diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,18826 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.870122878876536,
+  "global_step": 313500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-05,
+      "loss": 7.5249,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8e-05,
+      "loss": 7.4245,
+      "step": 200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012,
+      "loss": 7.2963,
+      "step": 300
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016,
+      "loss": 7.1522,
+      "step": 400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 7.0025,
+      "step": 500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00024,
+      "loss": 6.9945,
+      "step": 600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00028,
+      "loss": 6.8504,
+      "step": 700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00032,
+      "loss": 6.7442,
+      "step": 800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00036,
+      "loss": 6.6172,
+      "step": 900
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004,
+      "loss": 6.4995,
+      "step": 1000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003999971000413244,
+      "loss": 6.4012,
+      "step": 1100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00039999420008264883,
+      "loss": 6.3129,
+      "step": 1200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00039999130012397324,
+      "loss": 6.2215,
+      "step": 1300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003999884001652977,
+      "loss": 6.1408,
+      "step": 1400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003999855002066221,
+      "loss": 6.0646,
+      "step": 1500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003999826002479465,
+      "loss": 5.9936,
+      "step": 1600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003999797002892709,
+      "loss": 5.9392,
+      "step": 1700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00039997680033059533,
+      "loss": 5.8718,
+      "step": 1800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003999739003719197,
+      "loss": 5.8278,
+      "step": 1900
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003999710004132441,
+      "loss": 5.7876,
+      "step": 2000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039996810045456855,
+      "loss": 5.7336,
+      "step": 2100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039996520049589296,
+      "loss": 5.6916,
+      "step": 2200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039996230053721736,
+      "loss": 5.6522,
+      "step": 2300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039995940057854177,
+      "loss": 5.6134,
+      "step": 2400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999565006198662,
+      "loss": 5.5815,
+      "step": 2500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999536006611906,
+      "loss": 5.5551,
+      "step": 2600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000399950700702515,
+      "loss": 5.5124,
+      "step": 2700
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039994780074383945,
+      "loss": 5.4807,
+      "step": 2800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039994490078516386,
+      "loss": 5.4541,
+      "step": 2900
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039994200082648827,
+      "loss": 5.4356,
+      "step": 3000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00039993910086781267,
+      "loss": 5.4024,
+      "step": 3100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999362009091371,
+      "loss": 5.3989,
+      "step": 3200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999333009504615,
+      "loss": 5.3606,
+      "step": 3300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003999304009917859,
+      "loss": 5.347,
+      "step": 3400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999275010331103,
+      "loss": 5.323,
+      "step": 3500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999246010744347,
+      "loss": 5.3208,
+      "step": 3600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999217011157591,
+      "loss": 5.2871,
+      "step": 3700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999188011570835,
+      "loss": 5.281,
+      "step": 3800
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999159011984079,
+      "loss": 5.2553,
+      "step": 3900
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039991300123973233,
+      "loss": 5.2373,
+      "step": 4000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039991010128105674,
+      "loss": 5.2314,
+      "step": 4100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039990720132238115,
+      "loss": 5.22,
+      "step": 4200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999043013637056,
+      "loss": 5.2,
+      "step": 4300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039990140140503,
+      "loss": 5.1779,
+      "step": 4400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003998985014463544,
+      "loss": 5.1744,
+      "step": 4500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039989560148767883,
+      "loss": 5.1456,
+      "step": 4600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039989270152900324,
+      "loss": 5.1417,
+      "step": 4700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039988980157032764,
+      "loss": 5.1384,
+      "step": 4800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039988690161165205,
+      "loss": 5.1298,
+      "step": 4900
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039988400165297646,
+      "loss": 5.1061,
+      "step": 5000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039988110169430086,
+      "loss": 5.1056,
+      "step": 5100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039987820173562527,
+      "loss": 5.0904,
+      "step": 5200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003998753017769497,
+      "loss": 5.0778,
+      "step": 5300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003998724018182741,
+      "loss": 5.0676,
+      "step": 5400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003998695018595985,
+      "loss": 5.0624,
+      "step": 5500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003998666019009229,
+      "loss": 5.0432,
+      "step": 5600
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039986370194224736,
+      "loss": 5.0553,
+      "step": 5700
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039986080198357177,
+      "loss": 5.0277,
+      "step": 5800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003998579020248962,
+      "loss": 5.0281,
+      "step": 5900
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003998550020662206,
+      "loss": 5.0132,
+      "step": 6000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000399852102107545,
+      "loss": 5.0255,
+      "step": 6100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003998492021488694,
+      "loss": 5.0,
+      "step": 6200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003998463021901938,
+      "loss": 4.9881,
+      "step": 6300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039984340223151826,
+      "loss": 4.9915,
+      "step": 6400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039984050227284267,
+      "loss": 4.9707,
+      "step": 6500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000399837602314167,
+      "loss": 4.9711,
+      "step": 6600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039983470235549143,
+      "loss": 4.9566,
+      "step": 6700
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039983180239681584,
+      "loss": 4.9628,
+      "step": 6800
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039982890243814024,
+      "loss": 4.9445,
+      "step": 6900
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039982600247946465,
+      "loss": 4.9404,
+      "step": 7000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003998231025207891,
+      "loss": 4.947,
+      "step": 7100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003998202025621135,
+      "loss": 4.9271,
+      "step": 7200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003998173026034379,
+      "loss": 4.927,
+      "step": 7300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039981440264476233,
+      "loss": 4.9295,
+      "step": 7400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039981150268608674,
+      "loss": 4.9216,
+      "step": 7500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039980860272741114,
+      "loss": 4.8989,
+      "step": 7600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039980570276873555,
+      "loss": 4.8931,
+      "step": 7700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039980280281006,
+      "loss": 4.898,
+      "step": 7800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003997999028513844,
+      "loss": 4.8746,
+      "step": 7900
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003997970028927088,
+      "loss": 4.8949,
+      "step": 8000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039979410293403323,
+      "loss": 4.8863,
+      "step": 8100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039979120297535764,
+      "loss": 4.876,
+      "step": 8200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039978830301668205,
+      "loss": 4.8615,
+      "step": 8300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003997854030580064,
+      "loss": 4.8695,
+      "step": 8400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039978250309933086,
+      "loss": 4.8451,
+      "step": 8500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039977960314065527,
+      "loss": 4.8589,
+      "step": 8600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003997767031819797,
+      "loss": 4.8441,
+      "step": 8700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003997738032233041,
+      "loss": 4.8367,
+      "step": 8800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003997709032646285,
+      "loss": 4.8329,
+      "step": 8900
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003997680033059529,
+      "loss": 4.8386,
+      "step": 9000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003997651033472773,
+      "loss": 4.8264,
+      "step": 9100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039976220338860176,
+      "loss": 4.8295,
+      "step": 9200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039975930342992617,
+      "loss": 4.8141,
+      "step": 9300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003997564034712506,
+      "loss": 4.8151,
+      "step": 9400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000399753503512575,
+      "loss": 4.8054,
+      "step": 9500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003997506035538994,
+      "loss": 4.8111,
+      "step": 9600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003997477035952238,
+      "loss": 4.8031,
+      "step": 9700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003997448036365482,
+      "loss": 4.7996,
+      "step": 9800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003997419036778726,
+      "loss": 4.7805,
+      "step": 9900
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000399739003719197,
+      "loss": 4.7895,
+      "step": 10000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003997361037605214,
+      "loss": 4.7956,
+      "step": 10100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039973320380184583,
+      "loss": 4.786,
+      "step": 10200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039973030384317024,
+      "loss": 4.779,
+      "step": 10300
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039972740388449465,
+      "loss": 4.7627,
+      "step": 10400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039972450392581905,
+      "loss": 4.7751,
+      "step": 10500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039972160396714346,
+      "loss": 4.7646,
+      "step": 10600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003997187040084679,
+      "loss": 4.7597,
+      "step": 10700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003997158040497923,
+      "loss": 4.7488,
+      "step": 10800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039971290409111673,
+      "loss": 4.7669,
+      "step": 10900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039971000413244114,
+      "loss": 4.7648,
+      "step": 11000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039970710417376555,
+      "loss": 4.7447,
+      "step": 11100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039970420421508995,
+      "loss": 4.758,
+      "step": 11200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039970130425641436,
+      "loss": 4.7358,
+      "step": 11300
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039969840429773877,
+      "loss": 4.7366,
+      "step": 11400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003996955043390632,
+      "loss": 4.7382,
+      "step": 11500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003996926043803876,
+      "loss": 4.7178,
+      "step": 11600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000399689704421712,
+      "loss": 4.7148,
+      "step": 11700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003996868044630364,
+      "loss": 4.7175,
+      "step": 11800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003996839045043608,
+      "loss": 4.7243,
+      "step": 11900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003996810045456852,
+      "loss": 4.7101,
+      "step": 12000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039967810458700967,
+      "loss": 4.7186,
+      "step": 12100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003996752046283341,
+      "loss": 4.7145,
+      "step": 12200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003996723046696585,
+      "loss": 4.7153,
+      "step": 12300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003996694047109829,
+      "loss": 4.7011,
+      "step": 12400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003996665047523073,
+      "loss": 4.7046,
+      "step": 12500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003996636047936317,
+      "loss": 4.7093,
+      "step": 12600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003996607048349561,
+      "loss": 4.7058,
+      "step": 12700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003996578048762806,
+      "loss": 4.6925,
+      "step": 12800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000399654904917605,
+      "loss": 4.6863,
+      "step": 12900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003996520049589294,
+      "loss": 4.6947,
+      "step": 13000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039964910500025374,
+      "loss": 4.6748,
+      "step": 13100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039964620504157815,
+      "loss": 4.6839,
+      "step": 13200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039964330508290255,
+      "loss": 4.6826,
+      "step": 13300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039964040512422696,
+      "loss": 4.6741,
+      "step": 13400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003996375051655514,
+      "loss": 4.6974,
+      "step": 13500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039963460520687583,
+      "loss": 4.6824,
+      "step": 13600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039963170524820023,
+      "loss": 4.6565,
+      "step": 13700
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039962880528952464,
+      "loss": 4.6693,
+      "step": 13800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039962590533084905,
+      "loss": 4.6679,
+      "step": 13900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039962300537217346,
+      "loss": 4.675,
+      "step": 14000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039962010541349786,
+      "loss": 4.6514,
+      "step": 14100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000399617234454409,
+      "loss": 4.6673,
+      "step": 14200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003996143344957334,
+      "loss": 4.663,
+      "step": 14300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039961143453705783,
+      "loss": 4.6478,
+      "step": 14400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003996085345783823,
+      "loss": 4.6534,
+      "step": 14500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003996056346197067,
+      "loss": 4.6648,
+      "step": 14600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003996027346610311,
+      "loss": 4.6412,
+      "step": 14700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003995998347023555,
+      "loss": 4.6512,
+      "step": 14800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003995969347436799,
+      "loss": 4.643,
+      "step": 14900
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039959403478500433,
+      "loss": 4.646,
+      "step": 15000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039959113482632873,
+      "loss": 4.6289,
+      "step": 15100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003995882348676532,
+      "loss": 4.6295,
+      "step": 15200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003995853349089776,
+      "loss": 4.6296,
+      "step": 15300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000399582434950302,
+      "loss": 4.6334,
+      "step": 15400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003995795349916264,
+      "loss": 4.6282,
+      "step": 15500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003995766350329508,
+      "loss": 4.6187,
+      "step": 15600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003995737350742752,
+      "loss": 4.6249,
+      "step": 15700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003995708351155996,
+      "loss": 4.6235,
+      "step": 15800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000399567935156924,
+      "loss": 4.6273,
+      "step": 15900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039956503519824845,
+      "loss": 4.6227,
+      "step": 16000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003995621642391596,
+      "loss": 4.6248,
+      "step": 16100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000399559264280484,
+      "loss": 4.6205,
+      "step": 16200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003995563643218085,
+      "loss": 4.6148,
+      "step": 16300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003995534643631329,
+      "loss": 4.6072,
+      "step": 16400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003995505644044573,
+      "loss": 4.6021,
+      "step": 16500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039954766444578164,
+      "loss": 4.6062,
+      "step": 16600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039954476448710605,
+      "loss": 4.6003,
+      "step": 16700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039954186452843045,
+      "loss": 4.5972,
+      "step": 16800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039953896456975486,
+      "loss": 4.5961,
+      "step": 16900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003995360646110793,
+      "loss": 4.595,
+      "step": 17000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039953316465240373,
+      "loss": 4.6016,
+      "step": 17100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039953026469372814,
+      "loss": 4.5977,
+      "step": 17200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039952736473505254,
+      "loss": 4.5895,
+      "step": 17300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039952446477637695,
+      "loss": 4.5916,
+      "step": 17400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039952156481770136,
+      "loss": 4.5927,
+      "step": 17500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039951866485902576,
+      "loss": 4.5807,
+      "step": 17600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003995157649003502,
+      "loss": 4.583,
+      "step": 17700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039951286494167463,
+      "loss": 4.5926,
+      "step": 17800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039950996498299904,
+      "loss": 4.5893,
+      "step": 17900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039950706502432345,
+      "loss": 4.5956,
+      "step": 18000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039950416506564785,
+      "loss": 4.5921,
+      "step": 18100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039950126510697226,
+      "loss": 4.5907,
+      "step": 18200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003994983651482966,
+      "loss": 4.5894,
+      "step": 18300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039949546518962107,
+      "loss": 4.5749,
+      "step": 18400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003994925652309455,
+      "loss": 4.5697,
+      "step": 18500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003994896652722699,
+      "loss": 4.5698,
+      "step": 18600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003994867653135943,
+      "loss": 4.5787,
+      "step": 18700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003994838653549187,
+      "loss": 4.5683,
+      "step": 18800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003994809653962431,
+      "loss": 4.578,
+      "step": 18900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003994780654375675,
+      "loss": 4.5643,
+      "step": 19000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000399475165478892,
+      "loss": 4.5671,
+      "step": 19100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003994722655202164,
+      "loss": 4.5668,
+      "step": 19200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003994693655615408,
+      "loss": 4.5718,
+      "step": 19300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003994664656028652,
+      "loss": 4.5703,
+      "step": 19400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003994635656441896,
+      "loss": 4.5647,
+      "step": 19500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000399460665685514,
+      "loss": 4.5664,
+      "step": 19600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003994577657268384,
+      "loss": 4.5551,
+      "step": 19700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003994548657681628,
+      "loss": 4.5567,
+      "step": 19800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00039945196580948723,
+      "loss": 4.5506,
+      "step": 19900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00039944906585081164,
+      "loss": 4.554,
+      "step": 20000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003994461948917228,
+      "loss": 4.56,
+      "step": 20100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00039944329493304725,
+      "loss": 4.5506,
+      "step": 20200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00039944039497437166,
+      "loss": 4.5531,
+      "step": 20300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00039943752401528276,
+      "loss": 4.5437,
+      "step": 20400
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003994346240566072,
+      "loss": 4.5508,
+      "step": 20500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00039943172409793163,
+      "loss": 4.5415,
+      "step": 20600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00039942882413925604,
+      "loss": 4.5366,
+      "step": 20700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00039942592418058044,
+      "loss": 4.548,
+      "step": 20800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00039942302422190485,
+      "loss": 4.5344,
+      "step": 20900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00039942012426322926,
+      "loss": 4.5442,
+      "step": 21000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00039941722430455366,
+      "loss": 4.5482,
+      "step": 21100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003994143243458781,
+      "loss": 4.5418,
+      "step": 21200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00039941142438720253,
+      "loss": 4.5367,
+      "step": 21300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039940852442852694,
+      "loss": 4.5358,
+      "step": 21400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039940562446985135,
+      "loss": 4.5447,
+      "step": 21500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039940272451117575,
+      "loss": 4.526,
+      "step": 21600
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00039939982455250016,
+      "loss": 4.5345,
+      "step": 21700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003993969245938245,
+      "loss": 4.5279,
+      "step": 21800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000399394024635149,
+      "loss": 4.5342,
+      "step": 21900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003993911246764734,
+      "loss": 4.5548,
+      "step": 22000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003993882247177978,
+      "loss": 4.5387,
+      "step": 22100
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003993853247591222,
+      "loss": 4.5268,
+      "step": 22200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003993824248004466,
+      "loss": 4.5262,
+      "step": 22300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000399379524841771,
+      "loss": 4.5308,
+      "step": 22400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003993766248830954,
+      "loss": 4.5193,
+      "step": 22500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003993737249244199,
+      "loss": 4.5265,
+      "step": 22600
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003993708249657443,
+      "loss": 4.5237,
+      "step": 22700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003993679250070687,
+      "loss": 4.5195,
+      "step": 22800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003993650250483931,
+      "loss": 4.519,
+      "step": 22900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003993621250897175,
+      "loss": 4.5273,
+      "step": 23000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003993592251310419,
+      "loss": 4.5212,
+      "step": 23100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003993563251723663,
+      "loss": 4.5279,
+      "step": 23200
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003993534252136907,
+      "loss": 4.5119,
+      "step": 23300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00039935052525501513,
+      "loss": 4.521,
+      "step": 23400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00039934762529633954,
+      "loss": 4.5117,
+      "step": 23500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00039934472533766395,
+      "loss": 4.5088,
+      "step": 23600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00039934182537898835,
+      "loss": 4.5179,
+      "step": 23700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00039933892542031276,
+      "loss": 4.5149,
+      "step": 23800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00039933602546163717,
+      "loss": 4.5133,
+      "step": 23900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00039933312550296157,
+      "loss": 4.507,
+      "step": 24000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00039933022554428603,
+      "loss": 4.5053,
+      "step": 24100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00039932732558561044,
+      "loss": 4.5091,
+      "step": 24200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003993244546265216,
+      "loss": 4.5171,
+      "step": 24300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000399321554667846,
+      "loss": 4.5014,
+      "step": 24400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003993186547091704,
+      "loss": 4.4979,
+      "step": 24500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003993157547504948,
+      "loss": 4.5012,
+      "step": 24600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003993128547918192,
+      "loss": 4.4973,
+      "step": 24700
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00039930995483314363,
+      "loss": 4.5032,
+      "step": 24800
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00039930705487446804,
+      "loss": 4.499,
+      "step": 24900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00039930415491579244,
+      "loss": 4.504,
+      "step": 25000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003993012549571169,
+      "loss": 4.4902,
+      "step": 25100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003992983549984413,
+      "loss": 4.5014,
+      "step": 25200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003992954550397657,
+      "loss": 4.4936,
+      "step": 25300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003992925550810901,
+      "loss": 4.4869,
+      "step": 25400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00039928965512241453,
+      "loss": 4.4961,
+      "step": 25500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00039928675516373894,
+      "loss": 4.4884,
+      "step": 25600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00039928385520506335,
+      "loss": 4.4965,
+      "step": 25700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00039928095524638775,
+      "loss": 4.4952,
+      "step": 25800
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00039927805528771216,
+      "loss": 4.5,
+      "step": 25900
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00039927515532903657,
+      "loss": 4.4953,
+      "step": 26000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000399272255370361,
+      "loss": 4.4753,
+      "step": 26100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003992693554116854,
+      "loss": 4.4931,
+      "step": 26200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003992664554530098,
+      "loss": 4.4685,
+      "step": 26300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000399263584493921,
+      "loss": 4.475,
+      "step": 26400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003992606845352454,
+      "loss": 4.486,
+      "step": 26500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003992577845765698,
+      "loss": 4.4834,
+      "step": 26600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003992548846178942,
+      "loss": 4.4812,
+      "step": 26700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003992519846592186,
+      "loss": 4.481,
+      "step": 26800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00039924908470054303,
+      "loss": 4.4975,
+      "step": 26900
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00039924618474186744,
+      "loss": 4.4752,
+      "step": 27000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00039924328478319185,
+      "loss": 4.4828,
+      "step": 27100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00039924038482451625,
+      "loss": 4.4886,
+      "step": 27200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00039923748486584066,
+      "loss": 4.4786,
+      "step": 27300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00039923458490716507,
+      "loss": 4.4813,
+      "step": 27400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003992316849484895,
+      "loss": 4.4642,
+      "step": 27500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00039922878498981393,
+      "loss": 4.4782,
+      "step": 27600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00039922588503113834,
+      "loss": 4.4758,
+      "step": 27700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00039922298507246275,
+      "loss": 4.4762,
+      "step": 27800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00039922008511378716,
+      "loss": 4.4768,
+      "step": 27900
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00039921718515511156,
+      "loss": 4.4591,
+      "step": 28000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00039921428519643597,
+      "loss": 4.479,
+      "step": 28100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003992113852377604,
+      "loss": 4.4681,
+      "step": 28200
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003992084852790848,
+      "loss": 4.4625,
+      "step": 28300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003992055853204092,
+      "loss": 4.4815,
+      "step": 28400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003992026853617336,
+      "loss": 4.4657,
+      "step": 28500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000399199785403058,
+      "loss": 4.4603,
+      "step": 28600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003991968854443824,
+      "loss": 4.4598,
+      "step": 28700
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003991939854857068,
+      "loss": 4.4784,
+      "step": 28800
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003991910855270312,
+      "loss": 4.4623,
+      "step": 28900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003991881855683557,
+      "loss": 4.4726,
+      "step": 29000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003991852856096801,
+      "loss": 4.4721,
+      "step": 29100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003991823856510045,
+      "loss": 4.4544,
+      "step": 29200
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003991794856923289,
+      "loss": 4.4704,
+      "step": 29300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003991765857336533,
+      "loss": 4.4737,
+      "step": 29400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003991736857749777,
+      "loss": 4.4494,
+      "step": 29500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003991707858163021,
+      "loss": 4.4581,
+      "step": 29600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003991678858576266,
+      "loss": 4.4576,
+      "step": 29700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000399164985898951,
+      "loss": 4.4729,
+      "step": 29800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003991620859402754,
+      "loss": 4.4587,
+      "step": 29900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003991591859815998,
+      "loss": 4.4684,
+      "step": 30000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00039915628602292416,
+      "loss": 4.4572,
+      "step": 30100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00039915338606424857,
+      "loss": 4.454,
+      "step": 30200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000399150486105573,
+      "loss": 4.4692,
+      "step": 30300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003991476151464842,
+      "loss": 4.462,
+      "step": 30400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003991447151878086,
+      "loss": 4.4503,
+      "step": 30500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000399141815229133,
+      "loss": 4.4513,
+      "step": 30600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00039913891527045746,
+      "loss": 4.4542,
+      "step": 30700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00039913601531178187,
+      "loss": 4.4477,
+      "step": 30800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003991331153531062,
+      "loss": 4.4425,
+      "step": 30900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003991302153944306,
+      "loss": 4.4414,
+      "step": 31000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00039912734443534184,
+      "loss": 4.4521,
+      "step": 31100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00039912444447666624,
+      "loss": 4.4455,
+      "step": 31200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00039912154451799065,
+      "loss": 4.446,
+      "step": 31300
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00039911864455931506,
+      "loss": 4.453,
+      "step": 31400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00039911574460063946,
+      "loss": 4.4424,
+      "step": 31500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00039911284464196387,
+      "loss": 4.4522,
+      "step": 31600
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003991099446832883,
+      "loss": 4.4638,
+      "step": 31700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003991070447246127,
+      "loss": 4.453,
+      "step": 31800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003991041447659371,
+      "loss": 4.4428,
+      "step": 31900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003991012448072615,
+      "loss": 4.4468,
+      "step": 32000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003990983448485859,
+      "loss": 4.4442,
+      "step": 32100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003990954448899103,
+      "loss": 4.4541,
+      "step": 32200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003990925449312347,
+      "loss": 4.4468,
+      "step": 32300
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003990896449725591,
+      "loss": 4.4621,
+      "step": 32400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003990867450138836,
+      "loss": 4.4444,
+      "step": 32500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000399083845055208,
+      "loss": 4.4453,
+      "step": 32600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003990809450965324,
+      "loss": 4.4377,
+      "step": 32700
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003990780451378568,
+      "loss": 4.4415,
+      "step": 32800
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003990751451791812,
+      "loss": 4.4396,
+      "step": 32900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003990722452205056,
+      "loss": 4.4329,
+      "step": 33000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00039906934526183003,
+      "loss": 4.4411,
+      "step": 33100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003990664453031545,
+      "loss": 4.4406,
+      "step": 33200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003990635743440656,
+      "loss": 4.4348,
+      "step": 33300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00039906067438539,
+      "loss": 4.4598,
+      "step": 33400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00039905777442671446,
+      "loss": 4.4428,
+      "step": 33500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00039905487446803887,
+      "loss": 4.4196,
+      "step": 33600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00039905197450936327,
+      "loss": 4.444,
+      "step": 33700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003990490745506877,
+      "loss": 4.4447,
+      "step": 33800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003990461745920121,
+      "loss": 4.4396,
+      "step": 33900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003990432746333365,
+      "loss": 4.4386,
+      "step": 34000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003990403746746609,
+      "loss": 4.4236,
+      "step": 34100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00039903747471598536,
+      "loss": 4.4409,
+      "step": 34200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00039903457475730977,
+      "loss": 4.4217,
+      "step": 34300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003990316747986341,
+      "loss": 4.4228,
+      "step": 34400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00039902877483995853,
+      "loss": 4.4377,
+      "step": 34500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00039902587488128293,
+      "loss": 4.4298,
+      "step": 34600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00039902297492260734,
+      "loss": 4.4369,
+      "step": 34700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00039902007496393175,
+      "loss": 4.4349,
+      "step": 34800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003990171750052562,
+      "loss": 4.4325,
+      "step": 34900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003990142750465806,
+      "loss": 4.4373,
+      "step": 35000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000399011375087905,
+      "loss": 4.4358,
+      "step": 35100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00039900847512922943,
+      "loss": 4.433,
+      "step": 35200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00039900557517055384,
+      "loss": 4.4364,
+      "step": 35300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00039900267521187824,
+      "loss": 4.4277,
+      "step": 35400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00039899977525320265,
+      "loss": 4.4235,
+      "step": 35500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00039899687529452706,
+      "loss": 4.4316,
+      "step": 35600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003989939753358515,
+      "loss": 4.4371,
+      "step": 35700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003989910753771759,
+      "loss": 4.4453,
+      "step": 35800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00039898817541850033,
+      "loss": 4.4217,
+      "step": 35900
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00039898527545982474,
+      "loss": 4.4165,
+      "step": 36000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003989824045007359,
+      "loss": 4.4134,
+      "step": 36100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003989795045420603,
+      "loss": 4.4362,
+      "step": 36200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003989766045833847,
+      "loss": 4.4136,
+      "step": 36300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003989737046247091,
+      "loss": 4.424,
+      "step": 36400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003989708046660335,
+      "loss": 4.4294,
+      "step": 36500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039896790470735793,
+      "loss": 4.4104,
+      "step": 36600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003989650047486824,
+      "loss": 4.4121,
+      "step": 36700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003989621047900068,
+      "loss": 4.4182,
+      "step": 36800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003989592048313312,
+      "loss": 4.4178,
+      "step": 36900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003989563048726556,
+      "loss": 4.4155,
+      "step": 37000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039895340491397996,
+      "loss": 4.4153,
+      "step": 37100
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039895050495530437,
+      "loss": 4.4201,
+      "step": 37200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003989476049966288,
+      "loss": 4.405,
+      "step": 37300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039894470503795324,
+      "loss": 4.4177,
+      "step": 37400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039894180507927765,
+      "loss": 4.4167,
+      "step": 37500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039893890512060205,
+      "loss": 4.4321,
+      "step": 37600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039893600516192646,
+      "loss": 4.4311,
+      "step": 37700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039893310520325087,
+      "loss": 4.4272,
+      "step": 37800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00039893020524457527,
+      "loss": 4.4111,
+      "step": 37900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003989273052858997,
+      "loss": 4.4138,
+      "step": 38000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00039892440532722414,
+      "loss": 4.425,
+      "step": 38100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00039892150536854855,
+      "loss": 4.4153,
+      "step": 38200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00039891860540987295,
+      "loss": 4.4175,
+      "step": 38300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00039891570545119736,
+      "loss": 4.4099,
+      "step": 38400
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00039891280549252177,
+      "loss": 4.4074,
+      "step": 38500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003989099345334329,
+      "loss": 4.4198,
+      "step": 38600
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00039890703457475733,
+      "loss": 4.402,
+      "step": 38700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00039890413461608174,
+      "loss": 4.422,
+      "step": 38800
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00039890123465740614,
+      "loss": 4.4098,
+      "step": 38900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00039889833469873055,
+      "loss": 4.4247,
+      "step": 39000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00039889543474005496,
+      "loss": 4.4148,
+      "step": 39100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003988925347813794,
+      "loss": 4.4229,
+      "step": 39200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003988896348227038,
+      "loss": 4.4047,
+      "step": 39300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00039888673486402823,
+      "loss": 4.4182,
+      "step": 39400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00039888383490535264,
+      "loss": 4.4045,
+      "step": 39500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00039888093494667705,
+      "loss": 4.3967,
+      "step": 39600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003988780349880014,
+      "loss": 4.4076,
+      "step": 39700
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003988751350293258,
+      "loss": 4.4214,
+      "step": 39800
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00039887223507065027,
+      "loss": 4.4008,
+      "step": 39900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003988693351119747,
+      "loss": 4.4068,
