "step": 284800
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.906616017727129e-07,
+      "loss": 1.2228,
+      "step": 284900
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.890788224121558e-07,
+      "loss": 1.2014,
+      "step": 285000
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.874960430515986e-07,
+      "loss": 1.2248,
+      "step": 285100
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.859132636910415e-07,
+      "loss": 1.2065,
+      "step": 285200
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.843304843304843e-07,
+      "loss": 1.21,
+      "step": 285300
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.827477049699273e-07,
+      "loss": 1.2023,
+      "step": 285400
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.811649256093701e-07,
+      "loss": 1.2253,
+      "step": 285500
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.795821462488129e-07,
+      "loss": 1.2357,
+      "step": 285600
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.779993668882559e-07,
+      "loss": 1.1998,
+      "step": 285700
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.764165875276987e-07,
+      "loss": 1.2349,
+      "step": 285800
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.748338081671416e-07,
+      "loss": 1.2145,
+      "step": 285900
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.732510288065844e-07,
+      "loss": 1.2483,
+      "step": 286000
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.716682494460273e-07,
+      "loss": 1.2169,
+      "step": 286100
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.700854700854701e-07,
+      "loss": 1.218,
+      "step": 286200
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6850269072491304e-07,
+      "loss": 1.2338,
+      "step": 286300
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.669199113643558e-07,
+      "loss": 1.2377,
+      "step": 286400
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.653371320037987e-07,
+      "loss": 1.1896,
+      "step": 286500
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.6375435264324155e-07,
+      "loss": 1.2143,
+      "step": 286600
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.6217157328268444e-07,
+      "loss": 1.2243,
+      "step": 286700
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.605887939221273e-07,
+      "loss": 1.214,
+      "step": 286800
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.5900601456157016e-07,
+      "loss": 1.2217,
+      "step": 286900
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.57423235201013e-07,
+      "loss": 1.2288,
+      "step": 287000
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.558404558404559e-07,
+      "loss": 1.2313,
+      "step": 287100
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.542576764798987e-07,
+      "loss": 1.2183,
+      "step": 287200
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.526748971193416e-07,
+      "loss": 1.2458,
+      "step": 287300
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.5109211775878445e-07,
+      "loss": 1.2088,
+      "step": 287400
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.4950933839822734e-07,
+      "loss": 1.2247,
+      "step": 287500
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.479265590376702e-07,
+      "loss": 1.232,
+      "step": 287600
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.4634377967711306e-07,
+      "loss": 1.2408,
+      "step": 287700
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.447610003165559e-07,
+      "loss": 1.2542,
+      "step": 287800
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.431782209559988e-07,
+      "loss": 1.2232,
+      "step": 287900
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.415954415954416e-07,
+      "loss": 1.2252,
+      "step": 288000
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.400126622348845e-07,
+      "loss": 1.2167,
+      "step": 288100
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.3842988287432735e-07,
+      "loss": 1.2194,
+      "step": 288200
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.3684710351377024e-07,
+      "loss": 1.2245,
+      "step": 288300
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.352643241532131e-07,
+      "loss": 1.2244,
+      "step": 288400
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.3368154479265597e-07,
+      "loss": 1.2042,
+      "step": 288500
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.320987654320988e-07,
+      "loss": 1.2115,
+      "step": 288600
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.305159860715417e-07,
+      "loss": 1.2074,
+      "step": 288700
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.289332067109845e-07,
+      "loss": 1.233,
+      "step": 288800
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.273504273504274e-07,
+      "loss": 1.2184,
+      "step": 288900
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.257676479898702e-07,
+      "loss": 1.2244,
+      "step": 289000
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.2418486862931314e-07,
+      "loss": 1.23,
+      "step": 289100
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.226020892687559e-07,
+      "loss": 1.24,
+      "step": 289200
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.210193099081988e-07,
