diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,78536 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.897013328984666,
+  "global_step": 6500000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7664e-06,
+      "loss": 7.1306,
+      "step": 500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.5664e-06,
+      "loss": 4.7491,
+      "step": 1000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.3664e-06,
+      "loss": 4.5563,
+      "step": 1500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1166399999999999e-05,
+      "loss": 4.4179,
+      "step": 2000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.39664e-05,
+      "loss": 4.2868,
+      "step": 2500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.67664e-05,
+      "loss": 4.1711,
+      "step": 3000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.95664e-05,
+      "loss": 4.1089,
+      "step": 3500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.23664e-05,
+      "loss": 4.0387,
+      "step": 4000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.51664e-05,
+      "loss": 3.9442,
+      "step": 4500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.79664e-05,
+      "loss": 3.8552,
+      "step": 5000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.07664e-05,
+      "loss": 3.8383,
+      "step": 5500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3566400000000004e-05,
+      "loss": 3.7433,
+      "step": 6000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.63664e-05,
+      "loss": 3.672,
+      "step": 6500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.91664e-05,
+      "loss": 3.5705,
+      "step": 7000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.19664e-05,
+      "loss": 3.4977,
+      "step": 7500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.47664e-05,
+      "loss": 3.4228,
+      "step": 8000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.75664e-05,
+      "loss": 3.3206,
+      "step": 8500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.03664e-05,
+      "loss": 3.3024,
+      "step": 9000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.31608e-05,
+      "loss": 3.238,
+      "step": 9500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.5960799999999995e-05,
+      "loss": 3.1835,
+      "step": 10000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.599792943395728e-05,
+      "loss": 3.1321,
+      "step": 10500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.599582946839672e-05,
+      "loss": 2.9498,
+      "step": 11000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.5993733702767274e-05,
+      "loss": 2.9067,
+      "step": 11500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.599163373720671e-05,
+      "loss": 2.8201,
+      "step": 12000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.598953377164614e-05,
+      "loss": 2.7697,
+      "step": 12500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.598743380608558e-05,
+      "loss": 2.6956,
+      "step": 13000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.5985333840525015e-05,
+      "loss": 2.6767,
+      "step": 13500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.598323387496445e-05,
+      "loss": 2.6511,
+      "step": 14000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.598113390940389e-05,
+      "loss": 2.6005,
+      "step": 14500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.597903394384332e-05,
+      "loss": 2.5646,
+      "step": 15000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.5976933978282755e-05,
+      "loss": 2.5398,
+      "step": 15500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.5974838212653316e-05,
+      "loss": 2.5406,
+      "step": 16000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.597273824709275e-05,
+      "loss": 2.5244,
+      "step": 16500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.59706424814633e-05,
+      "loss": 2.4838,
+      "step": 17000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.5968542515902736e-05,
+      "loss": 2.4476,
+      "step": 17500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.5966442550342176e-05,
+      "loss": 2.3907,
+      "step": 18000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.596434258478161e-05,
+      "loss": 2.4071,
+      "step": 18500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.596224261922104e-05,
+      "loss": 2.3964,
+      "step": 19000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.5960142653660483e-05,
+      "loss": 2.3317,
+      "step": 19500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.595804268809992e-05,
+      "loss": 2.3686,
+      "step": 20000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.5955942722539344e-05,
+      "loss": 2.3225,
+      "step": 20500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.5953842756978784e-05,
+      "loss": 2.2592,
+      "step": 21000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.595174279141822e-05,
+      "loss": 2.2759,
+      "step": 21500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.594964282585766e-05,
+      "loss": 2.2664,
+      "step": 22000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.594754286029709e-05,
+      "loss": 2.281,
+      "step": 22500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.594545129459877e-05,
+      "loss": 2.239,
+      "step": 23000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.5943351329038205e-05,
+      "loss": 2.2078,
+      "step": 23500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.594125556340876e-05,
+      "loss": 2.2347,
+      "step": 24000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.593915559784819e-05,
+      "loss": 2.1976,
+      "step": 24500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.593705563228763e-05,
+      "loss": 2.1759,
+      "step": 25000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.5934955666727066e-05,
+      "loss": 2.1614,
+      "step": 25500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.593285990109762e-05,
+      "loss": 2.2002,
+      "step": 26000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.593075993553705e-05,
+      "loss": 2.167,
+      "step": 26500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.592865996997649e-05,
+      "loss": 2.1416,
+      "step": 27000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.5926560004415926e-05,
+      "loss": 2.1846,
+      "step": 27500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.592446003885537e-05,
+      "loss": 2.1457,
+      "step": 28000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.59223600732948e-05,
+      "loss": 2.1489,
+      "step": 28500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.5920260107734234e-05,
+      "loss": 2.0987,
+      "step": 29000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.5918160142173674e-05,
+      "loss": 2.102,
+      "step": 29500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.59160601766131e-05,
+      "loss": 2.1108,
+      "step": 30000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.5913960211052534e-05,
+      "loss": 2.1382,
+      "step": 30500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.5911864445423094e-05,
+      "loss": 2.0619,
+      "step": 31000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.590976867979365e-05,
+      "loss": 2.0802,
+      "step": 31500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.590766871423309e-05,
+      "loss": 2.0627,
+      "step": 32000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.590556874867252e-05,
+      "loss": 2.0826,
+      "step": 32500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.5903468783111955e-05,
+      "loss": 2.0486,
+      "step": 33000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.5901368817551395e-05,
+      "loss": 2.0652,
+      "step": 33500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.589926885199083e-05,
+      "loss": 2.048,
+      "step": 34000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.589716888643026e-05,
+      "loss": 2.0727,
+      "step": 34500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.5895068920869696e-05,
+      "loss": 2.0037,
+      "step": 35000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.589296895530913e-05,
+      "loss": 2.0548,
+      "step": 35500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.589086898974857e-05,
+      "loss": 1.9999,
+      "step": 36000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.5888769024188e-05,
+      "loss": 2.0511,
+      "step": 36500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.5886669058627436e-05,
+      "loss": 1.982,
+      "step": 37000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.588457749292912e-05,
+      "loss": 2.007,
+      "step": 37500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.588247752736855e-05,
+      "loss": 2.0074,
+      "step": 38000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.588037756180799e-05,
+      "loss": 2.0068,
+      "step": 38500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.5878277596247424e-05,
+      "loss": 2.0297,
+      "step": 39000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.587617763068685e-05,
+      "loss": 1.9801,
+      "step": 39500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.587408186505741e-05,
+      "loss": 1.968,
+      "step": 40000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.587198189949685e-05,
+      "loss": 1.9741,
+      "step": 40500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.5869881933936285e-05,
+      "loss": 2.0031,
+      "step": 41000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.586778196837572e-05,
+      "loss": 1.9725,
+      "step": 41500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.586568200281515e-05,
+      "loss": 1.9465,
+      "step": 42000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.586358623718571e-05,
+      "loss": 1.9521,
+      "step": 42500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.5861490471556265e-05,
+      "loss": 1.96,
+      "step": 43000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.58593905059957e-05,
+      "loss": 1.9815,
+      "step": 43500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.585729054043514e-05,
+      "loss": 1.9682,
+      "step": 44000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.585519057487457e-05,
+      "loss": 1.9458,
+      "step": 44500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.5853090609314006e-05,
+      "loss": 1.9562,
+      "step": 45000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.5850990643753446e-05,
+      "loss": 1.9166,
+      "step": 45500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.584889067819288e-05,
+      "loss": 1.9057,
+      "step": 46000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.5846790712632306e-05,
+      "loss": 1.9332,
+      "step": 46500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.584469074707175e-05,
+      "loss": 1.9453,
+      "step": 47000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.584259078151118e-05,
+      "loss": 1.9018,
+      "step": 47500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.5840490815950614e-05,
+      "loss": 1.9193,
+      "step": 48000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.5838390850390054e-05,
+      "loss": 1.9188,
+      "step": 48500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.583629088482949e-05,
+      "loss": 1.851,
+      "step": 49000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.583419511920004e-05,
+      "loss": 1.9075,
+      "step": 49500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.583209515363948e-05,
+      "loss": 1.8626,
+      "step": 50000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.5829995188078915e-05,
+      "loss": 1.9197,
+      "step": 50500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.582789522251835e-05,
+      "loss": 1.8625,
+      "step": 51000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.582579525695779e-05,
+      "loss": 1.903,
+      "step": 51500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.582369949132834e-05,
+      "loss": 1.8981,
+      "step": 52000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.58216037256989e-05,
+      "loss": 1.8916,
+      "step": 52500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.5819503760138336e-05,
+      "loss": 1.9053,
+      "step": 53000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.581740379457776e-05,
+      "loss": 1.8785,
+      "step": 53500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.58153038290172e-05,
+      "loss": 1.8709,
+      "step": 54000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.5813203863456636e-05,
+      "loss": 1.8518,
+      "step": 54500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.581110389789607e-05,
+      "loss": 1.8553,
+      "step": 55000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.580900393233551e-05,
+      "loss": 1.8435,
+      "step": 55500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.580690396677494e-05,
+      "loss": 1.8547,
+      "step": 56000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.5804804001214377e-05,
+      "loss": 1.8521,
+      "step": 56500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.580270403565382e-05,
+      "loss": 1.8389,
+      "step": 57000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.580060407009325e-05,
+      "loss": 1.8671,
+      "step": 57500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.5798508304463804e-05,
+      "loss": 1.8519,
+      "step": 58000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5796408338903244e-05,
+      "loss": 1.8339,
+      "step": 58500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.579430837334268e-05,
+      "loss": 1.8504,
+      "step": 59000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.579220840778211e-05,
+      "loss": 1.8243,
+      "step": 59500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.579010844222155e-05,
+      "loss": 1.8522,
+      "step": 60000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5788012676592105e-05,
+      "loss": 1.8346,
+      "step": 60500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.578591271103154e-05,
+      "loss": 1.8302,
+      "step": 61000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.578381694540209e-05,
+      "loss": 1.8138,
+      "step": 61500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5781716979841525e-05,
+      "loss": 1.8075,
+      "step": 62000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5779621214212086e-05,
+      "loss": 1.8539,
+      "step": 62500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.577752124865152e-05,
+      "loss": 1.8535,
+      "step": 63000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.577542128309095e-05,
+      "loss": 1.8144,
+      "step": 63500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.577332131753039e-05,
+      "loss": 1.8078,
+      "step": 64000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5771221351969826e-05,
+      "loss": 1.8121,
+      "step": 64500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.576912138640926e-05,
+      "loss": 1.819,
+      "step": 65000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.57670214208487e-05,
+      "loss": 1.7988,
+      "step": 65500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5764921455288133e-05,
+      "loss": 1.8449,
+      "step": 66000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.576282148972757e-05,
+      "loss": 1.8113,
+      "step": 66500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.576072152416701e-05,
+      "loss": 1.8032,
+      "step": 67000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.575862155860644e-05,
+      "loss": 1.7805,
+      "step": 67500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5756521593045874e-05,
+      "loss": 1.8332,
+      "step": 68000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.575442582741643e-05,
+      "loss": 1.8081,
+      "step": 68500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.575232586185587e-05,
+      "loss": 1.7867,
+      "step": 69000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.57502258962953e-05,
+      "loss": 1.8534,
+      "step": 69500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5748125930734735e-05,
+      "loss": 1.7946,
+      "step": 70000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.574603016510529e-05,
+      "loss": 1.7681,
+      "step": 70500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.574393439947585e-05,
+      "loss": 1.8173,
+      "step": 71000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.574183443391528e-05,
+      "loss": 1.7993,
+      "step": 71500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5739734468354716e-05,
+      "loss": 1.8121,
+      "step": 72000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5737634502794156e-05,
+      "loss": 1.7741,
+      "step": 72500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.573553453723359e-05,
+      "loss": 1.7632,
+      "step": 73000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.573343457167302e-05,
+      "loss": 1.7428,
+      "step": 73500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.573133460611246e-05,
+      "loss": 1.7824,
+      "step": 74000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5729234640551896e-05,
+      "loss": 1.7765,
+      "step": 74500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.572713467499133e-05,
+      "loss": 1.7981,
+      "step": 75000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.5725038909361883e-05,
+      "loss": 1.7925,
+      "step": 75500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.5722938943801324e-05,
+      "loss": 1.7825,
+      "step": 76000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.572083897824076e-05,
+      "loss": 1.7829,
+      "step": 76500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.571873901268019e-05,
+      "loss": 1.79,
+      "step": 77000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.571663904711963e-05,
+      "loss": 1.799,
+      "step": 77500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.5714543281490184e-05,
+      "loss": 1.7995,
+      "step": 78000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.571244331592962e-05,
+      "loss": 1.806,
+      "step": 78500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.571034335036905e-05,
+      "loss": 1.7516,
+      "step": 79000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.570824338480849e-05,
+      "loss": 1.7678,
+      "step": 79500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.5706143419247925e-05,
+      "loss": 1.7673,
+      "step": 80000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.570404345368736e-05,
+      "loss": 1.7944,
+      "step": 80500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.570194348812679e-05,
+      "loss": 1.8062,
+      "step": 81000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.5699843522566225e-05,
+      "loss": 1.7369,
+      "step": 81500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.5697747756936786e-05,
+      "loss": 1.7562,
+      "step": 82000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.5695647791376226e-05,
+      "loss": 1.7348,
+      "step": 82500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.569355622567789e-05,
+      "loss": 1.8007,
+      "step": 83000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.569145626011733e-05,
+      "loss": 1.7627,
+      "step": 83500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.568935629455677e-05,
+      "loss": 1.7367,
+      "step": 84000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.56872563289962e-05,
+      "loss": 1.7428,
+      "step": 84500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.568515636343564e-05,
+      "loss": 1.7706,
+      "step": 85000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.5683056397875074e-05,
+      "loss": 1.7478,
+      "step": 85500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.568095643231451e-05,
+      "loss": 1.7316,
+      "step": 86000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.567885646675395e-05,
+      "loss": 1.7544,
+      "step": 86500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.567675650119338e-05,
+      "loss": 1.7324,
+      "step": 87000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.5674660735563935e-05,
+      "loss": 1.7232,
+      "step": 87500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.5672560770003375e-05,
+      "loss": 1.742,
+      "step": 88000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.567046080444281e-05,
+      "loss": 1.7131,
+      "step": 88500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.566836083888224e-05,
+      "loss": 1.7744,
+      "step": 89000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.5666265073252795e-05,
+      "loss": 1.7089,
+      "step": 89500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.5664165107692235e-05,
+      "loss": 1.7256,
+      "step": 90000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.566206514213167e-05,
+      "loss": 1.7423,
+      "step": 90500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.56599651765711e-05,
+      "loss": 1.7465,
+      "step": 91000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.565786521101054e-05,
+      "loss": 1.7184,
+      "step": 91500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.5655769445381096e-05,
+      "loss": 1.7434,
+      "step": 92000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.565366947982053e-05,
+      "loss": 1.7604,
+      "step": 92500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.565157371419108e-05,
+      "loss": 1.7488,
+      "step": 93000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.5649473748630523e-05,
+      "loss": 1.697,
+      "step": 93500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.564737378306996e-05,
+      "loss": 1.7406,
+      "step": 94000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.564527381750939e-05,
+      "loss": 1.756,
+      "step": 94500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.564317385194883e-05,
+      "loss": 1.7186,
+      "step": 95000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.5641073886388264e-05,
+      "loss": 1.7215,
+      "step": 95500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.56389739208277e-05,
+      "loss": 1.7408,
+      "step": 96000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.563687395526714e-05,
+      "loss": 1.7173,
+      "step": 96500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.563477398970657e-05,
+      "loss": 1.7422,
+      "step": 97000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.5632674024146e-05,
+      "loss": 1.7511,
+      "step": 97500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.563057405858544e-05,
+      "loss": 1.7128,
+      "step": 98000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.562847409302487e-05,
+      "loss": 1.7366,
+      "step": 98500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.562638252732655e-05,
+      "loss": 1.7102,
+      "step": 99000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.5624282561765986e-05,
+      "loss": 1.6987,
+      "step": 99500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.5622182596205426e-05,
+      "loss": 1.7162,
+      "step": 100000
+    },
+    {
+      "epoch": 0.06,
+      "eval_loss": 1.6113003492355347,
+      "eval_runtime": 1095.3065,
+      "eval_samples_per_second": 480.888,
+      "eval_steps_per_second": 80.148,
+      "step": 100000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.562008263064486e-05,
+      "loss": 1.693,
+      "step": 100500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.561798266508429e-05,
+      "loss": 1.7246,
+      "step": 101000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.561588269952373e-05,
+      "loss": 1.7385,
+      "step": 101500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.561378273396316e-05,
+      "loss": 1.7152,
+      "step": 102000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.561168276840259e-05,
+      "loss": 1.6918,
+      "step": 102500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.5609587002773153e-05,
+      "loss": 1.6783,
+      "step": 103000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.5607487037212594e-05,
+      "loss": 1.6992,
+      "step": 103500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.560538707165203e-05,
+      "loss": 1.7065,
+      "step": 104000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.5603287106091454e-05,
+      "loss": 1.6912,
+      "step": 104500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.5601187140530894e-05,
+      "loss": 1.7006,
+      "step": 105000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.5599091374901454e-05,
+      "loss": 1.7121,
+      "step": 105500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.559699560927201e-05,
+      "loss": 1.6932,
+      "step": 106000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.559489564371144e-05,
+      "loss": 1.6908,
+      "step": 106500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.559279567815088e-05,
+      "loss": 1.7198,
+      "step": 107000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.5590695712590315e-05,
+      "loss": 1.7275,
+      "step": 107500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.558859574702975e-05,
+      "loss": 1.6924,
+      "step": 108000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.558649578146919e-05,
+      "loss": 1.6792,
+      "step": 108500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.5584395815908616e-05,
+      "loss": 1.6821,
+      "step": 109000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.558229585034805e-05,
+      "loss": 1.6619,
+      "step": 109500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.558020008471861e-05,
+      "loss": 1.6954,
+      "step": 110000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.557810011915805e-05,
+      "loss": 1.7016,
+      "step": 110500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.557600015359748e-05,
+      "loss": 1.6689,
+      "step": 111000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.557390018803691e-05,
+      "loss": 1.7108,
+      "step": 111500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.557180022247635e-05,
+      "loss": 1.7,
+      "step": 112000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.556970445684691e-05,
+      "loss": 1.7109,
+      "step": 112500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.5567608691217464e-05,
+      "loss": 1.6864,
+      "step": 113000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.55655087256569e-05,
+      "loss": 1.6809,
+      "step": 113500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.556341296002745e-05,
+      "loss": 1.7365,
+      "step": 114000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.556131299446689e-05,
+      "loss": 1.6716,
+      "step": 114500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.5559213028906325e-05,
+      "loss": 1.7088,
+      "step": 115000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.555711306334576e-05,
+      "loss": 1.6476,
+      "step": 115500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.555501729771631e-05,
+      "loss": 1.6933,
+      "step": 116000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.555291733215575e-05,
+      "loss": 1.6808,
+      "step": 116500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.5550817366595185e-05,
+      "loss": 1.7114,
+      "step": 117000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.554871740103462e-05,
+      "loss": 1.6573,
+      "step": 117500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.554661743547406e-05,
+      "loss": 1.6965,
+      "step": 118000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.554451746991349e-05,
+      "loss": 1.6986,
+      "step": 118500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.5542417504352926e-05,
+      "loss": 1.6969,
+      "step": 119000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.5540317538792366e-05,
+      "loss": 1.6895,
+      "step": 119500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.55382175732318e-05,
+      "loss": 1.6495,
+      "step": 120000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.553611760767123e-05,
+      "loss": 1.6495,
+      "step": 120500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.5534021842041793e-05,
+      "loss": 1.6729,
+      "step": 121000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.553192607641235e-05,
+      "loss": 1.706,
+      "step": 121500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.552982611085178e-05,
+      "loss": 1.672,
+      "step": 122000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.5527726145291214e-05,
+      "loss": 1.6255,
+      "step": 122500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.5525626179730654e-05,
+      "loss": 1.6458,
+      "step": 123000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.552352621417009e-05,
+      "loss": 1.6765,
+      "step": 123500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.552142624860952e-05,
+      "loss": 1.6828,
+      "step": 124000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.551932628304896e-05,
+      "loss": 1.6512,
+      "step": 124500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.5517226317488395e-05,
+      "loss": 1.6695,
+      "step": 125000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.551513055185895e-05,
+      "loss": 1.6789,
+      "step": 125500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.551303058629838e-05,
+      "loss": 1.6785,
+      "step": 126000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.551093062073782e-05,
+      "loss": 1.6435,
+      "step": 126500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.5508830655177255e-05,
+      "loss": 1.6531,
+      "step": 127000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.550673068961669e-05,
+      "loss": 1.6556,
+      "step": 127500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.550463072405613e-05,
+      "loss": 1.6563,
+      "step": 128000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.5502530758495556e-05,
+      "loss": 1.6816,
+      "step": 128500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.5500430792934996e-05,
+      "loss": 1.67,
+      "step": 129000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.549833082737443e-05,
+      "loss": 1.6965,
+      "step": 129500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.549623506174499e-05,
+      "loss": 1.6701,
+      "step": 130000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.5494139296115543e-05,
+      "loss": 1.6774,
+      "step": 130500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.549203933055498e-05,
+      "loss": 1.6849,
+      "step": 131000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.548993936499442e-05,
+      "loss": 1.6633,
+      "step": 131500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.548783939943385e-05,
+      "loss": 1.6135,
+      "step": 132000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.5485739433873284e-05,
+      "loss": 1.6785,
+      "step": 132500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.548363946831272e-05,
+      "loss": 1.6718,
+      "step": 133000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.548154370268328e-05,
+      "loss": 1.6594,
+      "step": 133500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.547944373712271e-05,
+      "loss": 1.6686,
+      "step": 134000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.5477343771562145e-05,
+      "loss": 1.6388,
+      "step": 134500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.5475243806001585e-05,
+      "loss": 1.6506,
+      "step": 135000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.547314804037214e-05,
+      "loss": 1.6689,
+      "step": 135500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.547104807481157e-05,
+      "loss": 1.6153,
+      "step": 136000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.546894810925101e-05,
+      "loss": 1.6636,
+      "step": 136500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.5466848143690446e-05,
+      "loss": 1.6691,
+      "step": 137000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.546474817812988e-05,
+      "loss": 1.6591,
+      "step": 137500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.546264821256931e-05,
+      "loss": 1.6408,
+      "step": 138000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.5460548247008746e-05,
+      "loss": 1.6484,
+      "step": 138500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.545844828144818e-05,
+      "loss": 1.6661,
+      "step": 139000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.545634831588762e-05,
+      "loss": 1.6503,
+      "step": 139500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.5454252550258173e-05,
+      "loss": 1.6475,
+      "step": 140000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.545215258469761e-05,
+      "loss": 1.6633,
+      "step": 140500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.545005261913704e-05,
+      "loss": 1.6189,
+      "step": 141000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.544795265357648e-05,
+      "loss": 1.6433,
+      "step": 141500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.544585688794704e-05,
+      "loss": 1.5953,
+      "step": 142000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.544375692238647e-05,
+      "loss": 1.6335,
+      "step": 142500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.544165695682591e-05,
+      "loss": 1.6495,
+      "step": 143000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.543955699126534e-05,
+      "loss": 1.6592,
+      "step": 143500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.54374612256359e-05,
+      "loss": 1.6293,
+      "step": 144000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.5435361260075335e-05,
+      "loss": 1.6412,
+      "step": 144500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.543326129451477e-05,
+      "loss": 1.655,
+      "step": 145000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.54311613289542e-05,
+      "loss": 1.654,
+      "step": 145500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.5429061363393636e-05,
+      "loss": 1.6185,
+      "step": 146000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.5426965597764196e-05,
+      "loss": 1.6205,
+      "step": 146500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.5424865632203636e-05,
+      "loss": 1.6426,
+      "step": 147000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.542276566664306e-05,
+      "loss": 1.6475,
+      "step": 147500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.5420665701082496e-05,
+      "loss": 1.6265,
+      "step": 148000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.5418565735521936e-05,
+      "loss": 1.6394,
+      "step": 148500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.541646576996137e-05,
+      "loss": 1.628,
+      "step": 149000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.54143658044008e-05,
+      "loss": 1.6245,
+      "step": 149500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.5412265838840244e-05,
+      "loss": 1.6298,
+      "step": 150000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.541016587327968e-05,
+      "loss": 1.632,
+      "step": 150500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.540807010765023e-05,
+      "loss": 1.6021,
+      "step": 151000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.540597014208967e-05,
+      "loss": 1.6157,
+      "step": 151500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.5403870176529104e-05,
+      "loss": 1.6264,
+      "step": 152000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.540177021096854e-05,
+      "loss": 1.6485,
+      "step": 152500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.539967444533909e-05,
+      "loss": 1.6493,
+      "step": 153000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.539757447977853e-05,
+      "loss": 1.6467,
+      "step": 153500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.5395474514217965e-05,
+      "loss": 1.6432,
+      "step": 154000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.539337874858852e-05,
+      "loss": 1.6397,
+      "step": 154500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.539127878302795e-05,
+      "loss": 1.6255,
+      "step": 155000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.538917881746739e-05,
+      "loss": 1.6372,
+      "step": 155500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.5387078851906826e-05,
+      "loss": 1.6155,
+      "step": 156000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.538497888634626e-05,
+      "loss": 1.6305,
+      "step": 156500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.538288312071682e-05,
+      "loss": 1.6082,
+      "step": 157000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.538078315515625e-05,
+      "loss": 1.6183,
+      "step": 157500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.5378683189595687e-05,
+      "loss": 1.6143,
+      "step": 158000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.537658322403513e-05,
+      "loss": 1.6299,
+      "step": 158500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.537448325847456e-05,
+      "loss": 1.6296,
+      "step": 159000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.5372383292913994e-05,
+      "loss": 1.6464,
+      "step": 159500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.5370283327353434e-05,
+      "loss": 1.6393,
+      "step": 160000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.536818336179287e-05,
+      "loss": 1.6345,
+      "step": 160500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.536608759616342e-05,
+      "loss": 1.6183,
+      "step": 161000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.5363987630602854e-05,
+      "loss": 1.62,
+      "step": 161500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.5361887665042295e-05,
+      "loss": 1.62,
+      "step": 162000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.535978769948173e-05,
+      "loss": 1.6003,
+      "step": 162500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.535769193385228e-05,
+      "loss": 1.6346,
+      "step": 163000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.5355591968291715e-05,
+      "loss": 1.6618,
+      "step": 163500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.5353492002731155e-05,
+      "loss": 1.5938,
+      "step": 164000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.535139203717059e-05,
+      "loss": 1.6081,
+      "step": 164500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.534929207161002e-05,
+      "loss": 1.6232,
+      "step": 165000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.534719210604946e-05,
+      "loss": 1.6045,
+      "step": 165500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.5345092140488896e-05,
+      "loss": 1.6061,
+      "step": 166000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.5342992174928336e-05,
+      "loss": 1.6491,
+      "step": 166500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.534089220936776e-05,
+      "loss": 1.6291,
+      "step": 167000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.533879644373832e-05,
+      "loss": 1.6282,
+      "step": 167500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.533669647817776e-05,
+      "loss": 1.602,
+      "step": 168000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.53345965126172e-05,
+      "loss": 1.6096,
+      "step": 168500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.533249654705663e-05,
+      "loss": 1.6116,
+      "step": 169000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.5330400781427184e-05,
+      "loss": 1.6204,
+      "step": 169500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.532830081586662e-05,
+      "loss": 1.6099,
+      "step": 170000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.532620085030606e-05,
+      "loss": 1.644,
+      "step": 170500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.532410088474549e-05,
+      "loss": 1.6269,
+      "step": 171000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.5322000919184925e-05,
+      "loss": 1.5776,
+      "step": 171500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.5319905153555485e-05,
+      "loss": 1.6387,
+      "step": 172000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.531780938792604e-05,
+      "loss": 1.6198,
+      "step": 172500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.531570942236547e-05,
+      "loss": 1.5853,
+      "step": 173000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.5313609456804905e-05,
+      "loss": 1.6199,
+      "step": 173500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.5311509491244346e-05,
+      "loss": 1.5931,
+      "step": 174000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.530940952568378e-05,
+      "loss": 1.6317,
+      "step": 174500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.530730956012321e-05,
+      "loss": 1.6126,
+      "step": 175000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.530520959456265e-05,
+      "loss": 1.6168,
+      "step": 175500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.5303113828933206e-05,
+      "loss": 1.5932,
+      "step": 176000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.530101386337264e-05,
+      "loss": 1.6496,
+      "step": 176500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.529891389781207e-05,
+      "loss": 1.6044,
+      "step": 177000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.5296813932251514e-05,
+      "loss": 1.6055,
+      "step": 177500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.529471816662207e-05,
+      "loss": 1.5994,
+      "step": 178000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.52926182010615e-05,
+      "loss": 1.5919,
+      "step": 178500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.529051823550094e-05,
+      "loss": 1.5779,
+      "step": 179000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.5288418269940374e-05,
+      "loss": 1.6163,
+      "step": 179500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.528631830437981e-05,
+      "loss": 1.6164,
+      "step": 180000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.528421833881925e-05,
+      "loss": 1.6308,
+      "step": 180500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.528211837325868e-05,
+      "loss": 1.6218,
+      "step": 181000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.528001840769811e-05,
+      "loss": 1.5762,
+      "step": 181500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.527792264206867e-05,
+      "loss": 1.6014,
+      "step": 182000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.527582267650811e-05,
+      "loss": 1.6038,
+      "step": 182500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.527372271094754e-05,
+      "loss": 1.6381,
+      "step": 183000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.5271622745386976e-05,
+      "loss": 1.6253,
+      "step": 183500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.526952277982641e-05,
+      "loss": 1.6,
+      "step": 184000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.526742281426584e-05,
+      "loss": 1.5856,
+      "step": 184500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.5265322848705276e-05,
+      "loss": 1.6109,
+      "step": 185000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.5263222883144716e-05,
+      "loss": 1.6121,
+      "step": 185500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.526112291758415e-05,
+      "loss": 1.6169,
+      "step": 186000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.52590271519547e-05,
+      "loss": 1.5976,
+      "step": 186500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.5256927186394144e-05,
+      "loss": 1.6129,
+      "step": 187000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.525482722083358e-05,
+      "loss": 1.5925,
+      "step": 187500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.525272725527301e-05,
+      "loss": 1.6216,
+      "step": 188000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.525062728971245e-05,
+      "loss": 1.6201,
+      "step": 188500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.5248531524083004e-05,
+      "loss": 1.6284,
+      "step": 189000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.5246435758453565e-05,
+      "loss": 1.5996,
+      "step": 189500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.5244335792893e-05,
+      "loss": 1.5812,
+      "step": 190000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.524224002726355e-05,
+      "loss": 1.586,
+      "step": 190500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.5240140061702985e-05,
+      "loss": 1.6248,
+      "step": 191000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.5238040096142425e-05,
+      "loss": 1.6143,
+      "step": 191500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.523594013058186e-05,
+      "loss": 1.581,
+      "step": 192000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.523384016502129e-05,
+      "loss": 1.6128,
+      "step": 192500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.523174019946073e-05,
+      "loss": 1.5824,
+      "step": 193000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.522964023390016e-05,
+      "loss": 1.6002,
+      "step": 193500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.52275402683396e-05,
+      "loss": 1.6012,
+      "step": 194000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.522544030277903e-05,
+      "loss": 1.5855,
+      "step": 194500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.522334453714959e-05,
+      "loss": 1.6029,
+      "step": 195000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.522124457158903e-05,
+      "loss": 1.6014,
+      "step": 195500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.521914460602846e-05,
+      "loss": 1.6003,
+      "step": 196000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.5217044640467894e-05,
+      "loss": 1.57,
+      "step": 196500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.521494467490733e-05,
+      "loss": 1.6085,
+      "step": 197000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.521284470934677e-05,
+      "loss": 1.6124,
+      "step": 197500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.52107447437862e-05,
+      "loss": 1.5989,
+      "step": 198000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.5208644778225634e-05,
+      "loss": 1.5842,
+      "step": 198500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.520654901259619e-05,
+      "loss": 1.6082,
+      "step": 199000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.520444904703563e-05,
+      "loss": 1.6194,
+      "step": 199500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.520235748133731e-05,
+      "loss": 1.5846,
+      "step": 200000
+    },
+    {
+      "epoch": 0.12,
+      "eval_loss": 1.5057743787765503,
+      "eval_runtime": 1098.6714,
+      "eval_samples_per_second": 479.415,
+      "eval_steps_per_second": 79.903,
+      "step": 200000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.520025751577674e-05,
+      "loss": 1.6172,
+      "step": 200500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.5198157550216175e-05,
+      "loss": 1.5929,
+      "step": 201000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.5196057584655616e-05,
+      "loss": 1.5772,
+      "step": 201500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.519395761909505e-05,
+      "loss": 1.5737,
+      "step": 202000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.519185765353448e-05,
+      "loss": 1.575,
+      "step": 202500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.5189757687973916e-05,
+      "loss": 1.6277,
+      "step": 203000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.518765772241335e-05,
+      "loss": 1.5874,
+      "step": 203500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.518555775685278e-05,
+      "loss": 1.5785,
+      "step": 204000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.518345779129222e-05,
+      "loss": 1.5956,
+      "step": 204500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.5181357825731657e-05,
+      "loss": 1.5306,
+      "step": 205000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.517925786017109e-05,
+      "loss": 1.6043,
+      "step": 205500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.517715789461053e-05,
+      "loss": 1.6062,
+      "step": 206000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.5175062128981084e-05,
+      "loss": 1.5506,
+      "step": 206500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.5172966363351644e-05,
+      "loss": 1.6059,
+      "step": 207000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.517086639779107e-05,
+      "loss": 1.6007,
+      "step": 207500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.516876643223051e-05,
+      "loss": 1.6055,
+      "step": 208000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.5166666466669945e-05,
+      "loss": 1.6001,
+      "step": 208500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.5164570701040505e-05,
+      "loss": 1.5949,
+      "step": 209000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.516247073547994e-05,
+      "loss": 1.599,
+      "step": 209500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.516037076991937e-05,
+      "loss": 1.5766,
+      "step": 210000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.5158270804358805e-05,
+      "loss": 1.6037,
+      "step": 210500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.515617083879824e-05,
+      "loss": 1.567,
+      "step": 211000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.515407087323768e-05,
+      "loss": 1.5633,
+      "step": 211500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.515197510760824e-05,
+      "loss": 1.5883,
+      "step": 212000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.5149875142047666e-05,
+      "loss": 1.5684,
+      "step": 212500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.51477751764871e-05,
+      "loss": 1.5685,
+      "step": 213000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.514567521092654e-05,
+      "loss": 1.5924,
+      "step": 213500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.514357524536597e-05,
+      "loss": 1.5858,
+      "step": 214000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.5141479479736534e-05,
+      "loss": 1.5703,
+      "step": 214500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.513937951417597e-05,
+      "loss": 1.5536,
+      "step": 215000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.51372795486154e-05,
+      "loss": 1.5849,
+      "step": 215500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.5135179583054834e-05,
+      "loss": 1.5829,
+      "step": 216000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.5133079617494274e-05,
+      "loss": 1.579,
+      "step": 216500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.513097965193371e-05,
+      "loss": 1.5838,
+      "step": 217000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.512887968637314e-05,
+      "loss": 1.5788,
+      "step": 217500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.512677972081258e-05,
+      "loss": 1.6009,
+      "step": 218000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.5124679755252015e-05,
+      "loss": 1.5677,
+      "step": 218500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.5122588189553695e-05,
+      "loss": 1.5503,
+      "step": 219000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.512048822399312e-05,
+      "loss": 1.5778,
+      "step": 219500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.5118388258432555e-05,
+      "loss": 1.5821,
+      "step": 220000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.5116288292871996e-05,
+      "loss": 1.5648,
+      "step": 220500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.5114192527242556e-05,
+      "loss": 1.5803,
+      "step": 221000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.511209256168199e-05,
+      "loss": 1.5933,
+      "step": 221500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.510999259612142e-05,
+      "loss": 1.6024,
+      "step": 222000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.510789683049198e-05,
+      "loss": 1.5776,
+      "step": 222500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.510579686493142e-05,
+      "loss": 1.6148,
+      "step": 223000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.510369689937085e-05,
+      "loss": 1.5659,
+      "step": 223500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.510159693381029e-05,
+      "loss": 1.5784,
+      "step": 224000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.509949696824972e-05,
+      "loss": 1.5568,
+      "step": 224500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.509739700268915e-05,
+      "loss": 1.5756,
+      "step": 225000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.509529703712859e-05,
+      "loss": 1.5393,
+      "step": 225500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.5093197071568024e-05,
+      "loss": 1.5623,
+      "step": 226000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.509109710600746e-05,
+      "loss": 1.5507,
+      "step": 226500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.50889971404469e-05,
+      "loss": 1.5647,
+      "step": 227000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.508689717488633e-05,
+      "loss": 1.5549,
+      "step": 227500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.5084797209325765e-05,
+      "loss": 1.5854,
+      "step": 228000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.508270144369632e-05,
+      "loss": 1.556,
+      "step": 228500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.508060147813576e-05,
+      "loss": 1.5822,
+      "step": 229000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.507850151257519e-05,
+      "loss": 1.5834,
+      "step": 229500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.5076401547014626e-05,
+      "loss": 1.6046,
+      "step": 230000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.5074301581454066e-05,
+      "loss": 1.6046,
+      "step": 230500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.5072210015755746e-05,
+      "loss": 1.5461,
+      "step": 231000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.507011005019517e-05,
+      "loss": 1.5611,
+      "step": 231500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.5068010084634606e-05,
+      "loss": 1.5498,
+      "step": 232000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.506591011907405e-05,
+      "loss": 1.5501,
+      "step": 232500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.506381015351348e-05,
+      "loss": 1.548,
+      "step": 233000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.5061710187952914e-05,
+      "loss": 1.5708,
+      "step": 233500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.505961442232347e-05,
+      "loss": 1.5545,
+      "step": 234000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.505751865669403e-05,
+      "loss": 1.555,
+      "step": 234500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.505541869113347e-05,
+      "loss": 1.5698,
+      "step": 235000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.50533187255729e-05,
+      "loss": 1.5515,
+      "step": 235500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.5051218760012335e-05,
+      "loss": 1.5406,
+      "step": 236000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.504911879445177e-05,
+      "loss": 1.5443,
+      "step": 236500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.50470188288912e-05,
+      "loss": 1.5536,
+      "step": 237000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.504491886333064e-05,
+      "loss": 1.5699,
+      "step": 237500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.5042818897770075e-05,
+      "loss": 1.5656,
+      "step": 238000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.504072313214063e-05,
+      "loss": 1.5824,
+      "step": 238500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.503862316658006e-05,
+      "loss": 1.5363,
+      "step": 239000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.50365232010195e-05,
+      "loss": 1.5354,
+      "step": 239500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.5034423235458936e-05,
+      "loss": 1.5804,
+      "step": 240000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.503232326989837e-05,
+      "loss": 1.5398,
+      "step": 240500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.503023170420005e-05,
+      "loss": 1.5531,
+      "step": 241000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.502813173863948e-05,
+      "loss": 1.5521,
+      "step": 241500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.5026031773078924e-05,
+      "loss": 1.5674,
+      "step": 242000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.502393180751836e-05,
+      "loss": 1.5475,
+      "step": 242500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.50218318419578e-05,
+      "loss": 1.549,
+      "step": 243000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.5019731876397224e-05,
+      "loss": 1.5572,
+      "step": 243500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.501763191083666e-05,
+      "loss": 1.5546,
+      "step": 244000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.50155319452761e-05,
+      "loss": 1.54,
+      "step": 244500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.501343197971553e-05,
+      "loss": 1.5534,
+      "step": 245000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.5011332014154965e-05,
+      "loss": 1.5464,
+      "step": 245500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.500923624852552e-05,
+      "loss": 1.5766,
+      "step": 246000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.500713628296496e-05,
+      "loss": 1.5508,
+      "step": 246500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.500503631740439e-05,
+      "loss": 1.5527,
+      "step": 247000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.5002936351843825e-05,
+      "loss": 1.5564,
+      "step": 247500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.5000836386283266e-05,
+      "loss": 1.5638,
+      "step": 248000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.49987364207227e-05,
+      "loss": 1.5471,
+      "step": 248500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.499663645516213e-05,
+      "loss": 1.5921,
+      "step": 249000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.499453648960157e-05,
+      "loss": 1.5616,
+      "step": 249500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.4992436524041006e-05,
+      "loss": 1.5572,
+      "step": 250000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.499034075841156e-05,
+      "loss": 1.5648,
+      "step": 250500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.4988240792851e-05,
+      "loss": 1.5501,
+      "step": 251000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.4986145027221554e-05,
+      "loss": 1.6062,
+      "step": 251500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.498404506166099e-05,
+      "loss": 1.5563,
+      "step": 252000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.498194509610042e-05,
+      "loss": 1.5646,
+      "step": 252500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.497984513053986e-05,
+      "loss": 1.5585,
+      "step": 253000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.4977745164979294e-05,
+      "loss": 1.5076,
+      "step": 253500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.497564519941873e-05,
+      "loss": 1.5712,
+      "step": 254000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.497354523385817e-05,
+      "loss": 1.5462,
+      "step": 254500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.49714452682976e-05,
+      "loss": 1.5614,
+      "step": 255000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.4969349502668155e-05,
+      "loss": 1.5738,
+      "step": 255500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.496724953710759e-05,
+      "loss": 1.5401,
+      "step": 256000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.496514957154703e-05,
+      "loss": 1.548,
+      "step": 256500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.496304960598646e-05,
+      "loss": 1.5519,
+      "step": 257000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.4960953840357016e-05,
+      "loss": 1.5678,
+      "step": 257500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.495885807472757e-05,
+      "loss": 1.5166,
+      "step": 258000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.495675810916701e-05,
+      "loss": 1.5571,
+      "step": 258500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.495465814360644e-05,
+      "loss": 1.5538,
+      "step": 259000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.4952558178045876e-05,
+      "loss": 1.5246,
+      "step": 259500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.4950458212485317e-05,
+      "loss": 1.5855,
+      "step": 260000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.494835824692475e-05,
+      "loss": 1.5427,
+      "step": 260500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.4946258281364184e-05,
+      "loss": 1.5513,
+      "step": 261000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.4944158315803624e-05,
+      "loss": 1.5516,
+      "step": 261500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.494205835024306e-05,
+      "loss": 1.5543,
+      "step": 262000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.493995838468249e-05,
+      "loss": 1.5701,
+      "step": 262500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.4937862619053044e-05,
+      "loss": 1.5495,
+      "step": 263000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.4935762653492484e-05,
+      "loss": 1.5427,
+      "step": 263500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.493366268793192e-05,
+      "loss": 1.5414,
+      "step": 264000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.493156272237135e-05,
+      "loss": 1.5612,
+      "step": 264500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.492946275681079e-05,
+      "loss": 1.5663,
+      "step": 265000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.4927366991181345e-05,
+      "loss": 1.5413,
+      "step": 265500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.492526702562078e-05,
+      "loss": 1.5568,
+      "step": 266000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.492316706006022e-05,
+      "loss": 1.5595,
+      "step": 266500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.492106709449965e-05,
+      "loss": 1.5211,
+      "step": 267000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.4918967128939086e-05,
+      "loss": 1.5112,
+      "step": 267500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.491686716337852e-05,
+      "loss": 1.5416,
+      "step": 268000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.491476719781795e-05,
+      "loss": 1.5689,
+      "step": 268500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.4912667232257386e-05,
+      "loss": 1.5339,
+      "step": 269000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.4910567266696826e-05,
+      "loss": 1.5484,
+      "step": 269500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.490847150106739e-05,
+      "loss": 1.5524,
+      "step": 270000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.490637573543794e-05,
+      "loss": 1.5528,
+      "step": 270500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.4904279969808494e-05,
+      "loss": 1.5559,
+      "step": 271000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.490218000424793e-05,
+      "loss": 1.5594,
+      "step": 271500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.490008003868737e-05,
+      "loss": 1.5377,
+      "step": 272000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.48979800731268e-05,
+      "loss": 1.537,
+      "step": 272500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.4895880107566235e-05,
+      "loss": 1.5462,
+      "step": 273000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.4893780142005675e-05,
+      "loss": 1.5742,
+      "step": 273500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.489168017644511e-05,
+      "loss": 1.5052,
+      "step": 274000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.488958021088454e-05,
+      "loss": 1.583,
+      "step": 274500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.4887480245323975e-05,
+      "loss": 1.5206,
+      "step": 275000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.4885384479694536e-05,
+      "loss": 1.576,
+      "step": 275500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.488328451413397e-05,
+      "loss": 1.5534,
+      "step": 276000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.488118874850452e-05,
+      "loss": 1.5095,
+      "step": 276500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.4879088782943956e-05,
+      "loss": 1.5479,
+      "step": 277000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.4876988817383396e-05,
+      "loss": 1.5154,
+      "step": 277500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.487488885182283e-05,
+      "loss": 1.5269,
+      "step": 278000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.487278888626226e-05,
+      "loss": 1.5386,
+      "step": 278500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.4870688920701703e-05,
+      "loss": 1.5105,
+      "step": 279000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.486858895514114e-05,
+      "loss": 1.5905,
+      "step": 279500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.486648898958057e-05,
+      "loss": 1.5191,
+      "step": 280000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.4864389024020004e-05,
+      "loss": 1.5372,
+      "step": 280500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.486228905845944e-05,
+      "loss": 1.5744,
+      "step": 281000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.486018909289888e-05,
+      "loss": 1.5372,
+      "step": 281500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.485809332726944e-05,
+      "loss": 1.5391,
+      "step": 282000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.4855993361708864e-05,
+      "loss": 1.5146,
+      "step": 282500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.48538933961483e-05,
+      "loss": 1.522,
+      "step": 283000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.485179343058774e-05,
+      "loss": 1.5382,
+      "step": 283500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.48496976649583e-05,
+      "loss": 1.5458,
+      "step": 284000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.4847597699397725e-05,
+      "loss": 1.5505,
+      "step": 284500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.484549773383716e-05,
+      "loss": 1.5513,
+      "step": 285000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.48433977682766e-05,
+      "loss": 1.5208,
+      "step": 285500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.484129780271603e-05,
+      "loss": 1.5379,
+      "step": 286000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.4839197837155466e-05,
+      "loss": 1.5663,
+      "step": 286500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.4837097871594906e-05,
+      "loss": 1.5263,
+      "step": 287000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.483499790603434e-05,
+      "loss": 1.5318,
+      "step": 287500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.483289794047377e-05,
+      "loss": 1.5053,
+      "step": 288000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.4830806374775453e-05,
+      "loss": 1.528,
+      "step": 288500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.4828706409214894e-05,
+      "loss": 1.5249,
+      "step": 289000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.482660644365432e-05,
+      "loss": 1.5616,
+      "step": 289500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.4824506478093754e-05,
+      "loss": 1.5175,
+      "step": 290000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.4822406512533194e-05,
+      "loss": 1.5088,
+      "step": 290500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.4820310746903754e-05,
+      "loss": 1.5718,
+      "step": 291000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.481821078134319e-05,
+      "loss": 1.5678,
+      "step": 291500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.4816110815782615e-05,
+      "loss": 1.5313,
+      "step": 292000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.4814010850222055e-05,
+      "loss": 1.5149,
+      "step": 292500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.481191088466149e-05,
+      "loss": 1.542,
+      "step": 293000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.480981091910092e-05,
+      "loss": 1.5365,
+      "step": 293500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.480771095354036e-05,
+      "loss": 1.5118,
+      "step": 294000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.4805610987979795e-05,
+      "loss": 1.5491,
+      "step": 294500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.480351102241923e-05,
+      "loss": 1.5336,
+      "step": 295000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.480141105685867e-05,
+      "loss": 1.5175,
+      "step": 295500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.479931529122922e-05,
+      "loss": 1.5407,
+      "step": 296000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.4797215325668656e-05,
+      "loss": 1.5076,
+      "step": 296500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.479511956003921e-05,
+      "loss": 1.5337,
+      "step": 297000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.479301959447865e-05,
+      "loss": 1.5312,
+      "step": 297500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.4790919628918083e-05,
+      "loss": 1.5084,
+      "step": 298000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.478881966335752e-05,
+      "loss": 1.5111,
+      "step": 298500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.478671969779696e-05,
+      "loss": 1.5434,
+      "step": 299000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.478461973223639e-05,
+      "loss": 1.5214,
+      "step": 299500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.4782519766675824e-05,
+      "loss": 1.54,
+      "step": 300000
+    },
+    {
+      "epoch": 0.18,
+      "eval_loss": 1.45011568069458,
+      "eval_runtime": 1097.0232,
+      "eval_samples_per_second": 480.136,
+      "eval_steps_per_second": 80.023,
+      "step": 300000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.478042400104638e-05,
+      "loss": 1.5421,
+      "step": 300500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.477832403548582e-05,
+      "loss": 1.5362,
+      "step": 301000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.477622406992525e-05,
+      "loss": 1.5563,
+      "step": 301500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.4774124104364685e-05,
+      "loss": 1.5506,
+      "step": 302000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.4772024138804125e-05,
+      "loss": 1.5259,
+      "step": 302500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.476992417324356e-05,
+      "loss": 1.5519,
+      "step": 303000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.476782840761411e-05,
+      "loss": 1.5289,
+      "step": 303500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.476572844205355e-05,
+      "loss": 1.536,
+      "step": 304000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.4763628476492986e-05,
+      "loss": 1.5506,
+      "step": 304500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.476152851093242e-05,
+      "loss": 1.4987,
+      "step": 305000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.475942854537186e-05,
+      "loss": 1.5034,
+      "step": 305500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.475732857981129e-05,
+      "loss": 1.5534,
+      "step": 306000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.4755228614250726e-05,
+      "loss": 1.5103,
+      "step": 306500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.475312864869016e-05,
+      "loss": 1.5517,
+      "step": 307000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.475103288306072e-05,
+      "loss": 1.5442,
+      "step": 307500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.4748937117431274e-05,
+      "loss": 1.5199,
+      "step": 308000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.474683715187071e-05,
+      "loss": 1.4951,
+      "step": 308500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.474473718631015e-05,
+      "loss": 1.5294,
+      "step": 309000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.474263722074958e-05,
+      "loss": 1.5289,
+      "step": 309500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.4740541455120134e-05,
+      "loss": 1.4918,
+      "step": 310000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.4738445689490695e-05,
+      "loss": 1.5376,
+      "step": 310500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.473634572393012e-05,
+      "loss": 1.5349,
+      "step": 311000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.473424575836956e-05,
+      "loss": 1.4939,
+      "step": 311500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.4732145792808995e-05,
+      "loss": 1.5329,
+      "step": 312000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.473004582724843e-05,
+      "loss": 1.5036,
+      "step": 312500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.472794586168787e-05,
+      "loss": 1.5368,
+      "step": 313000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.47258458961273e-05,
+      "loss": 1.5455,
+      "step": 313500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.4723745930566736e-05,
+      "loss": 1.5161,
+      "step": 314000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.4721645965006176e-05,
+      "loss": 1.4974,
+      "step": 314500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.471954599944561e-05,
+      "loss": 1.5229,
+      "step": 315000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.471744603388504e-05,
+      "loss": 1.5048,
+      "step": 315500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.471534606832448e-05,
+      "loss": 1.551,
+      "step": 316000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.471324610276391e-05,
+      "loss": 1.5474,
+      "step": 316500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.471114613720335e-05,
+      "loss": 1.5466,
+      "step": 317000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.470905037157391e-05,
+      "loss": 1.5042,
+      "step": 317500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.4706950406013344e-05,
+      "loss": 1.5298,
+      "step": 318000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.470485044045278e-05,
+      "loss": 1.5123,
+      "step": 318500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.470275047489221e-05,
+      "loss": 1.5178,
+      "step": 319000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.4700650509331644e-05,
+      "loss": 1.5563,
+      "step": 319500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.4698554743702205e-05,
+      "loss": 1.5241,
+      "step": 320000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.469645477814164e-05,
+      "loss": 1.5211,
+      "step": 320500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.469435901251219e-05,
+      "loss": 1.5086,
+      "step": 321000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.469225904695163e-05,
+      "loss": 1.5145,
+      "step": 321500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.4690159081391065e-05,
+      "loss": 1.5319,
+      "step": 322000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.46880591158305e-05,
+      "loss": 1.5272,
+      "step": 322500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.468595915026994e-05,
+      "loss": 1.5315,
+      "step": 323000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.4683859184709366e-05,
+      "loss": 1.5158,
+      "step": 323500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.4681759219148806e-05,
+      "loss": 1.5042,
+      "step": 324000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.467965925358824e-05,
+      "loss": 1.5135,
+      "step": 324500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.46775634879588e-05,
+      "loss": 1.5164,
+      "step": 325000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.467546352239823e-05,
+      "loss": 1.578,
+      "step": 325500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.467336355683767e-05,
+      "loss": 1.5135,
+      "step": 326000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.46712635912771e-05,
+      "loss": 1.5154,
+      "step": 326500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.4669163625716534e-05,
+      "loss": 1.5164,
+      "step": 327000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.4667063660155974e-05,
+      "loss": 1.5277,
+      "step": 327500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.466496369459541e-05,
+      "loss": 1.5319,
+      "step": 328000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.466286372903484e-05,
+      "loss": 1.517,
+      "step": 328500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.4660767963405394e-05,
+      "loss": 1.5135,
+      "step": 329000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.4658667997844835e-05,
+      "loss": 1.5282,
+      "step": 329500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.465656803228427e-05,
+      "loss": 1.5398,
+      "step": 330000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.465447226665483e-05,
+      "loss": 1.5406,
+      "step": 330500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.465237230109426e-05,
+      "loss": 1.4958,
+      "step": 331000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.4650272335533695e-05,
+      "loss": 1.5157,
+      "step": 331500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.464817236997313e-05,
+      "loss": 1.5218,
+      "step": 332000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.464607660434369e-05,
+      "loss": 1.5262,
+      "step": 332500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.464397663878312e-05,
+      "loss": 1.5212,
+      "step": 333000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.4641876673222556e-05,
+      "loss": 1.5404,
+      "step": 333500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.463977670766199e-05,
+      "loss": 1.5006,
+      "step": 334000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.463767674210143e-05,
+      "loss": 1.5156,
+      "step": 334500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.463558097647199e-05,
+      "loss": 1.5274,
+      "step": 335000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.463348101091142e-05,
+      "loss": 1.5069,
+      "step": 335500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.463138104535085e-05,
+      "loss": 1.5011,
+      "step": 336000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.462928107979029e-05,
+      "loss": 1.5012,
+      "step": 336500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.462718531416085e-05,
+      "loss": 1.4935,
+      "step": 337000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.4625085348600284e-05,
+      "loss": 1.5293,
+      "step": 337500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.462298538303972e-05,
+      "loss": 1.4987,
+      "step": 338000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.462088961741028e-05,
+      "loss": 1.5118,
+      "step": 338500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.461878965184971e-05,
+      "loss": 1.5351,
+      "step": 339000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.4616689686289145e-05,
+      "loss": 1.4981,
+      "step": 339500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.46145939206597e-05,
+      "loss": 1.4874,
+      "step": 340000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.461249395509914e-05,
+      "loss": 1.5241,
+      "step": 340500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.461039398953857e-05,
+      "loss": 1.5351,
+      "step": 341000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.4608294023978006e-05,
+      "loss": 1.5116,
+      "step": 341500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.4606194058417446e-05,
+      "loss": 1.4991,
+      "step": 342000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.460409409285687e-05,
+      "loss": 1.5235,
+      "step": 342500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.4601994127296306e-05,
+      "loss": 1.4995,
+      "step": 343000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.4599898361666866e-05,
+      "loss": 1.4979,
+      "step": 343500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.459779839610631e-05,
+      "loss": 1.4825,
+      "step": 344000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.459569843054574e-05,
+      "loss": 1.527,
+      "step": 344500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.4593598464985174e-05,
+      "loss": 1.5359,
+      "step": 345000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.4591502699355734e-05,
+      "loss": 1.4982,
+      "step": 345500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.458940273379517e-05,
+      "loss": 1.5081,
+      "step": 346000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.45873027682346e-05,
+      "loss": 1.5767,
+      "step": 346500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.458520280267404e-05,
+      "loss": 1.5586,
+      "step": 347000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.458310283711347e-05,
+      "loss": 1.4968,
+      "step": 347500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.45810028715529e-05,
+      "loss": 1.5282,
+      "step": 348000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.457890290599234e-05,
+      "loss": 1.5299,
+      "step": 348500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.4576802940431775e-05,
+      "loss": 1.5139,
+      "step": 349000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.457470297487121e-05,
+      "loss": 1.5233,
+      "step": 349500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.457260300931065e-05,
+      "loss": 1.5259,
+      "step": 350000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.457050304375008e-05,
+      "loss": 1.5051,
+      "step": 350500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.4568403078189516e-05,
+      "loss": 1.5282,
+      "step": 351000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.4566303112628956e-05,
+      "loss": 1.4965,
+      "step": 351500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.456420734699951e-05,
+      "loss": 1.5031,
+      "step": 352000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.456210738143894e-05,
+      "loss": 1.5164,
+      "step": 352500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.4560011615809496e-05,
+      "loss": 1.5002,
+      "step": 353000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.455791165024894e-05,
+      "loss": 1.5069,
+      "step": 353500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.455581168468837e-05,
+      "loss": 1.551,
+      "step": 354000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.4553711719127804e-05,
+      "loss": 1.5206,
+      "step": 354500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.4551611753567244e-05,
+      "loss": 1.505,
+      "step": 355000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.454951178800668e-05,
+      "loss": 1.4968,
+      "step": 355500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.454741182244611e-05,
+      "loss": 1.5227,
+      "step": 356000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.454531185688555e-05,
+      "loss": 1.5417,
+      "step": 356500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.4543211891324984e-05,
+      "loss": 1.4893,
+      "step": 357000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.454111612569554e-05,
+      "loss": 1.4669,
+      "step": 357500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.453901616013497e-05,
+      "loss": 1.4808,
+      "step": 358000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.453691619457441e-05,
+      "loss": 1.4919,
+      "step": 358500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.4534816229013845e-05,
+      "loss": 1.5133,
+      "step": 359000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.453272466331552e-05,
+      "loss": 1.5113,
+      "step": 359500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.453062889768608e-05,
+      "loss": 1.4929,
+      "step": 360000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.452852893212551e-05,
+      "loss": 1.5048,
+      "step": 360500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.452642896656495e-05,
+      "loss": 1.5149,
+      "step": 361000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.4524329001004386e-05,
+      "loss": 1.523,
+      "step": 361500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.452222903544381e-05,
+      "loss": 1.5208,
+      "step": 362000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.452013326981437e-05,
+      "loss": 1.5038,
+      "step": 362500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.4518033304253814e-05,
+      "loss": 1.5023,
+      "step": 363000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.451593333869325e-05,
+      "loss": 1.507,
+      "step": 363500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.4513833373132674e-05,
+      "loss": 1.508,
+      "step": 364000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.4511733407572114e-05,
+      "loss": 1.506,
+      "step": 364500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.450963344201155e-05,
+      "loss": 1.5085,
+      "step": 365000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.450753767638211e-05,
+      "loss": 1.5063,
+      "step": 365500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.450543771082154e-05,
+      "loss": 1.5293,
+      "step": 366000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.4503337745260975e-05,
+      "loss": 1.4851,
+      "step": 366500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.450123777970041e-05,
+      "loss": 1.5298,
+      "step": 367000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.449913781413985e-05,
+      "loss": 1.4931,
+      "step": 367500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.449703784857928e-05,
+      "loss": 1.5235,
+      "step": 368000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.4494937883018715e-05,
+      "loss": 1.5155,
+      "step": 368500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.4492837917458156e-05,
+      "loss": 1.4876,
+      "step": 369000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.449073795189759e-05,
+      "loss": 1.5209,
+      "step": 369500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.448864218626814e-05,
+      "loss": 1.5165,
+      "step": 370000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.4486542220707576e-05,
+      "loss": 1.5166,
+      "step": 370500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.4484442255147016e-05,
+      "loss": 1.4801,
+      "step": 371000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.448234648951757e-05,
+      "loss": 1.4763,
+      "step": 371500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.4480246523957e-05,
+      "loss": 1.4971,
+      "step": 372000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.447814655839644e-05,
+      "loss": 1.5724,
+      "step": 372500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.447604659283588e-05,
+      "loss": 1.5202,
+      "step": 373000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.447394662727531e-05,
+      "loss": 1.4959,
+      "step": 373500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.4471846661714744e-05,
+      "loss": 1.525,
+      "step": 374000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.4469746696154184e-05,
+      "loss": 1.5232,
+      "step": 374500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.446764673059362e-05,
+      "loss": 1.4812,
+      "step": 375000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.446554676503306e-05,
+      "loss": 1.5136,
+      "step": 375500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.446345099940361e-05,
+      "loss": 1.528,
+      "step": 376000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.4461351033843045e-05,
+      "loss": 1.4843,
+      "step": 376500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.44592552682136e-05,
+      "loss": 1.5356,
+      "step": 377000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.445715530265303e-05,
+      "loss": 1.4876,
+      "step": 377500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.445505533709247e-05,
+      "loss": 1.4893,
+      "step": 378000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.4452955371531906e-05,
+      "loss": 1.4816,
+      "step": 378500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.445085540597134e-05,
+      "loss": 1.4766,
+      "step": 379000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.444875544041078e-05,
+      "loss": 1.5296,
+      "step": 379500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.444665547485021e-05,
+      "loss": 1.4964,
+      "step": 380000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.4444555509289646e-05,
+      "loss": 1.4718,
+      "step": 380500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.4442455543729086e-05,
+      "loss": 1.5037,
+      "step": 381000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.444035977809964e-05,
+      "loss": 1.5162,
+      "step": 381500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.4438259812539073e-05,
+      "loss": 1.468,
+      "step": 382000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.4436159846978514e-05,
+      "loss": 1.5389,
+      "step": 382500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.443405988141795e-05,
+      "loss": 1.4856,
+      "step": 383000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.44319641157885e-05,
+      "loss": 1.518,
+      "step": 383500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.4429864150227934e-05,
+      "loss": 1.5001,
+      "step": 384000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.4427764184667374e-05,
+      "loss": 1.5253,
+      "step": 384500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.442566421910681e-05,
+      "loss": 1.5021,
+      "step": 385000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.442356845347736e-05,
+      "loss": 1.4786,
+      "step": 385500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.4421468487916795e-05,
+      "loss": 1.4512,
+      "step": 386000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.4419368522356235e-05,
+      "loss": 1.4892,
+      "step": 386500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.441726855679567e-05,
+      "loss": 1.4961,
+      "step": 387000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.44151685912351e-05,
+      "loss": 1.48,
+      "step": 387500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.441307282560566e-05,
+      "loss": 1.4948,
+      "step": 388000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.4410972860045096e-05,
+      "loss": 1.4842,
+      "step": 388500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.440887289448453e-05,
+      "loss": 1.4811,
+      "step": 389000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.440677292892397e-05,
+      "loss": 1.4699,
+      "step": 389500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.44046729633634e-05,
+      "loss": 1.5123,
+      "step": 390000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.4402572997802837e-05,
+      "loss": 1.4864,
+      "step": 390500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.440047303224227e-05,
+      "loss": 1.5081,
+      "step": 391000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.4398373066681703e-05,
+      "loss": 1.4964,
+      "step": 391500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.4396277301052264e-05,
+      "loss": 1.5164,
+      "step": 392000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.439418153542282e-05,
+      "loss": 1.4802,
+      "step": 392500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.439208156986225e-05,
+      "loss": 1.5002,
+      "step": 393000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.438998160430169e-05,
+      "loss": 1.5082,
+      "step": 393500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.4387881638741125e-05,
+      "loss": 1.5304,
+      "step": 394000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.438578167318056e-05,
+      "loss": 1.4962,
+      "step": 394500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.438368590755112e-05,
+      "loss": 1.4784,
+      "step": 395000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.438158594199055e-05,
+      "loss": 1.4968,
+      "step": 395500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.4379485976429985e-05,
+      "loss": 1.4957,
+      "step": 396000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.4377386010869425e-05,
+      "loss": 1.5053,
+      "step": 396500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.437528604530886e-05,
+      "loss": 1.5131,
+      "step": 397000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.437318607974829e-05,
+      "loss": 1.4995,
+      "step": 397500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.437108611418773e-05,
+      "loss": 1.4867,
+      "step": 398000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.436898614862716e-05,
+      "loss": 1.4864,
+      "step": 398500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.436688618306659e-05,
+      "loss": 1.485,
+      "step": 399000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.436478621750603e-05,
+      "loss": 1.5123,
+      "step": 399500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.436269045187659e-05,
+      "loss": 1.5177,
+      "step": 400000
+    },
+    {
+      "epoch": 0.24,
+      "eval_loss": 1.4208892583847046,
+      "eval_runtime": 1102.0508,
+      "eval_samples_per_second": 477.945,
+      "eval_steps_per_second": 79.658,
+      "step": 400000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.436059048631602e-05,
+      "loss": 1.4802,
+      "step": 400500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.4358490520755453e-05,
+      "loss": 1.4977,
+      "step": 401000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.4356394755126014e-05,
+      "loss": 1.5022,
+      "step": 401500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.4354294789565454e-05,
+      "loss": 1.4652,
+      "step": 402000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.435219902393601e-05,
+      "loss": 1.4737,
+      "step": 402500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.435009905837544e-05,
+      "loss": 1.5345,
+      "step": 403000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.434799909281488e-05,
+      "loss": 1.5055,
+      "step": 403500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.4345899127254315e-05,
+      "loss": 1.5234,
+      "step": 404000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.434379916169375e-05,
+      "loss": 1.4716,
+      "step": 404500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.434169919613319e-05,
+      "loss": 1.4981,
+      "step": 405000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.4339599230572615e-05,
+      "loss": 1.5184,
+      "step": 405500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.433749926501205e-05,
+      "loss": 1.4997,
+      "step": 406000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.433540349938261e-05,
+      "loss": 1.4904,
+      "step": 406500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.433330353382205e-05,
+      "loss": 1.476,
+      "step": 407000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.433120356826148e-05,
+      "loss": 1.4736,
+      "step": 407500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.432910360270091e-05,
+      "loss": 1.5078,
+      "step": 408000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.432700783707147e-05,
+      "loss": 1.4695,
+      "step": 408500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.432491207144203e-05,
+      "loss": 1.5021,
+      "step": 409000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.4322812105881464e-05,
+      "loss": 1.5098,
+      "step": 409500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.43207121403209e-05,
+      "loss": 1.491,
+      "step": 410000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.431861217476034e-05,
+      "loss": 1.4809,
+      "step": 410500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.431651220919977e-05,
+      "loss": 1.4879,
+      "step": 411000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.4314412243639204e-05,
+      "loss": 1.5112,
+      "step": 411500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.4312312278078644e-05,
+      "loss": 1.5208,
+      "step": 412000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.431021231251807e-05,
+      "loss": 1.4875,
+      "step": 412500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.4308112346957505e-05,
+      "loss": 1.4839,
+      "step": 413000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.4306012381396945e-05,
+      "loss": 1.5076,
+      "step": 413500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.430391241583638e-05,
+      "loss": 1.5046,
+      "step": 414000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.430181245027581e-05,
+      "loss": 1.5053,
+      "step": 414500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.429971248471525e-05,
+      "loss": 1.479,
+      "step": 415000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.4297616719085805e-05,
+      "loss": 1.5168,
+      "step": 415500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.429551675352524e-05,
+      "loss": 1.46,
+      "step": 416000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.429341678796467e-05,
+      "loss": 1.4813,
+      "step": 416500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.429131682240411e-05,
+      "loss": 1.4848,
+      "step": 417000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.4289221056774666e-05,
+      "loss": 1.504,
+      "step": 417500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.42871210912141e-05,
+      "loss": 1.4767,
+      "step": 418000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.428502532558466e-05,
+      "loss": 1.4702,
+      "step": 418500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.42829253600241e-05,
+      "loss": 1.5099,
+      "step": 419000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.428082539446353e-05,
+      "loss": 1.4795,
+      "step": 419500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.427872962883409e-05,
+      "loss": 1.5049,
+      "step": 420000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.427662966327352e-05,
+      "loss": 1.5336,
+      "step": 420500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.427452969771296e-05,
+      "loss": 1.5044,
+      "step": 421000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.4272429732152394e-05,
+      "loss": 1.5018,
+      "step": 421500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.427032976659182e-05,
+      "loss": 1.4736,
+      "step": 422000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.426822980103126e-05,
+      "loss": 1.4819,
+      "step": 422500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.4266129835470695e-05,
+      "loss": 1.4975,
+      "step": 423000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.426402986991013e-05,
+      "loss": 1.497,
+      "step": 423500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.426192990434957e-05,
+      "loss": 1.4882,
+      "step": 424000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.4259829938789e-05,
+      "loss": 1.4715,
+      "step": 424500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.4257729973228435e-05,
+      "loss": 1.4678,
+      "step": 425000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.4255630007667876e-05,
+      "loss": 1.5004,
+      "step": 425500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.425353004210731e-05,
+      "loss": 1.4857,
+      "step": 426000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.425143007654674e-05,
+      "loss": 1.4719,
+      "step": 426500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.424933011098618e-05,
+      "loss": 1.5231,
+      "step": 427000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.424723014542561e-05,
+      "loss": 1.51,
+      "step": 427500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.424513437979617e-05,
+      "loss": 1.4926,
+      "step": 428000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.424303441423561e-05,
+      "loss": 1.4765,
+      "step": 428500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.4240934448675044e-05,
+      "loss": 1.4609,
+      "step": 429000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.42388386830456e-05,
+      "loss": 1.5168,
+      "step": 429500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.423674291741615e-05,
+      "loss": 1.481,
+      "step": 430000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.4234642951855584e-05,
+      "loss": 1.4786,
+      "step": 430500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.4232542986295024e-05,
+      "loss": 1.4847,
+      "step": 431000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.423044302073446e-05,
+      "loss": 1.4862,
+      "step": 431500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.422834305517389e-05,
+      "loss": 1.4737,
+      "step": 432000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.422624308961333e-05,
+      "loss": 1.4638,
+      "step": 432500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.4224143124052765e-05,
+      "loss": 1.4685,
+      "step": 433000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.4222043158492205e-05,
+      "loss": 1.5195,
+      "step": 433500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.421994319293164e-05,
+      "loss": 1.5087,
+      "step": 434000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.421784322737107e-05,
+      "loss": 1.469,
+      "step": 434500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.4215743261810506e-05,
+      "loss": 1.4829,
+      "step": 435000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.421364329624994e-05,
+      "loss": 1.4669,
+      "step": 435500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.421154333068937e-05,
+      "loss": 1.4557,
+      "step": 436000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.420945176499105e-05,
+      "loss": 1.5271,
+      "step": 436500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.4207355999361607e-05,
+      "loss": 1.5119,
+      "step": 437000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.420525603380104e-05,
+      "loss": 1.4962,
+      "step": 437500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.420315606824048e-05,
+      "loss": 1.4752,
+      "step": 438000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.4201056102679914e-05,
+      "loss": 1.4952,
+      "step": 438500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.419895613711935e-05,
+      "loss": 1.5031,
+      "step": 439000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.419685617155879e-05,
+      "loss": 1.5013,
+      "step": 439500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.419476040592934e-05,
+      "loss": 1.4796,
+      "step": 440000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.4192660440368774e-05,
+      "loss": 1.4715,
+      "step": 440500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.4190560474808215e-05,
+      "loss": 1.5148,
+      "step": 441000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.418846050924765e-05,
+      "loss": 1.4945,
+      "step": 441500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.418636054368708e-05,
+      "loss": 1.4679,
+      "step": 442000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.418426057812652e-05,
+      "loss": 1.4923,
+      "step": 442500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.4182160612565955e-05,
+      "loss": 1.4764,
+      "step": 443000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.418006064700539e-05,
+      "loss": 1.4813,
+      "step": 443500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.417796488137594e-05,
+      "loss": 1.5003,
+      "step": 444000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.417586491581538e-05,
+      "loss": 1.4896,
+      "step": 444500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.4173764950254816e-05,
+      "loss": 1.458,
+      "step": 445000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.417166498469425e-05,
+      "loss": 1.4845,
+      "step": 445500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.416956501913369e-05,
+      "loss": 1.4785,
+      "step": 446000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.416746925350424e-05,
+      "loss": 1.4766,
+      "step": 446500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.416536928794368e-05,
+      "loss": 1.4629,
+      "step": 447000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.416327352231423e-05,
+      "loss": 1.4764,
+      "step": 447500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.416117355675367e-05,
+      "loss": 1.4984,
+      "step": 448000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.4159073591193104e-05,
+      "loss": 1.5047,
+      "step": 448500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.415697362563254e-05,
+      "loss": 1.483,
+      "step": 449000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.415487366007198e-05,
+      "loss": 1.4727,
+      "step": 449500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.415277369451141e-05,
+      "loss": 1.466,
+      "step": 450000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.4150673728950845e-05,
+      "loss": 1.4967,
+      "step": 450500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.4148573763390285e-05,
+      "loss": 1.5263,
+      "step": 451000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.414647379782971e-05,
+      "loss": 1.4971,
+      "step": 451500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.4144373832269145e-05,
+      "loss": 1.4895,
+      "step": 452000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.4142278066639705e-05,
+      "loss": 1.4727,
+      "step": 452500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.4140178101079146e-05,
+      "loss": 1.5039,
+      "step": 453000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.41380823354497e-05,
+      "loss": 1.4509,
+      "step": 453500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.413598236988913e-05,
+      "loss": 1.4789,
+      "step": 454000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.413388240432857e-05,
+      "loss": 1.4908,
+      "step": 454500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.4131782438768006e-05,
+      "loss": 1.4847,
+      "step": 455000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.412968247320744e-05,
+      "loss": 1.4833,