+      "step": 40000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003988664351532991,
+      "loss": 4.4005,
+      "step": 40100
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003988635351946235,
+      "loss": 4.4085,
+      "step": 40200
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003988606352359479,
+      "loss": 4.3981,
+      "step": 40300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003988577352772723,
+      "loss": 4.4089,
+      "step": 40400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003988548353185967,
+      "loss": 4.4073,
+      "step": 40500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00039885193535992117,
+      "loss": 4.3983,
+      "step": 40600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003988490354012456,
+      "loss": 4.4018,
+      "step": 40700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039884613544257,
+      "loss": 4.4078,
+      "step": 40800
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003988432354838944,
+      "loss": 4.4,
+      "step": 40900
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003988403355252188,
+      "loss": 4.4072,
+      "step": 41000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003988374355665432,
+      "loss": 4.4035,
+      "step": 41100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003988345356078676,
+      "loss": 4.4031,
+      "step": 41200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000398831635649192,
+      "loss": 4.4069,
+      "step": 41300
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003988287646901032,
+      "loss": 4.4018,
+      "step": 41400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003988258647314276,
+      "loss": 4.3995,
+      "step": 41500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039882296477275204,
+      "loss": 4.3974,
+      "step": 41600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039882006481407645,
+      "loss": 4.3838,
+      "step": 41700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039881716485540086,
+      "loss": 4.3974,
+      "step": 41800
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039881426489672526,
+      "loss": 4.3962,
+      "step": 41900
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039881136493804967,
+      "loss": 4.3919,
+      "step": 42000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003988084649793741,
+      "loss": 4.4028,
+      "step": 42100
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003988055650206985,
+      "loss": 4.3955,
+      "step": 42200
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039880269406160964,
+      "loss": 4.3963,
+      "step": 42300
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039879979410293405,
+      "loss": 4.3949,
+      "step": 42400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039879689414425845,
+      "loss": 4.4055,
+      "step": 42500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039879399418558286,
+      "loss": 4.4035,
+      "step": 42600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003987910942269073,
+      "loss": 4.3877,
+      "step": 42700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039878819426823173,
+      "loss": 4.3861,
+      "step": 42800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039878529430955613,
+      "loss": 4.3848,
+      "step": 42900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039878239435088054,
+      "loss": 4.3957,
+      "step": 43000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039877949439220495,
+      "loss": 4.3857,
+      "step": 43100
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003987765944335293,
+      "loss": 4.3898,
+      "step": 43200
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003987736944748537,
+      "loss": 4.3961,
+      "step": 43300
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039877079451617817,
+      "loss": 4.3806,
+      "step": 43400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003987678945575026,
+      "loss": 4.3897,
+      "step": 43500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000398764994598827,
+      "loss": 4.4047,
+      "step": 43600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003987620946401514,
+      "loss": 4.3974,
+      "step": 43700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003987591946814758,
+      "loss": 4.3858,
+      "step": 43800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003987562947228002,
+      "loss": 4.4,
+      "step": 43900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003987533947641246,
+      "loss": 4.3958,
+      "step": 44000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00039875049480544907,
+      "loss": 4.3926,
+      "step": 44100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003987475948467735,
+      "loss": 4.388,
+      "step": 44200
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003987446948880979,
+      "loss": 4.3938,
+      "step": 44300
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003987417949294223,
+      "loss": 4.3846,
+      "step": 44400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003987388949707467,
+      "loss": 4.389,
+      "step": 44500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003987359950120711,
+      "loss": 4.3866,
+      "step": 44600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003987330950533955,
+      "loss": 4.3773,
+      "step": 44700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003987301950947199,
+      "loss": 4.3949,
+      "step": 44800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003987272951360443,
+      "loss": 4.3797,
+      "step": 44900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00039872439517736873,
+      "loss": 4.384,
+      "step": 45000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00039872149521869314,
+      "loss": 4.3934,
+      "step": 45100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00039871859526001755,
+      "loss": 4.3909,
+      "step": 45200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00039871569530134195,
+      "loss": 4.3728,
+      "step": 45300
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00039871279534266636,
+      "loss": 4.395,
+      "step": 45400
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003987098953839908,
+      "loss": 4.3871,
+      "step": 45500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00039870699542531523,
+      "loss": 4.3832,
+      "step": 45600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00039870409546663964,
+      "loss": 4.3742,
+      "step": 45700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00039870119550796404,
+      "loss": 4.381,
+      "step": 45800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00039869829554928845,
+      "loss": 4.3971,
+      "step": 45900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00039869539559061286,
+      "loss": 4.3802,
+      "step": 46000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00039869249563193726,
+      "loss": 4.3745,
+      "step": 46100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00039868959567326167,
+      "loss": 4.3812,
+      "step": 46200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003986866957145861,
+      "loss": 4.3976,
+      "step": 46300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003986837957559105,
+      "loss": 4.386,
+      "step": 46400
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003986808957972349,
+      "loss": 4.3642,
+      "step": 46500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003986779958385593,
+      "loss": 4.3862,
+      "step": 46600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003986750958798837,
+      "loss": 4.3941,
+      "step": 46700
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003986721959212081,
+      "loss": 4.379,
+      "step": 46800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00039866929596253257,
+      "loss": 4.3804,
+      "step": 46900
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000398666396003857,
+      "loss": 4.3729,
+      "step": 47000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00039866352504476813,
+      "loss": 4.3822,
+      "step": 47100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00039866062508609254,
+      "loss": 4.3843,
+      "step": 47200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00039865772512741695,
+      "loss": 4.3838,
+      "step": 47300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00039865482516874136,
+      "loss": 4.3859,
+      "step": 47400
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00039865192521006576,
+      "loss": 4.3724,
+      "step": 47500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00039864902525139017,
+      "loss": 4.3892,
+      "step": 47600
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003986461252927146,
+      "loss": 4.3672,
+      "step": 47700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000398643225334039,
+      "loss": 4.3651,
+      "step": 47800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003986403253753634,
+      "loss": 4.3786,
+      "step": 47900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00039863742541668785,
+      "loss": 4.3967,
+      "step": 48000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00039863452545801226,
+      "loss": 4.385,
+      "step": 48100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00039863162549933666,
+      "loss": 4.3676,
+      "step": 48200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00039862872554066107,
+      "loss": 4.3751,
+      "step": 48300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003986258255819855,
+      "loss": 4.3833,
+      "step": 48400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003986229256233099,
+      "loss": 4.3806,
+      "step": 48500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003986200256646343,
+      "loss": 4.3697,
+      "step": 48600
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00039861712570595875,
+      "loss": 4.3775,
+      "step": 48700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003986142257472831,
+      "loss": 4.3717,
+      "step": 48800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003986113257886075,
+      "loss": 4.3814,
+      "step": 48900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003986084258299319,
+      "loss": 4.367,
+      "step": 49000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003986055258712563,
+      "loss": 4.3737,
+      "step": 49100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00039860265491216754,
+      "loss": 4.3704,
+      "step": 49200
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00039859975495349194,
+      "loss": 4.3675,
+      "step": 49300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00039859685499481635,
+      "loss": 4.3789,
+      "step": 49400
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00039859395503614076,
+      "loss": 4.3818,
+      "step": 49500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00039859105507746516,
+      "loss": 4.3611,
+      "step": 49600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00039858815511878957,
+      "loss": 4.367,
+      "step": 49700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000398585255160114,
+      "loss": 4.3723,
+      "step": 49800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003985823552014384,
+      "loss": 4.3899,
+      "step": 49900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003985794552427628,
+      "loss": 4.3712,
+      "step": 50000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003985765552840872,
+      "loss": 4.3684,
+      "step": 50100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003985736553254116,
+      "loss": 4.3575,
+      "step": 50200
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000398570755366736,
+      "loss": 4.3683,
+      "step": 50300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003985678554080605,
+      "loss": 4.3736,
+      "step": 50400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003985649554493849,
+      "loss": 4.3733,
+      "step": 50500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003985620554907093,
+      "loss": 4.371,
+      "step": 50600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003985591555320337,
+      "loss": 4.3776,
+      "step": 50700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003985562555733581,
+      "loss": 4.3645,
+      "step": 50800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003985533556146825,
+      "loss": 4.3786,
+      "step": 50900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003985504556560069,
+      "loss": 4.3601,
+      "step": 51000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003985475556973314,
+      "loss": 4.3573,
+      "step": 51100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003985446557386558,
+      "loss": 4.3657,
+      "step": 51200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003985417847795669,
+      "loss": 4.3718,
+      "step": 51300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003985388848208913,
+      "loss": 4.3632,
+      "step": 51400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00039853598486221575,
+      "loss": 4.3697,
+      "step": 51500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00039853308490354016,
+      "loss": 4.372,
+      "step": 51600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00039853018494486457,
+      "loss": 4.3683,
+      "step": 51700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00039852728498618897,
+      "loss": 4.3676,
+      "step": 51800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003985243850275134,
+      "loss": 4.3665,
+      "step": 51900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003985214850688378,
+      "loss": 4.3636,
+      "step": 52000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003985185851101622,
+      "loss": 4.3774,
+      "step": 52100
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00039851568515148665,
+      "loss": 4.3563,
+      "step": 52200
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000398512785192811,
+      "loss": 4.3693,
+      "step": 52300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003985098852341354,
+      "loss": 4.3636,
+      "step": 52400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003985069852754598,
+      "loss": 4.3713,
+      "step": 52500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00039850408531678423,
+      "loss": 4.361,
+      "step": 52600
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00039850118535810863,
+      "loss": 4.3664,
+      "step": 52700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00039849828539943304,
+      "loss": 4.3517,
+      "step": 52800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003984953854407575,
+      "loss": 4.3798,
+      "step": 52900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003984924854820819,
+      "loss": 4.3538,
+      "step": 53000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003984895855234063,
+      "loss": 4.3578,
+      "step": 53100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003984866855647307,
+      "loss": 4.3685,
+      "step": 53200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00039848378560605513,
+      "loss": 4.3532,
+      "step": 53300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003984809146469663,
+      "loss": 4.3596,
+      "step": 53400
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003984780146882907,
+      "loss": 4.3534,
+      "step": 53500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003984751147296151,
+      "loss": 4.3583,
+      "step": 53600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003984722147709395,
+      "loss": 4.3584,
+      "step": 53700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003984693148122639,
+      "loss": 4.3701,
+      "step": 53800
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003984664148535884,
+      "loss": 4.3599,
+      "step": 53900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003984635148949128,
+      "loss": 4.3645,
+      "step": 54000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003984606149362372,
+      "loss": 4.364,
+      "step": 54100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003984577149775616,
+      "loss": 4.3543,
+      "step": 54200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000398454815018886,
+      "loss": 4.3657,
+      "step": 54300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003984519150602104,
+      "loss": 4.366,
+      "step": 54400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003984490151015348,
+      "loss": 4.3526,
+      "step": 54500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003984461151428593,
+      "loss": 4.3572,
+      "step": 54600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003984432151841837,
+      "loss": 4.3721,
+      "step": 54700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003984403152255081,
+      "loss": 4.3658,
+      "step": 54800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039843741526683244,
+      "loss": 4.3704,
+      "step": 54900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039843451530815685,
+      "loss": 4.3646,
+      "step": 55000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039843161534948126,
+      "loss": 4.3688,
+      "step": 55100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039842871539080566,
+      "loss": 4.3529,
+      "step": 55200
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039842581543213007,
+      "loss": 4.3489,
+      "step": 55300
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039842291547345453,
+      "loss": 4.3561,
+      "step": 55400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039842001551477894,
+      "loss": 4.3474,
+      "step": 55500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039841711555610335,
+      "loss": 4.3559,
+      "step": 55600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00039841424459701456,
+      "loss": 4.3482,
+      "step": 55700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003984113446383389,
+      "loss": 4.3582,
+      "step": 55800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003984084446796633,
+      "loss": 4.3656,
+      "step": 55900
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003984055447209877,
+      "loss": 4.3684,
+      "step": 56000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00039840264476231213,
+      "loss": 4.3416,
+      "step": 56100
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00039839974480363654,
+      "loss": 4.3511,
+      "step": 56200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00039839684484496094,
+      "loss": 4.3609,
+      "step": 56300
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003983939448862854,
+      "loss": 4.3558,
+      "step": 56400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003983910449276098,
+      "loss": 4.3542,
+      "step": 56500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003983881449689342,
+      "loss": 4.3581,
+      "step": 56600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003983852450102586,
+      "loss": 4.3616,
+      "step": 56700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00039838234505158303,
+      "loss": 4.3459,
+      "step": 56800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00039837944509290744,
+      "loss": 4.3589,
+      "step": 56900
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00039837654513423185,
+      "loss": 4.3536,
+      "step": 57000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003983736451755563,
+      "loss": 4.357,
+      "step": 57100
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003983707742164674,
+      "loss": 4.3541,
+      "step": 57200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003983678742577918,
+      "loss": 4.3573,
+      "step": 57300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003983649742991163,
+      "loss": 4.3491,
+      "step": 57400
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003983620743404407,
+      "loss": 4.3425,
+      "step": 57500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003983591743817651,
+      "loss": 4.3395,
+      "step": 57600
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003983562744230895,
+      "loss": 4.354,
+      "step": 57700
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003983533744644139,
+      "loss": 4.348,
+      "step": 57800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003983504745057383,
+      "loss": 4.3514,
+      "step": 57900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003983475745470627,
+      "loss": 4.3522,
+      "step": 58000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003983446745883872,
+      "loss": 4.3412,
+      "step": 58100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003983417746297116,
+      "loss": 4.3544,
+      "step": 58200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000398338874671036,
+      "loss": 4.3416,
+      "step": 58300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00039833597471236034,
+      "loss": 4.3478,
+      "step": 58400
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00039833307475368475,
+      "loss": 4.3602,
+      "step": 58500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00039833017479500916,
+      "loss": 4.3517,
+      "step": 58600
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00039832727483633357,
+      "loss": 4.3461,
+      "step": 58700
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00039832437487765797,
+      "loss": 4.3473,
+      "step": 58800
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00039832147491898243,
+      "loss": 4.3622,
+      "step": 58900
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00039831857496030684,
+      "loss": 4.3519,
+      "step": 59000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00039831567500163125,
+      "loss": 4.3516,
+      "step": 59100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00039831277504295565,
+      "loss": 4.3558,
+      "step": 59200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00039830987508428006,
+      "loss": 4.3369,
+      "step": 59300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00039830697512560447,
+      "loss": 4.3411,
+      "step": 59400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003983040751669289,
+      "loss": 4.3516,
+      "step": 59500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00039830117520825334,
+      "loss": 4.3511,
+      "step": 59600
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00039829827524957774,
+      "loss": 4.3326,
+      "step": 59700
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00039829537529090215,
+      "loss": 4.3435,
+      "step": 59800
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00039829247533222656,
+      "loss": 4.3449,
+      "step": 59900
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00039828957537355096,
+      "loss": 4.3469,
+      "step": 60000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003982867044144621,
+      "loss": 4.3556,
+      "step": 60100
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003982838044557865,
+      "loss": 4.3438,
+      "step": 60200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00039828090449711093,
+      "loss": 4.3458,
+      "step": 60300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00039827800453843534,
+      "loss": 4.3492,
+      "step": 60400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00039827510457975975,
+      "loss": 4.3402,
+      "step": 60500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003982722046210842,
+      "loss": 4.347,
+      "step": 60600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003982693046624086,
+      "loss": 4.3419,
+      "step": 60700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000398266404703733,
+      "loss": 4.344,
+      "step": 60800
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00039826350474505743,
+      "loss": 4.3489,
+      "step": 60900
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00039826060478638183,
+      "loss": 4.3309,
+      "step": 61000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003982577048277062,
+      "loss": 4.3381,
+      "step": 61100
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003982548048690306,
+      "loss": 4.3312,
+      "step": 61200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00039825190491035506,
+      "loss": 4.3407,
+      "step": 61300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00039824900495167946,
+      "loss": 4.3458,
+      "step": 61400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00039824610499300387,
+      "loss": 4.3456,
+      "step": 61500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003982432050343283,
+      "loss": 4.3384,
+      "step": 61600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003982403050756527,
+      "loss": 4.3463,
+      "step": 61700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003982374051169771,
+      "loss": 4.3421,
+      "step": 61800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003982345051583015,
+      "loss": 4.3329,
+      "step": 61900
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00039823160519962596,
+      "loss": 4.3358,
+      "step": 62000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00039822870524095036,
+      "loss": 4.3414,
+      "step": 62100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00039822580528227477,
+      "loss": 4.3404,
+      "step": 62200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003982229053235992,
+      "loss": 4.3417,
+      "step": 62300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003982200053649236,
+      "loss": 4.3528,
+      "step": 62400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000398217105406248,
+      "loss": 4.3532,
+      "step": 62500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003982142054475724,
+      "loss": 4.3408,
+      "step": 62600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0003982113054888968,
+      "loss": 4.3464,
+      "step": 62700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003982084055302212,
+      "loss": 4.3349,
+      "step": 62800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003982055055715456,
+      "loss": 4.3414,
+      "step": 62900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00039820260561287,
+      "loss": 4.3405,
+      "step": 63000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00039819970565419443,
+      "loss": 4.3327,
+      "step": 63100
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00039819680569551884,
+      "loss": 4.3315,
+      "step": 63200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00039819390573684325,
+      "loss": 4.3471,
+      "step": 63300
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00039819100577816765,
+      "loss": 4.3362,
+      "step": 63400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003981881058194921,
+      "loss": 4.3402,
+      "step": 63500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003981852058608165,
+      "loss": 4.3534,
+      "step": 63600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00039818230590214093,
+      "loss": 4.3537,
+      "step": 63700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00039817940594346534,
+      "loss": 4.341,
+      "step": 63800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00039817650598478974,
+      "loss": 4.345,
+      "step": 63900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00039817360602611415,
+      "loss": 4.3254,
+      "step": 64000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0003981707350670253,
+      "loss": 4.3333,
+      "step": 64100
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003981678351083497,
+      "loss": 4.3414,
+      "step": 64200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003981649351496741,
+      "loss": 4.3459,
+      "step": 64300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003981620351909985,
+      "loss": 4.3469,
+      "step": 64400
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000398159135232323,
+      "loss": 4.334,
+      "step": 64500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003981562642732341,
+      "loss": 4.3412,
+      "step": 64600
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003981533643145585,
+      "loss": 4.3464,
+      "step": 64700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00039815046435588296,
+      "loss": 4.331,
+      "step": 64800
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00039814756439720736,
+      "loss": 4.3276,
+      "step": 64900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00039814466443853177,
+      "loss": 4.3396,
+      "step": 65000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003981417644798562,
+      "loss": 4.3285,
+      "step": 65100
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003981388645211806,
+      "loss": 4.332,
+      "step": 65200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000398135964562505,
+      "loss": 4.3383,
+      "step": 65300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0003981330646038294,
+      "loss": 4.3288,
+      "step": 65400
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00039813016464515386,
+      "loss": 4.3364,
+      "step": 65500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00039812726468647827,
+      "loss": 4.3462,
+      "step": 65600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00039812436472780267,
+      "loss": 4.3381,
+      "step": 65700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003981214647691271,
+      "loss": 4.3373,
+      "step": 65800
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003981185648104515,
+      "loss": 4.3279,
+      "step": 65900
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003981156648517759,
+      "loss": 4.3283,
+      "step": 66000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003981127648931003,
+      "loss": 4.3437,
+      "step": 66100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003981098649344247,
+      "loss": 4.3372,
+      "step": 66200
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003981069649757491,
+      "loss": 4.3405,
+      "step": 66300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0003981040650170735,
+      "loss": 4.3272,
+      "step": 66400
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00039810116505839793,
+      "loss": 4.3323,
+      "step": 66500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00039809826509972233,
+      "loss": 4.3339,
+      "step": 66600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00039809536514104674,
+      "loss": 4.3378,
+      "step": 66700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00039809246518237115,
+      "loss": 4.324,
+      "step": 66800
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00039808956522369556,
+      "loss": 4.3355,
+      "step": 66900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00039808666526502,
+      "loss": 4.3302,
+      "step": 67000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003980837653063444,
+      "loss": 4.3344,
+      "step": 67100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00039808086534766883,
+      "loss": 4.3298,
+      "step": 67200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00039807796538899324,
+      "loss": 4.3348,
+      "step": 67300
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00039807506543031764,
+      "loss": 4.3222,
+      "step": 67400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00039807216547164205,
+      "loss": 4.3414,
+      "step": 67500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00039806926551296646,
+      "loss": 4.3412,
+      "step": 67600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00039806636555429086,
+      "loss": 4.3311,
+      "step": 67700
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00039806346559561527,
+      "loss": 4.3226,
+      "step": 67800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003980605656369397,
+      "loss": 4.3339,
+      "step": 67900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003980576656782641,
+      "loss": 4.3287,
+      "step": 68000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003980547657195885,
+      "loss": 4.3376,
+      "step": 68100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0003980518657609129,
+      "loss": 4.3267,
+      "step": 68200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003980489658022373,
+      "loss": 4.3272,
+      "step": 68300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039804606584356177,
+      "loss": 4.3225,
+      "step": 68400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003980431658848862,
+      "loss": 4.3295,
+      "step": 68500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003980402659262106,
+      "loss": 4.3222,
+      "step": 68600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039803739496712174,
+      "loss": 4.3408,
+      "step": 68700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039803449500844614,
+      "loss": 4.3119,
+      "step": 68800
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039803159504977055,
+      "loss": 4.3384,
+      "step": 68900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039802869509109496,
+      "loss": 4.3343,
+      "step": 69000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039802579513241936,
+      "loss": 4.3281,
+      "step": 69100
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039802289517374377,
+      "loss": 4.3139,
+      "step": 69200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0003980199952150682,
+      "loss": 4.3301,
+      "step": 69300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039801709525639264,
+      "loss": 4.3332,
+      "step": 69400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039801419529771705,
+      "loss": 4.337,
+      "step": 69500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00039801129533904145,
+      "loss": 4.3271,
+      "step": 69600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00039800839538036586,
+      "loss": 4.3258,
+      "step": 69700
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00039800549542169027,
+      "loss": 4.33,
+      "step": 69800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003980025954630147,
+      "loss": 4.3234,
+      "step": 69900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003979996955043391,
+      "loss": 4.3194,
+      "step": 70000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00039799679554566354,
+      "loss": 4.3364,
+      "step": 70100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003979938955869879,
+      "loss": 4.3157,
+      "step": 70200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003979909956283123,
+      "loss": 4.3206,
+      "step": 70300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003979880956696367,
+      "loss": 4.3303,
+      "step": 70400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003979851957109611,
+      "loss": 4.3294,
+      "step": 70500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0003979822957522855,
+      "loss": 4.3323,
+      "step": 70600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00039797942479319673,
+      "loss": 4.3283,
+      "step": 70700
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00039797652483452114,
+      "loss": 4.3262,
+      "step": 70800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00039797362487584555,
+      "loss": 4.3394,
+      "step": 70900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00039797072491716995,
+      "loss": 4.3221,
+      "step": 71000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00039796782495849436,
+      "loss": 4.3174,
+      "step": 71100
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00039796492499981877,
+      "loss": 4.3255,
+      "step": 71200
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00039796202504114317,
+      "loss": 4.3246,
+      "step": 71300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003979591250824676,
+      "loss": 4.3381,
+      "step": 71400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000397956225123792,
+      "loss": 4.3167,
+      "step": 71500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003979533251651164,
+      "loss": 4.3265,
+      "step": 71600
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003979504252064408,
+      "loss": 4.3185,
+      "step": 71700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003979475252477652,
+      "loss": 4.3182,
+      "step": 71800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00039794462528908967,
+      "loss": 4.3236,
+      "step": 71900
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003979417253304141,
+      "loss": 4.3343,
+      "step": 72000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003979388253717385,
+      "loss": 4.3246,
+      "step": 72100
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003979359254130629,
+      "loss": 4.3288,
+      "step": 72200
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003979330254543873,
+      "loss": 4.3278,
+      "step": 72300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0003979301254957117,
+      "loss": 4.3328,
+      "step": 72400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003979272255370361,
+      "loss": 4.3192,
+      "step": 72500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00039792432557836057,
+      "loss": 4.3197,
+      "step": 72600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00039792145461927167,
+      "loss": 4.3286,
+      "step": 72700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003979185546605961,
+      "loss": 4.3167,
+      "step": 72800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00039791565470192054,
+      "loss": 4.3365,
+      "step": 72900
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00039791275474324495,
+      "loss": 4.3209,
+      "step": 73000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00039790985478456935,
+      "loss": 4.3276,
+      "step": 73100
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00039790695482589376,
+      "loss": 4.3296,
+      "step": 73200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00039790405486721817,
+      "loss": 4.3378,
+      "step": 73300
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003979011549085426,
+      "loss": 4.3275,
+      "step": 73400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000397898254949867,
+      "loss": 4.3136,
+      "step": 73500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00039789535499119144,
+      "loss": 4.3269,
+      "step": 73600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0003978924550325158,
+      "loss": 4.3162,
+      "step": 73700
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003978895550738402,
+      "loss": 4.3217,
+      "step": 73800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003978866551151646,
+      "loss": 4.3215,
+      "step": 73900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000397883755156489,
+      "loss": 4.3126,
+      "step": 74000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003978808551978134,
+      "loss": 4.323,
+      "step": 74100
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00039787795523913783,
+      "loss": 4.3177,
+      "step": 74200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003978750552804623,
+      "loss": 4.3188,
+      "step": 74300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003978721553217867,
+      "loss": 4.3199,
+      "step": 74400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003978692553631111,
+      "loss": 4.3167,
+      "step": 74500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003978663554044355,
+      "loss": 4.3257,
+      "step": 74600
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00039786348444534667,
+      "loss": 4.3207,
+      "step": 74700
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003978606134862578,
+      "loss": 4.3194,
+      "step": 74800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00039785771352758223,
+      "loss": 4.3163,
+      "step": 74900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003978548135689067,
+      "loss": 4.3241,
+      "step": 75000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0003978519136102311,
+      "loss": 4.3171,
+      "step": 75100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003978490136515555,
+      "loss": 4.3135,
+      "step": 75200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003978461136928799,
+      "loss": 4.2999,
+      "step": 75300
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003978432137342043,
+      "loss": 4.3271,
+      "step": 75400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003978403137755287,
+      "loss": 4.3209,
+      "step": 75500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00039783741381685313,
+      "loss": 4.3168,
+      "step": 75600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00039783451385817754,
+      "loss": 4.3166,
+      "step": 75700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00039783161389950195,
+      "loss": 4.3268,
+      "step": 75800
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00039782871394082635,
+      "loss": 4.3196,
+      "step": 75900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00039782581398215076,
+      "loss": 4.324,
+      "step": 76000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00039782291402347517,
+      "loss": 4.3112,
+      "step": 76100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0003978200140647996,
+      "loss": 4.3216,
+      "step": 76200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000397817114106124,
+      "loss": 4.3224,
+      "step": 76300
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00039781421414744844,
+      "loss": 4.3126,
+      "step": 76400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00039781131418877285,
+      "loss": 4.3174,
+      "step": 76500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00039780841423009725,
+      "loss": 4.3177,
+      "step": 76600
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00039780551427142166,
+      "loss": 4.3218,
+      "step": 76700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00039780261431274607,
+      "loss": 4.3055,
+      "step": 76800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003977997433536572,
+      "loss": 4.3088,
+      "step": 76900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00039779684339498163,
+      "loss": 4.3219,
+      "step": 77000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00039779394343630604,
+      "loss": 4.3148,
+      "step": 77100
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00039779104347763045,
+      "loss": 4.3246,
+      "step": 77200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00039778814351895485,
+      "loss": 4.321,
+      "step": 77300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00039778524356027926,
+      "loss": 4.3198,
+      "step": 77400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003977823436016037,
+      "loss": 4.3153,
+      "step": 77500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0003977794436429281,
+      "loss": 4.3154,
+      "step": 77600
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00039777654368425253,
+      "loss": 4.3111,
+      "step": 77700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00039777364372557694,
+      "loss": 4.3119,
+      "step": 77800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00039777074376690135,
+      "loss": 4.3128,
+      "step": 77900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00039776784380822575,
+      "loss": 4.3329,
+      "step": 78000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00039776494384955016,
+      "loss": 4.3164,
+      "step": 78100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00039776204389087457,
+      "loss": 4.3201,
+      "step": 78200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000397759143932199,
+      "loss": 4.3105,
+      "step": 78300
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003977562439735234,
+      "loss": 4.3049,
+      "step": 78400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003977533440148478,
+      "loss": 4.3109,
+      "step": 78500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003977504440561722,
+      "loss": 4.31,
+      "step": 78600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003977475440974966,
+      "loss": 4.3107,
+      "step": 78700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000397744644138821,
+      "loss": 4.3232,
+      "step": 78800
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00039774174418014547,
+      "loss": 4.3069,
+      "step": 78900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003977388442214699,
+      "loss": 4.3179,
+      "step": 79000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003977359442627943,
+      "loss": 4.3081,
+      "step": 79100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003977330443041187,
+      "loss": 4.3147,
+      "step": 79200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0003977301443454431,
+      "loss": 4.3211,
+      "step": 79300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003977272443867675,
+      "loss": 4.3235,
+      "step": 79400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003977243444280919,
+      "loss": 4.3216,
+      "step": 79500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039772144446941637,
+      "loss": 4.3162,
+      "step": 79600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003977185445107408,
+      "loss": 4.3045,
+      "step": 79700