+      "loss": 1.2348,
+      "step": 289300
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.1943653054764165e-07,
+      "loss": 1.2298,
+      "step": 289400
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.1785375118708454e-07,
+      "loss": 1.2101,
+      "step": 289500
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.162709718265274e-07,
+      "loss": 1.205,
+      "step": 289600
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.1468819246597026e-07,
+      "loss": 1.2398,
+      "step": 289700
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.131054131054131e-07,
+      "loss": 1.2069,
+      "step": 289800
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.11522633744856e-07,
+      "loss": 1.2107,
+      "step": 289900
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.099398543842989e-07,
+      "loss": 1.2189,
+      "step": 290000
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.083570750237417e-07,
+      "loss": 1.2256,
+      "step": 290100
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.067742956631846e-07,
+      "loss": 1.2452,
+      "step": 290200
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.0519151630262744e-07,
+      "loss": 1.2137,
+      "step": 290300
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.0360873694207033e-07,
+      "loss": 1.2218,
+      "step": 290400
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.0202595758151316e-07,
+      "loss": 1.2185,
+      "step": 290500
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.0044317822095605e-07,
+      "loss": 1.2252,
+      "step": 290600
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.988603988603989e-07,
+      "loss": 1.2214,
+      "step": 290700
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.972776194998418e-07,
+      "loss": 1.2231,
+      "step": 290800
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.956948401392846e-07,
+      "loss": 1.2375,
+      "step": 290900
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.941120607787275e-07,
+      "loss": 1.1954,
+      "step": 291000
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.9252928141817034e-07,
+      "loss": 1.2018,
+      "step": 291100
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.9094650205761323e-07,
+      "loss": 1.2031,
+      "step": 291200
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.8936372269705607e-07,
+      "loss": 1.2245,
+      "step": 291300
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.8778094333649895e-07,
+      "loss": 1.2096,
+      "step": 291400
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.861981639759418e-07,
+      "loss": 1.2666,
+      "step": 291500
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.846153846153847e-07,
+      "loss": 1.2485,
+      "step": 291600
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.830326052548275e-07,
+      "loss": 1.2282,
+      "step": 291700
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.814498258942704e-07,
+      "loss": 1.2175,
+      "step": 291800
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.7986704653371324e-07,
+      "loss": 1.2144,
+      "step": 291900
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.7828426717315613e-07,
+      "loss": 1.2098,
+      "step": 292000
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.767014878125989e-07,
+      "loss": 1.2274,
+      "step": 292100
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.7511870845204186e-07,
+      "loss": 1.2184,
+      "step": 292200
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.7353592909148464e-07,
+      "loss": 1.2081,
+      "step": 292300
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.719531497309276e-07,
+      "loss": 1.2077,
+      "step": 292400
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.7037037037037036e-07,
+      "loss": 1.2075,
+      "step": 292500
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.6878759100981325e-07,
+      "loss": 1.1956,
+      "step": 292600
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.672048116492561e-07,
+      "loss": 1.2263,
+      "step": 292700
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.65622032288699e-07,
+      "loss": 1.2194,
+      "step": 292800
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.640392529281418e-07,
+      "loss": 1.2004,
+      "step": 292900
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.624564735675847e-07,
+      "loss": 1.2458,
+      "step": 293000
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.6087369420702754e-07,
+      "loss": 1.2324,
+      "step": 293100
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.5929091484647043e-07,
+      "loss": 1.2312,
+      "step": 293200
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.5770813548591326e-07,
+      "loss": 1.212,
+      "step": 293300
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.5612535612535615e-07,
+      "loss": 1.1959,
+      "step": 293400
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.54542576764799e-07,
+      "loss": 1.1879,
+      "step": 293500
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.529597974042419e-07,
+      "loss": 1.2029,
+      "step": 293600
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.513770180436847e-07,
+      "loss": 1.223,
+      "step": 293700