+      "step": 455500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.412758250764687e-05,
+      "loss": 1.4864,
+      "step": 456000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.412548254208631e-05,
+      "loss": 1.4836,
+      "step": 456500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.412338257652574e-05,
+      "loss": 1.4851,
+      "step": 457000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.412128261096518e-05,
+      "loss": 1.4735,
+      "step": 457500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.4119182645404614e-05,
+      "loss": 1.4734,
+      "step": 458000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.411708267984405e-05,
+      "loss": 1.4755,
+      "step": 458500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.41149869142146e-05,
+      "loss": 1.4784,
+      "step": 459000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.411288694865404e-05,
+      "loss": 1.4712,
+      "step": 459500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.4110786983093475e-05,
+      "loss": 1.4509,
+      "step": 460000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.410868701753291e-05,
+      "loss": 1.4527,
+      "step": 460500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.410658705197235e-05,
+      "loss": 1.4796,
+      "step": 461000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.41044912863429e-05,
+      "loss": 1.4916,
+      "step": 461500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.4102391320782335e-05,
+      "loss": 1.4592,
+      "step": 462000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.4100291355221776e-05,
+      "loss": 1.5036,
+      "step": 462500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.409819138966121e-05,
+      "loss": 1.4846,
+      "step": 463000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.409609142410064e-05,
+      "loss": 1.4955,
+      "step": 463500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.4093995658471196e-05,
+      "loss": 1.4611,
+      "step": 464000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.4091895692910636e-05,
+      "loss": 1.4841,
+      "step": 464500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.408979572735007e-05,
+      "loss": 1.4605,
+      "step": 465000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.40876957617895e-05,
+      "loss": 1.4769,
+      "step": 465500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.4085595796228943e-05,
+      "loss": 1.4751,
+      "step": 466000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.408349583066838e-05,
+      "loss": 1.485,
+      "step": 466500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.408139586510781e-05,
+      "loss": 1.4905,
+      "step": 467000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.407929589954725e-05,
+      "loss": 1.4479,
+      "step": 467500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.4077195933986684e-05,
+      "loss": 1.4639,
+      "step": 468000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.407509596842612e-05,
+      "loss": 1.5038,
+      "step": 468500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.407300020279667e-05,
+      "loss": 1.4514,
+      "step": 469000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.407090443716723e-05,
+      "loss": 1.4981,
+      "step": 469500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.4068804471606665e-05,
+      "loss": 1.4475,
+      "step": 470000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.40667045060461e-05,
+      "loss": 1.4781,
+      "step": 470500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.406460454048554e-05,
+      "loss": 1.4666,
+      "step": 471000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.406250877485609e-05,
+      "loss": 1.4754,
+      "step": 471500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.406041300922665e-05,
+      "loss": 1.4858,
+      "step": 472000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.4058313043666086e-05,
+      "loss": 1.4444,
+      "step": 472500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.405621307810551e-05,
+      "loss": 1.5036,
+      "step": 473000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.405411311254495e-05,
+      "loss": 1.4703,
+      "step": 473500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.4052013146984386e-05,
+      "loss": 1.4395,
+      "step": 474000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.404991318142382e-05,
+      "loss": 1.458,
+      "step": 474500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.404781321586326e-05,
+      "loss": 1.4904,
+      "step": 475000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.4045713250302694e-05,
+      "loss": 1.4722,
+      "step": 475500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.404361328474213e-05,
+      "loss": 1.4849,
+      "step": 476000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.404151331918157e-05,
+      "loss": 1.4861,
+      "step": 476500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.4039413353621e-05,
+      "loss": 1.4941,
+      "step": 477000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.4037313388060434e-05,
+      "loss": 1.4892,
+      "step": 477500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.4035213422499874e-05,
+      "loss": 1.4417,
+      "step": 478000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.40331134569393e-05,
+      "loss": 1.4741,
+      "step": 478500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.403101349137874e-05,
+      "loss": 1.4821,
+      "step": 479000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.4028913525818175e-05,
+      "loss": 1.4759,
+      "step": 479500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.4026817760188735e-05,
+      "loss": 1.4759,
+      "step": 480000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.402472199455929e-05,
+      "loss": 1.4996,
+      "step": 480500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.402262202899872e-05,
+      "loss": 1.4596,
+      "step": 481000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.402052206343816e-05,
+      "loss": 1.4686,
+      "step": 481500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.4018422097877596e-05,
+      "loss": 1.4766,
+      "step": 482000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.401632633224815e-05,
+      "loss": 1.4807,
+      "step": 482500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.401422636668758e-05,
+      "loss": 1.4772,
+      "step": 483000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.401212640112702e-05,
+      "loss": 1.4749,
+      "step": 483500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.4010026435566457e-05,
+      "loss": 1.4796,
+      "step": 484000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.400792647000589e-05,
+      "loss": 1.4645,
+      "step": 484500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.400582650444533e-05,
+      "loss": 1.486,
+      "step": 485000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.400372653888476e-05,
+      "loss": 1.4682,
+      "step": 485500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.40016265733242e-05,
+      "loss": 1.4471,
+      "step": 486000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.399952660776363e-05,
+      "loss": 1.4638,
+      "step": 486500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.399743084213419e-05,
+      "loss": 1.4751,
+      "step": 487000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.3995330876573624e-05,
+      "loss": 1.4749,
+      "step": 487500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.399323511094418e-05,
+      "loss": 1.4473,
+      "step": 488000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.399113514538362e-05,
+      "loss": 1.4827,
+      "step": 488500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.398903517982305e-05,
+      "loss": 1.4956,
+      "step": 489000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.3986935214262485e-05,
+      "loss": 1.4725,
+      "step": 489500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.3984835248701925e-05,
+      "loss": 1.4544,
+      "step": 490000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.398273948307248e-05,
+      "loss": 1.4674,
+      "step": 490500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.398063951751191e-05,
+      "loss": 1.48,
+      "step": 491000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.3978539551951346e-05,
+      "loss": 1.4429,
+      "step": 491500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.3976439586390786e-05,
+      "loss": 1.4687,
+      "step": 492000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.397433962083021e-05,
+      "loss": 1.4786,
+      "step": 492500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.397223965526965e-05,
+      "loss": 1.4867,
+      "step": 493000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.3970139689709086e-05,
+      "loss": 1.5085,
+      "step": 493500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.396803972414852e-05,
+      "loss": 1.4532,
+      "step": 494000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.396594395851908e-05,
+      "loss": 1.4871,
+      "step": 494500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.3963843992958514e-05,
+      "loss": 1.4505,
+      "step": 495000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.3961748227329074e-05,
+      "loss": 1.485,
+      "step": 495500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.395964826176851e-05,
+      "loss": 1.4968,
+      "step": 496000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.395754829620794e-05,
+      "loss": 1.4476,
+      "step": 496500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.395544833064738e-05,
+      "loss": 1.4873,
+      "step": 497000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.395334836508681e-05,
+      "loss": 1.4653,
+      "step": 497500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.395124839952624e-05,
+      "loss": 1.4962,
+      "step": 498000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.394914843396568e-05,
+      "loss": 1.4837,
+      "step": 498500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.3947048468405115e-05,
+      "loss": 1.4668,
+      "step": 499000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.3944956902706796e-05,
+      "loss": 1.4675,
+      "step": 499500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.394285693714623e-05,
+      "loss": 1.4509,
+      "step": 500000
+    },
+    {
+      "epoch": 0.3,
+      "eval_loss": 1.39423406124115,
+      "eval_runtime": 1102.7794,
+      "eval_samples_per_second": 477.63,
+      "eval_steps_per_second": 79.605,
+      "step": 500000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.394075697158567e-05,
+      "loss": 1.4921,
+      "step": 500500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.39386570060251e-05,
+      "loss": 1.4708,
+      "step": 501000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.3936557040464536e-05,
+      "loss": 1.4328,
+      "step": 501500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.3934457074903976e-05,
+      "loss": 1.4618,
+      "step": 502000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.393236130927453e-05,
+      "loss": 1.469,
+      "step": 502500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.3930261343713963e-05,
+      "loss": 1.4599,
+      "step": 503000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.39281613781534e-05,
+      "loss": 1.4738,
+      "step": 503500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.392606141259284e-05,
+      "loss": 1.4419,
+      "step": 504000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.3923961447032264e-05,
+      "loss": 1.481,
+      "step": 504500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.39218614814717e-05,
+      "loss": 1.4842,
+      "step": 505000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.391976151591114e-05,
+      "loss": 1.4802,
+      "step": 505500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.391766155035057e-05,
+      "loss": 1.4596,
+      "step": 506000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.391556578472113e-05,
+      "loss": 1.4874,
+      "step": 506500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.3913465819160565e-05,
+      "loss": 1.4527,
+      "step": 507000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.39113658536e-05,
+      "loss": 1.483,
+      "step": 507500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.390926588803943e-05,
+      "loss": 1.4635,
+      "step": 508000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.390716592247887e-05,
+      "loss": 1.4686,
+      "step": 508500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.3905065956918305e-05,
+      "loss": 1.4861,
+      "step": 509000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.390296599135774e-05,
+      "loss": 1.4563,
+      "step": 509500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.390087022572829e-05,
+      "loss": 1.4874,
+      "step": 510000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.389877026016773e-05,
+      "loss": 1.4642,
+      "step": 510500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.3896670294607166e-05,
+      "loss": 1.4704,
+      "step": 511000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.38945703290466e-05,
+      "loss": 1.4708,
+      "step": 511500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.389247456341716e-05,
+      "loss": 1.464,
+      "step": 512000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.389037879778772e-05,
+      "loss": 1.459,
+      "step": 512500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.3888278832227154e-05,
+      "loss": 1.4629,
+      "step": 513000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.388618306659771e-05,
+      "loss": 1.4731,
+      "step": 513500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.388408310103714e-05,
+      "loss": 1.4593,
+      "step": 514000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.388198313547658e-05,
+      "loss": 1.4806,
+      "step": 514500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.3879883169916014e-05,
+      "loss": 1.4702,
+      "step": 515000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.387778320435545e-05,
+      "loss": 1.4674,
+      "step": 515500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.387568323879489e-05,
+      "loss": 1.4989,
+      "step": 516000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.3873583273234315e-05,
+      "loss": 1.4581,
+      "step": 516500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.387148330767375e-05,
+      "loss": 1.4312,
+      "step": 517000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.386938334211319e-05,
+      "loss": 1.4704,
+      "step": 517500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.386728337655262e-05,
+      "loss": 1.474,
+      "step": 518000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.3865183410992055e-05,
+      "loss": 1.4743,
+      "step": 518500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.3863083445431496e-05,
+      "loss": 1.4557,
+      "step": 519000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.386098347987093e-05,
+      "loss": 1.4193,
+      "step": 519500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.385888771424148e-05,
+      "loss": 1.4183,
+      "step": 520000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.385678774868092e-05,
+      "loss": 1.4506,
+      "step": 520500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.385469198305148e-05,
+      "loss": 1.4805,
+      "step": 521000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.385259201749091e-05,
+      "loss": 1.4719,
+      "step": 521500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.3850492051930343e-05,
+      "loss": 1.4597,
+      "step": 522000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.3848392086369784e-05,
+      "loss": 1.4465,
+      "step": 522500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.3846296320740344e-05,
+      "loss": 1.4688,
+      "step": 523000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.384419635517977e-05,
+      "loss": 1.4534,
+      "step": 523500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.3842096389619204e-05,
+      "loss": 1.4553,
+      "step": 524000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.3839996424058644e-05,
+      "loss": 1.4524,
+      "step": 524500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.383789645849808e-05,
+      "loss": 1.4509,
+      "step": 525000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.383579649293751e-05,
+      "loss": 1.5044,
+      "step": 525500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.383369652737695e-05,
+      "loss": 1.5006,
+      "step": 526000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.3831600761747505e-05,
+      "loss": 1.4834,
+      "step": 526500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.382950079618694e-05,
+      "loss": 1.4554,
+      "step": 527000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.382740083062638e-05,
+      "loss": 1.4694,
+      "step": 527500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.382530086506581e-05,
+      "loss": 1.4621,
+      "step": 528000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.3823200899505246e-05,
+      "loss": 1.4831,
+      "step": 528500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.3821100933944686e-05,
+      "loss": 1.4699,
+      "step": 529000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.381900096838412e-05,
+      "loss": 1.4635,
+      "step": 529500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.381690100282355e-05,
+      "loss": 1.4493,
+      "step": 530000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.381480103726299e-05,
+      "loss": 1.4588,
+      "step": 530500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.3812701071702427e-05,
+      "loss": 1.4337,
+      "step": 531000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.381060110614185e-05,
+      "loss": 1.4663,
+      "step": 531500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.3808505340512414e-05,
+      "loss": 1.4553,
+      "step": 532000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.3806405374951854e-05,
+      "loss": 1.4686,
+      "step": 532500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.380430540939129e-05,
+      "loss": 1.4886,
+      "step": 533000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.380220544383072e-05,
+      "loss": 1.4579,
+      "step": 533500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.3800105478270154e-05,
+      "loss": 1.4527,
+      "step": 534000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.379800551270959e-05,
+      "loss": 1.4672,
+      "step": 534500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.379590554714902e-05,
+      "loss": 1.4291,
+      "step": 535000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.379380978151958e-05,
+      "loss": 1.477,
+      "step": 535500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.379170981595902e-05,
+      "loss": 1.4374,
+      "step": 536000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.378960985039845e-05,
+      "loss": 1.4743,
+      "step": 536500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.378750988483789e-05,
+      "loss": 1.4532,
+      "step": 537000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.378540991927732e-05,
+      "loss": 1.4681,
+      "step": 537500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.3783309953716756e-05,
+      "loss": 1.4418,
+      "step": 538000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.3781214188087316e-05,
+      "loss": 1.4785,
+      "step": 538500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.377911422252675e-05,
+      "loss": 1.4813,
+      "step": 539000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.377701845689731e-05,
+      "loss": 1.4728,
+      "step": 539500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.377491849133674e-05,
+      "loss": 1.4488,
+      "step": 540000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.377281852577618e-05,
+      "loss": 1.4475,
+      "step": 540500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.377071856021561e-05,
+      "loss": 1.4984,
+      "step": 541000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.3768618594655044e-05,
+      "loss": 1.4632,
+      "step": 541500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.376651862909448e-05,
+      "loss": 1.418,
+      "step": 542000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.376441866353392e-05,
+      "loss": 1.4526,
+      "step": 542500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.376231869797335e-05,
+      "loss": 1.4693,
+      "step": 543000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.3760218732412784e-05,
+      "loss": 1.4522,
+      "step": 543500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.3758118766852224e-05,
+      "loss": 1.4922,
+      "step": 544000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.375601880129166e-05,
+      "loss": 1.433,
+      "step": 544500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.375391883573109e-05,
+      "loss": 1.4408,
+      "step": 545000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.375181887017053e-05,
+      "loss": 1.4421,
+      "step": 545500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.3749718904609965e-05,
+      "loss": 1.468,
+      "step": 546000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.37476189390494e-05,
+      "loss": 1.4545,
+      "step": 546500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.374552737335107e-05,
+      "loss": 1.4541,
+      "step": 547000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.374342740779051e-05,
+      "loss": 1.4912,
+      "step": 547500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.3741327442229946e-05,
+      "loss": 1.4484,
+      "step": 548000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.373922747666938e-05,
+      "loss": 1.4795,
+      "step": 548500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.373712751110882e-05,
+      "loss": 1.45,
+      "step": 549000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.373502754554825e-05,
+      "loss": 1.4788,
+      "step": 549500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.3732927579987687e-05,
+      "loss": 1.4774,
+      "step": 550000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.373082761442713e-05,
+      "loss": 1.4913,
+      "step": 550500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.372872764886656e-05,
+      "loss": 1.436,
+      "step": 551000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.3726636083168234e-05,
+      "loss": 1.4616,
+      "step": 551500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.372453611760767e-05,
+      "loss": 1.4495,
+      "step": 552000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.372243615204711e-05,
+      "loss": 1.4402,
+      "step": 552500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.372033618648654e-05,
+      "loss": 1.4483,
+      "step": 553000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.3718236220925975e-05,
+      "loss": 1.4489,
+      "step": 553500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.3716136255365415e-05,
+      "loss": 1.4528,
+      "step": 554000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.371403628980485e-05,
+      "loss": 1.4725,
+      "step": 554500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.371193632424428e-05,
+      "loss": 1.4451,
+      "step": 555000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.370983635868372e-05,
+      "loss": 1.4319,
+      "step": 555500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.370774479298539e-05,
+      "loss": 1.4718,
+      "step": 556000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.370564482742483e-05,
+      "loss": 1.4565,
+      "step": 556500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.370354486186426e-05,
+      "loss": 1.4427,
+      "step": 557000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.3701444896303696e-05,
+      "loss": 1.4279,
+      "step": 557500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.3699344930743136e-05,
+      "loss": 1.4681,
+      "step": 558000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.369724916511369e-05,
+      "loss": 1.4486,
+      "step": 558500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.369514919955312e-05,
+      "loss": 1.4583,
+      "step": 559000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.3693049233992563e-05,
+      "loss": 1.4572,
+      "step": 559500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.3690949268432e-05,
+      "loss": 1.4543,
+      "step": 560000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.368884930287143e-05,
+      "loss": 1.453,
+      "step": 560500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.3686753537241984e-05,
+      "loss": 1.4382,
+      "step": 561000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.3684653571681424e-05,
+      "loss": 1.4591,
+      "step": 561500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.3682557806051985e-05,
+      "loss": 1.4615,
+      "step": 562000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.368045784049141e-05,
+      "loss": 1.4834,
+      "step": 562500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.3678357874930845e-05,
+      "loss": 1.4664,
+      "step": 563000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.3676257909370285e-05,
+      "loss": 1.4678,
+      "step": 563500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.367415794380972e-05,
+      "loss": 1.4438,
+      "step": 564000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.367205797824916e-05,
+      "loss": 1.464,
+      "step": 564500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.366995801268859e-05,
+      "loss": 1.4623,
+      "step": 565000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.3667858047128026e-05,
+      "loss": 1.4471,
+      "step": 565500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.3665758081567466e-05,
+      "loss": 1.4515,
+      "step": 566000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.36636581160069e-05,
+      "loss": 1.4244,
+      "step": 566500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.366155815044633e-05,
+      "loss": 1.4558,
+      "step": 567000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.365945818488577e-05,
+      "loss": 1.4595,
+      "step": 567500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.36573582193252e-05,
+      "loss": 1.4305,
+      "step": 568000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.365526245369576e-05,
+      "loss": 1.4406,
+      "step": 568500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.3653162488135193e-05,
+      "loss": 1.462,
+      "step": 569000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.3651062522574634e-05,
+      "loss": 1.4595,
+      "step": 569500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.364896255701407e-05,
+      "loss": 1.4909,
+      "step": 570000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.364687099131574e-05,
+      "loss": 1.4719,
+      "step": 570500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.3644771025755174e-05,
+      "loss": 1.4807,
+      "step": 571000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.3642671060194614e-05,
+      "loss": 1.4544,
+      "step": 571500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.364057109463405e-05,
+      "loss": 1.4577,
+      "step": 572000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.363847112907348e-05,
+      "loss": 1.4659,
+      "step": 572500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.363637116351292e-05,
+      "loss": 1.454,
+      "step": 573000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.3634271197952355e-05,
+      "loss": 1.4702,
+      "step": 573500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.363217543232291e-05,
+      "loss": 1.4803,
+      "step": 574000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.363007546676234e-05,
+      "loss": 1.4818,
+      "step": 574500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.362797550120178e-05,
+      "loss": 1.4441,
+      "step": 575000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.3625875535641216e-05,
+      "loss": 1.4358,
+      "step": 575500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.362377557008065e-05,
+      "loss": 1.4538,
+      "step": 576000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.362167560452009e-05,
+      "loss": 1.469,
+      "step": 576500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.361957563895952e-05,
+      "loss": 1.4577,
+      "step": 577000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.361747567339895e-05,
+      "loss": 1.4498,
+      "step": 577500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.361537570783839e-05,
+      "loss": 1.4132,
+      "step": 578000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.361327994220895e-05,
+      "loss": 1.4081,
+      "step": 578500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.3611179976648384e-05,
+      "loss": 1.4546,
+      "step": 579000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.360908001108782e-05,
+      "loss": 1.4465,
+      "step": 579500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.360698004552725e-05,
+      "loss": 1.4338,
+      "step": 580000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.3604880079966684e-05,
+      "loss": 1.4582,
+      "step": 580500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.3602784314337244e-05,
+      "loss": 1.4748,
+      "step": 581000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.3600684348776685e-05,
+      "loss": 1.4371,
+      "step": 581500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.359858438321612e-05,
+      "loss": 1.4423,
+      "step": 582000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.359648861758667e-05,
+      "loss": 1.4646,
+      "step": 582500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.3594388652026105e-05,
+      "loss": 1.4183,
+      "step": 583000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.3592288686465545e-05,
+      "loss": 1.4732,
+      "step": 583500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.359018872090498e-05,
+      "loss": 1.4349,
+      "step": 584000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.358808875534441e-05,
+      "loss": 1.4537,
+      "step": 584500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.3585992989714966e-05,
+      "loss": 1.4631,
+      "step": 585000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.3583893024154406e-05,
+      "loss": 1.4499,
+      "step": 585500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.358179305859384e-05,
+      "loss": 1.4533,
+      "step": 586000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.357969309303327e-05,
+      "loss": 1.4466,
+      "step": 586500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.3577593127472707e-05,
+      "loss": 1.4415,
+      "step": 587000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.357549316191214e-05,
+      "loss": 1.4763,
+      "step": 587500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.35733973962827e-05,
+      "loss": 1.4567,
+      "step": 588000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.357129743072214e-05,
+      "loss": 1.4361,
+      "step": 588500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.3569197465161574e-05,
+      "loss": 1.4296,
+      "step": 589000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.3567097499601e-05,
+      "loss": 1.496,
+      "step": 589500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.356499753404044e-05,
+      "loss": 1.4276,
+      "step": 590000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.3562897568479874e-05,
+      "loss": 1.4369,
+      "step": 590500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.356079760291931e-05,
+      "loss": 1.4548,
+      "step": 591000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.355869763735875e-05,
+      "loss": 1.4422,
+      "step": 591500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.35566018717293e-05,
+      "loss": 1.4806,
+      "step": 592000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.3554501906168735e-05,
+      "loss": 1.4277,
+      "step": 592500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.355240194060817e-05,
+      "loss": 1.4337,
+      "step": 593000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.355030197504761e-05,
+      "loss": 1.4104,
+      "step": 593500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.354820620941817e-05,
+      "loss": 1.467,
+      "step": 594000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.354611044378872e-05,
+      "loss": 1.4678,
+      "step": 594500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.3544010478228156e-05,
+      "loss": 1.4167,
+      "step": 595000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.3541910512667596e-05,
+      "loss": 1.4224,
+      "step": 595500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.353981054710703e-05,
+      "loss": 1.4826,
+      "step": 596000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.3537710581546457e-05,
+      "loss": 1.4465,
+      "step": 596500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.35356106159859e-05,
+      "loss": 1.4431,
+      "step": 597000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.353351065042533e-05,
+      "loss": 1.4756,
+      "step": 597500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.3531410684864764e-05,
+      "loss": 1.4438,
+      "step": 598000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.3529314919235324e-05,
+      "loss": 1.45,
+      "step": 598500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.352721495367476e-05,
+      "loss": 1.445,
+      "step": 599000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.352511498811419e-05,
+      "loss": 1.4542,
+      "step": 599500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.3523015022553624e-05,
+      "loss": 1.4337,
+      "step": 600000
+    },
+    {
+      "epoch": 0.36,
+      "eval_loss": 1.3768744468688965,
+      "eval_runtime": 1111.0659,
+      "eval_samples_per_second": 474.067,
+      "eval_steps_per_second": 79.012,
+      "step": 600000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.3520915056993065e-05,
+      "loss": 1.4555,
+      "step": 600500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.35188150914325e-05,
+      "loss": 1.4661,
+      "step": 601000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.351671932580305e-05,
+      "loss": 1.4178,
+      "step": 601500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.351461936024249e-05,
+      "loss": 1.4682,
+      "step": 602000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.351252359461305e-05,
+      "loss": 1.427,
+      "step": 602500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.3510423629052486e-05,
+      "loss": 1.4369,
+      "step": 603000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.350832786342304e-05,
+      "loss": 1.4255,
+      "step": 603500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.350622789786247e-05,
+      "loss": 1.4287,
+      "step": 604000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.350412793230191e-05,
+      "loss": 1.4319,
+      "step": 604500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.3502027966741346e-05,
+      "loss": 1.4201,
+      "step": 605000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.349992800118078e-05,
+      "loss": 1.4515,
+      "step": 605500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.349782803562022e-05,
+      "loss": 1.4629,
+      "step": 606000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.349572807005965e-05,
+      "loss": 1.4473,
+      "step": 606500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.349362810449908e-05,
+      "loss": 1.4352,
+      "step": 607000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.349153233886964e-05,
+      "loss": 1.471,
+      "step": 607500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.348943237330908e-05,
+      "loss": 1.4547,
+      "step": 608000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.348733240774851e-05,
+      "loss": 1.4612,
+      "step": 608500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.348523244218795e-05,
+      "loss": 1.4406,
+      "step": 609000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.348313247662738e-05,
+      "loss": 1.4493,
+      "step": 609500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.3481032511066815e-05,
+      "loss": 1.451,
+      "step": 610000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.3478932545506255e-05,
+      "loss": 1.4441,
+      "step": 610500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.347683257994569e-05,
+      "loss": 1.4575,
+      "step": 611000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.347473261438512e-05,
+      "loss": 1.4635,
+      "step": 611500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.34726410486868e-05,
+      "loss": 1.4496,
+      "step": 612000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.3470541083126236e-05,
+      "loss": 1.469,
+      "step": 612500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.3468441117565676e-05,
+      "loss": 1.448,
+      "step": 613000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.34663411520051e-05,
+      "loss": 1.4622,
+      "step": 613500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.3464241186444536e-05,
+      "loss": 1.4492,
+      "step": 614000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.3462145420815097e-05,
+      "loss": 1.4249,
+      "step": 614500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.346004545525454e-05,
+      "loss": 1.4879,
+      "step": 615000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.345794548969397e-05,
+      "loss": 1.4732,
+      "step": 615500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.3455845524133404e-05,
+      "loss": 1.4502,
+      "step": 616000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.345374555857284e-05,
+      "loss": 1.423,
+      "step": 616500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.345164559301227e-05,
+      "loss": 1.4802,
+      "step": 617000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.344954562745171e-05,
+      "loss": 1.4899,
+      "step": 617500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.3447449861822264e-05,
+      "loss": 1.4651,
+      "step": 618000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.34453498962617e-05,
+      "loss": 1.4203,
+      "step": 618500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.344324993070113e-05,
+      "loss": 1.4236,
+      "step": 619000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.344115416507169e-05,
+      "loss": 1.4513,
+      "step": 619500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.343905419951113e-05,
+      "loss": 1.4665,
+      "step": 620000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.343695423395056e-05,
+      "loss": 1.4464,
+      "step": 620500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.343485426838999e-05,
+      "loss": 1.43,
+      "step": 621000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.343275430282943e-05,
+      "loss": 1.4221,
+      "step": 621500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.3430654337268866e-05,
+      "loss": 1.4269,
+      "step": 622000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.34285543717083e-05,
+      "loss": 1.4491,
+      "step": 622500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.342645440614774e-05,
+      "loss": 1.425,
+      "step": 623000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.342435444058717e-05,
+      "loss": 1.4478,
+      "step": 623500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.3422258674957727e-05,
+      "loss": 1.4586,
+      "step": 624000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.342015870939717e-05,
+      "loss": 1.448,
+      "step": 624500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.34180587438366e-05,
+      "loss": 1.4555,
+      "step": 625000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.3415958778276034e-05,
+      "loss": 1.4411,
+      "step": 625500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.3413858812715474e-05,
+      "loss": 1.4469,
+      "step": 626000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.341175884715491e-05,
+      "loss": 1.4492,
+      "step": 626500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.340965888159434e-05,
+      "loss": 1.4286,
+      "step": 627000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.340755891603378e-05,
+      "loss": 1.4145,
+      "step": 627500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.3405463150404335e-05,
+      "loss": 1.4191,
+      "step": 628000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.340336318484377e-05,
+      "loss": 1.4256,
+      "step": 628500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.34012632192832e-05,
+      "loss": 1.4367,
+      "step": 629000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.3399167453653755e-05,
+      "loss": 1.4325,
+      "step": 629500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.3397067488093195e-05,
+      "loss": 1.4389,
+      "step": 630000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.339496752253263e-05,
+      "loss": 1.452,
+      "step": 630500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.339286755697207e-05,
+      "loss": 1.4202,
+      "step": 631000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.339077179134262e-05,
+      "loss": 1.4379,
+      "step": 631500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.3388671825782056e-05,
+      "loss": 1.4422,
+      "step": 632000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.338657186022149e-05,
+      "loss": 1.4148,
+      "step": 632500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.338447189466093e-05,
+      "loss": 1.4406,
+      "step": 633000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.338237192910036e-05,
+      "loss": 1.4397,
+      "step": 633500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.33802719635398e-05,
+      "loss": 1.4582,
+      "step": 634000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.337817199797924e-05,
+      "loss": 1.4508,
+      "step": 634500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.337607203241867e-05,
+      "loss": 1.4748,
+      "step": 635000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.33739720668581e-05,
+      "loss": 1.4191,
+      "step": 635500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.337187210129754e-05,
+      "loss": 1.4539,
+      "step": 636000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.33697763356681e-05,
+      "loss": 1.4653,
+      "step": 636500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.336767637010753e-05,
+      "loss": 1.4614,
+      "step": 637000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.3365580604478085e-05,
+      "loss": 1.4223,
+      "step": 637500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.3363480638917525e-05,
+      "loss": 1.448,
+      "step": 638000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.336138067335696e-05,
+      "loss": 1.447,
+      "step": 638500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.335928070779639e-05,
+      "loss": 1.4191,
+      "step": 639000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.335718074223583e-05,
+      "loss": 1.4401,
+      "step": 639500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.3355080776675266e-05,
+      "loss": 1.4308,
+      "step": 640000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.335298081111469e-05,
+      "loss": 1.4421,
+      "step": 640500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.335088504548525e-05,
+      "loss": 1.4242,
+      "step": 641000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.3348789279855806e-05,
+      "loss": 1.4478,
+      "step": 641500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.3346689314295246e-05,
+      "loss": 1.4296,
+      "step": 642000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.334458934873468e-05,
+      "loss": 1.4458,
+      "step": 642500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.334248938317411e-05,
+      "loss": 1.4488,
+      "step": 643000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.3340389417613554e-05,
+      "loss": 1.4374,
+      "step": 643500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.333828945205299e-05,
+      "loss": 1.4245,
+      "step": 644000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.333618948649242e-05,
+      "loss": 1.4593,
+      "step": 644500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.3334089520931854e-05,
+      "loss": 1.431,
+      "step": 645000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.333198955537129e-05,
+      "loss": 1.4644,
+      "step": 645500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.332988958981073e-05,
+      "loss": 1.4543,
+      "step": 646000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.332778962425016e-05,
+      "loss": 1.442,
+      "step": 646500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.3325689658689595e-05,
+      "loss": 1.4349,
+      "step": 647000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.332359389306015e-05,
+      "loss": 1.4316,
+      "step": 647500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.332149392749959e-05,
+      "loss": 1.4708,
+      "step": 648000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.331939816187015e-05,
+      "loss": 1.4376,
+      "step": 648500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.331729819630958e-05,
+      "loss": 1.4279,
+      "step": 649000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.3315198230749016e-05,
+      "loss": 1.4284,
+      "step": 649500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.331309826518845e-05,
+      "loss": 1.4341,
+      "step": 650000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.331099829962788e-05,
+      "loss": 1.4804,
+      "step": 650500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.3308898334067316e-05,
+      "loss": 1.4413,
+      "step": 651000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.3306798368506756e-05,
+      "loss": 1.4023,
+      "step": 651500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.330469840294619e-05,
+      "loss": 1.43,
+      "step": 652000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.330260263731674e-05,
+      "loss": 1.4476,
+      "step": 652500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.3300506871687304e-05,
+      "loss": 1.4248,
+      "step": 653000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.3298406906126744e-05,
+      "loss": 1.4464,
+      "step": 653500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.329630694056618e-05,
+      "loss": 1.4334,
+      "step": 654000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.3294206975005604e-05,
+      "loss": 1.4424,
+      "step": 654500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.3292107009445044e-05,
+      "loss": 1.4482,
+      "step": 655000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.329000704388448e-05,
+      "loss": 1.4251,
+      "step": 655500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.328791127825504e-05,
+      "loss": 1.444,
+      "step": 656000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.328581131269447e-05,
+      "loss": 1.4468,
+      "step": 656500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.3283711347133905e-05,
+      "loss": 1.4211,
+      "step": 657000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.328161138157334e-05,
+      "loss": 1.4287,
+      "step": 657500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.327951141601277e-05,
+      "loss": 1.432,
+      "step": 658000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.327741145045221e-05,
+      "loss": 1.4243,
+      "step": 658500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.3275311484891646e-05,
+      "loss": 1.4279,
+      "step": 659000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.327321151933108e-05,
+      "loss": 1.471,
+      "step": 659500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.327111155377052e-05,
+      "loss": 1.4446,
+      "step": 660000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.326901578814107e-05,
+      "loss": 1.4615,
+      "step": 660500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.3266915822580506e-05,
+      "loss": 1.4471,
+      "step": 661000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.3264815857019947e-05,
+      "loss": 1.4619,
+      "step": 661500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.326271589145938e-05,
+      "loss": 1.4318,
+      "step": 662000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.3260615925898813e-05,
+      "loss": 1.4415,
+      "step": 662500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.3258515960338254e-05,
+      "loss": 1.4262,
+      "step": 663000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.325641599477769e-05,
+      "loss": 1.4446,
+      "step": 663500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.325431602921712e-05,
+      "loss": 1.4169,
+      "step": 664000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.3252220263587674e-05,
+      "loss": 1.4126,
+      "step": 664500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.3250120298027114e-05,
+      "loss": 1.4523,
+      "step": 665000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.324802033246655e-05,
+      "loss": 1.4505,
+      "step": 665500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.32459245668371e-05,
+      "loss": 1.4749,
+      "step": 666000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.3243824601276535e-05,
+      "loss": 1.4397,
+      "step": 666500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.3241724635715975e-05,
+      "loss": 1.4256,
+      "step": 667000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.323962467015541e-05,
+      "loss": 1.4611,
+      "step": 667500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.323752890452596e-05,
+      "loss": 1.4506,
+      "step": 668000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.32354289389654e-05,
+      "loss": 1.456,
+      "step": 668500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.3233337373267076e-05,
+      "loss": 1.4112,
+      "step": 669000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.3231237407706516e-05,
+      "loss": 1.4675,
+      "step": 669500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.322913744214595e-05,
+      "loss": 1.4507,
+      "step": 670000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.322703747658538e-05,
+      "loss": 1.4317,
+      "step": 670500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.3224937511024823e-05,
+      "loss": 1.4341,
+      "step": 671000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.322283754546425e-05,
+      "loss": 1.4129,
+      "step": 671500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.3220737579903684e-05,
+      "loss": 1.4284,
+      "step": 672000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.3218637614343124e-05,
+      "loss": 1.4382,
+      "step": 672500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.321653764878256e-05,
+      "loss": 1.4176,
+      "step": 673000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.321443768322199e-05,
+      "loss": 1.4286,
+      "step": 673500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.321233771766143e-05,
+      "loss": 1.4349,
+      "step": 674000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.3210237752100864e-05,
+      "loss": 1.4427,
+      "step": 674500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.32081377865403e-05,
+      "loss": 1.4347,
+      "step": 675000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.320604202091086e-05,
+      "loss": 1.4143,
+      "step": 675500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.320394625528141e-05,
+      "loss": 1.4515,
+      "step": 676000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.3201846289720845e-05,
+      "loss": 1.4268,
+      "step": 676500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.319974632416028e-05,
+      "loss": 1.4481,
+      "step": 677000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.319764635859972e-05,
+      "loss": 1.4725,
+      "step": 677500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.319554639303915e-05,
+      "loss": 1.4098,
+      "step": 678000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.3193446427478586e-05,
+      "loss": 1.4437,
+      "step": 678500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.3191346461918026e-05,
+      "loss": 1.3956,
+      "step": 679000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.318925069628858e-05,
+      "loss": 1.4491,
+      "step": 679500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.318715493065914e-05,
+      "loss": 1.4341,
+      "step": 680000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.3185054965098574e-05,
+      "loss": 1.4301,
+      "step": 680500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.318295499953801e-05,
+      "loss": 1.3966,
+      "step": 681000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.318085503397744e-05,
+      "loss": 1.4268,
+      "step": 681500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.3178755068416874e-05,
+      "loss": 1.4356,
+      "step": 682000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.3176655102856314e-05,
+      "loss": 1.4598,
+      "step": 682500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.317455513729575e-05,
+      "loss": 1.4122,
+      "step": 683000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.317245517173518e-05,
+      "loss": 1.4324,
+      "step": 683500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.317035520617462e-05,
+      "loss": 1.4231,
+      "step": 684000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.3168259440545175e-05,
+      "loss": 1.3964,
+      "step": 684500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.316615947498461e-05,
+      "loss": 1.4346,
+      "step": 685000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.316405950942404e-05,
+      "loss": 1.4072,
+      "step": 685500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.316195954386348e-05,
+      "loss": 1.4381,
+      "step": 686000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.3159859578302915e-05,
+      "loss": 1.4389,
+      "step": 686500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.315775961274235e-05,
+      "loss": 1.4436,
+      "step": 687000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.315565964718179e-05,
+      "loss": 1.4558,
+      "step": 687500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.315355968162122e-05,
+      "loss": 1.4319,
+      "step": 688000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.3151459716060656e-05,
+      "loss": 1.4506,
+      "step": 688500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.3149363950431216e-05,
+      "loss": 1.4629,
+      "step": 689000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.314726398487065e-05,
+      "loss": 1.4172,
+      "step": 689500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.314516401931008e-05,
+      "loss": 1.4572,
+      "step": 690000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.3143064053749524e-05,
+      "loss": 1.4266,
+      "step": 690500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.314096408818895e-05,
+      "loss": 1.4258,
+      "step": 691000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.3138864122628384e-05,
+      "loss": 1.4577,
+      "step": 691500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.3136764157067824e-05,
+      "loss": 1.4386,
+      "step": 692000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.313466419150726e-05,
+      "loss": 1.4349,
+      "step": 692500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.313257262580894e-05,
+      "loss": 1.4594,
+      "step": 693000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.313047266024837e-05,
+      "loss": 1.4586,
+      "step": 693500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.3128372694687805e-05,
+      "loss": 1.4089,
+      "step": 694000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.3126272729127245e-05,
+      "loss": 1.4087,
+      "step": 694500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.312417276356668e-05,
+      "loss": 1.4497,
+      "step": 695000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.312207279800611e-05,
+      "loss": 1.4195,
+      "step": 695500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.3119972832445545e-05,
+      "loss": 1.3999,
+      "step": 696000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.311787286688498e-05,
+      "loss": 1.429,
+      "step": 696500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.311577710125554e-05,
+      "loss": 1.4587,
+      "step": 697000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.311368133562609e-05,
+      "loss": 1.437,
+      "step": 697500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.311158137006553e-05,
+      "loss": 1.4342,
+      "step": 698000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.3109481404504967e-05,
+      "loss": 1.4075,
+      "step": 698500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.31073814389444e-05,
+      "loss": 1.4341,
+      "step": 699000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.3105285673314954e-05,
+      "loss": 1.4322,
+      "step": 699500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.3103185707754394e-05,
+      "loss": 1.4627,
+      "step": 700000
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 1.3590091466903687,
+      "eval_runtime": 1101.3908,
+      "eval_samples_per_second": 478.232,
+      "eval_steps_per_second": 79.706,
+      "step": 700000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.310108574219383e-05,
+      "loss": 1.408,
+      "step": 700500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.309898577663326e-05,
+      "loss": 1.4505,
+      "step": 701000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.30968858110727e-05,
+      "loss": 1.4231,
+      "step": 701500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.3094785845512134e-05,
+      "loss": 1.4554,
+      "step": 702000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.309268587995157e-05,
+      "loss": 1.4322,
+      "step": 702500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.3090585914391e-05,
+      "loss": 1.4357,
+      "step": 703000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.3088485948830435e-05,
+      "loss": 1.4412,
+      "step": 703500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.3086385983269875e-05,
+      "loss": 1.4213,
+      "step": 704000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.308428601770931e-05,
+      "loss": 1.4415,
+      "step": 704500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.308218605214874e-05,
+      "loss": 1.4051,
+      "step": 705000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.308008608658818e-05,
+      "loss": 1.4593,
+      "step": 705500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.3077994520889856e-05,
+      "loss": 1.4116,
+      "step": 706000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.3075894555329296e-05,
+      "loss": 1.3986,
+      "step": 706500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.307379458976873e-05,
+      "loss": 1.4291,
+      "step": 707000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.307169462420816e-05,
+      "loss": 1.4258,
+      "step": 707500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.3069594658647596e-05,
+      "loss": 1.43,
+      "step": 708000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.306749889301816e-05,
+      "loss": 1.4185,
+      "step": 708500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.306539892745759e-05,
+      "loss": 1.4218,
+      "step": 709000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.3063298961897024e-05,
+      "loss": 1.4296,
+      "step": 709500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.306119899633646e-05,
+      "loss": 1.425,
+      "step": 710000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.305909903077589e-05,
+      "loss": 1.4172,
+      "step": 710500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.305699906521533e-05,
+      "loss": 1.4091,
+      "step": 711000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.3054899099654764e-05,
+      "loss": 1.4459,
+      "step": 711500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.30527991340942e-05,
+      "loss": 1.4144,
+      "step": 712000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.305069916853364e-05,
+      "loss": 1.4102,
+      "step": 712500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.304860340290419e-05,
+      "loss": 1.4436,
+      "step": 713000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.3046503437343625e-05,
+      "loss": 1.4086,
+      "step": 713500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.3044407671714185e-05,
+      "loss": 1.4445,
+      "step": 714000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.304230770615362e-05,
+      "loss": 1.4127,
+      "step": 714500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.304020774059305e-05,
+      "loss": 1.4149,
+      "step": 715000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.3038107775032486e-05,
+      "loss": 1.4366,
+      "step": 715500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.303600780947192e-05,
+      "loss": 1.4374,
+      "step": 716000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.303390784391136e-05,
+      "loss": 1.4664,
+      "step": 716500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.303181207828192e-05,
+      "loss": 1.46,
+      "step": 717000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.3029712112721347e-05,
+      "loss": 1.4286,
+      "step": 717500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.302761214716079e-05,
+      "loss": 1.4548,
+      "step": 718000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.302551218160022e-05,
+      "loss": 1.4518,
+      "step": 718500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.3023412216039654e-05,
+      "loss": 1.4134,
+      "step": 719000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.3021312250479094e-05,
+      "loss": 1.4766,
+      "step": 719500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.301921648484965e-05,
+      "loss": 1.4393,
+      "step": 720000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.301711651928908e-05,
+      "loss": 1.4194,
+      "step": 720500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.3015016553728514e-05,
+      "loss": 1.4593,
+      "step": 721000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.3012916588167955e-05,
+      "loss": 1.4598,
+      "step": 721500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.301081662260739e-05,
+      "loss": 1.4184,
+      "step": 722000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.300872085697794e-05,
+      "loss": 1.4309,
+      "step": 722500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.3006620891417375e-05,
+      "loss": 1.4349,
+      "step": 723000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.3004520925856815e-05,
+      "loss": 1.4393,
+      "step": 723500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.300242096029625e-05,
+      "loss": 1.4495,
+      "step": 724000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.300032099473568e-05,
+      "loss": 1.433,
+      "step": 724500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.299822102917512e-05,
+      "loss": 1.4518,
+      "step": 725000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.2996121063614556e-05,
+      "loss": 1.4506,
+      "step": 725500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.299402109805399e-05,
+      "loss": 1.4156,
+      "step": 726000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.299192533242455e-05,
+      "loss": 1.4393,
+      "step": 726500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.298982536686398e-05,
+      "loss": 1.4054,
+      "step": 727000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.298772540130342e-05,
+      "loss": 1.4535,
+      "step": 727500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.298562543574286e-05,
+      "loss": 1.4119,
+      "step": 728000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.298352547018229e-05,
+      "loss": 1.4354,
+      "step": 728500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.298143390448397e-05,
+      "loss": 1.4135,
+      "step": 729000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.29793339389234e-05,
+      "loss": 1.4155,
+      "step": 729500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.297723397336283e-05,
+      "loss": 1.4293,
+      "step": 730000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.297513400780227e-05,
+      "loss": 1.4225,
+      "step": 730500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.297303824217283e-05,
+      "loss": 1.4443,
+      "step": 731000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.297093827661226e-05,
+      "loss": 1.4244,
+      "step": 731500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.29688383110517e-05,
+      "loss": 1.4241,
+      "step": 732000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.296673834549113e-05,
+      "loss": 1.4309,
+      "step": 732500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.2964638379930565e-05,
+      "loss": 1.4142,
+      "step": 733000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.2962542614301126e-05,
+      "loss": 1.4607,
+      "step": 733500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.296044264874056e-05,
+      "loss": 1.4232,
+      "step": 734000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.295834268317999e-05,
+      "loss": 1.4346,
+      "step": 734500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.2956242717619426e-05,
+      "loss": 1.429,
+      "step": 735000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.2954146951989987e-05,
+      "loss": 1.4275,
+      "step": 735500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.295204698642943e-05,
+      "loss": 1.4596,
+      "step": 736000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.2949947020868853e-05,
+      "loss": 1.4197,
+      "step": 736500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.294784705530829e-05,
+      "loss": 1.4253,
+      "step": 737000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.294574708974773e-05,
+      "loss": 1.4374,
+      "step": 737500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.294364712418716e-05,
+      "loss": 1.4415,
+      "step": 738000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.2941547158626594e-05,
+      "loss": 1.4352,
+      "step": 738500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.2939447193066034e-05,
+      "loss": 1.4322,
+      "step": 739000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.2937355627367715e-05,
+      "loss": 1.4446,
+      "step": 739500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.293525566180715e-05,
+      "loss": 1.4068,
+      "step": 740000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.293315569624658e-05,
+      "loss": 1.4331,
+      "step": 740500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.2931055730686015e-05,
+      "loss": 1.4115,
+      "step": 741000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.292895576512545e-05,
+      "loss": 1.4116,
+      "step": 741500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.292685579956488e-05,
+      "loss": 1.426,
+      "step": 742000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.292475583400432e-05,
+      "loss": 1.4512,
+      "step": 742500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.2922655868443756e-05,
+      "loss": 1.4143,
+      "step": 743000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.292056010281431e-05,
+      "loss": 1.4206,
+      "step": 743500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.291846013725374e-05,
+      "loss": 1.4405,
+      "step": 744000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.291636017169318e-05,
+      "loss": 1.4426,
+      "step": 744500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.291426440606374e-05,
+      "loss": 1.4342,
+      "step": 745000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.291216444050318e-05,
+      "loss": 1.389,
+      "step": 745500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.291006447494261e-05,
+      "loss": 1.3899,
+      "step": 746000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.2907964509382044e-05,
+      "loss": 1.4248,
+      "step": 746500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.2905868743752604e-05,
+      "loss": 1.4283,
+      "step": 747000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.290376877819204e-05,
+      "loss": 1.4424,
+      "step": 747500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.290167301256259e-05,
+      "loss": 1.4209,
+      "step": 748000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.289957304700203e-05,
+      "loss": 1.4141,
+      "step": 748500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.2897473081441465e-05,
+      "loss": 1.4175,
+      "step": 749000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.28953731158809e-05,
+      "loss": 1.4353,
+      "step": 749500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.289327315032034e-05,
+      "loss": 1.4151,
+      "step": 750000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.289117318475977e-05,
+      "loss": 1.432,
+      "step": 750500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.28890732191992e-05,
+      "loss": 1.4435,
+      "step": 751000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.288697325363864e-05,
+      "loss": 1.4726,
+      "step": 751500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.288487328807807e-05,
+      "loss": 1.4564,
+      "step": 752000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.2882773322517506e-05,
+      "loss": 1.4647,
+      "step": 752500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.2880673356956946e-05,
+      "loss": 1.4249,
+      "step": 753000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.287857339139638e-05,
+      "loss": 1.4816,
+      "step": 753500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.287647762576693e-05,
+      "loss": 1.422,
+      "step": 754000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.2874386060068614e-05,
+      "loss": 1.4455,
+      "step": 754500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.287228609450805e-05,
+      "loss": 1.421,
+      "step": 755000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.287018612894749e-05,
+      "loss": 1.4052,
+      "step": 755500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.286808616338692e-05,
+      "loss": 1.3963,
+      "step": 756000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.2865986197826354e-05,
+      "loss": 1.4512,
+      "step": 756500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.2863886232265794e-05,
+      "loss": 1.4022,
+      "step": 757000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.286178626670523e-05,
+      "loss": 1.4024,
+      "step": 757500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.285969050107578e-05,
+      "loss": 1.4019,
+      "step": 758000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.2857590535515215e-05,
+      "loss": 1.4412,
+      "step": 758500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.2855490569954655e-05,
+      "loss": 1.4478,
+      "step": 759000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.285339060439409e-05,
+      "loss": 1.4091,
+      "step": 759500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.285129063883352e-05,
+      "loss": 1.4316,
+      "step": 760000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.2849190673272955e-05,
+      "loss": 1.436,
+      "step": 760500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.284709070771239e-05,
+      "loss": 1.4195,
+      "step": 761000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.284499494208295e-05,
+      "loss": 1.3842,
+      "step": 761500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.284289497652239e-05,
+      "loss": 1.4408,
+      "step": 762000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.2840795010961816e-05,
+      "loss": 1.4332,
+      "step": 762500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.283869504540125e-05,
+      "loss": 1.3958,
+      "step": 763000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.283659507984069e-05,
+      "loss": 1.4334,
+      "step": 763500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.283449511428012e-05,
+      "loss": 1.4235,
+      "step": 764000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.283239514871956e-05,
+      "loss": 1.4217,
+      "step": 764500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.283029938309011e-05,
+      "loss": 1.4225,
+      "step": 765000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.282819941752955e-05,
+      "loss": 1.4362,
+      "step": 765500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.2826099451968984e-05,
+      "loss": 1.396,
+      "step": 766000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.282399948640842e-05,
+      "loss": 1.4001,
+      "step": 766500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.2821903720778985e-05,
+      "loss": 1.4292,
+      "step": 767000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.281980375521841e-05,
+      "loss": 1.4172,
+      "step": 767500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.2817703789657845e-05,
+      "loss": 1.4426,
+      "step": 768000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.2815603824097285e-05,
+      "loss": 1.4257,
+      "step": 768500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.281350385853672e-05,
+      "loss": 1.4549,
+      "step": 769000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.281140389297615e-05,
+      "loss": 1.429,
+      "step": 769500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.280930392741559e-05,
+      "loss": 1.4024,
+      "step": 770000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.2807203961855026e-05,
+      "loss": 1.3868,
+      "step": 770500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.280510399629446e-05,
+      "loss": 1.4544,
+      "step": 771000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.280300823066501e-05,
+      "loss": 1.402,
+      "step": 771500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.280090826510445e-05,
+      "loss": 1.4316,
+      "step": 772000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.2798808299543886e-05,
+      "loss": 1.409,
+      "step": 772500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.279670833398332e-05,
+      "loss": 1.4215,
+      "step": 773000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.279460836842276e-05,
+      "loss": 1.4543,
+      "step": 773500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.2792508402862194e-05,
+      "loss": 1.4367,
+      "step": 774000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.279040843730163e-05,
+      "loss": 1.4511,
+      "step": 774500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.278831267167219e-05,
+      "loss": 1.4217,
+      "step": 775000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.278621270611162e-05,
+      "loss": 1.423,
+      "step": 775500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2784112740551054e-05,
+      "loss": 1.4249,
+      "step": 776000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2782012774990495e-05,
+      "loss": 1.4343,
+      "step": 776500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.277991280942993e-05,
+      "loss": 1.4311,
+      "step": 777000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.277781704380048e-05,
+      "loss": 1.3929,
+      "step": 777500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2775721278171035e-05,
+      "loss": 1.384,
+      "step": 778000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.277362131261047e-05,
+      "loss": 1.4355,
+      "step": 778500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.277152134704991e-05,
+      "loss": 1.4185,
+      "step": 779000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.276942138148934e-05,
+      "loss": 1.4229,
+      "step": 779500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2767321415928776e-05,
+      "loss": 1.4343,
+      "step": 780000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2765221450368216e-05,
+      "loss": 1.4136,
+      "step": 780500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.276312148480765e-05,
+      "loss": 1.3813,
+      "step": 781000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.276102151924708e-05,
+      "loss": 1.4223,
+      "step": 781500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.275892155368652e-05,
+      "loss": 1.3987,
+      "step": 782000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.275682158812595e-05,
+      "loss": 1.4144,
+      "step": 782500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.275472162256539e-05,
+      "loss": 1.4169,
+      "step": 783000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2752621657004824e-05,
+      "loss": 1.4241,
+      "step": 783500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.275052169144426e-05,
+      "loss": 1.423,
+      "step": 784000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.274842592581482e-05,
+      "loss": 1.4311,
+      "step": 784500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.274633016018537e-05,
+      "loss": 1.4093,
+      "step": 785000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2744234394555924e-05,
+      "loss": 1.4348,
+      "step": 785500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2742134428995365e-05,
+      "loss": 1.4067,
+      "step": 786000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.27400344634348e-05,
+      "loss": 1.4248,
+      "step": 786500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.273793449787423e-05,
+      "loss": 1.4434,
+      "step": 787000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.273583453231367e-05,
+      "loss": 1.4103,
+      "step": 787500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2733734566753105e-05,
+      "loss": 1.4276,
+      "step": 788000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.273163460119254e-05,
+      "loss": 1.4342,
+      "step": 788500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.272953463563198e-05,
+      "loss": 1.4141,
+      "step": 789000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2727434670071406e-05,
+      "loss": 1.392,
+      "step": 789500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2725338904441966e-05,
+      "loss": 1.4136,
+      "step": 790000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2723238938881406e-05,
+      "loss": 1.3944,
+      "step": 790500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.272113897332084e-05,
+      "loss": 1.4292,
+      "step": 791000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.271903900776027e-05,
+      "loss": 1.4284,
+      "step": 791500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.271693904219971e-05,
+      "loss": 1.4478,
+      "step": 792000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.271483907663914e-05,
+      "loss": 1.4383,
+      "step": 792500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2712739111078574e-05,
+      "loss": 1.433,
+      "step": 793000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2710639145518014e-05,
+      "loss": 1.4345,
+      "step": 793500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2708543379888574e-05,
+      "loss": 1.4657,
+      "step": 794000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2706443414328e-05,
+      "loss": 1.4015,
+      "step": 794500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2704343448767434e-05,
+      "loss": 1.3839,
+      "step": 795000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2702243483206875e-05,
+      "loss": 1.4436,
+      "step": 795500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2700147717577435e-05,
+      "loss": 1.4201,
+      "step": 796000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.269805195194799e-05,
+      "loss": 1.4199,
+      "step": 796500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.269595198638742e-05,
+      "loss": 1.4173,
+      "step": 797000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.269385202082686e-05,
+      "loss": 1.4289,
+      "step": 797500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2691752055266296e-05,
+      "loss": 1.4226,
+      "step": 798000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.268965628963685e-05,
+      "loss": 1.3627,
+      "step": 798500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.268755632407628e-05,
+      "loss": 1.4062,
+      "step": 799000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.268545635851572e-05,
+      "loss": 1.4044,
+      "step": 799500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2683356392955156e-05,
+      "loss": 1.3971,
+      "step": 800000
+    },
+    {
+      "epoch": 0.48,
+      "eval_loss": 1.3476606607437134,
+      "eval_runtime": 1101.7337,
+      "eval_samples_per_second": 478.083,
+      "eval_steps_per_second": 79.681,
+      "step": 800000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.268125642739459e-05,
+      "loss": 1.4049,
+      "step": 800500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.267915646183403e-05,
+      "loss": 1.4219,
+      "step": 801000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.267705649627346e-05,
+      "loss": 1.4239,
+      "step": 801500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.267495653071289e-05,
+      "loss": 1.4172,
+      "step": 802000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.267285656515233e-05,
+      "loss": 1.4057,
+      "step": 802500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2670756599591764e-05,
+      "loss": 1.4081,
+      "step": 803000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.26686566340312e-05,
+      "loss": 1.4203,
+      "step": 803500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.266655666847064e-05,
+      "loss": 1.4057,
+      "step": 804000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.266446090284119e-05,
+      "loss": 1.415,
+      "step": 804500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2662360937280625e-05,
+      "loss": 1.4025,
+      "step": 805000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2660260971720065e-05,
+      "loss": 1.4336,
+      "step": 805500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.26581610061595e-05,
+      "loss": 1.3758,
+      "step": 806000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.265606944046118e-05,
+      "loss": 1.4085,
+      "step": 806500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.265396947490061e-05,
+      "loss": 1.402,
+      "step": 807000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2651869509340046e-05,
+      "loss": 1.4093,
+      "step": 807500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2649769543779486e-05,
+      "loss": 1.4118,
+      "step": 808000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.264766957821891e-05,
+      "loss": 1.4126,
+      "step": 808500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2645569612658346e-05,
+      "loss": 1.4374,
+      "step": 809000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2643469647097786e-05,
+      "loss": 1.4125,
+      "step": 809500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.264136968153722e-05,
+      "loss": 1.4105,
+      "step": 810000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.263926971597665e-05,
+      "loss": 1.4576,
+      "step": 810500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2637178150278334e-05,
+      "loss": 1.423,
+      "step": 811000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2635078184717774e-05,
+      "loss": 1.4182,
+      "step": 811500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.263297821915721e-05,
+      "loss": 1.4275,
+      "step": 812000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.263087825359664e-05,
+      "loss": 1.4098,
+      "step": 812500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.262877828803608e-05,
+      "loss": 1.4135,
+      "step": 813000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.262667832247551e-05,
+      "loss": 1.3993,
+      "step": 813500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.262457835691494e-05,
+      "loss": 1.4268,
+      "step": 814000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.262247839135438e-05,
+      "loss": 1.4105,
+      "step": 814500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.262038262572494e-05,
+      "loss": 1.4069,
+      "step": 815000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2618282660164375e-05,
+      "loss": 1.3967,
+      "step": 815500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.261618689453493e-05,
+      "loss": 1.3951,
+      "step": 816000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.261408692897436e-05,
+      "loss": 1.4378,
+      "step": 816500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.26119869634138e-05,
+      "loss": 1.3984,
+      "step": 817000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2609886997853236e-05,
+      "loss": 1.3945,
+      "step": 817500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.260779123222379e-05,
+      "loss": 1.4236,
+      "step": 818000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.260569126666323e-05,
+      "loss": 1.3945,
+      "step": 818500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.260359130110266e-05,
+      "loss": 1.4138,
+      "step": 819000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.26014913355421e-05,
+      "loss": 1.3774,
+      "step": 819500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.259939136998154e-05,
+      "loss": 1.4063,
+      "step": 820000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2597291404420964e-05,
+      "loss": 1.4424,
+      "step": 820500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.25951914388604e-05,
+      "loss": 1.4055,
+      "step": 821000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.259309147329984e-05,
+      "loss": 1.3729,
+      "step": 821500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.259099150773927e-05,
+      "loss": 1.4052,
+      "step": 822000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.258889574210983e-05,
+      "loss": 1.4193,
+      "step": 822500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.258679577654926e-05,
+      "loss": 1.4445,
+      "step": 823000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.25846958109887e-05,
+      "loss": 1.4077,
+      "step": 823500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.258260004535926e-05,
+      "loss": 1.4075,
+      "step": 824000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.258050007979869e-05,
+      "loss": 1.3979,
+      "step": 824500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2578400114238125e-05,
+      "loss": 1.4419,
+      "step": 825000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.257630014867756e-05,
+      "loss": 1.4302,
+      "step": 825500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.257420018311699e-05,
+      "loss": 1.4607,
+      "step": 826000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.257210441748755e-05,
+      "loss": 1.4044,
+      "step": 826500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.257000445192699e-05,
+      "loss": 1.414,
+      "step": 827000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2567908686297546e-05,
+      "loss": 1.412,
+      "step": 827500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.256580872073698e-05,
+      "loss": 1.4107,
+      "step": 828000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.256370875517641e-05,
+      "loss": 1.4276,
+      "step": 828500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2561608789615854e-05,
+      "loss": 1.4317,
+      "step": 829000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.255951302398641e-05,
+      "loss": 1.3977,
+      "step": 829500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.255741305842584e-05,
+      "loss": 1.4031,
+      "step": 830000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2555317292796394e-05,
+      "loss": 1.4017,
+      "step": 830500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2553217327235834e-05,
+      "loss": 1.4126,
+      "step": 831000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.255111736167527e-05,
+      "loss": 1.408,
+      "step": 831500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.25490173961147e-05,
+      "loss": 1.3863,
+      "step": 832000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.254691743055414e-05,
+      "loss": 1.4112,
+      "step": 832500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2544817464993575e-05,
+      "loss": 1.3932,
+      "step": 833000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.254271749943301e-05,
+      "loss": 1.429,
+      "step": 833500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.254061753387245e-05,
+      "loss": 1.4309,
+      "step": 834000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.253851756831188e-05,
+      "loss": 1.4212,
+      "step": 834500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.253641760275131e-05,
+      "loss": 1.4225,
+      "step": 835000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.253431763719075e-05,
+      "loss": 1.3975,
+      "step": 835500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.253221767163018e-05,
+      "loss": 1.4305,
+      "step": 836000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2530117706069616e-05,
+      "loss": 1.4204,
+      "step": 836500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2528017740509056e-05,
+      "loss": 1.4108,
+      "step": 837000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.252592197487961e-05,
+      "loss": 1.3828,
+      "step": 837500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.252382200931904e-05,
+      "loss": 1.4142,
+      "step": 838000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.252172204375848e-05,
+      "loss": 1.4294,
+      "step": 838500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.251962207819792e-05,
+      "loss": 1.4154,
+      "step": 839000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.251752211263735e-05,
+      "loss": 1.4359,
+      "step": 839500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.251542214707679e-05,
+      "loss": 1.4021,
+      "step": 840000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2513326381447344e-05,
+      "loss": 1.4131,
+      "step": 840500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.251122641588678e-05,
+      "loss": 1.4293,
+      "step": 841000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.250912645032621e-05,
+      "loss": 1.4133,
+      "step": 841500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.250702648476565e-05,
+      "loss": 1.441,
+      "step": 842000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2504926519205085e-05,
+      "loss": 1.4341,
+      "step": 842500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.250282655364452e-05,
+      "loss": 1.4481,
+      "step": 843000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.250073078801507e-05,
+      "loss": 1.3972,
+      "step": 843500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.249863082245451e-05,
+      "loss": 1.428,
+      "step": 844000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2496530856893946e-05,
+      "loss": 1.4243,
+      "step": 844500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.249443089133338e-05,
+      "loss": 1.4263,
+      "step": 845000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.249233512570393e-05,
+      "loss": 1.4228,
+      "step": 845500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.249023516014337e-05,
+      "loss": 1.3805,
+      "step": 846000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2488135194582806e-05,
+      "loss": 1.4233,
+      "step": 846500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2486035229022247e-05,
+      "loss": 1.4401,
+      "step": 847000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.248393526346168e-05,
+      "loss": 1.4353,
+      "step": 847500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2481839497832234e-05,
+      "loss": 1.4342,
+      "step": 848000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.247973953227167e-05,
+      "loss": 1.4169,
+      "step": 848500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.247763956671111e-05,
+      "loss": 1.4189,
+      "step": 849000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.247553960115054e-05,
+      "loss": 1.3913,
+      "step": 849500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2473439635589974e-05,
+      "loss": 1.4473,
+      "step": 850000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2471339670029414e-05,
+      "loss": 1.4138,
+      "step": 850500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.246924390439997e-05,
+      "loss": 1.4174,
+      "step": 851000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.24671439388394e-05,
+      "loss": 1.4409,
+      "step": 851500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2465043973278835e-05,
+      "loss": 1.4366,
+      "step": 852000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2462944007718275e-05,
+      "loss": 1.3988,
+      "step": 852500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.246084404215771e-05,
+      "loss": 1.4026,
+      "step": 853000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.245874407659714e-05,
+      "loss": 1.4102,
+      "step": 853500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.24566483109677e-05,
+      "loss": 1.4261,
+      "step": 854000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2454548345407136e-05,
+      "loss": 1.4146,
+      "step": 854500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.245244837984657e-05,
+      "loss": 1.4159,
+      "step": 855000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.245034841428601e-05,
+      "loss": 1.3839,
+      "step": 855500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.244824844872544e-05,
+      "loss": 1.4077,
+      "step": 856000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2446148483164876e-05,
+      "loss": 1.3936,
+      "step": 856500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.244404851760431e-05,
+      "loss": 1.3762,
+      "step": 857000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2441948552043743e-05,
+      "loss": 1.4056,
+      "step": 857500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.243984858648318e-05,
+      "loss": 1.4031,
+      "step": 858000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.243775702078486e-05,
+      "loss": 1.4041,
+      "step": 858500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.243565705522429e-05,
+      "loss": 1.4188,
+      "step": 859000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.243355708966373e-05,
+      "loss": 1.4126,
+      "step": 859500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.2431457124103164e-05,
+      "loss": 1.4103,
+      "step": 860000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.24293571585426e-05,
+      "loss": 1.4154,
+      "step": 860500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.242725719298204e-05,
+      "loss": 1.4055,
+      "step": 861000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.242516142735259e-05,
+      "loss": 1.3874,
+      "step": 861500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.2423061461792025e-05,
+      "loss": 1.387,
+      "step": 862000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.2420961496231465e-05,
+      "loss": 1.4242,
+      "step": 862500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.24188615306709e-05,
+      "loss": 1.4214,
+      "step": 863000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.241676156511033e-05,
+      "loss": 1.4212,
+      "step": 863500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.241466159954977e-05,
+      "loss": 1.4259,
+      "step": 864000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.24125616339892e-05,
+      "loss": 1.4294,
+      "step": 864500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.241046586835976e-05,
+      "loss": 1.3904,
+      "step": 865000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.240836590279919e-05,
+      "loss": 1.3872,
+      "step": 865500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.240626593723863e-05,
+      "loss": 1.415,
+      "step": 866000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.240416597167806e-05,
+      "loss": 1.4467,
+      "step": 866500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.2402066006117493e-05,
+      "loss": 1.3677,
+      "step": 867000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.2399966040556934e-05,
+      "loss": 1.3957,
+      "step": 867500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.2397870274927494e-05,
+      "loss": 1.4113,
+      "step": 868000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.239577450929805e-05,
+      "loss": 1.4239,
+      "step": 868500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.239367454373748e-05,
+      "loss": 1.3956,
+      "step": 869000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.239157457817692e-05,
+      "loss": 1.3879,
+      "step": 869500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.2389474612616355e-05,
+      "loss": 1.4304,
+      "step": 870000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.238737464705579e-05,
+      "loss": 1.4105,
+      "step": 870500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.238527468149523e-05,
+      "loss": 1.4289,
+      "step": 871000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.2383174715934655e-05,
+      "loss": 1.3973,
+      "step": 871500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.238107475037409e-05,
+      "loss": 1.3981,
+      "step": 872000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.237897478481353e-05,
+      "loss": 1.3973,
+      "step": 872500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.237687481925296e-05,
+      "loss": 1.4019,
+      "step": 873000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.2374774853692396e-05,
+      "loss": 1.4261,
+      "step": 873500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.2372674888131836e-05,
+      "loss": 1.4133,
+      "step": 874000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.237058332243351e-05,
+      "loss": 1.4314,
+      "step": 874500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.236848335687295e-05,
+      "loss": 1.3962,
+      "step": 875000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.2366383391312383e-05,
+      "loss": 1.4012,
+      "step": 875500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.236428762568294e-05,
+      "loss": 1.4301,
+      "step": 876000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.236218766012238e-05,
+      "loss": 1.4135,
+      "step": 876500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.236008769456181e-05,
+      "loss": 1.3894,
+      "step": 877000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.2357987729001244e-05,
+      "loss": 1.4161,
+      "step": 877500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.2355887763440684e-05,
+      "loss": 1.4214,
+      "step": 878000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.235378779788011e-05,
+      "loss": 1.4302,
+      "step": 878500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.2351687832319544e-05,
+      "loss": 1.4385,
+      "step": 879000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.2349587866758985e-05,
+      "loss": 1.4295,
+      "step": 879500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.2347492101129545e-05,
+      "loss": 1.4233,
+      "step": 880000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.234539213556898e-05,
+      "loss": 1.3985,
+      "step": 880500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.2343292170008405e-05,
+      "loss": 1.4143,
+      "step": 881000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.2341192204447845e-05,
+      "loss": 1.4033,
+      "step": 881500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.233909223888728e-05,
+      "loss": 1.3976,
+      "step": 882000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.233699227332671e-05,
+      "loss": 1.3902,
+      "step": 882500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.233489230776615e-05,
+      "loss": 1.4114,
+      "step": 883000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.2332792342205586e-05,
+      "loss": 1.4045,
+      "step": 883500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.233069657657614e-05,
+      "loss": 1.3903,
+      "step": 884000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.232859661101558e-05,
+      "loss": 1.4144,
+      "step": 884500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.232649664545501e-05,
+      "loss": 1.4088,
+      "step": 885000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.232439667989445e-05,
+      "loss": 1.364,
+      "step": 885500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.2322300914265e-05,
+      "loss": 1.3881,
+      "step": 886000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.232020514863556e-05,
+      "loss": 1.418,
+      "step": 886500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.2318105183075e-05,
+      "loss": 1.4143,
+      "step": 887000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.2316005217514434e-05,
+      "loss": 1.4366,
+      "step": 887500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.231390525195386e-05,
+      "loss": 1.4213,
+      "step": 888000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.23118052863933e-05,
+      "loss": 1.3788,
+      "step": 888500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.2309705320832735e-05,
+      "loss": 1.3851,
+      "step": 889000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.230760535527217e-05,
+      "loss": 1.3879,
+      "step": 889500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.230550958964273e-05,
+      "loss": 1.386,
+      "step": 890000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.230340962408216e-05,
+      "loss": 1.4073,
+      "step": 890500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.2301309658521596e-05,
+      "loss": 1.4101,
+      "step": 891000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.2299209692961036e-05,
+      "loss": 1.4105,
+      "step": 891500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.229710972740047e-05,
+      "loss": 1.4069,
+      "step": 892000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.22950097618399e-05,
+      "loss": 1.4192,
+      "step": 892500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.229290979627934e-05,
+      "loss": 1.3792,
+      "step": 893000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.2290809830718776e-05,
+      "loss": 1.3804,
+      "step": 893500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.228871406508933e-05,
+      "loss": 1.4097,
+      "step": 894000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.2286614099528763e-05,
+      "loss": 1.4191,
+      "step": 894500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.2284514133968204e-05,
+      "loss": 1.4171,
+      "step": 895000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.228241416840764e-05,
+      "loss": 1.4221,
+      "step": 895500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.228032260270932e-05,
+      "loss": 1.3626,
+      "step": 896000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.227822263714875e-05,
+      "loss": 1.3973,
+      "step": 896500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.2276122671588184e-05,
+      "loss": 1.3964,
+      "step": 897000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.2274026905958745e-05,
+      "loss": 1.4159,
+      "step": 897500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.227192694039818e-05,
+      "loss": 1.4249,
+      "step": 898000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.226982697483761e-05,
+      "loss": 1.4169,
+      "step": 898500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.226772700927705e-05,
+      "loss": 1.4336,
+      "step": 899000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.2265631243647606e-05,
+      "loss": 1.4122,
+      "step": 899500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.226353127808704e-05,
+      "loss": 1.4346,
+      "step": 900000
+    },
+    {
+      "epoch": 0.54,
+      "eval_loss": 1.338213324546814,
+      "eval_runtime": 1101.9411,
+      "eval_samples_per_second": 477.993,
+      "eval_steps_per_second": 79.666,
+      "step": 900000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.226143131252647e-05,
+      "loss": 1.4128,
+      "step": 900500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.225933134696591e-05,
+      "loss": 1.4096,
+      "step": 901000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.2257231381405346e-05,
+      "loss": 1.3951,
+      "step": 901500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.225513141584478e-05,
+      "loss": 1.4103,
+      "step": 902000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.225303145028421e-05,
+      "loss": 1.3779,
+      "step": 902500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.2250935684654773e-05,
+      "loss": 1.3767,
+      "step": 903000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.224883571909421e-05,
+      "loss": 1.3837,
+      "step": 903500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.224673575353364e-05,
+      "loss": 1.4177,
+      "step": 904000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.224463578797308e-05,
+      "loss": 1.3836,
+      "step": 904500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.224253582241251e-05,
+      "loss": 1.3993,
+      "step": 905000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.224043585685195e-05,
+      "loss": 1.4055,
+      "step": 905500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.223833589129138e-05,
+      "loss": 1.4267,
+      "step": 906000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.2236235925730814e-05,
+      "loss": 1.4204,
+      "step": 906500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.2234135960170255e-05,
+      "loss": 1.4353,
+      "step": 907000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.223203599460969e-05,
+      "loss": 1.3914,
+      "step": 907500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.222993602904912e-05,
+      "loss": 1.3852,
+      "step": 908000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.222783606348856e-05,
+      "loss": 1.3969,
+      "step": 908500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.2225740297859115e-05,
+      "loss": 1.4055,
+      "step": 909000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.222364033229855e-05,