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039771564455206513,
+      "loss": 4.3194,
+      "step": 79800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039771274459338954,
+      "loss": 4.3133,
+      "step": 79900
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039770984463471395,
+      "loss": 4.3221,
+      "step": 80000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039770694467603835,
+      "loss": 4.3041,
+      "step": 80100
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039770404471736276,
+      "loss": 4.3223,
+      "step": 80200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003977011447586872,
+      "loss": 4.2999,
+      "step": 80300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039769824480001163,
+      "loss": 4.3263,
+      "step": 80400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003976953738409228,
+      "loss": 4.305,
+      "step": 80500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039769247388224724,
+      "loss": 4.3165,
+      "step": 80600
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003976895739235716,
+      "loss": 4.3012,
+      "step": 80700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000397686673964896,
+      "loss": 4.3061,
+      "step": 80800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003976837740062204,
+      "loss": 4.3181,
+      "step": 80900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003976808740475448,
+      "loss": 4.3128,
+      "step": 81000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003976779740888692,
+      "loss": 4.3148,
+      "step": 81100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00039767507413019363,
+      "loss": 4.3188,
+      "step": 81200
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003976721741715181,
+      "loss": 4.3024,
+      "step": 81300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003976692742128425,
+      "loss": 4.3242,
+      "step": 81400
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003976663742541669,
+      "loss": 4.3085,
+      "step": 81500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003976634742954913,
+      "loss": 4.295,
+      "step": 81600
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003976605743368157,
+      "loss": 4.314,
+      "step": 81700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00039765767437814013,
+      "loss": 4.3074,
+      "step": 81800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00039765477441946453,
+      "loss": 4.3215,
+      "step": 81900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00039765187446078894,
+      "loss": 4.3204,
+      "step": 82000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003976489745021134,
+      "loss": 4.3203,
+      "step": 82100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003976460745434378,
+      "loss": 4.2888,
+      "step": 82200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003976431745847622,
+      "loss": 4.3087,
+      "step": 82300
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00039764027462608657,
+      "loss": 4.3037,
+      "step": 82400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000397637374667411,
+      "loss": 4.3128,
+      "step": 82500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003976344747087354,
+      "loss": 4.306,
+      "step": 82600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003976315747500598,
+      "loss": 4.3015,
+      "step": 82700
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000397628703790971,
+      "loss": 4.3198,
+      "step": 82800
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003976258038322954,
+      "loss": 4.3212,
+      "step": 82900
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003976229038736198,
+      "loss": 4.311,
+      "step": 83000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003976200039149443,
+      "loss": 4.3074,
+      "step": 83100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003976171039562687,
+      "loss": 4.3135,
+      "step": 83200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00039761420399759303,
+      "loss": 4.3029,
+      "step": 83300
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00039761130403891744,
+      "loss": 4.2986,
+      "step": 83400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00039760840408024185,
+      "loss": 4.3017,
+      "step": 83500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00039760550412156625,
+      "loss": 4.3036,
+      "step": 83600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00039760260416289066,
+      "loss": 4.3135,
+      "step": 83700
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003975997042042151,
+      "loss": 4.3113,
+      "step": 83800
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00039759680424553953,
+      "loss": 4.308,
+      "step": 83900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00039759390428686394,
+      "loss": 4.3196,
+      "step": 84000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00039759100432818834,
+      "loss": 4.2985,
+      "step": 84100
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00039758810436951275,
+      "loss": 4.3025,
+      "step": 84200
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00039758520441083716,
+      "loss": 4.3179,
+      "step": 84300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00039758230445216156,
+      "loss": 4.3193,
+      "step": 84400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000397579404493486,
+      "loss": 4.3009,
+      "step": 84500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00039757650453481043,
+      "loss": 4.3073,
+      "step": 84600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00039757360457613484,
+      "loss": 4.2989,
+      "step": 84700
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00039757070461745925,
+      "loss": 4.3057,
+      "step": 84800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00039756780465878365,
+      "loss": 4.3077,
+      "step": 84900
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000397564904700108,
+      "loss": 4.3297,
+      "step": 85000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003975620047414324,
+      "loss": 4.3152,
+      "step": 85100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00039755910478275687,
+      "loss": 4.3153,
+      "step": 85200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003975562048240813,
+      "loss": 4.3156,
+      "step": 85300
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003975533048654057,
+      "loss": 4.3197,
+      "step": 85400
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003975504049067301,
+      "loss": 4.3036,
+      "step": 85500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003975475049480545,
+      "loss": 4.3038,
+      "step": 85600
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003975446049893789,
+      "loss": 4.3113,
+      "step": 85700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003975417050307033,
+      "loss": 4.3148,
+      "step": 85800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003975388050720278,
+      "loss": 4.3065,
+      "step": 85900
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003975359051133522,
+      "loss": 4.3023,
+      "step": 86000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003975330051546766,
+      "loss": 4.3168,
+      "step": 86100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000397530105196001,
+      "loss": 4.2915,
+      "step": 86200
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003975272052373254,
+      "loss": 4.3124,
+      "step": 86300
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003975243052786498,
+      "loss": 4.3176,
+      "step": 86400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003975214053199742,
+      "loss": 4.3152,
+      "step": 86500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00039751853436088537,
+      "loss": 4.3189,
+      "step": 86600
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003975156344022098,
+      "loss": 4.3059,
+      "step": 86700
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003975127344435342,
+      "loss": 4.3068,
+      "step": 86800
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003975098344848586,
+      "loss": 4.302,
+      "step": 86900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00039750693452618305,
+      "loss": 4.3152,
+      "step": 87000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00039750403456750746,
+      "loss": 4.3042,
+      "step": 87100
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00039750113460883187,
+      "loss": 4.2914,
+      "step": 87200
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003974982346501563,
+      "loss": 4.3008,
+      "step": 87300
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003974953346914807,
+      "loss": 4.2996,
+      "step": 87400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003974924347328051,
+      "loss": 4.3072,
+      "step": 87500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003974895347741295,
+      "loss": 4.2957,
+      "step": 87600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003974866348154539,
+      "loss": 4.2925,
+      "step": 87700
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003974837348567783,
+      "loss": 4.3096,
+      "step": 87800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003974808348981027,
+      "loss": 4.2973,
+      "step": 87900
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003974779349394271,
+      "loss": 4.2975,
+      "step": 88000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00039747503498075153,
+      "loss": 4.2968,
+      "step": 88100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00039747213502207594,
+      "loss": 4.296,
+      "step": 88200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00039746923506340034,
+      "loss": 4.3037,
+      "step": 88300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003974663351047248,
+      "loss": 4.3106,
+      "step": 88400
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003974634351460492,
+      "loss": 4.3039,
+      "step": 88500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003974605351873736,
+      "loss": 4.3038,
+      "step": 88600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000397457635228698,
+      "loss": 4.3034,
+      "step": 88700
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00039745473527002243,
+      "loss": 4.3024,
+      "step": 88800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00039745183531134684,
+      "loss": 4.303,
+      "step": 88900
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00039744893535267125,
+      "loss": 4.3078,
+      "step": 89000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00039744603539399565,
+      "loss": 4.2963,
+      "step": 89100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00039744313543532006,
+      "loss": 4.3088,
+      "step": 89200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00039744023547664447,
+      "loss": 4.2959,
+      "step": 89300
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0003974373355179689,
+      "loss": 4.3016,
+      "step": 89400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0003974344355592933,
+      "loss": 4.3035,
+      "step": 89500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0003974315356006177,
+      "loss": 4.3175,
+      "step": 89600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0003974286356419421,
+      "loss": 4.3129,
+      "step": 89700
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00039742573568326655,
+      "loss": 4.3068,
+      "step": 89800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0003974228647241777,
+      "loss": 4.3021,
+      "step": 89900
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0003974199647655021,
+      "loss": 4.3034,
+      "step": 90000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0003974170648068265,
+      "loss": 4.286,
+      "step": 90100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00039741416484815093,
+      "loss": 4.3072,
+      "step": 90200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00039741126488947534,
+      "loss": 4.2942,
+      "step": 90300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00039740836493079975,
+      "loss": 4.2958,
+      "step": 90400
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00039740546497212415,
+      "loss": 4.3019,
+      "step": 90500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00039740256501344856,
+      "loss": 4.304,
+      "step": 90600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00039739966505477297,
+      "loss": 4.3096,
+      "step": 90700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00039739676509609737,
+      "loss": 4.2915,
+      "step": 90800
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00039739386513742183,
+      "loss": 4.2932,
+      "step": 90900
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00039739096517874624,
+      "loss": 4.2946,
+      "step": 91000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00039738806522007065,
+      "loss": 4.3032,
+      "step": 91100
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00039738516526139505,
+      "loss": 4.2987,
+      "step": 91200
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00039738226530271946,
+      "loss": 4.3038,
+      "step": 91300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00039737936534404387,
+      "loss": 4.3122,
+      "step": 91400
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0003973764653853683,
+      "loss": 4.2956,
+      "step": 91500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0003973735654266927,
+      "loss": 4.2915,
+      "step": 91600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0003973706654680171,
+      "loss": 4.3036,
+      "step": 91700
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003973677655093415,
+      "loss": 4.2947,
+      "step": 91800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003973648655506659,
+      "loss": 4.2907,
+      "step": 91900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003973619655919903,
+      "loss": 4.3012,
+      "step": 92000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003973590656333147,
+      "loss": 4.2921,
+      "step": 92100
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003973561656746391,
+      "loss": 4.2968,
+      "step": 92200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003973532657159636,
+      "loss": 4.3061,
+      "step": 92300
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000397350365757288,
+      "loss": 4.3078,
+      "step": 92400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003973474657986124,
+      "loss": 4.2969,
+      "step": 92500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003973445658399368,
+      "loss": 4.3104,
+      "step": 92600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003973416658812612,
+      "loss": 4.2987,
+      "step": 92700
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003973387659225856,
+      "loss": 4.3104,
+      "step": 92800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00039733586596391,
+      "loss": 4.2906,
+      "step": 92900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003973329660052345,
+      "loss": 4.2869,
+      "step": 93000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0003973300660465589,
+      "loss": 4.3019,
+      "step": 93100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0003973271660878833,
+      "loss": 4.3014,
+      "step": 93200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00039732426612920765,
+      "loss": 4.2934,
+      "step": 93300
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00039732136617053206,
+      "loss": 4.3007,
+      "step": 93400
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00039731846621185647,
+      "loss": 4.2998,
+      "step": 93500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0003973155662531809,
+      "loss": 4.292,
+      "step": 93600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00039731266629450533,
+      "loss": 4.2964,
+      "step": 93700
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00039730976633582974,
+      "loss": 4.2941,
+      "step": 93800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0003973068953767409,
+      "loss": 4.3129,
+      "step": 93900
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00039730399541806536,
+      "loss": 4.3095,
+      "step": 94000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00039730109545938977,
+      "loss": 4.3021,
+      "step": 94100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0003972981955007141,
+      "loss": 4.2855,
+      "step": 94200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0003972952955420385,
+      "loss": 4.2978,
+      "step": 94300
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00039729239558336293,
+      "loss": 4.2914,
+      "step": 94400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00039728949562468734,
+      "loss": 4.2905,
+      "step": 94500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00039728659566601175,
+      "loss": 4.3034,
+      "step": 94600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00039728369570733615,
+      "loss": 4.3056,
+      "step": 94700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0003972807957486606,
+      "loss": 4.2995,
+      "step": 94800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000397277895789985,
+      "loss": 4.2862,
+      "step": 94900
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00039727499583130943,
+      "loss": 4.2878,
+      "step": 95000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00039727209587263383,
+      "loss": 4.3099,
+      "step": 95100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00039726919591395824,
+      "loss": 4.3024,
+      "step": 95200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00039726629595528265,
+      "loss": 4.2892,
+      "step": 95300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00039726339599660705,
+      "loss": 4.293,
+      "step": 95400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0003972604960379315,
+      "loss": 4.2984,
+      "step": 95500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0003972575960792559,
+      "loss": 4.2892,
+      "step": 95600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00039725469612058033,
+      "loss": 4.3078,
+      "step": 95700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00039725179616190474,
+      "loss": 4.3032,
+      "step": 95800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00039724895420240264,
+      "loss": 4.2906,
+      "step": 95900
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00039724605424372705,
+      "loss": 4.3019,
+      "step": 96000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00039724315428505145,
+      "loss": 4.3107,
+      "step": 96100
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00039724025432637586,
+      "loss": 4.2904,
+      "step": 96200
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00039723735436770027,
+      "loss": 4.2902,
+      "step": 96300
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0003972344544090247,
+      "loss": 4.2956,
+      "step": 96400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0003972315544503491,
+      "loss": 4.3014,
+      "step": 96500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0003972286544916735,
+      "loss": 4.3016,
+      "step": 96600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0003972257545329979,
+      "loss": 4.2941,
+      "step": 96700
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00039722285457432236,
+      "loss": 4.2973,
+      "step": 96800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00039721995461564676,
+      "loss": 4.2949,
+      "step": 96900
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00039721705465697117,
+      "loss": 4.3019,
+      "step": 97000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0003972141546982956,
+      "loss": 4.3049,
+      "step": 97100
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00039721125473962,
+      "loss": 4.294,
+      "step": 97200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003972083547809444,
+      "loss": 4.2886,
+      "step": 97300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003972054548222688,
+      "loss": 4.2925,
+      "step": 97400
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00039720255486359326,
+      "loss": 4.284,
+      "step": 97500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00039719965490491767,
+      "loss": 4.2887,
+      "step": 97600
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000397196754946242,
+      "loss": 4.283,
+      "step": 97700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003971938549875664,
+      "loss": 4.285,
+      "step": 97800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00039719095502889083,
+      "loss": 4.2927,
+      "step": 97900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00039718805507021524,
+      "loss": 4.2888,
+      "step": 98000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00039718515511153965,
+      "loss": 4.305,
+      "step": 98100
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00039718225515286405,
+      "loss": 4.2909,
+      "step": 98200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003971793551941885,
+      "loss": 4.291,
+      "step": 98300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003971764552355129,
+      "loss": 4.2927,
+      "step": 98400
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00039717355527683733,
+      "loss": 4.2801,
+      "step": 98500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00039717065531816174,
+      "loss": 4.2853,
+      "step": 98600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00039716775535948614,
+      "loss": 4.2895,
+      "step": 98700
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003971648844003973,
+      "loss": 4.3037,
+      "step": 98800
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003971619844417217,
+      "loss": 4.2859,
+      "step": 98900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003971590844830461,
+      "loss": 4.2954,
+      "step": 99000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003971561845243705,
+      "loss": 4.2884,
+      "step": 99100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003971532845656949,
+      "loss": 4.2725,
+      "step": 99200
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003971503846070194,
+      "loss": 4.2974,
+      "step": 99300
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003971474846483438,
+      "loss": 4.2816,
+      "step": 99400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003971445846896682,
+      "loss": 4.2933,
+      "step": 99500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003971416847309926,
+      "loss": 4.2935,
+      "step": 99600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000397138784772317,
+      "loss": 4.2789,
+      "step": 99700
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003971358848136414,
+      "loss": 4.3014,
+      "step": 99800
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00039713298485496583,
+      "loss": 4.2906,
+      "step": 99900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003971300848962903,
+      "loss": 4.2927,
+      "step": 100000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003971271849376147,
+      "loss": 4.2868,
+      "step": 100100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003971242849789391,
+      "loss": 4.3011,
+      "step": 100200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00039712138502026346,
+      "loss": 4.2866,
+      "step": 100300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00039711848506158786,
+      "loss": 4.2973,
+      "step": 100400
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00039711558510291227,
+      "loss": 4.2933,
+      "step": 100500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003971126851442367,
+      "loss": 4.3056,
+      "step": 100600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00039710978518556114,
+      "loss": 4.3027,
+      "step": 100700
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00039710688522688554,
+      "loss": 4.2829,
+      "step": 100800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00039710398526820995,
+      "loss": 4.2927,
+      "step": 100900
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00039710108530953436,
+      "loss": 4.2806,
+      "step": 101000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00039709818535085876,
+      "loss": 4.2885,
+      "step": 101100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00039709534339135667,
+      "loss": 4.2879,
+      "step": 101200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003970924434326811,
+      "loss": 4.2853,
+      "step": 101300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00039708954347400554,
+      "loss": 4.2776,
+      "step": 101400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00039708664351532994,
+      "loss": 4.2923,
+      "step": 101500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00039708374355665435,
+      "loss": 4.3018,
+      "step": 101600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00039708084359797876,
+      "loss": 4.2882,
+      "step": 101700
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00039707794363930316,
+      "loss": 4.3042,
+      "step": 101800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00039707504368062757,
+      "loss": 4.2913,
+      "step": 101900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000397072143721952,
+      "loss": 4.2866,
+      "step": 102000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003970692437632764,
+      "loss": 4.2943,
+      "step": 102100
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003970663438046008,
+      "loss": 4.296,
+      "step": 102200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003970634438459252,
+      "loss": 4.3029,
+      "step": 102300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003970605438872496,
+      "loss": 4.2852,
+      "step": 102400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000397057643928574,
+      "loss": 4.2851,
+      "step": 102500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003970547439698984,
+      "loss": 4.2951,
+      "step": 102600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003970518440112228,
+      "loss": 4.2886,
+      "step": 102700
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003970489440525473,
+      "loss": 4.2816,
+      "step": 102800
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003970460440938717,
+      "loss": 4.2791,
+      "step": 102900
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003970431441351961,
+      "loss": 4.304,
+      "step": 103000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003970402441765205,
+      "loss": 4.3072,
+      "step": 103100
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003970373442178449,
+      "loss": 4.2974,
+      "step": 103200
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003970344442591693,
+      "loss": 4.2894,
+      "step": 103300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00039703154430049373,
+      "loss": 4.2919,
+      "step": 103400
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003970286443418182,
+      "loss": 4.2871,
+      "step": 103500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003970257443831426,
+      "loss": 4.2919,
+      "step": 103600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000397022844424467,
+      "loss": 4.2905,
+      "step": 103700
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00039701994446579136,
+      "loss": 4.2854,
+      "step": 103800
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00039701704450711576,
+      "loss": 4.2898,
+      "step": 103900
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00039701414454844017,
+      "loss": 4.2863,
+      "step": 104000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003970112445897646,
+      "loss": 4.28,
+      "step": 104100
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00039700834463108904,
+      "loss": 4.2931,
+      "step": 104200
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00039700544467241345,
+      "loss": 4.286,
+      "step": 104300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00039700254471373785,
+      "loss": 4.2823,
+      "step": 104400
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00039699964475506226,
+      "loss": 4.2969,
+      "step": 104500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00039699674479638667,
+      "loss": 4.2883,
+      "step": 104600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00039699384483771107,
+      "loss": 4.2882,
+      "step": 104700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0003969909448790355,
+      "loss": 4.2915,
+      "step": 104800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00039698804492035994,
+      "loss": 4.2948,
+      "step": 104900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00039698514496168435,
+      "loss": 4.2823,
+      "step": 105000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00039698224500300875,
+      "loss": 4.2887,
+      "step": 105100
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00039697934504433316,
+      "loss": 4.3017,
+      "step": 105200
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00039697644508565757,
+      "loss": 4.2847,
+      "step": 105300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000396973545126982,
+      "loss": 4.2847,
+      "step": 105400
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00039697064516830633,
+      "loss": 4.2762,
+      "step": 105500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003969677452096308,
+      "loss": 4.282,
+      "step": 105600
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003969648452509552,
+      "loss": 4.2733,
+      "step": 105700
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00039696197429186635,
+      "loss": 4.2831,
+      "step": 105800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00039695907433319076,
+      "loss": 4.2963,
+      "step": 105900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003969561743745152,
+      "loss": 4.2827,
+      "step": 106000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003969532744158396,
+      "loss": 4.2829,
+      "step": 106100
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00039695037445716403,
+      "loss": 4.3016,
+      "step": 106200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00039694747449848844,
+      "loss": 4.2758,
+      "step": 106300
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003969445745398128,
+      "loss": 4.285,
+      "step": 106400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003969416745811372,
+      "loss": 4.2768,
+      "step": 106500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003969387746224616,
+      "loss": 4.282,
+      "step": 106600
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00039693587466378607,
+      "loss": 4.2867,
+      "step": 106700
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003969329747051105,
+      "loss": 4.2834,
+      "step": 106800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003969300747464349,
+      "loss": 4.2813,
+      "step": 106900
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003969271747877593,
+      "loss": 4.2724,
+      "step": 107000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003969242748290837,
+      "loss": 4.2833,
+      "step": 107100
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003969213748704081,
+      "loss": 4.291,
+      "step": 107200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003969184749117325,
+      "loss": 4.3,
+      "step": 107300
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00039691557495305697,
+      "loss": 4.2695,
+      "step": 107400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003969126749943814,
+      "loss": 4.2923,
+      "step": 107500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003969097750357058,
+      "loss": 4.2882,
+      "step": 107600
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0003969068750770302,
+      "loss": 4.2899,
+      "step": 107700
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00039690400411794135,
+      "loss": 4.289,
+      "step": 107800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00039690110415926575,
+      "loss": 4.2907,
+      "step": 107900
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00039689820420059016,
+      "loss": 4.2824,
+      "step": 108000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00039689530424191457,
+      "loss": 4.2796,
+      "step": 108100
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000396892404283239,
+      "loss": 4.2789,
+      "step": 108200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003968895043245634,
+      "loss": 4.2819,
+      "step": 108300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00039688660436588784,
+      "loss": 4.2845,
+      "step": 108400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00039688370440721225,
+      "loss": 4.2861,
+      "step": 108500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00039688080444853666,
+      "loss": 4.2913,
+      "step": 108600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00039687790448986106,
+      "loss": 4.2836,
+      "step": 108700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00039687500453118547,
+      "loss": 4.2795,
+      "step": 108800
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003968721045725099,
+      "loss": 4.2933,
+      "step": 108900
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00039686920461383423,
+      "loss": 4.2742,
+      "step": 109000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003968663046551587,
+      "loss": 4.2866,
+      "step": 109100
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003968634046964831,
+      "loss": 4.282,
+      "step": 109200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003968605047378075,
+      "loss": 4.2888,
+      "step": 109300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003968576047791319,
+      "loss": 4.2737,
+      "step": 109400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003968547048204563,
+      "loss": 4.2662,
+      "step": 109500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0003968518048617807,
+      "loss": 4.277,
+      "step": 109600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00039684890490310513,
+      "loss": 4.2831,
+      "step": 109700
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00039684600494442954,
+      "loss": 4.2857,
+      "step": 109800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000396843104985754,
+      "loss": 4.2864,
+      "step": 109900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0003968402050270784,
+      "loss": 4.2856,
+      "step": 110000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0003968373340679895,
+      "loss": 4.2922,
+      "step": 110100
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00039683443410931397,
+      "loss": 4.2754,
+      "step": 110200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0003968315341506384,
+      "loss": 4.2915,
+      "step": 110300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0003968286341919628,
+      "loss": 4.2943,
+      "step": 110400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0003968257342332872,
+      "loss": 4.2867,
+      "step": 110500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0003968228342746116,
+      "loss": 4.2769,
+      "step": 110600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000396819934315936,
+      "loss": 4.2751,
+      "step": 110700
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0003968170343572604,
+      "loss": 4.2774,
+      "step": 110800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00039681416339817157,
+      "loss": 4.2794,
+      "step": 110900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00039681126343949597,
+      "loss": 4.2798,
+      "step": 111000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003968083634808204,
+      "loss": 4.2742,
+      "step": 111100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00039680546352214484,
+      "loss": 4.2794,
+      "step": 111200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00039680256356346925,
+      "loss": 4.2833,
+      "step": 111300
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00039679966360479365,
+      "loss": 4.2835,
+      "step": 111400
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00039679676364611806,
+      "loss": 4.2894,
+      "step": 111500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00039679386368744247,
+      "loss": 4.2781,
+      "step": 111600
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003967909637287669,
+      "loss": 4.2776,
+      "step": 111700
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003967880637700913,
+      "loss": 4.2741,
+      "step": 111800
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00039678516381141574,
+      "loss": 4.2827,
+      "step": 111900
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00039678226385274015,
+      "loss": 4.2771,
+      "step": 112000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00039677936389406456,
+      "loss": 4.2883,
+      "step": 112100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00039677646393538896,
+      "loss": 4.2786,
+      "step": 112200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00039677356397671337,
+      "loss": 4.2872,
+      "step": 112300
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0003967706640180378,
+      "loss": 4.2827,
+      "step": 112400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00039676776405936213,
+      "loss": 4.2862,
+      "step": 112500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003967648641006866,
+      "loss": 4.2799,
+      "step": 112600
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.000396761964142011,
+      "loss": 4.2776,
+      "step": 112700
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003967590641833354,
+      "loss": 4.2855,
+      "step": 112800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003967561642246598,
+      "loss": 4.2861,
+      "step": 112900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003967532642659842,
+      "loss": 4.2839,
+      "step": 113000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003967503643073086,
+      "loss": 4.2861,
+      "step": 113100
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00039674746434863303,
+      "loss": 4.2834,
+      "step": 113200
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00039674456438995744,
+      "loss": 4.278,
+      "step": 113300
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003967416644312819,
+      "loss": 4.2751,
+      "step": 113400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003967387644726063,
+      "loss": 4.2729,
+      "step": 113500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003967358645139307,
+      "loss": 4.2754,
+      "step": 113600
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0003967329645552551,
+      "loss": 4.277,
+      "step": 113700
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00039673006459657953,
+      "loss": 4.2856,
+      "step": 113800
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00039672716463790393,
+      "loss": 4.2828,
+      "step": 113900
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00039672426467922834,
+      "loss": 4.2771,
+      "step": 114000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00039672136472055275,
+      "loss": 4.2881,
+      "step": 114100
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00039671846476187716,
+      "loss": 4.2722,
+      "step": 114200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00039671556480320156,
+      "loss": 4.2664,
+      "step": 114300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00039671266484452597,
+      "loss": 4.2841,
+      "step": 114400
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0003967097648858504,
+      "loss": 4.2696,
+      "step": 114500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0003967068649271748,
+      "loss": 4.2749,
+      "step": 114600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0003967039649684992,
+      "loss": 4.2675,
+      "step": 114700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00039670106500982365,
+      "loss": 4.2747,
+      "step": 114800
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00039669816505114806,
+      "loss": 4.2868,
+      "step": 114900
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00039669526509247246,
+      "loss": 4.2695,
+      "step": 115000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00039669236513379687,
+      "loss": 4.2923,
+      "step": 115100
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00039668949417470803,
+      "loss": 4.2827,
+      "step": 115200
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00039668659421603243,
+      "loss": 4.2652,
+      "step": 115300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00039668369425735684,
+      "loss": 4.2896,
+      "step": 115400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00039668079429868125,
+      "loss": 4.276,
+      "step": 115500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00039667789434000565,
+      "loss": 4.2712,
+      "step": 115600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00039667499438133006,
+      "loss": 4.2788,
+      "step": 115700
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0003966720944226545,
+      "loss": 4.2892,
+      "step": 115800
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00039666919446397893,
+      "loss": 4.2666,
+      "step": 115900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00039666629450530334,
+      "loss": 4.2871,
+      "step": 116000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00039666339454662774,
+      "loss": 4.2708,
+      "step": 116100
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00039666049458795215,
+      "loss": 4.2699,
+      "step": 116200
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00039665759462927656,
+      "loss": 4.2756,
+      "step": 116300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00039665469467060096,
+      "loss": 4.2829,
+      "step": 116400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00039665179471192537,
+      "loss": 4.2591,
+      "step": 116500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0003966488947532498,
+      "loss": 4.2768,
+      "step": 116600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0003966459947945742,
+      "loss": 4.2712,
+      "step": 116700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0003966430948358986,
+      "loss": 4.2759,
+      "step": 116800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.000396640194877223,
+      "loss": 4.2799,
+      "step": 116900
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0003966372949185474,
+      "loss": 4.2729,
+      "step": 117000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0003966343949598718,
+      "loss": 4.2845,
+      "step": 117100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0003966314950011963,
+      "loss": 4.2895,
+      "step": 117200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0003966285950425207,
+      "loss": 4.2766,
+      "step": 117300
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0003966256950838451,
+      "loss": 4.2725,
+      "step": 117400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0003966227951251695,
+      "loss": 4.2705,
+      "step": 117500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00039661992416608065,
+      "loss": 4.2751,
+      "step": 117600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00039661702420740506,
+      "loss": 4.2885,
+      "step": 117700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00039661412424872946,
+      "loss": 4.2791,
+      "step": 117800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00039661122429005387,
+      "loss": 4.2657,
+      "step": 117900
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0003966083243313783,
+      "loss": 4.2747,
+      "step": 118000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0003966054243727027,
+      "loss": 4.2659,
+      "step": 118100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0003966025244140271,
+      "loss": 4.272,
+      "step": 118200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00039659962445535155,
+      "loss": 4.2731,
+      "step": 118300
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00039659672449667596,
+      "loss": 4.2728,
+      "step": 118400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00039659382453800037,
+      "loss": 4.2754,
+      "step": 118500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00039659092457932477,
+      "loss": 4.2579,
+      "step": 118600
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0003965880246206492,
+      "loss": 4.2682,
+      "step": 118700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0003965851246619736,
+      "loss": 4.2716,
+      "step": 118800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.000396582224703298,
+      "loss": 4.2762,
+      "step": 118900
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0003965793247446224,
+      "loss": 4.2748,
+      "step": 119000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0003965764247859468,
+      "loss": 4.2798,