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.497942386831276e-07,
+      "loss": 1.2122,
+      "step": 293800
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.4821145932257044e-07,
+      "loss": 1.2194,
+      "step": 293900
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.4662867996201333e-07,
+      "loss": 1.2188,
+      "step": 294000
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.4504590060145617e-07,
+      "loss": 1.233,
+      "step": 294100
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.4346312124089905e-07,
+      "loss": 1.2418,
+      "step": 294200
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.4188034188034194e-07,
+      "loss": 1.222,
+      "step": 294300
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.402975625197848e-07,
+      "loss": 1.2436,
+      "step": 294400
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.3871478315922767e-07,
+      "loss": 1.2186,
+      "step": 294500
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.371320037986705e-07,
+      "loss": 1.2061,
+      "step": 294600
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.355492244381134e-07,
+      "loss": 1.2245,
+      "step": 294700
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.3396644507755623e-07,
+      "loss": 1.2239,
+      "step": 294800
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.323836657169991e-07,
+      "loss": 1.2345,
+      "step": 294900
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.3080088635644195e-07,
+      "loss": 1.2085,
+      "step": 295000
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.2921810699588484e-07,
+      "loss": 1.1785,
+      "step": 295100
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.276353276353277e-07,
+      "loss": 1.2198,
+      "step": 295200
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.2605254827477057e-07,
+      "loss": 1.2277,
+      "step": 295300
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.2446976891421335e-07,
+      "loss": 1.1924,
+      "step": 295400
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.228869895536563e-07,
+      "loss": 1.1952,
+      "step": 295500
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.213042101930991e-07,
+      "loss": 1.2185,
+      "step": 295600
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.1972143083254197e-07,
+      "loss": 1.242,
+      "step": 295700
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.181386514719848e-07,
+      "loss": 1.2069,
+      "step": 295800
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.165558721114277e-07,
+      "loss": 1.2212,
+      "step": 295900
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.1497309275087053e-07,
+      "loss": 1.2264,
+      "step": 296000
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.133903133903134e-07,
+      "loss": 1.1997,
+      "step": 296100
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.1180753402975625e-07,
+      "loss": 1.2288,
+      "step": 296200
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.1022475466919914e-07,
+      "loss": 1.2201,
+      "step": 296300
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.08641975308642e-07,
+      "loss": 1.2367,
+      "step": 296400
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.0705919594808487e-07,
+      "loss": 1.238,
+      "step": 296500
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.0547641658752776e-07,
+      "loss": 1.2364,
+      "step": 296600
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.038936372269706e-07,
+      "loss": 1.2199,
+      "step": 296700
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.023108578664135e-07,
+      "loss": 1.202,
+      "step": 296800
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.007280785058563e-07,
+      "loss": 1.2126,
+      "step": 296900
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.991452991452992e-07,
+      "loss": 1.2282,
+      "step": 297000
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.9756251978474204e-07,
+      "loss": 1.2345,
+      "step": 297100
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.959797404241849e-07,
+      "loss": 1.2016,
+      "step": 297200
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.9439696106362777e-07,
+      "loss": 1.229,
+      "step": 297300
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.928141817030706e-07,
+      "loss": 1.2291,
+      "step": 297400
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.912314023425135e-07,
+      "loss": 1.2324,
+      "step": 297500
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.8964862298195633e-07,
+      "loss": 1.2067,
+      "step": 297600
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.880658436213992e-07,
+      "loss": 1.1926,
+      "step": 297700
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.8648306426084205e-07,
+      "loss": 1.2281,
+      "step": 297800
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.8490028490028494e-07,
+      "loss": 1.234,
+      "step": 297900
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.833175055397278e-07,
+      "loss": 1.2233,
+      "step": 298000
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.8173472617917067e-07,
+      "loss": 1.2169,
+      "step": 298100
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.801519468186135e-07,
+      "loss": 1.2032,
+      "step": 298200