+      "loss": 1.3707,
+      "step": 909500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.222154036673798e-05,
+      "loss": 1.3909,
+      "step": 910000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.221944040117742e-05,
+      "loss": 1.3987,
+      "step": 910500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.2217340435616856e-05,
+      "loss": 1.4216,
+      "step": 911000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.221524047005629e-05,
+      "loss": 1.4172,
+      "step": 911500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.221314470442685e-05,
+      "loss": 1.3801,
+      "step": 912000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.221104473886628e-05,
+      "loss": 1.4128,
+      "step": 912500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.220894477330572e-05,
+      "loss": 1.4,
+      "step": 913000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.220684480774516e-05,
+      "loss": 1.411,
+      "step": 913500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.220474484218459e-05,
+      "loss": 1.4058,
+      "step": 914000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.2202649076555144e-05,
+      "loss": 1.4101,
+      "step": 914500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.220054911099458e-05,
+      "loss": 1.406,
+      "step": 915000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.219844914543402e-05,
+      "loss": 1.4022,
+      "step": 915500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.219635337980457e-05,
+      "loss": 1.3702,
+      "step": 916000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.2194253414244005e-05,
+      "loss": 1.4078,
+      "step": 916500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.219215344868344e-05,
+      "loss": 1.3671,
+      "step": 917000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.219005348312288e-05,
+      "loss": 1.4178,
+      "step": 917500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.218795351756231e-05,
+      "loss": 1.4283,
+      "step": 918000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.2185853552001745e-05,
+      "loss": 1.3755,
+      "step": 918500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.2183753586441186e-05,
+      "loss": 1.4042,
+      "step": 919000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.218165362088062e-05,
+      "loss": 1.3998,
+      "step": 919500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.217955365532005e-05,
+      "loss": 1.4169,
+      "step": 920000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.2177453689759486e-05,
+      "loss": 1.4142,
+      "step": 920500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.217535372419892e-05,
+      "loss": 1.4231,
+      "step": 921000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.217325375863836e-05,
+      "loss": 1.393,
+      "step": 921500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.217115799300891e-05,
+      "loss": 1.3594,
+      "step": 922000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.216905802744835e-05,
+      "loss": 1.4134,
+      "step": 922500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.216695806188778e-05,
+      "loss": 1.3847,
+      "step": 923000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.216485809632722e-05,
+      "loss": 1.4215,
+      "step": 923500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.216276233069778e-05,
+      "loss": 1.3818,
+      "step": 924000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.216066236513721e-05,
+      "loss": 1.3955,
+      "step": 924500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.215856659950777e-05,
+      "loss": 1.3995,
+      "step": 925000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.21564666339472e-05,
+      "loss": 1.4,
+      "step": 925500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.215436666838664e-05,
+      "loss": 1.4416,
+      "step": 926000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.2152266702826075e-05,
+      "loss": 1.3824,
+      "step": 926500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.215017093719663e-05,
+      "loss": 1.3972,
+      "step": 927000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.214807097163607e-05,
+      "loss": 1.395,
+      "step": 927500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.21459710060755e-05,
+      "loss": 1.4043,
+      "step": 928000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.2143871040514936e-05,
+      "loss": 1.4319,
+      "step": 928500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.2141771074954376e-05,
+      "loss": 1.4073,
+      "step": 929000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.213967530932493e-05,
+      "loss": 1.4013,
+      "step": 929500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.213757534376436e-05,
+      "loss": 1.4218,
+      "step": 930000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.2135475378203796e-05,
+      "loss": 1.4094,
+      "step": 930500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.213337961257435e-05,
+      "loss": 1.4099,
+      "step": 931000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.213127964701379e-05,
+      "loss": 1.3959,
+      "step": 931500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.2129179681453224e-05,
+      "loss": 1.4247,
+      "step": 932000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.212707971589266e-05,
+      "loss": 1.4103,
+      "step": 932500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.21249797503321e-05,
+      "loss": 1.4104,
+      "step": 933000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.212287978477153e-05,
+      "loss": 1.3538,
+      "step": 933500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.2120779819210964e-05,
+      "loss": 1.4161,
+      "step": 934000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.21186798536504e-05,
+      "loss": 1.4061,
+      "step": 934500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.211657988808983e-05,
+      "loss": 1.3926,
+      "step": 935000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.211448412246039e-05,
+      "loss": 1.4,
+      "step": 935500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.211238415689983e-05,
+      "loss": 1.3897,
+      "step": 936000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.211028419133926e-05,
+      "loss": 1.3829,
+      "step": 936500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.210818422577869e-05,
+      "loss": 1.4139,
+      "step": 937000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.210608426021813e-05,
+      "loss": 1.4418,
+      "step": 937500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.2103984294657566e-05,
+      "loss": 1.4129,
+      "step": 938000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.2101884329097e-05,
+      "loss": 1.4144,
+      "step": 938500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.209978436353644e-05,
+      "loss": 1.3888,
+      "step": 939000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.209768859790699e-05,
+      "loss": 1.3965,
+      "step": 939500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.2095588632346426e-05,
+      "loss": 1.4176,
+      "step": 940000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.209348866678586e-05,
+      "loss": 1.3752,
+      "step": 940500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.20913887012253e-05,
+      "loss": 1.4028,
+      "step": 941000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.2089288735664733e-05,
+      "loss": 1.4089,
+      "step": 941500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.208718877010417e-05,
+      "loss": 1.4531,
+      "step": 942000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.208508880454361e-05,
+      "loss": 1.4071,
+      "step": 942500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.208298883898304e-05,
+      "loss": 1.3919,
+      "step": 943000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.2080893073353594e-05,
+      "loss": 1.3715,
+      "step": 943500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.2078793107793034e-05,
+      "loss": 1.4305,
+      "step": 944000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.207669314223247e-05,
+      "loss": 1.4003,
+      "step": 944500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.20745931766719e-05,
+      "loss": 1.3806,
+      "step": 945000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.207249321111134e-05,
+      "loss": 1.4218,
+      "step": 945500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.2070397445481895e-05,
+      "loss": 1.4129,
+      "step": 946000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.206829747992133e-05,
+      "loss": 1.3931,
+      "step": 946500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.206619751436076e-05,
+      "loss": 1.4204,
+      "step": 947000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.20640975488002e-05,
+      "loss": 1.431,
+      "step": 947500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.2062001783170756e-05,
+      "loss": 1.4018,
+      "step": 948000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.205990601754131e-05,
+      "loss": 1.3723,
+      "step": 948500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.205781445184299e-05,
+      "loss": 1.3877,
+      "step": 949000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.2055714486282423e-05,
+      "loss": 1.4077,
+      "step": 949500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.205361452072186e-05,
+      "loss": 1.4215,
+      "step": 950000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.20515145551613e-05,
+      "loss": 1.3674,
+      "step": 950500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.204941458960073e-05,
+      "loss": 1.4133,
+      "step": 951000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.2047314624040164e-05,
+      "loss": 1.4096,
+      "step": 951500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.2045214658479604e-05,
+      "loss": 1.3995,
+      "step": 952000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.204311469291904e-05,
+      "loss": 1.4209,
+      "step": 952500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.2041014727358464e-05,
+      "loss": 1.4311,
+      "step": 953000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.2038914761797905e-05,
+      "loss": 1.4108,
+      "step": 953500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.203681479623734e-05,
+      "loss": 1.3646,
+      "step": 954000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.203471483067677e-05,
+      "loss": 1.3945,
+      "step": 954500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.203261486511621e-05,
+      "loss": 1.3849,
+      "step": 955000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.2030519099486765e-05,
+      "loss": 1.3742,
+      "step": 955500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.20284191339262e-05,
+      "loss": 1.3872,
+      "step": 956000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.202631916836564e-05,
+      "loss": 1.4085,
+      "step": 956500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.202421920280507e-05,
+      "loss": 1.4047,
+      "step": 957000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.2022119237244506e-05,
+      "loss": 1.3947,
+      "step": 957500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.2020019271683946e-05,
+      "loss": 1.3956,
+      "step": 958000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.201791930612338e-05,
+      "loss": 1.421,
+      "step": 958500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.201582354049393e-05,
+      "loss": 1.3871,
+      "step": 959000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.201372357493337e-05,
+      "loss": 1.4278,
+      "step": 959500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.201162360937281e-05,
+      "loss": 1.3927,
+      "step": 960000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.200952364381224e-05,
+      "loss": 1.384,
+      "step": 960500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.2007423678251674e-05,
+      "loss": 1.4166,
+      "step": 961000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.200532791262223e-05,
+      "loss": 1.408,
+      "step": 961500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.200322794706167e-05,
+      "loss": 1.4496,
+      "step": 962000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.20011279815011e-05,
+      "loss": 1.4173,
+      "step": 962500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.1999028015940535e-05,
+      "loss": 1.3771,
+      "step": 963000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.1996928050379975e-05,
+      "loss": 1.4008,
+      "step": 963500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.199482808481941e-05,
+      "loss": 1.4063,
+      "step": 964000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.199272811925885e-05,
+      "loss": 1.4071,
+      "step": 964500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.199062815369828e-05,
+      "loss": 1.3668,
+      "step": 965000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.1988532388068836e-05,
+      "loss": 1.3878,
+      "step": 965500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.198643242250827e-05,
+      "loss": 1.4048,
+      "step": 966000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.198433665687882e-05,
+      "loss": 1.3872,
+      "step": 966500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.198224089124938e-05,
+      "loss": 1.4358,
+      "step": 967000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.1980140925688816e-05,
+      "loss": 1.3811,
+      "step": 967500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.197804096012825e-05,
+      "loss": 1.3875,
+      "step": 968000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.197594099456768e-05,
+      "loss": 1.3947,
+      "step": 968500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.1973841029007124e-05,
+      "loss": 1.4098,
+      "step": 969000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.197174106344656e-05,
+      "loss": 1.4383,
+      "step": 969500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.196964109788599e-05,
+      "loss": 1.4002,
+      "step": 970000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.196754113232543e-05,
+      "loss": 1.4456,
+      "step": 970500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.1965441166764864e-05,
+      "loss": 1.4,
+      "step": 971000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.1963341201204304e-05,
+      "loss": 1.377,
+      "step": 971500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.196124123564374e-05,
+      "loss": 1.3814,
+      "step": 972000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.195914127008317e-05,
+      "loss": 1.4064,
+      "step": 972500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.1957041304522605e-05,
+      "loss": 1.378,
+      "step": 973000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.195494133896204e-05,
+      "loss": 1.382,
+      "step": 973500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.195284977326372e-05,
+      "loss": 1.3616,
+      "step": 974000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.195074980770315e-05,
+      "loss": 1.4069,
+      "step": 974500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.1948649842142586e-05,
+      "loss": 1.3978,
+      "step": 975000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.1946549876582026e-05,
+      "loss": 1.3999,
+      "step": 975500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.194444991102146e-05,
+      "loss": 1.3601,
+      "step": 976000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.194234994546089e-05,
+      "loss": 1.3785,
+      "step": 976500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.194025417983145e-05,
+      "loss": 1.4267,
+      "step": 977000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.1938154214270887e-05,
+      "loss": 1.386,
+      "step": 977500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.193605424871032e-05,
+      "loss": 1.4166,
+      "step": 978000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.193395428314976e-05,
+      "loss": 1.3764,
+      "step": 978500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.1931854317589194e-05,
+      "loss": 1.3911,
+      "step": 979000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.192975435202863e-05,
+      "loss": 1.4092,
+      "step": 979500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.192765438646806e-05,
+      "loss": 1.3693,
+      "step": 980000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.1925554420907494e-05,
+      "loss": 1.4016,
+      "step": 980500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.1923458655278054e-05,
+      "loss": 1.4067,
+      "step": 981000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.192135868971749e-05,
+      "loss": 1.4018,
+      "step": 981500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.191925872415693e-05,
+      "loss": 1.3851,
+      "step": 982000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.1917158758596355e-05,
+      "loss": 1.4042,
+      "step": 982500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.191505879303579e-05,
+      "loss": 1.412,
+      "step": 983000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.191295882747523e-05,
+      "loss": 1.3879,
+      "step": 983500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.191086306184579e-05,
+      "loss": 1.4056,
+      "step": 984000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.190876309628522e-05,
+      "loss": 1.3936,
+      "step": 984500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.1906663130724656e-05,
+      "loss": 1.3976,
+      "step": 985000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.190456316516409e-05,
+      "loss": 1.3847,
+      "step": 985500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.190246319960352e-05,
+      "loss": 1.4073,
+      "step": 986000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.190036323404296e-05,
+      "loss": 1.4018,
+      "step": 986500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.1898263268482396e-05,
+      "loss": 1.4115,
+      "step": 987000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.189616330292183e-05,
+      "loss": 1.3592,
+      "step": 987500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.1894067537292383e-05,
+      "loss": 1.389,
+      "step": 988000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.1891967571731824e-05,
+      "loss": 1.3925,
+      "step": 988500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.188986760617126e-05,
+      "loss": 1.3924,
+      "step": 989000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.188777184054181e-05,
+      "loss": 1.3927,
+      "step": 989500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.1885671874981244e-05,
+      "loss": 1.3955,
+      "step": 990000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.1883571909420684e-05,
+      "loss": 1.3992,
+      "step": 990500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.188147194386012e-05,
+      "loss": 1.4168,
+      "step": 991000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.187937197829955e-05,
+      "loss": 1.3828,
+      "step": 991500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.187727621267011e-05,
+      "loss": 1.4009,
+      "step": 992000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.1875176247109545e-05,
+      "loss": 1.4023,
+      "step": 992500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.187307628154898e-05,
+      "loss": 1.3965,
+      "step": 993000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.187097631598842e-05,
+      "loss": 1.3953,
+      "step": 993500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.186888055035898e-05,
+      "loss": 1.3709,
+      "step": 994000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.1866780584798406e-05,
+      "loss": 1.3969,
+      "step": 994500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.186468061923784e-05,
+      "loss": 1.3666,
+      "step": 995000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.186258065367728e-05,
+      "loss": 1.3968,
+      "step": 995500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.186048068811671e-05,
+      "loss": 1.4132,
+      "step": 996000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.1858380722556146e-05,
+      "loss": 1.3949,
+      "step": 996500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.185628075699559e-05,
+      "loss": 1.3795,
+      "step": 997000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.185418499136614e-05,
+      "loss": 1.3726,
+      "step": 997500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.18520892257367e-05,
+      "loss": 1.3921,
+      "step": 998000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.1849989260176134e-05,
+      "loss": 1.3885,
+      "step": 998500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.184788929461557e-05,
+      "loss": 1.3876,
+      "step": 999000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.1845789329055e-05,
+      "loss": 1.3769,
+      "step": 999500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.1843689363494434e-05,
+      "loss": 1.4139,
+      "step": 1000000
+    },
+    {
+      "epoch": 0.6,
+      "eval_loss": 1.3278173208236694,
+      "eval_runtime": 1111.6546,
+      "eval_samples_per_second": 473.816,
+      "eval_steps_per_second": 78.97,
+      "step": 1000000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.1841589397933875e-05,
+      "loss": 1.3798,
+      "step": 1000500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.183948943237331e-05,
+      "loss": 1.4033,
+      "step": 1001000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.183738946681274e-05,
+      "loss": 1.4,
+      "step": 1001500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.183528950125218e-05,
+      "loss": 1.4045,
+      "step": 1002000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.1833193735622735e-05,
+      "loss": 1.4086,
+      "step": 1002500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.183109377006217e-05,
+      "loss": 1.3815,
+      "step": 1003000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.182899800443273e-05,
+      "loss": 1.3827,
+      "step": 1003500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.1826898038872156e-05,
+      "loss": 1.3701,
+      "step": 1004000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.1824798073311596e-05,
+      "loss": 1.3901,
+      "step": 1004500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.182269810775103e-05,
+      "loss": 1.3954,
+      "step": 1005000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.182059814219046e-05,
+      "loss": 1.3861,
+      "step": 1005500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.18184981766299e-05,
+      "loss": 1.385,
+      "step": 1006000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.181639821106934e-05,
+      "loss": 1.3888,
+      "step": 1006500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.181429824550877e-05,
+      "loss": 1.3929,
+      "step": 1007000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.181220247987933e-05,
+      "loss": 1.3979,
+      "step": 1007500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.181010671424989e-05,
+      "loss": 1.4017,
+      "step": 1008000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.1808010948620445e-05,
+      "loss": 1.3863,
+      "step": 1008500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.180591098305988e-05,
+      "loss": 1.4291,
+      "step": 1009000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.180381101749931e-05,
+      "loss": 1.4162,
+      "step": 1009500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.180171105193875e-05,
+      "loss": 1.3728,
+      "step": 1010000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.1799611086378185e-05,
+      "loss": 1.3718,
+      "step": 1010500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.179751112081761e-05,
+      "loss": 1.412,
+      "step": 1011000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.179541535518817e-05,
+      "loss": 1.3689,
+      "step": 1011500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.179331538962761e-05,
+      "loss": 1.3711,
+      "step": 1012000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.1791215424067046e-05,
+      "loss": 1.3999,
+      "step": 1012500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.178911545850648e-05,
+      "loss": 1.3754,
+      "step": 1013000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.178701549294591e-05,
+      "loss": 1.4067,
+      "step": 1013500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.1784915527385346e-05,
+      "loss": 1.4159,
+      "step": 1014000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.1782815561824786e-05,
+      "loss": 1.3835,
+      "step": 1014500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.178071559626422e-05,
+      "loss": 1.3775,
+      "step": 1015000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.177861563070365e-05,
+      "loss": 1.3898,
+      "step": 1015500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.177651986507421e-05,
+      "loss": 1.3919,
+      "step": 1016000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.177441989951365e-05,
+      "loss": 1.3715,
+      "step": 1016500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.177231993395308e-05,
+      "loss": 1.3596,
+      "step": 1017000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.1770219968392514e-05,
+      "loss": 1.3755,
+      "step": 1017500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.1768120002831954e-05,
+      "loss": 1.3783,
+      "step": 1018000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.176602003727139e-05,
+      "loss": 1.404,
+      "step": 1018500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.176392007171082e-05,
+      "loss": 1.4128,
+      "step": 1019000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.176182010615026e-05,
+      "loss": 1.4049,
+      "step": 1019500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.1759724340520815e-05,
+      "loss": 1.4072,
+      "step": 1020000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.175762857489137e-05,
+      "loss": 1.3885,
+      "step": 1020500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.17555286093308e-05,
+      "loss": 1.3825,
+      "step": 1021000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.175342864377024e-05,
+      "loss": 1.4055,
+      "step": 1021500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.1751328678209676e-05,
+      "loss": 1.3783,
+      "step": 1022000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.174922871264911e-05,
+      "loss": 1.3786,
+      "step": 1022500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.174712874708855e-05,
+      "loss": 1.4117,
+      "step": 1023000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.174502878152798e-05,
+      "loss": 1.3867,
+      "step": 1023500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.1742928815967416e-05,
+      "loss": 1.4044,
+      "step": 1024000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.174082885040686e-05,
+      "loss": 1.3679,
+      "step": 1024500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.173872888484629e-05,
+      "loss": 1.4149,
+      "step": 1025000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.1736633119216844e-05,
+      "loss": 1.3592,
+      "step": 1025500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.173453315365628e-05,
+      "loss": 1.408,
+      "step": 1026000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.173243318809572e-05,
+      "loss": 1.4198,
+      "step": 1026500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.173033322253515e-05,
+      "loss": 1.3621,
+      "step": 1027000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.1728233256974584e-05,
+      "loss": 1.3688,
+      "step": 1027500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.1726133291414025e-05,
+      "loss": 1.4138,
+      "step": 1028000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.172403332585345e-05,
+      "loss": 1.37,
+      "step": 1028500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.1721933360292885e-05,
+      "loss": 1.402,
+      "step": 1029000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.171983759466345e-05,
+      "loss": 1.4203,
+      "step": 1029500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.1717737629102885e-05,
+      "loss": 1.3896,
+      "step": 1030000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.171563766354232e-05,
+      "loss": 1.3662,
+      "step": 1030500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.171354189791287e-05,
+      "loss": 1.44,
+      "step": 1031000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.171144193235231e-05,
+      "loss": 1.384,
+      "step": 1031500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.1709341966791746e-05,
+      "loss": 1.3914,
+      "step": 1032000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.170724200123118e-05,
+      "loss": 1.3873,
+      "step": 1032500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.170514203567062e-05,
+      "loss": 1.4175,
+      "step": 1033000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.1703042070110046e-05,
+      "loss": 1.3952,
+      "step": 1033500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.170094210454948e-05,
+      "loss": 1.4053,
+      "step": 1034000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.169884213898892e-05,
+      "loss": 1.3834,
+      "step": 1034500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.169674637335948e-05,
+      "loss": 1.4075,
+      "step": 1035000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.1694646407798914e-05,
+      "loss": 1.3729,
+      "step": 1035500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.169254644223835e-05,
+      "loss": 1.3867,
+      "step": 1036000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.169044647667778e-05,
+      "loss": 1.4124,
+      "step": 1036500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.1688346511117214e-05,
+      "loss": 1.4109,
+      "step": 1037000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.1686246545556654e-05,
+      "loss": 1.4018,
+      "step": 1037500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.168414657999609e-05,
+      "loss": 1.402,
+      "step": 1038000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.168204661443552e-05,
+      "loss": 1.4015,
+      "step": 1038500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.1679950848806075e-05,
+      "loss": 1.366,
+      "step": 1039000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.1677850883245515e-05,
+      "loss": 1.4079,
+      "step": 1039500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.167575091768495e-05,
+      "loss": 1.368,
+      "step": 1040000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.167365095212438e-05,
+      "loss": 1.4189,
+      "step": 1040500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.167155098656382e-05,
+      "loss": 1.3794,
+      "step": 1041000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.1669455220934376e-05,
+      "loss": 1.3755,
+      "step": 1041500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.166735525537381e-05,
+      "loss": 1.3377,
+      "step": 1042000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.166525528981324e-05,
+      "loss": 1.3978,
+      "step": 1042500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.166315532425268e-05,
+      "loss": 1.4078,
+      "step": 1043000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.1661055358692117e-05,
+      "loss": 1.4194,
+      "step": 1043500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.165895539313155e-05,
+      "loss": 1.368,
+      "step": 1044000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.165685542757099e-05,
+      "loss": 1.3645,
+      "step": 1044500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.165476386187267e-05,
+      "loss": 1.3715,
+      "step": 1045000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.16526638963121e-05,
+      "loss": 1.3751,
+      "step": 1045500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.165056393075153e-05,
+      "loss": 1.4012,
+      "step": 1046000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.164846396519097e-05,
+      "loss": 1.3949,
+      "step": 1046500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.1646363999630405e-05,
+      "loss": 1.3993,
+      "step": 1047000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.164426403406984e-05,
+      "loss": 1.3863,
+      "step": 1047500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.164216406850928e-05,
+      "loss": 1.4035,
+      "step": 1048000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.164006410294871e-05,
+      "loss": 1.3584,
+      "step": 1048500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.1637968337319265e-05,
+      "loss": 1.3973,
+      "step": 1049000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.16358683717587e-05,
+      "loss": 1.3706,
+      "step": 1049500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.163376840619814e-05,
+      "loss": 1.4108,
+      "step": 1050000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.163166844063757e-05,
+      "loss": 1.3854,
+      "step": 1050500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.1629568475077006e-05,
+      "loss": 1.3789,
+      "step": 1051000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.1627468509516446e-05,
+      "loss": 1.3949,
+      "step": 1051500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.162536854395588e-05,
+      "loss": 1.4109,
+      "step": 1052000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.162326857839531e-05,
+      "loss": 1.3556,
+      "step": 1052500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.1621168612834746e-05,
+      "loss": 1.4084,
+      "step": 1053000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.161907284720531e-05,
+      "loss": 1.3938,
+      "step": 1053500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.161697288164474e-05,
+      "loss": 1.3817,
+      "step": 1054000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.1614877116015294e-05,
+      "loss": 1.3785,
+      "step": 1054500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.1612777150454734e-05,
+      "loss": 1.3916,
+      "step": 1055000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.161067718489417e-05,
+      "loss": 1.4128,
+      "step": 1055500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.16085772193336e-05,
+      "loss": 1.3998,
+      "step": 1056000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.160647725377304e-05,
+      "loss": 1.4013,
+      "step": 1056500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.1604381488143595e-05,
+      "loss": 1.3908,
+      "step": 1057000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.160228152258303e-05,
+      "loss": 1.3517,
+      "step": 1057500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.160018155702246e-05,
+      "loss": 1.3747,
+      "step": 1058000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.15980815914619e-05,
+      "loss": 1.4061,
+      "step": 1058500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.1595981625901335e-05,
+      "loss": 1.3756,
+      "step": 1059000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.159388166034077e-05,
+      "loss": 1.3746,
+      "step": 1059500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.159178169478021e-05,
+      "loss": 1.3892,
+      "step": 1060000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.158968592915076e-05,
+      "loss": 1.4025,
+      "step": 1060500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.1587585963590196e-05,
+      "loss": 1.3828,
+      "step": 1061000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.1585485998029636e-05,
+      "loss": 1.3817,
+      "step": 1061500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.158338603246907e-05,
+      "loss": 1.3874,
+      "step": 1062000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.1581286066908497e-05,
+      "loss": 1.4019,
+      "step": 1062500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.157918610134794e-05,
+      "loss": 1.3728,
+      "step": 1063000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.157708613578737e-05,
+      "loss": 1.4161,
+      "step": 1063500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.1574986170226804e-05,
+      "loss": 1.3771,
+      "step": 1064000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.1572890404597364e-05,
+      "loss": 1.413,
+      "step": 1064500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.157079463896792e-05,
+      "loss": 1.3853,
+      "step": 1065000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.156869467340736e-05,
+      "loss": 1.3814,
+      "step": 1065500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.156659470784679e-05,
+      "loss": 1.3648,
+      "step": 1066000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.1564494742286225e-05,
+      "loss": 1.3922,
+      "step": 1066500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.1562394776725665e-05,
+      "loss": 1.3924,
+      "step": 1067000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.156029481116509e-05,
+      "loss": 1.3848,
+      "step": 1067500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.155819904553565e-05,
+      "loss": 1.399,
+      "step": 1068000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.155609907997509e-05,
+      "loss": 1.3675,
+      "step": 1068500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.1553999114414526e-05,
+      "loss": 1.4029,
+      "step": 1069000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.155189914885396e-05,
+      "loss": 1.3658,
+      "step": 1069500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.154979918329339e-05,
+      "loss": 1.3809,
+      "step": 1070000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.1547699217732826e-05,
+      "loss": 1.3865,
+      "step": 1070500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.154559925217226e-05,
+      "loss": 1.3994,
+      "step": 1071000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.154350348654282e-05,
+      "loss": 1.4062,
+      "step": 1071500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.154140352098226e-05,
+      "loss": 1.3787,
+      "step": 1072000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.153930355542169e-05,
+      "loss": 1.3769,
+      "step": 1072500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.153720358986112e-05,
+      "loss": 1.3793,
+      "step": 1073000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.153510362430056e-05,
+      "loss": 1.3987,
+      "step": 1073500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.1533003658739994e-05,
+      "loss": 1.3735,
+      "step": 1074000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.153090369317943e-05,
+      "loss": 1.3841,
+      "step": 1074500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.152880372761887e-05,
+      "loss": 1.3733,
+      "step": 1075000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.15267037620583e-05,
+      "loss": 1.379,
+      "step": 1075500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.1524607996428855e-05,
+      "loss": 1.3643,
+      "step": 1076000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.1522508030868295e-05,
+      "loss": 1.4039,
+      "step": 1076500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.152041226523885e-05,
+      "loss": 1.422,
+      "step": 1077000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.151831229967828e-05,
+      "loss": 1.3997,
+      "step": 1077500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.1516212334117715e-05,
+      "loss": 1.3428,
+      "step": 1078000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.1514112368557156e-05,
+      "loss": 1.4137,
+      "step": 1078500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.1512016602927716e-05,
+      "loss": 1.3951,
+      "step": 1079000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.150991663736714e-05,
+      "loss": 1.4016,
+      "step": 1079500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.1507816671806576e-05,
+      "loss": 1.4085,
+      "step": 1080000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.1505716706246016e-05,
+      "loss": 1.3827,
+      "step": 1080500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.150361674068545e-05,
+      "loss": 1.349,
+      "step": 1081000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.150151677512488e-05,
+      "loss": 1.3886,
+      "step": 1081500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.1499416809564324e-05,
+      "loss": 1.414,
+      "step": 1082000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.149732104393488e-05,
+      "loss": 1.4363,
+      "step": 1082500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.149522107837431e-05,
+      "loss": 1.369,
+      "step": 1083000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.149312111281375e-05,
+      "loss": 1.3952,
+      "step": 1083500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.1491021147253184e-05,
+      "loss": 1.3957,
+      "step": 1084000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.148892118169262e-05,
+      "loss": 1.4046,
+      "step": 1084500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.148682121613206e-05,
+      "loss": 1.3808,
+      "step": 1085000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.148472545050261e-05,
+      "loss": 1.3724,
+      "step": 1085500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.1482625484942045e-05,
+      "loss": 1.402,
+      "step": 1086000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.148052551938148e-05,
+      "loss": 1.4139,
+      "step": 1086500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.147842555382092e-05,
+      "loss": 1.3749,
+      "step": 1087000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.147632558826035e-05,
+      "loss": 1.4002,
+      "step": 1087500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.1474225622699786e-05,
+      "loss": 1.3736,
+      "step": 1088000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.1472125657139226e-05,
+      "loss": 1.3747,
+      "step": 1088500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.147002569157866e-05,
+      "loss": 1.3658,
+      "step": 1089000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.146792992594921e-05,
+      "loss": 1.3934,
+      "step": 1089500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.146582996038865e-05,
+      "loss": 1.3778,
+      "step": 1090000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.146373419475921e-05,
+      "loss": 1.3887,
+      "step": 1090500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.146163422919864e-05,
+      "loss": 1.3824,
+      "step": 1091000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.1459534263638074e-05,
+      "loss": 1.41,
+      "step": 1091500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.1457434298077514e-05,
+      "loss": 1.4074,
+      "step": 1092000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.145533853244807e-05,
+      "loss": 1.4086,
+      "step": 1092500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.14532385668875e-05,
+      "loss": 1.3811,
+      "step": 1093000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1451138601326934e-05,
+      "loss": 1.3733,
+      "step": 1093500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1449038635766375e-05,
+      "loss": 1.4263,
+      "step": 1094000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.144693867020581e-05,
+      "loss": 1.3807,
+      "step": 1094500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.144483870464524e-05,
+      "loss": 1.3812,
+      "step": 1095000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.144273873908468e-05,
+      "loss": 1.3496,
+      "step": 1095500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1440638773524115e-05,
+      "loss": 1.3953,
+      "step": 1096000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.143853880796355e-05,
+      "loss": 1.3786,
+      "step": 1096500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.143644304233411e-05,
+      "loss": 1.3497,
+      "step": 1097000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.143434307677354e-05,
+      "loss": 1.3816,
+      "step": 1097500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1432243111212976e-05,
+      "loss": 1.3858,
+      "step": 1098000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1430143145652416e-05,
+      "loss": 1.3926,
+      "step": 1098500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.142804318009184e-05,
+      "loss": 1.4155,
+      "step": 1099000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.14259474144624e-05,
+      "loss": 1.378,
+      "step": 1099500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.142384744890184e-05,
+      "loss": 1.3969,
+      "step": 1100000
+    },
+    {
+      "epoch": 0.66,
+      "eval_loss": 1.3153480291366577,
+      "eval_runtime": 1111.5226,
+      "eval_samples_per_second": 473.873,
+      "eval_steps_per_second": 78.979,
+      "step": 1100000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.142174748334128e-05,
+      "loss": 1.3804,
+      "step": 1100500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.141964751778071e-05,
+      "loss": 1.3754,
+      "step": 1101000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.141754755222014e-05,
+      "loss": 1.3715,
+      "step": 1101500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.141544758665958e-05,
+      "loss": 1.3656,
+      "step": 1102000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.141335182103014e-05,
+      "loss": 1.3822,
+      "step": 1102500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.141125185546957e-05,
+      "loss": 1.3963,
+      "step": 1103000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1409151889909005e-05,
+      "loss": 1.3675,
+      "step": 1103500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.140705192434844e-05,
+      "loss": 1.415,
+      "step": 1104000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.140495195878787e-05,
+      "loss": 1.3495,
+      "step": 1104500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.140285199322731e-05,
+      "loss": 1.3917,
+      "step": 1105000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1400752027666745e-05,
+      "loss": 1.3534,
+      "step": 1105500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1398656262037306e-05,
+      "loss": 1.4196,
+      "step": 1106000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.139655629647673e-05,
+      "loss": 1.3883,
+      "step": 1106500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.139445633091617e-05,
+      "loss": 1.3661,
+      "step": 1107000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1392356365355606e-05,
+      "loss": 1.3944,
+      "step": 1107500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.139025639979504e-05,
+      "loss": 1.3743,
+      "step": 1108000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.138815643423448e-05,
+      "loss": 1.3976,
+      "step": 1108500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.138605646867391e-05,
+      "loss": 1.4238,
+      "step": 1109000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1383956503113347e-05,
+      "loss": 1.3618,
+      "step": 1109500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.138186493741503e-05,
+      "loss": 1.3853,
+      "step": 1110000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.137976497185446e-05,
+      "loss": 1.36,
+      "step": 1110500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1377665006293894e-05,
+      "loss": 1.4155,
+      "step": 1111000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1375569240664454e-05,
+      "loss": 1.3881,
+      "step": 1111500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.137346927510389e-05,
+      "loss": 1.3841,
+      "step": 1112000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.137136930954333e-05,
+      "loss": 1.3756,
+      "step": 1112500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.136926934398276e-05,
+      "loss": 1.4265,
+      "step": 1113000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.136716937842219e-05,
+      "loss": 1.3672,
+      "step": 1113500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.136506941286163e-05,
+      "loss": 1.3761,
+      "step": 1114000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.136296944730106e-05,
+      "loss": 1.3674,
+      "step": 1114500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1360869481740495e-05,
+      "loss": 1.3773,
+      "step": 1115000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1358769516179935e-05,
+      "loss": 1.3895,
+      "step": 1115500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.135666955061937e-05,
+      "loss": 1.3748,
+      "step": 1116000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.13545695850588e-05,
+      "loss": 1.4154,
+      "step": 1116500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.135246961949824e-05,
+      "loss": 1.401,
+      "step": 1117000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1350373853868796e-05,
+      "loss": 1.4003,
+      "step": 1117500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.134827388830823e-05,
+      "loss": 1.3916,
+      "step": 1118000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.134617392274766e-05,
+      "loss": 1.3667,
+      "step": 1118500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1344078157118223e-05,
+      "loss": 1.3732,
+      "step": 1119000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1341982391488784e-05,
+      "loss": 1.3752,
+      "step": 1119500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.133988242592822e-05,
+      "loss": 1.3812,
+      "step": 1120000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1337782460367644e-05,
+      "loss": 1.4159,
+      "step": 1120500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1335682494807084e-05,
+      "loss": 1.3821,
+      "step": 1121000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.133358252924652e-05,
+      "loss": 1.3776,
+      "step": 1121500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.133148256368595e-05,
+      "loss": 1.4106,
+      "step": 1122000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.132938259812539e-05,
+      "loss": 1.3728,
+      "step": 1122500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1327282632564825e-05,
+      "loss": 1.4052,
+      "step": 1123000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.132518266700426e-05,
+      "loss": 1.3816,
+      "step": 1123500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.13230827014437e-05,
+      "loss": 1.405,
+      "step": 1124000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.132098693581425e-05,
+      "loss": 1.3953,
+      "step": 1124500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1318886970253686e-05,
+      "loss": 1.3854,
+      "step": 1125000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.131678700469312e-05,
+      "loss": 1.3799,
+      "step": 1125500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.131468703913256e-05,
+      "loss": 1.3544,
+      "step": 1126000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.131258707357199e-05,
+      "loss": 1.3627,
+      "step": 1126500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1310487108011426e-05,
+      "loss": 1.3785,
+      "step": 1127000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1308387142450866e-05,
+      "loss": 1.4006,
+      "step": 1127500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.130629137682142e-05,
+      "loss": 1.3694,
+      "step": 1128000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1304191411260853e-05,
+      "loss": 1.3698,
+      "step": 1128500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1302091445700294e-05,
+      "loss": 1.4122,
+      "step": 1129000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.129999148013973e-05,
+      "loss": 1.3834,
+      "step": 1129500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.129789151457916e-05,
+      "loss": 1.3871,
+      "step": 1130000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1295795748949714e-05,
+      "loss": 1.3673,
+      "step": 1130500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1293695783389154e-05,
+      "loss": 1.4203,
+      "step": 1131000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.129159581782859e-05,
+      "loss": 1.3797,
+      "step": 1131500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.128949585226802e-05,
+      "loss": 1.3973,
+      "step": 1132000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1287400086638575e-05,
+      "loss": 1.3759,
+      "step": 1132500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1285300121078015e-05,
+      "loss": 1.4024,
+      "step": 1133000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.128320435544857e-05,
+      "loss": 1.3668,
+      "step": 1133500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1281104389888e-05,
+      "loss": 1.3785,
+      "step": 1134000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.127900442432744e-05,
+      "loss": 1.3942,
+      "step": 1134500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1276904458766876e-05,
+      "loss": 1.3998,
+      "step": 1135000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.127480449320631e-05,
+      "loss": 1.3801,
+      "step": 1135500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.127270452764575e-05,
+      "loss": 1.375,
+      "step": 1136000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.127060456208518e-05,
+      "loss": 1.3782,
+      "step": 1136500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1268508796455737e-05,
+      "loss": 1.3587,
+      "step": 1137000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.126640883089517e-05,
+      "loss": 1.3792,
+      "step": 1137500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.126430886533461e-05,
+      "loss": 1.3931,
+      "step": 1138000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1262208899774044e-05,
+      "loss": 1.3647,
+      "step": 1138500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.126010893421348e-05,
+      "loss": 1.3771,
+      "step": 1139000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.125800896865292e-05,
+      "loss": 1.4079,
+      "step": 1139500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.125590900309235e-05,
+      "loss": 1.3752,
+      "step": 1140000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.125380903753178e-05,
+      "loss": 1.3997,
+      "step": 1140500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.125170907197122e-05,
+      "loss": 1.4085,
+      "step": 1141000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.124960910641065e-05,
+      "loss": 1.3524,
+      "step": 1141500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.124750914085009e-05,
+      "loss": 1.3721,
+      "step": 1142000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1245409175289525e-05,
+      "loss": 1.3779,
+      "step": 1142500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.124331340966008e-05,
+      "loss": 1.3825,
+      "step": 1143000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.124121344409951e-05,
+      "loss": 1.4193,
+      "step": 1143500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.123911347853895e-05,
+      "loss": 1.3746,
+      "step": 1144000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.1237013512978386e-05,
+      "loss": 1.4017,
+      "step": 1144500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.123491354741782e-05,
+      "loss": 1.3472,
+      "step": 1145000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.123281358185726e-05,
+      "loss": 1.3713,
+      "step": 1145500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.123072201615893e-05,
+      "loss": 1.3892,
+      "step": 1146000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.122862205059837e-05,
+      "loss": 1.3863,
+      "step": 1146500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.122652208503781e-05,
+      "loss": 1.3757,
+      "step": 1147000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.1224422119477233e-05,
+      "loss": 1.3846,
+      "step": 1147500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.1222322153916674e-05,
+      "loss": 1.3781,
+      "step": 1148000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.122022218835611e-05,
+      "loss": 1.3856,
+      "step": 1148500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.121812222279555e-05,
+      "loss": 1.3714,
+      "step": 1149000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.121602225723498e-05,
+      "loss": 1.3471,
+      "step": 1149500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.1213922291674414e-05,
+      "loss": 1.3798,
+      "step": 1150000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.1211822326113855e-05,
+      "loss": 1.4012,
+      "step": 1150500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.120972236055329e-05,
+      "loss": 1.3677,
+      "step": 1151000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.120762239499272e-05,
+      "loss": 1.3716,
+      "step": 1151500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.1205526629363275e-05,
+      "loss": 1.3841,
+      "step": 1152000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.120343086373383e-05,
+      "loss": 1.3865,
+      "step": 1152500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.120133089817327e-05,
+      "loss": 1.3806,
+      "step": 1153000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.11992309326127e-05,
+      "loss": 1.3734,
+      "step": 1153500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.1197130967052136e-05,
+      "loss": 1.3676,
+      "step": 1154000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.1195031001491576e-05,
+      "loss": 1.3616,
+      "step": 1154500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.119293103593101e-05,
+      "loss": 1.3571,
+      "step": 1155000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.119083107037044e-05,
+      "loss": 1.3581,
+      "step": 1155500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.1188735304741e-05,
+      "loss": 1.4098,
+      "step": 1156000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.118663533918044e-05,
+      "loss": 1.3855,
+      "step": 1156500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.118453537361987e-05,
+      "loss": 1.3983,
+      "step": 1157000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.1182439607990424e-05,
+      "loss": 1.3756,
+      "step": 1157500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.1180339642429864e-05,
+      "loss": 1.3781,
+      "step": 1158000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.11782396768693e-05,
+      "loss": 1.3963,
+      "step": 1158500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.117613971130873e-05,
+      "loss": 1.3835,
+      "step": 1159000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.117403974574817e-05,
+      "loss": 1.3928,
+      "step": 1159500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.1171939780187605e-05,
+      "loss": 1.3423,
+      "step": 1160000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.116983981462704e-05,
+      "loss": 1.3655,
+      "step": 1160500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.116773984906648e-05,
+      "loss": 1.3685,
+      "step": 1161000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.116563988350591e-05,
+      "loss": 1.3539,
+      "step": 1161500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.1163539917945345e-05,
+      "loss": 1.3701,
+      "step": 1162000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.11614441523159e-05,
+      "loss": 1.3552,
+      "step": 1162500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.115934418675534e-05,
+      "loss": 1.4087,
+      "step": 1163000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.115724422119477e-05,
+      "loss": 1.3697,
+      "step": 1163500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.1155148455565326e-05,
+      "loss": 1.3504,
+      "step": 1164000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.1153048490004766e-05,
+      "loss": 1.3818,
+      "step": 1164500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.11509485244442e-05,
+      "loss": 1.3863,
+      "step": 1165000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.114884855888363e-05,
+      "loss": 1.3985,
+      "step": 1165500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.1146748593323073e-05,
+      "loss": 1.3701,
+      "step": 1166000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.114464862776251e-05,
+      "loss": 1.403,
+      "step": 1166500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.114254866220194e-05,
+      "loss": 1.3908,
+      "step": 1167000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.1140448696641374e-05,
+      "loss": 1.365,
+      "step": 1167500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.113834873108081e-05,
+      "loss": 1.3751,
+      "step": 1168000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.113625296545137e-05,
+      "loss": 1.4,
+      "step": 1168500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.11341529998908e-05,
+      "loss": 1.3502,
+      "step": 1169000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.113205303433024e-05,
+      "loss": 1.3935,
+      "step": 1169500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.112995306876967e-05,
+      "loss": 1.3827,
+      "step": 1170000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.112785730314023e-05,
+      "loss": 1.3787,
+      "step": 1170500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.112575733757966e-05,
+      "loss": 1.3628,
+      "step": 1171000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.11236573720191e-05,
+      "loss": 1.3505,
+      "step": 1171500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.1121557406458535e-05,
+      "loss": 1.3781,
+      "step": 1172000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.111945744089797e-05,
+      "loss": 1.3364,
+      "step": 1172500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.11173574753374e-05,
+      "loss": 1.3908,
+      "step": 1173000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.1115257509776836e-05,
+      "loss": 1.3812,
+      "step": 1173500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.1113161744147396e-05,
+      "loss": 1.4128,
+      "step": 1174000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.111106177858683e-05,
+      "loss": 1.3733,
+      "step": 1174500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.110896181302626e-05,
+      "loss": 1.3805,
+      "step": 1175000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.1106861847465697e-05,
+      "loss": 1.3571,
+      "step": 1175500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.110476188190514e-05,
+      "loss": 1.3847,
+      "step": 1176000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.110266191634457e-05,
+      "loss": 1.4097,
+      "step": 1176500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.1100566150715124e-05,
+      "loss": 1.4068,
+      "step": 1177000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.109846618515456e-05,
+      "loss": 1.3897,
+      "step": 1177500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.1096366219594e-05,
+      "loss": 1.3688,
+      "step": 1178000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.109426625403343e-05,
+      "loss": 1.3812,
+      "step": 1178500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.1092166288472864e-05,
+      "loss": 1.3976,
+      "step": 1179000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.1090066322912305e-05,
+      "loss": 1.3727,
+      "step": 1179500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.108796635735174e-05,
+      "loss": 1.3653,
+      "step": 1180000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.108587059172229e-05,
+      "loss": 1.3896,
+      "step": 1180500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.108377062616173e-05,
+      "loss": 1.3743,
+      "step": 1181000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.1081670660601165e-05,
+      "loss": 1.3712,
+      "step": 1181500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.10795706950406e-05,
+      "loss": 1.3768,
+      "step": 1182000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.107747072948004e-05,
+      "loss": 1.3707,
+      "step": 1182500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.107537076391947e-05,
+      "loss": 1.386,
+      "step": 1183000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.1073274998290026e-05,
+      "loss": 1.3595,
+      "step": 1183500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.107117503272946e-05,
+      "loss": 1.3561,
+      "step": 1184000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.10690750671689e-05,
+      "loss": 1.3615,
+      "step": 1184500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.106697510160833e-05,
+      "loss": 1.3894,
+      "step": 1185000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.106487513604777e-05,
+      "loss": 1.3899,
+      "step": 1185500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.106277517048721e-05,
+      "loss": 1.3547,
+      "step": 1186000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.106067940485776e-05,
+      "loss": 1.3683,
+      "step": 1186500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.1058579439297194e-05,
+      "loss": 1.3676,
+      "step": 1187000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.105647947373663e-05,
+      "loss": 1.3993,
+      "step": 1187500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.105437950817607e-05,
+      "loss": 1.3662,
+      "step": 1188000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.10522795426155e-05,
+      "loss": 1.3727,
+      "step": 1188500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.1050183776986055e-05,
+      "loss": 1.4103,
+      "step": 1189000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.1048083811425495e-05,
+      "loss": 1.3652,
+      "step": 1189500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.104598384586493e-05,
+      "loss": 1.3977,
+      "step": 1190000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.104388388030436e-05,
+      "loss": 1.3614,
+      "step": 1190500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.10417839147438e-05,
+      "loss": 1.414,
+      "step": 1191000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.1039683949183236e-05,
+      "loss": 1.3397,
+      "step": 1191500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.103758818355379e-05,
+      "loss": 1.3673,
+      "step": 1192000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.103548821799322e-05,
+      "loss": 1.3586,
+      "step": 1192500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.103338825243266e-05,
+      "loss": 1.3636,
+      "step": 1193000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.1031288286872096e-05,
+      "loss": 1.3844,
+      "step": 1193500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.102918832131153e-05,
+      "loss": 1.3879,
+      "step": 1194000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.102709255568209e-05,
+      "loss": 1.3613,
+      "step": 1194500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.1024992590121524e-05,
+      "loss": 1.3877,
+      "step": 1195000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.102289262456096e-05,
+      "loss": 1.3694,
+      "step": 1195500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.102079685893151e-05,
+      "loss": 1.3863,
+      "step": 1196000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.101869689337095e-05,
+      "loss": 1.3787,
+      "step": 1196500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.1016596927810384e-05,
+      "loss": 1.3816,
+      "step": 1197000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.101449696224982e-05,
+      "loss": 1.3487,
+      "step": 1197500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.101240119662037e-05,
+      "loss": 1.3569,
+      "step": 1198000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.101030123105981e-05,
+      "loss": 1.3908,
+      "step": 1198500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.1008201265499245e-05,
+      "loss": 1.3867,
+      "step": 1199000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.100610129993868e-05,
+      "loss": 1.3663,
+      "step": 1199500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.100400133437812e-05,
+      "loss": 1.3723,
+      "step": 1200000
+    },
+    {
+      "epoch": 0.72,
+      "eval_loss": 1.3151705265045166,
+      "eval_runtime": 1104.7712,
+      "eval_samples_per_second": 476.768,
+      "eval_steps_per_second": 79.462,
+      "step": 1200000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.100190136881755e-05,
+      "loss": 1.3804,
+      "step": 1200500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.0999801403256986e-05,
+      "loss": 1.3675,
+      "step": 1201000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.099770143769642e-05,
+      "loss": 1.3808,
+      "step": 1201500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.099560147213585e-05,
+      "loss": 1.3655,
+      "step": 1202000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.099350150657529e-05,
+      "loss": 1.3646,
+      "step": 1202500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.0991401541014726e-05,
+      "loss": 1.377,
+      "step": 1203000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.098930157545416e-05,
+      "loss": 1.3981,
+      "step": 1203500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.09872016098936e-05,
+      "loss": 1.3675,
+      "step": 1204000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.0985105844264154e-05,
+      "loss": 1.3753,
+      "step": 1204500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.098300587870359e-05,
+      "loss": 1.3472,
+      "step": 1205000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.098090591314302e-05,
+      "loss": 1.357,
+      "step": 1205500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.097880594758246e-05,
+      "loss": 1.3797,
+      "step": 1206000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.0976705982021894e-05,
+      "loss": 1.4025,
+      "step": 1206500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.097461021639245e-05,
+      "loss": 1.375,
+      "step": 1207000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.097251445076301e-05,
+      "loss": 1.3654,
+      "step": 1207500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.097041448520244e-05,
+      "loss": 1.3685,
+      "step": 1208000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.0968318719573e-05,
+      "loss": 1.3489,
+      "step": 1208500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.0966218754012435e-05,
+      "loss": 1.3542,
+      "step": 1209000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.096411878845187e-05,
+      "loss": 1.3754,
+      "step": 1209500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.096201882289131e-05,
+      "loss": 1.391,
+      "step": 1210000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.095991885733074e-05,
+      "loss": 1.379,
+      "step": 1210500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.095781889177017e-05,
+      "loss": 1.3446,
+      "step": 1211000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.095571892620961e-05,
+      "loss": 1.3724,
+      "step": 1211500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.095361896064904e-05,
+      "loss": 1.3668,
+      "step": 1212000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.0951518995088476e-05,
+      "loss": 1.3661,
+      "step": 1212500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.0949419029527917e-05,
+      "loss": 1.3654,
+      "step": 1213000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.094731906396735e-05,
+      "loss": 1.3541,
+      "step": 1213500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.0945219098406784e-05,
+      "loss": 1.3598,
+      "step": 1214000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.0943119132846224e-05,
+      "loss": 1.3503,
+      "step": 1214500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.094101916728566e-05,
+      "loss": 1.3971,
+      "step": 1215000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.093892340165621e-05,
+      "loss": 1.3821,
+      "step": 1215500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.0936823436095644e-05,
+      "loss": 1.377,
+      "step": 1216000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.0934723470535084e-05,
+      "loss": 1.356,
+      "step": 1216500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.093262350497452e-05,
+      "loss": 1.3789,
+      "step": 1217000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.093052353941395e-05,
+      "loss": 1.3561,
+      "step": 1217500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.092842357385339e-05,
+      "loss": 1.3694,
+      "step": 1218000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.0926327808223945e-05,
+      "loss": 1.3678,
+      "step": 1218500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.092422784266338e-05,
+      "loss": 1.3942,
+      "step": 1219000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.092212787710282e-05,
+      "loss": 1.3781,
+      "step": 1219500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.092002791154225e-05,
+      "loss": 1.4007,
+      "step": 1220000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.0917927945981686e-05,
+      "loss": 1.3577,
+      "step": 1220500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.091583218035224e-05,
+      "loss": 1.372,
+      "step": 1221000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.091373641472279e-05,
+      "loss": 1.3697,
+      "step": 1221500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.091164064909335e-05,
+      "loss": 1.3779,
+      "step": 1222000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.0909540683532794e-05,
+      "loss": 1.3555,
+      "step": 1222500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.090744071797222e-05,
+      "loss": 1.3911,
+      "step": 1223000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.090534075241166e-05,
+      "loss": 1.3517,
+      "step": 1223500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.0903240786851094e-05,
+      "loss": 1.3597,
+      "step": 1224000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.0901145021221654e-05,
+      "loss": 1.3848,
+      "step": 1224500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.089904505566109e-05,
+      "loss": 1.3372,
+      "step": 1225000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.089694509010052e-05,
+      "loss": 1.4007,
+      "step": 1225500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.0894845124539955e-05,
+      "loss": 1.3585,
+      "step": 1226000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.089274515897939e-05,
+      "loss": 1.3632,
+      "step": 1226500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.089064519341883e-05,
+      "loss": 1.3868,
+      "step": 1227000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.088854522785826e-05,
+      "loss": 1.3611,
+      "step": 1227500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.0886445262297695e-05,
+      "loss": 1.3871,
+      "step": 1228000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.088434949666825e-05,
+      "loss": 1.3915,
+      "step": 1228500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.088224953110769e-05,
+      "loss": 1.3762,
+      "step": 1229000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.088014956554712e-05,
+      "loss": 1.3556,
+      "step": 1229500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.0878049599986556e-05,
+      "loss": 1.3397,
+      "step": 1230000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.0875949634425996e-05,
+      "loss": 1.38,
+      "step": 1230500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.087384966886543e-05,
+      "loss": 1.3747,
+      "step": 1231000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.087174970330486e-05,
+      "loss": 1.359,
+      "step": 1231500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.08696497377443e-05,
+      "loss": 1.3434,
+      "step": 1232000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.086754977218374e-05,
+      "loss": 1.3631,
+      "step": 1232500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.086544980662317e-05,
+      "loss": 1.3705,
+      "step": 1233000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.0863349841062604e-05,
+      "loss": 1.3553,
+      "step": 1233500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.086124987550204e-05,
+      "loss": 1.3719,
+      "step": 1234000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.08591541098726e-05,
+      "loss": 1.3385,
+      "step": 1234500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.085705414431204e-05,
+      "loss": 1.3611,
+      "step": 1235000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.0854954178751464e-05,
+      "loss": 1.3961,
+      "step": 1235500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.08528542131909e-05,
+      "loss": 1.3432,
+      "step": 1236000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.085075424763034e-05,
+      "loss": 1.3536,
+      "step": 1236500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.084866268193201e-05,
+      "loss": 1.3644,
+      "step": 1237000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.084656271637145e-05,
+      "loss": 1.3721,
+      "step": 1237500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.0844466950742006e-05,
+      "loss": 1.4002,
+      "step": 1238000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.084236698518144e-05,
+      "loss": 1.3527,
+      "step": 1238500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.084026701962088e-05,
+      "loss": 1.3735,
+      "step": 1239000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.083816705406031e-05,
+      "loss": 1.3966,
+      "step": 1239500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.0836067088499746e-05,
+      "loss": 1.3662,
+      "step": 1240000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.0833967122939187e-05,
+      "loss": 1.3885,
+      "step": 1240500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.083186715737862e-05,
+      "loss": 1.3659,
+      "step": 1241000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.0829767191818053e-05,
+      "loss": 1.3429,
+      "step": 1241500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.0827667226257494e-05,
+      "loss": 1.3731,
+      "step": 1242000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.082556726069693e-05,
+      "loss": 1.3712,
+      "step": 1242500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0823467295136354e-05,
+      "loss": 1.3826,
+      "step": 1243000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0821371529506914e-05,
+      "loss": 1.3889,
+      "step": 1243500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0819271563946354e-05,
+      "loss": 1.3738,
+      "step": 1244000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.081717159838579e-05,
+      "loss": 1.3506,
+      "step": 1244500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.081507163282522e-05,
+      "loss": 1.3725,
+      "step": 1245000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0812971667264655e-05,
+      "loss": 1.3725,
+      "step": 1245500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.081087170170409e-05,
+      "loss": 1.3702,
+      "step": 1246000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.080877593607465e-05,
+      "loss": 1.3648,
+      "step": 1246500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.080667597051409e-05,
+      "loss": 1.3978,
+      "step": 1247000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0804576004953516e-05,
+      "loss": 1.3837,
+      "step": 1247500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.080247603939295e-05,
+      "loss": 1.3811,
+      "step": 1248000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.080037607383239e-05,
+      "loss": 1.3716,
+      "step": 1248500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.079827610827182e-05,
+      "loss": 1.3669,
+      "step": 1249000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0796176142711256e-05,
+      "loss": 1.3718,
+      "step": 1249500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0794076177150696e-05,
+      "loss": 1.3555,
+      "step": 1250000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.079198041152125e-05,
+      "loss": 1.3691,
+      "step": 1250500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0789880445960683e-05,
+      "loss": 1.3493,
+      "step": 1251000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.078778048040012e-05,
+      "loss": 1.4041,
+      "step": 1251500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.078568471477068e-05,
+      "loss": 1.4076,
+      "step": 1252000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.078358474921011e-05,
+      "loss": 1.3565,
+      "step": 1252500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0781484783649544e-05,
+      "loss": 1.3829,
+      "step": 1253000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0779384818088984e-05,
+      "loss": 1.3445,
+      "step": 1253500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.077728485252842e-05,
+      "loss": 1.3654,
+      "step": 1254000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.077518488696785e-05,
+      "loss": 1.3696,
+      "step": 1254500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0773089121338405e-05,
+      "loss": 1.3649,
+      "step": 1255000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0770989155777845e-05,
+      "loss": 1.3795,
+      "step": 1255500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.076888919021728e-05,
+      "loss": 1.3856,
+      "step": 1256000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.076678922465671e-05,
+      "loss": 1.3688,
+      "step": 1256500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.076468925909615e-05,
+      "loss": 1.346,
+      "step": 1257000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0762593493466706e-05,
+      "loss": 1.3966,
+      "step": 1257500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.076049352790614e-05,
+      "loss": 1.36,
+      "step": 1258000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.075839356234557e-05,
+      "loss": 1.3862,
+      "step": 1258500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.075629359678501e-05,
+      "loss": 1.3487,
+      "step": 1259000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.0754197831155567e-05,
+      "loss": 1.3504,
+      "step": 1259500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.0752097865595e-05,
+      "loss": 1.4048,
+      "step": 1260000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.074999790003444e-05,
+      "loss": 1.3684,
+      "step": 1260500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.0747902134405e-05,
+      "loss": 1.3706,
+      "step": 1261000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.0745802168844434e-05,
+      "loss": 1.327,
+      "step": 1261500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.074370220328386e-05,
+      "loss": 1.3419,
+      "step": 1262000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.07416022377233e-05,
+      "loss": 1.3675,
+      "step": 1262500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.0739502272162734e-05,
+      "loss": 1.3986,
+      "step": 1263000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.073740230660217e-05,
+      "loss": 1.388,
+      "step": 1263500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.073530234104161e-05,
+      "loss": 1.3511,
+      "step": 1264000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.073320237548104e-05,
+      "loss": 1.3755,
+      "step": 1264500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.0731102409920475e-05,
+      "loss": 1.3551,
+      "step": 1265000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.072900664429103e-05,
+      "loss": 1.358,
+      "step": 1265500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.072690667873047e-05,
+      "loss": 1.3489,
+      "step": 1266000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.07248067131699e-05,
+      "loss": 1.3711,
+      "step": 1266500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.0722706747609336e-05,
+      "loss": 1.3771,
+      "step": 1267000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.0720606782048776e-05,
+      "loss": 1.351,
+      "step": 1267500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.071851101641933e-05,
+      "loss": 1.3494,
+      "step": 1268000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.071641525078989e-05,
+      "loss": 1.3627,
+      "step": 1268500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.0714315285229317e-05,
+      "loss": 1.3813,
+      "step": 1269000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.071221531966876e-05,
+      "loss": 1.3699,
+      "step": 1269500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.071011535410819e-05,
+      "loss": 1.3926,
+      "step": 1270000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.0708015388547624e-05,
+      "loss": 1.3404,
+      "step": 1270500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.0705915422987064e-05,
+      "loss": 1.3685,
+      "step": 1271000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.07038154574265e-05,
+      "loss": 1.359,
+      "step": 1271500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.070171549186593e-05,
+      "loss": 1.3457,
+      "step": 1272000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.069961552630537e-05,
+      "loss": 1.3798,
+      "step": 1272500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.0697519760675925e-05,
+      "loss": 1.3773,
+      "step": 1273000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.069541979511536e-05,
+      "loss": 1.3593,
+      "step": 1273500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.069331982955479e-05,
+      "loss": 1.3903,
+      "step": 1274000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.069121986399423e-05,
+      "loss": 1.3781,
+      "step": 1274500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.0689119898433665e-05,
+      "loss": 1.3771,
+      "step": 1275000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.06870199328731e-05,
+      "loss": 1.3793,
+      "step": 1275500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.068492416724366e-05,
+      "loss": 1.3676,
+      "step": 1276000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.068282420168309e-05,
+      "loss": 1.3566,
+      "step": 1276500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.0680724236122526e-05,
+      "loss": 1.3834,
+      "step": 1277000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.0678632670424207e-05,
+      "loss": 1.3827,
+      "step": 1277500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.067653270486364e-05,
+      "loss": 1.3837,
+      "step": 1278000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.0674432739303073e-05,
+      "loss": 1.3729,
+      "step": 1278500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.067233277374251e-05,
+      "loss": 1.3562,
+      "step": 1279000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.067023280818194e-05,
+      "loss": 1.3532,
+      "step": 1279500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.066813284262138e-05,
+      "loss": 1.3822,
+      "step": 1280000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.0666032877060814e-05,
+      "loss": 1.3682,
+      "step": 1280500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.066393291150025e-05,
+      "loss": 1.3506,
+      "step": 1281000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.066183294593969e-05,
+      "loss": 1.3694,
+      "step": 1281500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.065973298037912e-05,
+      "loss": 1.3946,
+      "step": 1282000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.0657633014818555e-05,
+      "loss": 1.3872,
+      "step": 1282500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.0655533049257995e-05,
+      "loss": 1.3448,
+      "step": 1283000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.065343728362855e-05,
+      "loss": 1.3682,
+      "step": 1283500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.065133731806798e-05,
+      "loss": 1.4006,
+      "step": 1284000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.064923735250742e-05,
+      "loss": 1.3558,
+      "step": 1284500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.0647137386946856e-05,
+      "loss": 1.3758,
+      "step": 1285000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.064503742138629e-05,
+      "loss": 1.3791,
+      "step": 1285500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.064293745582573e-05,
+      "loss": 1.3498,
+      "step": 1286000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.0640837490265156e-05,
+      "loss": 1.3765,
+      "step": 1286500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.063873752470459e-05,
+      "loss": 1.3596,
+      "step": 1287000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.063664175907515e-05,
+      "loss": 1.3406,
+      "step": 1287500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.063454179351459e-05,
+      "loss": 1.3448,
+      "step": 1288000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.0632441827954024e-05,
+      "loss": 1.3607,
+      "step": 1288500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.063034186239345e-05,
+      "loss": 1.3638,
+      "step": 1289000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.062824609676401e-05,
+      "loss": 1.3756,
+      "step": 1289500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.062614613120345e-05,
+      "loss": 1.3478,
+      "step": 1290000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.0624050365574004e-05,
+      "loss": 1.3888,
+      "step": 1290500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.062195040001344e-05,
+      "loss": 1.3328,
+      "step": 1291000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.061985043445288e-05,
+      "loss": 1.3895,
+      "step": 1291500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.061775046889231e-05,
+      "loss": 1.3858,
+      "step": 1292000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.0615650503331745e-05,
+      "loss": 1.3637,
+      "step": 1292500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.0613550537771185e-05,
+      "loss": 1.3533,
+      "step": 1293000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.061145057221061e-05,
+      "loss": 1.3716,
+      "step": 1293500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.0609350606650045e-05,
+      "loss": 1.3715,
+      "step": 1294000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.0607254841020606e-05,
+      "loss": 1.3417,
+      "step": 1294500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.0605154875460046e-05,
+      "loss": 1.3617,
+      "step": 1295000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.06030591098306e-05,
+      "loss": 1.358,
+      "step": 1295500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.060095914427003e-05,
+      "loss": 1.353,
+      "step": 1296000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.0598859178709466e-05,
+      "loss": 1.3714,
+      "step": 1296500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.059675921314891e-05,
+      "loss": 1.3416,
+      "step": 1297000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.059465924758834e-05,
+      "loss": 1.3545,
+      "step": 1297500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.0592563481958894e-05,
+      "loss": 1.3627,
+      "step": 1298000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.0590463516398334e-05,
+      "loss": 1.3899,
+      "step": 1298500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.058836775076889e-05,
+      "loss": 1.3737,
+      "step": 1299000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.058626778520832e-05,
+      "loss": 1.3628,
+      "step": 1299500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.0584167819647754e-05,
+      "loss": 1.3439,
+      "step": 1300000
+    },
+    {
+      "epoch": 0.78,
+      "eval_loss": 1.3037538528442383,
+      "eval_runtime": 1107.3433,
+      "eval_samples_per_second": 475.661,
+      "eval_steps_per_second": 79.277,
+      "step": 1300000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.0582067854087195e-05,
+      "loss": 1.344,
+      "step": 1300500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.057996788852663e-05,
+      "loss": 1.3935,
+      "step": 1301000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.057786792296606e-05,
+      "loss": 1.3962,
+      "step": 1301500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.0575772157336615e-05,
+      "loss": 1.3614,
+      "step": 1302000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.0573672191776055e-05,
+      "loss": 1.3494,
+      "step": 1302500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.057157222621549e-05,
+      "loss": 1.3693,
+      "step": 1303000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.056947226065492e-05,
+      "loss": 1.3903,
+      "step": 1303500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.056737229509436e-05,
+      "loss": 1.3436,
+      "step": 1304000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.0565272329533796e-05,
+      "loss": 1.3662,
+      "step": 1304500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.056317236397323e-05,
+      "loss": 1.3452,
+      "step": 1305000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.056107659834379e-05,
+      "loss": 1.3836,
+      "step": 1305500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.055897663278322e-05,
+      "loss": 1.3745,
+      "step": 1306000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.055687666722266e-05,
+      "loss": 1.3352,
+      "step": 1306500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.05547767016621e-05,
+      "loss": 1.3591,
+      "step": 1307000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.055267673610153e-05,
+      "loss": 1.3743,
+      "step": 1307500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.055057677054096e-05,
+      "loss": 1.3339,
+      "step": 1308000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.054848520484264e-05,
+      "loss": 1.363,
+      "step": 1308500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.054638523928207e-05,
+      "loss": 1.3596,
+      "step": 1309000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.054428527372151e-05,
+      "loss": 1.356,
+      "step": 1309500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.0542185308160945e-05,
+      "loss": 1.3246,
+      "step": 1310000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.054008534260038e-05,
+      "loss": 1.3558,
+      "step": 1310500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.053798537703982e-05,
+      "loss": 1.3723,
+      "step": 1311000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.053588541147925e-05,
+      "loss": 1.379,
+      "step": 1311500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.0533789645849805e-05,
+      "loss": 1.3505,
+      "step": 1312000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.0531689680289246e-05,
+      "loss": 1.3125,
+      "step": 1312500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.052958971472868e-05,
+      "loss": 1.3397,
+      "step": 1313000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.052748974916811e-05,
+      "loss": 1.3812,
+      "step": 1313500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.052538978360755e-05,
+      "loss": 1.3588,
+      "step": 1314000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.0523289818046986e-05,
+      "loss": 1.3531,
+      "step": 1314500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.052118985248641e-05,
+      "loss": 1.3294,
+      "step": 1315000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.051908988692585e-05,
+      "loss": 1.3517,
+      "step": 1315500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.051698992136529e-05,
+      "loss": 1.3449,
+      "step": 1316000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.051488995580472e-05,
+      "loss": 1.3732,
+      "step": 1316500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.051278999024416e-05,
+      "loss": 1.3561,
+      "step": 1317000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.0510690024683594e-05,
+      "loss": 1.3669,
+      "step": 1317500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.050859005912303e-05,
+      "loss": 1.3783,
+      "step": 1318000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.050649009356247e-05,
+      "loss": 1.3262,
+      "step": 1318500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.05043901280019e-05,
+      "loss": 1.3714,
+      "step": 1319000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.0502294362372455e-05,
+      "loss": 1.3824,
+      "step": 1319500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.0500194396811895e-05,
+      "loss": 1.3397,
+      "step": 1320000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.049809443125133e-05,
+      "loss": 1.373,
+      "step": 1320500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.049599446569076e-05,
+      "loss": 1.3412,
+      "step": 1321000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.04938945001302e-05,
+      "loss": 1.3345,
+      "step": 1321500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.0491794534569635e-05,
+      "loss": 1.3698,
+      "step": 1322000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.048969456900907e-05,
+      "loss": 1.3573,
+      "step": 1322500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.04875946034485e-05,
+      "loss": 1.3603,
+      "step": 1323000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.048549883781906e-05,
+      "loss": 1.3738,
+      "step": 1323500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.0483403072189616e-05,
+      "loss": 1.3813,
+      "step": 1324000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.048130310662905e-05,
+      "loss": 1.3621,
+      "step": 1324500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.047920314106848e-05,
+      "loss": 1.3689,
+      "step": 1325000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.0477103175507923e-05,
+      "loss": 1.3878,
+      "step": 1325500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.047500740987848e-05,
+      "loss": 1.3583,
+      "step": 1326000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.047290744431791e-05,
+      "loss": 1.3273,
+      "step": 1326500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.047080747875735e-05,
+      "loss": 1.3863,
+      "step": 1327000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.0468707513196784e-05,
+      "loss": 1.3773,
+      "step": 1327500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.046660754763622e-05,
+      "loss": 1.3776,
+      "step": 1328000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.046450758207566e-05,
+      "loss": 1.3504,
+      "step": 1328500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.046240761651509e-05,
+      "loss": 1.3939,
+      "step": 1329000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.0460307650954525e-05,
+      "loss": 1.3695,
+      "step": 1329500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.045820768539396e-05,
+      "loss": 1.3504,
+      "step": 1330000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.045610771983339e-05,
+      "loss": 1.3887,
+      "step": 1330500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.0454007754272825e-05,
+      "loss": 1.3703,
+      "step": 1331000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.0451907788712265e-05,
+      "loss": 1.3466,
+      "step": 1331500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.0449812023082826e-05,
+      "loss": 1.3766,
+      "step": 1332000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.044771205752225e-05,
+      "loss": 1.3679,
+      "step": 1332500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.0445612091961686e-05,
+      "loss": 1.3516,
+      "step": 1333000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.0443516326332246e-05,
+      "loss": 1.3815,
+      "step": 1333500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.0441416360771686e-05,
+      "loss": 1.4009,
+      "step": 1334000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.043931639521112e-05,
+      "loss": 1.3667,
+      "step": 1334500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.043721642965055e-05,
+      "loss": 1.3997,
+      "step": 1335000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.043511646408999e-05,
+      "loss": 1.3618,
+      "step": 1335500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.043301649852942e-05,
+      "loss": 1.3986,
+      "step": 1336000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.043092073289998e-05,
+      "loss": 1.3771,
+      "step": 1336500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.042882076733942e-05,
+      "loss": 1.3695,
+      "step": 1337000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.0426725001709974e-05,
+      "loss": 1.3269,
+      "step": 1337500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.042462503614941e-05,
+      "loss": 1.3586,
+      "step": 1338000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.042252507058884e-05,
+      "loss": 1.357,
+      "step": 1338500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.042042510502828e-05,
+      "loss": 1.3575,
+      "step": 1339000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.041832513946771e-05,
+      "loss": 1.3306,
+      "step": 1339500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.041622517390714e-05,
+      "loss": 1.3897,
+      "step": 1340000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.041412520834658e-05,
+      "loss": 1.3571,
+      "step": 1340500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.0412025242786015e-05,
+      "loss": 1.3571,
+      "step": 1341000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.040992527722545e-05,
+      "loss": 1.3547,
+      "step": 1341500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.040782531166489e-05,
+      "loss": 1.382,
+      "step": 1342000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.040572534610432e-05,
+      "loss": 1.3588,
+      "step": 1342500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.0403629580474876e-05,
+      "loss": 1.3556,
+      "step": 1343000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.0401529614914316e-05,
+      "loss": 1.3782,
+      "step": 1343500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.039942964935375e-05,
+      "loss": 1.3421,
+      "step": 1344000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.039732968379318e-05,
+      "loss": 1.368,
+      "step": 1344500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.0395229718232624e-05,
+      "loss": 1.3513,
+      "step": 1345000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.039312975267206e-05,
+      "loss": 1.3517,
+      "step": 1345500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.039102978711149e-05,
+      "loss": 1.3703,
+      "step": 1346000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.0388934021482044e-05,
+      "loss": 1.3803,
+      "step": 1346500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.0386834055921484e-05,
+      "loss": 1.3719,
+      "step": 1347000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.038473409036092e-05,
+      "loss": 1.3595,
+      "step": 1347500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.038263832473147e-05,
+      "loss": 1.3555,
+      "step": 1348000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.0380538359170905e-05,
+      "loss": 1.368,
+      "step": 1348500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.0378438393610345e-05,
+      "loss": 1.3555,
+      "step": 1349000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.037633842804978e-05,
+      "loss": 1.3669,
+      "step": 1349500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.037423846248921e-05,
+      "loss": 1.3711,
+      "step": 1350000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.037213849692865e-05,
+      "loss": 1.3683,
+      "step": 1350500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.0370038531368086e-05,
+      "loss": 1.3621,
+      "step": 1351000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.036793856580752e-05,
+      "loss": 1.3727,
+      "step": 1351500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.036584280017808e-05,
+      "loss": 1.3427,
+      "step": 1352000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.036374283461751e-05,
+      "loss": 1.3463,
+      "step": 1352500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.0361642869056946e-05,
+      "loss": 1.3728,
+      "step": 1353000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.0359542903496387e-05,
+      "loss": 1.3597,
+      "step": 1353500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.035744713786694e-05,
+      "loss": 1.3237,
+      "step": 1354000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.0355347172306374e-05,
+      "loss": 1.3856,
+      "step": 1354500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.035324720674581e-05,
+      "loss": 1.3451,
+      "step": 1355000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.035114724118525e-05,
+      "loss": 1.3661,
+      "step": 1355500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.034904727562468e-05,
+      "loss": 1.3572,
+      "step": 1356000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.0346947310064114e-05,
+      "loss": 1.3456,
+      "step": 1356500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.034484734450355e-05,
+      "loss": 1.3603,
+      "step": 1357000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.034274737894298e-05,
+      "loss": 1.3491,
+      "step": 1357500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.0340647413382415e-05,
+      "loss": 1.3883,
+      "step": 1358000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.0338551647752975e-05,
+      "loss": 1.3518,
+      "step": 1358500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.0336451682192415e-05,
+      "loss": 1.3455,
+      "step": 1359000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.033435171663184e-05,
+      "loss": 1.3467,
+      "step": 1359500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.033225175107128e-05,
+      "loss": 1.3773,
+      "step": 1360000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.0330151785510716e-05,
+      "loss": 1.3622,
+      "step": 1360500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.032805181995015e-05,