+      "step": 119100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0003965735248272712,
+      "loss": 4.2757,
+      "step": 119200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0003965706248685956,
+      "loss": 4.2638,
+      "step": 119300
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00039656772490992003,
+      "loss": 4.2775,
+      "step": 119400
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00039656482495124443,
+      "loss": 4.2831,
+      "step": 119500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00039656192499256884,
+      "loss": 4.2815,
+      "step": 119600
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003965590250338933,
+      "loss": 4.2884,
+      "step": 119700
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003965561250752177,
+      "loss": 4.2711,
+      "step": 119800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003965532251165421,
+      "loss": 4.2808,
+      "step": 119900
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003965503251578665,
+      "loss": 4.2742,
+      "step": 120000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00039654742519919093,
+      "loss": 4.2935,
+      "step": 120100
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003965445542401021,
+      "loss": 4.2844,
+      "step": 120200
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003965416542814265,
+      "loss": 4.2746,
+      "step": 120300
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003965387543227509,
+      "loss": 4.2778,
+      "step": 120400
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003965358543640753,
+      "loss": 4.2756,
+      "step": 120500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003965329544053997,
+      "loss": 4.2768,
+      "step": 120600
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0003965300544467242,
+      "loss": 4.2817,
+      "step": 120700
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0003965271544880486,
+      "loss": 4.2652,
+      "step": 120800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.000396524254529373,
+      "loss": 4.2863,
+      "step": 120900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0003965213545706974,
+      "loss": 4.2822,
+      "step": 121000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0003965184546120218,
+      "loss": 4.2892,
+      "step": 121100
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0003965155546533462,
+      "loss": 4.2535,
+      "step": 121200
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0003965126546946706,
+      "loss": 4.2852,
+      "step": 121300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.000396509754735995,
+      "loss": 4.2759,
+      "step": 121400
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0003965068547773195,
+      "loss": 4.2878,
+      "step": 121500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0003965039548186439,
+      "loss": 4.268,
+      "step": 121600
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00039650105485996824,
+      "loss": 4.2701,
+      "step": 121700
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00039649815490129265,
+      "loss": 4.2734,
+      "step": 121800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00039649525494261706,
+      "loss": 4.2782,
+      "step": 121900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00039649235498394146,
+      "loss": 4.2584,
+      "step": 122000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00039648945502526587,
+      "loss": 4.2763,
+      "step": 122100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00039648655506659033,
+      "loss": 4.2663,
+      "step": 122200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0003964836841075015,
+      "loss": 4.284,
+      "step": 122300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0003964807841488259,
+      "loss": 4.2748,
+      "step": 122400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0003964778841901503,
+      "loss": 4.2781,
+      "step": 122500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0003964749842314747,
+      "loss": 4.2765,
+      "step": 122600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0003964720842727991,
+      "loss": 4.2673,
+      "step": 122700
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0003964691843141235,
+      "loss": 4.2709,
+      "step": 122800
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00039646628435544793,
+      "loss": 4.2804,
+      "step": 122900
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00039646338439677234,
+      "loss": 4.2731,
+      "step": 123000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00039646048443809674,
+      "loss": 4.273,
+      "step": 123100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0003964575844794212,
+      "loss": 4.2746,
+      "step": 123200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0003964546845207456,
+      "loss": 4.2671,
+      "step": 123300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00039645178456207,
+      "loss": 4.2716,
+      "step": 123400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0003964488846033944,
+      "loss": 4.2724,
+      "step": 123500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00039644598464471883,
+      "loss": 4.275,
+      "step": 123600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00039644308468604324,
+      "loss": 4.2616,
+      "step": 123700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00039644018472736764,
+      "loss": 4.2698,
+      "step": 123800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0003964372847686921,
+      "loss": 4.2842,
+      "step": 123900
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0003964343848100165,
+      "loss": 4.2763,
+      "step": 124000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0003964314848513409,
+      "loss": 4.266,
+      "step": 124100
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0003964285848926653,
+      "loss": 4.2588,
+      "step": 124200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0003964256849339897,
+      "loss": 4.2668,
+      "step": 124300
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0003964227849753141,
+      "loss": 4.2848,
+      "step": 124400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0003964198850166385,
+      "loss": 4.2707,
+      "step": 124500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00039641698505796295,
+      "loss": 4.2824,
+      "step": 124600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00039641408509928736,
+      "loss": 4.2885,
+      "step": 124700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00039641118514061177,
+      "loss": 4.2784,
+      "step": 124800
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0003964082851819362,
+      "loss": 4.2662,
+      "step": 124900
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0003964053852232606,
+      "loss": 4.2795,
+      "step": 125000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0003964025142641718,
+      "loss": 4.2695,
+      "step": 125100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00039639961430549614,
+      "loss": 4.263,
+      "step": 125200
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00039639671434682055,
+      "loss": 4.2742,
+      "step": 125300
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00039639381438814496,
+      "loss": 4.2688,
+      "step": 125400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00039639091442946937,
+      "loss": 4.2836,
+      "step": 125500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00039638801447079377,
+      "loss": 4.2806,
+      "step": 125600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00039638511451211823,
+      "loss": 4.2731,
+      "step": 125700
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00039638221455344264,
+      "loss": 4.2705,
+      "step": 125800
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00039637931459476705,
+      "loss": 4.2607,
+      "step": 125900
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00039637641463609145,
+      "loss": 4.2876,
+      "step": 126000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00039637351467741586,
+      "loss": 4.2794,
+      "step": 126100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00039637061471874027,
+      "loss": 4.2814,
+      "step": 126200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0003963677147600647,
+      "loss": 4.2742,
+      "step": 126300
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00039636481480138914,
+      "loss": 4.2698,
+      "step": 126400
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00039636191484271354,
+      "loss": 4.2679,
+      "step": 126500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00039635901488403795,
+      "loss": 4.2704,
+      "step": 126600
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00039635611492536236,
+      "loss": 4.2771,
+      "step": 126700
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00039635321496668676,
+      "loss": 4.2709,
+      "step": 126800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0003963503150080111,
+      "loss": 4.2732,
+      "step": 126900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0003963474150493355,
+      "loss": 4.2806,
+      "step": 127000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00039634451509066,
+      "loss": 4.2706,
+      "step": 127100
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0003963416151319844,
+      "loss": 4.2771,
+      "step": 127200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0003963387151733088,
+      "loss": 4.2728,
+      "step": 127300
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0003963358152146332,
+      "loss": 4.2712,
+      "step": 127400
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0003963329152559576,
+      "loss": 4.2618,
+      "step": 127500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.000396330015297282,
+      "loss": 4.2807,
+      "step": 127600
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0003963271153386064,
+      "loss": 4.2607,
+      "step": 127700
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0003963242153799309,
+      "loss": 4.2743,
+      "step": 127800
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0003963213154212553,
+      "loss": 4.2683,
+      "step": 127900
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0003963184154625797,
+      "loss": 4.2652,
+      "step": 128000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0003963155155039041,
+      "loss": 4.2703,
+      "step": 128100
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0003963126155452285,
+      "loss": 4.2654,
+      "step": 128200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0003963097155865529,
+      "loss": 4.2552,
+      "step": 128300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00039630681562787733,
+      "loss": 4.2764,
+      "step": 128400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00039630391566920173,
+      "loss": 4.2754,
+      "step": 128500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00039630101571052614,
+      "loss": 4.2797,
+      "step": 128600
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00039629811575185055,
+      "loss": 4.271,
+      "step": 128700
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00039629521579317495,
+      "loss": 4.2601,
+      "step": 128800
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00039629231583449936,
+      "loss": 4.271,
+      "step": 128900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00039628941587582377,
+      "loss": 4.2733,
+      "step": 129000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.000396286544916735,
+      "loss": 4.2675,
+      "step": 129100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0003962836449580594,
+      "loss": 4.2694,
+      "step": 129200
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0003962807449993838,
+      "loss": 4.2684,
+      "step": 129300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0003962778450407082,
+      "loss": 4.2599,
+      "step": 129400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00039627494508203255,
+      "loss": 4.2662,
+      "step": 129500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.000396272045123357,
+      "loss": 4.2849,
+      "step": 129600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0003962691451646814,
+      "loss": 4.2765,
+      "step": 129700
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0003962662452060058,
+      "loss": 4.2723,
+      "step": 129800
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00039626334524733023,
+      "loss": 4.2655,
+      "step": 129900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00039626044528865464,
+      "loss": 4.2715,
+      "step": 130000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00039625754532997905,
+      "loss": 4.2692,
+      "step": 130100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00039625464537130345,
+      "loss": 4.2724,
+      "step": 130200
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0003962517454126279,
+      "loss": 4.2797,
+      "step": 130300
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0003962488454539523,
+      "loss": 4.2683,
+      "step": 130400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0003962459744948634,
+      "loss": 4.2658,
+      "step": 130500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0003962430745361879,
+      "loss": 4.2783,
+      "step": 130600
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0003962401745775123,
+      "loss": 4.2649,
+      "step": 130700
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0003962372746188367,
+      "loss": 4.2539,
+      "step": 130800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0003962343746601611,
+      "loss": 4.2705,
+      "step": 130900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0003962314747014855,
+      "loss": 4.2675,
+      "step": 131000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0003962285747428099,
+      "loss": 4.2783,
+      "step": 131100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0003962256747841343,
+      "loss": 4.2692,
+      "step": 131200
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0003962227748254588,
+      "loss": 4.2667,
+      "step": 131300
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0003962198748667832,
+      "loss": 4.26,
+      "step": 131400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0003962169749081076,
+      "loss": 4.2503,
+      "step": 131500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00039621410394901876,
+      "loss": 4.283,
+      "step": 131600
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00039621120399034316,
+      "loss": 4.2748,
+      "step": 131700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00039620830403166757,
+      "loss": 4.2489,
+      "step": 131800
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.000396205404072992,
+      "loss": 4.2781,
+      "step": 131900
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0003962025041143164,
+      "loss": 4.2765,
+      "step": 132000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0003961996041556408,
+      "loss": 4.2661,
+      "step": 132100
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0003961967041969652,
+      "loss": 4.2691,
+      "step": 132200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00039619380423828966,
+      "loss": 4.275,
+      "step": 132300
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00039619090427961407,
+      "loss": 4.2648,
+      "step": 132400
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00039618800432093847,
+      "loss": 4.2667,
+      "step": 132500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0003961851043622629,
+      "loss": 4.2687,
+      "step": 132600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0003961822044035873,
+      "loss": 4.2784,
+      "step": 132700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0003961793044449117,
+      "loss": 4.2728,
+      "step": 132800
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0003961764044862361,
+      "loss": 4.2728,
+      "step": 132900
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00039617350452756045,
+      "loss": 4.2564,
+      "step": 133000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0003961706045688849,
+      "loss": 4.2728,
+      "step": 133100
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0003961677046102093,
+      "loss": 4.2606,
+      "step": 133200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00039616480465153373,
+      "loss": 4.2636,
+      "step": 133300
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00039616190469285813,
+      "loss": 4.2775,
+      "step": 133400
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00039615900473418254,
+      "loss": 4.2599,
+      "step": 133500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00039615610477550695,
+      "loss": 4.2682,
+      "step": 133600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00039615320481683136,
+      "loss": 4.2596,
+      "step": 133700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0003961503048581558,
+      "loss": 4.2736,
+      "step": 133800
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0003961474048994802,
+      "loss": 4.2766,
+      "step": 133900
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00039614450494080463,
+      "loss": 4.255,
+      "step": 134000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00039614160498212904,
+      "loss": 4.2703,
+      "step": 134100
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00039613870502345344,
+      "loss": 4.2619,
+      "step": 134200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00039613580506477785,
+      "loss": 4.2752,
+      "step": 134300
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00039613290510610226,
+      "loss": 4.2613,
+      "step": 134400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00039613000514742666,
+      "loss": 4.2564,
+      "step": 134500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00039612710518875107,
+      "loss": 4.2566,
+      "step": 134600
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0003961242052300755,
+      "loss": 4.2684,
+      "step": 134700
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0003961213052713999,
+      "loss": 4.2739,
+      "step": 134800
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0003961184053127243,
+      "loss": 4.2727,
+      "step": 134900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0003961155053540487,
+      "loss": 4.2665,
+      "step": 135000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0003961126053953731,
+      "loss": 4.2768,
+      "step": 135100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00039610970543669757,
+      "loss": 4.2743,
+      "step": 135200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.000396106805478022,
+      "loss": 4.2744,
+      "step": 135300
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0003961039055193464,
+      "loss": 4.2652,
+      "step": 135400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0003961010055606708,
+      "loss": 4.2782,
+      "step": 135500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0003960981056019952,
+      "loss": 4.2642,
+      "step": 135600
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0003960952056433196,
+      "loss": 4.2705,
+      "step": 135700
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00039609233468423076,
+      "loss": 4.2776,
+      "step": 135800
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00039608943472555516,
+      "loss": 4.2741,
+      "step": 135900
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00039608653476687957,
+      "loss": 4.2702,
+      "step": 136000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.000396083634808204,
+      "loss": 4.2594,
+      "step": 136100
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00039608073484952844,
+      "loss": 4.2726,
+      "step": 136200
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00039607783489085285,
+      "loss": 4.2659,
+      "step": 136300
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00039607493493217725,
+      "loss": 4.2722,
+      "step": 136400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00039607203497350166,
+      "loss": 4.2703,
+      "step": 136500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00039606913501482607,
+      "loss": 4.2611,
+      "step": 136600
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0003960662350561505,
+      "loss": 4.2531,
+      "step": 136700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0003960633350974749,
+      "loss": 4.2809,
+      "step": 136800
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0003960604351387993,
+      "loss": 4.2615,
+      "step": 136900
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0003960575351801237,
+      "loss": 4.2644,
+      "step": 137000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0003960546352214481,
+      "loss": 4.2657,
+      "step": 137100
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0003960517352627725,
+      "loss": 4.2759,
+      "step": 137200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0003960488353040969,
+      "loss": 4.2551,
+      "step": 137300
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0003960459353454213,
+      "loss": 4.2732,
+      "step": 137400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00039604303538674573,
+      "loss": 4.273,
+      "step": 137500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00039604013542807014,
+      "loss": 4.2673,
+      "step": 137600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0003960372354693946,
+      "loss": 4.2651,
+      "step": 137700
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000396034335510719,
+      "loss": 4.2722,
+      "step": 137800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0003960314355520434,
+      "loss": 4.2633,
+      "step": 137900
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00039602856459295457,
+      "loss": 4.279,
+      "step": 138000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00039602566463427897,
+      "loss": 4.2348,
+      "step": 138100
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0003960227646756034,
+      "loss": 4.2516,
+      "step": 138200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0003960198647169278,
+      "loss": 4.2515,
+      "step": 138300
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0003960169647582522,
+      "loss": 4.2477,
+      "step": 138400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0003960140647995766,
+      "loss": 4.2495,
+      "step": 138500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000396011164840901,
+      "loss": 4.2598,
+      "step": 138600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00039600826488222547,
+      "loss": 4.2388,
+      "step": 138700
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0003960053649235499,
+      "loss": 4.2453,
+      "step": 138800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0003960024649648743,
+      "loss": 4.2645,
+      "step": 138900
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0003959995650061987,
+      "loss": 4.2496,
+      "step": 139000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0003959966650475231,
+      "loss": 4.2489,
+      "step": 139100
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0003959937650888475,
+      "loss": 4.2465,
+      "step": 139200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0003959908651301719,
+      "loss": 4.2596,
+      "step": 139300
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00039598796517149637,
+      "loss": 4.2449,
+      "step": 139400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0003959850652128207,
+      "loss": 4.2567,
+      "step": 139500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00039598216525414513,
+      "loss": 4.2672,
+      "step": 139600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00039597926529546954,
+      "loss": 4.2543,
+      "step": 139700
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00039597636533679394,
+      "loss": 4.2485,
+      "step": 139800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00039597346537811835,
+      "loss": 4.2451,
+      "step": 139900
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00039597056541944276,
+      "loss": 4.2691,
+      "step": 140000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0003959676654607672,
+      "loss": 4.2628,
+      "step": 140100
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0003959647655020916,
+      "loss": 4.2664,
+      "step": 140200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00039596186554341603,
+      "loss": 4.2495,
+      "step": 140300
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0003959589945843272,
+      "loss": 4.2386,
+      "step": 140400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0003959560946256516,
+      "loss": 4.2492,
+      "step": 140500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.000395953194666976,
+      "loss": 4.2657,
+      "step": 140600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0003959502947083004,
+      "loss": 4.2436,
+      "step": 140700
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0003959473947496248,
+      "loss": 4.2594,
+      "step": 140800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0003959444947909492,
+      "loss": 4.2587,
+      "step": 140900
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00039594159483227363,
+      "loss": 4.2593,
+      "step": 141000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00039593869487359804,
+      "loss": 4.2574,
+      "step": 141100
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0003959357949149225,
+      "loss": 4.2377,
+      "step": 141200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0003959328949562469,
+      "loss": 4.2616,
+      "step": 141300
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0003959299949975713,
+      "loss": 4.248,
+      "step": 141400
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0003959270950388957,
+      "loss": 4.2529,
+      "step": 141500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0003959241950802201,
+      "loss": 4.255,
+      "step": 141600
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00039592129512154453,
+      "loss": 4.2466,
+      "step": 141700
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00039591839516286894,
+      "loss": 4.2584,
+      "step": 141800
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0003959154952041934,
+      "loss": 4.2561,
+      "step": 141900
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0003959125952455178,
+      "loss": 4.2589,
+      "step": 142000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00039590969528684216,
+      "loss": 4.2486,
+      "step": 142100
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00039590679532816657,
+      "loss": 4.2545,
+      "step": 142200
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.000395903895369491,
+      "loss": 4.2491,
+      "step": 142300
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0003959009954108154,
+      "loss": 4.2506,
+      "step": 142400
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0003958981244517266,
+      "loss": 4.2516,
+      "step": 142500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.000395895224493051,
+      "loss": 4.2553,
+      "step": 142600
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0003958923245343754,
+      "loss": 4.2579,
+      "step": 142700
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0003958894245756998,
+      "loss": 4.2466,
+      "step": 142800
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00039588652461702427,
+      "loss": 4.251,
+      "step": 142900
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0003958836246583486,
+      "loss": 4.265,
+      "step": 143000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00039588072469967303,
+      "loss": 4.2554,
+      "step": 143100
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00039587782474099744,
+      "loss": 4.2616,
+      "step": 143200
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00039587492478232184,
+      "loss": 4.2547,
+      "step": 143300
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00039587202482364625,
+      "loss": 4.2471,
+      "step": 143400
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00039586912486497066,
+      "loss": 4.2546,
+      "step": 143500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0003958662249062951,
+      "loss": 4.2579,
+      "step": 143600
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0003958633249476195,
+      "loss": 4.2548,
+      "step": 143700
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00039586042498894393,
+      "loss": 4.261,
+      "step": 143800
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00039585752503026834,
+      "loss": 4.2619,
+      "step": 143900
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00039585462507159275,
+      "loss": 4.2535,
+      "step": 144000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00039585172511291715,
+      "loss": 4.2474,
+      "step": 144100
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00039584882515424156,
+      "loss": 4.2622,
+      "step": 144200
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.000395845925195566,
+      "loss": 4.257,
+      "step": 144300
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00039584302523689043,
+      "loss": 4.2449,
+      "step": 144400
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00039584012527821484,
+      "loss": 4.2581,
+      "step": 144500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00039583722531953924,
+      "loss": 4.2738,
+      "step": 144600
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00039583432536086365,
+      "loss": 4.2582,
+      "step": 144700
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0003958314544017748,
+      "loss": 4.2646,
+      "step": 144800
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0003958285544430992,
+      "loss": 4.2498,
+      "step": 144900
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0003958256544844236,
+      "loss": 4.2492,
+      "step": 145000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.000395822754525748,
+      "loss": 4.2535,
+      "step": 145100
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00039581985456707243,
+      "loss": 4.2594,
+      "step": 145200
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00039581695460839684,
+      "loss": 4.2462,
+      "step": 145300
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0003958140546497213,
+      "loss": 4.261,
+      "step": 145400
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0003958111546910457,
+      "loss": 4.2398,
+      "step": 145500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00039580825473237006,
+      "loss": 4.2438,
+      "step": 145600
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00039580535477369447,
+      "loss": 4.2646,
+      "step": 145700
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0003958024548150189,
+      "loss": 4.2509,
+      "step": 145800
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0003957995548563433,
+      "loss": 4.2446,
+      "step": 145900
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0003957966548976677,
+      "loss": 4.2389,
+      "step": 146000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00039579375493899215,
+      "loss": 4.247,
+      "step": 146100
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00039579085498031656,
+      "loss": 4.2615,
+      "step": 146200
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00039578795502164096,
+      "loss": 4.2644,
+      "step": 146300
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00039578505506296537,
+      "loss": 4.2549,
+      "step": 146400
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0003957821551042898,
+      "loss": 4.2529,
+      "step": 146500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0003957792551456142,
+      "loss": 4.2562,
+      "step": 146600
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0003957763551869386,
+      "loss": 4.2523,
+      "step": 146700
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00039577345522826305,
+      "loss": 4.2565,
+      "step": 146800
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00039577055526958746,
+      "loss": 4.256,
+      "step": 146900
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00039576765531091187,
+      "loss": 4.2638,
+      "step": 147000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.000395764784351823,
+      "loss": 4.2532,
+      "step": 147100
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00039576188439314743,
+      "loss": 4.2485,
+      "step": 147200
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00039575898443447183,
+      "loss": 4.2545,
+      "step": 147300
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00039575608447579624,
+      "loss": 4.2574,
+      "step": 147400
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00039575318451712065,
+      "loss": 4.2408,
+      "step": 147500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00039575028455844506,
+      "loss": 4.2522,
+      "step": 147600
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00039574738459976946,
+      "loss": 4.2633,
+      "step": 147700
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0003957444846410939,
+      "loss": 4.2535,
+      "step": 147800
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00039574158468241833,
+      "loss": 4.2572,
+      "step": 147900
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00039573868472374274,
+      "loss": 4.2556,
+      "step": 148000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00039573578476506714,
+      "loss": 4.2544,
+      "step": 148100
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00039573288480639155,
+      "loss": 4.2551,
+      "step": 148200
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0003957299848477159,
+      "loss": 4.2632,
+      "step": 148300
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0003957270848890403,
+      "loss": 4.2524,
+      "step": 148400
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00039572418493036477,
+      "loss": 4.2406,
+      "step": 148500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0003957212849716892,
+      "loss": 4.2583,
+      "step": 148600
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0003957183850130136,
+      "loss": 4.2527,
+      "step": 148700
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.000395715485054338,
+      "loss": 4.2723,
+      "step": 148800
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0003957125850956624,
+      "loss": 4.2573,
+      "step": 148900
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0003957096851369868,
+      "loss": 4.264,
+      "step": 149000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.000395706814177898,
+      "loss": 4.2452,
+      "step": 149100
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00039570391421922237,
+      "loss": 4.2617,
+      "step": 149200
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0003957010142605468,
+      "loss": 4.2467,
+      "step": 149300
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0003956981143018712,
+      "loss": 4.2598,
+      "step": 149400
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0003956952143431956,
+      "loss": 4.268,
+      "step": 149500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00039569231438452005,
+      "loss": 4.2565,
+      "step": 149600
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00039568941442584446,
+      "loss": 4.2635,
+      "step": 149700
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00039568651446716886,
+      "loss": 4.2547,
+      "step": 149800
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00039568361450849327,
+      "loss": 4.2567,
+      "step": 149900
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0003956807145498177,
+      "loss": 4.251,
+      "step": 150000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0003956778145911421,
+      "loss": 4.2569,
+      "step": 150100
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0003956749146324665,
+      "loss": 4.2409,
+      "step": 150200
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00039567201467379095,
+      "loss": 4.2351,
+      "step": 150300
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00039566911471511536,
+      "loss": 4.2555,
+      "step": 150400
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00039566621475643977,
+      "loss": 4.2598,
+      "step": 150500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0003956633147977642,
+      "loss": 4.2591,
+      "step": 150600
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0003956604148390886,
+      "loss": 4.2627,
+      "step": 150700
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.000395657514880413,
+      "loss": 4.2596,
+      "step": 150800
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00039565461492173734,
+      "loss": 4.2493,
+      "step": 150900
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00039565174396264855,
+      "loss": 4.2587,
+      "step": 151000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00039564884400397296,
+      "loss": 4.2508,
+      "step": 151100
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00039564594404529736,
+      "loss": 4.2541,
+      "step": 151200
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0003956430440866218,
+      "loss": 4.2507,
+      "step": 151300
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00039564014412794623,
+      "loss": 4.2361,
+      "step": 151400
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00039563724416927064,
+      "loss": 4.2563,
+      "step": 151500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00039563434421059505,
+      "loss": 4.2622,
+      "step": 151600
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00039563144425191945,
+      "loss": 4.2608,
+      "step": 151700
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0003956285442932438,
+      "loss": 4.2494,
+      "step": 151800
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0003956256443345682,
+      "loss": 4.2624,
+      "step": 151900
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00039562274437589267,
+      "loss": 4.2411,
+      "step": 152000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0003956198444172171,
+      "loss": 4.2463,
+      "step": 152100
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0003956169444585415,
+      "loss": 4.2665,
+      "step": 152200
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0003956140444998659,
+      "loss": 4.2685,
+      "step": 152300
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0003956111445411903,
+      "loss": 4.2457,
+      "step": 152400
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0003956082445825147,
+      "loss": 4.2617,
+      "step": 152500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0003956053446238391,
+      "loss": 4.2457,
+      "step": 152600
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0003956024446651635,
+      "loss": 4.2505,
+      "step": 152700
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.000395599544706488,
+      "loss": 4.2533,
+      "step": 152800
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0003955966447478124,
+      "loss": 4.2467,
+      "step": 152900
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0003955937447891368,
+      "loss": 4.2475,
+      "step": 153000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0003955908448304612,
+      "loss": 4.265,
+      "step": 153100
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0003955879448717856,
+      "loss": 4.2647,
+      "step": 153200
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00039558504491311,
+      "loss": 4.2636,
+      "step": 153300
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0003955821449544344,
+      "loss": 4.2619,
+      "step": 153400
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00039557924499575883,
+      "loss": 4.2391,
+      "step": 153500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00039557634503708324,
+      "loss": 4.2457,
+      "step": 153600
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00039557344507840764,
+      "loss": 4.2502,
+      "step": 153700
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00039557054511973205,
+      "loss": 4.2532,
+      "step": 153800
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00039556764516105646,
+      "loss": 4.269,
+      "step": 153900
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00039556474520238086,
+      "loss": 4.2413,
+      "step": 154000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00039556184524370527,
+      "loss": 4.2538,
+      "step": 154100
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00039555894528502973,
+      "loss": 4.2508,
+      "step": 154200
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00039555604532635414,
+      "loss": 4.2591,
+      "step": 154300
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00039555314536767855,
+      "loss": 4.2544,
+      "step": 154400
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00039555024540900295,
+      "loss": 4.2476,
+      "step": 154500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00039554734545032736,
+      "loss": 4.2444,
+      "step": 154600
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00039554444549165177,
+      "loss": 4.2557,
+      "step": 154700
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0003955415455329762,
+      "loss": 4.2562,
+      "step": 154800
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0003955386455743006,
+      "loss": 4.2605,
+      "step": 154900
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.000395535745615625,
+      "loss": 4.2444,
+      "step": 155000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00039553287465653614,
+      "loss": 4.2384,
+      "step": 155100
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0003955299746978606,
+      "loss": 4.2535,
+      "step": 155200
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.000395527074739185,
+      "loss": 4.2563,
+      "step": 155300
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0003955241747805094,
+      "loss": 4.2638,
+      "step": 155400
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0003955212748218338,
+      "loss": 4.2588,
+      "step": 155500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00039551837486315823,
+      "loss": 4.2572,
+      "step": 155600
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00039551547490448264,
+      "loss": 4.2492,
+      "step": 155700
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00039551257494580705,
+      "loss": 4.264,
+      "step": 155800
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00039550967498713145,
+      "loss": 4.2591,
+      "step": 155900
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00039550677502845586,
+      "loss": 4.2505,
+      "step": 156000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00039550387506978027,
+      "loss": 4.2703,
+      "step": 156100
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0003955009751111047,
+      "loss": 4.2503,
+      "step": 156200
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0003954980751524291,
+      "loss": 4.252,
+      "step": 156300
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0003954951751937535,
+      "loss": 4.2568,
+      "step": 156400
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0003954922752350779,
+      "loss": 4.2623,
+      "step": 156500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00039548937527640235,
+      "loss": 4.2493,
+      "step": 156600
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00039548647531772676,
+      "loss": 4.2528,
+      "step": 156700
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00039548357535905117,
+      "loss": 4.2652,
+      "step": 156800
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0003954806754003756,
+      "loss": 4.2479,
+      "step": 156900
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0003954777754417,
+      "loss": 4.2426,
+      "step": 157000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00039547490448261114,
+      "loss": 4.2587,
+      "step": 157100
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00039547200452393554,
+      "loss": 4.2479,
+      "step": 157200
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00039546910456525995,
+      "loss": 4.2462,
+      "step": 157300
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00039546620460658436,
+      "loss": 4.2426,
+      "step": 157400
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00039546330464790877,
+      "loss": 4.2572,
+      "step": 157500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00039546040468923317,
+      "loss": 4.2526,
+      "step": 157600
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00039545750473055763,
+      "loss": 4.2522,
+      "step": 157700
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00039545460477188204,
+      "loss": 4.2531,
+      "step": 157800
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00039545170481320645,
+      "loss": 4.2479,
+      "step": 157900
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00039544880485453085,
+      "loss": 4.2396,
+      "step": 158000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00039544590489585526,
+      "loss": 4.2475,
+      "step": 158100