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.785691674580564e-07,
+      "loss": 1.2224,
+      "step": 298300
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.7698638809749923e-07,
+      "loss": 1.2104,
+      "step": 298400
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.754036087369421e-07,
+      "loss": 1.2418,
+      "step": 298500
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.7382082937638496e-07,
+      "loss": 1.221,
+      "step": 298600
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.722380500158278e-07,
+      "loss": 1.2161,
+      "step": 298700
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.706552706552707e-07,
+      "loss": 1.2257,
+      "step": 298800
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.690724912947135e-07,
+      "loss": 1.2346,
+      "step": 298900
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.674897119341564e-07,
+      "loss": 1.2346,
+      "step": 299000
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6590693257359924e-07,
+      "loss": 1.2251,
+      "step": 299100
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6432415321304213e-07,
+      "loss": 1.231,
+      "step": 299200
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6274137385248497e-07,
+      "loss": 1.2039,
+      "step": 299300
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6115859449192786e-07,
+      "loss": 1.2161,
+      "step": 299400
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.595758151313707e-07,
+      "loss": 1.2304,
+      "step": 299500
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.579930357708136e-07,
+      "loss": 1.2324,
+      "step": 299600
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.564102564102564e-07,
+      "loss": 1.2109,
+      "step": 299700
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.548274770496993e-07,
+      "loss": 1.2228,
+      "step": 299800
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.5324469768914214e-07,
+      "loss": 1.1968,
+      "step": 299900
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.51661918328585e-07,
+      "loss": 1.2128,
+      "step": 300000
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.5007913896802787e-07,
+      "loss": 1.212,
+      "step": 300100
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.484963596074707e-07,
+      "loss": 1.2323,
+      "step": 300200
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.469135802469136e-07,
+      "loss": 1.2144,
+      "step": 300300
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.4533080088635643e-07,
+      "loss": 1.2202,
+      "step": 300400
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.437480215257993e-07,
+      "loss": 1.2058,
+      "step": 300500
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.4216524216524215e-07,
+      "loss": 1.2323,
+      "step": 300600
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.4058246280468504e-07,
+      "loss": 1.2371,
+      "step": 300700
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.3899968344412793e-07,
+      "loss": 1.243,
+      "step": 300800
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.374169040835708e-07,
+      "loss": 1.2271,
+      "step": 300900
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.3583412472301366e-07,
+      "loss": 1.2325,
+      "step": 301000
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.3425134536245652e-07,
+      "loss": 1.2121,
+      "step": 301100
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.3266856600189936e-07,
+      "loss": 1.2104,
+      "step": 301200
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.3108578664134222e-07,
+      "loss": 1.2204,
+      "step": 301300
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.2950300728078508e-07,
+      "loss": 1.2291,
+      "step": 301400
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.2792022792022794e-07,
+      "loss": 1.2444,
+      "step": 301500
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.263374485596708e-07,
+      "loss": 1.2285,
+      "step": 301600
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.2475466919911367e-07,
+      "loss": 1.2038,
+      "step": 301700
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.2317188983855653e-07,
+      "loss": 1.217,
+      "step": 301800
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.215891104779994e-07,
+      "loss": 1.215,
+      "step": 301900
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.2000633111744226e-07,
+      "loss": 1.2345,
+      "step": 302000
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.1842355175688512e-07,
+      "loss": 1.2156,
+      "step": 302100
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.1684077239632798e-07,
+      "loss": 1.2218,
+      "step": 302200
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.1525799303577085e-07,
+      "loss": 1.2102,
+      "step": 302300
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.136752136752137e-07,
+      "loss": 1.2177,
+      "step": 302400
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.1209243431465657e-07,
+      "loss": 1.2422,
+      "step": 302500
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.105096549540994e-07,
+      "loss": 1.224,
+      "step": 302600
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.0892687559354227e-07,