+      "loss": 1.3367,
+      "step": 1361000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.032595185438959e-05,
+      "loss": 1.364,
+      "step": 1361500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.032385608876014e-05,
+      "loss": 1.3671,
+      "step": 1362000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.0321756123199576e-05,
+      "loss": 1.3471,
+      "step": 1362500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.031966035757014e-05,
+      "loss": 1.3458,
+      "step": 1363000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.031756039200957e-05,
+      "loss": 1.3724,
+      "step": 1363500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.031546042644901e-05,
+      "loss": 1.3923,
+      "step": 1364000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.031336046088844e-05,
+      "loss": 1.3735,
+      "step": 1364500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.031126049532787e-05,
+      "loss": 1.366,
+      "step": 1365000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.030916052976731e-05,
+      "loss": 1.3841,
+      "step": 1365500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.0307060564206744e-05,
+      "loss": 1.3393,
+      "step": 1366000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.0304960598646184e-05,
+      "loss": 1.3327,
+      "step": 1366500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.030286063308562e-05,
+      "loss": 1.3566,
+      "step": 1367000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.030076066752505e-05,
+      "loss": 1.3514,
+      "step": 1367500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.029866070196449e-05,
+      "loss": 1.3473,
+      "step": 1368000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.0296560736403925e-05,
+      "loss": 1.3618,
+      "step": 1368500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.029446497077448e-05,
+      "loss": 1.3708,
+      "step": 1369000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.029236920514503e-05,
+      "loss": 1.37,
+      "step": 1369500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.029027343951559e-05,
+      "loss": 1.3517,
+      "step": 1370000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.0288173473955026e-05,
+      "loss": 1.3619,
+      "step": 1370500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.0286073508394466e-05,
+      "loss": 1.3541,
+      "step": 1371000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.028397354283389e-05,
+      "loss": 1.339,
+      "step": 1371500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.028187357727333e-05,
+      "loss": 1.3613,
+      "step": 1372000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.0279773611712767e-05,
+      "loss": 1.3839,
+      "step": 1372500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.02776736461522e-05,
+      "loss": 1.3925,
+      "step": 1373000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.027557368059164e-05,
+      "loss": 1.372,
+      "step": 1373500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.0273473715031074e-05,
+      "loss": 1.3491,
+      "step": 1374000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.027137374947051e-05,
+      "loss": 1.3841,
+      "step": 1374500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.026927798384106e-05,
+      "loss": 1.3851,
+      "step": 1375000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.02671780182805e-05,
+      "loss": 1.3629,
+      "step": 1375500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.0265078052719934e-05,
+      "loss": 1.3333,
+      "step": 1376000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.026297808715937e-05,
+      "loss": 1.3728,
+      "step": 1376500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.026087812159881e-05,
+      "loss": 1.359,
+      "step": 1377000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.025877815603824e-05,
+      "loss": 1.3648,
+      "step": 1377500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.0256678190477675e-05,
+      "loss": 1.3634,
+      "step": 1378000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.0254578224917115e-05,
+      "loss": 1.3242,
+      "step": 1378500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.025248245928767e-05,
+      "loss": 1.3445,
+      "step": 1379000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.02503824937271e-05,
+      "loss": 1.3264,
+      "step": 1379500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.0248282528166536e-05,
+      "loss": 1.3571,
+      "step": 1380000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.0246182562605976e-05,
+      "loss": 1.3501,
+      "step": 1380500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.024408259704541e-05,
+      "loss": 1.3979,
+      "step": 1381000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.024198683141596e-05,
+      "loss": 1.3461,
+      "step": 1381500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.02398868658554e-05,
+      "loss": 1.3789,
+      "step": 1382000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.023778690029484e-05,
+      "loss": 1.365,
+      "step": 1382500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.023568693473427e-05,
+      "loss": 1.3791,
+      "step": 1383000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.023358696917371e-05,
+      "loss": 1.3781,
+      "step": 1383500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.023148700361314e-05,
+      "loss": 1.3572,
+      "step": 1384000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.022938703805257e-05,
+      "loss": 1.3567,
+      "step": 1384500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.022728707249201e-05,
+      "loss": 1.3838,
+      "step": 1385000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.022519130686257e-05,
+      "loss": 1.3742,
+      "step": 1385500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.0223095541233125e-05,
+      "loss": 1.3452,
+      "step": 1386000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.022099557567256e-05,
+      "loss": 1.3815,
+      "step": 1386500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.021889981004311e-05,
+      "loss": 1.3345,
+      "step": 1387000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.021679984448255e-05,
+      "loss": 1.3558,
+      "step": 1387500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.0214699878921985e-05,
+      "loss": 1.3787,
+      "step": 1388000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.021259991336142e-05,
+      "loss": 1.3451,
+      "step": 1388500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.021049994780086e-05,
+      "loss": 1.3576,
+      "step": 1389000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.020840418217141e-05,
+      "loss": 1.3378,
+      "step": 1389500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.0206304216610846e-05,
+      "loss": 1.3333,
+      "step": 1390000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.020420425105028e-05,
+      "loss": 1.3448,
+      "step": 1390500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.020210428548972e-05,
+      "loss": 1.3811,
+      "step": 1391000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.020000431992915e-05,
+      "loss": 1.3399,
+      "step": 1391500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.019790855429971e-05,
+      "loss": 1.3733,
+      "step": 1392000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.019580858873914e-05,
+      "loss": 1.3372,
+      "step": 1392500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.019370862317858e-05,
+      "loss": 1.3775,
+      "step": 1393000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.0191608657618014e-05,
+      "loss": 1.3937,
+      "step": 1393500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.018951289198857e-05,
+      "loss": 1.3529,
+      "step": 1394000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.018741292642801e-05,
+      "loss": 1.338,
+      "step": 1394500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.018531296086744e-05,
+      "loss": 1.3768,
+      "step": 1395000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.0183212995306875e-05,
+      "loss": 1.3556,
+      "step": 1395500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.0181113029746315e-05,
+      "loss": 1.345,
+      "step": 1396000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.017901306418575e-05,
+      "loss": 1.3234,
+      "step": 1396500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.017691309862518e-05,
+      "loss": 1.3617,
+      "step": 1397000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.017481313306462e-05,
+      "loss": 1.3563,
+      "step": 1397500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.0172713167504056e-05,
+      "loss": 1.3239,
+      "step": 1398000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.017061320194348e-05,
+      "loss": 1.3378,
+      "step": 1398500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.016851743631404e-05,
+      "loss": 1.354,
+      "step": 1399000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.016641747075348e-05,
+      "loss": 1.3327,
+      "step": 1399500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.0164317505192916e-05,
+      "loss": 1.3511,
+      "step": 1400000
+    },
+    {
+      "epoch": 0.84,
+      "eval_loss": 1.2957533597946167,
+      "eval_runtime": 1117.7468,
+      "eval_samples_per_second": 471.234,
+      "eval_steps_per_second": 78.539,
+      "step": 1400000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.016221753963235e-05,
+      "loss": 1.3559,
+      "step": 1400500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.016011757407178e-05,
+      "loss": 1.3687,
+      "step": 1401000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.015801760851122e-05,
+      "loss": 1.3554,
+      "step": 1401500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.015592184288178e-05,
+      "loss": 1.36,
+      "step": 1402000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.015382187732121e-05,
+      "loss": 1.3392,
+      "step": 1402500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.0151721911760644e-05,
+      "loss": 1.3502,
+      "step": 1403000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.014962194620008e-05,
+      "loss": 1.3557,
+      "step": 1403500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.014752198063952e-05,
+      "loss": 1.3468,
+      "step": 1404000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.014542201507895e-05,
+      "loss": 1.3427,
+      "step": 1404500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.0143322049518385e-05,
+      "loss": 1.3872,
+      "step": 1405000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.0141222083957825e-05,
+      "loss": 1.3528,
+      "step": 1405500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.013912631832838e-05,
+      "loss": 1.3865,
+      "step": 1406000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.013702635276781e-05,
+      "loss": 1.3521,
+      "step": 1406500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.0134926387207245e-05,
+      "loss": 1.369,
+      "step": 1407000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.0132826421646686e-05,
+      "loss": 1.3716,
+      "step": 1407500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.013072645608612e-05,
+      "loss": 1.3633,
+      "step": 1408000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.012862649052555e-05,
+      "loss": 1.3631,
+      "step": 1408500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.012652652496499e-05,
+      "loss": 1.367,
+      "step": 1409000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.0124426559404426e-05,
+      "loss": 1.3253,
+      "step": 1409500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.012232659384386e-05,
+      "loss": 1.3425,
+      "step": 1410000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.01202266282833e-05,
+      "loss": 1.3829,
+      "step": 1410500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.011812666272273e-05,
+      "loss": 1.3671,
+      "step": 1411000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.011602669716216e-05,
+      "loss": 1.3428,
+      "step": 1411500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.011393093153273e-05,
+      "loss": 1.356,
+      "step": 1412000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.011183096597216e-05,
+      "loss": 1.3526,
+      "step": 1412500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.0109731000411594e-05,
+      "loss": 1.3665,
+      "step": 1413000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.010763103485103e-05,
+      "loss": 1.3537,
+      "step": 1413500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.010553106929046e-05,
+      "loss": 1.3535,
+      "step": 1414000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.0103431103729894e-05,
+      "loss": 1.3784,
+      "step": 1414500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.0101335338100455e-05,
+      "loss": 1.3379,
+      "step": 1415000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.0099235372539895e-05,
+      "loss": 1.3678,
+      "step": 1415500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.009713540697932e-05,
+      "loss": 1.3478,
+      "step": 1416000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.0095035441418755e-05,
+      "loss": 1.3404,
+      "step": 1416500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.0092935475858195e-05,
+      "loss": 1.3638,
+      "step": 1417000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.009083551029763e-05,
+      "loss": 1.3543,
+      "step": 1417500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.008873974466818e-05,
+      "loss": 1.3728,
+      "step": 1418000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.008663977910762e-05,
+      "loss": 1.3341,
+      "step": 1418500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.0084539813547056e-05,
+      "loss": 1.3762,
+      "step": 1419000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.008243984798649e-05,
+      "loss": 1.3458,
+      "step": 1419500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.008034408235705e-05,
+      "loss": 1.3447,
+      "step": 1420000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.0078244116796483e-05,
+      "loss": 1.3587,
+      "step": 1420500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.007614415123592e-05,
+      "loss": 1.3616,
+      "step": 1421000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.007404418567535e-05,
+      "loss": 1.3597,
+      "step": 1421500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.007194842004591e-05,
+      "loss": 1.3673,
+      "step": 1422000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.0069852654416464e-05,
+      "loss": 1.3474,
+      "step": 1422500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.0067752688855905e-05,
+      "loss": 1.3718,
+      "step": 1423000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.006565272329534e-05,
+      "loss": 1.3493,
+      "step": 1423500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.006355275773477e-05,
+      "loss": 1.3535,
+      "step": 1424000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.006145279217421e-05,
+      "loss": 1.353,
+      "step": 1424500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.0059352826613645e-05,
+      "loss": 1.3666,
+      "step": 1425000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.005725286105308e-05,
+      "loss": 1.3343,
+      "step": 1425500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.005515709542364e-05,
+      "loss": 1.368,
+      "step": 1426000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.005305712986307e-05,
+      "loss": 1.3494,
+      "step": 1426500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0050957164302506e-05,
+      "loss": 1.3604,
+      "step": 1427000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0048857198741946e-05,
+      "loss": 1.3449,
+      "step": 1427500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.004675723318137e-05,
+      "loss": 1.3752,
+      "step": 1428000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0044657267620806e-05,
+      "loss": 1.3549,
+      "step": 1428500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0042557302060246e-05,
+      "loss": 1.3495,
+      "step": 1429000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.004045733649968e-05,
+      "loss": 1.3743,
+      "step": 1429500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0038357370939113e-05,
+      "loss": 1.3473,
+      "step": 1430000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.003626160530967e-05,
+      "loss": 1.3312,
+      "step": 1430500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.003416163974911e-05,
+      "loss": 1.3626,
+      "step": 1431000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.003206167418854e-05,
+      "loss": 1.3733,
+      "step": 1431500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0029961708627974e-05,
+      "loss": 1.3673,
+      "step": 1432000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0027865942998534e-05,
+      "loss": 1.3973,
+      "step": 1432500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.002576597743797e-05,
+      "loss": 1.3579,
+      "step": 1433000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.00236660118774e-05,
+      "loss": 1.3774,
+      "step": 1433500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.002157024624796e-05,
+      "loss": 1.3346,
+      "step": 1434000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.00194702806874e-05,
+      "loss": 1.3529,
+      "step": 1434500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.001737031512683e-05,
+      "loss": 1.3542,
+      "step": 1435000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.001527034956626e-05,
+      "loss": 1.3794,
+      "step": 1435500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.00131703840057e-05,
+      "loss": 1.3467,
+      "step": 1436000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.001107461837626e-05,
+      "loss": 1.3688,
+      "step": 1436500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0008974652815696e-05,
+      "loss": 1.3703,
+      "step": 1437000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.000687468725512e-05,
+      "loss": 1.3655,
+      "step": 1437500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.000477472169456e-05,
+      "loss": 1.3403,
+      "step": 1438000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0002674756133997e-05,
+      "loss": 1.3419,
+      "step": 1438500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.000057479057343e-05,
+      "loss": 1.3477,
+      "step": 1439000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.999847902494399e-05,
+      "loss": 1.3568,
+      "step": 1439500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9996379059383424e-05,
+      "loss": 1.3574,
+      "step": 1440000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.999427909382286e-05,
+      "loss": 1.3569,
+      "step": 1440500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.99921791282623e-05,
+      "loss": 1.3307,
+      "step": 1441000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.999007916270173e-05,
+      "loss": 1.3464,
+      "step": 1441500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9987983397072285e-05,
+      "loss": 1.3494,
+      "step": 1442000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.998588343151172e-05,
+      "loss": 1.3825,
+      "step": 1442500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.998378346595116e-05,
+      "loss": 1.3466,
+      "step": 1443000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.998168350039059e-05,
+      "loss": 1.3645,
+      "step": 1443500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.9979583534830025e-05,
+      "loss": 1.3432,
+      "step": 1444000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.9977483569269465e-05,
+      "loss": 1.3897,
+      "step": 1444500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.99753836037089e-05,
+      "loss": 1.3479,
+      "step": 1445000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.997328363814833e-05,
+      "loss": 1.3598,
+      "step": 1445500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.997118367258777e-05,
+      "loss": 1.3676,
+      "step": 1446000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.9969087906958326e-05,
+      "loss": 1.3521,
+      "step": 1446500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.996698794139776e-05,
+      "loss": 1.3406,
+      "step": 1447000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.996488797583719e-05,
+      "loss": 1.3737,
+      "step": 1447500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.996278801027663e-05,
+      "loss": 1.3665,
+      "step": 1448000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.996069224464719e-05,
+      "loss": 1.3931,
+      "step": 1448500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.995859227908662e-05,
+      "loss": 1.34,
+      "step": 1449000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.995649231352606e-05,
+      "loss": 1.3524,
+      "step": 1449500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.9954392347965494e-05,
+      "loss": 1.3403,
+      "step": 1450000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.995229238240493e-05,
+      "loss": 1.3441,
+      "step": 1450500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.995019241684437e-05,
+      "loss": 1.3675,
+      "step": 1451000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.99480924512838e-05,
+      "loss": 1.3558,
+      "step": 1451500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.9945992485723235e-05,
+      "loss": 1.3145,
+      "step": 1452000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.994389252016267e-05,
+      "loss": 1.3376,
+      "step": 1452500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.99417925546021e-05,
+      "loss": 1.3487,
+      "step": 1453000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.9939692589041535e-05,
+      "loss": 1.3964,
+      "step": 1453500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.9937596823412095e-05,
+      "loss": 1.384,
+      "step": 1454000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.993550105778265e-05,
+      "loss": 1.3288,
+      "step": 1454500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.993340109222209e-05,
+      "loss": 1.3658,
+      "step": 1455000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.993130112666152e-05,
+      "loss": 1.3944,
+      "step": 1455500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.9929201161100956e-05,
+      "loss": 1.3843,
+      "step": 1456000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.9927101195540396e-05,
+      "loss": 1.3401,
+      "step": 1456500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.992500122997982e-05,
+      "loss": 1.3714,
+      "step": 1457000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.992290126441926e-05,
+      "loss": 1.3549,
+      "step": 1457500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.99208012988587e-05,
+      "loss": 1.3391,
+      "step": 1458000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.991870133329813e-05,
+      "loss": 1.3782,
+      "step": 1458500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.991660136773757e-05,
+      "loss": 1.3308,
+      "step": 1459000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.9914501402177004e-05,
+      "loss": 1.3384,
+      "step": 1459500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.991240143661644e-05,
+      "loss": 1.3612,
+      "step": 1460000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.991030567098699e-05,
+      "loss": 1.3616,
+      "step": 1460500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.990820570542643e-05,
+      "loss": 1.3731,
+      "step": 1461000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.9906105739865865e-05,
+      "loss": 1.3501,
+      "step": 1461500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.99040057743053e-05,
+      "loss": 1.3622,
+      "step": 1462000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.990191000867585e-05,
+      "loss": 1.3907,
+      "step": 1462500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.989981424304641e-05,
+      "loss": 1.3486,
+      "step": 1463000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.989771427748585e-05,
+      "loss": 1.3768,
+      "step": 1463500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.9895614311925286e-05,
+      "loss": 1.3447,
+      "step": 1464000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.989351434636472e-05,
+      "loss": 1.3832,
+      "step": 1464500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.989141438080415e-05,
+      "loss": 1.3687,
+      "step": 1465000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.9889314415243586e-05,
+      "loss": 1.3254,
+      "step": 1465500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.9887218649614146e-05,
+      "loss": 1.3592,
+      "step": 1466000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.988511868405358e-05,
+      "loss": 1.3441,
+      "step": 1466500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.988301871849301e-05,
+      "loss": 1.3329,
+      "step": 1467000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.988091875293245e-05,
+      "loss": 1.327,
+      "step": 1467500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.987882298730301e-05,
+      "loss": 1.3498,
+      "step": 1468000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.987672302174245e-05,
+      "loss": 1.3766,
+      "step": 1468500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.9874623056181874e-05,
+      "loss": 1.3593,
+      "step": 1469000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.987252309062131e-05,
+      "loss": 1.3689,
+      "step": 1469500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.987042312506075e-05,
+      "loss": 1.3375,
+      "step": 1470000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.986832315950018e-05,
+      "loss": 1.3634,
+      "step": 1470500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.986622319393962e-05,
+      "loss": 1.3612,
+      "step": 1471000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.9864123228379055e-05,
+      "loss": 1.3619,
+      "step": 1471500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.986202746274961e-05,
+      "loss": 1.386,
+      "step": 1472000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.985992749718904e-05,
+      "loss": 1.343,
+      "step": 1472500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.985782753162848e-05,
+      "loss": 1.3386,
+      "step": 1473000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.9855727566067916e-05,
+      "loss": 1.3211,
+      "step": 1473500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.985363180043847e-05,
+      "loss": 1.365,
+      "step": 1474000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.98515318348779e-05,
+      "loss": 1.3443,
+      "step": 1474500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.984943186931734e-05,
+      "loss": 1.3525,
+      "step": 1475000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.9847331903756776e-05,
+      "loss": 1.3659,
+      "step": 1475500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.984523613812733e-05,
+      "loss": 1.3623,
+      "step": 1476000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.984314037249789e-05,
+      "loss": 1.3571,
+      "step": 1476500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9841040406937324e-05,
+      "loss": 1.3557,
+      "step": 1477000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9838940441376764e-05,
+      "loss": 1.3724,
+      "step": 1477500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.98368404758162e-05,
+      "loss": 1.3603,
+      "step": 1478000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.983474051025563e-05,
+      "loss": 1.3528,
+      "step": 1478500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.983264474462619e-05,
+      "loss": 1.35,
+      "step": 1479000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9830544779065625e-05,
+      "loss": 1.3552,
+      "step": 1479500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.982844481350506e-05,
+      "loss": 1.3568,
+      "step": 1480000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.98263448479445e-05,
+      "loss": 1.3534,
+      "step": 1480500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.982424908231505e-05,
+      "loss": 1.327,
+      "step": 1481000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9822149116754485e-05,
+      "loss": 1.357,
+      "step": 1481500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.982005335112504e-05,
+      "loss": 1.3553,
+      "step": 1482000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.981795338556447e-05,
+      "loss": 1.3514,
+      "step": 1482500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.981585342000391e-05,
+      "loss": 1.3264,
+      "step": 1483000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9813753454443346e-05,
+      "loss": 1.3313,
+      "step": 1483500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9811653488882786e-05,
+      "loss": 1.3577,
+      "step": 1484000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.980955352332222e-05,
+      "loss": 1.3812,
+      "step": 1484500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.980745355776165e-05,
+      "loss": 1.3652,
+      "step": 1485000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.980535359220109e-05,
+      "loss": 1.3653,
+      "step": 1485500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.980325362664052e-05,
+      "loss": 1.335,
+      "step": 1486000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.980115786101108e-05,
+      "loss": 1.3733,
+      "step": 1486500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9799062095381634e-05,
+      "loss": 1.3455,
+      "step": 1487000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.979696212982107e-05,
+      "loss": 1.3338,
+      "step": 1487500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.979486216426051e-05,
+      "loss": 1.3578,
+      "step": 1488000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.979276219869994e-05,
+      "loss": 1.3312,
+      "step": 1488500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9790662233139375e-05,
+      "loss": 1.3559,
+      "step": 1489000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9788562267578815e-05,
+      "loss": 1.3392,
+      "step": 1489500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.978646230201825e-05,
+      "loss": 1.3373,
+      "step": 1490000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.978436233645768e-05,
+      "loss": 1.3669,
+      "step": 1490500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9782262370897115e-05,
+      "loss": 1.3521,
+      "step": 1491000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.978016240533655e-05,
+      "loss": 1.3375,
+      "step": 1491500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.977806243977599e-05,
+      "loss": 1.324,
+      "step": 1492000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.977596667414655e-05,
+      "loss": 1.3571,
+      "step": 1492500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9773866708585976e-05,
+      "loss": 1.3361,
+      "step": 1493000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.977176674302541e-05,
+      "loss": 1.3423,
+      "step": 1493500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.976966677746485e-05,
+      "loss": 1.3543,
+      "step": 1494000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.976756681190428e-05,
+      "loss": 1.359,
+      "step": 1494500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9765471046274844e-05,
+      "loss": 1.3461,
+      "step": 1495000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.976337108071427e-05,
+      "loss": 1.3232,
+      "step": 1495500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.976127111515371e-05,
+      "loss": 1.3619,
+      "step": 1496000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9759171149593144e-05,
+      "loss": 1.3703,
+      "step": 1496500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9757075383963704e-05,
+      "loss": 1.3747,
+      "step": 1497000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.975497541840314e-05,
+      "loss": 1.36,
+      "step": 1497500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.975287545284257e-05,
+      "loss": 1.3668,
+      "step": 1498000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9750775487282005e-05,
+      "loss": 1.3609,
+      "step": 1498500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9748675521721445e-05,
+      "loss": 1.3692,
+      "step": 1499000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.974657555616088e-05,
+      "loss": 1.3724,
+      "step": 1499500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.974447559060031e-05,
+      "loss": 1.3617,
+      "step": 1500000
+    },
+    {
+      "epoch": 0.9,
+      "eval_loss": 1.2882256507873535,
+      "eval_runtime": 1113.7701,
+      "eval_samples_per_second": 472.916,
+      "eval_steps_per_second": 78.82,
+      "step": 1500000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.974237562503975e-05,
+      "loss": 1.348,
+      "step": 1500500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9740275659479186e-05,
+      "loss": 1.3631,
+      "step": 1501000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.973817989384974e-05,
+      "loss": 1.328,
+      "step": 1501500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.973607992828917e-05,
+      "loss": 1.3218,
+      "step": 1502000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.973397996272861e-05,
+      "loss": 1.3427,
+      "step": 1502500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9731879997168046e-05,
+      "loss": 1.3348,
+      "step": 1503000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.972978003160748e-05,
+      "loss": 1.3697,
+      "step": 1503500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.972768006604692e-05,
+      "loss": 1.3246,
+      "step": 1504000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9725580100486353e-05,
+      "loss": 1.3394,
+      "step": 1504500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.972348013492579e-05,
+      "loss": 1.3589,
+      "step": 1505000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.972138436929634e-05,
+      "loss": 1.3467,
+      "step": 1505500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.971928440373578e-05,
+      "loss": 1.3528,
+      "step": 1506000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9717188638106334e-05,
+      "loss": 1.3551,
+      "step": 1506500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.971508867254577e-05,
+      "loss": 1.3595,
+      "step": 1507000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.971298870698521e-05,
+      "loss": 1.3974,
+      "step": 1507500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.971088874142464e-05,
+      "loss": 1.3466,
+      "step": 1508000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9708792975795195e-05,
+      "loss": 1.3667,
+      "step": 1508500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9706697210165755e-05,
+      "loss": 1.3634,
+      "step": 1509000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.970459724460518e-05,
+      "loss": 1.3597,
+      "step": 1509500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.970249727904462e-05,
+      "loss": 1.354,
+      "step": 1510000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.9700397313484056e-05,
+      "loss": 1.3445,
+      "step": 1510500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.969829734792349e-05,
+      "loss": 1.3389,
+      "step": 1511000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.969619738236293e-05,
+      "loss": 1.3428,
+      "step": 1511500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.969409741680236e-05,
+      "loss": 1.3533,
+      "step": 1512000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.9691997451241796e-05,
+      "loss": 1.3077,
+      "step": 1512500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.968990168561236e-05,
+      "loss": 1.3885,
+      "step": 1513000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.968780172005179e-05,
+      "loss": 1.3356,
+      "step": 1513500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.9685701754491224e-05,
+      "loss": 1.3411,
+      "step": 1514000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.9683601788930664e-05,
+      "loss": 1.3309,
+      "step": 1514500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.96815018233701e-05,
+      "loss": 1.3334,
+      "step": 1515000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.967940185780953e-05,
+      "loss": 1.3377,
+      "step": 1515500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.9677306092180084e-05,
+      "loss": 1.3242,
+      "step": 1516000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.9675206126619525e-05,
+      "loss": 1.3217,
+      "step": 1516500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.967310616105896e-05,
+      "loss": 1.3435,
+      "step": 1517000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.967100619549839e-05,
+      "loss": 1.3899,
+      "step": 1517500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.9668910429868945e-05,
+      "loss": 1.3423,
+      "step": 1518000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.9666810464308385e-05,
+      "loss": 1.3375,
+      "step": 1518500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.966471469867894e-05,
+      "loss": 1.3677,
+      "step": 1519000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.966261473311837e-05,
+      "loss": 1.3255,
+      "step": 1519500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.966051476755781e-05,
+      "loss": 1.343,
+      "step": 1520000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.9658414801997246e-05,
+      "loss": 1.3466,
+      "step": 1520500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.9656319036367806e-05,
+      "loss": 1.3477,
+      "step": 1521000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.965421907080723e-05,
+      "loss": 1.3337,
+      "step": 1521500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.965211910524667e-05,
+      "loss": 1.3597,
+      "step": 1522000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.965001913968611e-05,
+      "loss": 1.3646,
+      "step": 1522500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.964791917412554e-05,
+      "loss": 1.3481,
+      "step": 1523000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.964581920856498e-05,
+      "loss": 1.3579,
+      "step": 1523500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.9643719243004414e-05,
+      "loss": 1.363,
+      "step": 1524000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.964161927744385e-05,
+      "loss": 1.336,
+      "step": 1524500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.963951931188329e-05,
+      "loss": 1.3627,
+      "step": 1525000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.963741934632272e-05,
+      "loss": 1.3363,
+      "step": 1525500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.9635319380762155e-05,
+      "loss": 1.3587,
+      "step": 1526000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.9633219415201595e-05,
+      "loss": 1.3497,
+      "step": 1526500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.963112364957215e-05,
+      "loss": 1.335,
+      "step": 1527000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.962902368401158e-05,
+      "loss": 1.3441,
+      "step": 1527500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.9626923718451015e-05,
+      "loss": 1.3459,
+      "step": 1528000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.9624823752890455e-05,
+      "loss": 1.3846,
+      "step": 1528500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.962273218719213e-05,
+      "loss": 1.351,
+      "step": 1529000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.962063222163156e-05,
+      "loss": 1.3635,
+      "step": 1529500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.961853645600212e-05,
+      "loss": 1.359,
+      "step": 1530000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.9616436490441556e-05,
+      "loss": 1.3364,
+      "step": 1530500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.961433652488099e-05,
+      "loss": 1.3353,
+      "step": 1531000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.961223655932042e-05,
+      "loss": 1.3298,
+      "step": 1531500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.961013659375986e-05,
+      "loss": 1.3628,
+      "step": 1532000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.960804082813042e-05,
+      "loss": 1.3335,
+      "step": 1532500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.960594086256986e-05,
+      "loss": 1.3212,
+      "step": 1533000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.9603840897009284e-05,
+      "loss": 1.3389,
+      "step": 1533500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.9601740931448724e-05,
+      "loss": 1.3497,
+      "step": 1534000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.959964096588816e-05,
+      "loss": 1.3317,
+      "step": 1534500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.959754100032759e-05,
+      "loss": 1.3454,
+      "step": 1535000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.959544103476703e-05,
+      "loss": 1.3644,
+      "step": 1535500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.9593341069206465e-05,
+      "loss": 1.3565,
+      "step": 1536000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.959124530357702e-05,
+      "loss": 1.3489,
+      "step": 1536500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.958914953794758e-05,
+      "loss": 1.3489,
+      "step": 1537000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.958704957238701e-05,
+      "loss": 1.37,
+      "step": 1537500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.9584949606826446e-05,
+      "loss": 1.3483,
+      "step": 1538000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.958284964126588e-05,
+      "loss": 1.3569,
+      "step": 1538500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.958074967570531e-05,
+      "loss": 1.3278,
+      "step": 1539000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.957864971014475e-05,
+      "loss": 1.3169,
+      "step": 1539500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.9576549744584186e-05,
+      "loss": 1.3644,
+      "step": 1540000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.957444977902362e-05,
+      "loss": 1.3393,
+      "step": 1540500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.957234981346306e-05,
+      "loss": 1.3195,
+      "step": 1541000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.9570254047833614e-05,
+      "loss": 1.3636,
+      "step": 1541500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.9568158282204174e-05,
+      "loss": 1.3549,
+      "step": 1542000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.956605831664361e-05,
+      "loss": 1.3589,
+      "step": 1542500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.956395835108304e-05,
+      "loss": 1.3544,
+      "step": 1543000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.9561858385522474e-05,
+      "loss": 1.36,
+      "step": 1543500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.955975841996191e-05,
+      "loss": 1.328,
+      "step": 1544000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.955765845440135e-05,
+      "loss": 1.3684,
+      "step": 1544500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.955555848884078e-05,
+      "loss": 1.3337,
+      "step": 1545000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.9553458523280215e-05,
+      "loss": 1.3604,
+      "step": 1545500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.9551358557719655e-05,
+      "loss": 1.3611,
+      "step": 1546000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.954926279209021e-05,
+      "loss": 1.3561,
+      "step": 1546500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.954716282652964e-05,
+      "loss": 1.3712,
+      "step": 1547000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.9545062860969076e-05,
+      "loss": 1.3572,
+      "step": 1547500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.9542962895408516e-05,
+      "loss": 1.3385,
+      "step": 1548000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.954086292984795e-05,
+      "loss": 1.33,
+      "step": 1548500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.953876296428739e-05,
+      "loss": 1.3408,
+      "step": 1549000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.953666719865794e-05,
+      "loss": 1.3574,
+      "step": 1549500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.953456723309738e-05,
+      "loss": 1.3448,
+      "step": 1550000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.953246726753681e-05,
+      "loss": 1.324,
+      "step": 1550500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.953036730197625e-05,
+      "loss": 1.3288,
+      "step": 1551000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.9528267336415684e-05,
+      "loss": 1.3402,
+      "step": 1551500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.952617157078624e-05,
+      "loss": 1.3622,
+      "step": 1552000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.952407160522567e-05,
+      "loss": 1.3224,
+      "step": 1552500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.952197163966511e-05,
+      "loss": 1.3556,
+      "step": 1553000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.9519871674104545e-05,
+      "loss": 1.34,
+      "step": 1553500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.95177759084751e-05,
+      "loss": 1.3238,
+      "step": 1554000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.951567594291453e-05,
+      "loss": 1.3197,
+      "step": 1554500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.951357597735397e-05,
+      "loss": 1.3351,
+      "step": 1555000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.9511476011793405e-05,
+      "loss": 1.3388,
+      "step": 1555500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.9509376046232846e-05,
+      "loss": 1.3504,
+      "step": 1556000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.950727608067228e-05,
+      "loss": 1.354,
+      "step": 1556500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.950517611511171e-05,
+      "loss": 1.3417,
+      "step": 1557000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.950307614955115e-05,
+      "loss": 1.3471,
+      "step": 1557500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.950097618399058e-05,
+      "loss": 1.3371,
+      "step": 1558000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.949888041836114e-05,
+      "loss": 1.4033,
+      "step": 1558500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.949678045280057e-05,
+      "loss": 1.3568,
+      "step": 1559000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.9494680487240013e-05,
+      "loss": 1.3091,
+      "step": 1559500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.949258052167945e-05,
+      "loss": 1.3257,
+      "step": 1560000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.949048475605e-05,
+      "loss": 1.3642,
+      "step": 1560500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.9488384790489434e-05,
+      "loss": 1.3463,
+      "step": 1561000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.9486284824928874e-05,
+      "loss": 1.352,
+      "step": 1561500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.948418485936831e-05,
+      "loss": 1.3293,
+      "step": 1562000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.948208489380774e-05,
+      "loss": 1.3767,
+      "step": 1562500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.9479984928247175e-05,
+      "loss": 1.3366,
+      "step": 1563000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.947788496268661e-05,
+      "loss": 1.3837,
+      "step": 1563500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.947578919705717e-05,
+      "loss": 1.3572,
+      "step": 1564000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.947368923149661e-05,
+      "loss": 1.3573,
+      "step": 1564500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.9471589265936035e-05,
+      "loss": 1.3428,
+      "step": 1565000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.946948930037547e-05,
+      "loss": 1.3316,
+      "step": 1565500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.946739353474603e-05,
+      "loss": 1.3365,
+      "step": 1566000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.946529356918547e-05,
+      "loss": 1.322,
+      "step": 1566500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.94631936036249e-05,
+      "loss": 1.3555,
+      "step": 1567000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.946109363806433e-05,
+      "loss": 1.3733,
+      "step": 1567500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.945899367250377e-05,
+      "loss": 1.3221,
+      "step": 1568000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.945689790687433e-05,
+      "loss": 1.3194,
+      "step": 1568500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.9454802141244884e-05,
+      "loss": 1.3709,
+      "step": 1569000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.945270217568432e-05,
+      "loss": 1.3719,
+      "step": 1569500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.945060221012376e-05,
+      "loss": 1.3441,
+      "step": 1570000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.944850224456319e-05,
+      "loss": 1.3369,
+      "step": 1570500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.9446402279002624e-05,
+      "loss": 1.3299,
+      "step": 1571000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.9444302313442064e-05,
+      "loss": 1.3348,
+      "step": 1571500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.94422023478815e-05,
+      "loss": 1.3646,
+      "step": 1572000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.9440102382320925e-05,
+      "loss": 1.3545,
+      "step": 1572500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.9438006616691485e-05,
+      "loss": 1.3508,
+      "step": 1573000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.9435906651130925e-05,
+      "loss": 1.3335,
+      "step": 1573500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.943380668557036e-05,
+      "loss": 1.3367,
+      "step": 1574000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.9431706720009785e-05,
+      "loss": 1.3435,
+      "step": 1574500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.9429606754449226e-05,
+      "loss": 1.3548,
+      "step": 1575000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.9427510988819786e-05,
+      "loss": 1.3785,
+      "step": 1575500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.942541102325922e-05,
+      "loss": 1.3359,
+      "step": 1576000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.942331525762977e-05,
+      "loss": 1.3524,
+      "step": 1576500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.942121529206921e-05,
+      "loss": 1.3399,
+      "step": 1577000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.941911532650865e-05,
+      "loss": 1.3354,
+      "step": 1577500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.941701536094808e-05,
+      "loss": 1.3445,
+      "step": 1578000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.941491539538752e-05,
+      "loss": 1.366,
+      "step": 1578500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.9412815429826954e-05,
+      "loss": 1.3466,
+      "step": 1579000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.941071546426638e-05,
+      "loss": 1.347,
+      "step": 1579500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.940861969863694e-05,
+      "loss": 1.3362,
+      "step": 1580000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.940651973307638e-05,
+      "loss": 1.3347,
+      "step": 1580500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.9404419767515815e-05,
+      "loss": 1.3625,
+      "step": 1581000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.940231980195525e-05,
+      "loss": 1.3505,
+      "step": 1581500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.940021983639468e-05,
+      "loss": 1.3196,
+      "step": 1582000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.9398119870834115e-05,
+      "loss": 1.3699,
+      "step": 1582500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.939601990527355e-05,
+      "loss": 1.3295,
+      "step": 1583000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.939391993971299e-05,
+      "loss": 1.3681,
+      "step": 1583500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.939181997415242e-05,
+      "loss": 1.3205,
+      "step": 1584000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.9389724208522976e-05,
+      "loss": 1.3392,
+      "step": 1584500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.9387624242962416e-05,
+      "loss": 1.3744,
+      "step": 1585000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.938552427740185e-05,
+      "loss": 1.3297,
+      "step": 1585500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.938342431184128e-05,
+      "loss": 1.3485,
+      "step": 1586000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.9381328546211836e-05,
+      "loss": 1.3344,
+      "step": 1586500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.9379228580651277e-05,
+      "loss": 1.3471,
+      "step": 1587000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.937712861509071e-05,
+      "loss": 1.3561,
+      "step": 1587500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.9375028649530143e-05,
+      "loss": 1.3412,
+      "step": 1588000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.9372928683969584e-05,
+      "loss": 1.3467,
+      "step": 1588500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.937083291834014e-05,
+      "loss": 1.3465,
+      "step": 1589000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.936873295277957e-05,
+      "loss": 1.3542,
+      "step": 1589500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.9366632987219004e-05,
+      "loss": 1.3381,
+      "step": 1590000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.9364533021658444e-05,
+      "loss": 1.3452,
+      "step": 1590500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.9362437256029005e-05,
+      "loss": 1.3143,
+      "step": 1591000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.936033729046843e-05,
+      "loss": 1.3282,
+      "step": 1591500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.935823732490787e-05,
+      "loss": 1.3443,
+      "step": 1592000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.9356137359347305e-05,
+      "loss": 1.3675,
+      "step": 1592500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.935403739378674e-05,
+      "loss": 1.3367,
+      "step": 1593000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.935194162815729e-05,
+      "loss": 1.3559,
+      "step": 1593500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.934984166259673e-05,
+      "loss": 1.3653,
+      "step": 1594000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.934774589696729e-05,
+      "loss": 1.3417,
+      "step": 1594500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.9345645931406726e-05,
+      "loss": 1.3747,
+      "step": 1595000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.934354596584616e-05,
+      "loss": 1.3429,
+      "step": 1595500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.934144600028559e-05,
+      "loss": 1.3333,
+      "step": 1596000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.933934603472503e-05,
+      "loss": 1.3325,
+      "step": 1596500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.933724606916446e-05,
+      "loss": 1.3429,
+      "step": 1597000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.93351461036039e-05,
+      "loss": 1.3447,
+      "step": 1597500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.9333046138043334e-05,
+      "loss": 1.3466,
+      "step": 1598000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.933094617248277e-05,
+      "loss": 1.3154,
+      "step": 1598500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.932885040685333e-05,
+      "loss": 1.3398,
+      "step": 1599000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.932675044129276e-05,
+      "loss": 1.3711,
+      "step": 1599500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.9324650475732195e-05,
+      "loss": 1.3525,
+      "step": 1600000
+    },
+    {
+      "epoch": 0.96,
+      "eval_loss": 1.2816615104675293,
+      "eval_runtime": 1114.3312,
+      "eval_samples_per_second": 472.678,
+      "eval_steps_per_second": 78.78,
+      "step": 1600000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.9322554710102755e-05,
+      "loss": 1.3228,
+      "step": 1600500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.932045474454219e-05,
+      "loss": 1.3275,
+      "step": 1601000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.931835477898162e-05,
+      "loss": 1.3305,
+      "step": 1601500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.9316254813421055e-05,
+      "loss": 1.3366,
+      "step": 1602000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.9314154847860495e-05,
+      "loss": 1.3326,
+      "step": 1602500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.931205488229993e-05,
+      "loss": 1.351,
+      "step": 1603000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.930995491673936e-05,
+      "loss": 1.3251,
+      "step": 1603500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.93078549511788e-05,
+      "loss": 1.3241,
+      "step": 1604000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.9305754985618236e-05,
+      "loss": 1.3535,
+      "step": 1604500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.930365502005767e-05,
+      "loss": 1.3476,
+      "step": 1605000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.930155505449711e-05,
+      "loss": 1.3506,
+      "step": 1605500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.929945508893654e-05,
+      "loss": 1.3257,
+      "step": 1606000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.929736352323822e-05,
+      "loss": 1.3537,
+      "step": 1606500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.929526355767765e-05,
+      "loss": 1.3489,
+      "step": 1607000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.929316359211709e-05,
+      "loss": 1.3398,
+      "step": 1607500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.9291063626556524e-05,
+      "loss": 1.3953,
+      "step": 1608000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.928896786092708e-05,
+      "loss": 1.38,
+      "step": 1608500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.928686789536651e-05,
+      "loss": 1.3583,
+      "step": 1609000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.928476792980595e-05,
+      "loss": 1.3318,
+      "step": 1609500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9282667964245385e-05,
+      "loss": 1.3325,
+      "step": 1610000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.928056799868482e-05,
+      "loss": 1.3764,
+      "step": 1610500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.927847223305537e-05,
+      "loss": 1.3424,
+      "step": 1611000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.927637226749481e-05,
+      "loss": 1.3613,
+      "step": 1611500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9274272301934246e-05,
+      "loss": 1.3303,
+      "step": 1612000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.927217233637368e-05,
+      "loss": 1.342,
+      "step": 1612500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.927007237081312e-05,
+      "loss": 1.3667,
+      "step": 1613000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.926797240525255e-05,
+      "loss": 1.3277,
+      "step": 1613500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9265876639623106e-05,
+      "loss": 1.3334,
+      "step": 1614000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9263776674062547e-05,
+      "loss": 1.3126,
+      "step": 1614500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.926167670850198e-05,
+      "loss": 1.3503,
+      "step": 1615000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9259580942872534e-05,
+      "loss": 1.3597,
+      "step": 1615500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.925748097731197e-05,
+      "loss": 1.3893,
+      "step": 1616000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.925538101175141e-05,
+      "loss": 1.3699,
+      "step": 1616500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.925328104619084e-05,
+      "loss": 1.3627,
+      "step": 1617000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9251181080630274e-05,
+      "loss": 1.3352,
+      "step": 1617500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9249081115069714e-05,
+      "loss": 1.3576,
+      "step": 1618000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.924698534944027e-05,
+      "loss": 1.3476,
+      "step": 1618500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.92448853838797e-05,
+      "loss": 1.3203,
+      "step": 1619000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9242785418319135e-05,
+      "loss": 1.3482,
+      "step": 1619500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9240685452758575e-05,
+      "loss": 1.3479,
+      "step": 1620000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.923858968712913e-05,
+      "loss": 1.3514,
+      "step": 1620500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.923648972156856e-05,
+      "loss": 1.3628,
+      "step": 1621000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9234389756008e-05,
+      "loss": 1.3282,
+      "step": 1621500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9232289790447436e-05,
+      "loss": 1.3338,
+      "step": 1622000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.923018982488687e-05,
+      "loss": 1.3206,
+      "step": 1622500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.922808985932631e-05,
+      "loss": 1.3188,
+      "step": 1623000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.922598989376574e-05,
+      "loss": 1.3437,
+      "step": 1623500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9223889928205176e-05,
+      "loss": 1.3719,
+      "step": 1624000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.922178996264462e-05,
+      "loss": 1.3251,
+      "step": 1624500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.921968999708405e-05,
+      "loss": 1.3605,
+      "step": 1625000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9217594231454604e-05,
+      "loss": 1.3332,
+      "step": 1625500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.921549426589404e-05,
+      "loss": 1.3278,
+      "step": 1626000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.921339430033348e-05,
+      "loss": 1.3342,
+      "step": 1626500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.921129433477291e-05,
+      "loss": 1.3519,
+      "step": 1627000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.9209194369212344e-05,
+      "loss": 1.3322,
+      "step": 1627500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.920709440365178e-05,
+      "loss": 1.3127,
+      "step": 1628000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.920499443809121e-05,
+      "loss": 1.3316,
+      "step": 1628500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.920289447253065e-05,
+      "loss": 1.3169,
+      "step": 1629000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.920079870690121e-05,
+      "loss": 1.3298,
+      "step": 1629500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.919869874134064e-05,
+      "loss": 1.327,
+      "step": 1630000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.919659877578007e-05,
+      "loss": 1.3527,
+      "step": 1630500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.919449881021951e-05,
+      "loss": 1.3395,
+      "step": 1631000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.9192398844658946e-05,
+      "loss": 1.3324,
+      "step": 1631500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.919029887909838e-05,
+      "loss": 1.3712,
+      "step": 1632000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.918819891353782e-05,
+      "loss": 1.3527,
+      "step": 1632500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.918609894797725e-05,
+      "loss": 1.3549,
+      "step": 1633000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.9184003182347806e-05,
+      "loss": 1.3562,
+      "step": 1633500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.918190321678724e-05,
+      "loss": 1.3464,
+      "step": 1634000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.91798074511578e-05,
+      "loss": 1.3637,
+      "step": 1634500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.9177707485597234e-05,
+      "loss": 1.3466,
+      "step": 1635000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.9175611719967794e-05,
+      "loss": 1.3584,
+      "step": 1635500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.917351175440723e-05,
+      "loss": 1.3563,
+      "step": 1636000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.917141178884667e-05,
+      "loss": 1.3437,
+      "step": 1636500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.91693118232861e-05,
+      "loss": 1.3621,
+      "step": 1637000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.916721185772553e-05,
+      "loss": 1.3542,
+      "step": 1637500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.916511189216497e-05,
+      "loss": 1.3303,
+      "step": 1638000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.916301612653553e-05,
+      "loss": 1.3375,
+      "step": 1638500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.916091616097496e-05,
+      "loss": 1.3842,
+      "step": 1639000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.9158816195414395e-05,
+      "loss": 1.3391,
+      "step": 1639500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.915671622985383e-05,
+      "loss": 1.3402,
+      "step": 1640000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.915461626429326e-05,
+      "loss": 1.3568,
+      "step": 1640500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.9152516298732696e-05,
+      "loss": 1.3437,
+      "step": 1641000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.9150416333172136e-05,
+      "loss": 1.3553,
+      "step": 1641500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.914832056754269e-05,
+      "loss": 1.3354,
+      "step": 1642000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.914622060198212e-05,
+      "loss": 1.3628,
+      "step": 1642500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.914412063642156e-05,
+      "loss": 1.3646,
+      "step": 1643000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.9142020670861e-05,
+      "loss": 1.3636,
+      "step": 1643500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.913992070530043e-05,
+      "loss": 1.3389,
+      "step": 1644000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.913782073973987e-05,
+      "loss": 1.3594,
+      "step": 1644500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.9135720774179304e-05,
+      "loss": 1.3287,
+      "step": 1645000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.913362500854986e-05,
+      "loss": 1.3302,
+      "step": 1645500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.913152504298929e-05,
+      "loss": 1.3227,
+      "step": 1646000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.912942507742873e-05,
+      "loss": 1.3276,
+      "step": 1646500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.9127325111868165e-05,
+      "loss": 1.3495,
+      "step": 1647000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.91252251463076e-05,
+      "loss": 1.3695,
+      "step": 1647500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.912312518074704e-05,
+      "loss": 1.3119,
+      "step": 1648000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.912102521518647e-05,
+      "loss": 1.3442,
+      "step": 1648500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.911893364948815e-05,
+      "loss": 1.3458,
+      "step": 1649000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.911683368392758e-05,
+      "loss": 1.3335,
+      "step": 1649500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.911473371836702e-05,
+      "loss": 1.3144,
+      "step": 1650000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.911263375280645e-05,
+      "loss": 1.3231,
+      "step": 1650500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.9110533787245886e-05,
+      "loss": 1.3279,
+      "step": 1651000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.9108433821685326e-05,
+      "loss": 1.3703,
+      "step": 1651500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.910633385612476e-05,
+      "loss": 1.3416,
+      "step": 1652000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.910423389056419e-05,
+      "loss": 1.314,
+      "step": 1652500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.9102133925003633e-05,
+      "loss": 1.3466,
+      "step": 1653000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.910003395944307e-05,
+      "loss": 1.374,
+      "step": 1653500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.90979339938825e-05,
+      "loss": 1.3301,
+      "step": 1654000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.909583402832194e-05,
+      "loss": 1.3464,
+      "step": 1654500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.9093738262692494e-05,
+      "loss": 1.3181,
+      "step": 1655000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.909163829713193e-05,
+      "loss": 1.3446,
+      "step": 1655500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.908953833157136e-05,
+      "loss": 1.3494,
+      "step": 1656000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.90874383660108e-05,
+      "loss": 1.3476,
+      "step": 1656500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.908533840045023e-05,
+      "loss": 1.3353,
+      "step": 1657000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.908323843488966e-05,
+      "loss": 1.3224,
+      "step": 1657500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.90811384693291e-05,
+      "loss": 1.3577,
+      "step": 1658000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.9079038503768535e-05,
+      "loss": 1.3249,
+      "step": 1658500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.9076942738139096e-05,
+      "loss": 1.3217,
+      "step": 1659000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.907484277257853e-05,
+      "loss": 1.3337,
+      "step": 1659500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.907274280701796e-05,
+      "loss": 1.3134,
+      "step": 1660000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.9070642841457396e-05,
+      "loss": 1.329,
+      "step": 1660500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.9068542875896836e-05,
+      "loss": 1.3373,
+      "step": 1661000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.906644291033627e-05,
+      "loss": 1.3225,
+      "step": 1661500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.906434714470682e-05,
+      "loss": 1.3423,
+      "step": 1662000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.9062247179146257e-05,
+      "loss": 1.3482,
+      "step": 1662500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.90601472135857e-05,
+      "loss": 1.3441,
+      "step": 1663000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.905804724802513e-05,
+      "loss": 1.3595,
+      "step": 1663500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.9055947282464564e-05,
+      "loss": 1.3429,
+      "step": 1664000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.9053847316904004e-05,
+      "loss": 1.3229,
+      "step": 1664500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.905174735134344e-05,
+      "loss": 1.3678,
+      "step": 1665000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.904965158571399e-05,
+      "loss": 1.3735,
+      "step": 1665500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.9047551620153424e-05,
+      "loss": 1.3369,
+      "step": 1666000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.9045451654592865e-05,
+      "loss": 1.3198,
+      "step": 1666500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.90433516890323e-05,
+      "loss": 1.3055,
+      "step": 1667000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.904125172347174e-05,
+      "loss": 1.3209,
+      "step": 1667500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.903915175791117e-05,
+      "loss": 1.3425,
+      "step": 1668000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.9037055992281725e-05,
+      "loss": 1.2968,
+      "step": 1668500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.903495602672116e-05,
+      "loss": 1.3499,
+      "step": 1669000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.90328560611606e-05,
+      "loss": 1.3247,
+      "step": 1669500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.903075609560003e-05,
+      "loss": 1.3059,
+      "step": 1670000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.9028656130039466e-05,
+      "loss": 1.3355,
+      "step": 1670500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.9026556164478906e-05,
+      "loss": 1.3115,
+      "step": 1671000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.902445619891834e-05,
+      "loss": 1.2839,
+      "step": 1671500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.902236043328889e-05,
+      "loss": 1.2893,
+      "step": 1672000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.902026046772833e-05,
+      "loss": 1.3132,
+      "step": 1672500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.901816050216777e-05,
+      "loss": 1.3137,
+      "step": 1673000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.90160605366072e-05,
+      "loss": 1.3299,
+      "step": 1673500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.9013960571046634e-05,
+      "loss": 1.3091,
+      "step": 1674000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.9011864805417194e-05,
+      "loss": 1.3247,
+      "step": 1674500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.900976483985663e-05,
+      "loss": 1.3192,
+      "step": 1675000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.900766487429606e-05,
+      "loss": 1.3255,
+      "step": 1675500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.90055649087355e-05,
+      "loss": 1.2757,
+      "step": 1676000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.9003464943174935e-05,
+      "loss": 1.3088,
+      "step": 1676500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.900136497761436e-05,
+      "loss": 1.3,
+      "step": 1677000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.899926921198492e-05,
+      "loss": 1.333,
+      "step": 1677500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.899716924642436e-05,
+      "loss": 1.3413,
+      "step": 1678000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8995069280863796e-05,
+      "loss": 1.3359,
+      "step": 1678500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.899296931530323e-05,
+      "loss": 1.3033,
+      "step": 1679000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.899086934974266e-05,
+      "loss": 1.3078,
+      "step": 1679500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.898877358411322e-05,
+      "loss": 1.3102,
+      "step": 1680000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8986673618552656e-05,
+      "loss": 1.3324,
+      "step": 1680500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.898457785292321e-05,
+      "loss": 1.3398,
+      "step": 1681000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.898247788736265e-05,
+      "loss": 1.2893,
+      "step": 1681500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8980377921802084e-05,
+      "loss": 1.2809,
+      "step": 1682000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.897827795624152e-05,
+      "loss": 1.319,
+      "step": 1682500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.897617799068096e-05,
+      "loss": 1.3058,
+      "step": 1683000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.897407802512039e-05,
+      "loss": 1.313,
+      "step": 1683500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.897197805955982e-05,
+      "loss": 1.2913,
+      "step": 1684000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.896987809399926e-05,
+      "loss": 1.2862,
+      "step": 1684500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.896777812843869e-05,
+      "loss": 1.3271,
+      "step": 1685000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.896568236280925e-05,
+      "loss": 1.3049,
+      "step": 1685500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8963582397248685e-05,
+      "loss": 1.3139,
+      "step": 1686000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.896148663161924e-05,
+      "loss": 1.3118,
+      "step": 1686500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.895938666605868e-05,
+      "loss": 1.3058,
+      "step": 1687000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.895728670049811e-05,
+      "loss": 1.3056,
+      "step": 1687500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8955186734937546e-05,
+      "loss": 1.2781,
+      "step": 1688000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8953086769376986e-05,
+      "loss": 1.3212,
+      "step": 1688500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.895098680381641e-05,
+      "loss": 1.3027,
+      "step": 1689000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.894888683825585e-05,
+      "loss": 1.3256,
+      "step": 1689500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8946786872695286e-05,
+      "loss": 1.2971,
+      "step": 1690000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.894468690713472e-05,
+      "loss": 1.3134,
+      "step": 1690500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.894258694157416e-05,
+      "loss": 1.3294,
+      "step": 1691000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8940486976013593e-05,
+      "loss": 1.2934,
+      "step": 1691500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.893839121038415e-05,
+      "loss": 1.2831,
+      "step": 1692000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.893629124482358e-05,
+      "loss": 1.344,
+      "step": 1692500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.893419127926302e-05,
+      "loss": 1.3321,
+      "step": 1693000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.8932091313702454e-05,
+      "loss": 1.3235,
+      "step": 1693500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.892999134814189e-05,
+      "loss": 1.3072,
+      "step": 1694000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.892789558251244e-05,
+      "loss": 1.2983,
+      "step": 1694500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.8925799816883e-05,
+      "loss": 1.3203,
+      "step": 1695000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.892369985132244e-05,
+      "loss": 1.3262,
+      "step": 1695500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.892159988576187e-05,
+      "loss": 1.3245,
+      "step": 1696000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.891949992020131e-05,
+      "loss": 1.3063,
+      "step": 1696500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.891739995464074e-05,
+      "loss": 1.3251,
+      "step": 1697000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.8915299989080176e-05,
+      "loss": 1.2963,
+      "step": 1697500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.8913200023519616e-05,
+      "loss": 1.315,
+      "step": 1698000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.891110005795905e-05,
+      "loss": 1.2855,
+      "step": 1698500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.890900009239848e-05,
+      "loss": 1.3112,
+      "step": 1699000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.8906904326769036e-05,
+      "loss": 1.3363,
+      "step": 1699500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.8904804361208477e-05,
+      "loss": 1.3065,
+      "step": 1700000
+    },
+    {
+      "epoch": 1.02,
+      "eval_loss": 1.2773289680480957,
+      "eval_runtime": 1103.5769,
+      "eval_samples_per_second": 477.284,
+      "eval_steps_per_second": 79.548,
+      "step": 1700000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.890270439564791e-05,
+      "loss": 1.3451,
+      "step": 1700500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.8900604430087344e-05,
+      "loss": 1.3006,
+      "step": 1701000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.8898504464526784e-05,
+      "loss": 1.2997,
+      "step": 1701500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.889640449896622e-05,
+      "loss": 1.3124,
+      "step": 1702000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.889430453340565e-05,
+      "loss": 1.2988,
+      "step": 1702500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.8892208767776204e-05,
+      "loss": 1.3237,
+      "step": 1703000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.8890108802215644e-05,
+      "loss": 1.3079,
+      "step": 1703500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.888800883665508e-05,
+      "loss": 1.3231,
+      "step": 1704000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.888590887109451e-05,
+      "loss": 1.3401,
+      "step": 1704500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.888380890553395e-05,
+      "loss": 1.2761,
+      "step": 1705000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.8881708939973385e-05,
+      "loss": 1.2937,
+      "step": 1705500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.887960897441282e-05,
+      "loss": 1.32,
+      "step": 1706000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.887751320878338e-05,
+      "loss": 1.3185,
+      "step": 1706500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.887541324322281e-05,
+      "loss": 1.3029,
+      "step": 1707000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.8873313277662246e-05,
+      "loss": 1.3474,
+      "step": 1707500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.8871213312101686e-05,
+      "loss": 1.302,
+      "step": 1708000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.886911754647224e-05,
+      "loss": 1.2993,
+      "step": 1708500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.886701758091167e-05,
+      "loss": 1.2953,
+      "step": 1709000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.886492181528223e-05,
+      "loss": 1.3249,
+      "step": 1709500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.886282184972166e-05,
+      "loss": 1.3487,
+      "step": 1710000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.88607218841611e-05,
+      "loss": 1.3395,
+      "step": 1710500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.8858621918600534e-05,
+      "loss": 1.3049,
+      "step": 1711000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.885652195303997e-05,
+      "loss": 1.3145,
+      "step": 1711500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.885442198747941e-05,
+      "loss": 1.3196,
+      "step": 1712000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.885232202191884e-05,
+      "loss": 1.3258,
+      "step": 1712500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.8850222056358274e-05,
+      "loss": 1.2732,
+      "step": 1713000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.884812209079771e-05,
+      "loss": 1.3186,
+      "step": 1713500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.884602212523714e-05,
+      "loss": 1.3046,
+      "step": 1714000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.884392215967658e-05,
+      "loss": 1.3048,
+      "step": 1714500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.8841822194116015e-05,
+      "loss": 1.3228,
+      "step": 1715000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.883972222855545e-05,
+      "loss": 1.3315,
+      "step": 1715500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.883762226299489e-05,
+      "loss": 1.3386,
+      "step": 1716000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.883552229743432e-05,
+      "loss": 1.3365,
+      "step": 1716500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.8833422331873756e-05,
+      "loss": 1.2803,
+      "step": 1717000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.8831322366313196e-05,
+      "loss": 1.2987,
+      "step": 1717500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.882923080061487e-05,
+      "loss": 1.3279,
+      "step": 1718000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.88271308350543e-05,
+      "loss": 1.3276,
+      "step": 1718500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.8825030869493737e-05,
+      "loss": 1.2691,
+      "step": 1719000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.882293090393317e-05,
+      "loss": 1.3368,
+      "step": 1719500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.882083093837261e-05,
+      "loss": 1.3238,
+      "step": 1720000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.8818735172743164e-05,
+      "loss": 1.335,
+      "step": 1720500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.88166352071826e-05,
+      "loss": 1.2921,
+      "step": 1721000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.881453524162204e-05,
+      "loss": 1.3196,
+      "step": 1721500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.881243527606147e-05,
+      "loss": 1.2938,
+      "step": 1722000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.8810335310500904e-05,
+      "loss": 1.3096,
+      "step": 1722500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.8808235344940345e-05,
+      "loss": 1.2825,
+      "step": 1723000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.88061395793109e-05,
+      "loss": 1.3051,
+      "step": 1723500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.880403961375033e-05,
+      "loss": 1.3161,
+      "step": 1724000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.8801939648189765e-05,
+      "loss": 1.2923,
+      "step": 1724500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.8799843882560325e-05,
+      "loss": 1.3401,
+      "step": 1725000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.879774391699976e-05,
+      "loss": 1.3084,
+      "step": 1725500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.879564395143919e-05,
+      "loss": 1.316,
+      "step": 1726000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.8793543985878626e-05,
+      "loss": 1.3236,
+      "step": 1726500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.8791444020318066e-05,
+      "loss": 1.3192,
+      "step": 1727000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.87893440547575e-05,
+      "loss": 1.298,
+      "step": 1727500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.878724828912805e-05,
+      "loss": 1.2932,
+      "step": 1728000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.878514832356749e-05,
+      "loss": 1.3344,
+      "step": 1728500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.878304835800693e-05,
+      "loss": 1.3266,
+      "step": 1729000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.878094839244636e-05,
+      "loss": 1.3025,
+      "step": 1729500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.87788484268858e-05,
+      "loss": 1.3232,
+      "step": 1730000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.8776752661256354e-05,
+      "loss": 1.3291,
+      "step": 1730500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.877465269569579e-05,
+      "loss": 1.3149,
+      "step": 1731000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.877255273013522e-05,
+      "loss": 1.3003,
+      "step": 1731500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.877045276457466e-05,
+      "loss": 1.3323,
+      "step": 1732000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.8768352799014095e-05,
+      "loss": 1.3121,
+      "step": 1732500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.876625283345353e-05,
+      "loss": 1.2915,
+      "step": 1733000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.876415286789297e-05,
+      "loss": 1.2812,
+      "step": 1733500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.87620529023324e-05,
+      "loss": 1.2925,
+      "step": 1734000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.8759952936771835e-05,
+      "loss": 1.2923,
+      "step": 1734500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.8757852971211276e-05,
+      "loss": 1.3062,
+      "step": 1735000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.875575720558183e-05,
+      "loss": 1.308,
+      "step": 1735500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.875365724002126e-05,
+      "loss": 1.2999,
+      "step": 1736000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.87515572744607e-05,
+      "loss": 1.3146,
+      "step": 1736500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.8749457308900136e-05,
+      "loss": 1.3186,
+      "step": 1737000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.874735734333957e-05,
+      "loss": 1.3461,
+      "step": 1737500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.874526157771012e-05,
+      "loss": 1.2961,
+      "step": 1738000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.8743161612149564e-05,
+      "loss": 1.2979,
+      "step": 1738500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.874106584652012e-05,
+      "loss": 1.3106,
+      "step": 1739000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.873896588095955e-05,
+      "loss": 1.3428,
+      "step": 1739500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.8736865915398984e-05,
+      "loss": 1.3389,
+      "step": 1740000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.8734765949838424e-05,
+      "loss": 1.3272,
+      "step": 1740500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.873266598427786e-05,
+      "loss": 1.3012,
+      "step": 1741000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.873056601871729e-05,
+      "loss": 1.3113,
+      "step": 1741500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.872846605315673e-05,
+      "loss": 1.3202,
+      "step": 1742000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.8726366087596165e-05,
+      "loss": 1.32,
+      "step": 1742500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.87242661220356e-05,
+      "loss": 1.303,
+      "step": 1743000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.872217035640616e-05,
+      "loss": 1.3142,
+      "step": 1743500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.872007039084559e-05,
+      "loss": 1.297,
+      "step": 1744000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.8717970425285026e-05,
+      "loss": 1.3163,
+      "step": 1744500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.8715870459724466e-05,
+      "loss": 1.3082,
+      "step": 1745000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.871377049416389e-05,
+      "loss": 1.2977,
+      "step": 1745500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.8711670528603326e-05,
+      "loss": 1.3109,
+      "step": 1746000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.8709574762973886e-05,
+      "loss": 1.309,
+      "step": 1746500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.8707474797413327e-05,
+      "loss": 1.2974,
+      "step": 1747000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.870537483185275e-05,
+      "loss": 1.3414,
+      "step": 1747500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.870327486629219e-05,
+      "loss": 1.341,
+      "step": 1748000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.870117490073163e-05,
+      "loss": 1.3161,
+      "step": 1748500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.869907493517106e-05,
+      "loss": 1.324,
+      "step": 1749000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.869697916954162e-05,
+      "loss": 1.3014,
+      "step": 1749500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.8694879203981054e-05,
+      "loss": 1.3037,
+      "step": 1750000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.869277923842049e-05,
+      "loss": 1.3504,
+      "step": 1750500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.869067927285992e-05,
+      "loss": 1.3312,
+      "step": 1751000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.868857930729936e-05,
+      "loss": 1.3136,
+      "step": 1751500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.868648354166992e-05,
+      "loss": 1.3492,
+      "step": 1752000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.868438357610935e-05,
+      "loss": 1.2909,
+      "step": 1752500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.868228361054878e-05,
+      "loss": 1.3058,
+      "step": 1753000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.868018364498822e-05,
+      "loss": 1.3117,
+      "step": 1753500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.867808787935878e-05,
+      "loss": 1.2739,
+      "step": 1754000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.8675987913798216e-05,
+      "loss": 1.3095,
+      "step": 1754500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.867388794823764e-05,
+      "loss": 1.308,
+      "step": 1755000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.867178798267708e-05,
+      "loss": 1.3201,
+      "step": 1755500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.8669688017116516e-05,
+      "loss": 1.2708,
+      "step": 1756000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.866759225148708e-05,
+      "loss": 1.3014,
+      "step": 1756500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.866549228592651e-05,
+      "loss": 1.3001,
+      "step": 1757000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.8663392320365944e-05,
+      "loss": 1.3107,
+      "step": 1757500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.866129235480538e-05,
+      "loss": 1.3307,
+      "step": 1758000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.865919238924482e-05,
+      "loss": 1.2898,
+      "step": 1758500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.865709242368425e-05,
+      "loss": 1.3414,
+      "step": 1759000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.8654996658054804e-05,
+      "loss": 1.3133,
+      "step": 1759500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.865289669249424e-05,
+      "loss": 1.3219,
+      "step": 1760000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.86508009268648e-05,
+      "loss": 1.3287,
+      "step": 1760500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.864870096130424e-05,
+      "loss": 1.3176,
+      "step": 1761000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.864660099574367e-05,
+      "loss": 1.3227,
+      "step": 1761500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.86445010301831e-05,
+      "loss": 1.3296,
+      "step": 1762000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.864240106462254e-05,
+      "loss": 1.3105,
+      "step": 1762500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.864030109906197e-05,
+      "loss": 1.3107,
+      "step": 1763000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.863820533343253e-05,
+      "loss": 1.3732,
+      "step": 1763500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.8636105367871966e-05,
+      "loss": 1.3197,
+      "step": 1764000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.86340054023114e-05,
+      "loss": 1.3488,
+      "step": 1764500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.863190543675083e-05,
+      "loss": 1.3445,
+      "step": 1765000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.862980547119027e-05,
+      "loss": 1.3561,
+      "step": 1765500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.862771390549195e-05,
+      "loss": 1.3153,
+      "step": 1766000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.862561393993139e-05,
+      "loss": 1.3101,
+      "step": 1766500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.862351397437082e-05,
+      "loss": 1.3025,
+      "step": 1767000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.8621414008810254e-05,
+      "loss": 1.3141,
+      "step": 1767500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.8619314043249694e-05,
+      "loss": 1.3163,
+      "step": 1768000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.861721407768913e-05,
+      "loss": 1.3264,
+      "step": 1768500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.8615114112128554e-05,
+      "loss": 1.2973,
+      "step": 1769000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.8613014146567995e-05,
+      "loss": 1.3279,
+      "step": 1769500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.8610918380938555e-05,
+      "loss": 1.3032,
+      "step": 1770000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.860881841537799e-05,
+      "loss": 1.3418,
+      "step": 1770500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.860671844981742e-05,
+      "loss": 1.3011,
+      "step": 1771000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.860462268418798e-05,
+      "loss": 1.3262,
+      "step": 1771500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.8602522718627416e-05,
+      "loss": 1.3351,
+      "step": 1772000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.860042275306685e-05,
+      "loss": 1.3056,
+      "step": 1772500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.859832278750629e-05,
+      "loss": 1.3281,
+      "step": 1773000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.859622282194572e-05,
+      "loss": 1.2955,
+      "step": 1773500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.859412285638515e-05,
+      "loss": 1.3369,
+      "step": 1774000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.859202289082459e-05,
+      "loss": 1.3167,
+      "step": 1774500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.858992292526402e-05,
+      "loss": 1.3314,
+      "step": 1775000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.858782295970346e-05,
+      "loss": 1.3354,
+      "step": 1775500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.85857229941429e-05,
+      "loss": 1.3033,
+      "step": 1776000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.858362302858233e-05,
+      "loss": 1.3206,
+      "step": 1776500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.8581523063021764e-05,
+      "loss": 1.348,
+      "step": 1777000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.8579423097461204e-05,
+      "loss": 1.3127,
+      "step": 1777500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.857732733183176e-05,
+      "loss": 1.3249,
+      "step": 1778000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.857523156620231e-05,
+      "loss": 1.2975,
+      "step": 1778500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.8573131600641745e-05,
+      "loss": 1.3334,
+      "step": 1779000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.8571031635081185e-05,
+      "loss": 1.3137,
+      "step": 1779500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.856893166952062e-05,
+      "loss": 1.3106,
+      "step": 1780000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.856683170396005e-05,
+      "loss": 1.3027,
+      "step": 1780500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.8564735938330605e-05,
+      "loss": 1.2983,
+      "step": 1781000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.8562635972770046e-05,
+      "loss": 1.2961,
+      "step": 1781500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.856053600720948e-05,
+      "loss": 1.3247,
+      "step": 1782000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.855843604164891e-05,
+      "loss": 1.3572,
+      "step": 1782500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.855633607608835e-05,
+      "loss": 1.3089,
+      "step": 1783000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.8554236110527786e-05,
+      "loss": 1.3045,
+      "step": 1783500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.855213614496722e-05,
+      "loss": 1.3004,
+      "step": 1784000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.855003617940666e-05,
+      "loss": 1.3203,
+      "step": 1784500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.854793621384609e-05,
+      "loss": 1.3277,
+      "step": 1785000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.854583624828553e-05,
+      "loss": 1.343,
+      "step": 1785500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.854373628272497e-05,
+      "loss": 1.3062,
+      "step": 1786000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.8541636317164394e-05,
+      "loss": 1.312,
+      "step": 1786500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.8539540551534954e-05,
+      "loss": 1.3076,
+      "step": 1787000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.853744478590551e-05,
+      "loss": 1.3162,
+      "step": 1787500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.853534482034495e-05,
+      "loss": 1.3062,
+      "step": 1788000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.853324485478438e-05,
+      "loss": 1.3056,
+      "step": 1788500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.8531144889223815e-05,
+      "loss": 1.3177,
+      "step": 1789000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.8529044923663255e-05,
+      "loss": 1.3211,
+      "step": 1789500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.852694495810269e-05,
+      "loss": 1.3162,
+      "step": 1790000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.852484499254212e-05,
+      "loss": 1.3199,
+      "step": 1790500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.8522749226912676e-05,
+      "loss": 1.3287,
+      "step": 1791000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.8520649261352116e-05,
+      "loss": 1.3033,
+      "step": 1791500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.851854929579155e-05,
+      "loss": 1.3347,
+      "step": 1792000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.851644933023098e-05,
+      "loss": 1.3048,
+      "step": 1792500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.851434936467042e-05,
+      "loss": 1.3333,
+      "step": 1793000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.851224939910985e-05,
+      "loss": 1.2907,
+      "step": 1793500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.851014943354929e-05,
+      "loss": 1.3106,
+      "step": 1794000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.850804946798872e-05,
+      "loss": 1.3237,
+      "step": 1794500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.8505953702359284e-05,
+      "loss": 1.3053,
+      "step": 1795000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.850385373679872e-05,
+      "loss": 1.3079,
+      "step": 1795500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.850175797116927e-05,
+      "loss": 1.3212,
+      "step": 1796000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.849965800560871e-05,
+      "loss": 1.3209,
+      "step": 1796500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.8497558040048144e-05,
+      "loss": 1.3323,
+      "step": 1797000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.849545807448758e-05,
+      "loss": 1.3424,
+      "step": 1797500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.849335810892702e-05,
+      "loss": 1.3093,
+      "step": 1798000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.8491258143366445e-05,
+      "loss": 1.2874,
+      "step": 1798500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.848915817780588e-05,
+      "loss": 1.338,
+      "step": 1799000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.848705821224532e-05,
+      "loss": 1.3179,
+      "step": 1799500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.848496244661588e-05,
+      "loss": 1.3104,
+      "step": 1800000
+    },
+    {
+      "epoch": 1.08,
+      "eval_loss": 1.270274043083191,
+      "eval_runtime": 1102.2881,
+      "eval_samples_per_second": 477.842,
+      "eval_steps_per_second": 79.641,
+      "step": 1800000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.848286668098643e-05,
+      "loss": 1.2991,
+      "step": 1800500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.8480766715425866e-05,
+      "loss": 1.3134,
+      "step": 1801000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.8478666749865306e-05,
+      "loss": 1.311,
+      "step": 1801500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.847656678430474e-05,
+      "loss": 1.3283,
+      "step": 1802000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.847447101867529e-05,
+      "loss": 1.3142,
+      "step": 1802500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.8472371053114727e-05,
+      "loss": 1.3167,
+      "step": 1803000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.847027108755417e-05,
+      "loss": 1.287,
+      "step": 1803500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.84681711219936e-05,
+      "loss": 1.3142,
+      "step": 1804000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.8466071156433034e-05,
+      "loss": 1.3273,
+      "step": 1804500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.8463971190872474e-05,
+      "loss": 1.2969,
+      "step": 1805000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.846187542524303e-05,
+      "loss": 1.2837,
+      "step": 1805500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.845977545968246e-05,
+      "loss": 1.3223,
+      "step": 1806000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.8457675494121894e-05,
+      "loss": 1.3271,
+      "step": 1806500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.8455575528561335e-05,
+      "loss": 1.3276,
+      "step": 1807000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.845347556300077e-05,
+      "loss": 1.3072,
+      "step": 1807500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.84513755974402e-05,
+      "loss": 1.2955,
+      "step": 1808000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.8449275631879635e-05,
+      "loss": 1.3455,
+      "step": 1808500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.8447179866250195e-05,
+      "loss": 1.3101,
+      "step": 1809000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.844507990068963e-05,
+      "loss": 1.3247,
+      "step": 1809500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.844297993512907e-05,
+      "loss": 1.3299,
+      "step": 1810000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.8440879969568496e-05,
+      "loss": 1.2931,
+      "step": 1810500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.843878000400793e-05,