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00039544300493717967,
+      "loss": 4.2521,
+      "step": 158200
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0003954401049785041,
+      "loss": 4.2605,
+      "step": 158300
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0003954372050198285,
+      "loss": 4.2524,
+      "step": 158400
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0003954343050611529,
+      "loss": 4.2557,
+      "step": 158500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0003954314051024773,
+      "loss": 4.2417,
+      "step": 158600
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0003954285051438017,
+      "loss": 4.2463,
+      "step": 158700
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0003954256051851261,
+      "loss": 4.2477,
+      "step": 158800
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0003954227052264505,
+      "loss": 4.2471,
+      "step": 158900
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0003954198052677749,
+      "loss": 4.2452,
+      "step": 159000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0003954169053090994,
+      "loss": 4.2543,
+      "step": 159100
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0003954140053504238,
+      "loss": 4.2592,
+      "step": 159200
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0003954111053917482,
+      "loss": 4.2399,
+      "step": 159300
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00039540823443265935,
+      "loss": 4.2619,
+      "step": 159400
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00039540533447398376,
+      "loss": 4.2622,
+      "step": 159500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00039540243451530817,
+      "loss": 4.2482,
+      "step": 159600
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0003953995345566326,
+      "loss": 4.2518,
+      "step": 159700
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.000395396634597957,
+      "loss": 4.2489,
+      "step": 159800
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0003953937346392814,
+      "loss": 4.2451,
+      "step": 159900
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0003953908346806058,
+      "loss": 4.2581,
+      "step": 160000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00039538793472193026,
+      "loss": 4.2473,
+      "step": 160100
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00039538503476325466,
+      "loss": 4.2599,
+      "step": 160200
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00039538213480457907,
+      "loss": 4.2561,
+      "step": 160300
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0003953792348459035,
+      "loss": 4.2408,
+      "step": 160400
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0003953763348872279,
+      "loss": 4.2504,
+      "step": 160500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0003953734349285523,
+      "loss": 4.2531,
+      "step": 160600
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0003953705349698767,
+      "loss": 4.2596,
+      "step": 160700
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0003953676350112011,
+      "loss": 4.2453,
+      "step": 160800
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0003953647350525255,
+      "loss": 4.2489,
+      "step": 160900
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0003953618350938499,
+      "loss": 4.2474,
+      "step": 161000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0003953589351351743,
+      "loss": 4.2477,
+      "step": 161100
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00039535603517649873,
+      "loss": 4.2698,
+      "step": 161200
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00039535313521782314,
+      "loss": 4.2399,
+      "step": 161300
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00039535023525914755,
+      "loss": 4.2424,
+      "step": 161400
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00039534733530047195,
+      "loss": 4.2503,
+      "step": 161500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0003953444353417964,
+      "loss": 4.2413,
+      "step": 161600
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00039534156438270757,
+      "loss": 4.2328,
+      "step": 161700
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.000395338664424032,
+      "loss": 4.2463,
+      "step": 161800
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0003953357644653564,
+      "loss": 4.265,
+      "step": 161900
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00039533289350626754,
+      "loss": 4.2594,
+      "step": 162000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00039532999354759195,
+      "loss": 4.2401,
+      "step": 162100
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0003953270935889164,
+      "loss": 4.2512,
+      "step": 162200
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0003953241936302408,
+      "loss": 4.2466,
+      "step": 162300
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0003953212936715652,
+      "loss": 4.2471,
+      "step": 162400
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00039531839371288963,
+      "loss": 4.2502,
+      "step": 162500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00039531549375421403,
+      "loss": 4.255,
+      "step": 162600
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00039531259379553844,
+      "loss": 4.2443,
+      "step": 162700
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00039530969383686285,
+      "loss": 4.2465,
+      "step": 162800
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00039530679387818725,
+      "loss": 4.2448,
+      "step": 162900
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00039530389391951166,
+      "loss": 4.2576,
+      "step": 163000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00039530099396083607,
+      "loss": 4.2583,
+      "step": 163100
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0003952980940021605,
+      "loss": 4.2422,
+      "step": 163200
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0003952951940434849,
+      "loss": 4.2529,
+      "step": 163300
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0003952922940848093,
+      "loss": 4.2453,
+      "step": 163400
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0003952893941261337,
+      "loss": 4.2632,
+      "step": 163500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00039528649416745816,
+      "loss": 4.2557,
+      "step": 163600
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00039528359420878256,
+      "loss": 4.2465,
+      "step": 163700
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00039528069425010697,
+      "loss": 4.2573,
+      "step": 163800
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0003952777942914314,
+      "loss": 4.258,
+      "step": 163900
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0003952748943327558,
+      "loss": 4.2399,
+      "step": 164000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0003952719943740802,
+      "loss": 4.2495,
+      "step": 164100
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0003952690944154046,
+      "loss": 4.2519,
+      "step": 164200
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.000395266194456729,
+      "loss": 4.2527,
+      "step": 164300
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0003952632944980534,
+      "loss": 4.2697,
+      "step": 164400
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0003952603945393778,
+      "loss": 4.2511,
+      "step": 164500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0003952574945807022,
+      "loss": 4.2514,
+      "step": 164600
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00039525459462202663,
+      "loss": 4.2456,
+      "step": 164700
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00039525169466335104,
+      "loss": 4.2442,
+      "step": 164800
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00039524879470467545,
+      "loss": 4.2546,
+      "step": 164900
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00039524589474599985,
+      "loss": 4.2513,
+      "step": 165000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0003952429947873243,
+      "loss": 4.2516,
+      "step": 165100
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0003952400948286487,
+      "loss": 4.2343,
+      "step": 165200
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00039523719486997313,
+      "loss": 4.2417,
+      "step": 165300
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00039523429491129754,
+      "loss": 4.252,
+      "step": 165400
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00039523139495262194,
+      "loss": 4.2574,
+      "step": 165500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00039522849499394635,
+      "loss": 4.2377,
+      "step": 165600
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00039522559503527076,
+      "loss": 4.2505,
+      "step": 165700
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0003952226950765952,
+      "loss": 4.2502,
+      "step": 165800
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0003952197951179196,
+      "loss": 4.2452,
+      "step": 165900
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0003952169241588307,
+      "loss": 4.2442,
+      "step": 166000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0003952140242001552,
+      "loss": 4.2651,
+      "step": 166100
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0003952111242414796,
+      "loss": 4.2394,
+      "step": 166200
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.000395208224282804,
+      "loss": 4.2423,
+      "step": 166300
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0003952053243241284,
+      "loss": 4.2465,
+      "step": 166400
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0003952024243654528,
+      "loss": 4.2552,
+      "step": 166500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0003951995244067772,
+      "loss": 4.2369,
+      "step": 166600
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00039519662444810163,
+      "loss": 4.2428,
+      "step": 166700
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0003951937244894261,
+      "loss": 4.2376,
+      "step": 166800
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0003951908245307505,
+      "loss": 4.2468,
+      "step": 166900
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00039518792457207485,
+      "loss": 4.2556,
+      "step": 167000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00039518502461339926,
+      "loss": 4.2601,
+      "step": 167100
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00039518212465472366,
+      "loss": 4.2398,
+      "step": 167200
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00039517922469604807,
+      "loss": 4.2503,
+      "step": 167300
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0003951763247373725,
+      "loss": 4.252,
+      "step": 167400
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00039517342477869694,
+      "loss": 4.2462,
+      "step": 167500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00039517052482002134,
+      "loss": 4.2549,
+      "step": 167600
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00039516762486134575,
+      "loss": 4.2526,
+      "step": 167700
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00039516472490267016,
+      "loss": 4.2472,
+      "step": 167800
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00039516182494399456,
+      "loss": 4.251,
+      "step": 167900
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0003951589539849057,
+      "loss": 4.2491,
+      "step": 168000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0003951560540262301,
+      "loss": 4.2455,
+      "step": 168100
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00039515315406755453,
+      "loss": 4.2562,
+      "step": 168200
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00039515025410887894,
+      "loss": 4.2486,
+      "step": 168300
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00039514735415020335,
+      "loss": 4.2489,
+      "step": 168400
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00039514445419152775,
+      "loss": 4.2482,
+      "step": 168500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0003951415542328522,
+      "loss": 4.2502,
+      "step": 168600
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0003951386542741766,
+      "loss": 4.2531,
+      "step": 168700
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00039513575431550103,
+      "loss": 4.2576,
+      "step": 168800
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00039513285435682544,
+      "loss": 4.2488,
+      "step": 168900
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00039512995439814984,
+      "loss": 4.2498,
+      "step": 169000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00039512705443947425,
+      "loss": 4.2599,
+      "step": 169100
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00039512415448079866,
+      "loss": 4.2467,
+      "step": 169200
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0003951212545221231,
+      "loss": 4.2617,
+      "step": 169300
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0003951183545634475,
+      "loss": 4.252,
+      "step": 169400
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00039511545460477193,
+      "loss": 4.2441,
+      "step": 169500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0003951125546460963,
+      "loss": 4.2484,
+      "step": 169600
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0003951096546874207,
+      "loss": 4.2389,
+      "step": 169700
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0003951067547287451,
+      "loss": 4.2483,
+      "step": 169800
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0003951038547700695,
+      "loss": 4.2404,
+      "step": 169900
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0003951009838109807,
+      "loss": 4.2543,
+      "step": 170000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0003950980838523051,
+      "loss": 4.2443,
+      "step": 170100
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00039509518389362953,
+      "loss": 4.2447,
+      "step": 170200
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.000395092283934954,
+      "loss": 4.2425,
+      "step": 170300
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0003950893839762784,
+      "loss": 4.2595,
+      "step": 170400
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00039508648401760275,
+      "loss": 4.2506,
+      "step": 170500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00039508358405892716,
+      "loss": 4.2452,
+      "step": 170600
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00039508068410025156,
+      "loss": 4.2309,
+      "step": 170700
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00039507778414157597,
+      "loss": 4.2503,
+      "step": 170800
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0003950748841829004,
+      "loss": 4.2473,
+      "step": 170900
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00039507198422422484,
+      "loss": 4.249,
+      "step": 171000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00039506908426554924,
+      "loss": 4.2266,
+      "step": 171100
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00039506618430687365,
+      "loss": 4.2446,
+      "step": 171200
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00039506328434819806,
+      "loss": 4.2349,
+      "step": 171300
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00039506038438952247,
+      "loss": 4.2511,
+      "step": 171400
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00039505748443084687,
+      "loss": 4.2544,
+      "step": 171500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0003950545844721713,
+      "loss": 4.2381,
+      "step": 171600
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00039505168451349574,
+      "loss": 4.2493,
+      "step": 171700
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00039504878455482015,
+      "loss": 4.2547,
+      "step": 171800
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00039504588459614455,
+      "loss": 4.2496,
+      "step": 171900
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00039504298463746896,
+      "loss": 4.2503,
+      "step": 172000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00039504008467879337,
+      "loss": 4.2579,
+      "step": 172100
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0003950372137197045,
+      "loss": 4.2527,
+      "step": 172200
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00039503431376102893,
+      "loss": 4.2438,
+      "step": 172300
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00039503141380235334,
+      "loss": 4.2481,
+      "step": 172400
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00039502851384367774,
+      "loss": 4.2443,
+      "step": 172500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00039502561388500215,
+      "loss": 4.2456,
+      "step": 172600
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00039502271392632656,
+      "loss": 4.243,
+      "step": 172700
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.000395019813967651,
+      "loss": 4.2417,
+      "step": 172800
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0003950169140089754,
+      "loss": 4.2558,
+      "step": 172900
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00039501401405029983,
+      "loss": 4.2401,
+      "step": 173000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0003950111140916242,
+      "loss": 4.238,
+      "step": 173100
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0003950082141329486,
+      "loss": 4.2407,
+      "step": 173200
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.000395005314174273,
+      "loss": 4.2553,
+      "step": 173300
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0003950024142155974,
+      "loss": 4.2581,
+      "step": 173400
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00039499951425692187,
+      "loss": 4.2497,
+      "step": 173500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0003949966142982463,
+      "loss": 4.2464,
+      "step": 173600
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0003949937143395707,
+      "loss": 4.2521,
+      "step": 173700
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0003949908143808951,
+      "loss": 4.2417,
+      "step": 173800
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0003949879144222195,
+      "loss": 4.2551,
+      "step": 173900
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0003949850144635439,
+      "loss": 4.2321,
+      "step": 174000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0003949821145048683,
+      "loss": 4.2372,
+      "step": 174100
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00039497924354577946,
+      "loss": 4.2426,
+      "step": 174200
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00039497634358710387,
+      "loss": 4.2481,
+      "step": 174300
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0003949734436284283,
+      "loss": 4.2485,
+      "step": 174400
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00039497054366975274,
+      "loss": 4.2411,
+      "step": 174500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00039496764371107715,
+      "loss": 4.2329,
+      "step": 174600
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00039496474375240155,
+      "loss": 4.2445,
+      "step": 174700
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00039496184379372596,
+      "loss": 4.2483,
+      "step": 174800
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00039495894383505037,
+      "loss": 4.2478,
+      "step": 174900
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0003949560438763748,
+      "loss": 4.2374,
+      "step": 175000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0003949531439176992,
+      "loss": 4.2471,
+      "step": 175100
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00039495024395902364,
+      "loss": 4.243,
+      "step": 175200
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00039494734400034805,
+      "loss": 4.2447,
+      "step": 175300
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00039494444404167246,
+      "loss": 4.2432,
+      "step": 175400
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00039494154408299686,
+      "loss": 4.2437,
+      "step": 175500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00039493864412432127,
+      "loss": 4.237,
+      "step": 175600
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0003949357441656457,
+      "loss": 4.2475,
+      "step": 175700
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00039493284420697003,
+      "loss": 4.2503,
+      "step": 175800
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00039492994424829444,
+      "loss": 4.2545,
+      "step": 175900
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0003949270442896189,
+      "loss": 4.2475,
+      "step": 176000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0003949241443309433,
+      "loss": 4.2422,
+      "step": 176100
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0003949212443722677,
+      "loss": 4.2499,
+      "step": 176200
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0003949183444135921,
+      "loss": 4.2583,
+      "step": 176300
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0003949154444549165,
+      "loss": 4.2398,
+      "step": 176400
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00039491254449624093,
+      "loss": 4.2461,
+      "step": 176500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0003949096735371521,
+      "loss": 4.2565,
+      "step": 176600
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0003949067735784765,
+      "loss": 4.2535,
+      "step": 176700
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0003949038736198009,
+      "loss": 4.247,
+      "step": 176800
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0003949009736611253,
+      "loss": 4.2469,
+      "step": 176900
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00039489807370244977,
+      "loss": 4.2396,
+      "step": 177000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0003948951737437742,
+      "loss": 4.2375,
+      "step": 177100
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0003948922737850986,
+      "loss": 4.2518,
+      "step": 177200
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.000394889373826423,
+      "loss": 4.2501,
+      "step": 177300
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0003948864738677474,
+      "loss": 4.2533,
+      "step": 177400
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0003948835739090718,
+      "loss": 4.2508,
+      "step": 177500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0003948806739503962,
+      "loss": 4.2594,
+      "step": 177600
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00039487777399172067,
+      "loss": 4.2615,
+      "step": 177700
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0003948748740330451,
+      "loss": 4.2342,
+      "step": 177800
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0003948719740743695,
+      "loss": 4.2494,
+      "step": 177900
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0003948690741156939,
+      "loss": 4.2422,
+      "step": 178000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0003948661741570183,
+      "loss": 4.2537,
+      "step": 178100
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0003948632741983427,
+      "loss": 4.252,
+      "step": 178200
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0003948603742396671,
+      "loss": 4.2387,
+      "step": 178300
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0003948574742809915,
+      "loss": 4.2387,
+      "step": 178400
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0003948545743223159,
+      "loss": 4.2423,
+      "step": 178500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00039485167436364033,
+      "loss": 4.2516,
+      "step": 178600
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00039484877440496474,
+      "loss": 4.2478,
+      "step": 178700
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00039484590344587595,
+      "loss": 4.2387,
+      "step": 178800
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00039484300348720036,
+      "loss": 4.2497,
+      "step": 178900
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00039484010352852476,
+      "loss": 4.2537,
+      "step": 179000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00039483720356984917,
+      "loss": 4.2445,
+      "step": 179100
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0003948343036111736,
+      "loss": 4.2379,
+      "step": 179200
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00039483140365249793,
+      "loss": 4.2404,
+      "step": 179300
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00039482850369382234,
+      "loss": 4.2511,
+      "step": 179400
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0003948256037351468,
+      "loss": 4.241,
+      "step": 179500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0003948227037764712,
+      "loss": 4.2438,
+      "step": 179600
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0003948198038177956,
+      "loss": 4.2575,
+      "step": 179700
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00039481690385912,
+      "loss": 4.2564,
+      "step": 179800
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0003948140039004444,
+      "loss": 4.2578,
+      "step": 179900
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00039481110394176883,
+      "loss": 4.2502,
+      "step": 180000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00039480820398309324,
+      "loss": 4.2363,
+      "step": 180100
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0003948053040244177,
+      "loss": 4.2503,
+      "step": 180200
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0003948024040657421,
+      "loss": 4.2465,
+      "step": 180300
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0003947995041070665,
+      "loss": 4.232,
+      "step": 180400
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0003947966041483909,
+      "loss": 4.2513,
+      "step": 180500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00039479370418971533,
+      "loss": 4.2469,
+      "step": 180600
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00039479080423103973,
+      "loss": 4.2332,
+      "step": 180700
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0003947879332719509,
+      "loss": 4.2381,
+      "step": 180800
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0003947850333132753,
+      "loss": 4.2465,
+      "step": 180900
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0003947821333545997,
+      "loss": 4.2518,
+      "step": 181000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0003947792333959241,
+      "loss": 4.2401,
+      "step": 181100
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00039477633343724857,
+      "loss": 4.2408,
+      "step": 181200
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.000394773433478573,
+      "loss": 4.2419,
+      "step": 181300
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0003947705335198974,
+      "loss": 4.2499,
+      "step": 181400
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0003947676335612218,
+      "loss": 4.2542,
+      "step": 181500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0003947647336025462,
+      "loss": 4.2349,
+      "step": 181600
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0003947618336438706,
+      "loss": 4.2343,
+      "step": 181700
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.000394758933685195,
+      "loss": 4.2483,
+      "step": 181800
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0003947560337265194,
+      "loss": 4.245,
+      "step": 181900
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00039475313376784383,
+      "loss": 4.2411,
+      "step": 182000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00039475023380916823,
+      "loss": 4.2443,
+      "step": 182100
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00039474733385049264,
+      "loss": 4.2458,
+      "step": 182200
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00039474443389181705,
+      "loss": 4.2314,
+      "step": 182300
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00039474153393314145,
+      "loss": 4.2527,
+      "step": 182400
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00039473863397446586,
+      "loss": 4.2434,
+      "step": 182500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0003947357340157903,
+      "loss": 4.2455,
+      "step": 182600
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00039473283405711473,
+      "loss": 4.2465,
+      "step": 182700
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00039472993409843914,
+      "loss": 4.235,
+      "step": 182800
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00039472706313935024,
+      "loss": 4.2439,
+      "step": 182900
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0003947241631806747,
+      "loss": 4.2463,
+      "step": 183000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0003947212632219991,
+      "loss": 4.2397,
+      "step": 183100
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0003947183632633235,
+      "loss": 4.2323,
+      "step": 183200
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0003947154633046479,
+      "loss": 4.2362,
+      "step": 183300
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0003947125633459723,
+      "loss": 4.2381,
+      "step": 183400
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00039470966338729673,
+      "loss": 4.2362,
+      "step": 183500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00039470676342862114,
+      "loss": 4.2463,
+      "step": 183600
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0003947038634699456,
+      "loss": 4.2469,
+      "step": 183700
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00039470096351127,
+      "loss": 4.2345,
+      "step": 183800
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0003946980635525944,
+      "loss": 4.2353,
+      "step": 183900
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0003946951635939188,
+      "loss": 4.2511,
+      "step": 184000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00039469226363524323,
+      "loss": 4.2479,
+      "step": 184100
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00039468936367656764,
+      "loss": 4.2371,
+      "step": 184200
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00039468646371789204,
+      "loss": 4.2472,
+      "step": 184300
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00039468356375921645,
+      "loss": 4.2206,
+      "step": 184400
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00039468066380054086,
+      "loss": 4.246,
+      "step": 184500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00039467776384186526,
+      "loss": 4.2367,
+      "step": 184600
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00039467486388318967,
+      "loss": 4.2343,
+      "step": 184700
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0003946719639245141,
+      "loss": 4.2406,
+      "step": 184800
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0003946690639658385,
+      "loss": 4.2375,
+      "step": 184900
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0003946661640071629,
+      "loss": 4.238,
+      "step": 185000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00039466326404848735,
+      "loss": 4.2403,
+      "step": 185100
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0003946603930893985,
+      "loss": 4.2545,
+      "step": 185200
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0003946574931307229,
+      "loss": 4.2324,
+      "step": 185300
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0003946545931720473,
+      "loss": 4.2307,
+      "step": 185400
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00039465169321337173,
+      "loss": 4.2425,
+      "step": 185500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00039464879325469614,
+      "loss": 4.2389,
+      "step": 185600
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00039464589329602054,
+      "loss": 4.2422,
+      "step": 185700
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00039464299333734495,
+      "loss": 4.2504,
+      "step": 185800
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00039464009337866936,
+      "loss": 4.2449,
+      "step": 185900
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00039463719341999376,
+      "loss": 4.2461,
+      "step": 186000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0003946342934613182,
+      "loss": 4.2384,
+      "step": 186100
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00039463139350264263,
+      "loss": 4.2495,
+      "step": 186200
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00039462849354396704,
+      "loss": 4.2372,
+      "step": 186300
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00039462559358529144,
+      "loss": 4.2443,
+      "step": 186400
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00039462269362661585,
+      "loss": 4.2348,
+      "step": 186500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00039461979366794026,
+      "loss": 4.2503,
+      "step": 186600
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00039461689370926467,
+      "loss": 4.2285,
+      "step": 186700
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00039461399375058907,
+      "loss": 4.2288,
+      "step": 186800
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0003946110937919135,
+      "loss": 4.2437,
+      "step": 186900
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0003946081938332379,
+      "loss": 4.2417,
+      "step": 187000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0003946052938745623,
+      "loss": 4.2582,
+      "step": 187100
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0003946023939158867,
+      "loss": 4.2373,
+      "step": 187200
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0003945994939572111,
+      "loss": 4.2327,
+      "step": 187300
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0003945966229981223,
+      "loss": 4.2433,
+      "step": 187400
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0003945937230394467,
+      "loss": 4.2409,
+      "step": 187500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00039459082308077113,
+      "loss": 4.2376,
+      "step": 187600
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00039458792312209554,
+      "loss": 4.2481,
+      "step": 187700
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00039458502316341994,
+      "loss": 4.2359,
+      "step": 187800
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00039458212320474435,
+      "loss": 4.2465,
+      "step": 187900
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00039457922324606876,
+      "loss": 4.2432,
+      "step": 188000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00039457632328739316,
+      "loss": 4.2548,
+      "step": 188100
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00039457342332871757,
+      "loss": 4.2376,
+      "step": 188200
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.000394570523370042,
+      "loss": 4.2375,
+      "step": 188300
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0003945676234113664,
+      "loss": 4.2419,
+      "step": 188400
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0003945647234526908,
+      "loss": 4.2399,
+      "step": 188500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00039456182349401525,
+      "loss": 4.2453,
+      "step": 188600
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00039455892353533966,
+      "loss": 4.245,
+      "step": 188700
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00039455602357666407,
+      "loss": 4.251,
+      "step": 188800
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0003945531236179885,
+      "loss": 4.2343,
+      "step": 188900
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0003945502236593129,
+      "loss": 4.251,
+      "step": 189000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0003945473237006373,
+      "loss": 4.248,
+      "step": 189100
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0003945444237419617,
+      "loss": 4.2409,
+      "step": 189200
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0003945415237832861,
+      "loss": 4.2433,
+      "step": 189300
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0003945386238246105,
+      "loss": 4.2262,
+      "step": 189400
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0003945357238659349,
+      "loss": 4.2595,
+      "step": 189500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0003945328239072593,
+      "loss": 4.2491,
+      "step": 189600
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00039452995294817053,
+      "loss": 4.2435,
+      "step": 189700
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00039452705298949494,
+      "loss": 4.251,
+      "step": 189800
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00039452415303081935,
+      "loss": 4.2327,
+      "step": 189900
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00039452125307214375,
+      "loss": 4.2446,
+      "step": 190000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00039451835311346816,
+      "loss": 4.2381,
+      "step": 190100
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00039451545315479257,
+      "loss": 4.2438,
+      "step": 190200
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.000394512553196117,
+      "loss": 4.2417,
+      "step": 190300
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0003945096532374414,
+      "loss": 4.2311,
+      "step": 190400
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0003945067532787658,
+      "loss": 4.2484,
+      "step": 190500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0003945038533200902,
+      "loss": 4.248,
+      "step": 190600
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0003945009533614146,
+      "loss": 4.2416,
+      "step": 190700
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.000394498053402739,
+      "loss": 4.2397,
+      "step": 190800
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0003944951534440634,
+      "loss": 4.2414,
+      "step": 190900
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0003944922534853878,
+      "loss": 4.2362,
+      "step": 191000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0003944893535267123,
+      "loss": 4.2453,
+      "step": 191100
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0003944864535680367,
+      "loss": 4.2485,
+      "step": 191200
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0003944835536093611,
+      "loss": 4.2283,
+      "step": 191300
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0003944806536506855,
+      "loss": 4.234,
+      "step": 191400
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0003944777536920099,
+      "loss": 4.2374,
+      "step": 191500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0003944748537333343,
+      "loss": 4.2372,
+      "step": 191600
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0003944719537746587,
+      "loss": 4.2455,
+      "step": 191700
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0003944690538159832,
+      "loss": 4.2393,
+      "step": 191800
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0003944661828568943,
+      "loss": 4.2455,
+      "step": 191900
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0003944632828982187,
+      "loss": 4.2444,
+      "step": 192000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00039446038293954315,
+      "loss": 4.2436,
+      "step": 192100
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00039445748298086756,
+      "loss": 4.2468,
+      "step": 192200
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00039445458302219197,
+      "loss": 4.2339,
+      "step": 192300
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0003944516830635164,
+      "loss": 4.2444,
+      "step": 192400
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0003944487831048408,
+      "loss": 4.2364,
+      "step": 192500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0003944458831461652,
+      "loss": 4.2414,
+      "step": 192600
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0003944429831874896,
+      "loss": 4.2432,
+      "step": 192700
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.000394440083228814,
+      "loss": 4.2395,
+      "step": 192800
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0003944371832701384,
+      "loss": 4.2493,
+      "step": 192900
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0003944342833114628,
+      "loss": 4.2294,
+      "step": 193000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0003944313833527872,
+      "loss": 4.2443,
+      "step": 193100
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00039442848339411163,
+      "loss": 4.2432,
+      "step": 193200
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00039442558343543604,
+      "loss": 4.2491,
+      "step": 193300
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00039442268347676044,
+      "loss": 4.2449,
+      "step": 193400
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0003944197835180849,
+      "loss": 4.2501,
+      "step": 193500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0003944168835594093,
+      "loss": 4.246,
+      "step": 193600
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0003944139836007337,
+      "loss": 4.2438,
+      "step": 193700
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0003944110836420581,
+      "loss": 4.2383,
+      "step": 193800
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00039440818368338253,
+      "loss": 4.2495,
+      "step": 193900
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00039440528372470694,
+      "loss": 4.2485,
+      "step": 194000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0003944024127656181,
+      "loss": 4.2516,
+      "step": 194100
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0003943995128069425,
+      "loss": 4.247,
+      "step": 194200
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0003943966128482669,
+      "loss": 4.2355,
+      "step": 194300
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0003943937128895913,
+      "loss": 4.2426,
+      "step": 194400
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0003943908129309157,
+      "loss": 4.2326,
+      "step": 194500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0003943879129722402,
+      "loss": 4.2439,
+      "step": 194600
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0003943850130135646,
+      "loss": 4.2482,
+      "step": 194700
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.000394382113054889,
+      "loss": 4.2383,
+      "step": 194800
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0003943792130962134,
+      "loss": 4.237,
+      "step": 194900
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0003943763131375378,
+      "loss": 4.2469,
+      "step": 195000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00039437344217844897,
+      "loss": 4.2335,
+      "step": 195100
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0003943705422197734,
+      "loss": 4.2309,
+      "step": 195200
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0003943676422610978,
+      "loss": 4.2539,
+      "step": 195300
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0003943647423024222,
+      "loss": 4.2412,
+      "step": 195400
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0003943618423437466,
+      "loss": 4.2427,
+      "step": 195500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00039435894238507106,
+      "loss": 4.2354,
+      "step": 195600
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00039435604242639546,
+      "loss": 4.2345,
+      "step": 195700
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00039435314246771987,
+      "loss": 4.2424,
+      "step": 195800
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0003943502425090443,
+      "loss": 4.222,
+      "step": 195900
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0003943473425503687,
+      "loss": 4.247,
+      "step": 196000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0003943444425916931,
+      "loss": 4.2434,
+      "step": 196100
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0003943415426330175,
+      "loss": 4.2339,
+      "step": 196200
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0003943386426743419,
+      "loss": 4.2446,
+      "step": 196300
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0003943357427156663,
+      "loss": 4.2428,
+      "step": 196400
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0003943328427569907,
+      "loss": 4.2415,
+      "step": 196500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0003943299427983151,
+      "loss": 4.2372,
+      "step": 196600
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00039432704283963953,
+      "loss": 4.2347,
+      "step": 196700
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00039432414288096394,
+      "loss": 4.2416,
+      "step": 196800
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00039432124292228834,
+      "loss": 4.234,
+      "step": 196900
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0003943183429636128,
+      "loss": 4.2457,
+      "step": 197000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0003943154430049372,