+      "loss": 1.2141,
+      "step": 302700
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.0734409623298513e-07,
+      "loss": 1.2344,
+      "step": 302800
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.05761316872428e-07,
+      "loss": 1.2134,
+      "step": 302900
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.0417853751187086e-07,
+      "loss": 1.2184,
+      "step": 303000
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.0259575815131372e-07,
+      "loss": 1.2245,
+      "step": 303100
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.0101297879075658e-07,
+      "loss": 1.2339,
+      "step": 303200
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.9943019943019944e-07,
+      "loss": 1.2247,
+      "step": 303300
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.978474200696423e-07,
+      "loss": 1.2184,
+      "step": 303400
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.9626464070908517e-07,
+      "loss": 1.2284,
+      "step": 303500
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.9468186134852803e-07,
+      "loss": 1.2347,
+      "step": 303600
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.930990819879709e-07,
+      "loss": 1.2117,
+      "step": 303700
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.9151630262741376e-07,
+      "loss": 1.2254,
+      "step": 303800
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.8993352326685662e-07,
+      "loss": 1.2108,
+      "step": 303900
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.8835074390629946e-07,
+      "loss": 1.2197,
+      "step": 304000
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.8676796454574232e-07,
+      "loss": 1.2218,
+      "step": 304100
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.8518518518518518e-07,
+      "loss": 1.2229,
+      "step": 304200
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.8360240582462804e-07,
+      "loss": 1.2485,
+      "step": 304300
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.820196264640709e-07,
+      "loss": 1.2303,
+      "step": 304400
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.8043684710351377e-07,
+      "loss": 1.2372,
+      "step": 304500
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.7885406774295663e-07,
+      "loss": 1.2463,
+      "step": 304600
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.772712883823995e-07,
+      "loss": 1.2267,
+      "step": 304700
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.7568850902184236e-07,
+      "loss": 1.2206,
+      "step": 304800
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.7410572966128522e-07,
+      "loss": 1.212,
+      "step": 304900
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.7252295030072808e-07,
+      "loss": 1.2316,
+      "step": 305000
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.7094017094017097e-07,
+      "loss": 1.2193,
+      "step": 305100
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.6935739157961383e-07,
+      "loss": 1.2505,
+      "step": 305200
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.677746122190567e-07,
+      "loss": 1.2173,
+      "step": 305300
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.6619183285849956e-07,
+      "loss": 1.2255,
+      "step": 305400
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.6460905349794242e-07,
+      "loss": 1.1953,
+      "step": 305500
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.6302627413738528e-07,
+      "loss": 1.2153,
+      "step": 305600
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.6144349477682815e-07,
+      "loss": 1.2301,
+      "step": 305700
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.5986071541627098e-07,
+      "loss": 1.2233,
+      "step": 305800
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.5827793605571385e-07,
+      "loss": 1.2252,
+      "step": 305900
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.566951566951567e-07,
+      "loss": 1.2092,
+      "step": 306000
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.5511237733459957e-07,
+      "loss": 1.2321,
+      "step": 306100
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.5352959797404243e-07,
+      "loss": 1.2337,
+      "step": 306200
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.519468186134853e-07,
+      "loss": 1.2172,
+      "step": 306300
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.5036403925292816e-07,
+      "loss": 1.2362,
+      "step": 306400
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.4878125989237102e-07,
+      "loss": 1.2429,
+      "step": 306500
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.4719848053181388e-07,
+      "loss": 1.2289,
+      "step": 306600
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.4561570117125675e-07,
+      "loss": 1.2219,
+      "step": 306700
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.440329218106996e-07,
+      "loss": 1.2242,
+      "step": 306800
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.4245014245014247e-07,
+      "loss": 1.1894,
+      "step": 306900
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.4086736308958533e-07,
+      "loss": 1.2284,
+      "step": 307000
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.392845837290282e-07,
+      "loss": 1.2378,
+      "step": 307100