+      "loss": 1.3211,
+      "step": 1811000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.843668423837849e-05,
+      "loss": 1.3119,
+      "step": 1811500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.843458427281793e-05,
+      "loss": 1.3255,
+      "step": 1812000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.843248430725736e-05,
+      "loss": 1.2941,
+      "step": 1812500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.843038434169679e-05,
+      "loss": 1.2795,
+      "step": 1813000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.842828437613623e-05,
+      "loss": 1.3158,
+      "step": 1813500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.842618861050679e-05,
+      "loss": 1.3327,
+      "step": 1814000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.8424088644946224e-05,
+      "loss": 1.2865,
+      "step": 1814500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.842198867938566e-05,
+      "loss": 1.343,
+      "step": 1815000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.841988871382509e-05,
+      "loss": 1.3282,
+      "step": 1815500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.8417788748264524e-05,
+      "loss": 1.3172,
+      "step": 1816000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.8415688782703965e-05,
+      "loss": 1.3135,
+      "step": 1816500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.84135888171434e-05,
+      "loss": 1.3162,
+      "step": 1817000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.841149305151395e-05,
+      "loss": 1.2869,
+      "step": 1817500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.8409393085953385e-05,
+      "loss": 1.3241,
+      "step": 1818000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.8407293120392825e-05,
+      "loss": 1.3076,
+      "step": 1818500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.8405197354763386e-05,
+      "loss": 1.3266,
+      "step": 1819000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.840309738920282e-05,
+      "loss": 1.3018,
+      "step": 1819500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.8400997423642246e-05,
+      "loss": 1.3167,
+      "step": 1820000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.8398897458081686e-05,
+      "loss": 1.328,
+      "step": 1820500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.839679749252112e-05,
+      "loss": 1.3197,
+      "step": 1821000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.839469752696055e-05,
+      "loss": 1.3096,
+      "step": 1821500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.839259756139999e-05,
+      "loss": 1.3213,
+      "step": 1822000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.839049759583943e-05,
+      "loss": 1.3035,
+      "step": 1822500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.838839763027886e-05,
+      "loss": 1.3237,
+      "step": 1823000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.83862976647183e-05,
+      "loss": 1.3145,
+      "step": 1823500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.8384201899088854e-05,
+      "loss": 1.342,
+      "step": 1824000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.838210193352829e-05,
+      "loss": 1.3149,
+      "step": 1824500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.838000196796773e-05,
+      "loss": 1.3287,
+      "step": 1825000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.837790200240716e-05,
+      "loss": 1.3375,
+      "step": 1825500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.8375802036846595e-05,
+      "loss": 1.3295,
+      "step": 1826000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.837370627121715e-05,
+      "loss": 1.3244,
+      "step": 1826500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.837160630565659e-05,
+      "loss": 1.3153,
+      "step": 1827000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.836950634009602e-05,
+      "loss": 1.3397,
+      "step": 1827500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.8367406374535455e-05,
+      "loss": 1.2851,
+      "step": 1828000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.8365306408974896e-05,
+      "loss": 1.3149,
+      "step": 1828500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.836321064334545e-05,
+      "loss": 1.3114,
+      "step": 1829000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.836111067778488e-05,
+      "loss": 1.3317,
+      "step": 1829500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.8359010712224316e-05,
+      "loss": 1.3038,
+      "step": 1830000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.8356910746663756e-05,
+      "loss": 1.3123,
+      "step": 1830500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.835481078110319e-05,
+      "loss": 1.3107,
+      "step": 1831000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.835271081554262e-05,
+      "loss": 1.2752,
+      "step": 1831500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.8350610849982063e-05,
+      "loss": 1.2903,
+      "step": 1832000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.834851088442149e-05,
+      "loss": 1.2883,
+      "step": 1832500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.834641511879205e-05,
+      "loss": 1.3159,
+      "step": 1833000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.834431515323149e-05,
+      "loss": 1.3215,
+      "step": 1833500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.8342215187670924e-05,
+      "loss": 1.3161,
+      "step": 1834000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.834011522211036e-05,
+      "loss": 1.3116,
+      "step": 1834500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.833801525654979e-05,
+      "loss": 1.3314,
+      "step": 1835000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.833591949092035e-05,
+      "loss": 1.2894,
+      "step": 1835500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.8333819525359785e-05,
+      "loss": 1.3186,
+      "step": 1836000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.833171955979922e-05,
+      "loss": 1.3133,
+      "step": 1836500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.832961959423866e-05,
+      "loss": 1.2812,
+      "step": 1837000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.832752802854033e-05,
+      "loss": 1.3165,
+      "step": 1837500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.8325428062979766e-05,
+      "loss": 1.3302,
+      "step": 1838000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.83233280974192e-05,
+      "loss": 1.2881,
+      "step": 1838500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.832122813185864e-05,
+      "loss": 1.2915,
+      "step": 1839000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.831912816629807e-05,
+      "loss": 1.3224,
+      "step": 1839500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.8317028200737506e-05,
+      "loss": 1.3222,
+      "step": 1840000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.8314928235176947e-05,
+      "loss": 1.3011,
+      "step": 1840500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.831282826961638e-05,
+      "loss": 1.2879,
+      "step": 1841000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.8310732503986934e-05,
+      "loss": 1.2982,
+      "step": 1841500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.830863253842637e-05,
+      "loss": 1.3242,
+      "step": 1842000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.830653677279692e-05,
+      "loss": 1.3228,
+      "step": 1842500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.830443680723636e-05,
+      "loss": 1.3108,
+      "step": 1843000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.8302336841675794e-05,
+      "loss": 1.3055,
+      "step": 1843500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.830024107604635e-05,
+      "loss": 1.2989,
+      "step": 1844000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.829814111048579e-05,
+      "loss": 1.3046,
+      "step": 1844500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.829604114492522e-05,
+      "loss": 1.3159,
+      "step": 1845000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.8293941179364655e-05,
+      "loss": 1.3203,
+      "step": 1845500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.8291841213804095e-05,
+      "loss": 1.3156,
+      "step": 1846000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.828974124824353e-05,
+      "loss": 1.3104,
+      "step": 1846500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.828764128268296e-05,
+      "loss": 1.3332,
+      "step": 1847000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.82855413171224e-05,
+      "loss": 1.3068,
+      "step": 1847500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.8283441351561836e-05,
+      "loss": 1.3318,
+      "step": 1848000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.828134138600127e-05,
+      "loss": 1.327,
+      "step": 1848500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.827924142044071e-05,
+      "loss": 1.3263,
+      "step": 1849000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.8277141454880136e-05,
+      "loss": 1.2962,
+      "step": 1849500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.82750456892507e-05,
+      "loss": 1.2958,
+      "step": 1850000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.827294992362125e-05,
+      "loss": 1.307,
+      "step": 1850500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.8270854157991804e-05,
+      "loss": 1.3069,
+      "step": 1851000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.8268754192431244e-05,
+      "loss": 1.3428,
+      "step": 1851500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.826665422687068e-05,
+      "loss": 1.3033,
+      "step": 1852000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.826455426131011e-05,
+      "loss": 1.3086,
+      "step": 1852500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.826245429574955e-05,
+      "loss": 1.2995,
+      "step": 1853000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.8260354330188985e-05,
+      "loss": 1.3417,
+      "step": 1853500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.825825436462842e-05,
+      "loss": 1.3133,
+      "step": 1854000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.825615439906786e-05,
+      "loss": 1.2966,
+      "step": 1854500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.825405863343841e-05,
+      "loss": 1.2957,
+      "step": 1855000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.8251958667877845e-05,
+      "loss": 1.3265,
+      "step": 1855500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.824985870231728e-05,
+      "loss": 1.325,
+      "step": 1856000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.824775873675672e-05,
+      "loss": 1.3063,
+      "step": 1856500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.824565877119615e-05,
+      "loss": 1.3177,
+      "step": 1857000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.8243558805635586e-05,
+      "loss": 1.3145,
+      "step": 1857500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.8241458840075026e-05,
+      "loss": 1.3118,
+      "step": 1858000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.823936307444558e-05,
+      "loss": 1.3287,
+      "step": 1858500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.823726310888501e-05,
+      "loss": 1.3177,
+      "step": 1859000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.8235163143324453e-05,
+      "loss": 1.338,
+      "step": 1859500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.823306317776389e-05,
+      "loss": 1.297,
+      "step": 1860000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.823096321220332e-05,
+      "loss": 1.3539,
+      "step": 1860500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.8228863246642754e-05,
+      "loss": 1.3059,
+      "step": 1861000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.822676328108219e-05,
+      "loss": 1.2969,
+      "step": 1861500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.822466331552162e-05,
+      "loss": 1.2838,
+      "step": 1862000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.822256334996106e-05,
+      "loss": 1.282,
+      "step": 1862500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.822046758433162e-05,
+      "loss": 1.3411,
+      "step": 1863000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.821836761877105e-05,
+      "loss": 1.3203,
+      "step": 1863500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.821626765321048e-05,
+      "loss": 1.3352,
+      "step": 1864000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.821416768764992e-05,
+      "loss": 1.3241,
+      "step": 1864500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.8212067722089355e-05,
+      "loss": 1.3355,
+      "step": 1865000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.820996775652879e-05,
+      "loss": 1.3013,
+      "step": 1865500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.820786779096823e-05,
+      "loss": 1.3189,
+      "step": 1866000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.820577202533878e-05,
+      "loss": 1.2824,
+      "step": 1866500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.820367625970934e-05,
+      "loss": 1.3141,
+      "step": 1867000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.8201576294148776e-05,
+      "loss": 1.3096,
+      "step": 1867500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.8199476328588217e-05,
+      "loss": 1.2998,
+      "step": 1868000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.819737636302764e-05,
+      "loss": 1.3126,
+      "step": 1868500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.819527639746708e-05,
+      "loss": 1.3126,
+      "step": 1869000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.819317643190652e-05,
+      "loss": 1.3258,
+      "step": 1869500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.819107646634595e-05,
+      "loss": 1.2946,
+      "step": 1870000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.8188976500785384e-05,
+      "loss": 1.3413,
+      "step": 1870500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.8186876535224824e-05,
+      "loss": 1.3415,
+      "step": 1871000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.818477656966426e-05,
+      "loss": 1.3305,
+      "step": 1871500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.818267660410369e-05,
+      "loss": 1.3167,
+      "step": 1872000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.8180580838474245e-05,
+      "loss": 1.2958,
+      "step": 1872500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.8178480872913685e-05,
+      "loss": 1.3007,
+      "step": 1873000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.817638090735312e-05,
+      "loss": 1.2843,
+      "step": 1873500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.817428094179255e-05,
+      "loss": 1.3086,
+      "step": 1874000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.817218097623199e-05,
+      "loss": 1.2905,
+      "step": 1874500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.8170081010671425e-05,
+      "loss": 1.3166,
+      "step": 1875000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.816798104511086e-05,
+      "loss": 1.2994,
+      "step": 1875500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.816588107955029e-05,
+      "loss": 1.3281,
+      "step": 1876000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.816378531392085e-05,
+      "loss": 1.2857,
+      "step": 1876500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.8161689548291406e-05,
+      "loss": 1.3176,
+      "step": 1877000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.815958958273084e-05,
+      "loss": 1.2986,
+      "step": 1877500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.815748961717028e-05,
+      "loss": 1.3129,
+      "step": 1878000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.8155393851540834e-05,
+      "loss": 1.3136,
+      "step": 1878500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.8153298085911394e-05,
+      "loss": 1.3019,
+      "step": 1879000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.815119812035083e-05,
+      "loss": 1.3076,
+      "step": 1879500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.814909815479026e-05,
+      "loss": 1.297,
+      "step": 1880000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.8146998189229694e-05,
+      "loss": 1.3049,
+      "step": 1880500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.814489822366913e-05,
+      "loss": 1.3001,
+      "step": 1881000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.814279825810857e-05,
+      "loss": 1.3266,
+      "step": 1881500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.8140698292548e-05,
+      "loss": 1.3026,
+      "step": 1882000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.8138598326987435e-05,
+      "loss": 1.3462,
+      "step": 1882500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.8136498361426875e-05,
+      "loss": 1.326,
+      "step": 1883000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.813439839586631e-05,
+      "loss": 1.3078,
+      "step": 1883500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.813229843030574e-05,
+      "loss": 1.3208,
+      "step": 1884000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.813019846474518e-05,
+      "loss": 1.3004,
+      "step": 1884500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.8128098499184616e-05,
+      "loss": 1.316,
+      "step": 1885000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.812599853362405e-05,
+      "loss": 1.2925,
+      "step": 1885500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.812389856806348e-05,
+      "loss": 1.3095,
+      "step": 1886000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.8121798602502916e-05,
+      "loss": 1.3264,
+      "step": 1886500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.8119702836873476e-05,
+      "loss": 1.2915,
+      "step": 1887000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.811760707124403e-05,
+      "loss": 1.3195,
+      "step": 1887500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.8115511305614584e-05,
+      "loss": 1.2971,
+      "step": 1888000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.8113411340054024e-05,
+      "loss": 1.2841,
+      "step": 1888500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.811131137449346e-05,
+      "loss": 1.329,
+      "step": 1889000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.810921140893289e-05,
+      "loss": 1.2787,
+      "step": 1889500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.810711144337233e-05,
+      "loss": 1.3185,
+      "step": 1890000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.8105015677742885e-05,
+      "loss": 1.3155,
+      "step": 1890500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.810291571218232e-05,
+      "loss": 1.3171,
+      "step": 1891000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.810081574662175e-05,
+      "loss": 1.2975,
+      "step": 1891500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.809871578106119e-05,
+      "loss": 1.3036,
+      "step": 1892000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.8096615815500625e-05,
+      "loss": 1.3092,
+      "step": 1892500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.809452004987118e-05,
+      "loss": 1.3449,
+      "step": 1893000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.809242428424174e-05,
+      "loss": 1.3265,
+      "step": 1893500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.809032431868117e-05,
+      "loss": 1.3099,
+      "step": 1894000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.8088224353120606e-05,
+      "loss": 1.3159,
+      "step": 1894500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.808612438756004e-05,
+      "loss": 1.3047,
+      "step": 1895000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.808402442199948e-05,
+      "loss": 1.3174,
+      "step": 1895500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.808192445643891e-05,
+      "loss": 1.3332,
+      "step": 1896000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.8079824490878347e-05,
+      "loss": 1.2987,
+      "step": 1896500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.807772452531779e-05,
+      "loss": 1.2993,
+      "step": 1897000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.807562455975722e-05,
+      "loss": 1.3207,
+      "step": 1897500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.8073524594196654e-05,
+      "loss": 1.275,
+      "step": 1898000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.807142882856721e-05,
+      "loss": 1.3002,
+      "step": 1898500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.806932886300665e-05,
+      "loss": 1.3351,
+      "step": 1899000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.806722889744608e-05,
+      "loss": 1.2986,
+      "step": 1899500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.8065128931885514e-05,
+      "loss": 1.3284,
+      "step": 1900000
+    },
+    {
+      "epoch": 1.14,
+      "eval_loss": 1.2703109979629517,
+      "eval_runtime": 1101.196,
+      "eval_samples_per_second": 478.316,
+      "eval_steps_per_second": 79.72,
+      "step": 1900000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.8063028966324955e-05,
+      "loss": 1.2887,
+      "step": 1900500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.806092900076439e-05,
+      "loss": 1.3282,
+      "step": 1901000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.805882903520382e-05,
+      "loss": 1.3289,
+      "step": 1901500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.8056733269574375e-05,
+      "loss": 1.3267,
+      "step": 1902000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.8054633304013815e-05,
+      "loss": 1.3294,
+      "step": 1902500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.805253333845325e-05,
+      "loss": 1.2941,
+      "step": 1903000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.805043337289268e-05,
+      "loss": 1.2935,
+      "step": 1903500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.804833340733212e-05,
+      "loss": 1.3163,
+      "step": 1904000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.8046233441771556e-05,
+      "loss": 1.3318,
+      "step": 1904500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.804413767614211e-05,
+      "loss": 1.3048,
+      "step": 1905000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.804203771058155e-05,
+      "loss": 1.334,
+      "step": 1905500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.803993774502098e-05,
+      "loss": 1.2887,
+      "step": 1906000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.803783777946042e-05,
+      "loss": 1.292,
+      "step": 1906500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.803573781389985e-05,
+      "loss": 1.3307,
+      "step": 1907000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.8033637848339284e-05,
+      "loss": 1.2953,
+      "step": 1907500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.803153788277872e-05,
+      "loss": 1.2881,
+      "step": 1908000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.802944211714928e-05,
+      "loss": 1.3179,
+      "step": 1908500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.802734635151983e-05,
+      "loss": 1.3172,
+      "step": 1909000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.802524638595927e-05,
+      "loss": 1.3109,
+      "step": 1909500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.8023146420398705e-05,
+      "loss": 1.289,
+      "step": 1910000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.8021046454838145e-05,
+      "loss": 1.2974,
+      "step": 1910500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.801894648927758e-05,
+      "loss": 1.3036,
+      "step": 1911000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.801684652371701e-05,
+      "loss": 1.2854,
+      "step": 1911500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.8014746558156445e-05,
+      "loss": 1.2929,
+      "step": 1912000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.801264659259588e-05,
+      "loss": 1.3062,
+      "step": 1912500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.801054662703531e-05,
+      "loss": 1.3065,
+      "step": 1913000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.800844666147475e-05,
+      "loss": 1.2574,
+      "step": 1913500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.800635089584531e-05,
+      "loss": 1.3001,
+      "step": 1914000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.800425093028474e-05,
+      "loss": 1.2838,
+      "step": 1914500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.800215096472417e-05,
+      "loss": 1.306,
+      "step": 1915000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.800005099916361e-05,
+      "loss": 1.3196,
+      "step": 1915500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.799795103360305e-05,
+      "loss": 1.3106,
+      "step": 1916000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.799585106804248e-05,
+      "loss": 1.3306,
+      "step": 1916500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.799375110248192e-05,
+      "loss": 1.3111,
+      "step": 1917000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.7991651136921354e-05,
+      "loss": 1.3301,
+      "step": 1917500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.798955117136079e-05,
+      "loss": 1.2936,
+      "step": 1918000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.798745120580023e-05,
+      "loss": 1.3296,
+      "step": 1918500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.798535124023966e-05,
+      "loss": 1.3108,
+      "step": 1919000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.7983251274679095e-05,
+      "loss": 1.3224,
+      "step": 1919500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.7981155509049655e-05,
+      "loss": 1.2901,
+      "step": 1920000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.797905554348909e-05,
+      "loss": 1.3028,
+      "step": 1920500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.797695977785964e-05,
+      "loss": 1.3334,
+      "step": 1921000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.7974859812299075e-05,
+      "loss": 1.3049,
+      "step": 1921500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.7972759846738516e-05,
+      "loss": 1.2857,
+      "step": 1922000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.797065988117795e-05,
+      "loss": 1.3086,
+      "step": 1922500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.796855991561738e-05,
+      "loss": 1.3116,
+      "step": 1923000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.796645995005682e-05,
+      "loss": 1.3488,
+      "step": 1923500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.7964359984496256e-05,
+      "loss": 1.3126,
+      "step": 1924000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.796226421886681e-05,
+      "loss": 1.3128,
+      "step": 1924500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.796016425330624e-05,
+      "loss": 1.3525,
+      "step": 1925000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.7958064287745683e-05,
+      "loss": 1.3269,
+      "step": 1925500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.795596852211624e-05,
+      "loss": 1.3106,
+      "step": 1926000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.795386855655567e-05,
+      "loss": 1.3043,
+      "step": 1926500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.795176859099511e-05,
+      "loss": 1.3259,
+      "step": 1927000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.7949668625434544e-05,
+      "loss": 1.3182,
+      "step": 1927500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.794756865987398e-05,
+      "loss": 1.3094,
+      "step": 1928000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.794547289424453e-05,
+      "loss": 1.3304,
+      "step": 1928500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.794337292868397e-05,
+      "loss": 1.3227,
+      "step": 1929000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.7941272963123405e-05,
+      "loss": 1.3197,
+      "step": 1929500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.793917719749396e-05,
+      "loss": 1.3179,
+      "step": 1930000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.793707723193339e-05,
+      "loss": 1.3055,
+      "step": 1930500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.793497726637283e-05,
+      "loss": 1.3084,
+      "step": 1931000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.7932877300812266e-05,
+      "loss": 1.2987,
+      "step": 1931500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.79307773352517e-05,
+      "loss": 1.3032,
+      "step": 1932000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.792867736969114e-05,
+      "loss": 1.3378,
+      "step": 1932500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.792657740413057e-05,
+      "loss": 1.3174,
+      "step": 1933000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.7924481638501126e-05,
+      "loss": 1.3101,
+      "step": 1933500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.792238167294057e-05,
+      "loss": 1.3142,
+      "step": 1934000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.792028170738e-05,
+      "loss": 1.3066,
+      "step": 1934500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.7918181741819434e-05,
+      "loss": 1.3203,
+      "step": 1935000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.7916081776258874e-05,
+      "loss": 1.3089,
+      "step": 1935500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.791398181069831e-05,
+      "loss": 1.3238,
+      "step": 1936000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.7911881845137734e-05,
+      "loss": 1.3104,
+      "step": 1936500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.7909781879577174e-05,
+      "loss": 1.3061,
+      "step": 1937000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.790768191401661e-05,
+      "loss": 1.3394,
+      "step": 1937500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.790558194845604e-05,
+      "loss": 1.3253,
+      "step": 1938000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.790348198289548e-05,
+      "loss": 1.3198,
+      "step": 1938500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.7901386217266035e-05,
+      "loss": 1.307,
+      "step": 1939000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.789928625170547e-05,
+      "loss": 1.2937,
+      "step": 1939500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.78971862861449e-05,
+      "loss": 1.2981,
+      "step": 1940000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.789508632058434e-05,
+      "loss": 1.3396,
+      "step": 1940500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.7892986355023775e-05,
+      "loss": 1.3233,
+      "step": 1941000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.789089058939433e-05,
+      "loss": 1.2959,
+      "step": 1941500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.788879062383377e-05,
+      "loss": 1.3036,
+      "step": 1942000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.78866906582732e-05,
+      "loss": 1.319,
+      "step": 1942500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.7884590692712636e-05,
+      "loss": 1.2911,
+      "step": 1943000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.7882490727152076e-05,
+      "loss": 1.3313,
+      "step": 1943500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.788039076159151e-05,
+      "loss": 1.308,
+      "step": 1944000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.787829079603094e-05,
+      "loss": 1.3128,
+      "step": 1944500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.7876190830470384e-05,
+      "loss": 1.2849,
+      "step": 1945000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.787409506484094e-05,
+      "loss": 1.3079,
+      "step": 1945500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.787199509928037e-05,
+      "loss": 1.2978,
+      "step": 1946000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.7869895133719804e-05,
+      "loss": 1.3392,
+      "step": 1946500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.7867795168159244e-05,
+      "loss": 1.303,
+      "step": 1947000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.786569520259868e-05,
+      "loss": 1.2798,
+      "step": 1947500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.786359523703811e-05,
+      "loss": 1.3084,
+      "step": 1948000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.786149527147755e-05,
+      "loss": 1.3264,
+      "step": 1948500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.785939530591698e-05,
+      "loss": 1.2991,
+      "step": 1949000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.785729954028754e-05,
+      "loss": 1.3265,
+      "step": 1949500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.785519957472697e-05,
+      "loss": 1.2822,
+      "step": 1950000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.785309960916641e-05,
+      "loss": 1.3209,
+      "step": 1950500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.7851003843536966e-05,
+      "loss": 1.3073,
+      "step": 1951000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.78489038779764e-05,
+      "loss": 1.3184,
+      "step": 1951500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.784680391241584e-05,
+      "loss": 1.2972,
+      "step": 1952000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.784470394685527e-05,
+      "loss": 1.3109,
+      "step": 1952500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.7842603981294706e-05,
+      "loss": 1.3136,
+      "step": 1953000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.784050401573415e-05,
+      "loss": 1.2922,
+      "step": 1953500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.78384082501047e-05,
+      "loss": 1.326,
+      "step": 1954000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.7836308284544134e-05,
+      "loss": 1.3251,
+      "step": 1954500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.783420831898357e-05,
+      "loss": 1.3091,
+      "step": 1955000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.783210835342301e-05,
+      "loss": 1.2829,
+      "step": 1955500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.783001258779356e-05,
+      "loss": 1.3083,
+      "step": 1956000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.7827912622232994e-05,
+      "loss": 1.2976,
+      "step": 1956500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.7825812656672435e-05,
+      "loss": 1.3159,
+      "step": 1957000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.782371269111187e-05,
+      "loss": 1.318,
+      "step": 1957500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.78216127255513e-05,
+      "loss": 1.3248,
+      "step": 1958000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.781951275999074e-05,
+      "loss": 1.3238,
+      "step": 1958500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.781741279443017e-05,
+      "loss": 1.31,
+      "step": 1959000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.781531702880073e-05,
+      "loss": 1.2805,
+      "step": 1959500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.781321706324016e-05,
+      "loss": 1.3078,
+      "step": 1960000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.78111170976796e-05,
+      "loss": 1.3041,
+      "step": 1960500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.780901713211903e-05,
+      "loss": 1.3363,
+      "step": 1961000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.780691716655846e-05,
+      "loss": 1.3228,
+      "step": 1961500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.78048172009979e-05,
+      "loss": 1.302,
+      "step": 1962000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.7802717235437336e-05,
+      "loss": 1.3133,
+      "step": 1962500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.780061726987677e-05,
+      "loss": 1.3215,
+      "step": 1963000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.779851730431621e-05,
+      "loss": 1.3024,
+      "step": 1963500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.7796421538686764e-05,
+      "loss": 1.3371,
+      "step": 1964000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.77943215731262e-05,
+      "loss": 1.3041,
+      "step": 1964500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.779222160756564e-05,
+      "loss": 1.3172,
+      "step": 1965000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.779012164200507e-05,
+      "loss": 1.3548,
+      "step": 1965500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.7788021676444504e-05,
+      "loss": 1.3079,
+      "step": 1966000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.7785921710883944e-05,
+      "loss": 1.3098,
+      "step": 1966500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.77838259452545e-05,
+      "loss": 1.2921,
+      "step": 1967000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.778172597969393e-05,
+      "loss": 1.2957,
+      "step": 1967500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.7779626014133365e-05,
+      "loss": 1.3378,
+      "step": 1968000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.777753024850392e-05,
+      "loss": 1.306,
+      "step": 1968500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.777543028294336e-05,
+      "loss": 1.3239,
+      "step": 1969000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.777333451731392e-05,
+      "loss": 1.3367,
+      "step": 1969500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.777123455175335e-05,
+      "loss": 1.3068,
+      "step": 1970000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.7769134586192786e-05,
+      "loss": 1.3171,
+      "step": 1970500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.776703462063222e-05,
+      "loss": 1.2998,
+      "step": 1971000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.776493465507165e-05,
+      "loss": 1.3171,
+      "step": 1971500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.776283468951109e-05,
+      "loss": 1.31,
+      "step": 1972000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.776073472395053e-05,
+      "loss": 1.2945,
+      "step": 1972500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.775863475838996e-05,
+      "loss": 1.336,
+      "step": 1973000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.7756538992760514e-05,
+      "loss": 1.2934,
+      "step": 1973500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.7754439027199954e-05,
+      "loss": 1.2931,
+      "step": 1974000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.775233906163939e-05,
+      "loss": 1.2957,
+      "step": 1974500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.775023909607882e-05,
+      "loss": 1.2815,
+      "step": 1975000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.7748143330449374e-05,
+      "loss": 1.3078,
+      "step": 1975500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.7746043364888815e-05,
+      "loss": 1.3086,
+      "step": 1976000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.774394339932825e-05,
+      "loss": 1.3303,
+      "step": 1976500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.774184343376768e-05,
+      "loss": 1.2859,
+      "step": 1977000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.773974346820712e-05,
+      "loss": 1.3045,
+      "step": 1977500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.7737643502646555e-05,
+      "loss": 1.2856,
+      "step": 1978000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.773554353708599e-05,
+      "loss": 1.3258,
+      "step": 1978500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.773344357152543e-05,
+      "loss": 1.3055,
+      "step": 1979000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.773134360596486e-05,
+      "loss": 1.3127,
+      "step": 1979500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.7729243640404296e-05,
+      "loss": 1.3333,
+      "step": 1980000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.7727143674843736e-05,
+      "loss": 1.3097,
+      "step": 1980500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.772504370928316e-05,
+      "loss": 1.3109,
+      "step": 1981000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.772294794365372e-05,
+      "loss": 1.3105,
+      "step": 1981500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.772084797809316e-05,
+      "loss": 1.3171,
+      "step": 1982000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.77187480125326e-05,
+      "loss": 1.3272,
+      "step": 1982500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.771664804697203e-05,
+      "loss": 1.3086,
+      "step": 1983000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.7714548081411464e-05,
+      "loss": 1.3165,
+      "step": 1983500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.7712452315782024e-05,
+      "loss": 1.298,
+      "step": 1984000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.771035235022146e-05,
+      "loss": 1.3231,
+      "step": 1984500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.770825238466089e-05,
+      "loss": 1.3212,
+      "step": 1985000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.7706152419100324e-05,
+      "loss": 1.2968,
+      "step": 1985500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.7704056653470885e-05,
+      "loss": 1.3224,
+      "step": 1986000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.770195668791032e-05,
+      "loss": 1.3082,
+      "step": 1986500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.769985672234975e-05,
+      "loss": 1.2747,
+      "step": 1987000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.769776095672031e-05,
+      "loss": 1.2951,
+      "step": 1987500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.7695660991159746e-05,
+      "loss": 1.2899,
+      "step": 1988000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.769356102559918e-05,
+      "loss": 1.3238,
+      "step": 1988500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.769146106003862e-05,
+      "loss": 1.3103,
+      "step": 1989000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.768936109447805e-05,
+      "loss": 1.3129,
+      "step": 1989500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.7687261128917486e-05,
+      "loss": 1.3034,
+      "step": 1990000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.768516116335692e-05,
+      "loss": 1.3063,
+      "step": 1990500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.768306119779635e-05,
+      "loss": 1.2932,
+      "step": 1991000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.7680961232235787e-05,
+      "loss": 1.3289,
+      "step": 1991500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.767886126667523e-05,
+      "loss": 1.3119,
+      "step": 1992000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.767676130111466e-05,
+      "loss": 1.2969,
+      "step": 1992500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.7674665535485214e-05,
+      "loss": 1.3076,
+      "step": 1993000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.767256556992465e-05,
+      "loss": 1.3159,
+      "step": 1993500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.767046980429521e-05,
+      "loss": 1.2829,
+      "step": 1994000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.766836983873465e-05,
+      "loss": 1.2993,
+      "step": 1994500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.766626987317408e-05,
+      "loss": 1.3246,
+      "step": 1995000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.7664169907613515e-05,
+      "loss": 1.2838,
+      "step": 1995500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.766206994205295e-05,
+      "loss": 1.3062,
+      "step": 1996000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.765997417642351e-05,
+      "loss": 1.3042,
+      "step": 1996500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.765787421086294e-05,
+      "loss": 1.3223,
+      "step": 1997000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.7655774245302376e-05,
+      "loss": 1.2662,
+      "step": 1997500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.765367427974181e-05,
+      "loss": 1.3111,
+      "step": 1998000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.765157431418124e-05,
+      "loss": 1.3282,
+      "step": 1998500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.764947434862068e-05,
+      "loss": 1.2805,
+      "step": 1999000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.7647374383060116e-05,
+      "loss": 1.2951,
+      "step": 1999500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.764527441749955e-05,
+      "loss": 1.3019,
+      "step": 2000000
+    },
+    {
+      "epoch": 1.2,
+      "eval_loss": 1.2635042667388916,
+      "eval_runtime": 1110.2941,
+      "eval_samples_per_second": 474.397,
+      "eval_steps_per_second": 79.066,
+      "step": 2000000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.764317445193899e-05,
+      "loss": 1.3184,
+      "step": 2000500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.764107868630954e-05,
+      "loss": 1.3177,
+      "step": 2001000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.763897872074898e-05,
+      "loss": 1.2882,
+      "step": 2001500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.763687875518841e-05,
+      "loss": 1.2976,
+      "step": 2002000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.763477878962785e-05,
+      "loss": 1.3246,
+      "step": 2002500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.7632678824067284e-05,
+      "loss": 1.3032,
+      "step": 2003000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.7630578858506724e-05,
+      "loss": 1.3143,
+      "step": 2003500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.762847889294616e-05,
+      "loss": 1.3001,
+      "step": 2004000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.762637892738559e-05,
+      "loss": 1.3211,
+      "step": 2004500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.7624283161756145e-05,
+      "loss": 1.3073,
+      "step": 2005000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.7622183196195585e-05,
+      "loss": 1.309,
+      "step": 2005500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.762008743056614e-05,
+      "loss": 1.3084,
+      "step": 2006000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.761798746500557e-05,
+      "loss": 1.3333,
+      "step": 2006500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.7615887499445005e-05,
+      "loss": 1.2914,
+      "step": 2007000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.7613787533884446e-05,
+      "loss": 1.2925,
+      "step": 2007500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.761168756832388e-05,
+      "loss": 1.309,
+      "step": 2008000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.760958760276331e-05,
+      "loss": 1.3079,
+      "step": 2008500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.760748763720275e-05,
+      "loss": 1.3071,
+      "step": 2009000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.7605391871573306e-05,
+      "loss": 1.3132,
+      "step": 2009500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.760329610594386e-05,
+      "loss": 1.2999,
+      "step": 2010000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.7601196140383293e-05,
+      "loss": 1.3272,
+      "step": 2010500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.7599096174822734e-05,
+      "loss": 1.3097,
+      "step": 2011000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.759699620926217e-05,
+      "loss": 1.2843,
+      "step": 2011500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.75948962437016e-05,
+      "loss": 1.3241,
+      "step": 2012000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.759279627814104e-05,
+      "loss": 1.3017,
+      "step": 2012500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.7590696312580474e-05,
+      "loss": 1.3028,
+      "step": 2013000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.758859634701991e-05,
+      "loss": 1.2988,
+      "step": 2013500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.758649638145935e-05,
+      "loss": 1.3152,
+      "step": 2014000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.75844006158299e-05,
+      "loss": 1.3362,
+      "step": 2014500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.7582300650269335e-05,
+      "loss": 1.3022,
+      "step": 2015000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.758020068470877e-05,
+      "loss": 1.2848,
+      "step": 2015500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.757810071914821e-05,
+      "loss": 1.3017,
+      "step": 2016000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.757600075358764e-05,
+      "loss": 1.3187,
+      "step": 2016500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.7573900788027076e-05,
+      "loss": 1.2904,
+      "step": 2017000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.757180082246651e-05,
+      "loss": 1.3214,
+      "step": 2017500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.756970085690594e-05,
+      "loss": 1.2859,
+      "step": 2018000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.756760929120762e-05,
+      "loss": 1.3173,
+      "step": 2018500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.7565509325647056e-05,
+      "loss": 1.3088,
+      "step": 2019000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.75634093600865e-05,
+      "loss": 1.3237,
+      "step": 2019500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.756130939452593e-05,
+      "loss": 1.3135,
+      "step": 2020000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.7559209428965364e-05,
+      "loss": 1.315,
+      "step": 2020500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.7557109463404804e-05,
+      "loss": 1.3039,
+      "step": 2021000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.755500949784424e-05,
+      "loss": 1.3214,
+      "step": 2021500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.755291373221479e-05,
+      "loss": 1.2855,
+      "step": 2022000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.7550813766654224e-05,
+      "loss": 1.3199,
+      "step": 2022500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.7548713801093665e-05,
+      "loss": 1.3314,
+      "step": 2023000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.75466138355331e-05,
+      "loss": 1.3271,
+      "step": 2023500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.754451386997253e-05,
+      "loss": 1.3423,
+      "step": 2024000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.7542413904411965e-05,
+      "loss": 1.323,
+      "step": 2024500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.75403139388514e-05,
+      "loss": 1.3235,
+      "step": 2025000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.753821397329084e-05,
+      "loss": 1.3155,
+      "step": 2025500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.75361182076614e-05,
+      "loss": 1.293,
+      "step": 2026000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.753402244203195e-05,
+      "loss": 1.3104,
+      "step": 2026500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.7531922476471386e-05,
+      "loss": 1.3364,
+      "step": 2027000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.752982251091082e-05,
+      "loss": 1.279,
+      "step": 2027500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.752772254535026e-05,
+      "loss": 1.3082,
+      "step": 2028000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.752562257978969e-05,
+      "loss": 1.3133,
+      "step": 2028500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.752353101409137e-05,
+      "loss": 1.2863,
+      "step": 2029000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.75214310485308e-05,
+      "loss": 1.2735,
+      "step": 2029500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.751933108297024e-05,
+      "loss": 1.3269,
+      "step": 2030000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.7517231117409674e-05,
+      "loss": 1.3441,
+      "step": 2030500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.751513115184911e-05,
+      "loss": 1.3173,
+      "step": 2031000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.751303118628855e-05,
+      "loss": 1.328,
+      "step": 2031500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.751093122072798e-05,
+      "loss": 1.2935,
+      "step": 2032000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.7508831255167415e-05,
+      "loss": 1.2998,
+      "step": 2032500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.7506731289606855e-05,
+      "loss": 1.2958,
+      "step": 2033000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.750463132404629e-05,
+      "loss": 1.293,
+      "step": 2033500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.750253975834796e-05,
+      "loss": 1.3184,
+      "step": 2034000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.7500439792787396e-05,
+      "loss": 1.3178,
+      "step": 2034500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.749833982722683e-05,
+      "loss": 1.2973,
+      "step": 2035000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.749623986166627e-05,
+      "loss": 1.3073,
+      "step": 2035500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.74941398961057e-05,
+      "loss": 1.3038,
+      "step": 2036000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.7492039930545136e-05,
+      "loss": 1.3156,
+      "step": 2036500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.7489939964984576e-05,
+      "loss": 1.3153,
+      "step": 2037000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.748783999942401e-05,
+      "loss": 1.2881,
+      "step": 2037500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.748574003386344e-05,
+      "loss": 1.3233,
+      "step": 2038000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.7483640068302884e-05,
+      "loss": 1.3069,
+      "step": 2038500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.748154010274231e-05,
+      "loss": 1.2989,
+      "step": 2039000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.747944013718175e-05,
+      "loss": 1.3174,
+      "step": 2039500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.747734437155231e-05,
+      "loss": 1.2975,
+      "step": 2040000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.7475244405991744e-05,
+      "loss": 1.3095,
+      "step": 2040500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.747314444043118e-05,
+      "loss": 1.296,
+      "step": 2041000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.747104867480173e-05,
+      "loss": 1.3202,
+      "step": 2041500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.746894870924117e-05,
+      "loss": 1.3006,
+      "step": 2042000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.7466848743680605e-05,
+      "loss": 1.3178,
+      "step": 2042500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.746474877812004e-05,
+      "loss": 1.3315,
+      "step": 2043000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.746264881255947e-05,
+      "loss": 1.326,
+      "step": 2043500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.7460548846998905e-05,
+      "loss": 1.3472,
+      "step": 2044000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.7458453081369466e-05,
+      "loss": 1.2894,
+      "step": 2044500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.74563531158089e-05,
+      "loss": 1.3077,
+      "step": 2045000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.745425315024834e-05,
+      "loss": 1.3049,
+      "step": 2045500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.7452153184687766e-05,
+      "loss": 1.3121,
+      "step": 2046000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.7450053219127206e-05,
+      "loss": 1.3135,
+      "step": 2046500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.744795325356664e-05,
+      "loss": 1.2927,
+      "step": 2047000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.744585328800607e-05,
+      "loss": 1.3167,
+      "step": 2047500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.7443757522376634e-05,
+      "loss": 1.3012,
+      "step": 2048000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.744165755681607e-05,
+      "loss": 1.2855,
+      "step": 2048500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.74395575912555e-05,
+      "loss": 1.3016,
+      "step": 2049000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.7437457625694934e-05,
+      "loss": 1.2813,
+      "step": 2049500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.7435357660134374e-05,
+      "loss": 1.304,
+      "step": 2050000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.743325769457381e-05,
+      "loss": 1.3497,
+      "step": 2050500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.743116192894436e-05,
+      "loss": 1.3306,
+      "step": 2051000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.7429061963383795e-05,
+      "loss": 1.3091,
+      "step": 2051500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.7426961997823235e-05,
+      "loss": 1.3144,
+      "step": 2052000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.742486203226267e-05,
+      "loss": 1.303,
+      "step": 2052500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.74227620667021e-05,
+      "loss": 1.3204,
+      "step": 2053000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.742066210114154e-05,
+      "loss": 1.2898,
+      "step": 2053500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.7418562135580976e-05,
+      "loss": 1.3483,
+      "step": 2054000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.741646217002041e-05,
+      "loss": 1.2907,
+      "step": 2054500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.741436640439097e-05,
+      "loss": 1.3236,
+      "step": 2055000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.74122664388304e-05,
+      "loss": 1.3411,
+      "step": 2055500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.7410166473269836e-05,
+      "loss": 1.3246,
+      "step": 2056000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.740807070764039e-05,
+      "loss": 1.2847,
+      "step": 2056500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.740597074207983e-05,
+      "loss": 1.3262,
+      "step": 2057000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.7403870776519264e-05,
+      "loss": 1.2978,
+      "step": 2057500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.74017708109587e-05,
+      "loss": 1.3048,
+      "step": 2058000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.739967084539814e-05,
+      "loss": 1.3204,
+      "step": 2058500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.739757507976869e-05,
+      "loss": 1.3146,
+      "step": 2059000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.7395475114208124e-05,
+      "loss": 1.2996,
+      "step": 2059500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.739337514864756e-05,
+      "loss": 1.3284,
+      "step": 2060000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.7391275183087e-05,
+      "loss": 1.3154,
+      "step": 2060500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.738917941745755e-05,
+      "loss": 1.2837,
+      "step": 2061000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.7387079451896985e-05,
+      "loss": 1.2931,
+      "step": 2061500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.7384979486336425e-05,
+      "loss": 1.285,
+      "step": 2062000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.738287952077586e-05,
+      "loss": 1.2819,
+      "step": 2062500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.738077955521529e-05,
+      "loss": 1.328,
+      "step": 2063000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.737867958965473e-05,
+      "loss": 1.3267,
+      "step": 2063500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.7376579624094166e-05,
+      "loss": 1.3084,
+      "step": 2064000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.73744796585336e-05,
+      "loss": 1.3122,
+      "step": 2064500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.737237969297304e-05,
+      "loss": 1.3371,
+      "step": 2065000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.737028392734359e-05,
+      "loss": 1.3154,
+      "step": 2065500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.736818816171415e-05,
+      "loss": 1.3268,
+      "step": 2066000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.736608819615358e-05,
+      "loss": 1.2935,
+      "step": 2066500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.7363988230593014e-05,
+      "loss": 1.3219,
+      "step": 2067000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.7361888265032454e-05,
+      "loss": 1.3166,
+      "step": 2067500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.735979249940301e-05,
+      "loss": 1.3136,
+      "step": 2068000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.735769253384244e-05,
+      "loss": 1.2952,
+      "step": 2068500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.735559256828188e-05,
+      "loss": 1.2956,
+      "step": 2069000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.735349680265244e-05,
+      "loss": 1.3205,
+      "step": 2069500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.735139683709187e-05,
+      "loss": 1.3191,
+      "step": 2070000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.73492968715313e-05,
+      "loss": 1.2867,
+      "step": 2070500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.734719690597074e-05,
+      "loss": 1.3288,
+      "step": 2071000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.7345096940410175e-05,
+      "loss": 1.2941,
+      "step": 2071500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.734299697484961e-05,
+      "loss": 1.2976,
+      "step": 2072000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.734089700928905e-05,
+      "loss": 1.3322,
+      "step": 2072500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.733879704372848e-05,
+      "loss": 1.3032,
+      "step": 2073000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.7336697078167916e-05,
+      "loss": 1.3135,
+      "step": 2073500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.7334597112607356e-05,
+      "loss": 1.2917,
+      "step": 2074000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.733249714704679e-05,
+      "loss": 1.3004,
+      "step": 2074500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.733039718148622e-05,
+      "loss": 1.3116,
+      "step": 2075000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.7328301415856783e-05,
+      "loss": 1.2761,
+      "step": 2075500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.732620145029622e-05,
+      "loss": 1.343,
+      "step": 2076000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.732410148473565e-05,
+      "loss": 1.3075,
+      "step": 2076500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.732200151917509e-05,
+      "loss": 1.2946,
+      "step": 2077000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.7319901553614524e-05,
+      "loss": 1.3313,
+      "step": 2077500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.731780158805395e-05,
+      "loss": 1.3118,
+      "step": 2078000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.731570162249339e-05,
+      "loss": 1.307,
+      "step": 2078500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.7313601656932824e-05,
+      "loss": 1.2916,
+      "step": 2079000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.7311510091234505e-05,
+      "loss": 1.3316,
+      "step": 2079500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.730941012567394e-05,
+      "loss": 1.2948,
+      "step": 2080000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.730731016011337e-05,
+      "loss": 1.2968,
+      "step": 2080500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.730521019455281e-05,
+      "loss": 1.2793,
+      "step": 2081000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.7303110228992245e-05,
+      "loss": 1.3118,
+      "step": 2081500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.730101026343168e-05,
+      "loss": 1.2868,
+      "step": 2082000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.729891029787111e-05,
+      "loss": 1.2819,
+      "step": 2082500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.729681453224167e-05,
+      "loss": 1.2951,
+      "step": 2083000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.7294714566681106e-05,
+      "loss": 1.2996,
+      "step": 2083500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.7292614601120546e-05,
+      "loss": 1.2857,
+      "step": 2084000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.729051463555998e-05,
+      "loss": 1.2822,
+      "step": 2084500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.7288414669999407e-05,
+      "loss": 1.3085,
+      "step": 2085000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.728631470443885e-05,
+      "loss": 1.2799,
+      "step": 2085500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.728421473887828e-05,
+      "loss": 1.3053,
+      "step": 2086000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.7282114773317714e-05,
+      "loss": 1.2769,
+      "step": 2086500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.7280019007688274e-05,
+      "loss": 1.3155,
+      "step": 2087000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.727791904212771e-05,
+      "loss": 1.2979,
+      "step": 2087500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.727581907656714e-05,
+      "loss": 1.3011,
+      "step": 2088000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.7273719111006574e-05,
+      "loss": 1.2925,
+      "step": 2088500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.7271627545308255e-05,
+      "loss": 1.3319,
+      "step": 2089000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.7269527579747695e-05,
+      "loss": 1.3107,
+      "step": 2089500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.726742761418713e-05,
+      "loss": 1.3185,
+      "step": 2090000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.726532764862656e-05,
+      "loss": 1.2748,
+      "step": 2090500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.7263227683066e-05,
+      "loss": 1.3045,
+      "step": 2091000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.7261127717505436e-05,
+      "loss": 1.2925,
+      "step": 2091500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.725902775194486e-05,
+      "loss": 1.3007,
+      "step": 2092000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.725693198631542e-05,
+      "loss": 1.3188,
+      "step": 2092500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.725483202075486e-05,
+      "loss": 1.314,
+      "step": 2093000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.7252732055194296e-05,
+      "loss": 1.3045,
+      "step": 2093500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.725063208963373e-05,
+      "loss": 1.2829,
+      "step": 2094000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.7248536324004284e-05,
+      "loss": 1.2943,
+      "step": 2094500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.7246436358443724e-05,
+      "loss": 1.2947,
+      "step": 2095000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.724433639288316e-05,
+      "loss": 1.306,
+      "step": 2095500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.724223642732259e-05,
+      "loss": 1.2993,
+      "step": 2096000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.724013646176203e-05,
+      "loss": 1.3149,
+      "step": 2096500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.723803649620146e-05,
+      "loss": 1.2976,
+      "step": 2097000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.723594073057202e-05,
+      "loss": 1.3189,
+      "step": 2097500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.723384076501146e-05,
+      "loss": 1.3227,
+      "step": 2098000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.723174079945089e-05,
+      "loss": 1.2949,
+      "step": 2098500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.7229640833890325e-05,
+      "loss": 1.2889,
+      "step": 2099000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.722754086832976e-05,
+      "loss": 1.3133,
+      "step": 2099500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.722544090276919e-05,
+      "loss": 1.2928,
+      "step": 2100000
+    },
+    {
+      "epoch": 1.26,
+      "eval_loss": 1.2566254138946533,
+      "eval_runtime": 1098.8033,
+      "eval_samples_per_second": 479.358,
+      "eval_steps_per_second": 79.893,
+      "step": 2100000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.722334513713975e-05,
+      "loss": 1.3077,
+      "step": 2100500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.7221245171579186e-05,
+      "loss": 1.2959,
+      "step": 2101000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.721914520601862e-05,
+      "loss": 1.3162,
+      "step": 2101500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.721704524045805e-05,
+      "loss": 1.2964,
+      "step": 2102000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.7214945274897486e-05,
+      "loss": 1.2677,
+      "step": 2102500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.7212845309336926e-05,
+      "loss": 1.3008,
+      "step": 2103000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.721074954370749e-05,
+      "loss": 1.2803,
+      "step": 2103500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.7208649578146913e-05,
+      "loss": 1.3284,
+      "step": 2104000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.7206549612586354e-05,
+      "loss": 1.3021,
+      "step": 2104500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.720444964702579e-05,
+      "loss": 1.319,
+      "step": 2105000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.720234968146522e-05,
+      "loss": 1.3188,
+      "step": 2105500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.720025391583578e-05,
+      "loss": 1.299,
+      "step": 2106000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.7198153950275214e-05,
+      "loss": 1.292,
+      "step": 2106500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.719605398471465e-05,
+      "loss": 1.2733,
+      "step": 2107000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.719395401915408e-05,
+      "loss": 1.2849,
+      "step": 2107500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.719185405359352e-05,
+      "loss": 1.3065,
+      "step": 2108000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.7189754088032955e-05,
+      "loss": 1.2831,
+      "step": 2108500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.718765412247239e-05,
+      "loss": 1.2901,
+      "step": 2109000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.718555415691183e-05,
+      "loss": 1.2861,
+      "step": 2109500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.71834625912135e-05,
+      "loss": 1.308,
+      "step": 2110000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.718136262565294e-05,
+      "loss": 1.308,
+      "step": 2110500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.717926266009237e-05,
+      "loss": 1.3069,
+      "step": 2111000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.717716269453181e-05,
+      "loss": 1.2779,
+      "step": 2111500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.717506272897124e-05,
+      "loss": 1.2905,
+      "step": 2112000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.7172962763410677e-05,
+      "loss": 1.3082,
+      "step": 2112500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.717086279785012e-05,
+      "loss": 1.3273,
+      "step": 2113000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.716876283228955e-05,
+      "loss": 1.3164,
+      "step": 2113500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.7166662866728984e-05,
+      "loss": 1.31,
+      "step": 2114000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.7164562901168424e-05,
+      "loss": 1.2888,
+      "step": 2114500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.716246713553898e-05,
+      "loss": 1.3108,
+      "step": 2115000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.716036716997841e-05,
+      "loss": 1.2832,
+      "step": 2115500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.7158267204417844e-05,
+      "loss": 1.2932,
+      "step": 2116000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.7156167238857285e-05,
+      "loss": 1.3313,
+      "step": 2116500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.715406727329672e-05,
+      "loss": 1.3278,
+      "step": 2117000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.715196730773615e-05,
+      "loss": 1.304,
+      "step": 2117500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.714986734217559e-05,
+      "loss": 1.313,
+      "step": 2118000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.7147767376615025e-05,
+      "loss": 1.2911,
+      "step": 2118500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.714567161098558e-05,
+      "loss": 1.3017,
+      "step": 2119000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.714357164542501e-05,
+      "loss": 1.3371,
+      "step": 2119500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.714147167986445e-05,
+      "loss": 1.289,
+      "step": 2120000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.7139371714303886e-05,
+      "loss": 1.3024,
+      "step": 2120500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.713727174874332e-05,
+      "loss": 1.3215,
+      "step": 2121000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.713517178318275e-05,
+      "loss": 1.2975,
+      "step": 2121500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.7133071817622186e-05,
+      "loss": 1.299,
+      "step": 2122000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.7130971852061627e-05,
+      "loss": 1.291,
+      "step": 2122500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.71288802863633e-05,
+      "loss": 1.3185,
+      "step": 2123000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.712678032080274e-05,
+      "loss": 1.3085,
+      "step": 2123500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.7124680355242174e-05,
+      "loss": 1.2923,
+      "step": 2124000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.712258038968161e-05,
+      "loss": 1.2792,
+      "step": 2124500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.712048042412105e-05,
+      "loss": 1.307,
+      "step": 2125000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.711838045856048e-05,
+      "loss": 1.2888,
+      "step": 2125500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.711628049299991e-05,
+      "loss": 1.3012,
+      "step": 2126000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.711418052743935e-05,
+      "loss": 1.303,
+      "step": 2126500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.711208056187878e-05,
+      "loss": 1.3162,
+      "step": 2127000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.710998479624934e-05,
+      "loss": 1.2963,
+      "step": 2127500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.710788483068878e-05,
+      "loss": 1.2893,
+      "step": 2128000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.710578486512821e-05,
+      "loss": 1.2947,
+      "step": 2128500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.710368489956764e-05,
+      "loss": 1.3196,
+      "step": 2129000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.71015891339382e-05,
+      "loss": 1.3049,
+      "step": 2129500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.7099493368308756e-05,
+      "loss": 1.3145,
+      "step": 2130000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.7097393402748196e-05,
+      "loss": 1.2462,
+      "step": 2130500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.709529343718763e-05,
+      "loss": 1.3073,
+      "step": 2131000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.709319347162706e-05,
+      "loss": 1.3123,
+      "step": 2131500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.7091093506066504e-05,
+      "loss": 1.3297,
+      "step": 2132000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.708899354050594e-05,
+      "loss": 1.2877,
+      "step": 2132500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.708689357494537e-05,
+      "loss": 1.3054,
+      "step": 2133000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.7084793609384804e-05,
+      "loss": 1.3155,
+      "step": 2133500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.7082697843755364e-05,
+      "loss": 1.2698,
+      "step": 2134000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.70805978781948e-05,
+      "loss": 1.2962,
+      "step": 2134500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.707850631249647e-05,
+      "loss": 1.3119,
+      "step": 2135000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.7076406346935905e-05,
+      "loss": 1.3021,
+      "step": 2135500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.7074306381375345e-05,
+      "loss": 1.2672,
+      "step": 2136000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.707220641581478e-05,
+      "loss": 1.3078,
+      "step": 2136500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.707010645025421e-05,
+      "loss": 1.2866,
+      "step": 2137000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.706800648469365e-05,
+      "loss": 1.3061,
+      "step": 2137500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.7065906519133086e-05,
+      "loss": 1.3172,
+      "step": 2138000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.706380655357252e-05,
+      "loss": 1.3106,
+      "step": 2138500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.706170658801196e-05,
+      "loss": 1.2807,
+      "step": 2139000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.705960662245139e-05,
+      "loss": 1.3079,
+      "step": 2139500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.7057506656890826e-05,
+      "loss": 1.2682,
+      "step": 2140000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.705540669133026e-05,
+      "loss": 1.2924,
+      "step": 2140500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.705331092570082e-05,
+      "loss": 1.3214,
+      "step": 2141000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.7051210960140254e-05,
+      "loss": 1.2939,
+      "step": 2141500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.7049110994579694e-05,
+      "loss": 1.2926,
+      "step": 2142000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.704701942888136e-05,
+      "loss": 1.2906,
+      "step": 2142500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.70449194633208e-05,
+      "loss": 1.3183,
+      "step": 2143000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.7042819497760234e-05,
+      "loss": 1.2804,
+      "step": 2143500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.704071953219967e-05,
+      "loss": 1.3177,
+      "step": 2144000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.703861956663911e-05,
+      "loss": 1.2715,
+      "step": 2144500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.703651960107854e-05,
+      "loss": 1.2879,
+      "step": 2145000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.7034423835449095e-05,
+      "loss": 1.3203,
+      "step": 2145500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.703232386988853e-05,
+      "loss": 1.2726,
+      "step": 2146000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.703022390432797e-05,
+      "loss": 1.3169,
+      "step": 2146500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.70281239387674e-05,
+      "loss": 1.3098,
+      "step": 2147000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.702602397320684e-05,
+      "loss": 1.2992,
+      "step": 2147500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.7023924007646276e-05,
+      "loss": 1.2911,
+      "step": 2148000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.702182404208571e-05,
+      "loss": 1.2933,
+      "step": 2148500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.701972407652515e-05,
+      "loss": 1.2892,
+      "step": 2149000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.701762411096458e-05,
+      "loss": 1.3169,
+      "step": 2149500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.701552834533514e-05,
+      "loss": 1.2979,
+      "step": 2150000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.701342837977457e-05,
+      "loss": 1.3174,
+      "step": 2150500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.701132841421401e-05,
+      "loss": 1.295,
+      "step": 2151000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.7009228448653444e-05,
+      "loss": 1.2988,
+      "step": 2151500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.700712848309288e-05,
+      "loss": 1.2717,
+      "step": 2152000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.700502851753231e-05,
+      "loss": 1.3021,
+      "step": 2152500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.7002928551971744e-05,
+      "loss": 1.2963,
+      "step": 2153000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.7000832786342305e-05,
+      "loss": 1.3104,
+      "step": 2153500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.699873282078174e-05,
+      "loss": 1.2788,
+      "step": 2154000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.699663285522118e-05,
+      "loss": 1.3219,
+      "step": 2154500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.6994532889660605e-05,
+      "loss": 1.3133,
+      "step": 2155000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.6992432924100045e-05,
+      "loss": 1.3099,
+      "step": 2155500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.699033295853948e-05,
+      "loss": 1.31,
+      "step": 2156000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.698823299297891e-05,
+      "loss": 1.3022,
+      "step": 2156500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.6986137227349466e-05,
+      "loss": 1.2902,
+      "step": 2157000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.6984037261788906e-05,
+      "loss": 1.3182,
+      "step": 2157500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.698193729622834e-05,
+      "loss": 1.2903,
+      "step": 2158000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.697983733066777e-05,
+      "loss": 1.3087,
+      "step": 2158500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.697773736510721e-05,
+      "loss": 1.3112,
+      "step": 2159000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.6975637399546647e-05,
+      "loss": 1.3376,
+      "step": 2159500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.69735416339172e-05,
+      "loss": 1.2639,
+      "step": 2160000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.6971441668356634e-05,
+      "loss": 1.3126,
+      "step": 2160500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.6969341702796074e-05,
+      "loss": 1.3024,
+      "step": 2161000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.696724173723551e-05,
+      "loss": 1.2967,
+      "step": 2161500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.696514597160606e-05,
+      "loss": 1.2999,
+      "step": 2162000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.696305020597662e-05,
+      "loss": 1.2965,
+      "step": 2162500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.6960954440347175e-05,
+      "loss": 1.2943,
+      "step": 2163000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.6958854474786615e-05,
+      "loss": 1.2958,
+      "step": 2163500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.695675450922605e-05,
+      "loss": 1.2983,
+      "step": 2164000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.695465454366548e-05,
+      "loss": 1.2821,
+      "step": 2164500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.695255457810492e-05,
+      "loss": 1.2944,
+      "step": 2165000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.6950454612544356e-05,
+      "loss": 1.2745,
+      "step": 2165500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.694835464698379e-05,
+      "loss": 1.324,
+      "step": 2166000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.694625468142322e-05,
+      "loss": 1.3051,
+      "step": 2166500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.6944154715862656e-05,
+      "loss": 1.3191,
+      "step": 2167000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.694205475030209e-05,
+      "loss": 1.3141,
+      "step": 2167500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.693995478474153e-05,
+      "loss": 1.3012,
+      "step": 2168000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.693785481918096e-05,
+      "loss": 1.3137,
+      "step": 2168500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.69357548536204e-05,
+      "loss": 1.2997,
+      "step": 2169000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.693365488805984e-05,
+      "loss": 1.2951,
+      "step": 2169500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.693155912243039e-05,
+      "loss": 1.2867,
+      "step": 2170000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.6929459156869824e-05,
+      "loss": 1.2777,
+      "step": 2170500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.6927359191309264e-05,
+      "loss": 1.3149,
+      "step": 2171000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.69252592257487e-05,
+      "loss": 1.2897,
+      "step": 2171500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.692316346011925e-05,
+      "loss": 1.3101,
+      "step": 2172000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.6921063494558685e-05,
+      "loss": 1.3044,
+      "step": 2172500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.6918963528998125e-05,
+      "loss": 1.3205,
+      "step": 2173000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.691686356343756e-05,
+      "loss": 1.3072,
+      "step": 2173500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.691476359787699e-05,
+      "loss": 1.3045,
+      "step": 2174000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.6912667832247545e-05,
+      "loss": 1.2946,
+      "step": 2174500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.6910567866686986e-05,
+      "loss": 1.3025,
+      "step": 2175000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.6908472101057546e-05,
+      "loss": 1.2733,
+      "step": 2175500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.690637213549698e-05,
+      "loss": 1.3078,
+      "step": 2176000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.690427216993641e-05,
+      "loss": 1.27,
+      "step": 2176500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.6902172204375846e-05,
+      "loss": 1.2954,
+      "step": 2177000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.690007223881528e-05,
+      "loss": 1.3015,
+      "step": 2177500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.689797227325472e-05,
+      "loss": 1.3106,
+      "step": 2178000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.6895872307694153e-05,
+      "loss": 1.3247,
+      "step": 2178500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.689377234213359e-05,
+      "loss": 1.3034,
+      "step": 2179000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.689168077643527e-05,
+      "loss": 1.2953,
+      "step": 2179500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.68895808108747e-05,
+      "loss": 1.3089,
+      "step": 2180000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.688748084531414e-05,
+      "loss": 1.2996,
+      "step": 2180500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.688538087975357e-05,
+      "loss": 1.3184,
+      "step": 2181000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.6883280914193e-05,
+      "loss": 1.3064,
+      "step": 2181500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.688118094863244e-05,
+      "loss": 1.2797,
+      "step": 2182000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.6879080983071875e-05,
+      "loss": 1.2958,
+      "step": 2182500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.687698101751131e-05,
+      "loss": 1.2903,
+      "step": 2183000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.687488105195075e-05,
+      "loss": 1.2999,
+      "step": 2183500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.687278108639018e-05,
+      "loss": 1.2913,
+      "step": 2184000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.6870681120829616e-05,
+      "loss": 1.2997,
+      "step": 2184500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.6868581155269056e-05,
+      "loss": 1.2939,
+      "step": 2185000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.686648118970849e-05,
+      "loss": 1.3071,
+      "step": 2185500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.686438962401016e-05,
+      "loss": 1.3262,
+      "step": 2186000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.6862289658449596e-05,
+      "loss": 1.3023,
+      "step": 2186500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.686018969288904e-05,
+      "loss": 1.2758,
+      "step": 2187000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.685808972732847e-05,
+      "loss": 1.3133,
+      "step": 2187500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.6855989761767904e-05,
+      "loss": 1.2848,
+      "step": 2188000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.6853889796207344e-05,
+      "loss": 1.3079,
+      "step": 2188500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.685178983064678e-05,
+      "loss": 1.2885,
+      "step": 2189000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.684968986508621e-05,
+      "loss": 1.3003,
+      "step": 2189500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.684758989952565e-05,
+      "loss": 1.2812,
+      "step": 2190000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.6845489933965084e-05,
+      "loss": 1.3098,
+      "step": 2190500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.684338996840452e-05,
+      "loss": 1.2912,
+      "step": 2191000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.684129000284395e-05,
+      "loss": 1.3051,
+      "step": 2191500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.683919423721451e-05,
+      "loss": 1.2818,
+      "step": 2192000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.6837094271653945e-05,
+      "loss": 1.2831,
+      "step": 2192500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.683499430609338e-05,
+      "loss": 1.3283,
+      "step": 2193000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.683289434053281e-05,
+      "loss": 1.2793,
+      "step": 2193500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.6830794374972245e-05,
+      "loss": 1.2845,
+      "step": 2194000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.6828694409411686e-05,
+      "loss": 1.2864,
+      "step": 2194500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.682659444385112e-05,
+      "loss": 1.3281,
+      "step": 2195000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.682449447829055e-05,
+      "loss": 1.2927,
+      "step": 2195500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.6822398712661106e-05,
+      "loss": 1.2917,
+      "step": 2196000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.6820298747100546e-05,
+      "loss": 1.2737,
+      "step": 2196500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.681820298147111e-05,
+      "loss": 1.2818,
+      "step": 2197000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.681610301591054e-05,
+      "loss": 1.2887,
+      "step": 2197500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.6814003050349974e-05,
+      "loss": 1.3135,
+      "step": 2198000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.681190308478941e-05,
+      "loss": 1.3057,
+      "step": 2198500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.680980311922884e-05,
+      "loss": 1.3412,
+      "step": 2199000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.6807703153668274e-05,
+      "loss": 1.31,
+      "step": 2199500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.680560738803884e-05,
+      "loss": 1.2881,
+      "step": 2200000
+    },
+    {
+      "epoch": 1.32,
+      "eval_loss": 1.2527239322662354,
+      "eval_runtime": 1100.921,
+      "eval_samples_per_second": 478.436,
+      "eval_steps_per_second": 79.74,
+      "step": 2200000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.6803507422478275e-05,
+      "loss": 1.2734,
+      "step": 2200500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.68014074569177e-05,
+      "loss": 1.3167,
+      "step": 2201000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.679930749135714e-05,
+      "loss": 1.295,
+      "step": 2201500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.6797207525796575e-05,
+      "loss": 1.2774,
+      "step": 2202000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.6795111760167135e-05,
+      "loss": 1.3323,
+      "step": 2202500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.679301179460656e-05,
+      "loss": 1.2898,
+      "step": 2203000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.6790911829046e-05,
+      "loss": 1.301,
+      "step": 2203500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.6788811863485436e-05,
+      "loss": 1.3103,
+      "step": 2204000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.678671189792487e-05,
+      "loss": 1.2631,
+      "step": 2204500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.678461193236431e-05,
+      "loss": 1.2943,
+      "step": 2205000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.678251616673486e-05,
+      "loss": 1.2944,
+      "step": 2205500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.6780420401105423e-05,
+      "loss": 1.2924,
+      "step": 2206000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.677832043554486e-05,
+      "loss": 1.2832,
+      "step": 2206500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.67762204699843e-05,
+      "loss": 1.3144,
+      "step": 2207000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.677412050442373e-05,
+      "loss": 1.3082,
+      "step": 2207500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.677202053886316e-05,
+      "loss": 1.3081,
+      "step": 2208000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.67699205733026e-05,
+      "loss": 1.302,
+      "step": 2208500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.676782060774203e-05,
+      "loss": 1.314,
+      "step": 2209000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.6765720642181464e-05,
+      "loss": 1.2998,
+      "step": 2209500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.6763624876552025e-05,
+      "loss": 1.294,
+      "step": 2210000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.676152911092258e-05,
+      "loss": 1.3207,
+      "step": 2210500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.675942914536202e-05,
+      "loss": 1.2986,
+      "step": 2211000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.675732917980145e-05,
+      "loss": 1.306,
+      "step": 2211500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.6755229214240885e-05,
+      "loss": 1.2873,
+      "step": 2212000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.6753129248680326e-05,
+      "loss": 1.276,
+      "step": 2212500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.675102928311975e-05,
+      "loss": 1.3286,
+      "step": 2213000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.674892931755919e-05,
+      "loss": 1.3012,
+      "step": 2213500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.674683355192975e-05,
+      "loss": 1.2965,
+      "step": 2214000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.6744733586369186e-05,
+      "loss": 1.2722,
+      "step": 2214500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.674263362080861e-05,
+      "loss": 1.2788,
+      "step": 2215000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.674053365524805e-05,
+      "loss": 1.2766,
+      "step": 2215500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.673843368968749e-05,
+      "loss": 1.3094,
+      "step": 2216000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.673633372412692e-05,
+      "loss": 1.292,
+      "step": 2216500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.673423375856636e-05,
+      "loss": 1.2821,
+      "step": 2217000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.6732133793005794e-05,
+      "loss": 1.3001,
+      "step": 2217500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.673003382744523e-05,
+      "loss": 1.307,
+      "step": 2218000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.672793806181578e-05,
+      "loss": 1.3112,
+      "step": 2218500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.672583809625522e-05,
+      "loss": 1.3469,
+      "step": 2219000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.6723738130694655e-05,
+      "loss": 1.3156,
+      "step": 2219500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.672163816513409e-05,
+      "loss": 1.2869,
+      "step": 2220000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.671954239950465e-05,
+      "loss": 1.301,
+      "step": 2220500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.671744243394408e-05,
+      "loss": 1.2725,
+      "step": 2221000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.6715342468383515e-05,
+      "loss": 1.2962,
+      "step": 2221500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.6713246702754076e-05,
+      "loss": 1.2724,
+      "step": 2222000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.671114673719351e-05,
+      "loss": 1.3134,
+      "step": 2222500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.670904677163294e-05,
+      "loss": 1.3162,
+      "step": 2223000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.6706946806072376e-05,
+      "loss": 1.2715,
+      "step": 2223500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.6704846840511816e-05,
+      "loss": 1.2806,
+      "step": 2224000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.670274687495125e-05,
+      "loss": 1.2666,
+      "step": 2224500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.670064690939068e-05,
+      "loss": 1.3119,
+      "step": 2225000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.6698546943830124e-05,
+      "loss": 1.2865,
+      "step": 2225500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.669645117820068e-05,
+      "loss": 1.3324,
+      "step": 2226000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.669435121264011e-05,
+      "loss": 1.2923,
+      "step": 2226500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.6692251247079544e-05,
+      "loss": 1.3143,
+      "step": 2227000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.6690151281518984e-05,
+      "loss": 1.2843,
+      "step": 2227500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.668805131595842e-05,
+      "loss": 1.3038,
+      "step": 2228000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.668595135039785e-05,
+      "loss": 1.322,
+      "step": 2228500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.668385138483729e-05,
+      "loss": 1.286,
+      "step": 2229000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.6681751419276725e-05,
+      "loss": 1.3026,
+      "step": 2229500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.667965565364728e-05,
+      "loss": 1.3323,
+      "step": 2230000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.667755568808672e-05,
+      "loss": 1.3133,
+      "step": 2230500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.667545572252615e-05,
+      "loss": 1.3191,
+      "step": 2231000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.667336415682783e-05,
+      "loss": 1.2958,
+      "step": 2231500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.667126419126726e-05,
+      "loss": 1.3135,
+      "step": 2232000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.666916422570669e-05,
+      "loss": 1.2853,
+      "step": 2232500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.666706426014613e-05,
+      "loss": 1.2745,
+      "step": 2233000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.6664964294585566e-05,
+      "loss": 1.2897,
+      "step": 2233500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.6662864329025e-05,
+      "loss": 1.296,
+      "step": 2234000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.666076436346444e-05,
+      "loss": 1.3027,
+      "step": 2234500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.6658664397903874e-05,
+      "loss": 1.319,
+      "step": 2235000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.665656443234331e-05,
+      "loss": 1.2971,
+      "step": 2235500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.665446446678275e-05,
+      "loss": 1.2938,
+      "step": 2236000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.665236450122218e-05,
+      "loss": 1.3074,
+      "step": 2236500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.6650268735592734e-05,
+      "loss": 1.2965,
+      "step": 2237000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.6648168770032175e-05,
+      "loss": 1.2943,
+      "step": 2237500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.664606880447161e-05,
+      "loss": 1.2951,
+      "step": 2238000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.664396883891104e-05,
+      "loss": 1.2736,
+      "step": 2238500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.6641873073281595e-05,
+      "loss": 1.3118,
+      "step": 2239000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.6639773107721035e-05,
+      "loss": 1.299,
+      "step": 2239500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.663767734209159e-05,
+      "loss": 1.3115,
+      "step": 2240000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.663557737653102e-05,
+      "loss": 1.2997,
+      "step": 2240500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.6633477410970456e-05,
+      "loss": 1.3099,
+      "step": 2241000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.6631377445409896e-05,
+      "loss": 1.2724,
+      "step": 2241500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.662927747984933e-05,
+      "loss": 1.3063,
+      "step": 2242000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.662717751428876e-05,
+      "loss": 1.2978,
+      "step": 2242500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.66250775487282e-05,
+      "loss": 1.2564,
+      "step": 2243000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.662297758316764e-05,
+      "loss": 1.3206,
+      "step": 2243500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.662088181753819e-05,
+      "loss": 1.2842,
+      "step": 2244000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.661878185197763e-05,
+      "loss": 1.3035,
+      "step": 2244500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.6616681886417064e-05,
+      "loss": 1.3145,
+      "step": 2245000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.66145819208565e-05,
+      "loss": 1.3327,
+      "step": 2245500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.661248195529594e-05,
+      "loss": 1.2972,
+      "step": 2246000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.661038198973537e-05,
+      "loss": 1.3036,
+      "step": 2246500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.66082820241748e-05,
+      "loss": 1.2864,
+      "step": 2247000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.660618205861424e-05,
+      "loss": 1.3089,
+      "step": 2247500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.66040862929848e-05,
+      "loss": 1.2754,
+      "step": 2248000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.660198632742423e-05,
+      "loss": 1.3042,
+      "step": 2248500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.6599886361863665e-05,
+      "loss": 1.292,
+      "step": 2249000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.65977863963031e-05,
+      "loss": 1.2758,
+      "step": 2249500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.659568643074253e-05,
+      "loss": 1.3118,
+      "step": 2250000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.6593586465181966e-05,
+      "loss": 1.3109,
+      "step": 2250500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.6591486499621406e-05,
+      "loss": 1.3248,
+      "step": 2251000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.658939073399196e-05,
+      "loss": 1.2969,
+      "step": 2251500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.658729076843139e-05,
+      "loss": 1.2684,
+      "step": 2252000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.658519500280195e-05,
+      "loss": 1.3061,
+      "step": 2252500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.6583095037241393e-05,
+      "loss": 1.3306,
+      "step": 2253000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.658099507168083e-05,
+      "loss": 1.2987,
+      "step": 2253500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.6578895106120254e-05,
+      "loss": 1.3017,
+      "step": 2254000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.6576795140559694e-05,
+      "loss": 1.2799,
+      "step": 2254500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.657469517499913e-05,
+      "loss": 1.2884,
+      "step": 2255000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.657259520943856e-05,
+      "loss": 1.3019,
+      "step": 2255500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.657049944380912e-05,
+      "loss": 1.2894,
+      "step": 2256000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.6568399478248555e-05,
+      "loss": 1.2969,
+      "step": 2256500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.656629951268799e-05,
+      "loss": 1.2884,
+      "step": 2257000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.656419954712742e-05,
+      "loss": 1.2966,
+      "step": 2257500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.656209958156686e-05,
+      "loss": 1.2782,
+      "step": 2258000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.6559999616006295e-05,
+      "loss": 1.3302,
+      "step": 2258500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.6557899650445735e-05,
+      "loss": 1.2937,
+      "step": 2259000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.655579968488517e-05,
+      "loss": 1.2745,
+      "step": 2259500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.65536997193246e-05,
+      "loss": 1.3018,
+      "step": 2260000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.6551603953695156e-05,
+      "loss": 1.3177,
+      "step": 2260500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.6549503988134596e-05,
+      "loss": 1.2754,
+      "step": 2261000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.654740402257403e-05,
+      "loss": 1.3223,