+      "loss": 4.2304,
+      "step": 197100
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0003943125430462616,
+      "loss": 4.2424,
+      "step": 197200
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.000394309643087586,
+      "loss": 4.2469,
+      "step": 197300
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00039430674312891043,
+      "loss": 4.246,
+      "step": 197400
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00039430384317023484,
+      "loss": 4.2446,
+      "step": 197500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00039430094321155925,
+      "loss": 4.2214,
+      "step": 197600
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0003942980432528837,
+      "loss": 4.2304,
+      "step": 197700
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0003942951432942081,
+      "loss": 4.2399,
+      "step": 197800
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0003942922433355325,
+      "loss": 4.2276,
+      "step": 197900
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0003942893433768569,
+      "loss": 4.2269,
+      "step": 198000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0003942864434181813,
+      "loss": 4.2476,
+      "step": 198100
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0003942835434595057,
+      "loss": 4.2372,
+      "step": 198200
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0003942806435008301,
+      "loss": 4.2312,
+      "step": 198300
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00039427774354215456,
+      "loss": 4.2332,
+      "step": 198400
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00039427484358347896,
+      "loss": 4.2284,
+      "step": 198500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0003942719726243901,
+      "loss": 4.2366,
+      "step": 198600
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0003942690726657145,
+      "loss": 4.2303,
+      "step": 198700
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.000394266172707039,
+      "loss": 4.2492,
+      "step": 198800
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00039426327274836334,
+      "loss": 4.2397,
+      "step": 198900
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00039426037278968775,
+      "loss": 4.2377,
+      "step": 199000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00039425747283101215,
+      "loss": 4.2412,
+      "step": 199100
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00039425457287233656,
+      "loss": 4.2473,
+      "step": 199200
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00039425167291366097,
+      "loss": 4.2414,
+      "step": 199300
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0003942487729549854,
+      "loss": 4.2353,
+      "step": 199400
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00039424587299630984,
+      "loss": 4.2498,
+      "step": 199500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00039424297303763424,
+      "loss": 4.2437,
+      "step": 199600
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00039424007307895865,
+      "loss": 4.2436,
+      "step": 199700
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00039423717312028306,
+      "loss": 4.2223,
+      "step": 199800
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00039423427316160746,
+      "loss": 4.2379,
+      "step": 199900
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00039423137320293187,
+      "loss": 4.241,
+      "step": 200000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0003942284732442563,
+      "loss": 4.2418,
+      "step": 200100
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00039422557328558074,
+      "loss": 4.232,
+      "step": 200200
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00039422267332690514,
+      "loss": 4.2411,
+      "step": 200300
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00039421977336822955,
+      "loss": 4.224,
+      "step": 200400
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00039421687340955396,
+      "loss": 4.2369,
+      "step": 200500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0003942139734508783,
+      "loss": 4.2296,
+      "step": 200600
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0003942110734922027,
+      "loss": 4.2469,
+      "step": 200700
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0003942081735335271,
+      "loss": 4.239,
+      "step": 200800
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0003942052735748516,
+      "loss": 4.2417,
+      "step": 200900
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.000394202373616176,
+      "loss": 4.2231,
+      "step": 201000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0003941994736575004,
+      "loss": 4.2336,
+      "step": 201100
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0003941965736988248,
+      "loss": 4.2381,
+      "step": 201200
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0003941936737401492,
+      "loss": 4.2288,
+      "step": 201300
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0003941907737814736,
+      "loss": 4.2203,
+      "step": 201400
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.000394187873822798,
+      "loss": 4.2356,
+      "step": 201500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0003941849738641225,
+      "loss": 4.2297,
+      "step": 201600
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0003941820739054469,
+      "loss": 4.2499,
+      "step": 201700
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0003941791739467713,
+      "loss": 4.2437,
+      "step": 201800
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0003941762739880957,
+      "loss": 4.244,
+      "step": 201900
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0003941733740294201,
+      "loss": 4.2298,
+      "step": 202000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0003941704740707445,
+      "loss": 4.2339,
+      "step": 202100
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00039416757411206893,
+      "loss": 4.2423,
+      "step": 202200
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00039416467415339334,
+      "loss": 4.2519,
+      "step": 202300
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00039416177419471774,
+      "loss": 4.2359,
+      "step": 202400
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00039415887423604215,
+      "loss": 4.2326,
+      "step": 202500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00039415597427736656,
+      "loss": 4.2396,
+      "step": 202600
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00039415310331827777,
+      "loss": 4.2349,
+      "step": 202700
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0003941502033596022,
+      "loss": 4.2423,
+      "step": 202800
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0003941473034009266,
+      "loss": 4.238,
+      "step": 202900
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.000394144403442251,
+      "loss": 4.229,
+      "step": 203000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0003941415034835754,
+      "loss": 4.2441,
+      "step": 203100
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0003941386035248998,
+      "loss": 4.2345,
+      "step": 203200
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00039413570356622415,
+      "loss": 4.2351,
+      "step": 203300
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0003941328036075486,
+      "loss": 4.2282,
+      "step": 203400
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.000394129903648873,
+      "loss": 4.2345,
+      "step": 203500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00039412700369019743,
+      "loss": 4.2342,
+      "step": 203600
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00039412410373152184,
+      "loss": 4.2292,
+      "step": 203700
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00039412120377284624,
+      "loss": 4.236,
+      "step": 203800
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00039411830381417065,
+      "loss": 4.2461,
+      "step": 203900
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00039411540385549506,
+      "loss": 4.2365,
+      "step": 204000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0003941125038968195,
+      "loss": 4.2419,
+      "step": 204100
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0003941096039381439,
+      "loss": 4.2321,
+      "step": 204200
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00039410670397946833,
+      "loss": 4.2413,
+      "step": 204300
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00039410380402079274,
+      "loss": 4.2307,
+      "step": 204400
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00039410090406211714,
+      "loss": 4.239,
+      "step": 204500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00039409800410344155,
+      "loss": 4.2339,
+      "step": 204600
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0003940951331443527,
+      "loss": 4.2236,
+      "step": 204700
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0003940922331856771,
+      "loss": 4.2333,
+      "step": 204800
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0003940893332270015,
+      "loss": 4.2352,
+      "step": 204900
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00039408643326832593,
+      "loss": 4.2356,
+      "step": 205000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0003940835333096504,
+      "loss": 4.2215,
+      "step": 205100
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0003940806333509748,
+      "loss": 4.2324,
+      "step": 205200
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0003940777333922992,
+      "loss": 4.224,
+      "step": 205300
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0003940748334336236,
+      "loss": 4.2293,
+      "step": 205400
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.000394071933474948,
+      "loss": 4.2332,
+      "step": 205500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0003940690335162724,
+      "loss": 4.231,
+      "step": 205600
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00039406613355759683,
+      "loss": 4.243,
+      "step": 205700
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00039406323359892124,
+      "loss": 4.2456,
+      "step": 205800
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00039406033364024564,
+      "loss": 4.2336,
+      "step": 205900
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00039405743368157005,
+      "loss": 4.2268,
+      "step": 206000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00039405453372289446,
+      "loss": 4.2345,
+      "step": 206100
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00039405163376421886,
+      "loss": 4.245,
+      "step": 206200
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00039404873380554327,
+      "loss": 4.231,
+      "step": 206300
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0003940458338468677,
+      "loss": 4.2425,
+      "step": 206400
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00039404293388819214,
+      "loss": 4.2317,
+      "step": 206500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00039404003392951655,
+      "loss": 4.2469,
+      "step": 206600
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00039403713397084095,
+      "loss": 4.2289,
+      "step": 206700
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00039403426301175206,
+      "loss": 4.2345,
+      "step": 206800
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0003940313630530765,
+      "loss": 4.2437,
+      "step": 206900
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0003940284630944009,
+      "loss": 4.2487,
+      "step": 207000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00039402556313572533,
+      "loss": 4.2278,
+      "step": 207100
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00039402266317704974,
+      "loss": 4.2416,
+      "step": 207200
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00039401976321837414,
+      "loss": 4.2374,
+      "step": 207300
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00039401686325969855,
+      "loss": 4.2125,
+      "step": 207400
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00039401396330102296,
+      "loss": 4.2443,
+      "step": 207500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0003940110633423474,
+      "loss": 4.2315,
+      "step": 207600
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0003940081633836718,
+      "loss": 4.2317,
+      "step": 207700
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00039400526342499623,
+      "loss": 4.247,
+      "step": 207800
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0003940023924659074,
+      "loss": 4.2553,
+      "step": 207900
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0003939994925072318,
+      "loss": 4.2419,
+      "step": 208000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0003939965925485562,
+      "loss": 4.2336,
+      "step": 208100
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0003939936925898806,
+      "loss": 4.2333,
+      "step": 208200
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.000393990792631205,
+      "loss": 4.2327,
+      "step": 208300
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0003939878926725294,
+      "loss": 4.2234,
+      "step": 208400
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00039398499271385383,
+      "loss": 4.2528,
+      "step": 208500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0003939820927551783,
+      "loss": 4.2527,
+      "step": 208600
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0003939791927965027,
+      "loss": 4.2363,
+      "step": 208700
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0003939762928378271,
+      "loss": 4.2296,
+      "step": 208800
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0003939733928791515,
+      "loss": 4.2398,
+      "step": 208900
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0003939704929204759,
+      "loss": 4.2339,
+      "step": 209000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0003939675929618003,
+      "loss": 4.2326,
+      "step": 209100
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00039396469300312473,
+      "loss": 4.2301,
+      "step": 209200
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00039396179304444914,
+      "loss": 4.2418,
+      "step": 209300
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00039395889308577355,
+      "loss": 4.2343,
+      "step": 209400
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00039395599312709795,
+      "loss": 4.2402,
+      "step": 209500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00039395309316842236,
+      "loss": 4.235,
+      "step": 209600
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00039395019320974677,
+      "loss": 4.2345,
+      "step": 209700
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0003939472932510712,
+      "loss": 4.2195,
+      "step": 209800
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0003939443932923956,
+      "loss": 4.2402,
+      "step": 209900
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00039394149333372004,
+      "loss": 4.2376,
+      "step": 210000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00039393859337504445,
+      "loss": 4.2339,
+      "step": 210100
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00039393569341636885,
+      "loss": 4.2332,
+      "step": 210200
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00039393279345769326,
+      "loss": 4.2462,
+      "step": 210300
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00039392989349901767,
+      "loss": 4.2319,
+      "step": 210400
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0003939269935403421,
+      "loss": 4.2314,
+      "step": 210500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0003939240935816665,
+      "loss": 4.2521,
+      "step": 210600
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0003939211936229909,
+      "loss": 4.2572,
+      "step": 210700
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0003939182936643153,
+      "loss": 4.2359,
+      "step": 210800
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0003939153937056397,
+      "loss": 4.2251,
+      "step": 210900
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0003939124937469641,
+      "loss": 4.224,
+      "step": 211000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0003939095937882885,
+      "loss": 4.2269,
+      "step": 211100
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0003939066938296129,
+      "loss": 4.2426,
+      "step": 211200
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00039390379387093733,
+      "loss": 4.2305,
+      "step": 211300
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00039390089391226174,
+      "loss": 4.2288,
+      "step": 211400
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0003938979939535862,
+      "loss": 4.2415,
+      "step": 211500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0003938950939949106,
+      "loss": 4.2227,
+      "step": 211600
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.000393892194036235,
+      "loss": 4.2342,
+      "step": 211700
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0003938892940775594,
+      "loss": 4.2318,
+      "step": 211800
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0003938863941188838,
+      "loss": 4.2499,
+      "step": 211900
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00039388349416020823,
+      "loss": 4.2326,
+      "step": 212000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00039388059420153264,
+      "loss": 4.2411,
+      "step": 212100
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0003938776942428571,
+      "loss": 4.2339,
+      "step": 212200
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0003938747942841815,
+      "loss": 4.2337,
+      "step": 212300
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0003938719233250926,
+      "loss": 4.2333,
+      "step": 212400
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00039386902336641707,
+      "loss": 4.2422,
+      "step": 212500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0003938661234077415,
+      "loss": 4.233,
+      "step": 212600
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0003938632234490659,
+      "loss": 4.234,
+      "step": 212700
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0003938603234903903,
+      "loss": 4.2346,
+      "step": 212800
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0003938574235317147,
+      "loss": 4.2429,
+      "step": 212900
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0003938545235730391,
+      "loss": 4.237,
+      "step": 213000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0003938516236143635,
+      "loss": 4.2218,
+      "step": 213100
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00039384872365568797,
+      "loss": 4.2279,
+      "step": 213200
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0003938458236970123,
+      "loss": 4.2316,
+      "step": 213300
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00039384292373833673,
+      "loss": 4.2474,
+      "step": 213400
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00039384002377966114,
+      "loss": 4.2375,
+      "step": 213500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00039383712382098555,
+      "loss": 4.237,
+      "step": 213600
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00039383422386230995,
+      "loss": 4.2349,
+      "step": 213700
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00039383132390363436,
+      "loss": 4.2437,
+      "step": 213800
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0003938284239449588,
+      "loss": 4.2312,
+      "step": 213900
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00039382552398628323,
+      "loss": 4.2361,
+      "step": 214000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00039382262402760763,
+      "loss": 4.2213,
+      "step": 214100
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00039381972406893204,
+      "loss": 4.2217,
+      "step": 214200
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00039381682411025645,
+      "loss": 4.2384,
+      "step": 214300
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0003938139531511676,
+      "loss": 4.2504,
+      "step": 214400
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.000393811053192492,
+      "loss": 4.231,
+      "step": 214500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0003938081532338164,
+      "loss": 4.233,
+      "step": 214600
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0003938052532751408,
+      "loss": 4.2442,
+      "step": 214700
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00039380235331646523,
+      "loss": 4.2481,
+      "step": 214800
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00039379945335778964,
+      "loss": 4.2263,
+      "step": 214900
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0003937965533991141,
+      "loss": 4.2362,
+      "step": 215000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0003937936534404385,
+      "loss": 4.2322,
+      "step": 215100
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0003937907534817629,
+      "loss": 4.2345,
+      "step": 215200
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0003937878535230873,
+      "loss": 4.2298,
+      "step": 215300
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0003937849535644117,
+      "loss": 4.2262,
+      "step": 215400
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00039378205360573613,
+      "loss": 4.2196,
+      "step": 215500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00039377915364706054,
+      "loss": 4.234,
+      "step": 215600
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.000393776253688385,
+      "loss": 4.2379,
+      "step": 215700
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0003937733537297094,
+      "loss": 4.2434,
+      "step": 215800
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00039377045377103376,
+      "loss": 4.2361,
+      "step": 215900
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00039376755381235817,
+      "loss": 4.2286,
+      "step": 216000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0003937646538536826,
+      "loss": 4.2322,
+      "step": 216100
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.000393761753895007,
+      "loss": 4.2278,
+      "step": 216200
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0003937588539363314,
+      "loss": 4.2328,
+      "step": 216300
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00039375595397765585,
+      "loss": 4.2343,
+      "step": 216400
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00039375305401898026,
+      "loss": 4.2419,
+      "step": 216500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0003937501830598914,
+      "loss": 4.2303,
+      "step": 216600
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0003937472831012159,
+      "loss": 4.2266,
+      "step": 216700
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0003937443831425402,
+      "loss": 4.2287,
+      "step": 216800
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00039374148318386463,
+      "loss": 4.2296,
+      "step": 216900
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00039373858322518904,
+      "loss": 4.2313,
+      "step": 217000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00039373568326651345,
+      "loss": 4.2308,
+      "step": 217100
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00039373278330783785,
+      "loss": 4.2106,
+      "step": 217200
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00039372988334916226,
+      "loss": 4.2291,
+      "step": 217300
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0003937269833904867,
+      "loss": 4.2263,
+      "step": 217400
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00039372408343181113,
+      "loss": 4.2415,
+      "step": 217500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00039372118347313554,
+      "loss": 4.2271,
+      "step": 217600
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00039371828351445994,
+      "loss": 4.2382,
+      "step": 217700
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00039371538355578435,
+      "loss": 4.2244,
+      "step": 217800
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00039371248359710876,
+      "loss": 4.245,
+      "step": 217900
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00039370958363843316,
+      "loss": 4.2287,
+      "step": 218000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0003937066836797576,
+      "loss": 4.2371,
+      "step": 218100
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00039370378372108203,
+      "loss": 4.2309,
+      "step": 218200
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00039370088376240644,
+      "loss": 4.2295,
+      "step": 218300
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00039369798380373084,
+      "loss": 4.2329,
+      "step": 218400
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0003936950838450552,
+      "loss": 4.2414,
+      "step": 218500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0003936921838863796,
+      "loss": 4.2279,
+      "step": 218600
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0003936893129272908,
+      "loss": 4.2257,
+      "step": 218700
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0003936864129686152,
+      "loss": 4.2244,
+      "step": 218800
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00039368351300993963,
+      "loss": 4.2432,
+      "step": 218900
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00039368061305126404,
+      "loss": 4.2399,
+      "step": 219000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00039367771309258844,
+      "loss": 4.2405,
+      "step": 219100
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0003936748131339129,
+      "loss": 4.2273,
+      "step": 219200
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0003936719131752373,
+      "loss": 4.2454,
+      "step": 219300
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00039366901321656166,
+      "loss": 4.2263,
+      "step": 219400
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00039366611325788607,
+      "loss": 4.2314,
+      "step": 219500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0003936632132992105,
+      "loss": 4.2361,
+      "step": 219600
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0003936603133405349,
+      "loss": 4.233,
+      "step": 219700
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0003936574133818593,
+      "loss": 4.2425,
+      "step": 219800
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00039365451342318375,
+      "loss": 4.2325,
+      "step": 219900
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00039365161346450816,
+      "loss": 4.237,
+      "step": 220000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00039364871350583256,
+      "loss": 4.2324,
+      "step": 220100
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00039364581354715697,
+      "loss": 4.2249,
+      "step": 220200
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0003936429135884814,
+      "loss": 4.2393,
+      "step": 220300
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0003936400136298058,
+      "loss": 4.2343,
+      "step": 220400
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0003936371136711302,
+      "loss": 4.2241,
+      "step": 220500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00039363421371245465,
+      "loss": 4.2289,
+      "step": 220600
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00039363131375377906,
+      "loss": 4.2338,
+      "step": 220700
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00039362844279469016,
+      "loss": 4.2318,
+      "step": 220800
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0003936255428360146,
+      "loss": 4.2356,
+      "step": 220900
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00039362264287733903,
+      "loss": 4.2347,
+      "step": 221000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00039361974291866344,
+      "loss": 4.235,
+      "step": 221100
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00039361684295998784,
+      "loss": 4.2321,
+      "step": 221200
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00039361394300131225,
+      "loss": 4.2346,
+      "step": 221300
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00039361104304263666,
+      "loss": 4.2392,
+      "step": 221400
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00039360814308396106,
+      "loss": 4.2293,
+      "step": 221500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0003936052431252855,
+      "loss": 4.2344,
+      "step": 221600
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00039360234316660993,
+      "loss": 4.2169,
+      "step": 221700
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00039359944320793434,
+      "loss": 4.2392,
+      "step": 221800
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00039359654324925875,
+      "loss": 4.2456,
+      "step": 221900
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0003935936432905831,
+      "loss": 4.2333,
+      "step": 222000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0003935907433319075,
+      "loss": 4.2249,
+      "step": 222100
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0003935878433732319,
+      "loss": 4.2384,
+      "step": 222200
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0003935849434145563,
+      "loss": 4.2257,
+      "step": 222300
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0003935820434558808,
+      "loss": 4.242,
+      "step": 222400
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0003935791434972052,
+      "loss": 4.2391,
+      "step": 222500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0003935762435385296,
+      "loss": 4.2357,
+      "step": 222600
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.000393573343579854,
+      "loss": 4.2296,
+      "step": 222700
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0003935704436211784,
+      "loss": 4.2468,
+      "step": 222800
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0003935675436625028,
+      "loss": 4.2324,
+      "step": 222900
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0003935646437038272,
+      "loss": 4.2309,
+      "step": 223000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0003935617727447384,
+      "loss": 4.2184,
+      "step": 223100
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0003935588727860628,
+      "loss": 4.2363,
+      "step": 223200
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0003935559728273872,
+      "loss": 4.2352,
+      "step": 223300
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00039355307286871165,
+      "loss": 4.2389,
+      "step": 223400
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00039355017291003606,
+      "loss": 4.2157,
+      "step": 223500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00039354727295136047,
+      "loss": 4.234,
+      "step": 223600
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0003935443729926849,
+      "loss": 4.2363,
+      "step": 223700
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0003935414730340093,
+      "loss": 4.2245,
+      "step": 223800
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0003935385730753337,
+      "loss": 4.2468,
+      "step": 223900
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0003935356731166581,
+      "loss": 4.2331,
+      "step": 224000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00039353277315798255,
+      "loss": 4.2273,
+      "step": 224100
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00039352987319930696,
+      "loss": 4.2436,
+      "step": 224200
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00039352700224021806,
+      "loss": 4.2224,
+      "step": 224300
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0003935241022815425,
+      "loss": 4.2405,
+      "step": 224400
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00039352120232286693,
+      "loss": 4.239,
+      "step": 224500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00039351830236419134,
+      "loss": 4.2395,
+      "step": 224600
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00039351540240551574,
+      "loss": 4.2378,
+      "step": 224700
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00039351250244684015,
+      "loss": 4.2323,
+      "step": 224800
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00039350960248816456,
+      "loss": 4.2291,
+      "step": 224900
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0003935067315290757,
+      "loss": 4.2433,
+      "step": 225000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0003935038315704001,
+      "loss": 4.2343,
+      "step": 225100
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00039350093161172453,
+      "loss": 4.2314,
+      "step": 225200
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00039349803165304894,
+      "loss": 4.2277,
+      "step": 225300
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00039349513169437334,
+      "loss": 4.2319,
+      "step": 225400
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0003934922317356978,
+      "loss": 4.2411,
+      "step": 225500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0003934893317770222,
+      "loss": 4.2248,
+      "step": 225600
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0003934864318183466,
+      "loss": 4.2377,
+      "step": 225700
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.000393483531859671,
+      "loss": 4.221,
+      "step": 225800
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00039348063190099543,
+      "loss": 4.2296,
+      "step": 225900
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00039347773194231984,
+      "loss": 4.2163,
+      "step": 226000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00039347483198364424,
+      "loss": 4.2379,
+      "step": 226100
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0003934719320249687,
+      "loss": 4.2291,
+      "step": 226200
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0003934690320662931,
+      "loss": 4.2275,
+      "step": 226300
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00039346613210761746,
+      "loss": 4.2278,
+      "step": 226400
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00039346323214894187,
+      "loss": 4.239,
+      "step": 226500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0003934603321902663,
+      "loss": 4.2242,
+      "step": 226600
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0003934574322315907,
+      "loss": 4.2281,
+      "step": 226700
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0003934545322729151,
+      "loss": 4.2382,
+      "step": 226800
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00039345163231423955,
+      "loss": 4.2375,
+      "step": 226900
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00039344873235556396,
+      "loss": 4.2289,
+      "step": 227000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00039344583239688837,
+      "loss": 4.2295,
+      "step": 227100
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0003934429324382128,
+      "loss": 4.2306,
+      "step": 227200
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0003934400324795372,
+      "loss": 4.2231,
+      "step": 227300
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0003934371325208616,
+      "loss": 4.2139,
+      "step": 227400
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.000393434232562186,
+      "loss": 4.2382,
+      "step": 227500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00039343133260351046,
+      "loss": 4.2306,
+      "step": 227600
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00039342843264483486,
+      "loss": 4.2278,
+      "step": 227700
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00039342553268615927,
+      "loss": 4.2354,
+      "step": 227800
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0003934226327274837,
+      "loss": 4.2277,
+      "step": 227900
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0003934197327688081,
+      "loss": 4.2312,
+      "step": 228000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00039341683281013244,
+      "loss": 4.2366,
+      "step": 228100
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00039341393285145684,
+      "loss": 4.2361,
+      "step": 228200
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0003934110328927813,
+      "loss": 4.2283,
+      "step": 228300
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0003934081329341057,
+      "loss": 4.2341,
+      "step": 228400
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0003934052329754301,
+      "loss": 4.2399,
+      "step": 228500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0003934023330167545,
+      "loss": 4.2355,
+      "step": 228600
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00039339943305807893,
+      "loss": 4.225,
+      "step": 228700
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00039339653309940334,
+      "loss": 4.2189,
+      "step": 228800
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00039339363314072775,
+      "loss": 4.2301,
+      "step": 228900
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0003933907331820522,
+      "loss": 4.2339,
+      "step": 229000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0003933878332233766,
+      "loss": 4.245,
+      "step": 229100
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.000393384933264701,
+      "loss": 4.2378,
+      "step": 229200
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0003933820333060254,
+      "loss": 4.2221,
+      "step": 229300
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00039337913334734983,
+      "loss": 4.2356,
+      "step": 229400
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00039337623338867424,
+      "loss": 4.2163,
+      "step": 229500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00039337333342999865,
+      "loss": 4.2289,
+      "step": 229600
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00039337043347132305,
+      "loss": 4.2284,
+      "step": 229700
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00039336753351264746,
+      "loss": 4.2371,
+      "step": 229800
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0003933646625535586,
+      "loss": 4.2255,
+      "step": 229900
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.000393361762594883,
+      "loss": 4.2097,
+      "step": 230000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0003933588626362075,
+      "loss": 4.2304,
+      "step": 230100
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0003933559626775319,
+      "loss": 4.2308,
+      "step": 230200
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0003933530627188563,
+      "loss": 4.2288,
+      "step": 230300
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0003933501627601807,
+      "loss": 4.2227,
+      "step": 230400
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0003933472628015051,
+      "loss": 4.2218,
+      "step": 230500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0003933443628428295,
+      "loss": 4.2297,
+      "step": 230600
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0003933414628841539,
+      "loss": 4.2276,
+      "step": 230700
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00039333856292547833,
+      "loss": 4.2367,
+      "step": 230800
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00039333566296680274,
+      "loss": 4.23,
+      "step": 230900
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00039333276300812715,
+      "loss": 4.2248,
+      "step": 231000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00039332986304945155,
+      "loss": 4.239,
+      "step": 231100
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00039332696309077596,
+      "loss": 4.2214,
+      "step": 231200
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00039332406313210037,
+      "loss": 4.2281,
+      "step": 231300
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0003933211631734248,
+      "loss": 4.2197,
+      "step": 231400
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00039331826321474924,
+      "loss": 4.2234,
+      "step": 231500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00039331536325607364,
+      "loss": 4.2262,
+      "step": 231600
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00039331246329739805,
+      "loss": 4.2361,
+      "step": 231700
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00039330956333872246,
+      "loss": 4.2284,
+      "step": 231800
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00039330666338004686,
+      "loss": 4.2252,
+      "step": 231900
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00039330376342137127,
+      "loss": 4.2284,
+      "step": 232000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0003933008634626957,
+      "loss": 4.2286,
+      "step": 232100
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0003932979635040201,
+      "loss": 4.2302,
+      "step": 232200
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0003932950635453445,
+      "loss": 4.2365,
+      "step": 232300
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0003932921635866689,
+      "loss": 4.231,
+      "step": 232400
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0003932892636279933,
+      "loss": 4.2278,
+      "step": 232500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0003932863636693177,
+      "loss": 4.216,
+      "step": 232600
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0003932834637106421,
+      "loss": 4.239,
+      "step": 232700
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0003932805637519665,
+      "loss": 4.2304,
+      "step": 232800
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.000393277663793291,
+      "loss": 4.2207,
+      "step": 232900
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0003932747638346154,
+      "loss": 4.2322,
+      "step": 233000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0003932718638759398,
+      "loss": 4.2309,
+      "step": 233100
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0003932689639172642,
+      "loss": 4.226,
+      "step": 233200
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0003932660639585886,
+      "loss": 4.2244,
+      "step": 233300
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.000393263163999913,
+      "loss": 4.231,
+      "step": 233400
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00039326026404123743,
+      "loss": 4.226,
+      "step": 233500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0003932573640825619,
+      "loss": 4.2258,
+      "step": 233600
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00039325446412388624,
+      "loss": 4.2397,
+      "step": 233700
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00039325156416521065,
+      "loss": 4.2206,
+      "step": 233800
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00039324866420653506,
+      "loss": 4.2324,
+      "step": 233900
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00039324579324744627,
+      "loss": 4.2359,
+      "step": 234000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00039324289328877067,
+      "loss": 4.2297,
+      "step": 234100
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0003932399933300951,
+      "loss": 4.2206,
+      "step": 234200
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0003932370933714195,
+      "loss": 4.224,
+      "step": 234300
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0003932341934127439,
+      "loss": 4.2439,
+      "step": 234400
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0003932312934540683,
+      "loss": 4.2251,
+      "step": 234500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0003932283934953927,
+      "loss": 4.2433,
+      "step": 234600
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0003932254935367171,
+      "loss": 4.2327,
+      "step": 234700
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0003932225935780415,
+      "loss": 4.2289,
+      "step": 234800
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0003932196936193659,
+      "loss": 4.2257,
+      "step": 234900
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00039321679366069033,
+      "loss": 4.2312,
+      "step": 235000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00039321389370201474,
+      "loss": 4.2348,
+      "step": 235100
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00039321099374333915,
+      "loss": 4.2285,
+      "step": 235200
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00039320809378466355,
+      "loss": 4.2272,
+      "step": 235300
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.000393205193825988,
+      "loss": 4.2333,
+      "step": 235400
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0003932022938673124,
+      "loss": 4.2256,
+      "step": 235500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00039319939390863683,
+      "loss": 4.2377,
+      "step": 235600
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00039319649394996124,
+      "loss": 4.2303,
+      "step": 235700
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0003931936229908724,
+      "loss": 4.2184,
+      "step": 235800
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0003931907230321968,
+      "loss": 4.2228,
+      "step": 235900
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0003931878230735212,
+      "loss": 4.2351,