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.3770180436847106e-07,
+      "loss": 1.2331,
+      "step": 307200
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.361190250079139e-07,
+      "loss": 1.2469,
+      "step": 307300
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.3453624564735676e-07,
+      "loss": 1.2334,
+      "step": 307400
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.3295346628679962e-07,
+      "loss": 1.2104,
+      "step": 307500
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.3137068692624248e-07,
+      "loss": 1.2208,
+      "step": 307600
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.2978790756568535e-07,
+      "loss": 1.2203,
+      "step": 307700
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.282051282051282e-07,
+      "loss": 1.219,
+      "step": 307800
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.2662234884457107e-07,
+      "loss": 1.2067,
+      "step": 307900
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.2503956948401393e-07,
+      "loss": 1.2382,
+      "step": 308000
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.234567901234568e-07,
+      "loss": 1.2283,
+      "step": 308100
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.2187401076289966e-07,
+      "loss": 1.2223,
+      "step": 308200
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.2029123140234252e-07,
+      "loss": 1.2019,
+      "step": 308300
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.187084520417854e-07,
+      "loss": 1.2453,
+      "step": 308400
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.1712567268122826e-07,
+      "loss": 1.2536,
+      "step": 308500
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.1554289332067111e-07,
+      "loss": 1.2293,
+      "step": 308600
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.1396011396011397e-07,
+      "loss": 1.2118,
+      "step": 308700
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.1237733459955683e-07,
+      "loss": 1.2164,
+      "step": 308800
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.107945552389997e-07,
+      "loss": 1.234,
+      "step": 308900
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.0921177587844256e-07,
+      "loss": 1.2312,
+      "step": 309000
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.0762899651788542e-07,
+      "loss": 1.2062,
+      "step": 309100
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.0604621715732829e-07,
+      "loss": 1.2238,
+      "step": 309200
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0446343779677113e-07,
+      "loss": 1.222,
+      "step": 309300
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.02880658436214e-07,
+      "loss": 1.2275,
+      "step": 309400
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0129787907565686e-07,
+      "loss": 1.2366,
+      "step": 309500
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 9.971509971509972e-08,
+      "loss": 1.2173,
+      "step": 309600
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 9.813232035454258e-08,
+      "loss": 1.2097,
+      "step": 309700
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 9.654954099398545e-08,
+      "loss": 1.2409,
+      "step": 309800
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 9.496676163342831e-08,
+      "loss": 1.2271,
+      "step": 309900
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 9.338398227287116e-08,
+      "loss": 1.2154,
+      "step": 310000
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 9.180120291231402e-08,
+      "loss": 1.2535,
+      "step": 310100
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 9.021842355175688e-08,
+      "loss": 1.2182,
+      "step": 310200
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 8.863564419119975e-08,
+      "loss": 1.2073,
+      "step": 310300
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 8.705286483064261e-08,
+      "loss": 1.2331,
+      "step": 310400
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 8.547008547008549e-08,
+      "loss": 1.2228,
+      "step": 310500
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 8.388730610952835e-08,
+      "loss": 1.2424,
+      "step": 310600
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 8.230452674897121e-08,
+      "loss": 1.2394,
+      "step": 310700
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 8.072174738841407e-08,
+      "loss": 1.2078,
+      "step": 310800
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.913896802785692e-08,
+      "loss": 1.2206,
+      "step": 310900
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.755618866729979e-08,
+      "loss": 1.2444,
+      "step": 311000
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.597340930674265e-08,
+      "loss": 1.2195,
+      "step": 311100
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.439062994618551e-08,
+      "loss": 1.2308,
+      "step": 311200
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.280785058562837e-08,
+      "loss": 1.2342,
+      "step": 311300
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.122507122507124e-08,
+      "loss": 1.2097,
+      "step": 311400
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 6.96422918645141e-08,
+      "loss": 1.2102,
+      "step": 311500
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 6.805951250395695e-08,