+      "step": 2261500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.654530405701346e-05,
+      "loss": 1.2784,
+      "step": 2262000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.65432040914529e-05,
+      "loss": 1.2722,
+      "step": 2262500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.654110412589234e-05,
+      "loss": 1.3009,
+      "step": 2263000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.653900836026289e-05,
+      "loss": 1.3102,
+      "step": 2263500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.6536908394702324e-05,
+      "loss": 1.3325,
+      "step": 2264000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.6534808429141764e-05,
+      "loss": 1.2946,
+      "step": 2264500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.65327084635812e-05,
+      "loss": 1.284,
+      "step": 2265000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.653060849802063e-05,
+      "loss": 1.3078,
+      "step": 2265500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.652850853246007e-05,
+      "loss": 1.2931,
+      "step": 2266000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.65264085668995e-05,
+      "loss": 1.2773,
+      "step": 2266500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.652430860133894e-05,
+      "loss": 1.2688,
+      "step": 2267000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.65222128357095e-05,
+      "loss": 1.3088,
+      "step": 2267500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.652011287014893e-05,
+      "loss": 1.2823,
+      "step": 2268000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.6518017104519486e-05,
+      "loss": 1.2822,
+      "step": 2268500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.651591713895892e-05,
+      "loss": 1.302,
+      "step": 2269000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.651381717339836e-05,
+      "loss": 1.3335,
+      "step": 2269500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.651171720783779e-05,
+      "loss": 1.2964,
+      "step": 2270000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.6509621442208346e-05,
+      "loss": 1.3176,
+      "step": 2270500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.650752147664778e-05,
+      "loss": 1.3005,
+      "step": 2271000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.650542151108722e-05,
+      "loss": 1.2932,
+      "step": 2271500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.650332154552665e-05,
+      "loss": 1.2737,
+      "step": 2272000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.650122157996609e-05,
+      "loss": 1.2979,
+      "step": 2272500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.649912161440553e-05,
+      "loss": 1.3064,
+      "step": 2273000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.649702164884496e-05,
+      "loss": 1.3267,
+      "step": 2273500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.6494921683284394e-05,
+      "loss": 1.3039,
+      "step": 2274000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.6492825917654954e-05,
+      "loss": 1.2954,
+      "step": 2274500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.649072595209439e-05,
+      "loss": 1.2922,
+      "step": 2275000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.648863018646494e-05,
+      "loss": 1.3065,
+      "step": 2275500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.6486534420835495e-05,
+      "loss": 1.3031,
+      "step": 2276000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.648443445527493e-05,
+      "loss": 1.2907,
+      "step": 2276500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.648233448971437e-05,
+      "loss": 1.2877,
+      "step": 2277000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.64802345241538e-05,
+      "loss": 1.2736,
+      "step": 2277500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.6478134558593236e-05,
+      "loss": 1.2866,
+      "step": 2278000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.6476034593032676e-05,
+      "loss": 1.2804,
+      "step": 2278500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.647393462747211e-05,
+      "loss": 1.3138,
+      "step": 2279000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.647183466191154e-05,
+      "loss": 1.2649,
+      "step": 2279500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.646973469635098e-05,
+      "loss": 1.3112,
+      "step": 2280000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.6467634730790416e-05,
+      "loss": 1.3056,
+      "step": 2280500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.646553476522985e-05,
+      "loss": 1.3003,
+      "step": 2281000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.646343479966928e-05,
+      "loss": 1.2893,
+      "step": 2281500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.6461339034039844e-05,
+      "loss": 1.2977,
+      "step": 2282000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.64592432684104e-05,
+      "loss": 1.2905,
+      "step": 2282500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.645714330284983e-05,
+      "loss": 1.3007,
+      "step": 2283000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.6455047537220384e-05,
+      "loss": 1.2901,
+      "step": 2283500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.6452947571659825e-05,
+      "loss": 1.2815,
+      "step": 2284000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.645084760609926e-05,
+      "loss": 1.3198,
+      "step": 2284500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.644874764053869e-05,
+      "loss": 1.2933,
+      "step": 2285000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.644664767497813e-05,
+      "loss": 1.2763,
+      "step": 2285500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.6444547709417565e-05,
+      "loss": 1.2977,
+      "step": 2286000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.644245194378812e-05,
+      "loss": 1.3025,
+      "step": 2286500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.644035197822756e-05,
+      "loss": 1.2982,
+      "step": 2287000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.643825201266699e-05,
+      "loss": 1.2912,
+      "step": 2287500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.6436152047106426e-05,
+      "loss": 1.2818,
+      "step": 2288000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.6434052081545866e-05,
+      "loss": 1.3461,
+      "step": 2288500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.64319521159853e-05,
+      "loss": 1.2826,
+      "step": 2289000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.642985215042473e-05,
+      "loss": 1.3084,
+      "step": 2289500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.642775218486417e-05,
+      "loss": 1.2781,
+      "step": 2290000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.64256522193036e-05,
+      "loss": 1.2973,
+      "step": 2290500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.642355645367416e-05,
+      "loss": 1.2925,
+      "step": 2291000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.6421456488113594e-05,
+      "loss": 1.2886,
+      "step": 2291500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.641936072248415e-05,
+      "loss": 1.2997,
+      "step": 2292000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.641726075692359e-05,
+      "loss": 1.2762,
+      "step": 2292500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.641516079136302e-05,
+      "loss": 1.28,
+      "step": 2293000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6413060825802454e-05,
+      "loss": 1.3279,
+      "step": 2293500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6410965060173015e-05,
+      "loss": 1.2638,
+      "step": 2294000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.640886509461245e-05,
+      "loss": 1.2722,
+      "step": 2294500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.640676512905188e-05,
+      "loss": 1.3138,
+      "step": 2295000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.640466516349132e-05,
+      "loss": 1.2903,
+      "step": 2295500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6402565197930755e-05,
+      "loss": 1.2959,
+      "step": 2296000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.640046523237019e-05,
+      "loss": 1.2991,
+      "step": 2296500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.639836526680963e-05,
+      "loss": 1.2732,
+      "step": 2297000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6396265301249056e-05,
+      "loss": 1.3255,
+      "step": 2297500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.639416533568849e-05,
+      "loss": 1.3055,
+      "step": 2298000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.639206537012793e-05,
+      "loss": 1.3088,
+      "step": 2298500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.638996540456736e-05,
+      "loss": 1.2932,
+      "step": 2299000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6387865439006796e-05,
+      "loss": 1.2821,
+      "step": 2299500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.638576547344624e-05,
+      "loss": 1.288,
+      "step": 2300000
+    },
+    {
+      "epoch": 1.38,
+      "eval_loss": 1.2455518245697021,
+      "eval_runtime": 1106.2825,
+      "eval_samples_per_second": 476.117,
+      "eval_steps_per_second": 79.353,
+      "step": 2300000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.638366970781679e-05,
+      "loss": 1.2798,
+      "step": 2300500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6381569742256224e-05,
+      "loss": 1.3179,
+      "step": 2301000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.637946977669566e-05,
+      "loss": 1.3051,
+      "step": 2301500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.63773698111351e-05,
+      "loss": 1.2896,
+      "step": 2302000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.637526984557453e-05,
+      "loss": 1.2863,
+      "step": 2302500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6373169880013964e-05,
+      "loss": 1.3223,
+      "step": 2303000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6371069914453405e-05,
+      "loss": 1.3026,
+      "step": 2303500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.636897414882396e-05,
+      "loss": 1.2974,
+      "step": 2304000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.636687418326339e-05,
+      "loss": 1.302,
+      "step": 2304500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.636477421770283e-05,
+      "loss": 1.2879,
+      "step": 2305000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6362674252142265e-05,
+      "loss": 1.2889,
+      "step": 2305500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.636057848651282e-05,
+      "loss": 1.2862,
+      "step": 2306000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.635847852095225e-05,
+      "loss": 1.2743,
+      "step": 2306500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.635637855539169e-05,
+      "loss": 1.2755,
+      "step": 2307000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6354278589831126e-05,
+      "loss": 1.2925,
+      "step": 2307500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.635217862427056e-05,
+      "loss": 1.2772,
+      "step": 2308000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.635007865871e-05,
+      "loss": 1.2943,
+      "step": 2308500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.634797869314943e-05,
+      "loss": 1.3097,
+      "step": 2309000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.634587872758887e-05,
+      "loss": 1.2934,
+      "step": 2309500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.634377876202831e-05,
+      "loss": 1.2788,
+      "step": 2310000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.634168299639886e-05,
+      "loss": 1.304,
+      "step": 2310500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.6339587230769414e-05,
+      "loss": 1.2732,
+      "step": 2311000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.633748726520885e-05,
+      "loss": 1.3261,
+      "step": 2311500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.633538729964829e-05,
+      "loss": 1.3063,
+      "step": 2312000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.633328733408772e-05,
+      "loss": 1.3185,
+      "step": 2312500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.6331187368527155e-05,
+      "loss": 1.2942,
+      "step": 2313000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.6329087402966595e-05,
+      "loss": 1.2738,
+      "step": 2313500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.632698743740603e-05,
+      "loss": 1.3169,
+      "step": 2314000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.632488747184546e-05,
+      "loss": 1.2872,
+      "step": 2314500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.6322791706216015e-05,
+      "loss": 1.3158,
+      "step": 2315000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.6320691740655456e-05,
+      "loss": 1.2949,
+      "step": 2315500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.631859177509489e-05,
+      "loss": 1.2925,
+      "step": 2316000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.631649180953432e-05,
+      "loss": 1.3328,
+      "step": 2316500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.631439184397376e-05,
+      "loss": 1.296,
+      "step": 2317000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.631229187841319e-05,
+      "loss": 1.2819,
+      "step": 2317500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.631019191285262e-05,
+      "loss": 1.276,
+      "step": 2318000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.630809194729206e-05,
+      "loss": 1.2908,
+      "step": 2318500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.6305996181662623e-05,
+      "loss": 1.3047,
+      "step": 2319000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.630389621610206e-05,
+      "loss": 1.2927,
+      "step": 2319500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.630180045047261e-05,
+      "loss": 1.2969,
+      "step": 2320000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.629970048491205e-05,
+      "loss": 1.3405,
+      "step": 2320500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.6297600519351484e-05,
+      "loss": 1.3179,
+      "step": 2321000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.629550055379092e-05,
+      "loss": 1.3056,
+      "step": 2321500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.629340058823036e-05,
+      "loss": 1.2859,
+      "step": 2322000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.6291300622669785e-05,
+      "loss": 1.2527,
+      "step": 2322500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.628920065710922e-05,
+      "loss": 1.2885,
+      "step": 2323000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.628710069154866e-05,
+      "loss": 1.2838,
+      "step": 2323500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.628500072598809e-05,
+      "loss": 1.3033,
+      "step": 2324000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.6282900760427525e-05,
+      "loss": 1.2698,
+      "step": 2324500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.6280800794866965e-05,
+      "loss": 1.3107,
+      "step": 2325000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.62787008293064e-05,
+      "loss": 1.2818,
+      "step": 2325500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.627660506367695e-05,
+      "loss": 1.2776,
+      "step": 2326000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.627450509811639e-05,
+      "loss": 1.3122,
+      "step": 2326500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.6272405132555826e-05,
+      "loss": 1.2634,
+      "step": 2327000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.627030516699526e-05,
+      "loss": 1.2735,
+      "step": 2327500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.626820940136581e-05,
+      "loss": 1.3148,
+      "step": 2328000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.6266109435805253e-05,
+      "loss": 1.3328,
+      "step": 2328500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.626400947024469e-05,
+      "loss": 1.2974,
+      "step": 2329000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.626190950468412e-05,
+      "loss": 1.2861,
+      "step": 2329500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.625980953912356e-05,
+      "loss": 1.3123,
+      "step": 2330000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.6257709573562994e-05,
+      "loss": 1.2748,
+      "step": 2330500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.625560960800243e-05,
+      "loss": 1.2806,
+      "step": 2331000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.625350964244187e-05,
+      "loss": 1.3202,
+      "step": 2331500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.625141387681242e-05,
+      "loss": 1.2773,
+      "step": 2332000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.6249313911251855e-05,
+      "loss": 1.3239,
+      "step": 2332500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.624721394569129e-05,
+      "loss": 1.29,
+      "step": 2333000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.624511818006185e-05,
+      "loss": 1.3223,
+      "step": 2333500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.624301821450128e-05,
+      "loss": 1.3051,
+      "step": 2334000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.6240918248940715e-05,
+      "loss": 1.306,
+      "step": 2334500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.6238818283380156e-05,
+      "loss": 1.3051,
+      "step": 2335000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.623671831781959e-05,
+      "loss": 1.2673,
+      "step": 2335500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.623461835225902e-05,
+      "loss": 1.3023,
+      "step": 2336000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.623251838669846e-05,
+      "loss": 1.2617,
+      "step": 2336500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.6230418421137896e-05,
+      "loss": 1.285,
+      "step": 2337000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.622831845557732e-05,
+      "loss": 1.2997,
+      "step": 2337500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.622621849001676e-05,
+      "loss": 1.2867,
+      "step": 2338000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.6224122724387324e-05,
+      "loss": 1.3164,
+      "step": 2338500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.622202695875788e-05,
+      "loss": 1.2961,
+      "step": 2339000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.621992699319731e-05,
+      "loss": 1.323,
+      "step": 2339500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.6217827027636744e-05,
+      "loss": 1.3008,
+      "step": 2340000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.6215731262007304e-05,
+      "loss": 1.3047,
+      "step": 2340500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.621363129644674e-05,
+      "loss": 1.2989,
+      "step": 2341000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.621153133088617e-05,
+      "loss": 1.3063,
+      "step": 2341500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.620943136532561e-05,
+      "loss": 1.3167,
+      "step": 2342000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.6207331399765045e-05,
+      "loss": 1.3272,
+      "step": 2342500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.620523143420448e-05,
+      "loss": 1.3168,
+      "step": 2343000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.620313566857503e-05,
+      "loss": 1.308,
+      "step": 2343500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.620103570301447e-05,
+      "loss": 1.3284,
+      "step": 2344000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.6198935737453906e-05,
+      "loss": 1.3037,
+      "step": 2344500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.619683997182446e-05,
+      "loss": 1.2934,
+      "step": 2345000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.619474000626389e-05,
+      "loss": 1.2741,
+      "step": 2345500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.619264004070333e-05,
+      "loss": 1.2789,
+      "step": 2346000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.6190540075142767e-05,
+      "loss": 1.3141,
+      "step": 2346500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.61884401095822e-05,
+      "loss": 1.3117,
+      "step": 2347000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.618634014402164e-05,
+      "loss": 1.2914,
+      "step": 2347500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.6184240178461074e-05,
+      "loss": 1.2901,
+      "step": 2348000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.618214021290051e-05,
+      "loss": 1.2943,
+      "step": 2348500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.618004024733995e-05,
+      "loss": 1.2785,
+      "step": 2349000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.6177940281779374e-05,
+      "loss": 1.2823,
+      "step": 2349500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.6175840316218814e-05,
+      "loss": 1.2861,
+      "step": 2350000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.617374035065825e-05,
+      "loss": 1.2859,
+      "step": 2350500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.617164038509768e-05,
+      "loss": 1.3077,
+      "step": 2351000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.616954041953712e-05,
+      "loss": 1.2524,
+      "step": 2351500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.6167440453976555e-05,
+      "loss": 1.2938,
+      "step": 2352000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.616534048841599e-05,
+      "loss": 1.3115,
+      "step": 2352500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.616324052285543e-05,
+      "loss": 1.3106,
+      "step": 2353000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.616114475722598e-05,
+      "loss": 1.3199,
+      "step": 2353500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.6159044791665416e-05,
+      "loss": 1.2777,
+      "step": 2354000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.615694482610485e-05,
+      "loss": 1.2608,
+      "step": 2354500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.615484486054429e-05,
+      "loss": 1.3005,
+      "step": 2355000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.615274909491484e-05,
+      "loss": 1.2807,
+      "step": 2355500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.6150649129354276e-05,
+      "loss": 1.2883,
+      "step": 2356000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.614854916379371e-05,
+      "loss": 1.2886,
+      "step": 2356500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.614644919823315e-05,
+      "loss": 1.3124,
+      "step": 2357000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.6144353432603704e-05,
+      "loss": 1.2828,
+      "step": 2357500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.614225346704314e-05,
+      "loss": 1.253,
+      "step": 2358000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.614015350148258e-05,
+      "loss": 1.2813,
+      "step": 2358500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.613805353592201e-05,
+      "loss": 1.2631,
+      "step": 2359000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.6135953570361444e-05,
+      "loss": 1.2863,
+      "step": 2359500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.6133857804732e-05,
+      "loss": 1.3079,
+      "step": 2360000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.613175783917144e-05,
+      "loss": 1.294,
+      "step": 2360500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.612965787361087e-05,
+      "loss": 1.2917,
+      "step": 2361000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.6127557908050305e-05,
+      "loss": 1.289,
+      "step": 2361500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.6125457942489745e-05,
+      "loss": 1.2959,
+      "step": 2362000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.612335797692918e-05,
+      "loss": 1.3066,
+      "step": 2362500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.612125801136861e-05,
+      "loss": 1.3018,
+      "step": 2363000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.611915804580805e-05,
+      "loss": 1.2903,
+      "step": 2363500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.6117058080247486e-05,
+      "loss": 1.2832,
+      "step": 2364000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.611496231461804e-05,
+      "loss": 1.2831,
+      "step": 2364500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.611286234905748e-05,
+      "loss": 1.314,
+      "step": 2365000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.611076238349691e-05,
+      "loss": 1.2875,
+      "step": 2365500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.6108662417936347e-05,
+      "loss": 1.3031,
+      "step": 2366000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.610656245237578e-05,
+      "loss": 1.2956,
+      "step": 2366500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.6104462486815213e-05,
+      "loss": 1.3006,
+      "step": 2367000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.610236252125465e-05,
+      "loss": 1.2941,
+      "step": 2367500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.610026675562521e-05,
+      "loss": 1.3108,
+      "step": 2368000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.609816679006465e-05,
+      "loss": 1.2935,
+      "step": 2368500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.6096066824504074e-05,
+      "loss": 1.2873,
+      "step": 2369000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.609396685894351e-05,
+      "loss": 1.2912,
+      "step": 2369500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.609186689338295e-05,
+      "loss": 1.2927,
+      "step": 2370000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.608976692782238e-05,
+      "loss": 1.2987,
+      "step": 2370500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.608767116219294e-05,
+      "loss": 1.2951,
+      "step": 2371000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.6085571196632375e-05,
+      "loss": 1.301,
+      "step": 2371500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.608347123107181e-05,
+      "loss": 1.2768,
+      "step": 2372000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.608137126551124e-05,
+      "loss": 1.31,
+      "step": 2372500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.60792754998818e-05,
+      "loss": 1.3129,
+      "step": 2373000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.607717553432124e-05,
+      "loss": 1.3254,
+      "step": 2373500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.6075079768691796e-05,
+      "loss": 1.3078,
+      "step": 2374000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.607297980313123e-05,
+      "loss": 1.2997,
+      "step": 2374500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.607087983757066e-05,
+      "loss": 1.3192,
+      "step": 2375000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.60687798720101e-05,
+      "loss": 1.3055,
+      "step": 2375500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.606667990644953e-05,
+      "loss": 1.2904,
+      "step": 2376000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.6064579940888964e-05,
+      "loss": 1.2697,
+      "step": 2376500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.6062479975328404e-05,
+      "loss": 1.2983,
+      "step": 2377000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.6060384209698964e-05,
+      "loss": 1.2853,
+      "step": 2377500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.60582842441384e-05,
+      "loss": 1.2846,
+      "step": 2378000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.605618427857783e-05,
+      "loss": 1.2802,
+      "step": 2378500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.6054084313017264e-05,
+      "loss": 1.2968,
+      "step": 2379000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.60519843474567e-05,
+      "loss": 1.2991,
+      "step": 2379500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.604988438189614e-05,
+      "loss": 1.2938,
+      "step": 2380000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.604778441633557e-05,
+      "loss": 1.294,
+      "step": 2380500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.6045684450775005e-05,
+      "loss": 1.3013,
+      "step": 2381000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.6043584485214445e-05,
+      "loss": 1.2893,
+      "step": 2381500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.6041488719585e-05,
+      "loss": 1.2824,
+      "step": 2382000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.603939715388667e-05,
+      "loss": 1.2879,
+      "step": 2382500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.603729718832611e-05,
+      "loss": 1.2562,
+      "step": 2383000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.6035197222765546e-05,
+      "loss": 1.2687,
+      "step": 2383500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.603309725720498e-05,
+      "loss": 1.2971,
+      "step": 2384000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.603099729164442e-05,
+      "loss": 1.3108,
+      "step": 2384500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.6028897326083853e-05,
+      "loss": 1.2912,
+      "step": 2385000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.602679736052329e-05,
+      "loss": 1.292,
+      "step": 2385500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.602469739496272e-05,
+      "loss": 1.2852,
+      "step": 2386000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.6022597429402154e-05,
+      "loss": 1.2846,
+      "step": 2386500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.6020497463841594e-05,
+      "loss": 1.2607,
+      "step": 2387000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.601839749828103e-05,
+      "loss": 1.2958,
+      "step": 2387500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.601629753272046e-05,
+      "loss": 1.2838,
+      "step": 2388000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.60141975671599e-05,
+      "loss": 1.2743,
+      "step": 2388500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.6012097601599335e-05,
+      "loss": 1.2688,
+      "step": 2389000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.600999763603877e-05,
+      "loss": 1.2871,
+      "step": 2389500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.600790187040932e-05,
+      "loss": 1.3003,
+      "step": 2390000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.600580190484876e-05,
+      "loss": 1.2955,
+      "step": 2390500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.6003701939288195e-05,
+      "loss": 1.3004,
+      "step": 2391000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.600160197372763e-05,
+      "loss": 1.2977,
+      "step": 2391500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.599950200816707e-05,
+      "loss": 1.3495,
+      "step": 2392000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.59974020426065e-05,
+      "loss": 1.2895,
+      "step": 2392500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.5995302077045936e-05,
+      "loss": 1.2887,
+      "step": 2393000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.599320631141649e-05,
+      "loss": 1.2873,
+      "step": 2393500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.599110634585593e-05,
+      "loss": 1.2978,
+      "step": 2394000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.598900638029536e-05,
+      "loss": 1.2998,
+      "step": 2394500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.59869064147348e-05,
+      "loss": 1.2726,
+      "step": 2395000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.598480644917424e-05,
+      "loss": 1.3146,
+      "step": 2395500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.5982706483613664e-05,
+      "loss": 1.3099,
+      "step": 2396000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.5980606518053104e-05,
+      "loss": 1.293,
+      "step": 2396500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.597850655249254e-05,
+      "loss": 1.2994,
+      "step": 2397000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.59764107868631e-05,
+      "loss": 1.2754,
+      "step": 2397500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.597431502123365e-05,
+      "loss": 1.3155,
+      "step": 2398000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.5972215055673085e-05,
+      "loss": 1.2789,
+      "step": 2398500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.5970115090112525e-05,
+      "loss": 1.2833,
+      "step": 2399000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.596801512455196e-05,
+      "loss": 1.295,
+      "step": 2399500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.596591515899139e-05,
+      "loss": 1.2783,
+      "step": 2400000
+    },
+    {
+      "epoch": 1.44,
+      "eval_loss": 1.2449374198913574,
+      "eval_runtime": 1111.8994,
+      "eval_samples_per_second": 473.712,
+      "eval_steps_per_second": 78.952,
+      "step": 2400000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.596381519343083e-05,
+      "loss": 1.3102,
+      "step": 2400500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.5961719427801386e-05,
+      "loss": 1.279,
+      "step": 2401000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.595961946224082e-05,
+      "loss": 1.3264,
+      "step": 2401500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.595751949668025e-05,
+      "loss": 1.2797,
+      "step": 2402000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.595541953111969e-05,
+      "loss": 1.2995,
+      "step": 2402500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.595331956555912e-05,
+      "loss": 1.3128,
+      "step": 2403000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.595121959999856e-05,
+      "loss": 1.3285,
+      "step": 2403500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.594911963443799e-05,
+      "loss": 1.2796,
+      "step": 2404000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.594701966887743e-05,
+      "loss": 1.2816,
+      "step": 2404500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.594492390324799e-05,
+      "loss": 1.281,
+      "step": 2405000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.594282393768742e-05,
+      "loss": 1.2861,
+      "step": 2405500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.5940723972126854e-05,
+      "loss": 1.3202,
+      "step": 2406000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.593862400656629e-05,
+      "loss": 1.2889,
+      "step": 2406500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.593652404100573e-05,
+      "loss": 1.3099,
+      "step": 2407000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.593442827537629e-05,
+      "loss": 1.2828,
+      "step": 2407500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.593233250974684e-05,
+      "loss": 1.3041,
+      "step": 2408000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.5930232544186275e-05,
+      "loss": 1.29,
+      "step": 2408500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.592813677855683e-05,
+      "loss": 1.2607,
+      "step": 2409000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.592603681299627e-05,
+      "loss": 1.3062,
+      "step": 2409500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.59239368474357e-05,
+      "loss": 1.3198,
+      "step": 2410000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.5921836881875136e-05,
+      "loss": 1.3009,
+      "step": 2410500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.5919736916314576e-05,
+      "loss": 1.2754,
+      "step": 2411000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.591763695075401e-05,
+      "loss": 1.2774,
+      "step": 2411500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.591553698519344e-05,
+      "loss": 1.2868,
+      "step": 2412000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.5913437019632876e-05,
+      "loss": 1.2933,
+      "step": 2412500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.591133705407231e-05,
+      "loss": 1.2822,
+      "step": 2413000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.590924128844287e-05,
+      "loss": 1.3011,
+      "step": 2413500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.5907141322882304e-05,
+      "loss": 1.2893,
+      "step": 2414000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.5905041357321744e-05,
+      "loss": 1.3064,
+      "step": 2414500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.590294139176117e-05,
+      "loss": 1.2824,
+      "step": 2415000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.5900841426200604e-05,
+      "loss": 1.2849,
+      "step": 2415500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.5898741460640044e-05,
+      "loss": 1.3107,
+      "step": 2416000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.589664149507948e-05,
+      "loss": 1.2817,
+      "step": 2416500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.589454572945004e-05,
+      "loss": 1.328,
+      "step": 2417000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.589244576388947e-05,
+      "loss": 1.3029,
+      "step": 2417500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.5890345798328905e-05,
+      "loss": 1.2561,
+      "step": 2418000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.588824583276834e-05,
+      "loss": 1.2795,
+      "step": 2418500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.588614586720778e-05,
+      "loss": 1.297,
+      "step": 2419000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.588404590164721e-05,
+      "loss": 1.3036,
+      "step": 2419500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.5881950136017766e-05,
+      "loss": 1.3244,
+      "step": 2420000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.58798501704572e-05,
+      "loss": 1.2817,
+      "step": 2420500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.587775020489664e-05,
+      "loss": 1.2823,
+      "step": 2421000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.587565023933607e-05,
+      "loss": 1.2691,
+      "step": 2421500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.5873550273775506e-05,
+      "loss": 1.3004,
+      "step": 2422000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.5871450308214947e-05,
+      "loss": 1.2823,
+      "step": 2422500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.586935034265438e-05,
+      "loss": 1.2543,
+      "step": 2423000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.5867250377093813e-05,
+      "loss": 1.2933,
+      "step": 2423500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.5865154611464374e-05,
+      "loss": 1.279,
+      "step": 2424000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.586305884583493e-05,
+      "loss": 1.2722,
+      "step": 2424500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.58609672801366e-05,
+      "loss": 1.3164,
+      "step": 2425000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.585886731457604e-05,
+      "loss": 1.2859,
+      "step": 2425500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.5856767349015475e-05,
+      "loss": 1.3089,
+      "step": 2426000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.585466738345491e-05,
+      "loss": 1.2864,
+      "step": 2426500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.585256741789435e-05,
+      "loss": 1.2942,
+      "step": 2427000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.585046745233378e-05,
+      "loss": 1.2887,
+      "step": 2427500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.5848367486773215e-05,
+      "loss": 1.3303,
+      "step": 2428000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.5846267521212656e-05,
+      "loss": 1.2841,
+      "step": 2428500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.584416755565209e-05,
+      "loss": 1.285,
+      "step": 2429000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.5842067590091516e-05,
+      "loss": 1.2766,
+      "step": 2429500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.5839967624530956e-05,
+      "loss": 1.2875,
+      "step": 2430000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.583786765897039e-05,
+      "loss": 1.2884,
+      "step": 2430500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.583576769340983e-05,
+      "loss": 1.2892,
+      "step": 2431000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.583366772784926e-05,
+      "loss": 1.2817,
+      "step": 2431500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.5831567762288697e-05,
+      "loss": 1.2835,
+      "step": 2432000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.582946779672814e-05,
+      "loss": 1.2694,
+      "step": 2432500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.582737203109869e-05,
+      "loss": 1.3126,
+      "step": 2433000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.5825272065538124e-05,
+      "loss": 1.2653,
+      "step": 2433500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.582317209997756e-05,
+      "loss": 1.2622,
+      "step": 2434000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.5821072134417e-05,
+      "loss": 1.2731,
+      "step": 2434500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.581897216885643e-05,
+      "loss": 1.2944,
+      "step": 2435000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.5816872203295864e-05,
+      "loss": 1.2903,
+      "step": 2435500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.581477643766642e-05,
+      "loss": 1.2766,
+      "step": 2436000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.581268067203697e-05,
+      "loss": 1.2762,
+      "step": 2436500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.581058070647641e-05,
+      "loss": 1.3124,
+      "step": 2437000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.5808480740915845e-05,
+      "loss": 1.2866,
+      "step": 2437500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.5806380775355286e-05,
+      "loss": 1.2678,
+      "step": 2438000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.580428080979472e-05,
+      "loss": 1.2629,
+      "step": 2438500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.580218084423415e-05,
+      "loss": 1.2938,
+      "step": 2439000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.580008087867359e-05,
+      "loss": 1.2974,
+      "step": 2439500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.5797980913113026e-05,
+      "loss": 1.271,
+      "step": 2440000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.579588094755246e-05,
+      "loss": 1.3093,
+      "step": 2440500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.579378518192301e-05,
+      "loss": 1.3068,
+      "step": 2441000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.5791685216362453e-05,
+      "loss": 1.2741,
+      "step": 2441500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.578958525080189e-05,
+      "loss": 1.283,
+      "step": 2442000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.578748948517244e-05,
+      "loss": 1.2647,
+      "step": 2442500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.5785389519611874e-05,
+      "loss": 1.2893,
+      "step": 2443000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.5783293753982434e-05,
+      "loss": 1.2831,
+      "step": 2443500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.578119378842187e-05,
+      "loss": 1.2566,
+      "step": 2444000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.57790938228613e-05,
+      "loss": 1.2799,
+      "step": 2444500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.577699385730074e-05,
+      "loss": 1.3023,
+      "step": 2445000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.5774893891740175e-05,
+      "loss": 1.3199,
+      "step": 2445500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.577279392617961e-05,
+      "loss": 1.2825,
+      "step": 2446000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.577069396061905e-05,
+      "loss": 1.2875,
+      "step": 2446500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.576859399505848e-05,
+      "loss": 1.289,
+      "step": 2447000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.5766494029497916e-05,
+      "loss": 1.3108,
+      "step": 2447500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.5764394063937356e-05,
+      "loss": 1.2929,
+      "step": 2448000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.576229409837679e-05,
+      "loss": 1.2773,
+      "step": 2448500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.5760194132816216e-05,
+      "loss": 1.2939,
+      "step": 2449000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.5758094167255656e-05,
+      "loss": 1.2857,
+      "step": 2449500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.575599420169509e-05,
+      "loss": 1.2969,
+      "step": 2450000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.575389423613452e-05,
+      "loss": 1.2902,
+      "step": 2450500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.575179427057396e-05,
+      "loss": 1.2707,
+      "step": 2451000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.574969850494452e-05,
+      "loss": 1.3025,
+      "step": 2451500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.574759853938395e-05,
+      "loss": 1.3052,
+      "step": 2452000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.574550277375451e-05,
+      "loss": 1.2797,
+      "step": 2452500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.5743402808193944e-05,
+      "loss": 1.2865,
+      "step": 2453000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.5741302842633384e-05,
+      "loss": 1.2825,
+      "step": 2453500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.573920287707281e-05,
+      "loss": 1.3003,
+      "step": 2454000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.573710711144337e-05,
+      "loss": 1.316,
+      "step": 2454500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.573500714588281e-05,
+      "loss": 1.3039,
+      "step": 2455000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.5732907180322245e-05,
+      "loss": 1.2965,
+      "step": 2455500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.573080721476168e-05,
+      "loss": 1.2978,
+      "step": 2456000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.572870724920111e-05,
+      "loss": 1.2954,
+      "step": 2456500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.5726607283640545e-05,
+      "loss": 1.2858,
+      "step": 2457000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.572450731807998e-05,
+      "loss": 1.2854,
+      "step": 2457500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.572240735251942e-05,
+      "loss": 1.2753,
+      "step": 2458000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.572030738695885e-05,
+      "loss": 1.2986,
+      "step": 2458500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.5718211621329406e-05,
+      "loss": 1.283,
+      "step": 2459000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.571611165576884e-05,
+      "loss": 1.2909,
+      "step": 2459500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.571401169020828e-05,
+      "loss": 1.2943,
+      "step": 2460000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.571191172464771e-05,
+      "loss": 1.2873,
+      "step": 2460500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.570981175908715e-05,
+      "loss": 1.2884,
+      "step": 2461000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.570771179352659e-05,
+      "loss": 1.3129,
+      "step": 2461500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.570561182796602e-05,
+      "loss": 1.2774,
+      "step": 2462000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.5703511862405454e-05,
+      "loss": 1.2387,
+      "step": 2462500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.5701416096776014e-05,
+      "loss": 1.2889,
+      "step": 2463000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.569932033114657e-05,
+      "loss": 1.2891,
+      "step": 2463500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.5697220365586e-05,
+      "loss": 1.314,
+      "step": 2464000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.5695120400025435e-05,
+      "loss": 1.2999,
+      "step": 2464500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.5693020434464875e-05,
+      "loss": 1.2791,
+      "step": 2465000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.569092046890431e-05,
+      "loss": 1.2661,
+      "step": 2465500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.568882470327486e-05,
+      "loss": 1.2753,
+      "step": 2466000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.5686724737714296e-05,
+      "loss": 1.2481,
+      "step": 2466500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.5684624772153736e-05,
+      "loss": 1.3074,
+      "step": 2467000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.568252480659317e-05,
+      "loss": 1.2951,
+      "step": 2467500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.568042904096373e-05,
+      "loss": 1.2984,
+      "step": 2468000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.567832907540316e-05,
+      "loss": 1.2969,
+      "step": 2468500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.5676229109842596e-05,
+      "loss": 1.285,
+      "step": 2469000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.567412914428203e-05,
+      "loss": 1.2838,
+      "step": 2469500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.567202917872147e-05,
+      "loss": 1.31,
+      "step": 2470000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.5669929213160904e-05,
+      "loss": 1.3301,
+      "step": 2470500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.566782924760034e-05,
+      "loss": 1.3014,
+      "step": 2471000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.566572928203978e-05,
+      "loss": 1.2799,
+      "step": 2471500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.566362931647921e-05,
+      "loss": 1.3107,
+      "step": 2472000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.5661529350918644e-05,
+      "loss": 1.3026,
+      "step": 2472500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.5659429385358084e-05,
+      "loss": 1.2913,
+      "step": 2473000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.565732941979752e-05,
+      "loss": 1.3244,
+      "step": 2473500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.565523365416807e-05,
+      "loss": 1.3016,
+      "step": 2474000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.5653133688607505e-05,
+      "loss": 1.3108,
+      "step": 2474500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.5651033723046945e-05,
+      "loss": 1.2615,
+      "step": 2475000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.564893375748638e-05,
+      "loss": 1.3088,
+      "step": 2475500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.564683799185693e-05,
+      "loss": 1.302,
+      "step": 2476000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.5644742226227486e-05,
+      "loss": 1.271,
+      "step": 2476500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.5642642260666926e-05,
+      "loss": 1.2775,
+      "step": 2477000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.564054229510636e-05,
+      "loss": 1.2906,
+      "step": 2477500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.563844232954579e-05,
+      "loss": 1.2826,
+      "step": 2478000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.563634236398523e-05,
+      "loss": 1.3045,
+      "step": 2478500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.563424659835579e-05,
+      "loss": 1.2764,
+      "step": 2479000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.563214663279522e-05,
+      "loss": 1.2918,
+      "step": 2479500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.5630046667234654e-05,
+      "loss": 1.2914,
+      "step": 2480000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.5627946701674094e-05,
+      "loss": 1.2884,
+      "step": 2480500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.562584673611353e-05,
+      "loss": 1.3056,
+      "step": 2481000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.562374677055296e-05,
+      "loss": 1.2832,
+      "step": 2481500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.56216468049924e-05,
+      "loss": 1.2507,
+      "step": 2482000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.5619546839431835e-05,
+      "loss": 1.2789,
+      "step": 2482500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.561745107380239e-05,
+      "loss": 1.2784,
+      "step": 2483000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.561535530817294e-05,
+      "loss": 1.2693,
+      "step": 2483500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.561325534261238e-05,
+      "loss": 1.3018,
+      "step": 2484000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.5611155377051815e-05,
+      "loss": 1.2829,
+      "step": 2484500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.560905541149125e-05,
+      "loss": 1.2979,
+      "step": 2485000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.560695544593069e-05,
+      "loss": 1.2843,
+      "step": 2485500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.560485548037012e-05,
+      "loss": 1.2841,
+      "step": 2486000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.5602759714740676e-05,
+      "loss": 1.2954,
+      "step": 2486500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.5600663949111236e-05,
+      "loss": 1.2777,
+      "step": 2487000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.559856398355066e-05,
+      "loss": 1.3195,
+      "step": 2487500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.5596464017990103e-05,
+      "loss": 1.2912,
+      "step": 2488000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.559436405242954e-05,
+      "loss": 1.2836,
+      "step": 2488500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.559226408686897e-05,
+      "loss": 1.2996,
+      "step": 2489000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.559016412130841e-05,
+      "loss": 1.2988,
+      "step": 2489500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.5588064155747844e-05,
+      "loss": 1.2854,
+      "step": 2490000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.5585964190187284e-05,
+      "loss": 1.2872,
+      "step": 2490500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.558386842455784e-05,
+      "loss": 1.2494,
+      "step": 2491000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.558176845899727e-05,
+      "loss": 1.3069,
+      "step": 2491500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.5579668493436705e-05,
+      "loss": 1.2806,
+      "step": 2492000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.5577568527876145e-05,
+      "loss": 1.2924,
+      "step": 2492500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.55754727622467e-05,
+      "loss": 1.2733,
+      "step": 2493000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.557337279668613e-05,
+      "loss": 1.2912,
+      "step": 2493500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.5571272831125565e-05,
+      "loss": 1.3073,
+      "step": 2494000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.5569172865565006e-05,
+      "loss": 1.2972,
+      "step": 2494500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.556707290000444e-05,
+      "loss": 1.2806,
+      "step": 2495000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.556497293444387e-05,
+      "loss": 1.2528,
+      "step": 2495500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.556287296888331e-05,
+      "loss": 1.2791,
+      "step": 2496000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.5560773003322746e-05,
+      "loss": 1.2974,
+      "step": 2496500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.555867303776218e-05,
+      "loss": 1.2761,
+      "step": 2497000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.555657727213274e-05,
+      "loss": 1.2989,
+      "step": 2497500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.5554477306572174e-05,
+      "loss": 1.288,
+      "step": 2498000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.555237734101161e-05,
+      "loss": 1.2736,
+      "step": 2498500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.555027737545105e-05,
+      "loss": 1.2777,
+      "step": 2499000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.55481816098216e-05,
+      "loss": 1.2958,
+      "step": 2499500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.5546081644261034e-05,
+      "loss": 1.2887,
+      "step": 2500000
+    },
+    {
+      "epoch": 1.5,
+      "eval_loss": 1.2355479001998901,
+      "eval_runtime": 1109.1586,
+      "eval_samples_per_second": 474.883,
+      "eval_steps_per_second": 79.147,
+      "step": 2500000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.554398167870047e-05,
+      "loss": 1.2709,
+      "step": 2500500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.554188171313991e-05,
+      "loss": 1.2705,
+      "step": 2501000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.553978174757934e-05,
+      "loss": 1.2786,
+      "step": 2501500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.5537681782018775e-05,
+      "loss": 1.2982,
+      "step": 2502000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.553558181645821e-05,
+      "loss": 1.2999,
+      "step": 2502500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.553348605082877e-05,
+      "loss": 1.2726,
+      "step": 2503000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.55313860852682e-05,
+      "loss": 1.2899,
+      "step": 2503500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.5529286119707636e-05,
+      "loss": 1.2713,
+      "step": 2504000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.5527186154147076e-05,
+      "loss": 1.2629,
+      "step": 2504500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.55250861885865e-05,
+      "loss": 1.2921,
+      "step": 2505000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.552298622302594e-05,
+      "loss": 1.2762,
+      "step": 2505500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.55208904573965e-05,
+      "loss": 1.2605,
+      "step": 2506000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.551879889169817e-05,
+      "loss": 1.3046,
+      "step": 2506500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.551669892613761e-05,
+      "loss": 1.2998,
+      "step": 2507000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.5514598960577044e-05,
+      "loss": 1.3053,
+      "step": 2507500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.551249899501648e-05,
+      "loss": 1.2953,
+      "step": 2508000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.551039902945592e-05,
+      "loss": 1.2653,
+      "step": 2508500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.550829906389535e-05,
+      "loss": 1.301,
+      "step": 2509000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.5506199098334784e-05,
+      "loss": 1.2972,
+      "step": 2509500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.5504099132774225e-05,
+      "loss": 1.2996,
+      "step": 2510000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.550199916721366e-05,
+      "loss": 1.2793,
+      "step": 2510500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.549989920165309e-05,
+      "loss": 1.3054,
+      "step": 2511000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.549779923609253e-05,
+      "loss": 1.2679,
+      "step": 2511500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.549569927053196e-05,
+      "loss": 1.3086,
+      "step": 2512000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.549360350490252e-05,
+      "loss": 1.2989,
+      "step": 2512500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.549150353934196e-05,
+      "loss": 1.3138,
+      "step": 2513000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.548940777371251e-05,
+      "loss": 1.2683,
+      "step": 2513500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.5487307808151946e-05,
+      "loss": 1.3015,
+      "step": 2514000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.548520784259138e-05,
+      "loss": 1.3071,
+      "step": 2514500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.548310787703082e-05,
+      "loss": 1.283,
+      "step": 2515000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.548100791147025e-05,
+      "loss": 1.2628,
+      "step": 2515500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.547890794590969e-05,
+      "loss": 1.2838,
+      "step": 2516000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.547680798034912e-05,
+      "loss": 1.2543,
+      "step": 2516500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.5474708014788554e-05,
+      "loss": 1.2399,
+      "step": 2517000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.547260804922799e-05,
+      "loss": 1.3154,
+      "step": 2517500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.547050808366743e-05,
+      "loss": 1.277,
+      "step": 2518000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.546841231803799e-05,
+      "loss": 1.2831,
+      "step": 2518500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.5466312352477414e-05,
+      "loss": 1.3303,
+      "step": 2519000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.5464212386916855e-05,
+      "loss": 1.2873,
+      "step": 2519500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.546211242135629e-05,
+      "loss": 1.2857,
+      "step": 2520000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.546001245579572e-05,
+      "loss": 1.283,
+      "step": 2520500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.545791249023516e-05,
+      "loss": 1.2871,
+      "step": 2521000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.5455816724605715e-05,
+      "loss": 1.2865,
+      "step": 2521500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.545371675904515e-05,
+      "loss": 1.268,
+      "step": 2522000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.545161679348458e-05,
+      "loss": 1.2916,
+      "step": 2522500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.544951682792402e-05,
+      "loss": 1.3,
+      "step": 2523000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.5447416862363456e-05,
+      "loss": 1.3141,
+      "step": 2523500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.544532109673401e-05,
+      "loss": 1.2835,
+      "step": 2524000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.544322113117344e-05,
+      "loss": 1.2896,
+      "step": 2524500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.544112116561288e-05,
+      "loss": 1.2659,
+      "step": 2525000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.543902120005232e-05,
+      "loss": 1.276,
+      "step": 2525500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.543692123449175e-05,
+      "loss": 1.2937,
+      "step": 2526000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.543482126893119e-05,
+      "loss": 1.2792,
+      "step": 2526500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.5432721303370624e-05,
+      "loss": 1.2954,
+      "step": 2527000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.543062133781006e-05,
+      "loss": 1.2844,
+      "step": 2527500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.542852557218062e-05,
+      "loss": 1.2766,
+      "step": 2528000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.542642560662005e-05,
+      "loss": 1.3086,
+      "step": 2528500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.5424325641059485e-05,
+      "loss": 1.2646,
+      "step": 2529000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.5422225675498925e-05,
+      "loss": 1.2701,
+      "step": 2529500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.542012990986948e-05,
+      "loss": 1.2832,
+      "step": 2530000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.541803414424004e-05,
+      "loss": 1.2783,
+      "step": 2530500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.5415934178679465e-05,
+      "loss": 1.2968,
+      "step": 2531000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.54138342131189e-05,
+      "loss": 1.2805,
+      "step": 2531500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.541173424755834e-05,
+      "loss": 1.2604,
+      "step": 2532000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.540963428199777e-05,
+      "loss": 1.2809,
+      "step": 2532500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.540753851636833e-05,
+      "loss": 1.2952,
+      "step": 2533000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.5405438550807766e-05,
+      "loss": 1.3006,
+      "step": 2533500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.54033385852472e-05,
+      "loss": 1.2728,
+      "step": 2534000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.540123861968663e-05,
+      "loss": 1.2973,
+      "step": 2534500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.5399138654126073e-05,
+      "loss": 1.28,
+      "step": 2535000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.539703868856551e-05,
+      "loss": 1.2993,
+      "step": 2535500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.539493872300494e-05,
+      "loss": 1.276,
+      "step": 2536000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.539283875744438e-05,
+      "loss": 1.2823,
+      "step": 2536500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.5390742991814934e-05,
+      "loss": 1.262,
+      "step": 2537000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.538864302625437e-05,
+      "loss": 1.2664,
+      "step": 2537500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.53865430606938e-05,
+      "loss": 1.3091,
+      "step": 2538000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.538444309513324e-05,
+      "loss": 1.2859,
+      "step": 2538500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.5382347329503795e-05,
+      "loss": 1.2933,
+      "step": 2539000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.538024736394323e-05,
+      "loss": 1.2675,
+      "step": 2539500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.537815159831379e-05,
+      "loss": 1.2796,
+      "step": 2540000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.537605163275322e-05,
+      "loss": 1.3124,
+      "step": 2540500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.5373951667192656e-05,
+      "loss": 1.2767,
+      "step": 2541000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.537185170163209e-05,
+      "loss": 1.2893,
+      "step": 2541500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.536975173607153e-05,
+      "loss": 1.2997,
+      "step": 2542000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.536765597044209e-05,
+      "loss": 1.2888,
+      "step": 2542500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.5365556004881516e-05,
+      "loss": 1.3313,
+      "step": 2543000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.536345603932095e-05,
+      "loss": 1.2738,
+      "step": 2543500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.536135607376039e-05,
+      "loss": 1.2988,
+      "step": 2544000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.535926030813095e-05,
+      "loss": 1.2847,
+      "step": 2544500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.5357160342570384e-05,
+      "loss": 1.2844,
+      "step": 2545000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.535506037700981e-05,
+      "loss": 1.2749,
+      "step": 2545500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.535296041144925e-05,
+      "loss": 1.289,
+      "step": 2546000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.5350860445888684e-05,
+      "loss": 1.2605,
+      "step": 2546500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.5348768880190365e-05,
+      "loss": 1.324,
+      "step": 2547000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.53466689146298e-05,
+      "loss": 1.2942,
+      "step": 2547500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.534456894906924e-05,
+      "loss": 1.2843,
+      "step": 2548000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.534246898350867e-05,
+      "loss": 1.2994,
+      "step": 2548500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.5340369017948105e-05,
+      "loss": 1.2747,
+      "step": 2549000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.5338269052387546e-05,
+      "loss": 1.332,
+      "step": 2549500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.533616908682697e-05,
+      "loss": 1.2875,
+      "step": 2550000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.5334069121266406e-05,
+      "loss": 1.2939,
+      "step": 2550500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.5331969155705846e-05,
+      "loss": 1.2803,
+      "step": 2551000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.5329873390076406e-05,
+      "loss": 1.3065,
+      "step": 2551500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.532777342451584e-05,
+      "loss": 1.3055,
+      "step": 2552000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.5325673458955266e-05,
+      "loss": 1.2939,
+      "step": 2552500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.532357349339471e-05,
+      "loss": 1.3022,
+      "step": 2553000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.532147772776527e-05,
+      "loss": 1.2936,
+      "step": 2553500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.531938196213582e-05,
+      "loss": 1.2541,
+      "step": 2554000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.5317281996575254e-05,
+      "loss": 1.2981,
+      "step": 2554500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.5315182031014694e-05,
+      "loss": 1.2905,
+      "step": 2555000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.531308206545413e-05,
+      "loss": 1.3155,
+      "step": 2555500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.531098629982468e-05,
+      "loss": 1.2903,
+      "step": 2556000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.5308886334264115e-05,
+      "loss": 1.2703,
+      "step": 2556500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.5306786368703555e-05,
+      "loss": 1.2927,
+      "step": 2557000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.530468640314299e-05,
+      "loss": 1.2701,
+      "step": 2557500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.530258643758242e-05,
+      "loss": 1.289,
+      "step": 2558000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.530048647202186e-05,
+      "loss": 1.2629,
+      "step": 2558500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.5298390706392416e-05,
+      "loss": 1.2639,
+      "step": 2559000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.529629074083185e-05,
+      "loss": 1.3199,
+      "step": 2559500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.529419077527128e-05,
+      "loss": 1.2753,
+      "step": 2560000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.529209080971072e-05,
+      "loss": 1.2861,
+      "step": 2560500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.5289990844150156e-05,
+      "loss": 1.2496,
+      "step": 2561000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.5287890878589597e-05,
+      "loss": 1.2569,
+      "step": 2561500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.528579091302902e-05,
+      "loss": 1.3157,
+      "step": 2562000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.5283695147399584e-05,
+      "loss": 1.2619,
+      "step": 2562500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.528159518183902e-05,
+      "loss": 1.3042,
+      "step": 2563000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.527949521627846e-05,
+      "loss": 1.2966,
+      "step": 2563500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.527739525071789e-05,
+      "loss": 1.3004,
+      "step": 2564000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.527529528515732e-05,
+      "loss": 1.3088,
+      "step": 2564500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.527319531959676e-05,
+      "loss": 1.3073,
+      "step": 2565000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.527109535403619e-05,
+      "loss": 1.2696,
+      "step": 2565500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.5268995388475625e-05,
+      "loss": 1.294,
+      "step": 2566000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.5266895422915065e-05,
+      "loss": 1.2843,
+      "step": 2566500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.52647954573545e-05,
+      "loss": 1.2865,
+      "step": 2567000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.526269549179393e-05,
+      "loss": 1.2804,
+      "step": 2567500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.526059972616449e-05,
+      "loss": 1.2992,
+      "step": 2568000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.5258499760603926e-05,
+      "loss": 1.3119,
+      "step": 2568500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.525639979504336e-05,
+      "loss": 1.2871,
+      "step": 2569000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.52542998294828e-05,
+      "loss": 1.2824,
+      "step": 2569500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.525219986392223e-05,
+      "loss": 1.281,
+      "step": 2570000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.5250099898361666e-05,
+      "loss": 1.2778,
+      "step": 2570500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.5247999932801106e-05,
+      "loss": 1.2915,
+      "step": 2571000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.524589996724054e-05,
+      "loss": 1.2814,
+      "step": 2571500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.524380000167997e-05,
+      "loss": 1.2775,
+      "step": 2572000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.524170003611941e-05,
+      "loss": 1.2922,
+      "step": 2572500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.523960007055884e-05,
+      "loss": 1.302,
+      "step": 2573000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.5237500104998274e-05,
+      "loss": 1.324,
+      "step": 2573500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.5235404339368834e-05,
+      "loss": 1.3116,
+      "step": 2574000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.523330437380827e-05,
+      "loss": 1.2843,
+      "step": 2574500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.52312044082477e-05,
+      "loss": 1.2948,
+      "step": 2575000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.5229104442687134e-05,
+      "loss": 1.2783,
+      "step": 2575500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.5227004477126575e-05,
+      "loss": 1.2708,
+      "step": 2576000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.522490451156601e-05,
+      "loss": 1.2467,
+      "step": 2576500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.522280454600544e-05,
+      "loss": 1.2924,
+      "step": 2577000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.5220708780376e-05,
+      "loss": 1.3003,
+      "step": 2577500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.521861301474656e-05,
+      "loss": 1.2745,
+      "step": 2578000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.5216513049185996e-05,
+      "loss": 1.2873,
+      "step": 2578500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.521441308362543e-05,
+      "loss": 1.2833,
+      "step": 2579000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.521231311806486e-05,
+      "loss": 1.2841,
+      "step": 2579500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.5210213152504296e-05,
+      "loss": 1.3092,
+      "step": 2580000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.520811318694373e-05,
+      "loss": 1.2864,
+      "step": 2580500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.520601322138317e-05,
+      "loss": 1.2591,
+      "step": 2581000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.52039132558226e-05,
+      "loss": 1.2806,
+      "step": 2581500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.520181329026204e-05,
+      "loss": 1.2873,
+      "step": 2582000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.519971752463259e-05,
+      "loss": 1.2471,
+      "step": 2582500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.519761755907203e-05,
+      "loss": 1.2981,
+      "step": 2583000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.5195517593511464e-05,
+      "loss": 1.2825,
+      "step": 2583500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.51934176279509e-05,
+      "loss": 1.2977,
+      "step": 2584000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.519131766239034e-05,
+      "loss": 1.2983,
+      "step": 2584500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.518921769682977e-05,
+      "loss": 1.2695,
+      "step": 2585000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.5187117731269205e-05,
+      "loss": 1.2862,
+      "step": 2585500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.5185017765708645e-05,
+      "loss": 1.2896,
+      "step": 2586000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.51829220000792e-05,
+      "loss": 1.3089,
+      "step": 2586500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.518082203451863e-05,
+      "loss": 1.2799,
+      "step": 2587000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.517872206895807e-05,
+      "loss": 1.2714,
+      "step": 2587500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.5176622103397506e-05,
+      "loss": 1.2882,
+      "step": 2588000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.517452213783694e-05,
+      "loss": 1.3135,
+      "step": 2588500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.517242217227638e-05,
+      "loss": 1.2768,
+      "step": 2589000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.5170322206715806e-05,
+      "loss": 1.2649,
+      "step": 2589500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.5168230641017486e-05,
+      "loss": 1.2943,
+      "step": 2590000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.516613067545692e-05,
+      "loss": 1.2447,
+      "step": 2590500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.516403070989635e-05,
+      "loss": 1.2678,
+      "step": 2591000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.5161930744335794e-05,
+      "loss": 1.2842,
+      "step": 2591500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.515983077877523e-05,
+      "loss": 1.2881,
+      "step": 2592000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.515773081321466e-05,
+      "loss": 1.3077,
+      "step": 2592500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.51556308476541e-05,
+      "loss": 1.2986,
+      "step": 2593000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.5153530882093534e-05,
+      "loss": 1.2869,
+      "step": 2593500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.515143091653297e-05,
+      "loss": 1.2916,
+      "step": 2594000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.51493309509724e-05,
+      "loss": 1.3003,
+      "step": 2594500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.5147230985411835e-05,
+      "loss": 1.2656,
+      "step": 2595000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.5145131019851275e-05,
+      "loss": 1.2334,
+      "step": 2595500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.5143035254221835e-05,
+      "loss": 1.2851,
+      "step": 2596000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.514093528866127e-05,
+      "loss": 1.2723,
+      "step": 2596500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.5138835323100695e-05,
+      "loss": 1.2863,
+      "step": 2597000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.5136735357540136e-05,
+      "loss": 1.2901,
+      "step": 2597500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.513463539197957e-05,
+      "loss": 1.3008,
+      "step": 2598000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.5132535426419e-05,
+      "loss": 1.3031,
+      "step": 2598500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.513043546085844e-05,
+      "loss": 1.2627,
+      "step": 2599000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.5128335495297876e-05,
+      "loss": 1.2682,
+      "step": 2599500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.512623972966843e-05,
+      "loss": 1.3191,
+      "step": 2600000
+    },
+    {
+      "epoch": 1.56,
+      "eval_loss": 1.2294633388519287,
+      "eval_runtime": 1103.4979,
+      "eval_samples_per_second": 477.319,
+      "eval_steps_per_second": 79.553,
+      "step": 2600000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.512413976410786e-05,
+      "loss": 1.2858,
+      "step": 2600500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.5122039798547303e-05,
+      "loss": 1.2558,
+      "step": 2601000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.511993983298674e-05,
+      "loss": 1.2744,
+      "step": 2601500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.511783986742618e-05,
+      "loss": 1.2931,
+      "step": 2602000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.511573990186561e-05,
+      "loss": 1.2964,
+      "step": 2602500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.5113639936305044e-05,
+      "loss": 1.2838,
+      "step": 2603000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.5111539970744484e-05,
+      "loss": 1.3014,
+      "step": 2603500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.510944420511504e-05,
+      "loss": 1.2888,
+      "step": 2604000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.510734843948559e-05,
+      "loss": 1.2512,
+      "step": 2604500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.5105248473925025e-05,
+      "loss": 1.283,
+      "step": 2605000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.510314850836446e-05,
+      "loss": 1.3043,
+      "step": 2605500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.51010485428039e-05,
+      "loss": 1.2737,
+      "step": 2606000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.509895277717445e-05,
+      "loss": 1.277,
+      "step": 2606500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.5096852811613886e-05,
+      "loss": 1.3011,
+      "step": 2607000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.509475284605332e-05,
+      "loss": 1.3037,
+      "step": 2607500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.5092657080423886e-05,
+      "loss": 1.2544,
+      "step": 2608000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.509055711486332e-05,
+      "loss": 1.2733,
+      "step": 2608500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.5088457149302746e-05,
+      "loss": 1.2428,
+      "step": 2609000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.5086357183742187e-05,
+      "loss": 1.2869,
+      "step": 2609500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.508425721818162e-05,
+      "loss": 1.3026,
+      "step": 2610000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.5082157252621054e-05,
+      "loss": 1.2608,
+      "step": 2610500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.5080057287060494e-05,
+      "loss": 1.3032,
+      "step": 2611000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.507795732149993e-05,
+      "loss": 1.2973,
+      "step": 2611500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.507585735593936e-05,
+      "loss": 1.3113,
+      "step": 2612000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.50737573903788e-05,
+      "loss": 1.2821,
+      "step": 2612500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.5071661624749354e-05,
+      "loss": 1.2639,
+      "step": 2613000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.506956165918879e-05,
+      "loss": 1.2903,
+      "step": 2613500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.506746169362822e-05,
+      "loss": 1.2997,
+      "step": 2614000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.506536172806766e-05,
+      "loss": 1.2777,
+      "step": 2614500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.5063265962438215e-05,
+      "loss": 1.2692,
+      "step": 2615000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.506116599687765e-05,
+      "loss": 1.2957,
+      "step": 2615500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.505906603131709e-05,
+      "loss": 1.2923,
+      "step": 2616000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.505696606575652e-05,
+      "loss": 1.2605,
+      "step": 2616500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.5054866100195956e-05,
+      "loss": 1.3037,
+      "step": 2617000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.5052766134635396e-05,
+      "loss": 1.3016,
+      "step": 2617500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.505067036900595e-05,
+      "loss": 1.2866,
+      "step": 2618000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.504857040344538e-05,
+      "loss": 1.278,
+      "step": 2618500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.5046470437884817e-05,
+      "loss": 1.2771,
+      "step": 2619000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.504437047232426e-05,
+      "loss": 1.2907,
+      "step": 2619500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.504227050676369e-05,
+      "loss": 1.2826,
+      "step": 2620000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.5040170541203124e-05,
+      "loss": 1.3166,
+      "step": 2620500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.503807477557368e-05,
+      "loss": 1.3075,
+      "step": 2621000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.503597481001312e-05,
+      "loss": 1.2728,
+      "step": 2621500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.503387484445255e-05,
+      "loss": 1.312,
+      "step": 2622000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.5031774878891984e-05,
+      "loss": 1.2831,
+      "step": 2622500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.5029674913331425e-05,
+      "loss": 1.2878,
+      "step": 2623000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.502757494777086e-05,
+      "loss": 1.3032,
+      "step": 2623500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.502547498221029e-05,
+      "loss": 1.2846,
+      "step": 2624000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.502337921658085e-05,
+      "loss": 1.3033,
+      "step": 2624500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.5021279251020285e-05,
+      "loss": 1.3147,
+      "step": 2625000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.501917928545972e-05,
+      "loss": 1.2934,
+      "step": 2625500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.501707931989915e-05,
+      "loss": 1.282,
+      "step": 2626000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.5014979354338586e-05,
+      "loss": 1.2801,
+      "step": 2626500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.5012883588709146e-05,
+      "loss": 1.3081,
+      "step": 2627000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.50107878230797e-05,
+      "loss": 1.291,
+      "step": 2627500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.500868785751913e-05,
+      "loss": 1.2816,
+      "step": 2628000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.500658789195857e-05,
+      "loss": 1.3186,
+      "step": 2628500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.500448792639801e-05,
+      "loss": 1.2771,
+      "step": 2629000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.500238796083744e-05,
+      "loss": 1.281,
+      "step": 2629500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.500028799527688e-05,
+      "loss": 1.2775,
+      "step": 2630000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.4998188029716314e-05,
+      "loss": 1.2693,
+      "step": 2630500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.499608806415575e-05,
+      "loss": 1.2985,
+      "step": 2631000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.499398809859518e-05,
+      "loss": 1.2897,
+      "step": 2631500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.4991888133034614e-05,
+      "loss": 1.2609,
+      "step": 2632000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.4989788167474055e-05,
+      "loss": 1.2735,
+      "step": 2632500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.4987692401844615e-05,
+      "loss": 1.2951,
+      "step": 2633000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.498559243628404e-05,
+      "loss": 1.2729,
+      "step": 2633500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.49834966706546e-05,
+      "loss": 1.2747,
+      "step": 2634000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.4981396705094035e-05,
+      "loss": 1.2496,
+      "step": 2634500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.497930093946459e-05,
+      "loss": 1.2857,
+      "step": 2635000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.497720097390403e-05,
+      "loss": 1.252,
+      "step": 2635500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.497510100834346e-05,
+      "loss": 1.2492,
+      "step": 2636000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.4973001042782896e-05,
+      "loss": 1.2699,
+      "step": 2636500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.4970901077222336e-05,
+      "loss": 1.3128,
+      "step": 2637000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.496880111166177e-05,
+      "loss": 1.3217,
+      "step": 2637500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.49667011461012e-05,
+      "loss": 1.2634,
+      "step": 2638000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.496460118054064e-05,
+      "loss": 1.3003,
+      "step": 2638500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.49625054149112e-05,
+      "loss": 1.2762,
+      "step": 2639000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.496040544935063e-05,
+      "loss": 1.2938,
+      "step": 2639500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.495830548379007e-05,
+      "loss": 1.2752,
+      "step": 2640000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.49562055182295e-05,
+      "loss": 1.2598,
+      "step": 2640500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.495410555266893e-05,
+      "loss": 1.2733,
+      "step": 2641000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.495200558710837e-05,
+      "loss": 1.2814,
+      "step": 2641500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.4949905621547805e-05,
+      "loss": 1.2651,
+      "step": 2642000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.494780565598724e-05,
+      "loss": 1.2976,
+      "step": 2642500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.494570569042668e-05,
+      "loss": 1.2862,
+      "step": 2643000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.494360572486611e-05,
+      "loss": 1.3144,
+      "step": 2643500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.4941509959236665e-05,
+      "loss": 1.2564,
+      "step": 2644000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.49394099936761e-05,
+      "loss": 1.2608,
+      "step": 2644500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.493731002811554e-05,
+      "loss": 1.2598,
+      "step": 2645000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.493521006255497e-05,
+      "loss": 1.2775,
+      "step": 2645500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.4933110096994406e-05,
+      "loss": 1.2591,
+      "step": 2646000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.4931014331364966e-05,
+      "loss": 1.3068,
+      "step": 2646500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.49289143658044e-05,
+      "loss": 1.3016,
+      "step": 2647000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.492681440024383e-05,
+      "loss": 1.2448,
+      "step": 2647500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.4924714434683274e-05,
+      "loss": 1.3006,
+      "step": 2648000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.492261446912271e-05,
+      "loss": 1.258,
+      "step": 2648500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.492051450356214e-05,
+      "loss": 1.2787,
+      "step": 2649000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.4918418737932694e-05,
+      "loss": 1.2741,
+      "step": 2649500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.4916318772372134e-05,
+      "loss": 1.2789,
+      "step": 2650000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.491421880681157e-05,
+      "loss": 1.2427,
+      "step": 2650500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.4912118841251e-05,
+      "loss": 1.2882,
+      "step": 2651000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.4910023075621555e-05,
+      "loss": 1.2815,
+      "step": 2651500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.4907923110060995e-05,
+      "loss": 1.262,
+      "step": 2652000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.490582314450043e-05,
+      "loss": 1.3025,
+      "step": 2652500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.490372317893986e-05,
+      "loss": 1.2743,
+      "step": 2653000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.490162741331042e-05,
+      "loss": 1.3134,
+      "step": 2653500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.4899527447749856e-05,
+      "loss": 1.2732,
+      "step": 2654000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.489742748218929e-05,
+      "loss": 1.2708,
+      "step": 2654500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.489532751662873e-05,
+      "loss": 1.2945,
+      "step": 2655000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.489322755106816e-05,
+      "loss": 1.2918,
+      "step": 2655500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.4891127585507596e-05,
+      "loss": 1.2728,
+      "step": 2656000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.488903181987815e-05,
+      "loss": 1.2737,
+      "step": 2656500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.488693185431759e-05,
+      "loss": 1.2728,
+      "step": 2657000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.4884831888757024e-05,
+      "loss": 1.2816,
+      "step": 2657500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.488273612312758e-05,
+      "loss": 1.2608,
+      "step": 2658000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.488063615756701e-05,
+      "loss": 1.3047,
+      "step": 2658500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.487853619200645e-05,
+      "loss": 1.2897,
+      "step": 2659000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.4876436226445884e-05,
+      "loss": 1.2787,
+      "step": 2659500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.487433626088532e-05,
+      "loss": 1.2716,
+      "step": 2660000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.487223629532476e-05,
+      "loss": 1.296,
+      "step": 2660500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.487013632976419e-05,
+      "loss": 1.274,
+      "step": 2661000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.486803636420363e-05,
+      "loss": 1.2686,
+      "step": 2661500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.4865936398643065e-05,
+      "loss": 1.2757,
+      "step": 2662000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.486383643308249e-05,
+      "loss": 1.2919,
+      "step": 2662500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.486173646752193e-05,
+      "loss": 1.2609,
+      "step": 2663000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.4859636501961366e-05,
+      "loss": 1.2737,
+      "step": 2663500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.4857544936263046e-05,
+      "loss": 1.2938,
+      "step": 2664000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.485544497070248e-05,
+      "loss": 1.2725,
+      "step": 2664500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.485334500514191e-05,
+      "loss": 1.3034,
+      "step": 2665000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.4851249239512466e-05,
+      "loss": 1.256,
+      "step": 2665500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.484914927395191e-05,
+      "loss": 1.2894,
+      "step": 2666000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.484704930839134e-05,
+      "loss": 1.2734,
+      "step": 2666500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.484494934283078e-05,
+      "loss": 1.2838,
+      "step": 2667000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.4842849377270214e-05,
+      "loss": 1.2808,
+      "step": 2667500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.484074941170965e-05,
+      "loss": 1.2721,
+      "step": 2668000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.483864944614909e-05,
+      "loss": 1.2806,
+      "step": 2668500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.483654948058852e-05,
+      "loss": 1.2849,
+      "step": 2669000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.4834453714959075e-05,
+      "loss": 1.2884,
+      "step": 2669500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.483235374939851e-05,
+      "loss": 1.28,
+      "step": 2670000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.483025378383795e-05,
+      "loss": 1.2608,
+      "step": 2670500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.482815381827738e-05,
+      "loss": 1.2964,
+      "step": 2671000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.4826053852716815e-05,
+      "loss": 1.2929,
+      "step": 2671500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.482395808708737e-05,
+      "loss": 1.2757,
+      "step": 2672000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.482185812152681e-05,
+      "loss": 1.2923,
+      "step": 2672500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.481975815596624e-05,
+      "loss": 1.2737,
+      "step": 2673000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.4817658190405676e-05,
+      "loss": 1.2722,
+      "step": 2673500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.4815558224845116e-05,
+      "loss": 1.2952,
+      "step": 2674000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.481345825928454e-05,
+      "loss": 1.3059,
+      "step": 2674500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.481135829372398e-05,
+      "loss": 1.2653,
+      "step": 2675000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.4809258328163417e-05,
+      "loss": 1.2836,
+      "step": 2675500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.480716256253398e-05,
+      "loss": 1.298,
+      "step": 2676000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.480506259697341e-05,
+      "loss": 1.2785,
+      "step": 2676500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.4802966831343964e-05,
+      "loss": 1.2634,
+      "step": 2677000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.4800866865783404e-05,
+      "loss": 1.2827,
+      "step": 2677500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.479877110015396e-05,
+      "loss": 1.2763,
+      "step": 2678000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.479667113459339e-05,
+      "loss": 1.2864,
+      "step": 2678500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.4794571169032825e-05,
+      "loss": 1.2704,
+      "step": 2679000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.4792471203472265e-05,
+      "loss": 1.2752,
+      "step": 2679500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.47903712379117e-05,
+      "loss": 1.2607,
+      "step": 2680000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.478827127235113e-05,
+      "loss": 1.3022,
+      "step": 2680500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.478617130679057e-05,
+      "loss": 1.2871,
+      "step": 2681000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.4784071341230006e-05,
+      "loss": 1.2727,
+      "step": 2681500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.478197137566944e-05,
+      "loss": 1.276,
+      "step": 2682000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.477987141010887e-05,
+      "loss": 1.2688,
+      "step": 2682500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.477777564447943e-05,
+      "loss": 1.3122,
+      "step": 2683000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.4775675678918866e-05,
+      "loss": 1.2712,
+      "step": 2683500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.47735757133583e-05,
+      "loss": 1.2611,
+      "step": 2684000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.477147574779773e-05,
+      "loss": 1.2871,
+      "step": 2684500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.476937578223717e-05,
+      "loss": 1.2672,
+      "step": 2685000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.476727581667661e-05,
+      "loss": 1.2632,
+      "step": 2685500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.476517585111604e-05,
+      "loss": 1.2822,
+      "step": 2686000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.4763075885555474e-05,
+      "loss": 1.2799,
+      "step": 2686500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.476098011992603e-05,
+      "loss": 1.2917,
+      "step": 2687000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.475888015436547e-05,
+      "loss": 1.2349,
+      "step": 2687500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.47567801888049e-05,
+      "loss": 1.2994,
+      "step": 2688000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.4754680223244335e-05,
+      "loss": 1.2875,
+      "step": 2688500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.4752584457614895e-05,
+      "loss": 1.2925,
+      "step": 2689000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.475048449205433e-05,
+      "loss": 1.3024,
+      "step": 2689500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.474838452649376e-05,
+      "loss": 1.2844,
+      "step": 2690000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.47462845609332e-05,
+      "loss": 1.2586,
+      "step": 2690500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.474418879530376e-05,
+      "loss": 1.2971,
+      "step": 2691000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.4742093029674316e-05,
+      "loss": 1.2734,
+      "step": 2691500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.473999306411375e-05,
+      "loss": 1.2926,
+      "step": 2692000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.473789309855318e-05,
+      "loss": 1.2958,
+      "step": 2692500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.473579313299262e-05,
+      "loss": 1.2833,
+      "step": 2693000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.473369736736318e-05,
+      "loss": 1.2844,
+      "step": 2693500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.473159740180261e-05,
+      "loss": 1.2831,
+      "step": 2694000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.4729497436242044e-05,
+      "loss": 1.2842,
+      "step": 2694500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.4727397470681484e-05,
+      "loss": 1.2921,
+      "step": 2695000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.472529750512092e-05,
+      "loss": 1.2868,
+      "step": 2695500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.472319753956035e-05,
+      "loss": 1.3167,
+      "step": 2696000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.4721097573999784e-05,
+      "loss": 1.2865,
+      "step": 2696500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.4719001808370345e-05,
+      "loss": 1.2542,
+      "step": 2697000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.47169060427409e-05,
+      "loss": 1.2872,
+      "step": 2697500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.471480607718033e-05,
+      "loss": 1.2858,
+      "step": 2698000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.471270611161977e-05,
+      "loss": 1.2537,
+      "step": 2698500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.4710606146059205e-05,
+      "loss": 1.2622,
+      "step": 2699000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.470850618049864e-05,
+      "loss": 1.2992,
+      "step": 2699500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.470640621493808e-05,
+      "loss": 1.2761,
+      "step": 2700000
+    },
+    {
+      "epoch": 1.62,
+      "eval_loss": 1.2280163764953613,
+      "eval_runtime": 1111.7976,
+      "eval_samples_per_second": 473.755,
+      "eval_steps_per_second": 78.96,
+      "step": 2700000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.470430624937751e-05,
+      "loss": 1.281,
+      "step": 2700500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.470220628381694e-05,
+      "loss": 1.2715,
+      "step": 2701000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.47001105181875e-05,
+      "loss": 1.2663,
+      "step": 2701500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.469801055262694e-05,
+      "loss": 1.2829,
+      "step": 2702000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.469591058706637e-05,
+      "loss": 1.3043,
+      "step": 2702500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.4693810621505807e-05,
+      "loss": 1.3026,
+      "step": 2703000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.469171065594524e-05,
+      "loss": 1.288,
+      "step": 2703500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.4689610690384674e-05,
+      "loss": 1.2662,
+      "step": 2704000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.4687514924755234e-05,
+      "loss": 1.2436,
+      "step": 2704500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.4685414959194674e-05,
+      "loss": 1.2721,
+      "step": 2705000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.46833149936341e-05,
+      "loss": 1.2826,
+      "step": 2705500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.4681215028073534e-05,
+      "loss": 1.2634,
+      "step": 2706000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.4679119262444095e-05,
+      "loss": 1.2816,
+      "step": 2706500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.4677019296883535e-05,
+      "loss": 1.2865,
+      "step": 2707000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.467491933132297e-05,
+      "loss": 1.2966,
+      "step": 2707500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.467282356569352e-05,
+      "loss": 1.3025,
+      "step": 2708000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.4670723600132955e-05,
+      "loss": 1.2434,
+      "step": 2708500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.4668623634572396e-05,
+      "loss": 1.2705,
+      "step": 2709000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.466652366901183e-05,
+      "loss": 1.2885,
+      "step": 2709500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.466442370345126e-05,
+      "loss": 1.2539,
+      "step": 2710000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.4662323737890696e-05,
+      "loss": 1.2856,
+      "step": 2710500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.466022377233013e-05,
+      "loss": 1.2649,
+      "step": 2711000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.465812380676957e-05,