+      "step": 236000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0003931849231148456,
+      "loss": 4.2311,
+      "step": 236100
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00039318202315617,
+      "loss": 4.2358,
+      "step": 236200
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0003931791231974944,
+      "loss": 4.2354,
+      "step": 236300
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0003931762232388189,
+      "loss": 4.23,
+      "step": 236400
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0003931733232801433,
+      "loss": 4.2267,
+      "step": 236500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0003931704233214677,
+      "loss": 4.2271,
+      "step": 236600
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0003931675233627921,
+      "loss": 4.2358,
+      "step": 236700
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0003931646234041165,
+      "loss": 4.2196,
+      "step": 236800
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0003931617234454409,
+      "loss": 4.2224,
+      "step": 236900
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00039315882348676533,
+      "loss": 4.2144,
+      "step": 237000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0003931559235280898,
+      "loss": 4.2322,
+      "step": 237100
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00039315302356941414,
+      "loss": 4.2376,
+      "step": 237200
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00039315012361073855,
+      "loss": 4.225,
+      "step": 237300
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00039314722365206296,
+      "loss": 4.2386,
+      "step": 237400
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00039314432369338736,
+      "loss": 4.2344,
+      "step": 237500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00039314142373471177,
+      "loss": 4.2215,
+      "step": 237600
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0003931385237760362,
+      "loss": 4.2233,
+      "step": 237700
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00039313562381736064,
+      "loss": 4.2414,
+      "step": 237800
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00039313272385868504,
+      "loss": 4.2174,
+      "step": 237900
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00039312982390000945,
+      "loss": 4.2208,
+      "step": 238000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00039312692394133386,
+      "loss": 4.2431,
+      "step": 238100
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00039312402398265827,
+      "loss": 4.2315,
+      "step": 238200
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00039312112402398267,
+      "loss": 4.2365,
+      "step": 238300
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0003931182240653071,
+      "loss": 4.2297,
+      "step": 238400
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0003931153241066315,
+      "loss": 4.2325,
+      "step": 238500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00039311242414795595,
+      "loss": 4.2347,
+      "step": 238600
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00039310952418928035,
+      "loss": 4.2263,
+      "step": 238700
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00039310662423060476,
+      "loss": 4.2114,
+      "step": 238800
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00039310372427192917,
+      "loss": 4.2334,
+      "step": 238900
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0003931008243132535,
+      "loss": 4.2277,
+      "step": 239000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00039309792435457793,
+      "loss": 4.231,
+      "step": 239100
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00039309502439590233,
+      "loss": 4.2326,
+      "step": 239200
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0003930921244372268,
+      "loss": 4.2242,
+      "step": 239300
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0003930892244785512,
+      "loss": 4.2293,
+      "step": 239400
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00039308635351946236,
+      "loss": 4.2328,
+      "step": 239500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0003930834535607868,
+      "loss": 4.2252,
+      "step": 239600
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0003930805536021112,
+      "loss": 4.2187,
+      "step": 239700
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00039307765364343563,
+      "loss": 4.2292,
+      "step": 239800
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00039307475368476,
+      "loss": 4.22,
+      "step": 239900
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0003930718537260844,
+      "loss": 4.2233,
+      "step": 240000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0003930689537674088,
+      "loss": 4.22,
+      "step": 240100
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0003930660538087332,
+      "loss": 4.2232,
+      "step": 240200
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00039306315385005767,
+      "loss": 4.2287,
+      "step": 240300
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0003930602538913821,
+      "loss": 4.2309,
+      "step": 240400
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0003930573539327065,
+      "loss": 4.2298,
+      "step": 240500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0003930544539740309,
+      "loss": 4.2257,
+      "step": 240600
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0003930515540153553,
+      "loss": 4.2141,
+      "step": 240700
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0003930486540566797,
+      "loss": 4.2245,
+      "step": 240800
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0003930457540980041,
+      "loss": 4.2244,
+      "step": 240900
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00039304285413932857,
+      "loss": 4.2254,
+      "step": 241000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.000393039954180653,
+      "loss": 4.2043,
+      "step": 241100
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0003930370542219774,
+      "loss": 4.2302,
+      "step": 241200
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0003930341542633018,
+      "loss": 4.225,
+      "step": 241300
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0003930312543046262,
+      "loss": 4.2296,
+      "step": 241400
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0003930283543459506,
+      "loss": 4.2264,
+      "step": 241500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00039302545438727496,
+      "loss": 4.2262,
+      "step": 241600
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0003930225544285994,
+      "loss": 4.2385,
+      "step": 241700
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0003930196544699238,
+      "loss": 4.2271,
+      "step": 241800
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00039301675451124823,
+      "loss": 4.2234,
+      "step": 241900
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00039301385455257264,
+      "loss": 4.2112,
+      "step": 242000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00039301095459389705,
+      "loss": 4.2178,
+      "step": 242100
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00039300805463522145,
+      "loss": 4.2258,
+      "step": 242200
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00039300515467654586,
+      "loss": 4.2245,
+      "step": 242300
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0003930022547178703,
+      "loss": 4.223,
+      "step": 242400
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0003929993547591947,
+      "loss": 4.2436,
+      "step": 242500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00039299645480051913,
+      "loss": 4.2258,
+      "step": 242600
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00039299355484184354,
+      "loss": 4.2238,
+      "step": 242700
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00039299065488316795,
+      "loss": 4.2189,
+      "step": 242800
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00039298775492449235,
+      "loss": 4.2298,
+      "step": 242900
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00039298485496581676,
+      "loss": 4.2431,
+      "step": 243000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00039298195500714117,
+      "loss": 4.2353,
+      "step": 243100
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0003929790550484656,
+      "loss": 4.2219,
+      "step": 243200
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00039297615508979,
+      "loss": 4.2176,
+      "step": 243300
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0003929732551311144,
+      "loss": 4.2246,
+      "step": 243400
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0003929703551724388,
+      "loss": 4.2195,
+      "step": 243500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0003929674552137632,
+      "loss": 4.2338,
+      "step": 243600
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0003929645552550876,
+      "loss": 4.2263,
+      "step": 243700
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0003929616842959988,
+      "loss": 4.224,
+      "step": 243800
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0003929587843373232,
+      "loss": 4.2195,
+      "step": 243900
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00039295588437864763,
+      "loss": 4.2309,
+      "step": 244000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00039295298441997204,
+      "loss": 4.2238,
+      "step": 244100
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00039295008446129645,
+      "loss": 4.2373,
+      "step": 244200
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00039294718450262085,
+      "loss": 4.2218,
+      "step": 244300
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00039294428454394526,
+      "loss": 4.2175,
+      "step": 244400
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00039294138458526967,
+      "loss": 4.224,
+      "step": 244500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0003929384846265941,
+      "loss": 4.2303,
+      "step": 244600
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0003929355846679185,
+      "loss": 4.2257,
+      "step": 244700
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0003929326847092429,
+      "loss": 4.2332,
+      "step": 244800
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00039292978475056735,
+      "loss": 4.2152,
+      "step": 244900
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00039292688479189176,
+      "loss": 4.2379,
+      "step": 245000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00039292398483321616,
+      "loss": 4.2345,
+      "step": 245100
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00039292108487454057,
+      "loss": 4.2207,
+      "step": 245200
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.000392918184915865,
+      "loss": 4.2377,
+      "step": 245300
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0003929152849571894,
+      "loss": 4.2335,
+      "step": 245400
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0003929123849985138,
+      "loss": 4.2273,
+      "step": 245500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0003929094850398382,
+      "loss": 4.2263,
+      "step": 245600
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0003929065850811626,
+      "loss": 4.2214,
+      "step": 245700
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00039290371412207376,
+      "loss": 4.2245,
+      "step": 245800
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0003929008141633982,
+      "loss": 4.2241,
+      "step": 245900
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00039289791420472263,
+      "loss": 4.2278,
+      "step": 246000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00039289501424604704,
+      "loss": 4.2162,
+      "step": 246100
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00039289211428737144,
+      "loss": 4.2221,
+      "step": 246200
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00039288921432869585,
+      "loss": 4.2138,
+      "step": 246300
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00039288631437002026,
+      "loss": 4.224,
+      "step": 246400
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00039288341441134466,
+      "loss": 4.2243,
+      "step": 246500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00039288051445266907,
+      "loss": 4.2204,
+      "step": 246600
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0003928776144939935,
+      "loss": 4.2208,
+      "step": 246700
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0003928747145353179,
+      "loss": 4.228,
+      "step": 246800
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0003928718145766423,
+      "loss": 4.2242,
+      "step": 246900
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0003928689146179667,
+      "loss": 4.2235,
+      "step": 247000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0003928660146592911,
+      "loss": 4.2304,
+      "step": 247100
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0003928631147006155,
+      "loss": 4.235,
+      "step": 247200
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0003928602147419399,
+      "loss": 4.2272,
+      "step": 247300
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0003928573147832644,
+      "loss": 4.2281,
+      "step": 247400
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0003928544148245888,
+      "loss": 4.2351,
+      "step": 247500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0003928515148659132,
+      "loss": 4.2236,
+      "step": 247600
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0003928486149072376,
+      "loss": 4.2227,
+      "step": 247700
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00039284574394814876,
+      "loss": 4.2412,
+      "step": 247800
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00039284284398947316,
+      "loss": 4.2321,
+      "step": 247900
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00039283994403079757,
+      "loss": 4.2221,
+      "step": 248000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.000392837044072122,
+      "loss": 4.2251,
+      "step": 248100
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0003928341441134464,
+      "loss": 4.2212,
+      "step": 248200
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0003928312441547708,
+      "loss": 4.2191,
+      "step": 248300
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00039282834419609525,
+      "loss": 4.2285,
+      "step": 248400
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00039282544423741966,
+      "loss": 4.221,
+      "step": 248500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00039282254427874406,
+      "loss": 4.2246,
+      "step": 248600
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00039281964432006847,
+      "loss": 4.2064,
+      "step": 248700
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0003928167443613929,
+      "loss": 4.2218,
+      "step": 248800
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0003928138444027173,
+      "loss": 4.2196,
+      "step": 248900
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0003928109444440417,
+      "loss": 4.2296,
+      "step": 249000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0003928080444853661,
+      "loss": 4.2191,
+      "step": 249100
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0003928051445266905,
+      "loss": 4.2281,
+      "step": 249200
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0003928022445680149,
+      "loss": 4.2367,
+      "step": 249300
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0003927993446093393,
+      "loss": 4.2158,
+      "step": 249400
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0003927964446506637,
+      "loss": 4.233,
+      "step": 249500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00039279354469198813,
+      "loss": 4.2208,
+      "step": 249600
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00039279064473331254,
+      "loss": 4.238,
+      "step": 249700
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.000392787744774637,
+      "loss": 4.2319,
+      "step": 249800
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0003927848448159614,
+      "loss": 4.2339,
+      "step": 249900
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0003927819448572858,
+      "loss": 4.2201,
+      "step": 250000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00039277907389819697,
+      "loss": 4.2269,
+      "step": 250100
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0003927761739395214,
+      "loss": 4.2325,
+      "step": 250200
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0003927732739808458,
+      "loss": 4.2191,
+      "step": 250300
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0003927703740221702,
+      "loss": 4.2182,
+      "step": 250400
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0003927674740634946,
+      "loss": 4.2323,
+      "step": 250500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.000392764574104819,
+      "loss": 4.2055,
+      "step": 250600
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0003927616741461434,
+      "loss": 4.2199,
+      "step": 250700
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0003927587741874678,
+      "loss": 4.22,
+      "step": 250800
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0003927558742287923,
+      "loss": 4.2247,
+      "step": 250900
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0003927529742701167,
+      "loss": 4.2144,
+      "step": 251000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0003927500743114411,
+      "loss": 4.2157,
+      "step": 251100
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0003927471743527655,
+      "loss": 4.2259,
+      "step": 251200
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0003927442743940899,
+      "loss": 4.2176,
+      "step": 251300
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0003927413744354143,
+      "loss": 4.2097,
+      "step": 251400
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0003927384744767387,
+      "loss": 4.228,
+      "step": 251500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00039273557451806313,
+      "loss": 4.2199,
+      "step": 251600
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00039273267455938753,
+      "loss": 4.2343,
+      "step": 251700
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00039272977460071194,
+      "loss": 4.2322,
+      "step": 251800
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00039272687464203635,
+      "loss": 4.2139,
+      "step": 251900
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00039272397468336076,
+      "loss": 4.2159,
+      "step": 252000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00039272110372427197,
+      "loss": 4.2161,
+      "step": 252100
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00039271820376559637,
+      "loss": 4.2156,
+      "step": 252200
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0003927153038069208,
+      "loss": 4.2229,
+      "step": 252300
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0003927124038482452,
+      "loss": 4.2403,
+      "step": 252400
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0003927095038895696,
+      "loss": 4.2282,
+      "step": 252500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.000392706603930894,
+      "loss": 4.2188,
+      "step": 252600
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0003927037039722184,
+      "loss": 4.2273,
+      "step": 252700
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0003927008040135428,
+      "loss": 4.2255,
+      "step": 252800
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0003926979040548672,
+      "loss": 4.2204,
+      "step": 252900
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00039269500409619163,
+      "loss": 4.2294,
+      "step": 253000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00039269210413751603,
+      "loss": 4.2034,
+      "step": 253100
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00039268920417884044,
+      "loss": 4.231,
+      "step": 253200
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0003926863042201649,
+      "loss": 4.2291,
+      "step": 253300
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0003926834042614893,
+      "loss": 4.2436,
+      "step": 253400
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0003926805043028137,
+      "loss": 4.2281,
+      "step": 253500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0003926776043441381,
+      "loss": 4.2249,
+      "step": 253600
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00039267470438546253,
+      "loss": 4.2182,
+      "step": 253700
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00039267180442678694,
+      "loss": 4.2127,
+      "step": 253800
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00039266890446811134,
+      "loss": 4.2165,
+      "step": 253900
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0003926660045094358,
+      "loss": 4.2194,
+      "step": 254000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0003926631045507602,
+      "loss": 4.2148,
+      "step": 254100
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00039266020459208456,
+      "loss": 4.2077,
+      "step": 254200
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0003926573336329957,
+      "loss": 4.2151,
+      "step": 254300
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0003926544336743202,
+      "loss": 4.2235,
+      "step": 254400
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0003926515337156446,
+      "loss": 4.2213,
+      "step": 254500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.000392648633756969,
+      "loss": 4.2239,
+      "step": 254600
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0003926457337982934,
+      "loss": 4.226,
+      "step": 254700
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0003926428338396178,
+      "loss": 4.2269,
+      "step": 254800
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0003926399338809422,
+      "loss": 4.213,
+      "step": 254900
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0003926370339222666,
+      "loss": 4.2274,
+      "step": 255000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00039263413396359103,
+      "loss": 4.2176,
+      "step": 255100
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00039263123400491544,
+      "loss": 4.2138,
+      "step": 255200
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00039262833404623984,
+      "loss": 4.2083,
+      "step": 255300
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00039262543408756425,
+      "loss": 4.2196,
+      "step": 255400
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00039262253412888866,
+      "loss": 4.2281,
+      "step": 255500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00039261963417021306,
+      "loss": 4.2215,
+      "step": 255600
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00039261673421153747,
+      "loss": 4.2318,
+      "step": 255700
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00039261383425286193,
+      "loss": 4.2221,
+      "step": 255800
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00039261093429418634,
+      "loss": 4.2227,
+      "step": 255900
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00039260803433551075,
+      "loss": 4.2153,
+      "step": 256000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00039260513437683515,
+      "loss": 4.2244,
+      "step": 256100
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00039260223441815956,
+      "loss": 4.2196,
+      "step": 256200
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0003925993634590707,
+      "loss": 4.2343,
+      "step": 256300
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0003925964635003951,
+      "loss": 4.2156,
+      "step": 256400
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00039259356354171953,
+      "loss": 4.2363,
+      "step": 256500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00039259066358304394,
+      "loss": 4.2211,
+      "step": 256600
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00039258776362436834,
+      "loss": 4.2262,
+      "step": 256700
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0003925848636656928,
+      "loss": 4.2162,
+      "step": 256800
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0003925819637070172,
+      "loss": 4.2257,
+      "step": 256900
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0003925790637483416,
+      "loss": 4.2235,
+      "step": 257000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.000392576163789666,
+      "loss": 4.2267,
+      "step": 257100
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00039257326383099043,
+      "loss": 4.2233,
+      "step": 257200
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00039257036387231484,
+      "loss": 4.2195,
+      "step": 257300
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00039256746391363924,
+      "loss": 4.2235,
+      "step": 257400
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0003925645639549637,
+      "loss": 4.2104,
+      "step": 257500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0003925616639962881,
+      "loss": 4.225,
+      "step": 257600
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00039255876403761247,
+      "loss": 4.2281,
+      "step": 257700
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00039255586407893687,
+      "loss": 4.2072,
+      "step": 257800
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0003925529641202613,
+      "loss": 4.2268,
+      "step": 257900
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0003925500641615857,
+      "loss": 4.217,
+      "step": 258000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0003925471642029101,
+      "loss": 4.2218,
+      "step": 258100
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0003925442642442345,
+      "loss": 4.2229,
+      "step": 258200
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00039254136428555896,
+      "loss": 4.214,
+      "step": 258300
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0003925384933264701,
+      "loss": 4.2095,
+      "step": 258400
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0003925355933677945,
+      "loss": 4.2265,
+      "step": 258500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00039253269340911893,
+      "loss": 4.2298,
+      "step": 258600
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00039252979345044334,
+      "loss": 4.222,
+      "step": 258700
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00039252689349176774,
+      "loss": 4.2284,
+      "step": 258800
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00039252399353309215,
+      "loss": 4.2266,
+      "step": 258900
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00039252109357441656,
+      "loss": 4.2377,
+      "step": 259000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00039251819361574096,
+      "loss": 4.2169,
+      "step": 259100
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00039251529365706537,
+      "loss": 4.2367,
+      "step": 259200
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00039251239369838983,
+      "loss": 4.2234,
+      "step": 259300
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00039250949373971424,
+      "loss": 4.2108,
+      "step": 259400
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00039250659378103865,
+      "loss": 4.2261,
+      "step": 259500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00039250369382236305,
+      "loss": 4.218,
+      "step": 259600
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00039250079386368746,
+      "loss": 4.2251,
+      "step": 259700
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00039249789390501187,
+      "loss": 4.2242,
+      "step": 259800
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0003924949939463363,
+      "loss": 4.2211,
+      "step": 259900
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00039249209398766074,
+      "loss": 4.2192,
+      "step": 260000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00039248919402898514,
+      "loss": 4.2197,
+      "step": 260100
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00039248629407030955,
+      "loss": 4.226,
+      "step": 260200
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0003924833941116339,
+      "loss": 4.2269,
+      "step": 260300
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0003924805231525451,
+      "loss": 4.2264,
+      "step": 260400
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0003924776231938695,
+      "loss": 4.2213,
+      "step": 260500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0003924747232351939,
+      "loss": 4.2132,
+      "step": 260600
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00039247182327651833,
+      "loss": 4.2262,
+      "step": 260700
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00039246892331784274,
+      "loss": 4.2176,
+      "step": 260800
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00039246602335916715,
+      "loss": 4.2177,
+      "step": 260900
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0003924631234004916,
+      "loss": 4.2446,
+      "step": 261000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.000392460223441816,
+      "loss": 4.2204,
+      "step": 261100
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00039245732348314037,
+      "loss": 4.2091,
+      "step": 261200
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0003924544235244648,
+      "loss": 4.2261,
+      "step": 261300
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0003924515235657892,
+      "loss": 4.2094,
+      "step": 261400
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0003924486236071136,
+      "loss": 4.2255,
+      "step": 261500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.000392445723648438,
+      "loss": 4.2216,
+      "step": 261600
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0003924428236897624,
+      "loss": 4.2154,
+      "step": 261700
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00039243992373108686,
+      "loss": 4.2286,
+      "step": 261800
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00039243702377241127,
+      "loss": 4.2273,
+      "step": 261900
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0003924341238137357,
+      "loss": 4.2227,
+      "step": 262000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0003924312238550601,
+      "loss": 4.2204,
+      "step": 262100
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0003924283238963845,
+      "loss": 4.2383,
+      "step": 262200
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0003924254239377089,
+      "loss": 4.22,
+      "step": 262300
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0003924225239790333,
+      "loss": 4.2198,
+      "step": 262400
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00039241962402035776,
+      "loss": 4.225,
+      "step": 262500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00039241675306126887,
+      "loss": 4.2176,
+      "step": 262600
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00039241385310259327,
+      "loss": 4.2182,
+      "step": 262700
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00039241095314391773,
+      "loss": 4.2311,
+      "step": 262800
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00039240805318524214,
+      "loss": 4.2204,
+      "step": 262900
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00039240515322656655,
+      "loss": 4.24,
+      "step": 263000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00039240225326789095,
+      "loss": 4.2335,
+      "step": 263100
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00039239935330921536,
+      "loss": 4.2178,
+      "step": 263200
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00039239645335053977,
+      "loss": 4.2299,
+      "step": 263300
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0003923935533918642,
+      "loss": 4.2193,
+      "step": 263400
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00039239065343318864,
+      "loss": 4.2231,
+      "step": 263500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00039238775347451304,
+      "loss": 4.2163,
+      "step": 263600
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00039238488251542414,
+      "loss": 4.2219,
+      "step": 263700
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0003923819825567486,
+      "loss": 4.2264,
+      "step": 263800
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.000392379082598073,
+      "loss": 4.233,
+      "step": 263900
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0003923761826393974,
+      "loss": 4.2206,
+      "step": 264000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0003923732826807218,
+      "loss": 4.2091,
+      "step": 264100
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00039237038272204623,
+      "loss": 4.2234,
+      "step": 264200
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00039236748276337064,
+      "loss": 4.2216,
+      "step": 264300
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00039236458280469505,
+      "loss": 4.2315,
+      "step": 264400
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0003923616828460195,
+      "loss": 4.217,
+      "step": 264500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0003923587828873439,
+      "loss": 4.2229,
+      "step": 264600
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00039235588292866827,
+      "loss": 4.2251,
+      "step": 264700
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0003923529829699927,
+      "loss": 4.213,
+      "step": 264800
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0003923500830113171,
+      "loss": 4.2129,
+      "step": 264900
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0003923471830526415,
+      "loss": 4.2318,
+      "step": 265000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0003923442830939659,
+      "loss": 4.2269,
+      "step": 265100
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0003923413831352903,
+      "loss": 4.2136,
+      "step": 265200
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00039233848317661476,
+      "loss": 4.224,
+      "step": 265300
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00039233558321793917,
+      "loss": 4.223,
+      "step": 265400
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0003923326832592636,
+      "loss": 4.2245,
+      "step": 265500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.000392329783300588,
+      "loss": 4.223,
+      "step": 265600
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0003923268833419124,
+      "loss": 4.2298,
+      "step": 265700
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0003923239833832368,
+      "loss": 4.2261,
+      "step": 265800
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0003923210834245612,
+      "loss": 4.2268,
+      "step": 265900
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00039231818346588567,
+      "loss": 4.2167,
+      "step": 266000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00039231528350721007,
+      "loss": 4.2257,
+      "step": 266100
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0003923123835485345,
+      "loss": 4.2211,
+      "step": 266200
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0003923094835898589,
+      "loss": 4.2092,
+      "step": 266300
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0003923065836311833,
+      "loss": 4.2233,
+      "step": 266400
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00039230368367250765,
+      "loss": 4.2249,
+      "step": 266500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00039230078371383205,
+      "loss": 4.2279,
+      "step": 266600
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0003922978837551565,
+      "loss": 4.2239,
+      "step": 266700
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0003922949837964809,
+      "loss": 4.2179,
+      "step": 266800
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00039229208383780533,
+      "loss": 4.208,
+      "step": 266900
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00039228918387912973,
+      "loss": 4.2217,
+      "step": 267000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00039228628392045414,
+      "loss": 4.2343,
+      "step": 267100
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00039228338396177855,
+      "loss": 4.2151,
+      "step": 267200
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00039228048400310295,
+      "loss": 4.2092,
+      "step": 267300
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0003922775840444274,
+      "loss": 4.2285,
+      "step": 267400
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0003922746840857518,
+      "loss": 4.2173,
+      "step": 267500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00039227178412707623,
+      "loss": 4.2263,
+      "step": 267600
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00039226888416840064,
+      "loss": 4.229,
+      "step": 267700
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00039226604220889854,
+      "loss": 4.2184,
+      "step": 267800
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00039226314225022295,
+      "loss": 4.2282,
+      "step": 267900
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0003922602422915474,
+      "loss": 4.2076,
+      "step": 268000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0003922573423328718,
+      "loss": 4.2101,
+      "step": 268100
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0003922544423741962,
+      "loss": 4.2248,
+      "step": 268200
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0003922515424155206,
+      "loss": 4.212,
+      "step": 268300
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.000392248642456845,
+      "loss": 4.2165,
+      "step": 268400
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0003922457424981694,
+      "loss": 4.2231,
+      "step": 268500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0003922428425394938,
+      "loss": 4.214,
+      "step": 268600
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0003922399425808182,
+      "loss": 4.2167,
+      "step": 268700
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00039223704262214266,
+      "loss": 4.2236,
+      "step": 268800
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00039223414266346707,
+      "loss": 4.2216,
+      "step": 268900
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0003922312427047915,
+      "loss": 4.2232,
+      "step": 269000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0003922283427461159,
+      "loss": 4.219,
+      "step": 269100
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0003922254427874403,
+      "loss": 4.2218,
+      "step": 269200
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0003922225428287647,
+      "loss": 4.2224,
+      "step": 269300
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0003922196428700891,
+      "loss": 4.2287,
+      "step": 269400
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00039221674291141357,
+      "loss": 4.2211,
+      "step": 269500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.000392213842952738,
+      "loss": 4.2253,
+      "step": 269600
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0003922109429940624,
+      "loss": 4.2249,
+      "step": 269700
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0003922080430353868,
+      "loss": 4.2232,
+      "step": 269800
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00039220517207629794,
+      "loss": 4.2212,
+      "step": 269900
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00039220227211762235,
+      "loss": 4.2321,
+      "step": 270000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00039219937215894676,
+      "loss": 4.2208,
+      "step": 270100
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00039219647220027116,
+      "loss": 4.2151,
+      "step": 270200
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00039219357224159557,
+      "loss": 4.2184,
+      "step": 270300
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00039219067228292,
+      "loss": 4.2156,
+      "step": 270400
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00039218777232424444,
+      "loss": 4.2068,
+      "step": 270500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00039218487236556885,
+      "loss": 4.2301,
+      "step": 270600
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00039218197240689325,
+      "loss": 4.2293,
+      "step": 270700
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00039217907244821766,
+      "loss": 4.2246,
+      "step": 270800
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.000392176172489542,
+      "loss": 4.2144,
+      "step": 270900
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0003921732725308664,
+      "loss": 4.219,
+      "step": 271000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0003921703725721908,
+      "loss": 4.2181,
+      "step": 271100
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0003921674726135153,
+      "loss": 4.2233,
+      "step": 271200
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0003921645726548397,
+      "loss": 4.2309,
+      "step": 271300
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0003921616726961641,
+      "loss": 4.2189,
+      "step": 271400
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0003921587727374885,
+      "loss": 4.2297,
+      "step": 271500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0003921558727788129,
+      "loss": 4.217,
+      "step": 271600
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0003921529728201373,
+      "loss": 4.2307,
+      "step": 271700
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00039215007286146173,
+      "loss": 4.2162,
+      "step": 271800
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0003921471729027862,
+      "loss": 4.2079,
+      "step": 271900
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0003921442729441106,
+      "loss": 4.2162,
+      "step": 272000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.000392141372985435,
+      "loss": 4.2281,
+      "step": 272100
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0003921385020263461,
+      "loss": 4.2151,
+      "step": 272200
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00039213560206767057,
+      "loss": 4.2153,
+      "step": 272300
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00039213270210899497,
+      "loss": 4.2061,
+      "step": 272400
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0003921298021503194,
+      "loss": 4.2186,
+      "step": 272500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0003921269021916438,
+      "loss": 4.2271,
+      "step": 272600
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0003921240022329682,
+      "loss": 4.2153,
+      "step": 272700
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0003921211022742926,
+      "loss": 4.2191,
+      "step": 272800
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.000392118202315617,
+      "loss": 4.2198,
+      "step": 272900
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00039211530235694147,
+      "loss": 4.2024,
+      "step": 273000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0003921124023982659,
+      "loss": 4.2215,
+      "step": 273100
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0003921095024395903,
+      "loss": 4.2231,
+      "step": 273200
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0003921066024809147,
+      "loss": 4.2272,
+      "step": 273300
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0003921037025222391,
+      "loss": 4.22,
+      "step": 273400
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00039210080256356345,
+      "loss": 4.2274,
+      "step": 273500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00039209790260488785,
+      "loss": 4.2421,
+      "step": 273600
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0003920950026462123,
+      "loss": 4.2291,
+      "step": 273700
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0003920921026875367,
+      "loss": 4.1995,
+      "step": 273800
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00039208920272886113,
+      "loss": 4.2161,
+      "step": 273900
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00039208630277018554,
+      "loss": 4.2099,
+      "step": 274000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00039208340281150994,
+      "loss": 4.228,
+      "step": 274100
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00039208050285283435,
+      "loss": 4.2214,
+      "step": 274200
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00039207760289415876,
+      "loss": 4.2242,
+      "step": 274300
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0003920747029354832,
+      "loss": 4.2018,
+      "step": 274400
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0003920718029768076,
+      "loss": 4.2253,
+      "step": 274500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00039206890301813203,
+      "loss": 4.2283,
+      "step": 274600
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00039206600305945644,
+      "loss": 4.233,
+      "step": 274700
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00039206310310078085,
+      "loss": 4.2274,
+      "step": 274800
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00039206020314210525,
+      "loss": 4.2205,
+      "step": 274900
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00039205730318342966,
+      "loss": 4.2085,