+      "loss": 1.2217,
+      "step": 311600
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 6.647673314339981e-08,
+      "loss": 1.1993,
+      "step": 311700
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.489395378284267e-08,
+      "loss": 1.2376,
+      "step": 311800
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.331117442228554e-08,
+      "loss": 1.2212,
+      "step": 311900
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.17283950617284e-08,
+      "loss": 1.1984,
+      "step": 312000
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.014561570117126e-08,
+      "loss": 1.2196,
+      "step": 312100
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 5.856283634061413e-08,
+      "loss": 1.2022,
+      "step": 312200
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 5.6980056980056986e-08,
+      "loss": 1.2208,
+      "step": 312300
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 5.539727761949985e-08,
+      "loss": 1.2318,
+      "step": 312400
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.381449825894271e-08,
+      "loss": 1.2267,
+      "step": 312500
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.223171889838557e-08,
+      "loss": 1.1789,
+      "step": 312600
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.064893953782843e-08,
+      "loss": 1.2309,
+      "step": 312700
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.906616017727129e-08,
+      "loss": 1.2244,
+      "step": 312800
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.7483380816714155e-08,
+      "loss": 1.2148,
+      "step": 312900
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.590060145615701e-08,
+      "loss": 1.2476,
+      "step": 313000
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.4317822095599874e-08,
+      "loss": 1.2182,
+      "step": 313100
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.273504273504274e-08,
+      "loss": 1.2218,
+      "step": 313200
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.1152263374485605e-08,
+      "loss": 1.2346,
+      "step": 313300
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.956948401392846e-08,
+      "loss": 1.2107,
+      "step": 313400
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.7986704653371324e-08,
+      "loss": 1.2277,
+      "step": 313500
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.6403925292814187e-08,
+      "loss": 1.2254,
+      "step": 313600
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.482114593225705e-08,
+      "loss": 1.2188,
+      "step": 313700
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.3238366571699905e-08,
+      "loss": 1.2252,
+      "step": 313800
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.165558721114277e-08,
+      "loss": 1.2335,
+      "step": 313900
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.007280785058563e-08,
+      "loss": 1.2149,
+      "step": 314000
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.8490028490028493e-08,
+      "loss": 1.2068,
+      "step": 314100
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.6907249129471356e-08,
+      "loss": 1.2225,
+      "step": 314200
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5324469768914215e-08,
+      "loss": 1.2119,
+      "step": 314300
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.3741690408357078e-08,
+      "loss": 1.2168,
+      "step": 314400
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.2158911047799937e-08,
+      "loss": 1.2255,
+      "step": 314500
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.0576131687242803e-08,
+      "loss": 1.2219,
+      "step": 314600
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.8993352326685662e-08,
+      "loss": 1.2177,
+      "step": 314700
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.7410572966128525e-08,
+      "loss": 1.2205,
+      "step": 314800
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.5827793605571384e-08,
+      "loss": 1.2058,
+      "step": 314900
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.4245014245014247e-08,
+      "loss": 1.2225,
+      "step": 315000
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.2662234884457107e-08,
+      "loss": 1.2069,
+      "step": 315100
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.1079455523899968e-08,
+      "loss": 1.2139,
+      "step": 315200
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 9.496676163342831e-09,
+      "loss": 1.2175,
+      "step": 315300
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.913896802785692e-09,
+      "loss": 1.2145,
+      "step": 315400
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.331117442228554e-09,
+      "loss": 1.1867,
+      "step": 315500
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.7483380816714155e-09,
+      "loss": 1.2111,
+      "step": 315600
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.165558721114277e-09,
+      "loss": 1.2163,
+      "step": 315700
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.5827793605571384e-09,
+      "loss": 1.2097,
+      "step": 315800
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0,
+      "loss": 1.2281,
+      "step": 315900
+    }
+  ],
+  "max_steps": 315900,
+  "num_train_epochs": 5,
+  "total_flos": 9.28756872184293e+18,
+  "trial_name": null,
+  "trial_params": null
+}