+      "loss": 1.2792,
+      "step": 2711500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.4656023841209e-05,
+      "loss": 1.2829,
+      "step": 2712000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.4653928075579563e-05,
+      "loss": 1.2671,
+      "step": 2712500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.465182811001899e-05,
+      "loss": 1.2669,
+      "step": 2713000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.464972814445843e-05,
+      "loss": 1.2659,
+      "step": 2713500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.4647628178897864e-05,
+      "loss": 1.2569,
+      "step": 2714000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.4645532413268424e-05,
+      "loss": 1.2823,
+      "step": 2714500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.464343244770785e-05,
+      "loss": 1.2627,
+      "step": 2715000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.464133248214729e-05,
+      "loss": 1.2723,
+      "step": 2715500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.4639232516586725e-05,
+      "loss": 1.282,
+      "step": 2716000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.463713255102616e-05,
+      "loss": 1.3103,
+      "step": 2716500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.463503678539672e-05,
+      "loss": 1.2907,
+      "step": 2717000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.463293681983615e-05,
+      "loss": 1.2643,
+      "step": 2717500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.4630836854275585e-05,
+      "loss": 1.299,
+      "step": 2718000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.4628736888715026e-05,
+      "loss": 1.2953,
+      "step": 2718500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.462663692315446e-05,
+      "loss": 1.2915,
+      "step": 2719000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.462453695759389e-05,
+      "loss": 1.2624,
+      "step": 2719500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.462243699203333e-05,
+      "loss": 1.2789,
+      "step": 2720000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.4620337026472766e-05,
+      "loss": 1.283,
+      "step": 2720500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.46182370609122e-05,
+      "loss": 1.2595,
+      "step": 2721000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.461614129528275e-05,
+      "loss": 1.2692,
+      "step": 2721500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.4614041329722193e-05,
+      "loss": 1.2892,
+      "step": 2722000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.461194556409275e-05,
+      "loss": 1.2747,
+      "step": 2722500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.460984559853218e-05,
+      "loss": 1.2646,
+      "step": 2723000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.4607745632971614e-05,
+      "loss": 1.2666,
+      "step": 2723500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.4605645667411054e-05,
+      "loss": 1.2622,
+      "step": 2724000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.460354570185049e-05,
+      "loss": 1.2708,
+      "step": 2724500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.460144993622104e-05,
+      "loss": 1.2832,
+      "step": 2725000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.459934997066048e-05,
+      "loss": 1.2835,
+      "step": 2725500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.4597250005099915e-05,
+      "loss": 1.2776,
+      "step": 2726000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.459515003953935e-05,
+      "loss": 1.2853,
+      "step": 2726500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.459305007397879e-05,
+      "loss": 1.3128,
+      "step": 2727000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.459095010841822e-05,
+      "loss": 1.252,
+      "step": 2727500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.4588850142857655e-05,
+      "loss": 1.2898,
+      "step": 2728000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.458675437722821e-05,
+      "loss": 1.2854,
+      "step": 2728500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.458465441166765e-05,
+      "loss": 1.2467,
+      "step": 2729000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.458255444610708e-05,
+      "loss": 1.2758,
+      "step": 2729500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.4580454480546516e-05,
+      "loss": 1.294,
+      "step": 2730000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.4578354514985956e-05,
+      "loss": 1.2922,
+      "step": 2730500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.457625454942539e-05,
+      "loss": 1.2706,
+      "step": 2731000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.457415458386482e-05,
+      "loss": 1.2872,
+      "step": 2731500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.457205881823538e-05,
+      "loss": 1.2639,
+      "step": 2732000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.456995885267482e-05,
+      "loss": 1.2723,
+      "step": 2732500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.456785888711425e-05,
+      "loss": 1.2789,
+      "step": 2733000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.456575892155369e-05,
+      "loss": 1.2482,
+      "step": 2733500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.4563663155924244e-05,
+      "loss": 1.2766,
+      "step": 2734000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.45615673902948e-05,
+      "loss": 1.2582,
+      "step": 2734500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.455946742473423e-05,
+      "loss": 1.2885,
+      "step": 2735000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.4557367459173665e-05,
+      "loss": 1.2748,
+      "step": 2735500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.4555267493613105e-05,
+      "loss": 1.2998,
+      "step": 2736000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.455316752805254e-05,
+      "loss": 1.309,
+      "step": 2736500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.455106756249197e-05,
+      "loss": 1.2683,
+      "step": 2737000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.454896759693141e-05,
+      "loss": 1.2889,
+      "step": 2737500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.4546867631370846e-05,
+      "loss": 1.2577,
+      "step": 2738000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.454476766581028e-05,
+      "loss": 1.2825,
+      "step": 2738500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.454266770024972e-05,
+      "loss": 1.2675,
+      "step": 2739000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.454056773468915e-05,
+      "loss": 1.2951,
+      "step": 2739500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.4538467769128586e-05,
+      "loss": 1.2578,
+      "step": 2740000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.453637200349915e-05,
+      "loss": 1.2886,
+      "step": 2740500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.453427203793858e-05,
+      "loss": 1.2841,
+      "step": 2741000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.4532176272309134e-05,
+      "loss": 1.2679,
+      "step": 2741500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.453007630674857e-05,
+      "loss": 1.2747,
+      "step": 2742000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.452798054111912e-05,
+      "loss": 1.2718,
+      "step": 2742500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.452588057555856e-05,
+      "loss": 1.2925,
+      "step": 2743000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.4523780609997994e-05,
+      "loss": 1.2809,
+      "step": 2743500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.452168064443743e-05,
+      "loss": 1.2814,
+      "step": 2744000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.451958067887687e-05,
+      "loss": 1.2981,
+      "step": 2744500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.45174807133163e-05,
+      "loss": 1.2569,
+      "step": 2745000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.4515380747755735e-05,
+      "loss": 1.2387,
+      "step": 2745500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.4513280782195175e-05,
+      "loss": 1.2727,
+      "step": 2746000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.451118081663461e-05,
+      "loss": 1.3173,
+      "step": 2746500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.450908085107404e-05,
+      "loss": 1.2991,
+      "step": 2747000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.45069850854446e-05,
+      "loss": 1.2762,
+      "step": 2747500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.4504885119884036e-05,
+      "loss": 1.2756,
+      "step": 2748000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.450278515432347e-05,
+      "loss": 1.2669,
+      "step": 2748500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.450068518876291e-05,
+      "loss": 1.2771,
+      "step": 2749000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.4498585223202336e-05,
+      "loss": 1.3009,
+      "step": 2749500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.449648525764177e-05,
+      "loss": 1.283,
+      "step": 2750000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.449438529208121e-05,
+      "loss": 1.2454,
+      "step": 2750500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.449228952645177e-05,
+      "loss": 1.2566,
+      "step": 2751000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.44901895608912e-05,
+      "loss": 1.3038,
+      "step": 2751500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.448808959533063e-05,
+      "loss": 1.3018,
+      "step": 2752000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.448598962977007e-05,
+      "loss": 1.2697,
+      "step": 2752500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.448389386414063e-05,
+      "loss": 1.2759,
+      "step": 2753000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.4481793898580065e-05,
+      "loss": 1.2569,
+      "step": 2753500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.44796939330195e-05,
+      "loss": 1.2954,
+      "step": 2754000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.447759396745893e-05,
+      "loss": 1.2901,
+      "step": 2754500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.4475494001898365e-05,
+      "loss": 1.2459,
+      "step": 2755000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.4473394036337805e-05,
+      "loss": 1.2932,
+      "step": 2755500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.447129407077724e-05,
+      "loss": 1.3017,
+      "step": 2756000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.446919410521667e-05,
+      "loss": 1.2887,
+      "step": 2756500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.446710253951835e-05,
+      "loss": 1.2613,
+      "step": 2757000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.4465006773888906e-05,
+      "loss": 1.2699,
+      "step": 2757500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.446290680832834e-05,
+      "loss": 1.2701,
+      "step": 2758000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.446080684276778e-05,
+      "loss": 1.2729,
+      "step": 2758500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.4458706877207213e-05,
+      "loss": 1.2623,
+      "step": 2759000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.445660691164665e-05,
+      "loss": 1.2595,
+      "step": 2759500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.445450694608609e-05,
+      "loss": 1.2573,
+      "step": 2760000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.445240698052552e-05,
+      "loss": 1.2304,
+      "step": 2760500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.4450307014964954e-05,
+      "loss": 1.2761,
+      "step": 2761000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.444820704940439e-05,
+      "loss": 1.2885,
+      "step": 2761500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.444611128377495e-05,
+      "loss": 1.2958,
+      "step": 2762000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.444401131821438e-05,
+      "loss": 1.2791,
+      "step": 2762500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.444191135265382e-05,
+      "loss": 1.2702,
+      "step": 2763000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.443981138709325e-05,
+      "loss": 1.2463,
+      "step": 2763500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.443771142153268e-05,
+      "loss": 1.2878,
+      "step": 2764000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.443561145597212e-05,
+      "loss": 1.3098,
+      "step": 2764500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.4433511490411555e-05,
+      "loss": 1.295,
+      "step": 2765000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.443141152485099e-05,
+      "loss": 1.2844,
+      "step": 2765500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.442931575922154e-05,
+      "loss": 1.2834,
+      "step": 2766000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.442721579366098e-05,
+      "loss": 1.3025,
+      "step": 2766500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.4425115828100416e-05,
+      "loss": 1.2601,
+      "step": 2767000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.442301586253985e-05,
+      "loss": 1.2807,
+      "step": 2767500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.442092009691041e-05,
+      "loss": 1.2902,
+      "step": 2768000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.441882013134984e-05,
+      "loss": 1.2739,
+      "step": 2768500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.441672016578928e-05,
+      "loss": 1.249,
+      "step": 2769000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.441462020022872e-05,
+      "loss": 1.2955,
+      "step": 2769500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.441252443459928e-05,
+      "loss": 1.2952,
+      "step": 2770000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.4410424469038704e-05,
+      "loss": 1.2536,
+      "step": 2770500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.440832450347814e-05,
+      "loss": 1.284,
+      "step": 2771000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.440622453791758e-05,
+      "loss": 1.2813,
+      "step": 2771500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.440412457235701e-05,
+      "loss": 1.2933,
+      "step": 2772000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.4402024606796445e-05,
+      "loss": 1.2654,
+      "step": 2772500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.4399924641235885e-05,
+      "loss": 1.2739,
+      "step": 2773000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.439782887560644e-05,
+      "loss": 1.2778,
+      "step": 2773500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.439572891004587e-05,
+      "loss": 1.2719,
+      "step": 2774000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.4393628944485305e-05,
+      "loss": 1.3077,
+      "step": 2774500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.4391528978924746e-05,
+      "loss": 1.2711,
+      "step": 2775000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.438942901336418e-05,
+      "loss": 1.2713,
+      "step": 2775500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.438732904780361e-05,
+      "loss": 1.2693,
+      "step": 2776000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.438522908224305e-05,
+      "loss": 1.2937,
+      "step": 2776500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.4383129116682486e-05,
+      "loss": 1.265,
+      "step": 2777000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.438103755098417e-05,
+      "loss": 1.2554,
+      "step": 2777500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.4378937585423593e-05,
+      "loss": 1.2583,
+      "step": 2778000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.4376837619863034e-05,
+      "loss": 1.3052,
+      "step": 2778500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.437473765430247e-05,
+      "loss": 1.29,
+      "step": 2779000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.43726376887419e-05,
+      "loss": 1.2633,
+      "step": 2779500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.437054192311246e-05,
+      "loss": 1.2774,
+      "step": 2780000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.4368441957551894e-05,
+      "loss": 1.2989,
+      "step": 2780500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.436634199199133e-05,
+      "loss": 1.2444,
+      "step": 2781000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.436424202643076e-05,
+      "loss": 1.2488,
+      "step": 2781500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.436214626080132e-05,
+      "loss": 1.2693,
+      "step": 2782000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.4360046295240755e-05,
+      "loss": 1.2537,
+      "step": 2782500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.435794632968019e-05,
+      "loss": 1.2616,
+      "step": 2783000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.435584636411963e-05,
+      "loss": 1.2884,
+      "step": 2783500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.435374639855906e-05,
+      "loss": 1.2757,
+      "step": 2784000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.435165063292962e-05,
+      "loss": 1.2979,
+      "step": 2784500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.434955066736905e-05,
+      "loss": 1.2576,
+      "step": 2785000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.434745070180849e-05,
+      "loss": 1.3013,
+      "step": 2785500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.434535073624792e-05,
+      "loss": 1.2771,
+      "step": 2786000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.4343250770687356e-05,
+      "loss": 1.2668,
+      "step": 2786500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.43411508051268e-05,
+      "loss": 1.2539,
+      "step": 2787000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.433905503949735e-05,
+      "loss": 1.3054,
+      "step": 2787500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.4336955073936784e-05,
+      "loss": 1.2678,
+      "step": 2788000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.433485510837622e-05,
+      "loss": 1.2617,
+      "step": 2788500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.433275514281566e-05,
+      "loss": 1.248,
+      "step": 2789000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.433065937718622e-05,
+      "loss": 1.268,
+      "step": 2789500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.432856361155677e-05,
+      "loss": 1.2703,
+      "step": 2790000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.4326463645996205e-05,
+      "loss": 1.2807,
+      "step": 2790500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.4324363680435645e-05,
+      "loss": 1.2749,
+      "step": 2791000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.432226371487508e-05,
+      "loss": 1.262,
+      "step": 2791500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.4320163749314505e-05,
+      "loss": 1.2939,
+      "step": 2792000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.4318063783753945e-05,
+      "loss": 1.2863,
+      "step": 2792500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.431596381819338e-05,
+      "loss": 1.2631,
+      "step": 2793000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.431386385263281e-05,
+      "loss": 1.292,
+      "step": 2793500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.431176388707225e-05,
+      "loss": 1.2809,
+      "step": 2794000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.4309663921511686e-05,
+      "loss": 1.2875,
+      "step": 2794500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.430756815588224e-05,
+      "loss": 1.2951,
+      "step": 2795000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.430546819032167e-05,
+      "loss": 1.2674,
+      "step": 2795500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.430336822476111e-05,
+      "loss": 1.2705,
+      "step": 2796000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.430126825920055e-05,
+      "loss": 1.2844,
+      "step": 2796500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.42991724935711e-05,
+      "loss": 1.2606,
+      "step": 2797000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.429707252801054e-05,
+      "loss": 1.2584,
+      "step": 2797500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.4294972562449974e-05,
+      "loss": 1.2507,
+      "step": 2798000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.429287259688941e-05,
+      "loss": 1.2705,
+      "step": 2798500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.429077263132885e-05,
+      "loss": 1.2818,
+      "step": 2799000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.42886768656994e-05,
+      "loss": 1.2817,
+      "step": 2799500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.4286576900138835e-05,
+      "loss": 1.2715,
+      "step": 2800000
+    },
+    {
+      "epoch": 1.68,
+      "eval_loss": 1.2226179838180542,
+      "eval_runtime": 1103.9672,
+      "eval_samples_per_second": 477.116,
+      "eval_steps_per_second": 79.52,
+      "step": 2800000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.428447693457827e-05,
+      "loss": 1.294,
+      "step": 2800500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.428237696901771e-05,
+      "loss": 1.2937,
+      "step": 2801000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.428028120338826e-05,
+      "loss": 1.3009,
+      "step": 2801500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.4278181237827695e-05,
+      "loss": 1.2765,
+      "step": 2802000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.427608127226713e-05,
+      "loss": 1.2804,
+      "step": 2802500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.427398130670657e-05,
+      "loss": 1.2421,
+      "step": 2803000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.427188554107713e-05,
+      "loss": 1.2775,
+      "step": 2803500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.4269785575516556e-05,
+      "loss": 1.2546,
+      "step": 2804000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.4267685609955996e-05,
+      "loss": 1.2518,
+      "step": 2804500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.426558564439543e-05,
+      "loss": 1.2727,
+      "step": 2805000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.426348567883486e-05,
+      "loss": 1.2878,
+      "step": 2805500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.4261385713274304e-05,
+      "loss": 1.2705,
+      "step": 2806000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.425928994764486e-05,
+      "loss": 1.2553,
+      "step": 2806500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.425718998208429e-05,
+      "loss": 1.2811,
+      "step": 2807000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.4255090016523724e-05,
+      "loss": 1.2443,
+      "step": 2807500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.4252990050963164e-05,
+      "loss": 1.296,
+      "step": 2808000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.4250894285333725e-05,
+      "loss": 1.2692,
+      "step": 2808500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.424879851970428e-05,
+      "loss": 1.2825,
+      "step": 2809000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.424669855414371e-05,
+      "loss": 1.3024,
+      "step": 2809500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.4244598588583145e-05,
+      "loss": 1.2659,
+      "step": 2810000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.4242498623022585e-05,
+      "loss": 1.274,
+      "step": 2810500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.424039865746202e-05,
+      "loss": 1.2771,
+      "step": 2811000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.423829869190145e-05,
+      "loss": 1.2545,
+      "step": 2811500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.4236198726340886e-05,
+      "loss": 1.3149,
+      "step": 2812000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.4234102960711446e-05,
+      "loss": 1.2534,
+      "step": 2812500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.423200299515088e-05,
+      "loss": 1.2642,
+      "step": 2813000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.422990722952143e-05,
+      "loss": 1.2629,
+      "step": 2813500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.4227807263960873e-05,
+      "loss": 1.2385,
+      "step": 2814000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.422570729840031e-05,
+      "loss": 1.2798,
+      "step": 2814500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.422361153277086e-05,
+      "loss": 1.2952,
+      "step": 2815000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.4221511567210294e-05,
+      "loss": 1.261,
+      "step": 2815500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.4219411601649734e-05,
+      "loss": 1.2523,
+      "step": 2816000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.421731163608917e-05,
+      "loss": 1.2845,
+      "step": 2816500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.421521167052861e-05,
+      "loss": 1.2959,
+      "step": 2817000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.421311170496804e-05,
+      "loss": 1.2691,
+      "step": 2817500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.4211011739407475e-05,
+      "loss": 1.2564,
+      "step": 2818000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.420891177384691e-05,
+      "loss": 1.2879,
+      "step": 2818500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.420681180828634e-05,
+      "loss": 1.2757,
+      "step": 2819000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.4204711842725775e-05,
+      "loss": 1.2955,
+      "step": 2819500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.4202611877165215e-05,
+      "loss": 1.2819,
+      "step": 2820000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.420051191160465e-05,
+      "loss": 1.264,
+      "step": 2820500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.41984161459752e-05,
+      "loss": 1.2737,
+      "step": 2821000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.4196316180414636e-05,
+      "loss": 1.2553,
+      "step": 2821500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.4194216214854076e-05,
+      "loss": 1.2577,
+      "step": 2822000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.419211624929351e-05,
+      "loss": 1.2754,
+      "step": 2822500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.419001628373294e-05,
+      "loss": 1.2544,
+      "step": 2823000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.41879205181035e-05,
+      "loss": 1.279,
+      "step": 2823500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.418582055254294e-05,
+      "loss": 1.2617,
+      "step": 2824000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.418372058698237e-05,
+      "loss": 1.2698,
+      "step": 2824500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.418162062142181e-05,
+      "loss": 1.2622,
+      "step": 2825000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.4179520655861244e-05,
+      "loss": 1.2624,
+      "step": 2825500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.417742069030068e-05,
+      "loss": 1.2539,
+      "step": 2826000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.417532072474012e-05,
+      "loss": 1.2666,
+      "step": 2826500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.417322075917955e-05,
+      "loss": 1.3247,
+      "step": 2827000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.4171120793618985e-05,
+      "loss": 1.3033,
+      "step": 2827500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.416902502798954e-05,
+      "loss": 1.2853,
+      "step": 2828000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.416692506242898e-05,
+      "loss": 1.2842,
+      "step": 2828500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.416482509686841e-05,
+      "loss": 1.2583,
+      "step": 2829000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.4162725131307845e-05,
+      "loss": 1.2951,
+      "step": 2829500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.41606293656784e-05,
+      "loss": 1.2889,
+      "step": 2830000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.415852940011784e-05,
+      "loss": 1.296,
+      "step": 2830500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.415642943455727e-05,
+      "loss": 1.2616,
+      "step": 2831000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.4154329468996706e-05,
+      "loss": 1.2624,
+      "step": 2831500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.4152229503436146e-05,
+      "loss": 1.2785,
+      "step": 2832000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.415012953787558e-05,
+      "loss": 1.2784,
+      "step": 2832500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.414803377224613e-05,
+      "loss": 1.2761,
+      "step": 2833000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.4145933806685574e-05,
+      "loss": 1.2756,
+      "step": 2833500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.414383384112501e-05,
+      "loss": 1.2913,
+      "step": 2834000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.414173387556444e-05,
+      "loss": 1.2844,
+      "step": 2834500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.4139638109934994e-05,
+      "loss": 1.2809,
+      "step": 2835000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.413754234430555e-05,
+      "loss": 1.2538,
+      "step": 2835500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.413544237874499e-05,
+      "loss": 1.2484,
+      "step": 2836000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.413334241318442e-05,
+      "loss": 1.2803,
+      "step": 2836500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.4131242447623855e-05,
+      "loss": 1.2876,
+      "step": 2837000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.4129142482063295e-05,
+      "loss": 1.2538,
+      "step": 2837500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.412704251650273e-05,
+      "loss": 1.2659,
+      "step": 2838000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.412494255094216e-05,
+      "loss": 1.3031,
+      "step": 2838500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.41228425853816e-05,
+      "loss": 1.285,
+      "step": 2839000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.4120742619821036e-05,
+      "loss": 1.2525,
+      "step": 2839500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.411864265426047e-05,
+      "loss": 1.2602,
+      "step": 2840000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.41165426886999e-05,
+      "loss": 1.297,
+      "step": 2840500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.4114442723139336e-05,
+      "loss": 1.2712,
+      "step": 2841000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.4112346957509896e-05,
+      "loss": 1.2449,
+      "step": 2841500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.411025119188045e-05,
+      "loss": 1.2277,
+      "step": 2842000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.4108155426251003e-05,
+      "loss": 1.2735,
+      "step": 2842500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.4106055460690444e-05,
+      "loss": 1.3002,
+      "step": 2843000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.410395549512988e-05,
+      "loss": 1.2743,
+      "step": 2843500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.410185552956931e-05,
+      "loss": 1.2622,
+      "step": 2844000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.409975556400875e-05,
+      "loss": 1.2638,
+      "step": 2844500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.4097655598448184e-05,
+      "loss": 1.2857,
+      "step": 2845000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.409555563288762e-05,
+      "loss": 1.2835,
+      "step": 2845500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.409345566732706e-05,
+      "loss": 1.2801,
+      "step": 2846000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.409135570176649e-05,
+      "loss": 1.2927,
+      "step": 2846500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.4089255736205925e-05,
+      "loss": 1.3046,
+      "step": 2847000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.4087159970576485e-05,
+      "loss": 1.2726,
+      "step": 2847500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.408506000501592e-05,
+      "loss": 1.2844,
+      "step": 2848000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.408296003945535e-05,
+      "loss": 1.2841,
+      "step": 2848500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.408086007389479e-05,
+      "loss": 1.283,
+      "step": 2849000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.4078760108334226e-05,
+      "loss": 1.2581,
+      "step": 2849500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.407666434270478e-05,
+      "loss": 1.287,
+      "step": 2850000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.407456437714421e-05,
+      "loss": 1.276,
+      "step": 2850500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.407246441158365e-05,
+      "loss": 1.2657,
+      "step": 2851000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.407036444602309e-05,
+      "loss": 1.274,
+      "step": 2851500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.406826868039364e-05,
+      "loss": 1.2848,
+      "step": 2852000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.4066168714833074e-05,
+      "loss": 1.2651,
+      "step": 2852500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.4064068749272514e-05,
+      "loss": 1.2794,
+      "step": 2853000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.406196878371195e-05,
+      "loss": 1.3028,
+      "step": 2853500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.405986881815138e-05,
+      "loss": 1.3084,
+      "step": 2854000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.405777305252194e-05,
+      "loss": 1.2676,
+      "step": 2854500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.4055673086961375e-05,
+      "loss": 1.2972,
+      "step": 2855000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.405357312140081e-05,
+      "loss": 1.2309,
+      "step": 2855500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.405147315584025e-05,
+      "loss": 1.2657,
+      "step": 2856000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.40493773902108e-05,
+      "loss": 1.2729,
+      "step": 2856500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.4047277424650235e-05,
+      "loss": 1.2866,
+      "step": 2857000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.404517745908967e-05,
+      "loss": 1.2765,
+      "step": 2857500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.404307749352911e-05,
+      "loss": 1.2708,
+      "step": 2858000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.404097752796854e-05,
+      "loss": 1.2642,
+      "step": 2858500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.4038877562407976e-05,
+      "loss": 1.2957,
+      "step": 2859000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.403677759684741e-05,
+      "loss": 1.2841,
+      "step": 2859500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.403468183121797e-05,
+      "loss": 1.2551,
+      "step": 2860000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.40325818656574e-05,
+      "loss": 1.2733,
+      "step": 2860500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.403048190009684e-05,
+      "loss": 1.2524,
+      "step": 2861000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.402838193453628e-05,
+      "loss": 1.2945,
+      "step": 2861500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.4026281968975704e-05,
+      "loss": 1.2743,
+      "step": 2862000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.4024182003415144e-05,
+      "loss": 1.2744,
+      "step": 2862500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.402208203785458e-05,
+      "loss": 1.2906,
+      "step": 2863000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.401998627222514e-05,
+      "loss": 1.2877,
+      "step": 2863500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.401788630666457e-05,
+      "loss": 1.2748,
+      "step": 2864000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.4015786341104005e-05,
+      "loss": 1.244,
+      "step": 2864500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.401368637554344e-05,
+      "loss": 1.2454,
+      "step": 2865000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.401158640998287e-05,
+      "loss": 1.2843,
+      "step": 2865500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.400948644442231e-05,
+      "loss": 1.2632,
+      "step": 2866000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.4007386478861745e-05,
+      "loss": 1.2558,
+      "step": 2866500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.400528651330118e-05,
+      "loss": 1.2715,
+      "step": 2867000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.400319074767173e-05,
+      "loss": 1.2719,
+      "step": 2867500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.400109078211117e-05,
+      "loss": 1.299,
+      "step": 2868000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.3998990816550606e-05,
+      "loss": 1.2808,
+      "step": 2868500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.399689505092116e-05,
+      "loss": 1.2802,
+      "step": 2869000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.39947950853606e-05,
+      "loss": 1.2669,
+      "step": 2869500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.399269511980003e-05,
+      "loss": 1.2659,
+      "step": 2870000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.399059515423947e-05,
+      "loss": 1.2638,
+      "step": 2870500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.398849518867891e-05,
+      "loss": 1.2692,
+      "step": 2871000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.398639522311834e-05,
+      "loss": 1.2922,
+      "step": 2871500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.3984295257557774e-05,
+      "loss": 1.2768,
+      "step": 2872000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.3982195291997214e-05,
+      "loss": 1.2906,
+      "step": 2872500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.398009952636777e-05,
+      "loss": 1.2858,
+      "step": 2873000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.39779995608072e-05,
+      "loss": 1.2792,
+      "step": 2873500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.3975903795177755e-05,
+      "loss": 1.292,
+      "step": 2874000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.397380382961719e-05,
+      "loss": 1.2844,
+      "step": 2874500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.397170386405663e-05,
+      "loss": 1.2774,
+      "step": 2875000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.396960389849606e-05,
+      "loss": 1.3109,
+      "step": 2875500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.396750813286662e-05,
+      "loss": 1.2479,
+      "step": 2876000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.3965408167306056e-05,
+      "loss": 1.2731,
+      "step": 2876500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.396330820174549e-05,
+      "loss": 1.2692,
+      "step": 2877000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.396121243611605e-05,
+      "loss": 1.2584,
+      "step": 2877500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.395911247055548e-05,
+      "loss": 1.2503,
+      "step": 2878000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.395701250499492e-05,
+      "loss": 1.2696,
+      "step": 2878500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.395491253943435e-05,
+      "loss": 1.2599,
+      "step": 2879000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.395281257387378e-05,
+      "loss": 1.279,
+      "step": 2879500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.3950716808244344e-05,
+      "loss": 1.2657,
+      "step": 2880000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.3948616842683784e-05,
+      "loss": 1.2749,
+      "step": 2880500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.394651687712321e-05,
+      "loss": 1.2687,
+      "step": 2881000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.394442111149377e-05,
+      "loss": 1.241,
+      "step": 2881500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.3942321145933204e-05,
+      "loss": 1.2776,
+      "step": 2882000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.3940221180372645e-05,
+      "loss": 1.2631,
+      "step": 2882500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.393812121481208e-05,
+      "loss": 1.2985,
+      "step": 2883000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.393602124925151e-05,
+      "loss": 1.3127,
+      "step": 2883500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.3933921283690945e-05,
+      "loss": 1.2485,
+      "step": 2884000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.393182131813038e-05,
+      "loss": 1.2889,
+      "step": 2884500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.392972135256982e-05,
+      "loss": 1.2742,
+      "step": 2885000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.392762138700925e-05,
+      "loss": 1.2753,
+      "step": 2885500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.3925525621379806e-05,
+      "loss": 1.252,
+      "step": 2886000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.392342565581924e-05,
+      "loss": 1.2911,
+      "step": 2886500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.392132569025868e-05,
+      "loss": 1.2844,
+      "step": 2887000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.391922572469811e-05,
+      "loss": 1.2594,
+      "step": 2887500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.3917125759137546e-05,
+      "loss": 1.2723,
+      "step": 2888000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.3915025793576987e-05,
+      "loss": 1.2955,
+      "step": 2888500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.391293002794754e-05,
+      "loss": 1.2501,
+      "step": 2889000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.3910830062386974e-05,
+      "loss": 1.2553,
+      "step": 2889500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.3908730096826414e-05,
+      "loss": 1.2625,
+      "step": 2890000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.390663013126585e-05,
+      "loss": 1.3029,
+      "step": 2890500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.390453016570528e-05,
+      "loss": 1.2679,
+      "step": 2891000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.390243020014472e-05,
+      "loss": 1.2776,
+      "step": 2891500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.3900330234584154e-05,
+      "loss": 1.2888,
+      "step": 2892000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.389823026902359e-05,
+      "loss": 1.2898,
+      "step": 2892500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.389613030346303e-05,
+      "loss": 1.3113,
+      "step": 2893000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.389403033790246e-05,
+      "loss": 1.268,
+      "step": 2893500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.389193037234189e-05,
+      "loss": 1.296,
+      "step": 2894000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.388983040678133e-05,
+      "loss": 1.2824,
+      "step": 2894500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.388773464115189e-05,
+      "loss": 1.2835,
+      "step": 2895000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.388563467559132e-05,
+      "loss": 1.2843,
+      "step": 2895500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.3883538909961876e-05,
+      "loss": 1.2494,
+      "step": 2896000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.388143894440131e-05,
+      "loss": 1.2731,
+      "step": 2896500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.387933897884075e-05,
+      "loss": 1.2725,
+      "step": 2897000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.387723901328018e-05,
+      "loss": 1.2774,
+      "step": 2897500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.3875139047719616e-05,
+      "loss": 1.2902,
+      "step": 2898000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.387303908215905e-05,
+      "loss": 1.2858,
+      "step": 2898500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.387093911659848e-05,
+      "loss": 1.2664,
+      "step": 2899000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.3868839151037924e-05,
+      "loss": 1.3078,
+      "step": 2899500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.3866743385408484e-05,
+      "loss": 1.2707,
+      "step": 2900000
+    },
+    {
+      "epoch": 1.74,
+      "eval_loss": 1.2214449644088745,
+      "eval_runtime": 1124.1633,
+      "eval_samples_per_second": 468.544,
+      "eval_steps_per_second": 78.091,
+      "step": 2900000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.386464341984792e-05,
+      "loss": 1.2383,
+      "step": 2900500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.3862543454287344e-05,
+      "loss": 1.2507,
+      "step": 2901000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.3860443488726784e-05,
+      "loss": 1.2754,
+      "step": 2901500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.385834352316622e-05,
+      "loss": 1.2775,
+      "step": 2902000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.385624355760565e-05,
+      "loss": 1.286,
+      "step": 2902500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.385414359204509e-05,
+      "loss": 1.2687,
+      "step": 2903000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.3852043626484525e-05,
+      "loss": 1.2695,
+      "step": 2903500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.384994786085508e-05,
+      "loss": 1.2489,
+      "step": 2904000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.384785209522564e-05,
+      "loss": 1.285,
+      "step": 2904500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.384575212966507e-05,
+      "loss": 1.2849,
+      "step": 2905000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.3843652164104506e-05,
+      "loss": 1.2667,
+      "step": 2905500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.384155219854394e-05,
+      "loss": 1.253,
+      "step": 2906000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.383945223298338e-05,
+      "loss": 1.2526,
+      "step": 2906500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.383735226742281e-05,
+      "loss": 1.2656,
+      "step": 2907000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.3835252301862246e-05,
+      "loss": 1.2558,
+      "step": 2907500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.38331565362328e-05,
+      "loss": 1.2571,
+      "step": 2908000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.383105657067224e-05,
+      "loss": 1.2562,
+      "step": 2908500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.3828956605111674e-05,
+      "loss": 1.272,
+      "step": 2909000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.382685663955111e-05,
+      "loss": 1.241,
+      "step": 2909500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.382475667399055e-05,
+      "loss": 1.2726,
+      "step": 2910000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.38226609083611e-05,
+      "loss": 1.2858,
+      "step": 2910500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.3820560942800534e-05,
+      "loss": 1.2924,
+      "step": 2911000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.381846097723997e-05,
+      "loss": 1.2769,
+      "step": 2911500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.381636101167941e-05,
+      "loss": 1.2921,
+      "step": 2912000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.381426104611884e-05,
+      "loss": 1.2759,
+      "step": 2912500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.3812161080558275e-05,
+      "loss": 1.2482,
+      "step": 2913000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.3810061114997715e-05,
+      "loss": 1.2885,
+      "step": 2913500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.3807969549299396e-05,
+      "loss": 1.2662,
+      "step": 2914000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.380586958373883e-05,
+      "loss": 1.2619,
+      "step": 2914500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.380376961817826e-05,
+      "loss": 1.2657,
+      "step": 2915000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.3801669652617696e-05,
+      "loss": 1.2536,
+      "step": 2915500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.379956968705713e-05,
+      "loss": 1.267,
+      "step": 2916000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.379746972149656e-05,
+      "loss": 1.2333,
+      "step": 2916500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.3795369755936e-05,
+      "loss": 1.2517,
+      "step": 2917000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.379326979037544e-05,
+      "loss": 1.2552,
+      "step": 2917500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.379116982481487e-05,
+      "loss": 1.2464,
+      "step": 2918000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.378906985925431e-05,
+      "loss": 1.2496,
+      "step": 2918500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.3786969893693744e-05,
+      "loss": 1.2622,
+      "step": 2919000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.37848741280643e-05,
+      "loss": 1.2754,
+      "step": 2919500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.378277416250373e-05,
+      "loss": 1.2571,
+      "step": 2920000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.378067419694317e-05,
+      "loss": 1.2673,
+      "step": 2920500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.3778574231382605e-05,
+      "loss": 1.2946,
+      "step": 2921000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.377647426582204e-05,
+      "loss": 1.2752,
+      "step": 2921500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.37743785001926e-05,
+      "loss": 1.2904,
+      "step": 2922000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.377228273456315e-05,
+      "loss": 1.2781,
+      "step": 2922500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.3770182769002585e-05,
+      "loss": 1.2698,
+      "step": 2923000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.376808280344202e-05,
+      "loss": 1.2699,
+      "step": 2923500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.376598283788146e-05,
+      "loss": 1.2648,
+      "step": 2924000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.376388287232089e-05,
+      "loss": 1.2872,
+      "step": 2924500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.3761782906760326e-05,
+      "loss": 1.2993,
+      "step": 2925000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.3759682941199766e-05,
+      "loss": 1.2596,
+      "step": 2925500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.37575829756392e-05,
+      "loss": 1.244,
+      "step": 2926000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.375548301007863e-05,
+      "loss": 1.2727,
+      "step": 2926500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.375338724444919e-05,
+      "loss": 1.2797,
+      "step": 2927000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.375128727888863e-05,
+      "loss": 1.2758,
+      "step": 2927500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.374918731332806e-05,
+      "loss": 1.237,
+      "step": 2928000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.3747087347767494e-05,
+      "loss": 1.2569,
+      "step": 2928500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.3744987382206934e-05,
+      "loss": 1.2649,
+      "step": 2929000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.374288741664637e-05,
+      "loss": 1.2924,
+      "step": 2929500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.374079165101692e-05,
+      "loss": 1.2451,
+      "step": 2930000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.373869168545636e-05,
+      "loss": 1.3132,
+      "step": 2930500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.3736591719895795e-05,
+      "loss": 1.2719,
+      "step": 2931000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.373449175433523e-05,
+      "loss": 1.2473,
+      "step": 2931500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.373239178877467e-05,
+      "loss": 1.2582,
+      "step": 2932000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.3730291823214095e-05,
+      "loss": 1.2387,
+      "step": 2932500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.372819185765353e-05,
+      "loss": 1.3085,
+      "step": 2933000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.372609189209297e-05,
+      "loss": 1.2424,
+      "step": 2933500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.372399612646353e-05,
+      "loss": 1.2901,
+      "step": 2934000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.372189616090296e-05,
+      "loss": 1.2433,
+      "step": 2934500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.371979619534239e-05,
+      "loss": 1.2585,
+      "step": 2935000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.371769622978183e-05,
+      "loss": 1.273,
+      "step": 2935500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.371560046415239e-05,
+      "loss": 1.2416,
+      "step": 2936000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.3713500498591824e-05,
+      "loss": 1.2494,
+      "step": 2936500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.3711400533031264e-05,
+      "loss": 1.2569,
+      "step": 2937000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.370930056747069e-05,
+      "loss": 1.2505,
+      "step": 2937500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.370720480184125e-05,
+      "loss": 1.2539,
+      "step": 2938000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.3705104836280684e-05,
+      "loss": 1.252,
+      "step": 2938500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.3703004870720124e-05,
+      "loss": 1.2508,
+      "step": 2939000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.370090490515956e-05,
+      "loss": 1.2588,
+      "step": 2939500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.3698804939598985e-05,
+      "loss": 1.2657,
+      "step": 2940000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.3696709173969545e-05,
+      "loss": 1.2612,
+      "step": 2940500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.3694609208408985e-05,
+      "loss": 1.2671,
+      "step": 2941000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.369250924284842e-05,
+      "loss": 1.2621,
+      "step": 2941500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.369040927728785e-05,
+      "loss": 1.3084,
+      "step": 2942000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.3688309311727286e-05,
+      "loss": 1.2611,
+      "step": 2942500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.3686213546097846e-05,
+      "loss": 1.2815,
+      "step": 2943000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.368411358053728e-05,
+      "loss": 1.2471,
+      "step": 2943500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.368201361497672e-05,
+      "loss": 1.2752,
+      "step": 2944000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.3679913649416146e-05,
+      "loss": 1.2855,
+      "step": 2944500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.367781368385558e-05,
+      "loss": 1.2629,
+      "step": 2945000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.367571371829502e-05,
+      "loss": 1.2581,
+      "step": 2945500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.3673613752734453e-05,
+      "loss": 1.2516,
+      "step": 2946000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.367151378717389e-05,
+      "loss": 1.2788,
+      "step": 2946500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.366941802154444e-05,
+      "loss": 1.281,
+      "step": 2947000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.366731805598388e-05,
+      "loss": 1.2421,
+      "step": 2947500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.366522229035444e-05,
+      "loss": 1.2779,
+      "step": 2948000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.3663122324793875e-05,
+      "loss": 1.2748,
+      "step": 2948500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.366102235923331e-05,
+      "loss": 1.2938,
+      "step": 2949000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.365892239367274e-05,
+      "loss": 1.2804,
+      "step": 2949500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.3656822428112175e-05,
+      "loss": 1.2656,
+      "step": 2950000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.3654722462551615e-05,
+      "loss": 1.2637,
+      "step": 2950500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.365262249699105e-05,
+      "loss": 1.2759,
+      "step": 2951000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.365052253143048e-05,
+      "loss": 1.3018,
+      "step": 2951500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.3648426765801036e-05,
+      "loss": 1.2506,
+      "step": 2952000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.3646326800240476e-05,
+      "loss": 1.2807,
+      "step": 2952500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.364422683467991e-05,
+      "loss": 1.2556,
+      "step": 2953000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.364212686911934e-05,
+      "loss": 1.2495,
+      "step": 2953500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.364002690355878e-05,
+      "loss": 1.2583,
+      "step": 2954000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.3637926937998216e-05,
+      "loss": 1.2634,
+      "step": 2954500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.363582697243765e-05,
+      "loss": 1.2818,
+      "step": 2955000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.363372700687709e-05,
+      "loss": 1.2658,
+      "step": 2955500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.3631627041316524e-05,
+      "loss": 1.2508,
+      "step": 2956000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.362952707575596e-05,
+      "loss": 1.2946,
+      "step": 2956500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.362743131012651e-05,
+      "loss": 1.2787,
+      "step": 2957000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.362533134456595e-05,
+      "loss": 1.2815,
+      "step": 2957500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.3623231379005384e-05,
+      "loss": 1.2718,
+      "step": 2958000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.362113141344482e-05,
+      "loss": 1.2848,
+      "step": 2958500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.361903144788426e-05,
+      "loss": 1.2711,
+      "step": 2959000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.361693568225481e-05,
+      "loss": 1.2805,
+      "step": 2959500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.3614835716694245e-05,
+      "loss": 1.2787,
+      "step": 2960000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.3612735751133685e-05,
+      "loss": 1.235,
+      "step": 2960500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.361063578557312e-05,
+      "loss": 1.2918,
+      "step": 2961000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.360853582001255e-05,
+      "loss": 1.3014,
+      "step": 2961500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.3606435854451986e-05,
+      "loss": 1.2773,
+      "step": 2962000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.3604340088822546e-05,
+      "loss": 1.2502,
+      "step": 2962500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.360224012326198e-05,
+      "loss": 1.2761,
+      "step": 2963000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.360014015770141e-05,
+      "loss": 1.2191,
+      "step": 2963500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.359804019214085e-05,
+      "loss": 1.2778,
+      "step": 2964000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.359594022658028e-05,
+      "loss": 1.2618,
+      "step": 2964500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.359384026101971e-05,
+      "loss": 1.2734,
+      "step": 2965000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.3591740295459154e-05,
+      "loss": 1.2641,
+      "step": 2965500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.358964032989859e-05,
+      "loss": 1.274,
+      "step": 2966000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.358754036433802e-05,
+      "loss": 1.2451,
+      "step": 2966500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.35854487986397e-05,
+      "loss": 1.2825,
+      "step": 2967000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.358334883307914e-05,
+      "loss": 1.2664,
+      "step": 2967500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.3581248867518575e-05,
+      "loss": 1.2477,
+      "step": 2968000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.357914890195801e-05,
+      "loss": 1.279,
+      "step": 2968500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.357704893639744e-05,
+      "loss": 1.2749,
+      "step": 2969000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.3574948970836875e-05,
+      "loss": 1.2427,
+      "step": 2969500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.357284900527631e-05,
+      "loss": 1.2901,
+      "step": 2970000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.357074903971575e-05,
+      "loss": 1.2501,
+      "step": 2970500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.356865327408631e-05,
+      "loss": 1.269,
+      "step": 2971000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.3566553308525736e-05,
+      "loss": 1.2711,
+      "step": 2971500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.3564457542896296e-05,
+      "loss": 1.2674,
+      "step": 2972000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.356235757733573e-05,
+      "loss": 1.2632,
+      "step": 2972500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.356025761177517e-05,
+      "loss": 1.2817,
+      "step": 2973000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.35581576462146e-05,
+      "loss": 1.24,
+      "step": 2973500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.355605768065404e-05,
+      "loss": 1.2769,
+      "step": 2974000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.355395771509347e-05,
+      "loss": 1.2673,
+      "step": 2974500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.3551857749532904e-05,
+      "loss": 1.2529,
+      "step": 2975000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.3549757783972344e-05,
+      "loss": 1.2799,
+      "step": 2975500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.3547662018342904e-05,
+      "loss": 1.2601,
+      "step": 2976000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.354556205278233e-05,
+      "loss": 1.2751,
+      "step": 2976500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.3543462087221764e-05,
+      "loss": 1.2592,
+      "step": 2977000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.3541362121661205e-05,
+      "loss": 1.2692,
+      "step": 2977500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.3539266356031765e-05,
+      "loss": 1.2813,
+      "step": 2978000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.353717059040232e-05,
+      "loss": 1.2671,
+      "step": 2978500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.353507062484175e-05,
+      "loss": 1.2601,
+      "step": 2979000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.3532970659281185e-05,
+      "loss": 1.2517,
+      "step": 2979500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.3530870693720626e-05,
+      "loss": 1.2983,
+      "step": 2980000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.352877072816006e-05,
+      "loss": 1.2892,
+      "step": 2980500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.352667076259949e-05,
+      "loss": 1.2624,
+      "step": 2981000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.3524570797038926e-05,
+      "loss": 1.2708,
+      "step": 2981500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.352247083147836e-05,
+      "loss": 1.2479,
+      "step": 2982000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.352037506584892e-05,
+      "loss": 1.2673,
+      "step": 2982500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.351827510028836e-05,
+      "loss": 1.2788,
+      "step": 2983000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.351617513472779e-05,
+      "loss": 1.2673,
+      "step": 2983500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.351407936909835e-05,
+      "loss": 1.262,
+      "step": 2984000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.351197940353778e-05,
+      "loss": 1.2664,
+      "step": 2984500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.350987943797722e-05,
+      "loss": 1.2552,
+      "step": 2985000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.3507779472416654e-05,
+      "loss": 1.2421,
+      "step": 2985500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.350567950685608e-05,
+      "loss": 1.2544,
+      "step": 2986000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.350357954129552e-05,
+      "loss": 1.2573,
+      "step": 2986500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.3501479575734955e-05,
+      "loss": 1.276,
+      "step": 2987000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.3499383810105515e-05,
+      "loss": 1.2476,
+      "step": 2987500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.3497283844544955e-05,
+      "loss": 1.2706,
+      "step": 2988000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.349518387898438e-05,
+      "loss": 1.2864,
+      "step": 2988500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.3493083913423815e-05,
+      "loss": 1.2899,
+      "step": 2989000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.3490983947863256e-05,
+      "loss": 1.2559,
+      "step": 2989500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.348888398230269e-05,
+      "loss": 1.2899,
+      "step": 2990000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.348678401674212e-05,
+      "loss": 1.2857,
+      "step": 2990500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.348468405118156e-05,
+      "loss": 1.2683,
+      "step": 2991000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.3482584085620996e-05,
+      "loss": 1.2764,
+      "step": 2991500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.348048412006043e-05,
+      "loss": 1.2795,
+      "step": 2992000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.347838415449987e-05,
+      "loss": 1.2778,
+      "step": 2992500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.3476284188939303e-05,
+      "loss": 1.2778,
+      "step": 2993000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.347418842330986e-05,
+      "loss": 1.2904,
+      "step": 2993500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.347208845774929e-05,
+      "loss": 1.2703,
+      "step": 2994000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.346998849218873e-05,
+      "loss": 1.2864,
+      "step": 2994500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.3467892726559284e-05,
+      "loss": 1.2469,
+      "step": 2995000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.346579276099872e-05,
+      "loss": 1.2605,
+      "step": 2995500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.346369279543816e-05,
+      "loss": 1.2682,
+      "step": 2996000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.346159282987759e-05,
+      "loss": 1.2911,
+      "step": 2996500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.3459497064248145e-05,
+      "loss": 1.2421,
+      "step": 2997000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.345739709868758e-05,
+      "loss": 1.2776,
+      "step": 2997500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.345529713312702e-05,
+      "loss": 1.2784,
+      "step": 2998000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.345319716756645e-05,
+      "loss": 1.2667,
+      "step": 2998500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.3451097202005886e-05,
+      "loss": 1.2566,
+      "step": 2999000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.3448997236445326e-05,
+      "loss": 1.2322,
+      "step": 2999500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.344690147081588e-05,
+      "loss": 1.2769,
+      "step": 3000000
+    },
+    {
+      "epoch": 1.8,
+      "eval_loss": 1.2166739702224731,
+      "eval_runtime": 1103.532,
+      "eval_samples_per_second": 477.304,
+      "eval_steps_per_second": 79.551,
+      "step": 3000000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.344480150525531e-05,
+      "loss": 1.285,
+      "step": 3000500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.3442701539694746e-05,
+      "loss": 1.2474,
+      "step": 3001000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.3440601574134187e-05,
+      "loss": 1.2763,
+      "step": 3001500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.343850580850474e-05,
+      "loss": 1.2567,
+      "step": 3002000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.3436405842944174e-05,
+      "loss": 1.2724,
+      "step": 3002500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.3434305877383614e-05,
+      "loss": 1.2459,
+      "step": 3003000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.343220591182305e-05,
+      "loss": 1.2671,
+      "step": 3003500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.34301101461936e-05,
+      "loss": 1.2762,
+      "step": 3004000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.3428010180633034e-05,
+      "loss": 1.2305,
+      "step": 3004500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.3425910215072475e-05,
+      "loss": 1.2917,
+      "step": 3005000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.342381024951191e-05,
+      "loss": 1.245,
+      "step": 3005500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.342171028395134e-05,
+      "loss": 1.246,
+      "step": 3006000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.341961031839078e-05,
+      "loss": 1.2648,
+      "step": 3006500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.3417510352830215e-05,
+      "loss": 1.2847,
+      "step": 3007000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.341541038726965e-05,
+      "loss": 1.2734,
+      "step": 3007500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.341331042170908e-05,
+      "loss": 1.2603,
+      "step": 3008000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.3411210456148516e-05,
+      "loss": 1.244,
+      "step": 3008500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.3409114690519076e-05,
+      "loss": 1.262,
+      "step": 3009000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.340701472495851e-05,
+      "loss": 1.2634,
+      "step": 3009500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.340491475939795e-05,
+      "loss": 1.2499,
+      "step": 3010000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.3402814793837376e-05,
+      "loss": 1.2837,
+      "step": 3010500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.3400714828276817e-05,
+      "loss": 1.2401,
+      "step": 3011000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.339861906264738e-05,
+      "loss": 1.2328,
+      "step": 3011500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.339651909708681e-05,
+      "loss": 1.2649,
+      "step": 3012000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.3394419131526244e-05,
+      "loss": 1.2763,
+      "step": 3012500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.339231916596568e-05,
+      "loss": 1.2608,
+      "step": 3013000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.339022340033624e-05,
+      "loss": 1.2733,
+      "step": 3013500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.338812343477567e-05,
+      "loss": 1.2699,
+      "step": 3014000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.3386023469215105e-05,
+      "loss": 1.2651,
+      "step": 3014500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.338392350365454e-05,
+      "loss": 1.2768,
+      "step": 3015000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.33818277380251e-05,
+      "loss": 1.26,
+      "step": 3015500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.337972777246453e-05,
+      "loss": 1.2685,
+      "step": 3016000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.3377627806903965e-05,
+      "loss": 1.2775,
+      "step": 3016500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.3375532041274526e-05,
+      "loss": 1.2761,
+      "step": 3017000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.337343207571396e-05,
+      "loss": 1.2665,
+      "step": 3017500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.337133211015339e-05,
+      "loss": 1.2405,
+      "step": 3018000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.336923214459283e-05,
+      "loss": 1.2575,
+      "step": 3018500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.3367132179032266e-05,
+      "loss": 1.2699,
+      "step": 3019000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.33650322134717e-05,
+      "loss": 1.2717,
+      "step": 3019500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.336293224791113e-05,
+      "loss": 1.2667,
+      "step": 3020000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.3360832282350567e-05,
+      "loss": 1.2688,
+      "step": 3020500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.335873651672113e-05,
+      "loss": 1.2488,
+      "step": 3021000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.335664075109168e-05,
+      "loss": 1.2516,
+      "step": 3021500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.3354540785531114e-05,
+      "loss": 1.2612,
+      "step": 3022000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.3352440819970554e-05,
+      "loss": 1.2439,
+      "step": 3022500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.335034085440999e-05,
+      "loss": 1.2796,
+      "step": 3023000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.334824088884942e-05,
+      "loss": 1.2655,
+      "step": 3023500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.334614092328886e-05,
+      "loss": 1.2487,
+      "step": 3024000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.3344040957728295e-05,
+      "loss": 1.2884,
+      "step": 3024500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.334194099216773e-05,
+      "loss": 1.2638,
+      "step": 3025000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.333984102660716e-05,
+      "loss": 1.2818,
+      "step": 3025500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.3337741061046595e-05,
+      "loss": 1.2924,
+      "step": 3026000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.3335641095486035e-05,
+      "loss": 1.2554,
+      "step": 3026500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.333354112992547e-05,
+      "loss": 1.279,
+      "step": 3027000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.333144536429602e-05,
+      "loss": 1.2362,
+      "step": 3027500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.332934959866658e-05,
+      "loss": 1.2775,
+      "step": 3028000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3327249633106016e-05,
+      "loss": 1.252,
+      "step": 3028500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3325149667545456e-05,
+      "loss": 1.2668,
+      "step": 3029000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.332304970198488e-05,
+      "loss": 1.2745,
+      "step": 3029500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.332094973642432e-05,
+      "loss": 1.2744,
+      "step": 3030000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.331885397079488e-05,
+      "loss": 1.2852,
+      "step": 3030500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.331675400523432e-05,
+      "loss": 1.2807,
+      "step": 3031000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.331465403967375e-05,
+      "loss": 1.2418,
+      "step": 3031500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3312554074113184e-05,
+      "loss": 1.2644,
+      "step": 3032000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.331045410855262e-05,
+      "loss": 1.225,
+      "step": 3032500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.330835414299205e-05,
+      "loss": 1.2714,
+      "step": 3033000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.330625417743149e-05,
+      "loss": 1.2526,
+      "step": 3033500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.330415841180205e-05,
+      "loss": 1.2309,
+      "step": 3034000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.330205844624148e-05,
+      "loss": 1.2763,
+      "step": 3034500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.329995848068091e-05,
+      "loss": 1.2683,
+      "step": 3035000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.329785851512035e-05,
+      "loss": 1.2836,
+      "step": 3035500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3295758549559785e-05,
+      "loss": 1.2767,
+      "step": 3036000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.329365858399922e-05,
+      "loss": 1.2516,
+      "step": 3036500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.329155861843866e-05,
+      "loss": 1.2932,
+      "step": 3037000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.328945865287809e-05,
+      "loss": 1.2552,
+      "step": 3037500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3287362887248646e-05,
+      "loss": 1.2731,
+      "step": 3038000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.328526292168808e-05,
+      "loss": 1.272,
+      "step": 3038500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.328316295612752e-05,
+      "loss": 1.2585,
+      "step": 3039000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.328106299056695e-05,
+      "loss": 1.269,
+      "step": 3039500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3278971424868634e-05,
+      "loss": 1.265,
+      "step": 3040000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.327687145930807e-05,
+      "loss": 1.2741,
+      "step": 3040500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.327477149374751e-05,
+      "loss": 1.2566,
+      "step": 3041000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3272671528186934e-05,
+      "loss": 1.2786,
+      "step": 3041500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.327057156262637e-05,
+      "loss": 1.2618,
+      "step": 3042000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.326847159706581e-05,
+      "loss": 1.2394,
+      "step": 3042500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.326637163150524e-05,
+      "loss": 1.279,
+      "step": 3043000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3264271665944675e-05,
+      "loss": 1.29,
+      "step": 3043500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.3262171700384115e-05,
+      "loss": 1.2744,
+      "step": 3044000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.326007593475467e-05,
+      "loss": 1.268,
+      "step": 3044500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.32579759691941e-05,
+      "loss": 1.2659,
+      "step": 3045000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.3255876003633536e-05,
+      "loss": 1.2498,
+      "step": 3045500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.3253776038072976e-05,
+      "loss": 1.2531,
+      "step": 3046000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.325167607251241e-05,
+      "loss": 1.2629,
+      "step": 3046500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.324957610695184e-05,
+      "loss": 1.2382,
+      "step": 3047000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.32474803413224e-05,
+      "loss": 1.2571,
+      "step": 3047500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.3245380375761837e-05,
+      "loss": 1.2552,
+      "step": 3048000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.324328041020127e-05,
+      "loss": 1.2866,
+      "step": 3048500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.324118044464071e-05,
+      "loss": 1.278,
+      "step": 3049000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.3239080479080144e-05,
+      "loss": 1.2773,
+      "step": 3049500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.323698051351958e-05,
+      "loss": 1.2513,
+      "step": 3050000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.323488054795902e-05,
+      "loss": 1.2832,
+      "step": 3050500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.323278058239845e-05,
+      "loss": 1.2647,
+      "step": 3051000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.3230684816769004e-05,
+      "loss": 1.2592,
+      "step": 3051500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.322858485120844e-05,
+      "loss": 1.2351,
+      "step": 3052000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.322648488564788e-05,
+      "loss": 1.2721,
+      "step": 3052500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.322438492008731e-05,
+      "loss": 1.2302,
+      "step": 3053000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.3222289154457865e-05,
+      "loss": 1.2492,
+      "step": 3053500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.3220189188897305e-05,
+      "loss": 1.2827,
+      "step": 3054000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.321808922333674e-05,
+      "loss": 1.264,
+      "step": 3054500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.321598925777617e-05,
+      "loss": 1.2814,
+      "step": 3055000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.321389769207785e-05,
+      "loss": 1.261,
+      "step": 3055500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.321179772651728e-05,
+      "loss": 1.2875,
+      "step": 3056000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.320969776095672e-05,
+      "loss": 1.2747,
+      "step": 3056500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.320759779539615e-05,
+      "loss": 1.2918,
+      "step": 3057000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.3205497829835587e-05,
+      "loss": 1.256,
+      "step": 3057500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.320339786427503e-05,
+      "loss": 1.2927,
+      "step": 3058000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.320129789871446e-05,
+      "loss": 1.2408,
+      "step": 3058500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.3199197933153894e-05,
+      "loss": 1.2707,
+      "step": 3059000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.319710216752445e-05,
+      "loss": 1.2634,
+      "step": 3059500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.319500220196389e-05,
+      "loss": 1.2753,
+      "step": 3060000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.319290643633444e-05,
+      "loss": 1.2585,
+      "step": 3060500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.3190806470773875e-05,
+      "loss": 1.2788,
+      "step": 3061000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.3188706505213315e-05,
+      "loss": 1.2403,
+      "step": 3061500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.318660653965275e-05,
+      "loss": 1.2662,
+      "step": 3062000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.318451077402331e-05,
+      "loss": 1.2324,
+      "step": 3062500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.3182410808462735e-05,
+      "loss": 1.2866,
+      "step": 3063000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.3180310842902176e-05,
+      "loss": 1.2782,
+      "step": 3063500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.3178215077272736e-05,
+      "loss": 1.251,
+      "step": 3064000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.317611511171217e-05,
+      "loss": 1.2853,
+      "step": 3064500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.31740151461516e-05,
+      "loss": 1.2673,
+      "step": 3065000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.3171919380522156e-05,
+      "loss": 1.254,
+      "step": 3065500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.31698194149616e-05,
+      "loss": 1.3198,
+      "step": 3066000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.316771944940103e-05,
+      "loss": 1.253,
+      "step": 3066500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.316561948384047e-05,
+      "loss": 1.2679,
+      "step": 3067000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.31635195182799e-05,
+      "loss": 1.2806,
+      "step": 3067500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.316141955271933e-05,
+      "loss": 1.2512,
+      "step": 3068000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.315931958715877e-05,
+      "loss": 1.2734,
+      "step": 3068500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.3157219621598204e-05,
+      "loss": 1.2736,
+      "step": 3069000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.315511965603764e-05,
+      "loss": 1.2782,
+      "step": 3069500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.315301969047708e-05,
+      "loss": 1.2426,
+      "step": 3070000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.315091972491651e-05,
+      "loss": 1.2562,
+      "step": 3070500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.3148819759355945e-05,
+      "loss": 1.2655,
+      "step": 3071000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.3146719793795385e-05,
+      "loss": 1.2834,
+      "step": 3071500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.314461982823482e-05,
+      "loss": 1.2869,
+      "step": 3072000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.314251986267425e-05,
+      "loss": 1.2349,
+      "step": 3072500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.3140419897113685e-05,
+      "loss": 1.2657,
+      "step": 3073000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.3138324131484246e-05,
+      "loss": 1.2556,
+      "step": 3073500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.313622416592368e-05,
+      "loss": 1.2601,
+      "step": 3074000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.313412840029423e-05,
+      "loss": 1.2684,
+      "step": 3074500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.313202843473367e-05,
+      "loss": 1.2592,
+      "step": 3075000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.3129928469173106e-05,
+      "loss": 1.2785,
+      "step": 3075500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.312782850361254e-05,
+      "loss": 1.2401,
+      "step": 3076000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.312572853805198e-05,
+      "loss": 1.2561,
+      "step": 3076500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.3123628572491414e-05,
+      "loss": 1.2457,
+      "step": 3077000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.312152860693085e-05,
+      "loss": 1.2543,
+      "step": 3077500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.31194328413014e-05,
+      "loss": 1.2555,
+      "step": 3078000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.311733287574084e-05,
+      "loss": 1.2487,
+      "step": 3078500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.3115232910180274e-05,
+      "loss": 1.2905,
+      "step": 3079000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.311313294461971e-05,
+      "loss": 1.242,
+      "step": 3079500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.311103297905915e-05,
+      "loss": 1.2668,
+      "step": 3080000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.3108933013498575e-05,
+      "loss": 1.2938,
+      "step": 3080500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.310683304793801e-05,
+      "loss": 1.2484,
+      "step": 3081000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.310473308237745e-05,
+      "loss": 1.2249,
+      "step": 3081500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.310263731674801e-05,
+      "loss": 1.2519,
+      "step": 3082000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.3100537351187435e-05,
+      "loss": 1.2675,
+      "step": 3082500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.3098437385626876e-05,
+      "loss": 1.2619,
+      "step": 3083000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.3096341619997436e-05,
+      "loss": 1.2876,
+      "step": 3083500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.309424165443687e-05,
+      "loss": 1.2662,
+      "step": 3084000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.30921416888763e-05,
+      "loss": 1.2465,
+      "step": 3084500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.3090041723315736e-05,
+      "loss": 1.2585,
+      "step": 3085000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.308794175775517e-05,
+      "loss": 1.2758,
+      "step": 3085500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.30858417921946e-05,
+      "loss": 1.2778,
+      "step": 3086000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.3083741826634044e-05,
+      "loss": 1.2478,
+      "step": 3086500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.308164186107348e-05,
+      "loss": 1.2621,
+      "step": 3087000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.307955029537516e-05,
+      "loss": 1.2696,
+      "step": 3087500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.307745032981459e-05,
+      "loss": 1.2838,
+      "step": 3088000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.3075350364254024e-05,
+      "loss": 1.2666,
+      "step": 3088500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.3073250398693465e-05,
+      "loss": 1.2525,
+      "step": 3089000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.30711504331329e-05,
+      "loss": 1.257,
+      "step": 3089500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.306905046757233e-05,
+      "loss": 1.2452,
+      "step": 3090000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.3066950502011765e-05,
+      "loss": 1.2539,
+      "step": 3090500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.30648505364512e-05,
+      "loss": 1.2632,
+      "step": 3091000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.306275057089064e-05,
+      "loss": 1.2429,
+      "step": 3091500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.30606548052612e-05,
+      "loss": 1.2784,
+      "step": 3092000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.3058554839700626e-05,
+      "loss": 1.2769,
+      "step": 3092500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.305645487414006e-05,
+      "loss": 1.272,
+      "step": 3093000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.30543549085795e-05,
+      "loss": 1.2478,
+      "step": 3093500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.305225494301893e-05,
+      "loss": 1.2507,
+      "step": 3094000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.3050159177389486e-05,
+      "loss": 1.2705,
+      "step": 3094500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.304805921182892e-05,
+      "loss": 1.2597,
+      "step": 3095000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.304595924626836e-05,
+      "loss": 1.2493,
+      "step": 3095500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.3043859280707794e-05,
+      "loss": 1.2844,
+      "step": 3096000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.304175931514723e-05,
+      "loss": 1.2512,
+      "step": 3096500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.303965934958667e-05,
+      "loss": 1.2771,
+      "step": 3097000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.30375593840261e-05,
+      "loss": 1.2674,
+      "step": 3097500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.3035459418465534e-05,
+      "loss": 1.2445,
+      "step": 3098000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.3033359452904974e-05,
+      "loss": 1.2614,
+      "step": 3098500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.303126368727553e-05,
+      "loss": 1.253,
+      "step": 3099000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.302916372171496e-05,
+      "loss": 1.2486,
+      "step": 3099500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.30270637561544e-05,
+      "loss": 1.2472,
+      "step": 3100000
+    },
+    {
+      "epoch": 1.86,
+      "eval_loss": 1.209123134613037,
+      "eval_runtime": 1124.5793,
+      "eval_samples_per_second": 468.371,
+      "eval_steps_per_second": 78.062,
+      "step": 3100000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.3024963790593835e-05,
+      "loss": 1.2343,
+      "step": 3100500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.302286802496439e-05,
+      "loss": 1.224,
+      "step": 3101000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.302076805940382e-05,
+      "loss": 1.26,
+      "step": 3101500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.301866809384326e-05,
+      "loss": 1.2775,
+      "step": 3102000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.3016568128282696e-05,
+      "loss": 1.2587,
+      "step": 3102500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.301447236265325e-05,
+      "loss": 1.2448,
+      "step": 3103000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.301237659702381e-05,
+      "loss": 1.2553,
+      "step": 3103500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.3010280831394363e-05,
+      "loss": 1.2446,
+      "step": 3104000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.3008180865833804e-05,
+      "loss": 1.29,
+      "step": 3104500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.300608090027324e-05,
+      "loss": 1.2545,
+      "step": 3105000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.300398093471267e-05,
+      "loss": 1.2646,
+      "step": 3105500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.300188096915211e-05,
+      "loss": 1.2599,
+      "step": 3106000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.299978100359154e-05,
+      "loss": 1.2752,
+      "step": 3106500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.299768103803097e-05,
+      "loss": 1.276,
+      "step": 3107000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.299558107247041e-05,
+      "loss": 1.2405,
+      "step": 3107500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.2993481106909845e-05,
+      "loss": 1.2672,
+      "step": 3108000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.299138114134928e-05,
+      "loss": 1.2679,
+      "step": 3108500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.298928537571983e-05,
+      "loss": 1.2633,
+      "step": 3109000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.298718541015927e-05,
+      "loss": 1.2684,
+      "step": 3109500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.2985085444598705e-05,
+      "loss": 1.2687,
+      "step": 3110000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.298298547903814e-05,
+      "loss": 1.2805,
+      "step": 3110500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.298088551347758e-05,
+      "loss": 1.2818,
+      "step": 3111000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.297878974784813e-05,
+      "loss": 1.2645,
+      "step": 3111500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.2976689782287566e-05,
+      "loss": 1.2534,
+      "step": 3112000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.2974589816727006e-05,
+      "loss": 1.2574,
+      "step": 3112500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.297248985116644e-05,
+      "loss": 1.2234,
+      "step": 3113000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.297038988560587e-05,
+      "loss": 1.2387,
+      "step": 3113500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.2968289920045313e-05,
+      "loss": 1.2629,
+      "step": 3114000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.296619415441587e-05,
+      "loss": 1.2619,
+      "step": 3114500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.29640941888553e-05,
+      "loss": 1.282,
+      "step": 3115000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.2961994223294734e-05,
+      "loss": 1.2529,
+      "step": 3115500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.2959894257734174e-05,
+      "loss": 1.2745,
+      "step": 3116000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.295779429217361e-05,
+      "loss": 1.2258,
+      "step": 3116500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.295569432661304e-05,
+      "loss": 1.2688,
+      "step": 3117000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.2953598560983595e-05,
+      "loss": 1.2593,
+      "step": 3117500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.2951498595423035e-05,
+      "loss": 1.2499,
+      "step": 3118000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.294939862986247e-05,
+      "loss": 1.2717,
+      "step": 3118500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.294729866430191e-05,
+      "loss": 1.2887,
+      "step": 3119000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.294519869874134e-05,
+      "loss": 1.2704,
+      "step": 3119500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.2943098733180776e-05,
+      "loss": 1.2706,
+      "step": 3120000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.294100296755133e-05,
+      "loss": 1.2656,
+      "step": 3120500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.293890300199077e-05,
+      "loss": 1.2632,
+      "step": 3121000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.29368030364302e-05,
+      "loss": 1.3073,
+      "step": 3121500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.2934703070869636e-05,
+      "loss": 1.2898,
+      "step": 3122000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.2932603105309077e-05,
+      "loss": 1.2434,
+      "step": 3122500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.293050733967963e-05,
+      "loss": 1.2646,
+      "step": 3123000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.2928407374119064e-05,
+      "loss": 1.2725,
+      "step": 3123500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.29263074085585e-05,
+      "loss": 1.2688,
+      "step": 3124000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.292420744299794e-05,
+      "loss": 1.2712,
+      "step": 3124500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.292211167736849e-05,
+      "loss": 1.2777,
+      "step": 3125000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.2920011711807924e-05,
+      "loss": 1.2765,
+      "step": 3125500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.2917911746247365e-05,
+      "loss": 1.2509,
+      "step": 3126000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.29158117806868e-05,
+      "loss": 1.2731,
+      "step": 3126500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.291371181512623e-05,
+      "loss": 1.2582,
+      "step": 3127000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.2911616049496785e-05,
+      "loss": 1.2426,
+      "step": 3127500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.2909516083936225e-05,
+      "loss": 1.2886,
+      "step": 3128000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.290741611837566e-05,
+      "loss": 1.2447,
+      "step": 3128500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.290531615281509e-05,
+      "loss": 1.2977,
+      "step": 3129000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.290321618725453e-05,
+      "loss": 1.292,
+      "step": 3129500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.2901116221693966e-05,
+      "loss": 1.2684,
+      "step": 3130000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.28990162561334e-05,
+      "loss": 1.2826,
+      "step": 3130500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.289691629057283e-05,
+      "loss": 1.2486,
+      "step": 3131000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.289482052494339e-05,
+      "loss": 1.2773,
+      "step": 3131500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.2892720559382827e-05,
+      "loss": 1.2591,
+      "step": 3132000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.289062059382226e-05,
+      "loss": 1.2362,
+      "step": 3132500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.288852482819282e-05,
+      "loss": 1.2681,
+      "step": 3133000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.2886429062563374e-05,
+      "loss": 1.2386,
+      "step": 3133500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.288432909700281e-05,
+      "loss": 1.2811,
+      "step": 3134000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.288223333137337e-05,
+      "loss": 1.2685,
+      "step": 3134500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.2880133365812794e-05,
+      "loss": 1.2437,
+      "step": 3135000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.2878033400252235e-05,
+      "loss": 1.2618,
+      "step": 3135500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.287593343469167e-05,
+      "loss": 1.2622,
+      "step": 3136000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.28738334691311e-05,
+      "loss": 1.2779,
+      "step": 3136500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.287173350357054e-05,
+      "loss": 1.2474,
+      "step": 3137000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.2869633538009975e-05,
+      "loss": 1.2797,
+      "step": 3137500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.286753357244941e-05,
+      "loss": 1.2861,
+      "step": 3138000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.286543360688885e-05,
+      "loss": 1.2461,
+      "step": 3138500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.286333364132828e-05,
+      "loss": 1.2612,
+      "step": 3139000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.2861233675767716e-05,
+      "loss": 1.2474,
+      "step": 3139500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.2859133710207156e-05,
+      "loss": 1.2752,
+      "step": 3140000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.285703374464658e-05,
+      "loss": 1.2569,
+      "step": 3140500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.285493377908602e-05,
+      "loss": 1.2494,
+      "step": 3141000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.2852833813525457e-05,
+      "loss": 1.2851,
+      "step": 3141500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.285073384796489e-05,
+      "loss": 1.2685,
+      "step": 3142000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.284863388240433e-05,
+      "loss": 1.2703,
+      "step": 3142500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.2846542316706004e-05,
+      "loss": 1.2539,
+      "step": 3143000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.2844442351145444e-05,
+      "loss": 1.2712,
+      "step": 3143500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.284234238558488e-05,
+      "loss": 1.2302,
+      "step": 3144000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.284024661995543e-05,
+      "loss": 1.2721,
+      "step": 3144500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.2838146654394865e-05,
+      "loss": 1.2504,
+      "step": 3145000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.2836046688834305e-05,
+      "loss": 1.2896,
+      "step": 3145500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.283394672327374e-05,
+      "loss": 1.2657,
+      "step": 3146000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.283184675771317e-05,
+      "loss": 1.2266,
+      "step": 3146500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.282974679215261e-05,
+      "loss": 1.2686,
+      "step": 3147000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.2827646826592045e-05,
+      "loss": 1.2506,
+      "step": 3147500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.282554686103148e-05,
+      "loss": 1.2483,
+      "step": 3148000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.282344689547091e-05,
+      "loss": 1.2456,
+      "step": 3148500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.2821346929910346e-05,
+      "loss": 1.2521,
+      "step": 3149000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.2819251164280906e-05,
+      "loss": 1.2561,
+      "step": 3149500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.281715539865146e-05,
+      "loss": 1.25,
+      "step": 3150000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.281505963302201e-05,
+      "loss": 1.2924,
+      "step": 3150500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.2812959667461454e-05,
+      "loss": 1.243,
+      "step": 3151000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.281085970190089e-05,
+      "loss": 1.2674,
+      "step": 3151500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.280875973634032e-05,
+      "loss": 1.2568,
+      "step": 3152000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.280665977077976e-05,
+      "loss": 1.2808,
+      "step": 3152500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.2804559805219194e-05,
+      "loss": 1.2469,
+      "step": 3153000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.280245983965863e-05,
+      "loss": 1.2478,
+      "step": 3153500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.280035987409807e-05,
+      "loss": 1.275,
+      "step": 3154000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.27982599085375e-05,
+      "loss": 1.2479,
+      "step": 3154500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.2796159942976935e-05,
+      "loss": 1.2588,
+      "step": 3155000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.279405997741637e-05,
+      "loss": 1.2645,
+      "step": 3155500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.27919600118558e-05,
+      "loss": 1.2647,
+      "step": 3156000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.278986004629524e-05,
+      "loss": 1.266,
+      "step": 3156500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.27877642806658e-05,
+      "loss": 1.2583,
+      "step": 3157000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.278566431510523e-05,
+      "loss": 1.2339,
+      "step": 3157500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.278356434954466e-05,
+      "loss": 1.299,
+      "step": 3158000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.27814643839841e-05,
+      "loss": 1.2854,
+      "step": 3158500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.2779364418423536e-05,
+      "loss": 1.2619,
+      "step": 3159000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.277726445286297e-05,
+      "loss": 1.2424,
+      "step": 3159500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.277516448730241e-05,
+      "loss": 1.2615,
+      "step": 3160000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.277306452174184e-05,
+      "loss": 1.2373,
+      "step": 3160500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.27709687561124e-05,
+      "loss": 1.2556,
+      "step": 3161000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.276886879055183e-05,
+      "loss": 1.229,
+      "step": 3161500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.276676882499127e-05,