+      "step": 275000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00039205440322475407,
+      "loss": 4.2232,
+      "step": 275100
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0003920515032660785,
+      "loss": 4.2215,
+      "step": 275200
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0003920486033074029,
+      "loss": 4.2211,
+      "step": 275300
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0003920457033487273,
+      "loss": 4.2232,
+      "step": 275400
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0003920428033900517,
+      "loss": 4.2255,
+      "step": 275500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0003920399034313761,
+      "loss": 4.2182,
+      "step": 275600
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0003920370034727005,
+      "loss": 4.2203,
+      "step": 275700
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00039203410351402497,
+      "loss": 4.2106,
+      "step": 275800
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0003920312035553494,
+      "loss": 4.2315,
+      "step": 275900
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0003920283035966738,
+      "loss": 4.2141,
+      "step": 276000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0003920254036379982,
+      "loss": 4.2144,
+      "step": 276100
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0003920225036793226,
+      "loss": 4.2067,
+      "step": 276200
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.000392019603720647,
+      "loss": 4.2042,
+      "step": 276300
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0003920167037619714,
+      "loss": 4.1975,
+      "step": 276400
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00039201383280288257,
+      "loss": 4.1942,
+      "step": 276500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00039201093284420697,
+      "loss": 4.2142,
+      "step": 276600
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0003920080328855314,
+      "loss": 4.2113,
+      "step": 276700
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0003920051329268558,
+      "loss": 4.2034,
+      "step": 276800
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.000392002261967767,
+      "loss": 4.2075,
+      "step": 276900
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00039199936200909135,
+      "loss": 4.2128,
+      "step": 277000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00039199646205041576,
+      "loss": 4.2109,
+      "step": 277100
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0003919935620917402,
+      "loss": 4.201,
+      "step": 277200
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0003919906621330646,
+      "loss": 4.2107,
+      "step": 277300
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00039198776217438903,
+      "loss": 4.2123,
+      "step": 277400
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00039198486221571344,
+      "loss": 4.2211,
+      "step": 277500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00039198196225703784,
+      "loss": 4.2054,
+      "step": 277600
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00039197906229836225,
+      "loss": 4.2116,
+      "step": 277700
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00039197616233968666,
+      "loss": 4.2041,
+      "step": 277800
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0003919732623810111,
+      "loss": 4.2127,
+      "step": 277900
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0003919703624223355,
+      "loss": 4.2014,
+      "step": 278000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00039196746246365993,
+      "loss": 4.1906,
+      "step": 278100
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00039196456250498434,
+      "loss": 4.195,
+      "step": 278200
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00039196166254630875,
+      "loss": 4.2181,
+      "step": 278300
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00039195876258763315,
+      "loss": 4.2025,
+      "step": 278400
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00039195586262895756,
+      "loss": 4.2052,
+      "step": 278500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00039195296267028197,
+      "loss": 4.1958,
+      "step": 278600
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0003919500627116064,
+      "loss": 4.204,
+      "step": 278700
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0003919471627529308,
+      "loss": 4.2088,
+      "step": 278800
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0003919442627942552,
+      "loss": 4.2088,
+      "step": 278900
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0003919413628355796,
+      "loss": 4.1875,
+      "step": 279000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.000391938462876904,
+      "loss": 4.2045,
+      "step": 279100
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0003919355629182284,
+      "loss": 4.2043,
+      "step": 279200
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00039193266295955287,
+      "loss": 4.2066,
+      "step": 279300
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0003919297630008773,
+      "loss": 4.2061,
+      "step": 279400
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0003919268630422017,
+      "loss": 4.2047,
+      "step": 279500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0003919239630835261,
+      "loss": 4.2167,
+      "step": 279600
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0003919210631248505,
+      "loss": 4.2112,
+      "step": 279700
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0003919181631661749,
+      "loss": 4.2074,
+      "step": 279800
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0003919152632074993,
+      "loss": 4.1958,
+      "step": 279900
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0003919123632488237,
+      "loss": 4.2226,
+      "step": 280000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0003919094632901481,
+      "loss": 4.2057,
+      "step": 280100
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0003919065923310593,
+      "loss": 4.2067,
+      "step": 280200
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0003919036923723837,
+      "loss": 4.2104,
+      "step": 280300
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00039190079241370815,
+      "loss": 4.2195,
+      "step": 280400
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00039189789245503256,
+      "loss": 4.2094,
+      "step": 280500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00039189499249635696,
+      "loss": 4.2143,
+      "step": 280600
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00039189209253768137,
+      "loss": 4.2286,
+      "step": 280700
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0003918891925790058,
+      "loss": 4.2196,
+      "step": 280800
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0003918862926203302,
+      "loss": 4.2148,
+      "step": 280900
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0003918833926616546,
+      "loss": 4.2066,
+      "step": 281000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.000391880492702979,
+      "loss": 4.2122,
+      "step": 281100
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0003918775927443034,
+      "loss": 4.2182,
+      "step": 281200
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0003918746927856278,
+      "loss": 4.2209,
+      "step": 281300
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0003918717928269522,
+      "loss": 4.2138,
+      "step": 281400
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0003918688928682766,
+      "loss": 4.2253,
+      "step": 281500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00039186599290960103,
+      "loss": 4.2116,
+      "step": 281600
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00039186309295092544,
+      "loss": 4.2121,
+      "step": 281700
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0003918601929922499,
+      "loss": 4.2125,
+      "step": 281800
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0003918572930335743,
+      "loss": 4.2132,
+      "step": 281900
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0003918543930748987,
+      "loss": 4.2163,
+      "step": 282000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0003918514931162231,
+      "loss": 4.236,
+      "step": 282100
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0003918485931575475,
+      "loss": 4.2116,
+      "step": 282200
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00039184569319887193,
+      "loss": 4.2176,
+      "step": 282300
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00039184279324019634,
+      "loss": 4.2081,
+      "step": 282400
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0003918398932815208,
+      "loss": 4.2181,
+      "step": 282500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00039183699332284515,
+      "loss": 4.2031,
+      "step": 282600
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00039183409336416956,
+      "loss": 4.2157,
+      "step": 282700
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00039183119340549397,
+      "loss": 4.2025,
+      "step": 282800
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0003918282934468184,
+      "loss": 4.2214,
+      "step": 282900
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0003918253934881428,
+      "loss": 4.2118,
+      "step": 283000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0003918224935294672,
+      "loss": 4.2191,
+      "step": 283100
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00039181959357079165,
+      "loss": 4.2182,
+      "step": 283200
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00039181669361211606,
+      "loss": 4.208,
+      "step": 283300
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00039181379365344046,
+      "loss": 4.2125,
+      "step": 283400
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00039181089369476487,
+      "loss": 4.2205,
+      "step": 283500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0003918079937360893,
+      "loss": 4.2023,
+      "step": 283600
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0003918050937774137,
+      "loss": 4.2153,
+      "step": 283700
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0003918021938187381,
+      "loss": 4.2064,
+      "step": 283800
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00039179929386006255,
+      "loss": 4.2145,
+      "step": 283900
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00039179639390138696,
+      "loss": 4.2062,
+      "step": 284000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00039179349394271137,
+      "loss": 4.2143,
+      "step": 284100
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0003917906229836225,
+      "loss": 4.2092,
+      "step": 284200
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00039178772302494693,
+      "loss": 4.207,
+      "step": 284300
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00039178482306627134,
+      "loss": 4.2156,
+      "step": 284400
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00039178192310759574,
+      "loss": 4.1908,
+      "step": 284500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00039177902314892015,
+      "loss": 4.2183,
+      "step": 284600
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00039177612319024456,
+      "loss": 4.2154,
+      "step": 284700
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00039177322323156896,
+      "loss": 4.2106,
+      "step": 284800
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00039177032327289337,
+      "loss": 4.2177,
+      "step": 284900
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00039176742331421783,
+      "loss": 4.2029,
+      "step": 285000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00039176452335554224,
+      "loss": 4.2224,
+      "step": 285100
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0003917616233968666,
+      "loss": 4.2057,
+      "step": 285200
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.000391758723438191,
+      "loss": 4.2025,
+      "step": 285300
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0003917558234795154,
+      "loss": 4.2207,
+      "step": 285400
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0003917529235208398,
+      "loss": 4.2098,
+      "step": 285500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0003917500235621642,
+      "loss": 4.2034,
+      "step": 285600
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0003917471236034887,
+      "loss": 4.223,
+      "step": 285700
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0003917442236448131,
+      "loss": 4.2229,
+      "step": 285800
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0003917413236861375,
+      "loss": 4.2172,
+      "step": 285900
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0003917384237274619,
+      "loss": 4.2123,
+      "step": 286000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0003917355237687863,
+      "loss": 4.2229,
+      "step": 286100
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0003917326238101107,
+      "loss": 4.2138,
+      "step": 286200
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00039172975285102187,
+      "loss": 4.2063,
+      "step": 286300
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0003917268528923463,
+      "loss": 4.2174,
+      "step": 286400
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0003917239529336707,
+      "loss": 4.2105,
+      "step": 286500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0003917210529749951,
+      "loss": 4.213,
+      "step": 286600
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00039171815301631955,
+      "loss": 4.2194,
+      "step": 286700
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00039171525305764396,
+      "loss": 4.218,
+      "step": 286800
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00039171235309896836,
+      "loss": 4.2288,
+      "step": 286900
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00039170945314029277,
+      "loss": 4.2122,
+      "step": 287000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0003917065531816172,
+      "loss": 4.2163,
+      "step": 287100
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0003917036532229416,
+      "loss": 4.2165,
+      "step": 287200
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.000391700753264266,
+      "loss": 4.1967,
+      "step": 287300
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00039169785330559045,
+      "loss": 4.2219,
+      "step": 287400
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00039169495334691486,
+      "loss": 4.2042,
+      "step": 287500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00039169205338823927,
+      "loss": 4.2058,
+      "step": 287600
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0003916891534295637,
+      "loss": 4.1974,
+      "step": 287700
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.000391686253470888,
+      "loss": 4.2147,
+      "step": 287800
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00039168335351221243,
+      "loss": 4.2123,
+      "step": 287900
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00039168045355353684,
+      "loss": 4.2215,
+      "step": 288000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0003916775535948613,
+      "loss": 4.2083,
+      "step": 288100
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0003916746536361857,
+      "loss": 4.2136,
+      "step": 288200
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0003916717536775101,
+      "loss": 4.2113,
+      "step": 288300
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0003916688537188345,
+      "loss": 4.2087,
+      "step": 288400
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00039166598275974573,
+      "loss": 4.2118,
+      "step": 288500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00039166308280107014,
+      "loss": 4.2195,
+      "step": 288600
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0003916601828423945,
+      "loss": 4.2155,
+      "step": 288700
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0003916572828837189,
+      "loss": 4.2165,
+      "step": 288800
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0003916543829250433,
+      "loss": 4.2192,
+      "step": 288900
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0003916514829663677,
+      "loss": 4.2171,
+      "step": 289000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0003916485830076921,
+      "loss": 4.2086,
+      "step": 289100
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0003916456830490166,
+      "loss": 4.1975,
+      "step": 289200
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.000391642783090341,
+      "loss": 4.2182,
+      "step": 289300
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0003916398831316654,
+      "loss": 4.2132,
+      "step": 289400
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0003916369831729898,
+      "loss": 4.2096,
+      "step": 289500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0003916340832143142,
+      "loss": 4.1979,
+      "step": 289600
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0003916311832556386,
+      "loss": 4.2094,
+      "step": 289700
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.000391628283296963,
+      "loss": 4.2136,
+      "step": 289800
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0003916253833382875,
+      "loss": 4.2068,
+      "step": 289900
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0003916224833796119,
+      "loss": 4.2044,
+      "step": 290000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0003916195834209363,
+      "loss": 4.2138,
+      "step": 290100
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0003916166834622607,
+      "loss": 4.2077,
+      "step": 290200
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0003916137835035851,
+      "loss": 4.2194,
+      "step": 290300
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0003916108835449095,
+      "loss": 4.2096,
+      "step": 290400
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00039160798358623387,
+      "loss": 4.2156,
+      "step": 290500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0003916051126271451,
+      "loss": 4.2132,
+      "step": 290600
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0003916022126684695,
+      "loss": 4.2108,
+      "step": 290700
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0003915993127097939,
+      "loss": 4.211,
+      "step": 290800
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00039159641275111835,
+      "loss": 4.2155,
+      "step": 290900
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00039159351279244276,
+      "loss": 4.2061,
+      "step": 291000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00039159061283376717,
+      "loss": 4.2115,
+      "step": 291100
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0003915877128750916,
+      "loss": 4.219,
+      "step": 291200
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.000391584812916416,
+      "loss": 4.2155,
+      "step": 291300
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00039158191295774033,
+      "loss": 4.2115,
+      "step": 291400
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00039157901299906474,
+      "loss": 4.2122,
+      "step": 291500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0003915761130403892,
+      "loss": 4.2201,
+      "step": 291600
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0003915732130817136,
+      "loss": 4.21,
+      "step": 291700
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.000391570313123038,
+      "loss": 4.2204,
+      "step": 291800
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0003915674131643624,
+      "loss": 4.2111,
+      "step": 291900
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00039156451320568683,
+      "loss": 4.2064,
+      "step": 292000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00039156161324701124,
+      "loss": 4.2227,
+      "step": 292100
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00039155871328833564,
+      "loss": 4.2178,
+      "step": 292200
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0003915558133296601,
+      "loss": 4.2136,
+      "step": 292300
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0003915529133709845,
+      "loss": 4.224,
+      "step": 292400
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0003915500134123089,
+      "loss": 4.2163,
+      "step": 292500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0003915471134536333,
+      "loss": 4.2174,
+      "step": 292600
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00039154421349495773,
+      "loss": 4.2162,
+      "step": 292700
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0003915413425358689,
+      "loss": 4.2061,
+      "step": 292800
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0003915384425771933,
+      "loss": 4.2318,
+      "step": 292900
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0003915355426185177,
+      "loss": 4.2119,
+      "step": 293000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0003915326426598421,
+      "loss": 4.2155,
+      "step": 293100
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0003915297427011665,
+      "loss": 4.2035,
+      "step": 293200
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0003915268427424909,
+      "loss": 4.2092,
+      "step": 293300
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0003915239427838154,
+      "loss": 4.2091,
+      "step": 293400
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0003915210428251398,
+      "loss": 4.2041,
+      "step": 293500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0003915181428664642,
+      "loss": 4.2137,
+      "step": 293600
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00039151527190737535,
+      "loss": 4.2248,
+      "step": 293700
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00039151237194869976,
+      "loss": 4.2038,
+      "step": 293800
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00039150947199002417,
+      "loss": 4.2137,
+      "step": 293900
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0003915065720313486,
+      "loss": 4.2205,
+      "step": 294000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.000391503672072673,
+      "loss": 4.2066,
+      "step": 294100
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0003915007721139974,
+      "loss": 4.2235,
+      "step": 294200
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0003914978721553218,
+      "loss": 4.2146,
+      "step": 294300
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00039149497219664626,
+      "loss": 4.2142,
+      "step": 294400
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00039149207223797066,
+      "loss": 4.2103,
+      "step": 294500
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 0.0003928901392440118,
+      "loss": 4.1411,
+      "step": 294600
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 0.00039288532216332194,
+      "loss": 4.0957,
+      "step": 294700
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 0.0003928805034808982,
+      "loss": 4.0677,
+      "step": 294800
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 0.0003928756831967806,
+      "loss": 4.0663,
+      "step": 294900
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 0.00039287086131100903,
+      "loss": 4.0514,
+      "step": 295000
+    },
+    {
+      "epoch": 21.39,
+      "learning_rate": 0.0003567606794640303,
+      "loss": 4.0252,
+      "step": 295100
+    },
+    {
+      "epoch": 21.4,
+      "learning_rate": 0.00035673236782667294,
+      "loss": 4.0012,
+      "step": 295200
+    },
+    {
+      "epoch": 21.41,
+      "learning_rate": 0.00035670404804775035,
+      "loss": 3.9916,
+      "step": 295300
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 0.00035667572012873354,
+      "loss": 3.9865,
+      "step": 295400
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 0.000356647384071094,
+      "loss": 3.9842,
+      "step": 295500
+    },
+    {
+      "epoch": 21.43,
+      "learning_rate": 0.00035661903987630383,
+      "loss": 3.9721,
+      "step": 295600
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 0.0003565906875458352,
+      "loss": 3.9662,
+      "step": 295700
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 0.00035656232708116103,
+      "loss": 3.965,
+      "step": 295800
+    },
+    {
+      "epoch": 21.45,
+      "learning_rate": 0.00035653395848375436,
+      "loss": 3.9588,
+      "step": 295900
+    },
+    {
+      "epoch": 21.46,
+      "learning_rate": 0.000356505581755089,
+      "loss": 3.9579,
+      "step": 296000
+    },
+    {
+      "epoch": 21.47,
+      "learning_rate": 0.00035647719689663887,
+      "loss": 3.9556,
+      "step": 296100
+    },
+    {
+      "epoch": 21.47,
+      "learning_rate": 0.0003564488039098785,
+      "loss": 3.9527,
+      "step": 296200
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 0.0003564204027962828,
+      "loss": 3.9508,
+      "step": 296300
+    },
+    {
+      "epoch": 21.49,
+      "learning_rate": 0.00035639199355732697,
+      "loss": 3.9482,
+      "step": 296400
+    },
+    {
+      "epoch": 21.49,
+      "learning_rate": 0.0003563635761944869,
+      "loss": 3.9479,
+      "step": 296500
+    },
+    {
+      "epoch": 21.5,
+      "learning_rate": 0.0003563351507092386,
+      "loss": 3.9478,
+      "step": 296600
+    },
+    {
+      "epoch": 21.51,
+      "learning_rate": 0.00035630671710305876,
+      "loss": 3.9425,
+      "step": 296700
+    },
+    {
+      "epoch": 21.52,
+      "learning_rate": 0.00035627827537742433,
+      "loss": 3.9425,
+      "step": 296800
+    },
+    {
+      "epoch": 21.52,
+      "learning_rate": 0.00035624982553381286,
+      "loss": 3.9379,
+      "step": 296900
+    },
+    {
+      "epoch": 21.53,
+      "learning_rate": 0.00035622136757370193,
+      "loss": 3.9366,
+      "step": 297000
+    },
+    {
+      "epoch": 21.54,
+      "learning_rate": 0.00035619290149857004,
+      "loss": 3.9392,
+      "step": 297100
+    },
+    {
+      "epoch": 21.55,
+      "learning_rate": 0.0003561644273098958,
+      "loss": 3.9372,
+      "step": 297200
+    },
+    {
+      "epoch": 21.55,
+      "learning_rate": 0.0003561359450091583,
+      "loss": 3.9414,
+      "step": 297300
+    },
+    {
+      "epoch": 21.56,
+      "learning_rate": 0.00035610745459783717,
+      "loss": 3.9406,
+      "step": 297400
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 0.00035607895607741224,
+      "loss": 3.939,
+      "step": 297500
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.0003969648894231196,
+      "loss": 3.9775,
+      "step": 297600
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.0003969628672822069,
+      "loss": 4.0215,
+      "step": 297700
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.0003969608240372223,
+      "loss": 4.0282,
+      "step": 297800
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00039695878011042964,
+      "loss": 4.0355,
+      "step": 297900
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00039695673550183604,
+      "loss": 4.0281,
+      "step": 298000
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.0003969546902114484,
+      "loss": 4.0262,
+      "step": 298100
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00039695264423927397,
+      "loss": 4.0216,
+      "step": 298200
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00039695059758531974,
+      "loss": 4.0174,
+      "step": 298300
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00039694855024959285,
+      "loss": 4.0109,
+      "step": 298400
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.0003969465022321003,
+      "loss": 4.0171,
+      "step": 298500
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.0003969444535328493,
+      "loss": 4.0163,
+      "step": 298600
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00039694240415184687,
+      "loss": 4.0065,
+      "step": 298700
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00039694035408910014,
+      "loss": 4.0091,
+      "step": 298800
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00039693830334461614,
+      "loss": 3.9978,
+      "step": 298900
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00039693625191840196,
+      "loss": 4.0069,
+      "step": 299000
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.0003969341998104648,
+      "loss": 4.0029,
+      "step": 299100
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00039693214702081177,
+      "loss": 3.9981,
+      "step": 299200
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00039693009354944985,
+      "loss": 4.0088,
+      "step": 299300
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.0003969280393963863,
+      "loss": 4.0046,
+      "step": 299400
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00039692598456162804,
+      "loss": 4.0022,
+      "step": 299500
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00039692392904518235,
+      "loss": 4.0023,
+      "step": 299600
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.0003969218728470563,
+      "loss": 4.0058,
+      "step": 299700
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00039691981596725695,
+      "loss": 4.0067,
+      "step": 299800
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.0003969177789847804,
+      "loss": 3.9932,
+      "step": 299900
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.0003969157207484725,
+      "loss": 3.9903,
+      "step": 300000
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00039691366183051263,
+      "loss": 3.9929,
+      "step": 300100
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00039691160223090795,
+      "loss": 3.9917,
+      "step": 300200
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00039690954194966557,
+      "loss": 3.9997,
+      "step": 300300
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.0003969074809867926,
+      "loss": 3.992,
+      "step": 300400
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00039690541934229615,
+      "loss": 3.9897,
+      "step": 300500
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.0003969033570161835,
+      "loss": 3.9984,
+      "step": 300600
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00039690129400846163,
+      "loss": 3.9943,
+      "step": 300700
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00039689923031913777,
+      "loss": 3.9959,
+      "step": 300800
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00039689716594821904,
+      "loss": 4.0,
+      "step": 300900
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00039689510089571255,
+      "loss": 3.9895,
+      "step": 301000
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.0003968930351616255,
+      "loss": 3.9888,
+      "step": 301100
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.000396890968745965,
+      "loss": 3.9896,
+      "step": 301200
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.0003968889016487383,
+      "loss": 3.9938,
+      "step": 301300
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.0003968868338699524,
+      "loss": 3.9886,
+      "step": 301400
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00039688476540961456,
+      "loss": 3.9892,
+      "step": 301500
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00039688269626773193,
+      "loss": 3.9876,
+      "step": 301600
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00039688062644431174,
+      "loss": 3.9915,
+      "step": 301700
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00039687855593936093,
+      "loss": 3.9936,
+      "step": 301800
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00039687648475288687,
+      "loss": 3.9879,
+      "step": 301900
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.0003968744128848967,
+      "loss": 3.9987,
+      "step": 302000
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.0003968723403353976,
+      "loss": 3.982,
+      "step": 302100
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.0003968702671043966,
+      "loss": 3.9934,
+      "step": 302200
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.0003968681931919011,
+      "loss": 3.9886,
+      "step": 302300
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.0003968661185979181,
+      "loss": 3.985,
+      "step": 302400
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00039686404332245486,
+      "loss": 3.9879,
+      "step": 302500
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00039686196736551856,
+      "loss": 3.9972,
+      "step": 302600
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.0003968598907271164,
+      "loss": 3.9931,
+      "step": 302700
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.0003968578134072555,
+      "loss": 3.9896,
+      "step": 302800
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00039685573540594317,
+      "loss": 3.9886,
+      "step": 302900
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.0003968536567231865,
+      "loss": 3.9854,
+      "step": 303000
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00039685157735899263,
+      "loss": 3.9959,
+      "step": 303100
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00039684949731336893,
+      "loss": 3.9813,
+      "step": 303200
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00039684741658632254,
+      "loss": 3.9894,
+      "step": 303300
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00039684533517786056,
+      "loss": 3.9883,
+      "step": 303400
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.0003968432530879903,
+      "loss": 3.9907,
+      "step": 303500
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00039684117031671894,
+      "loss": 3.9927,
+      "step": 303600
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00039683908686405367,
+      "loss": 3.985,
+      "step": 303700
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00039683700273000175,
+      "loss": 3.9813,
+      "step": 303800
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00039683493876609756,
+      "loss": 3.9809,
+      "step": 303900
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.0003968328532761076,
+      "loss": 3.9944,
+      "step": 304000
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00039683076710475253,
+      "loss": 3.9808,
+      "step": 304100
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.0003968286802520396,
+      "loss": 3.9819,
+      "step": 304200
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.000396826592717976,
+      "loss": 3.9916,
+      "step": 304300
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00039682452538809576,
+      "loss": 3.9807,
+      "step": 304400
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00039682243649816586,
+      "loss": 3.9853,
+      "step": 304500
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.000396820346926907,
+      "loss": 3.9882,
+      "step": 304600
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.0003968182566743262,
+      "loss": 3.9923,
+      "step": 304700
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.0003968161657404309,
+      "loss": 3.9872,
+      "step": 304800
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.0003968140741252283,
+      "loss": 3.9955,
+      "step": 304900
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.0003968119818287256,
+      "loss": 3.9846,
+      "step": 305000
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00039680988885093006,
+      "loss": 3.9832,
+      "step": 305100
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.0003968077951918489,
+      "loss": 3.9816,
+      "step": 305200
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00039680570085148933,
+      "loss": 3.9895,
+      "step": 305300
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.0003968036058298587,
+      "loss": 3.9753,
+      "step": 305400
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00039680151012696423,
+      "loss": 3.987,
+      "step": 305500
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00039679941374281316,
+      "loss": 3.9875,
+      "step": 305600
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.0003967973166774127,
+      "loss": 3.9895,
+      "step": 305700
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.0003967952189307702,
+      "loss": 3.9815,
+      "step": 305800
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.0003967931205028928,
+      "loss": 3.9851,
+      "step": 305900
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.0003967910213937879,
+      "loss": 3.9814,
+      "step": 306000
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00039678892160346273,
+      "loss": 3.9794,
+      "step": 306100
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.0003967868211319245,
+      "loss": 3.9927,
+      "step": 306200
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00039678471997918056,
+      "loss": 3.9825,
+      "step": 306300
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.0003967826181452381,
+      "loss": 3.9864,
+      "step": 306400
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00039678051563010443,
+      "loss": 3.9817,
+      "step": 306500
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00039677841243378686,
+      "loss": 3.9732,
+      "step": 306600
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00039677630855629266,
+      "loss": 3.9814,
+      "step": 306700
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00039677420399762903,
+      "loss": 3.9757,
+      "step": 306800
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.0003967720987578034,
+      "loss": 3.9838,
+      "step": 306900
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00039676999283682295,
+      "loss": 3.9841,
+      "step": 307000
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00039676788623469497,
+      "loss": 3.9811,
+      "step": 307100
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00039676577895142684,
+      "loss": 3.986,
+      "step": 307200
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00039676367098702577,
+      "loss": 3.9778,
+      "step": 307300
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00039676156234149914,
+      "loss": 3.9862,
+      "step": 307400
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00039675945301485415,
+      "loss": 3.984,
+      "step": 307500
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.0003967573430070981,
+      "loss": 3.985,
+      "step": 307600
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00039675523231823837,
+      "loss": 3.9858,
+      "step": 307700
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.0003967531209482823,
+      "loss": 3.9826,
+      "step": 307800
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00039675100889723704,
+      "loss": 3.9826,
+      "step": 307900
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00039674889616511005,
+      "loss": 3.9791,
+      "step": 308000
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00039674678275190855,
+      "loss": 3.9895,
+      "step": 308100
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.0003967446686576399,
+      "loss": 3.9848,
+      "step": 308200
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.0003967425538823115,
+      "loss": 3.9773,
+      "step": 308300
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.0003967404595838655,
+      "loss": 3.9734,
+      "step": 308400
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.0003967383434532498,
+      "loss": 3.9831,
+      "step": 308500
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.0003967362266415961,
+      "loss": 3.9703,
+      "step": 308600
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00039673410914891174,
+      "loss": 3.982,
+      "step": 308700
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.0003967319909752042,
+      "loss": 3.9836,
+      "step": 308800
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.0003967298721204806,
+      "loss": 3.9769,
+      "step": 308900
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.0003967277525847485,
+      "loss": 3.9794,
+      "step": 309000
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00039672563236801506,
+      "loss": 3.9806,
+      "step": 309100
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.0003967235114702877,
+      "loss": 3.9794,
+      "step": 309200
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.0003967213898915737,
+      "loss": 3.9802,
+      "step": 309300
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.0003967192676318805,
+      "loss": 3.9863,
+      "step": 309400
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.0003967171446912154,
+      "loss": 3.9831,
+      "step": 309500
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00039671502106958565,
+      "loss": 3.9835,
+      "step": 309600
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.0003967128967669988,
+      "loss": 3.9819,
+      "step": 309700
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00039671077178346205,
+      "loss": 3.9792,
+      "step": 309800
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.0003967086461189828,
+      "loss": 3.9784,
+      "step": 309900
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00039670651977356846,
+      "loss": 3.9825,
+      "step": 310000
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.0003967043927472263,
+      "loss": 3.9785,
+      "step": 310100
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00039670226503996374,
+      "loss": 3.9841,
+      "step": 310200
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.0003967001366517881,
+      "loss": 3.9748,
+      "step": 310300
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00039669802887676807,
+      "loss": 3.9815,
+      "step": 310400
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00039669589913359746,
+      "loss": 3.9862,
+      "step": 310500
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00039669376870953574,
+      "loss": 3.9805,
+      "step": 310600
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00039669163760459045,
+      "loss": 3.9813,
+      "step": 310700
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00039668950581876885,
+      "loss": 3.9775,
+      "step": 310800
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.0003966873946801156,
+      "loss": 3.9825,
+      "step": 310900
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00039668526153937216,
+      "loss": 3.9799,
+      "step": 311000
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00039668312771777456,
+      "loss": 3.9781,
+      "step": 311100
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00039668099321533024,
+      "loss": 3.9819,
+      "step": 311200
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00039667885803204643,
+      "loss": 3.98,
+      "step": 311300
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.0003966767221679306,
+      "loss": 3.976,
+      "step": 311400
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00039667458562299017,
+      "loss": 3.9793,
+      "step": 311500
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.0003966724483972325,
+      "loss": 3.9745,
+      "step": 311600
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00039667031049066503,
+      "loss": 3.9834,
+      "step": 311700
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00039666817190329507,
+      "loss": 3.9682,
+      "step": 311800
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.0003966660326351301,
+      "loss": 3.9841,
+      "step": 311900
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.0003966638926861775,
+      "loss": 3.9771,
+      "step": 312000
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00039666175205644466,
+      "loss": 3.9711,
+      "step": 312100
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00039665961074593906,
+      "loss": 3.992,
+      "step": 312200
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.000396657468754668,
+      "loss": 3.9794,
+      "step": 312300
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.000396655326082639,
+      "loss": 3.9697,
+      "step": 312400
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.0003966531827298595,
+      "loss": 3.9766,
+      "step": 312500
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00039665103869633675,
+      "loss": 3.9824,
+      "step": 312600
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.0003966488939820783,
+      "loss": 3.9837,
+      "step": 312700
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.0003966467485870916,
+      "loss": 3.9797,
+      "step": 312800
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.000396644602511384,
+      "loss": 3.975,
+      "step": 312900
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00039664245575496296,
+      "loss": 3.9725,
+      "step": 313000
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.0003966403083178359,
+      "loss": 3.9709,
+      "step": 313100
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00039663816020001026,
+      "loss": 3.9748,
+      "step": 313200
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.0003966360114014935,
+      "loss": 3.9732,
+      "step": 313300
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00039663386192229306,
+      "loss": 3.9755,
+      "step": 313400
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.0003966317117624163,
+      "loss": 3.9727,
+      "step": 313500
+    }
+  ],
+  "max_steps": 5340600,
+  "num_train_epochs": 100,
+  "total_flos": 2.1723600144992256e+19,
+  "trial_name": null,
+  "trial_params": null
+}