+      "loss": 1.2517,
+      "step": 3162000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.2764668859430704e-05,
+      "loss": 1.266,
+      "step": 3162500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.276256889387014e-05,
+      "loss": 1.2544,
+      "step": 3163000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.27604731282407e-05,
+      "loss": 1.2833,
+      "step": 3163500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.275837316268013e-05,
+      "loss": 1.2409,
+      "step": 3164000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.2756273197119565e-05,
+      "loss": 1.3014,
+      "step": 3164500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.2754173231559005e-05,
+      "loss": 1.2382,
+      "step": 3165000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.275207326599844e-05,
+      "loss": 1.2918,
+      "step": 3165500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.274997330043787e-05,
+      "loss": 1.2665,
+      "step": 3166000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.274787333487731e-05,
+      "loss": 1.2504,
+      "step": 3166500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.2745773369316746e-05,
+      "loss": 1.2563,
+      "step": 3167000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.27436776036873e-05,
+      "loss": 1.2843,
+      "step": 3167500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.274157763812673e-05,
+      "loss": 1.256,
+      "step": 3168000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.273947767256617e-05,
+      "loss": 1.2548,
+      "step": 3168500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.2737377707005606e-05,
+      "loss": 1.2641,
+      "step": 3169000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.273528194137616e-05,
+      "loss": 1.2389,
+      "step": 3169500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.273318197581559e-05,
+      "loss": 1.2574,
+      "step": 3170000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.2731082010255034e-05,
+      "loss": 1.2612,
+      "step": 3170500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.272898204469447e-05,
+      "loss": 1.2486,
+      "step": 3171000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.272688627906502e-05,
+      "loss": 1.218,
+      "step": 3171500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.2724790513435574e-05,
+      "loss": 1.3072,
+      "step": 3172000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.2722690547875014e-05,
+      "loss": 1.2725,
+      "step": 3172500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.272059058231445e-05,
+      "loss": 1.2654,
+      "step": 3173000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.271849061675388e-05,
+      "loss": 1.2903,
+      "step": 3173500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.271639065119332e-05,
+      "loss": 1.2635,
+      "step": 3174000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.2714290685632755e-05,
+      "loss": 1.2605,
+      "step": 3174500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.271219072007219e-05,
+      "loss": 1.2348,
+      "step": 3175000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.271009075451163e-05,
+      "loss": 1.2429,
+      "step": 3175500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.270799498888218e-05,
+      "loss": 1.2244,
+      "step": 3176000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.2705895023321616e-05,
+      "loss": 1.2714,
+      "step": 3176500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.270379925769217e-05,
+      "loss": 1.2345,
+      "step": 3177000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.270169929213161e-05,
+      "loss": 1.2598,
+      "step": 3177500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.269959932657104e-05,
+      "loss": 1.2585,
+      "step": 3178000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.2697499361010477e-05,
+      "loss": 1.2597,
+      "step": 3178500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.269539939544992e-05,
+      "loss": 1.2622,
+      "step": 3179000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.269330362982047e-05,
+      "loss": 1.2243,
+      "step": 3179500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.2691203664259904e-05,
+      "loss": 1.2413,
+      "step": 3180000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.268910369869934e-05,
+      "loss": 1.2682,
+      "step": 3180500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.268700373313878e-05,
+      "loss": 1.2487,
+      "step": 3181000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.268490376757821e-05,
+      "loss": 1.2745,
+      "step": 3181500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.2682808001948765e-05,
+      "loss": 1.2335,
+      "step": 3182000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.26807080363882e-05,
+      "loss": 1.2597,
+      "step": 3182500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.267860807082764e-05,
+      "loss": 1.2745,
+      "step": 3183000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.267651230519819e-05,
+      "loss": 1.2564,
+      "step": 3183500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.2674412339637625e-05,
+      "loss": 1.2421,
+      "step": 3184000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.2672312374077065e-05,
+      "loss": 1.253,
+      "step": 3184500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.26702124085165e-05,
+      "loss": 1.2387,
+      "step": 3185000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.266811244295593e-05,
+      "loss": 1.2667,
+      "step": 3185500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.266601247739537e-05,
+      "loss": 1.2406,
+      "step": 3186000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.2663912511834806e-05,
+      "loss": 1.2563,
+      "step": 3186500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.266181254627424e-05,
+      "loss": 1.2614,
+      "step": 3187000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.265971258071368e-05,
+      "loss": 1.2685,
+      "step": 3187500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.265761681508423e-05,
+      "loss": 1.2725,
+      "step": 3188000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.265551684952367e-05,
+      "loss": 1.2487,
+      "step": 3188500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.26534168839631e-05,
+      "loss": 1.2513,
+      "step": 3189000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.265131691840254e-05,
+      "loss": 1.2562,
+      "step": 3189500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.2649216952841974e-05,
+      "loss": 1.2393,
+      "step": 3190000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.264711698728141e-05,
+      "loss": 1.2384,
+      "step": 3190500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.264502122165197e-05,
+      "loss": 1.2399,
+      "step": 3191000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.26429212560914e-05,
+      "loss": 1.2503,
+      "step": 3191500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.2640821290530835e-05,
+      "loss": 1.2649,
+      "step": 3192000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.2638721324970275e-05,
+      "loss": 1.248,
+      "step": 3192500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.263662135940971e-05,
+      "loss": 1.2149,
+      "step": 3193000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.263452139384914e-05,
+      "loss": 1.2251,
+      "step": 3193500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.2632421428288575e-05,
+      "loss": 1.2697,
+      "step": 3194000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.263032146272801e-05,
+      "loss": 1.2772,
+      "step": 3194500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.262822989702969e-05,
+      "loss": 1.2437,
+      "step": 3195000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.262612993146912e-05,
+      "loss": 1.2492,
+      "step": 3195500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.2624029965908556e-05,
+      "loss": 1.2316,
+      "step": 3196000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.2621930000347996e-05,
+      "loss": 1.2439,
+      "step": 3196500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.261983003478743e-05,
+      "loss": 1.2871,
+      "step": 3197000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.261773006922686e-05,
+      "loss": 1.2511,
+      "step": 3197500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.2615630103666304e-05,
+      "loss": 1.2308,
+      "step": 3198000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.261353013810573e-05,
+      "loss": 1.2586,
+      "step": 3198500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.261143437247629e-05,
+      "loss": 1.253,
+      "step": 3199000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.2609338606846844e-05,
+      "loss": 1.2774,
+      "step": 3199500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.2607238641286284e-05,
+      "loss": 1.2039,
+      "step": 3200000
+    },
+    {
+      "epoch": 1.92,
+      "eval_loss": 1.2074214220046997,
+      "eval_runtime": 1108.4818,
+      "eval_samples_per_second": 475.172,
+      "eval_steps_per_second": 79.196,
+      "step": 3200000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.260513867572572e-05,
+      "loss": 1.2663,
+      "step": 3200500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.260303871016515e-05,
+      "loss": 1.253,
+      "step": 3201000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.260093874460459e-05,
+      "loss": 1.2607,
+      "step": 3201500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.2598838779044025e-05,
+      "loss": 1.2493,
+      "step": 3202000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.259674301341458e-05,
+      "loss": 1.2476,
+      "step": 3202500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.259464304785401e-05,
+      "loss": 1.2731,
+      "step": 3203000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.259254308229345e-05,
+      "loss": 1.2501,
+      "step": 3203500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.2590443116732886e-05,
+      "loss": 1.2505,
+      "step": 3204000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.258834315117232e-05,
+      "loss": 1.2692,
+      "step": 3204500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.258624318561176e-05,
+      "loss": 1.245,
+      "step": 3205000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.258414741998231e-05,
+      "loss": 1.2613,
+      "step": 3205500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.2582047454421746e-05,
+      "loss": 1.2524,
+      "step": 3206000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.257994748886119e-05,
+      "loss": 1.2695,
+      "step": 3206500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.257784752330062e-05,
+      "loss": 1.2779,
+      "step": 3207000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.2575747557740054e-05,
+      "loss": 1.2607,
+      "step": 3207500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.257364759217949e-05,
+      "loss": 1.2864,
+      "step": 3208000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.257155602648116e-05,
+      "loss": 1.2378,
+      "step": 3208500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.25694560609206e-05,
+      "loss": 1.2748,
+      "step": 3209000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.2567356095360034e-05,
+      "loss": 1.3,
+      "step": 3209500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.256525612979947e-05,
+      "loss": 1.2768,
+      "step": 3210000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.256315616423891e-05,
+      "loss": 1.2322,
+      "step": 3210500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.256105619867834e-05,
+      "loss": 1.2621,
+      "step": 3211000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.2558956233117775e-05,
+      "loss": 1.2457,
+      "step": 3211500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.2556860467488335e-05,
+      "loss": 1.2913,
+      "step": 3212000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.255476050192777e-05,
+      "loss": 1.2595,
+      "step": 3212500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.25526605363672e-05,
+      "loss": 1.3004,
+      "step": 3213000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.255056057080664e-05,
+      "loss": 1.2453,
+      "step": 3213500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.2548460605246076e-05,
+      "loss": 1.2538,
+      "step": 3214000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.254636063968551e-05,
+      "loss": 1.2464,
+      "step": 3214500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.254426067412495e-05,
+      "loss": 1.28,
+      "step": 3215000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.2542160708564376e-05,
+      "loss": 1.2426,
+      "step": 3215500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.254006074300381e-05,
+      "loss": 1.2328,
+      "step": 3216000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.253796077744325e-05,
+      "loss": 1.2959,
+      "step": 3216500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.2535860811882684e-05,
+      "loss": 1.2548,
+      "step": 3217000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.253376504625324e-05,
+      "loss": 1.2665,
+      "step": 3217500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.253166508069267e-05,
+      "loss": 1.2709,
+      "step": 3218000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.252956511513211e-05,
+      "loss": 1.2179,
+      "step": 3218500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.2527465149571544e-05,
+      "loss": 1.2469,
+      "step": 3219000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.252536518401098e-05,
+      "loss": 1.2408,
+      "step": 3219500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.252326941838154e-05,
+      "loss": 1.2658,
+      "step": 3220000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.252116945282097e-05,
+      "loss": 1.2749,
+      "step": 3220500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.2519069487260405e-05,
+      "loss": 1.259,
+      "step": 3221000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.2516969521699845e-05,
+      "loss": 1.2758,
+      "step": 3221500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.251486955613928e-05,
+      "loss": 1.2234,
+      "step": 3222000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.251276959057871e-05,
+      "loss": 1.258,
+      "step": 3222500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.251066962501815e-05,
+      "loss": 1.258,
+      "step": 3223000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.2508569659457586e-05,
+      "loss": 1.2398,
+      "step": 3223500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.250647389382814e-05,
+      "loss": 1.228,
+      "step": 3224000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.250437392826757e-05,
+      "loss": 1.2617,
+      "step": 3224500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.250227396270701e-05,
+      "loss": 1.2707,
+      "step": 3225000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.2500173997146447e-05,
+      "loss": 1.2718,
+      "step": 3225500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.2498078231517e-05,
+      "loss": 1.2469,
+      "step": 3226000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.249598246588756e-05,
+      "loss": 1.2453,
+      "step": 3226500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.2493882500326994e-05,
+      "loss": 1.2633,
+      "step": 3227000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.249178253476643e-05,
+      "loss": 1.2703,
+      "step": 3227500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.248968256920586e-05,
+      "loss": 1.2821,
+      "step": 3228000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.24875826036453e-05,
+      "loss": 1.2644,
+      "step": 3228500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.248548683801586e-05,
+      "loss": 1.245,
+      "step": 3229000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.248338687245529e-05,
+      "loss": 1.2813,
+      "step": 3229500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.248128690689472e-05,
+      "loss": 1.241,
+      "step": 3230000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.247918694133416e-05,
+      "loss": 1.2352,
+      "step": 3230500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.2477086975773595e-05,
+      "loss": 1.2492,
+      "step": 3231000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.247498701021303e-05,
+      "loss": 1.2615,
+      "step": 3231500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.247288704465247e-05,
+      "loss": 1.2399,
+      "step": 3232000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.24707870790919e-05,
+      "loss": 1.2631,
+      "step": 3232500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.2468691313462456e-05,
+      "loss": 1.2476,
+      "step": 3233000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.2466595547833016e-05,
+      "loss": 1.2551,
+      "step": 3233500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.246449558227245e-05,
+      "loss": 1.2667,
+      "step": 3234000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.246239561671188e-05,
+      "loss": 1.2636,
+      "step": 3234500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.246029565115132e-05,
+      "loss": 1.219,
+      "step": 3235000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.245819568559076e-05,
+      "loss": 1.2677,
+      "step": 3235500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.245609572003019e-05,
+      "loss": 1.2509,
+      "step": 3236000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.2453995754469624e-05,
+      "loss": 1.2561,
+      "step": 3236500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.2451895788909064e-05,
+      "loss": 1.2597,
+      "step": 3237000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.24497958233485e-05,
+      "loss": 1.2341,
+      "step": 3237500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.244769585778793e-05,
+      "loss": 1.2512,
+      "step": 3238000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.244559589222737e-05,
+      "loss": 1.2507,
+      "step": 3238500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.2443495926666805e-05,
+      "loss": 1.2449,
+      "step": 3239000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.244139596110624e-05,
+      "loss": 1.2726,
+      "step": 3239500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.243930019547679e-05,
+      "loss": 1.2887,
+      "step": 3240000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.243720022991623e-05,
+      "loss": 1.2233,
+      "step": 3240500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.2435100264355666e-05,
+      "loss": 1.2724,
+      "step": 3241000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.24330002987951e-05,
+      "loss": 1.2573,
+      "step": 3241500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.243090453316565e-05,
+      "loss": 1.2553,
+      "step": 3242000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.242880876753621e-05,
+      "loss": 1.2407,
+      "step": 3242500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.2426708801975646e-05,
+      "loss": 1.286,
+      "step": 3243000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.242460883641508e-05,
+      "loss": 1.2374,
+      "step": 3243500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.2422513070785633e-05,
+      "loss": 1.2559,
+      "step": 3244000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.2420413105225074e-05,
+      "loss": 1.2762,
+      "step": 3244500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.241831313966451e-05,
+      "loss": 1.2666,
+      "step": 3245000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.241621317410394e-05,
+      "loss": 1.28,
+      "step": 3245500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.24141174084745e-05,
+      "loss": 1.2758,
+      "step": 3246000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.2412017442913934e-05,
+      "loss": 1.2608,
+      "step": 3246500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.240991747735337e-05,
+      "loss": 1.2418,
+      "step": 3247000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.24078175117928e-05,
+      "loss": 1.2353,
+      "step": 3247500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.240571754623224e-05,
+      "loss": 1.2717,
+      "step": 3248000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.2403617580671675e-05,
+      "loss": 1.2264,
+      "step": 3248500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.240152181504223e-05,
+      "loss": 1.2419,
+      "step": 3249000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.239942184948167e-05,
+      "loss": 1.2698,
+      "step": 3249500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.23973218839211e-05,
+      "loss": 1.2194,
+      "step": 3250000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.2395221918360536e-05,
+      "loss": 1.2654,
+      "step": 3250500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.239312615273109e-05,
+      "loss": 1.2396,
+      "step": 3251000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.239102618717053e-05,
+      "loss": 1.2614,
+      "step": 3251500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.238892622160996e-05,
+      "loss": 1.2914,
+      "step": 3252000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.2386826256049396e-05,
+      "loss": 1.2677,
+      "step": 3252500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.238472629048884e-05,
+      "loss": 1.2343,
+      "step": 3253000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.238262632492827e-05,
+      "loss": 1.2466,
+      "step": 3253500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.2380526359367704e-05,
+      "loss": 1.2603,
+      "step": 3254000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.2378426393807144e-05,
+      "loss": 1.249,
+      "step": 3254500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.237632642824658e-05,
+      "loss": 1.262,
+      "step": 3255000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.237422646268601e-05,
+      "loss": 1.2699,
+      "step": 3255500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.237212649712545e-05,
+      "loss": 1.2444,
+      "step": 3256000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.237002653156488e-05,
+      "loss": 1.2671,
+      "step": 3256500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.236793076593544e-05,
+      "loss": 1.2631,
+      "step": 3257000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.236583080037488e-05,
+      "loss": 1.2807,
+      "step": 3257500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.236373083481431e-05,
+      "loss": 1.2529,
+      "step": 3258000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.2361630869253745e-05,
+      "loss": 1.2813,
+      "step": 3258500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.235953090369318e-05,
+      "loss": 1.2389,
+      "step": 3259000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.235743093813261e-05,
+      "loss": 1.2514,
+      "step": 3259500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.235533517250317e-05,
+      "loss": 1.2191,
+      "step": 3260000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.2353235206942606e-05,
+      "loss": 1.2344,
+      "step": 3260500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.2351135241382046e-05,
+      "loss": 1.2406,
+      "step": 3261000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.234903527582147e-05,
+      "loss": 1.2533,
+      "step": 3261500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.2346935310260906e-05,
+      "loss": 1.2339,
+      "step": 3262000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.2344835344700346e-05,
+      "loss": 1.256,
+      "step": 3262500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.234273537913978e-05,
+      "loss": 1.231,
+      "step": 3263000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.2340635413579213e-05,
+      "loss": 1.2715,
+      "step": 3263500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.2338539647949774e-05,
+      "loss": 1.2553,
+      "step": 3264000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.233643968238921e-05,
+      "loss": 1.2449,
+      "step": 3264500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.233433971682864e-05,
+      "loss": 1.2567,
+      "step": 3265000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.233223975126808e-05,
+      "loss": 1.2715,
+      "step": 3265500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.2330143985638634e-05,
+      "loss": 1.2638,
+      "step": 3266000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.232804402007807e-05,
+      "loss": 1.2496,
+      "step": 3266500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.23259440545175e-05,
+      "loss": 1.2638,
+      "step": 3267000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.232384408895694e-05,
+      "loss": 1.2339,
+      "step": 3267500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.23217483233275e-05,
+      "loss": 1.2569,
+      "step": 3268000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.231964835776693e-05,
+      "loss": 1.2453,
+      "step": 3268500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.231755259213749e-05,
+      "loss": 1.2279,
+      "step": 3269000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.231545262657692e-05,
+      "loss": 1.2554,
+      "step": 3269500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.231335266101636e-05,
+      "loss": 1.2364,
+      "step": 3270000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.2311252695455796e-05,
+      "loss": 1.2518,
+      "step": 3270500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.230915272989523e-05,
+      "loss": 1.2416,
+      "step": 3271000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.230705276433466e-05,
+      "loss": 1.2475,
+      "step": 3271500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.2304956998705223e-05,
+      "loss": 1.2499,
+      "step": 3272000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.230285703314466e-05,
+      "loss": 1.2287,
+      "step": 3272500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.230076126751521e-05,
+      "loss": 1.2467,
+      "step": 3273000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.229866130195465e-05,
+      "loss": 1.2494,
+      "step": 3273500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.2296561336394084e-05,
+      "loss": 1.2477,
+      "step": 3274000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.229446137083352e-05,
+      "loss": 1.257,
+      "step": 3274500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.229236140527296e-05,
+      "loss": 1.2359,
+      "step": 3275000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.229026563964351e-05,
+      "loss": 1.259,
+      "step": 3275500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.2288165674082945e-05,
+      "loss": 1.2504,
+      "step": 3276000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.228606570852238e-05,
+      "loss": 1.2543,
+      "step": 3276500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.228396994289294e-05,
+      "loss": 1.2482,
+      "step": 3277000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.228186997733237e-05,
+      "loss": 1.2365,
+      "step": 3277500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.2279770011771806e-05,
+      "loss": 1.273,
+      "step": 3278000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.2277670046211246e-05,
+      "loss": 1.2801,
+      "step": 3278500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.227557008065068e-05,
+      "loss": 1.2342,
+      "step": 3279000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.227347011509011e-05,
+      "loss": 1.2276,
+      "step": 3279500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.227137014952955e-05,
+      "loss": 1.262,
+      "step": 3280000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.226927018396898e-05,
+      "loss": 1.2404,
+      "step": 3280500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.226717021840841e-05,
+      "loss": 1.2392,
+      "step": 3281000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.2265070252847853e-05,
+      "loss": 1.2491,
+      "step": 3281500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.226297028728729e-05,
+      "loss": 1.2557,
+      "step": 3282000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.226087032172672e-05,
+      "loss": 1.2356,
+      "step": 3282500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.225877035616616e-05,
+      "loss": 1.252,
+      "step": 3283000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.2256670390605594e-05,
+      "loss": 1.272,
+      "step": 3283500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.225457462497615e-05,
+      "loss": 1.2863,
+      "step": 3284000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.225247885934671e-05,
+      "loss": 1.2522,
+      "step": 3284500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.225037889378614e-05,
+      "loss": 1.2599,
+      "step": 3285000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.2248278928225575e-05,
+      "loss": 1.2393,
+      "step": 3285500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.224617896266501e-05,
+      "loss": 1.2691,
+      "step": 3286000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.224407899710445e-05,
+      "loss": 1.2463,
+      "step": 3286500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.224197903154388e-05,
+      "loss": 1.2306,
+      "step": 3287000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.2239879065983315e-05,
+      "loss": 1.2334,
+      "step": 3287500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.2237779100422756e-05,
+      "loss": 1.2817,
+      "step": 3288000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.223567913486219e-05,
+      "loss": 1.261,
+      "step": 3288500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.223357916930162e-05,
+      "loss": 1.2443,
+      "step": 3289000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.2231483403672176e-05,
+      "loss": 1.2694,
+      "step": 3289500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.2229383438111616e-05,
+      "loss": 1.2468,
+      "step": 3290000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.222728347255105e-05,
+      "loss": 1.2669,
+      "step": 3290500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.222518350699048e-05,
+      "loss": 1.2399,
+      "step": 3291000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.2223083541429924e-05,
+      "loss": 1.2915,
+      "step": 3291500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.222098357586936e-05,
+      "loss": 1.2601,
+      "step": 3292000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.221888361030879e-05,
+      "loss": 1.2774,
+      "step": 3292500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.2216783644748224e-05,
+      "loss": 1.2671,
+      "step": 3293000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.2214687879118784e-05,
+      "loss": 1.2529,
+      "step": 3293500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.221259211348934e-05,
+      "loss": 1.2297,
+      "step": 3294000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.221049214792877e-05,
+      "loss": 1.249,
+      "step": 3294500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.220839218236821e-05,
+      "loss": 1.215,
+      "step": 3295000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.2206292216807645e-05,
+      "loss": 1.2759,
+      "step": 3295500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.220419225124708e-05,
+      "loss": 1.2869,
+      "step": 3296000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.220209228568652e-05,
+      "loss": 1.2703,
+      "step": 3296500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.219999652005707e-05,
+      "loss": 1.2599,
+      "step": 3297000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.2197896554496506e-05,
+      "loss": 1.251,
+      "step": 3297500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.219579658893594e-05,
+      "loss": 1.2561,
+      "step": 3298000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.219369662337538e-05,
+      "loss": 1.2464,
+      "step": 3298500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.219160085774593e-05,
+      "loss": 1.2714,
+      "step": 3299000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.218950509211649e-05,
+      "loss": 1.2492,
+      "step": 3299500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.218740512655592e-05,
+      "loss": 1.2767,
+      "step": 3300000
+    },
+    {
+      "epoch": 1.98,
+      "eval_loss": 1.2033525705337524,
+      "eval_runtime": 1102.9314,
+      "eval_samples_per_second": 477.564,
+      "eval_steps_per_second": 79.594,
+      "step": 3300000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.218530516099536e-05,
+      "loss": 1.2708,
+      "step": 3300500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.2183205195434794e-05,
+      "loss": 1.2561,
+      "step": 3301000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.218110522987423e-05,
+      "loss": 1.2472,
+      "step": 3301500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.217900526431367e-05,
+      "loss": 1.2677,
+      "step": 3302000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.21769052987531e-05,
+      "loss": 1.2653,
+      "step": 3302500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.2174805333192534e-05,
+      "loss": 1.2271,
+      "step": 3303000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.217270956756309e-05,
+      "loss": 1.2333,
+      "step": 3303500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.217060960200253e-05,
+      "loss": 1.2417,
+      "step": 3304000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.216850963644196e-05,
+      "loss": 1.2627,
+      "step": 3304500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.2166409670881395e-05,
+      "loss": 1.257,
+      "step": 3305000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.2164309705320835e-05,
+      "loss": 1.2694,
+      "step": 3305500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.216220973976027e-05,
+      "loss": 1.2381,
+      "step": 3306000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.21601097741997e-05,
+      "loss": 1.2477,
+      "step": 3306500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.215800980863914e-05,
+      "loss": 1.2748,
+      "step": 3307000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.215590984307857e-05,
+      "loss": 1.2409,
+      "step": 3307500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.215381407744913e-05,
+      "loss": 1.2652,
+      "step": 3308000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.215171411188856e-05,
+      "loss": 1.2544,
+      "step": 3308500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.2149614146328e-05,
+      "loss": 1.2546,
+      "step": 3309000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.214751418076744e-05,
+      "loss": 1.2503,
+      "step": 3309500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.214541421520687e-05,
+      "loss": 1.2463,
+      "step": 3310000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.2143314249646304e-05,
+      "loss": 1.2593,
+      "step": 3310500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.214121428408574e-05,
+      "loss": 1.2696,
+      "step": 3311000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.213911431852518e-05,
+      "loss": 1.2522,
+      "step": 3311500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.213701855289573e-05,
+      "loss": 1.2546,
+      "step": 3312000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.213492278726629e-05,
+      "loss": 1.2457,
+      "step": 3312500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.2132822821705725e-05,
+      "loss": 1.2387,
+      "step": 3313000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.213072285614516e-05,
+      "loss": 1.2456,
+      "step": 3313500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.21286228905846e-05,
+      "loss": 1.2579,
+      "step": 3314000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.2126522925024025e-05,
+      "loss": 1.2592,
+      "step": 3314500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.212442295946346e-05,
+      "loss": 1.2537,
+      "step": 3315000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.212233139376514e-05,
+      "loss": 1.2857,
+      "step": 3315500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.212023142820458e-05,
+      "loss": 1.2259,
+      "step": 3316000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.211813146264401e-05,
+      "loss": 1.27,
+      "step": 3316500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.2116031497083446e-05,
+      "loss": 1.2551,
+      "step": 3317000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.2113931531522886e-05,
+      "loss": 1.2545,
+      "step": 3317500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.211183156596232e-05,
+      "loss": 1.2653,
+      "step": 3318000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.210973160040175e-05,
+      "loss": 1.2679,
+      "step": 3318500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.2107631634841194e-05,
+      "loss": 1.2723,
+      "step": 3319000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.210553166928062e-05,
+      "loss": 1.2721,
+      "step": 3319500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.2103431703720054e-05,
+      "loss": 1.2557,
+      "step": 3320000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.2101331738159494e-05,
+      "loss": 1.2296,
+      "step": 3320500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.209923177259893e-05,
+      "loss": 1.281,
+      "step": 3321000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.209713600696948e-05,
+      "loss": 1.2263,
+      "step": 3321500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.209503604140892e-05,
+      "loss": 1.2488,
+      "step": 3322000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.2092936075848355e-05,
+      "loss": 1.2555,
+      "step": 3322500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.209083611028779e-05,
+      "loss": 1.238,
+      "step": 3323000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.208874034465835e-05,
+      "loss": 1.2498,
+      "step": 3323500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.208664037909778e-05,
+      "loss": 1.2839,
+      "step": 3324000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.2084540413537215e-05,
+      "loss": 1.2234,
+      "step": 3324500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.2082453047770016e-05,
+      "loss": 1.2577,
+      "step": 3325000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.208035308220945e-05,
+      "loss": 1.2438,
+      "step": 3325500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.207825311664888e-05,
+      "loss": 1.2413,
+      "step": 3326000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.2076153151088316e-05,
+      "loss": 1.2751,
+      "step": 3326500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.2074053185527757e-05,
+      "loss": 1.2194,
+      "step": 3327000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.207195321996719e-05,
+      "loss": 1.2353,
+      "step": 3327500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.2069853254406623e-05,
+      "loss": 1.2602,
+      "step": 3328000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.2067753288846064e-05,
+      "loss": 1.236,
+      "step": 3328500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.20656533232855e-05,
+      "loss": 1.2609,
+      "step": 3329000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.206355335772494e-05,
+      "loss": 1.2641,
+      "step": 3329500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.206145339216437e-05,
+      "loss": 1.2569,
+      "step": 3330000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.2059353426603804e-05,
+      "loss": 1.2481,
+      "step": 3330500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.205725346104324e-05,
+      "loss": 1.2419,
+      "step": 3331000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.205515349548267e-05,
+      "loss": 1.2608,
+      "step": 3331500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.2053053529922105e-05,
+      "loss": 1.2305,
+      "step": 3332000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.2050953564361545e-05,
+      "loss": 1.2409,
+      "step": 3332500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.2048857798732105e-05,
+      "loss": 1.2405,
+      "step": 3333000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.204676203310266e-05,
+      "loss": 1.2761,
+      "step": 3333500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.204466206754209e-05,
+      "loss": 1.2441,
+      "step": 3334000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.2042562101981526e-05,
+      "loss": 1.2889,
+      "step": 3334500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.2040462136420966e-05,
+      "loss": 1.2861,
+      "step": 3335000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.20383621708604e-05,
+      "loss": 1.2382,
+      "step": 3335500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.203627060516207e-05,
+      "loss": 1.2626,
+      "step": 3336000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.203417063960151e-05,
+      "loss": 1.2198,
+      "step": 3336500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.203207067404095e-05,
+      "loss": 1.2162,
+      "step": 3337000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.202997070848038e-05,
+      "loss": 1.2305,
+      "step": 3337500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.2027870742919814e-05,
+      "loss": 1.2166,
+      "step": 3338000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.2025770777359254e-05,
+      "loss": 1.2541,
+      "step": 3338500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.202367081179869e-05,
+      "loss": 1.2032,
+      "step": 3339000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.202157084623812e-05,
+      "loss": 1.209,
+      "step": 3339500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.201947088067756e-05,
+      "loss": 1.1905,
+      "step": 3340000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.2017370915116995e-05,
+      "loss": 1.2406,
+      "step": 3340500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.201527094955642e-05,
+      "loss": 1.1992,
+      "step": 3341000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.201317518392698e-05,
+      "loss": 1.2321,
+      "step": 3341500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.201107521836642e-05,
+      "loss": 1.2198,
+      "step": 3342000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.2008975252805855e-05,
+      "loss": 1.2011,
+      "step": 3342500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.200687528724529e-05,
+      "loss": 1.2217,
+      "step": 3343000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.200477952161585e-05,
+      "loss": 1.2124,
+      "step": 3343500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.200267955605528e-05,
+      "loss": 1.2352,
+      "step": 3344000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.2000579590494716e-05,
+      "loss": 1.2127,
+      "step": 3344500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.1998479624934156e-05,
+      "loss": 1.22,
+      "step": 3345000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.199638385930471e-05,
+      "loss": 1.237,
+      "step": 3345500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.199428389374414e-05,
+      "loss": 1.2126,
+      "step": 3346000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.199218392818358e-05,
+      "loss": 1.2094,
+      "step": 3346500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.199008396262302e-05,
+      "loss": 1.231,
+      "step": 3347000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.198798399706245e-05,
+      "loss": 1.2546,
+      "step": 3347500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.1985888231433004e-05,
+      "loss": 1.2435,
+      "step": 3348000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.198378826587244e-05,
+      "loss": 1.2206,
+      "step": 3348500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.1981692500243e-05,
+      "loss": 1.1801,
+      "step": 3349000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.197959253468243e-05,
+      "loss": 1.2382,
+      "step": 3349500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.1977492569121865e-05,
+      "loss": 1.2125,
+      "step": 3350000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.1975392603561305e-05,
+      "loss": 1.2069,
+      "step": 3350500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.197329263800074e-05,
+      "loss": 1.1793,
+      "step": 3351000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.197119267244017e-05,
+      "loss": 1.2416,
+      "step": 3351500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.196909270687961e-05,
+      "loss": 1.2327,
+      "step": 3352000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.196699274131904e-05,
+      "loss": 1.2248,
+      "step": 3352500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.196489277575847e-05,
+      "loss": 1.2351,
+      "step": 3353000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.196279281019791e-05,
+      "loss": 1.2355,
+      "step": 3353500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.1960692844637346e-05,
+      "loss": 1.1975,
+      "step": 3354000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.195859287907678e-05,
+      "loss": 1.2151,
+      "step": 3354500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.195650131337846e-05,
+      "loss": 1.2216,
+      "step": 3355000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.1954401347817893e-05,
+      "loss": 1.2075,
+      "step": 3355500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.1952301382257334e-05,
+      "loss": 1.2321,
+      "step": 3356000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.195020141669677e-05,
+      "loss": 1.2194,
+      "step": 3356500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.19481014511362e-05,
+      "loss": 1.2372,
+      "step": 3357000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.1946001485575634e-05,
+      "loss": 1.2411,
+      "step": 3357500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.194390152001507e-05,
+      "loss": 1.2438,
+      "step": 3358000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.194180155445451e-05,
+      "loss": 1.2259,
+      "step": 3358500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.193970578882507e-05,
+      "loss": 1.2535,
+      "step": 3359000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.19376058232645e-05,
+      "loss": 1.2145,
+      "step": 3359500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.1935510057635055e-05,
+      "loss": 1.2192,
+      "step": 3360000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.193341009207449e-05,
+      "loss": 1.2145,
+      "step": 3360500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.193131012651393e-05,
+      "loss": 1.2104,
+      "step": 3361000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.192921016095336e-05,
+      "loss": 1.2195,
+      "step": 3361500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.1927114395323916e-05,
+      "loss": 1.1939,
+      "step": 3362000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.192501442976335e-05,
+      "loss": 1.2449,
+      "step": 3362500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.192291446420279e-05,
+      "loss": 1.2233,
+      "step": 3363000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.192081449864222e-05,
+      "loss": 1.2038,
+      "step": 3363500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.1918714533081656e-05,
+      "loss": 1.2076,
+      "step": 3364000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.191661456752109e-05,
+      "loss": 1.2652,
+      "step": 3364500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.191451460196052e-05,
+      "loss": 1.219,
+      "step": 3365000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.1912414636399964e-05,
+      "loss": 1.1913,
+      "step": 3365500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.19103146708394e-05,
+      "loss": 1.2265,
+      "step": 3366000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.190821470527883e-05,
+      "loss": 1.2151,
+      "step": 3366500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.190611473971827e-05,
+      "loss": 1.2252,
+      "step": 3367000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.1904014774157704e-05,
+      "loss": 1.2086,
+      "step": 3367500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.1901923208459385e-05,
+      "loss": 1.215,
+      "step": 3368000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.189982324289882e-05,
+      "loss": 1.2167,
+      "step": 3368500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.189772327733825e-05,
+      "loss": 1.2496,
+      "step": 3369000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.1895623311777685e-05,
+      "loss": 1.2426,
+      "step": 3369500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.189352334621712e-05,
+      "loss": 1.2192,
+      "step": 3370000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.189142338065655e-05,
+      "loss": 1.2157,
+      "step": 3370500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.188932341509599e-05,
+      "loss": 1.2326,
+      "step": 3371000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.1887227649466546e-05,
+      "loss": 1.2165,
+      "step": 3371500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.188512768390598e-05,
+      "loss": 1.2458,
+      "step": 3372000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.188302771834542e-05,
+      "loss": 1.2398,
+      "step": 3372500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.188092775278485e-05,
+      "loss": 1.2033,
+      "step": 3373000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.1878827787224286e-05,
+      "loss": 1.2292,
+      "step": 3373500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.187673202159484e-05,
+      "loss": 1.2279,
+      "step": 3374000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.187463205603428e-05,
+      "loss": 1.2338,
+      "step": 3374500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.1872532090473714e-05,
+      "loss": 1.1969,
+      "step": 3375000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.187043212491315e-05,
+      "loss": 1.1877,
+      "step": 3375500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.186833215935259e-05,
+      "loss": 1.2651,
+      "step": 3376000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.186623639372314e-05,
+      "loss": 1.2203,
+      "step": 3376500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.18641406280937e-05,
+      "loss": 1.2326,
+      "step": 3377000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.1862040662533135e-05,
+      "loss": 1.226,
+      "step": 3377500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.185994069697257e-05,
+      "loss": 1.2216,
+      "step": 3378000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.185784073141201e-05,
+      "loss": 1.2174,
+      "step": 3378500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.1855740765851435e-05,
+      "loss": 1.2111,
+      "step": 3379000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.1853640800290875e-05,
+      "loss": 1.2416,
+      "step": 3379500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.185154083473031e-05,
+      "loss": 1.2204,
+      "step": 3380000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.184944086916974e-05,
+      "loss": 1.2264,
+      "step": 3380500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.184734090360918e-05,
+      "loss": 1.224,
+      "step": 3381000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.1845245137979736e-05,
+      "loss": 1.199,
+      "step": 3381500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.184314517241917e-05,
+      "loss": 1.2083,
+      "step": 3382000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.18410452068586e-05,
+      "loss": 1.2069,
+      "step": 3382500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.183894524129804e-05,
+      "loss": 1.2393,
+      "step": 3383000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.183684527573748e-05,
+      "loss": 1.2264,
+      "step": 3383500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.183474531017691e-05,
+      "loss": 1.2552,
+      "step": 3384000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.1832649544547464e-05,
+      "loss": 1.2178,
+      "step": 3384500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.1830549578986904e-05,
+      "loss": 1.2385,
+      "step": 3385000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.182844961342634e-05,
+      "loss": 1.2373,
+      "step": 3385500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.182634964786577e-05,
+      "loss": 1.198,
+      "step": 3386000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.182424968230521e-05,
+      "loss": 1.2394,
+      "step": 3386500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.1822149716744645e-05,
+      "loss": 1.2254,
+      "step": 3387000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.1820049751184085e-05,
+      "loss": 1.2216,
+      "step": 3387500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.181794978562352e-05,
+      "loss": 1.2435,
+      "step": 3388000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.181585401999407e-05,
+      "loss": 1.1905,
+      "step": 3388500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.1813758254364625e-05,
+      "loss": 1.2216,
+      "step": 3389000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.181165828880406e-05,
+      "loss": 1.2113,
+      "step": 3389500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.18095583232435e-05,
+      "loss": 1.2289,
+      "step": 3390000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.180745835768293e-05,
+      "loss": 1.2403,
+      "step": 3390500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.1805358392122366e-05,
+      "loss": 1.2307,
+      "step": 3391000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.1803258426561806e-05,
+      "loss": 1.2269,
+      "step": 3391500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.180115846100124e-05,
+      "loss": 1.2084,
+      "step": 3392000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.179905849544067e-05,
+      "loss": 1.2356,
+      "step": 3392500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.179696272981123e-05,
+      "loss": 1.2204,
+      "step": 3393000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.179486276425067e-05,
+      "loss": 1.2372,
+      "step": 3393500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.179276699862122e-05,
+      "loss": 1.2304,
+      "step": 3394000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.1790667033060654e-05,
+      "loss": 1.2305,
+      "step": 3394500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.1788567067500094e-05,
+      "loss": 1.2504,
+      "step": 3395000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.178646710193953e-05,
+      "loss": 1.2269,
+      "step": 3395500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.178437133631008e-05,
+      "loss": 1.2155,
+      "step": 3396000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.1782271370749515e-05,
+      "loss": 1.2351,
+      "step": 3396500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.1780171405188955e-05,
+      "loss": 1.2375,
+      "step": 3397000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.177807143962839e-05,
+      "loss": 1.2498,
+      "step": 3397500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.177597567399894e-05,
+      "loss": 1.2327,
+      "step": 3398000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.1773875708438375e-05,
+      "loss": 1.2327,
+      "step": 3398500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.1771775742877816e-05,
+      "loss": 1.2099,
+      "step": 3399000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.176967577731725e-05,
+      "loss": 1.232,
+      "step": 3399500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.176757581175669e-05,
+      "loss": 1.2407,
+      "step": 3400000
+    },
+    {
+      "epoch": 2.04,
+      "eval_loss": 1.1995346546173096,
+      "eval_runtime": 1104.2629,
+      "eval_samples_per_second": 476.988,
+      "eval_steps_per_second": 79.498,
+      "step": 3400000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.176547584619612e-05,
+      "loss": 1.2394,
+      "step": 3400500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.1763375880635556e-05,
+      "loss": 1.2466,
+      "step": 3401000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.1761275915074997e-05,
+      "loss": 1.2169,
+      "step": 3401500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.175917594951443e-05,
+      "loss": 1.2255,
+      "step": 3402000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.1757075983953863e-05,
+      "loss": 1.2384,
+      "step": 3402500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.175498441825554e-05,
+      "loss": 1.2456,
+      "step": 3403000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.17528886526261e-05,
+      "loss": 1.248,
+      "step": 3403500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.175078868706553e-05,
+      "loss": 1.2369,
+      "step": 3404000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.174868872150497e-05,
+      "loss": 1.2359,
+      "step": 3404500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.17465887559444e-05,
+      "loss": 1.2531,
+      "step": 3405000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.174448879038383e-05,
+      "loss": 1.2493,
+      "step": 3405500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.174238882482327e-05,
+      "loss": 1.2164,
+      "step": 3406000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.1740288859262705e-05,
+      "loss": 1.2284,
+      "step": 3406500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.1738188893702145e-05,
+      "loss": 1.2662,
+      "step": 3407000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.173608892814158e-05,
+      "loss": 1.2377,
+      "step": 3407500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.173398896258101e-05,
+      "loss": 1.223,
+      "step": 3408000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.173188899702045e-05,
+      "loss": 1.2476,
+      "step": 3408500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.1729789031459886e-05,
+      "loss": 1.2279,
+      "step": 3409000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.172768906589932e-05,
+      "loss": 1.2808,
+      "step": 3409500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.172558910033876e-05,
+      "loss": 1.2192,
+      "step": 3410000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.172349333470931e-05,
+      "loss": 1.2598,
+      "step": 3410500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.172139336914875e-05,
+      "loss": 1.2182,
+      "step": 3411000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.17192976035193e-05,
+      "loss": 1.2103,
+      "step": 3411500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.1717197637958734e-05,
+      "loss": 1.2197,
+      "step": 3412000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.1715097672398174e-05,
+      "loss": 1.2106,
+      "step": 3412500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.171299770683761e-05,
+      "loss": 1.2403,
+      "step": 3413000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.171090194120816e-05,
+      "loss": 1.2443,
+      "step": 3413500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.17088019756476e-05,
+      "loss": 1.1998,
+      "step": 3414000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.1706706210018155e-05,
+      "loss": 1.231,
+      "step": 3414500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.170460624445759e-05,
+      "loss": 1.2469,
+      "step": 3415000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.170250627889702e-05,
+      "loss": 1.2463,
+      "step": 3415500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.170040631333646e-05,
+      "loss": 1.2177,
+      "step": 3416000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.1698306347775895e-05,
+      "loss": 1.2238,
+      "step": 3416500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.169620638221533e-05,
+      "loss": 1.2112,
+      "step": 3417000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.169410641665477e-05,
+      "loss": 1.2427,
+      "step": 3417500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.16920064510942e-05,
+      "loss": 1.2545,
+      "step": 3418000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.1689906485533636e-05,
+      "loss": 1.2304,
+      "step": 3418500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.1687806519973076e-05,
+      "loss": 1.2084,
+      "step": 3419000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.168570655441251e-05,
+      "loss": 1.2293,
+      "step": 3419500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.1683606588851936e-05,
+      "loss": 1.222,
+      "step": 3420000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.1681506623291377e-05,
+      "loss": 1.2257,
+      "step": 3420500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.167940665773081e-05,
+      "loss": 1.1888,
+      "step": 3421000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.1677306692170243e-05,
+      "loss": 1.2141,
+      "step": 3421500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.1675210926540804e-05,
+      "loss": 1.2359,
+      "step": 3422000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.167311096098024e-05,
+      "loss": 1.2536,
+      "step": 3422500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.167101099541967e-05,
+      "loss": 1.2324,
+      "step": 3423000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.166891102985911e-05,
+      "loss": 1.222,
+      "step": 3423500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.1666811064298544e-05,
+      "loss": 1.2551,
+      "step": 3424000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.166471109873798e-05,
+      "loss": 1.2209,
+      "step": 3424500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.166261113317742e-05,
+      "loss": 1.2397,
+      "step": 3425000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.166051956747909e-05,
+      "loss": 1.2731,
+      "step": 3425500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.165841960191853e-05,
+      "loss": 1.2385,
+      "step": 3426000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.1656319636357966e-05,
+      "loss": 1.242,
+      "step": 3426500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.16542196707974e-05,
+      "loss": 1.2372,
+      "step": 3427000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.165211970523683e-05,
+      "loss": 1.1994,
+      "step": 3427500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.1650019739676266e-05,
+      "loss": 1.2346,
+      "step": 3428000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.16479197741157e-05,
+      "loss": 1.2372,
+      "step": 3428500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.164581980855514e-05,
+      "loss": 1.2293,
+      "step": 3429000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.164371984299457e-05,
+      "loss": 1.2457,
+      "step": 3429500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.1641619877434007e-05,
+      "loss": 1.2465,
+      "step": 3430000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.163951991187345e-05,
+      "loss": 1.2104,
+      "step": 3430500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.163741994631288e-05,
+      "loss": 1.2047,
+      "step": 3431000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.1635324180683434e-05,
+      "loss": 1.2,
+      "step": 3431500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.1633224215122874e-05,
+      "loss": 1.2296,
+      "step": 3432000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.163112424956231e-05,
+      "loss": 1.2321,
+      "step": 3432500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.162902428400174e-05,
+      "loss": 1.2377,
+      "step": 3433000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.162692431844118e-05,
+      "loss": 1.228,
+      "step": 3433500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.1624828552811735e-05,
+      "loss": 1.2278,
+      "step": 3434000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.162272858725117e-05,
+      "loss": 1.2536,
+      "step": 3434500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.16206286216906e-05,
+      "loss": 1.2067,
+      "step": 3435000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.161852865613004e-05,
+      "loss": 1.2149,
+      "step": 3435500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.1616428690569475e-05,
+      "loss": 1.2283,
+      "step": 3436000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.161433292494003e-05,
+      "loss": 1.2661,
+      "step": 3436500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.161223295937946e-05,
+      "loss": 1.2302,
+      "step": 3437000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.16101329938189e-05,
+      "loss": 1.2363,
+      "step": 3437500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.1608033028258336e-05,
+      "loss": 1.2443,
+      "step": 3438000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.160593306269777e-05,
+      "loss": 1.2265,
+      "step": 3438500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.160383729706833e-05,
+      "loss": 1.2134,
+      "step": 3439000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.160173733150776e-05,
+      "loss": 1.2167,
+      "step": 3439500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.15996373659472e-05,
+      "loss": 1.2021,
+      "step": 3440000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.159753740038664e-05,
+      "loss": 1.2388,
+      "step": 3440500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.159544163475719e-05,
+      "loss": 1.2328,
+      "step": 3441000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.1593341669196624e-05,
+      "loss": 1.2251,
+      "step": 3441500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.159124170363606e-05,
+      "loss": 1.2456,
+      "step": 3442000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.15891417380755e-05,
+      "loss": 1.2272,
+      "step": 3442500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.158704597244605e-05,
+      "loss": 1.1844,
+      "step": 3443000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.1584946006885485e-05,
+      "loss": 1.2297,
+      "step": 3443500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.158284604132492e-05,
+      "loss": 1.2249,
+      "step": 3444000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.158075027569548e-05,
+      "loss": 1.254,
+      "step": 3444500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.157865031013491e-05,
+      "loss": 1.2145,
+      "step": 3445000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.1576550344574346e-05,
+      "loss": 1.2321,
+      "step": 3445500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.1574450379013786e-05,
+      "loss": 1.2322,
+      "step": 3446000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.157235041345322e-05,
+      "loss": 1.2327,
+      "step": 3446500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.157025044789265e-05,
+      "loss": 1.2095,
+      "step": 3447000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.156815048233209e-05,
+      "loss": 1.2214,
+      "step": 3447500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.1566050516771526e-05,
+      "loss": 1.262,
+      "step": 3448000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.156395475114208e-05,
+      "loss": 1.2392,
+      "step": 3448500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.1561854785581513e-05,
+      "loss": 1.1848,
+      "step": 3449000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.1559754820020954e-05,
+      "loss": 1.2267,
+      "step": 3449500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.155765485446039e-05,
+      "loss": 1.2404,
+      "step": 3450000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.155555488889982e-05,
+      "loss": 1.221,
+      "step": 3450500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.1553459123270374e-05,
+      "loss": 1.2428,
+      "step": 3451000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.1551359157709814e-05,
+      "loss": 1.2228,
+      "step": 3451500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.154925919214925e-05,
+      "loss": 1.2029,
+      "step": 3452000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.154715922658868e-05,
+      "loss": 1.2183,
+      "step": 3452500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.154505926102812e-05,
+      "loss": 1.2054,
+      "step": 3453000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.1542959295467555e-05,
+      "loss": 1.2406,
+      "step": 3453500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.154086352983811e-05,
+      "loss": 1.2163,
+      "step": 3454000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.153876356427755e-05,
+      "loss": 1.2391,
+      "step": 3454500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.153666359871698e-05,
+      "loss": 1.2256,
+      "step": 3455000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.1534563633156416e-05,
+      "loss": 1.2156,
+      "step": 3455500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.1532463667595856e-05,
+      "loss": 1.2426,
+      "step": 3456000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.153036370203528e-05,
+      "loss": 1.2507,
+      "step": 3456500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.1528263736474716e-05,
+      "loss": 1.2142,
+      "step": 3457000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.1526163770914156e-05,
+      "loss": 1.2403,
+      "step": 3457500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.152406800528472e-05,
+      "loss": 1.2124,
+      "step": 3458000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.152196803972415e-05,
+      "loss": 1.2475,
+      "step": 3458500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.1519872274094704e-05,
+      "loss": 1.2159,
+      "step": 3459000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.1517772308534144e-05,
+      "loss": 1.2058,
+      "step": 3459500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.151567234297358e-05,
+      "loss": 1.2364,
+      "step": 3460000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.151357237741301e-05,
+      "loss": 1.1895,
+      "step": 3460500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.151147241185245e-05,
+      "loss": 1.2301,
+      "step": 3461000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.150937244629188e-05,
+      "loss": 1.2055,
+      "step": 3461500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.150727248073131e-05,
+      "loss": 1.2452,
+      "step": 3462000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.150517251517075e-05,
+      "loss": 1.2129,
+      "step": 3462500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.1503080949472425e-05,
+      "loss": 1.2424,
+      "step": 3463000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.1500980983911865e-05,
+      "loss": 1.2232,
+      "step": 3463500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.14988810183513e-05,
+      "loss": 1.2265,
+      "step": 3464000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.149678105279073e-05,
+      "loss": 1.2374,
+      "step": 3464500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.149468948709241e-05,
+      "loss": 1.2224,
+      "step": 3465000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.1492589521531846e-05,
+      "loss": 1.2308,
+      "step": 3465500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.149048955597128e-05,
+      "loss": 1.2467,
+      "step": 3466000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.148838959041071e-05,
+      "loss": 1.2269,
+      "step": 3466500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.1486289624850153e-05,
+      "loss": 1.2319,
+      "step": 3467000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.148418965928959e-05,
+      "loss": 1.222,
+      "step": 3467500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.148208969372902e-05,
+      "loss": 1.231,
+      "step": 3468000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.147998972816846e-05,
+      "loss": 1.2495,
+      "step": 3468500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.1477889762607894e-05,
+      "loss": 1.2213,
+      "step": 3469000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.147578979704733e-05,
+      "loss": 1.235,
+      "step": 3469500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.147368983148677e-05,
+      "loss": 1.2433,
+      "step": 3470000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.14715898659262e-05,
+      "loss": 1.2384,
+      "step": 3470500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.146948990036563e-05,
+      "loss": 1.2304,
+      "step": 3471000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.146738993480507e-05,
+      "loss": 1.2309,
+      "step": 3471500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.146529416917563e-05,
+      "loss": 1.2203,
+      "step": 3472000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.146319420361506e-05,
+      "loss": 1.2341,
+      "step": 3472500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.1461094238054495e-05,
+      "loss": 1.2289,
+      "step": 3473000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.145899427249393e-05,
+      "loss": 1.2271,
+      "step": 3473500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.145689430693336e-05,
+      "loss": 1.2185,
+      "step": 3474000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.14547943413728e-05,
+      "loss": 1.2475,
+      "step": 3474500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.1452694375812236e-05,
+      "loss": 1.2306,
+      "step": 3475000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.145059441025167e-05,
+      "loss": 1.2225,
+      "step": 3475500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.144849864462222e-05,
+      "loss": 1.2461,
+      "step": 3476000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.144639867906166e-05,
+      "loss": 1.244,
+      "step": 3476500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.1444302913432224e-05,
+      "loss": 1.2447,
+      "step": 3477000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.144220294787166e-05,
+      "loss": 1.2357,
+      "step": 3477500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.1440102982311084e-05,
+      "loss": 1.2141,
+      "step": 3478000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.1438003016750524e-05,
+      "loss": 1.2331,
+      "step": 3478500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.143590305118996e-05,
+      "loss": 1.2179,
+      "step": 3479000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.143380728556052e-05,
+      "loss": 1.234,
+      "step": 3479500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.143170731999995e-05,
+      "loss": 1.2484,
+      "step": 3480000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.1429607354439385e-05,
+      "loss": 1.2231,
+      "step": 3480500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.142750738887882e-05,
+      "loss": 1.228,
+      "step": 3481000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.142540742331826e-05,
+      "loss": 1.2168,
+      "step": 3481500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.142330745775769e-05,
+      "loss": 1.2402,
+      "step": 3482000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.1421207492197125e-05,
+      "loss": 1.2272,
+      "step": 3482500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.1419107526636566e-05,
+      "loss": 1.2321,
+      "step": 3483000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.1417007561076e-05,
+      "loss": 1.2245,
+      "step": 3483500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.141490759551543e-05,
+      "loss": 1.227,
+      "step": 3484000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.141280762995487e-05,
+      "loss": 1.244,
+      "step": 3484500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.1410707664394306e-05,
+      "loss": 1.2173,
+      "step": 3485000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.140861189876486e-05,
+      "loss": 1.2061,
+      "step": 3485500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.140651613313541e-05,
+      "loss": 1.235,
+      "step": 3486000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.140441616757485e-05,
+      "loss": 1.2211,
+      "step": 3486500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.140231620201429e-05,
+      "loss": 1.2069,
+      "step": 3487000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.140021623645372e-05,
+      "loss": 1.2444,
+      "step": 3487500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.1398120470824274e-05,
+      "loss": 1.2429,
+      "step": 3488000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.1396020505263714e-05,
+      "loss": 1.2227,
+      "step": 3488500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.139392053970315e-05,
+      "loss": 1.2269,
+      "step": 3489000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.139182057414258e-05,
+      "loss": 1.2472,
+      "step": 3489500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.138972060858202e-05,
+      "loss": 1.2061,
+      "step": 3490000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.1387620643021455e-05,
+      "loss": 1.24,
+      "step": 3490500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.138552487739201e-05,
+      "loss": 1.2378,
+      "step": 3491000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.138342491183144e-05,
+      "loss": 1.2219,
+      "step": 3491500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.138132494627088e-05,
+      "loss": 1.2121,
+      "step": 3492000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.1379224980710316e-05,
+      "loss": 1.2498,
+      "step": 3492500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.137712501514975e-05,
+      "loss": 1.2114,
+      "step": 3493000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.137502504958919e-05,
+      "loss": 1.1985,
+      "step": 3493500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.137293348389086e-05,
+      "loss": 1.2494,
+      "step": 3494000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.13708335183303e-05,
+      "loss": 1.2352,
+      "step": 3494500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.136873355276973e-05,
+      "loss": 1.2027,
+      "step": 3495000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.136663358720917e-05,
+      "loss": 1.2658,
+      "step": 3495500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.1364533621648604e-05,
+      "loss": 1.2395,
+      "step": 3496000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.136243365608804e-05,
+      "loss": 1.2204,
+      "step": 3496500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.136033369052748e-05,
+      "loss": 1.2042,
+      "step": 3497000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.135823372496691e-05,
+      "loss": 1.2156,
+      "step": 3497500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.1356133759406344e-05,
+      "loss": 1.2516,
+      "step": 3498000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.13540379937769e-05,
+      "loss": 1.2226,
+      "step": 3498500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.135193802821634e-05,
+      "loss": 1.2301,
+      "step": 3499000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.134983806265577e-05,
+      "loss": 1.2171,
+      "step": 3499500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.1347738097095205e-05,
+      "loss": 1.2252,
+      "step": 3500000
+    },
+    {
+      "epoch": 2.1,
+      "eval_loss": 1.1996427774429321,
+      "eval_runtime": 1100.4525,
+      "eval_samples_per_second": 478.639,
+      "eval_steps_per_second": 79.774,
+      "step": 3500000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.1345638131534645e-05,
+      "loss": 1.248,
+      "step": 3500500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.13435423659052e-05,
+      "loss": 1.235,
+      "step": 3501000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.134144240034463e-05,
+      "loss": 1.2172,
+      "step": 3501500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.1339342434784066e-05,
+      "loss": 1.2192,
+      "step": 3502000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.1337242469223506e-05,
+      "loss": 1.2215,
+      "step": 3502500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.133514250366294e-05,
+      "loss": 1.2154,
+      "step": 3503000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.133304253810237e-05,
+      "loss": 1.255,
+      "step": 3503500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.133094677247293e-05,
+      "loss": 1.2364,
+      "step": 3504000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.132884680691237e-05,
+      "loss": 1.2171,
+      "step": 3504500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.13267468413518e-05,
+      "loss": 1.2397,
+      "step": 3505000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.132464687579124e-05,
+      "loss": 1.226,
+      "step": 3505500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.1322546910230674e-05,
+      "loss": 1.2268,
+      "step": 3506000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.132045114460123e-05,
+      "loss": 1.2206,
+      "step": 3506500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.131835537897178e-05,
+      "loss": 1.2344,
+      "step": 3507000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.1316255413411214e-05,
+      "loss": 1.2253,
+      "step": 3507500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.1314155447850655e-05,
+      "loss": 1.2526,
+      "step": 3508000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.131205548229009e-05,
+      "loss": 1.2111,
+      "step": 3508500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.130995551672952e-05,
+      "loss": 1.2455,
+      "step": 3509000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.130785555116896e-05,
+      "loss": 1.2818,
+      "step": 3509500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.1305755585608395e-05,
+      "loss": 1.2294,
+      "step": 3510000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.130365562004783e-05,
+      "loss": 1.1845,
+      "step": 3510500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.130155565448727e-05,
+      "loss": 1.2215,
+      "step": 3511000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.12994556889267e-05,
+      "loss": 1.2234,
+      "step": 3511500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.1297355723366136e-05,
+      "loss": 1.2079,
+      "step": 3512000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.1295259957736696e-05,
+      "loss": 1.2202,
+      "step": 3512500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.129315999217613e-05,
+      "loss": 1.2175,
+      "step": 3513000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.129106002661556e-05,
+      "loss": 1.2392,
+      "step": 3513500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.1288960061055e-05,
+      "loss": 1.2231,
+      "step": 3514000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.128686009549443e-05,
+      "loss": 1.2203,
+      "step": 3514500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.1284760129933864e-05,
+      "loss": 1.2551,
+      "step": 3515000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.1282660164373304e-05,
+      "loss": 1.2142,
+      "step": 3515500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.128056019881274e-05,
+      "loss": 1.2432,
+      "step": 3516000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.127846863311442e-05,
+      "loss": 1.2502,
+      "step": 3516500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.127636866755385e-05,
+      "loss": 1.2369,
+      "step": 3517000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.1274268701993285e-05,
+      "loss": 1.2499,
+      "step": 3517500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.1272168736432725e-05,
+      "loss": 1.2356,
+      "step": 3518000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.127006877087216e-05,
+      "loss": 1.2287,
+      "step": 3518500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.12679688053116e-05,
+      "loss": 1.2071,
+      "step": 3519000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.126587303968215e-05,
+      "loss": 1.2358,
+      "step": 3519500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.1263773074121586e-05,
+      "loss": 1.2269,
+      "step": 3520000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.126167310856102e-05,
+      "loss": 1.2406,
+      "step": 3520500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.125957314300046e-05,
+      "loss": 1.2383,
+      "step": 3521000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.125747317743989e-05,
+      "loss": 1.2097,
+      "step": 3521500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.1255377411810446e-05,
+      "loss": 1.2258,
+      "step": 3522000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.125327744624988e-05,
+      "loss": 1.2323,
+      "step": 3522500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.125117748068932e-05,
+      "loss": 1.2419,
+      "step": 3523000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.1249077515128753e-05,
+      "loss": 1.2142,
+      "step": 3523500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.124698174949931e-05,
+      "loss": 1.2518,
+      "step": 3524000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.124488178393875e-05,
+      "loss": 1.2314,
+      "step": 3524500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.124278181837818e-05,
+      "loss": 1.2163,
+      "step": 3525000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.1240681852817614e-05,
+      "loss": 1.2036,
+      "step": 3525500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.1238581887257054e-05,
+      "loss": 1.2061,
+      "step": 3526000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.123648192169648e-05,
+      "loss": 1.24,
+      "step": 3526500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.1234381956135915e-05,
+      "loss": 1.2288,
+      "step": 3527000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.1232281990575355e-05,
+      "loss": 1.2306,
+      "step": 3527500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1230186224945915e-05,
+      "loss": 1.2451,
+      "step": 3528000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.122809045931647e-05,
+      "loss": 1.2354,
+      "step": 3528500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.12259904937559e-05,
+      "loss": 1.2141,
+      "step": 3529000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1223890528195336e-05,
+      "loss": 1.233,
+      "step": 3529500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1221790562634776e-05,
+      "loss": 1.2152,
+      "step": 3530000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.121969059707421e-05,
+      "loss": 1.2528,
+      "step": 3530500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.121759063151364e-05,
+      "loss": 1.2356,
+      "step": 3531000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1215490665953076e-05,
+      "loss": 1.2097,
+      "step": 3531500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1213394900323637e-05,
+      "loss": 1.2307,
+      "step": 3532000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.121129493476307e-05,
+      "loss": 1.2151,
+      "step": 3532500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.120919496920251e-05,
+      "loss": 1.2007,
+      "step": 3533000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.120709500364194e-05,
+      "loss": 1.2277,
+      "step": 3533500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.12049992380125e-05,
+      "loss": 1.225,
+      "step": 3534000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.120289927245193e-05,
+      "loss": 1.2351,
+      "step": 3534500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.120079930689137e-05,
+      "loss": 1.2407,
+      "step": 3535000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1198699341330804e-05,
+      "loss": 1.2415,
+      "step": 3535500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.119659937577023e-05,
+      "loss": 1.2279,
+      "step": 3536000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.119449941020967e-05,
+      "loss": 1.2467,
+      "step": 3536500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.119240364458023e-05,
+      "loss": 1.2529,
+      "step": 3537000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1190303679019665e-05,
+      "loss": 1.2579,
+      "step": 3537500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.11882037134591e-05,
+      "loss": 1.243,
+      "step": 3538000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.118610374789853e-05,
+      "loss": 1.2212,
+      "step": 3538500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1184003782337966e-05,
+      "loss": 1.2285,
+      "step": 3539000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1181903816777406e-05,
+      "loss": 1.2538,
+      "step": 3539500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.117980385121684e-05,
+      "loss": 1.2445,
+      "step": 3540000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.11777080855874e-05,
+      "loss": 1.2141,
+      "step": 3540500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1175608120026826e-05,
+      "loss": 1.2616,
+      "step": 3541000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1173508154466267e-05,
+      "loss": 1.2623,
+      "step": 3541500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.11714081889057e-05,
+      "loss": 1.221,
+      "step": 3542000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1169308223345133e-05,
+      "loss": 1.241,
+      "step": 3542500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.116721245771569e-05,
+      "loss": 1.2284,
+      "step": 3543000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.116511249215513e-05,
+      "loss": 1.2147,
+      "step": 3543500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.116301252659456e-05,
+      "loss": 1.2416,
+      "step": 3544000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1160912561033994e-05,
+      "loss": 1.2338,
+      "step": 3544500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1158812595473434e-05,
+      "loss": 1.2398,
+      "step": 3545000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.115671682984399e-05,
+      "loss": 1.2359,
+      "step": 3545500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.115461686428342e-05,
+      "loss": 1.2084,
+      "step": 3546000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.115251689872286e-05,
+      "loss": 1.2191,
+      "step": 3546500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1150416933162295e-05,
+      "loss": 1.2479,
+      "step": 3547000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.114831696760173e-05,
+      "loss": 1.2123,
+      "step": 3547500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.114621700204117e-05,
+      "loss": 1.2314,
+      "step": 3548000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.114412123641172e-05,
+      "loss": 1.2204,
+      "step": 3548500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1142021270851156e-05,
+      "loss": 1.2198,
+      "step": 3549000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.113992130529059e-05,
+      "loss": 1.21,
+      "step": 3549500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.113782133973003e-05,
+      "loss": 1.2273,
+      "step": 3550000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.113572137416946e-05,
+      "loss": 1.2323,
+      "step": 3550500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1133625608540017e-05,
+      "loss": 1.2359,
+      "step": 3551000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.113152564297945e-05,
+      "loss": 1.2505,
+      "step": 3551500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.112942987735001e-05,
+      "loss": 1.2459,
+      "step": 3552000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.112732991178945e-05,
+      "loss": 1.2046,
+      "step": 3552500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1125234146160004e-05,
+      "loss": 1.2202,
+      "step": 3553000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.112313418059944e-05,
+      "loss": 1.2222,
+      "step": 3553500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.112103421503888e-05,
+      "loss": 1.251,
+      "step": 3554000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.111893424947831e-05,
+      "loss": 1.2437,
+      "step": 3554500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.111683428391774e-05,
+      "loss": 1.2565,
+      "step": 3555000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.111473431835718e-05,
+      "loss": 1.2552,
+      "step": 3555500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.111263435279661e-05,
+      "loss": 1.2169,
+      "step": 3556000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1110534387236045e-05,
+      "loss": 1.2241,
+      "step": 3556500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1108434421675485e-05,
+      "loss": 1.239,
+      "step": 3557000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.110633445611492e-05,
+      "loss": 1.2272,
+      "step": 3557500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.110423449055435e-05,
+      "loss": 1.2569,
+      "step": 3558000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.110213452499379e-05,
+      "loss": 1.2195,
+      "step": 3558500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1100034559433226e-05,
+      "loss": 1.2572,
+      "step": 3559000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.109793879380378e-05,
+      "loss": 1.2303,
+      "step": 3559500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.109583882824321e-05,
+      "loss": 1.2465,
+      "step": 3560000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.109373886268265e-05,
+      "loss": 1.2202,
+      "step": 3560500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.109163889712209e-05,
+      "loss": 1.239,
+      "step": 3561000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.108954313149264e-05,
+      "loss": 1.2675,
+      "step": 3561500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.108744316593208e-05,
+      "loss": 1.2567,
+      "step": 3562000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.1085343200371514e-05,
+      "loss": 1.2044,
+      "step": 3562500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.108324323481095e-05,
+      "loss": 1.2437,
+      "step": 3563000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.108114326925039e-05,
+      "loss": 1.2077,
+      "step": 3563500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.107904330368982e-05,
+      "loss": 1.2339,
+      "step": 3564000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.1076943338129255e-05,
+      "loss": 1.2318,
+      "step": 3564500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.1074843372568695e-05,
+      "loss": 1.2257,
+      "step": 3565000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.107275180687036e-05,
+      "loss": 1.2363,
+      "step": 3565500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.10706518413098e-05,
+      "loss": 1.234,
+      "step": 3566000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.1068551875749236e-05,
+      "loss": 1.2525,
+      "step": 3566500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.106645191018867e-05,
+      "loss": 1.2372,
+      "step": 3567000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.106435194462811e-05,
+      "loss": 1.2331,
+      "step": 3567500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.106225197906754e-05,
+      "loss": 1.2273,
+      "step": 3568000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.1060156213438096e-05,
+      "loss": 1.2108,
+      "step": 3568500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.1058056247877536e-05,
+      "loss": 1.2362,
+      "step": 3569000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.105596048224809e-05,
+      "loss": 1.2525,
+      "step": 3569500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.1053860516687524e-05,
+      "loss": 1.2313,
+      "step": 3570000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.105176055112696e-05,
+      "loss": 1.2501,
+      "step": 3570500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.10496605855664e-05,
+      "loss": 1.2583,
+      "step": 3571000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.104756062000583e-05,
+      "loss": 1.2126,
+      "step": 3571500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.1045460654445264e-05,
+      "loss": 1.2128,
+      "step": 3572000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.1043360688884704e-05,
+      "loss": 1.2059,
+      "step": 3572500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.104126072332414e-05,
+      "loss": 1.1771,
+      "step": 3573000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.103916075776357e-05,
+      "loss": 1.2223,
+      "step": 3573500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.103706079220301e-05,
+      "loss": 1.2126,
+      "step": 3574000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.1034960826642445e-05,
+      "loss": 1.2197,
+      "step": 3574500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.103286086108187e-05,
+      "loss": 1.229,
+      "step": 3575000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.103076509545243e-05,
+      "loss": 1.1865,
+      "step": 3575500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.102866512989187e-05,
+      "loss": 1.2268,
+      "step": 3576000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.1026569364262426e-05,
+      "loss": 1.2483,
+      "step": 3576500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.102446939870186e-05,
+      "loss": 1.2043,
+      "step": 3577000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.10223694331413e-05,
+      "loss": 1.2436,
+      "step": 3577500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.102026946758073e-05,
+      "loss": 1.2041,
+      "step": 3578000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.1018173701951287e-05,
+      "loss": 1.1972,
+      "step": 3578500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.101607373639072e-05,
+      "loss": 1.2306,
+      "step": 3579000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.101397377083016e-05,
+      "loss": 1.2229,
+      "step": 3579500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.1011873805269594e-05,
+      "loss": 1.2086,
+      "step": 3580000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.100977803964015e-05,
+      "loss": 1.2452,
+      "step": 3580500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.100767807407958e-05,
+      "loss": 1.2577,
+      "step": 3581000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.100557810851902e-05,
+      "loss": 1.2387,
+      "step": 3581500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.1003478142958454e-05,
+      "loss": 1.2188,
+      "step": 3582000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.100137817739789e-05,
+      "loss": 1.2117,
+      "step": 3582500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.099927821183733e-05,
+      "loss": 1.2442,
+      "step": 3583000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.099717824627676e-05,
+      "loss": 1.2268,
+      "step": 3583500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.09950782807162e-05,
+      "loss": 1.1932,
+      "step": 3584000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.099297831515563e-05,
+      "loss": 1.2111,
+      "step": 3584500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.099089094938843e-05,
+      "loss": 1.2372,
+      "step": 3585000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.098879098382787e-05,
+      "loss": 1.2462,
+      "step": 3585500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.0986691018267296e-05,
+      "loss": 1.2332,
+      "step": 3586000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.098459105270673e-05,
+      "loss": 1.2067,
+      "step": 3586500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.098249108714617e-05,
+      "loss": 1.2383,
+      "step": 3587000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.09803911215856e-05,
+      "loss": 1.2644,
+      "step": 3587500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.0978291156025037e-05,
+      "loss": 1.2061,
+      "step": 3588000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.097619119046448e-05,
+      "loss": 1.2285,
+      "step": 3588500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.097409122490391e-05,
+      "loss": 1.2196,
+      "step": 3589000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.0971991259343344e-05,
+      "loss": 1.2348,
+      "step": 3589500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.0969891293782784e-05,
+      "loss": 1.202,
+      "step": 3590000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.096779132822222e-05,
+      "loss": 1.2456,
+      "step": 3590500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.096569136266166e-05,
+      "loss": 1.1937,
+      "step": 3591000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.096359559703221e-05,
+      "loss": 1.1942,
+      "step": 3591500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.0961495631471645e-05,
+      "loss": 1.2394,
+      "step": 3592000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.095939566591108e-05,
+      "loss": 1.2142,
+      "step": 3592500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.095729570035052e-05,
+      "loss": 1.244,
+      "step": 3593000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.095519993472107e-05,
+      "loss": 1.2574,
+      "step": 3593500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.0953104169091626e-05,
+      "loss": 1.2328,
+      "step": 3594000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.095100420353106e-05,
+      "loss": 1.2613,
+      "step": 3594500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.094890423797049e-05,
+      "loss": 1.2338,
+      "step": 3595000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.094680427240993e-05,
+      "loss": 1.2463,
+      "step": 3595500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.0944704306849366e-05,
+      "loss": 1.2282,
+      "step": 3596000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.0942604341288806e-05,
+      "loss": 1.1977,
+      "step": 3596500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.094050437572824e-05,
+      "loss": 1.2276,
+      "step": 3597000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.093840441016767e-05,
+      "loss": 1.2254,
+      "step": 3597500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.0936304444607114e-05,
+      "loss": 1.2333,
+      "step": 3598000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.093420867897767e-05,
+      "loss": 1.2006,
+      "step": 3598500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.09321087134171e-05,
+      "loss": 1.2201,
+      "step": 3599000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.0930012947787654e-05,
+      "loss": 1.2306,
+      "step": 3599500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.092791298222709e-05,
+      "loss": 1.2166,
+      "step": 3600000
+    },
+    {
+      "epoch": 2.16,
+      "eval_loss": 1.1922718286514282,
+      "eval_runtime": 1098.7908,
+      "eval_samples_per_second": 479.363,
+      "eval_steps_per_second": 79.894,
+      "step": 3600000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.092581301666653e-05,
+      "loss": 1.2243,
+      "step": 3600500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.092371305110596e-05,
+      "loss": 1.2198,
+      "step": 3601000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.0921613085545395e-05,
+      "loss": 1.2253,
+      "step": 3601500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.0919513119984835e-05,
+      "loss": 1.2125,
+      "step": 3602000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.091741315442427e-05,
+      "loss": 1.2157,
+      "step": 3602500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.09153131888637e-05,
+      "loss": 1.2382,
+      "step": 3603000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.091321742323426e-05,
+      "loss": 1.2563,
+      "step": 3603500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.0911121657604816e-05,
+      "loss": 1.1968,
+      "step": 3604000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.090902169204425e-05,
+      "loss": 1.2192,
+      "step": 3604500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.090692172648368e-05,
+      "loss": 1.2573,
+      "step": 3605000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.090482176092312e-05,
+      "loss": 1.2138,
+      "step": 3605500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.0902721795362556e-05,
+      "loss": 1.2522,
+      "step": 3606000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.090062182980199e-05,
+      "loss": 1.2643,
+      "step": 3606500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.089852186424143e-05,
+      "loss": 1.2317,
+      "step": 3607000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.0896421898680864e-05,
+      "loss": 1.2595,
+      "step": 3607500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.089432613305142e-05,
+      "loss": 1.2503,
+      "step": 3608000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.089222616749085e-05,
+      "loss": 1.2369,
+      "step": 3608500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.0890130401861404e-05,
+      "loss": 1.2174,
+      "step": 3609000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.0888030436300844e-05,
+      "loss": 1.1942,
+      "step": 3609500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.088593047074028e-05,
+      "loss": 1.1876,
+      "step": 3610000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.088383050517972e-05,
+      "loss": 1.2287,
+      "step": 3610500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.088173053961915e-05,
+      "loss": 1.234,
+      "step": 3611000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.0879630574058585e-05,
+      "loss": 1.2279,
+      "step": 3611500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.0877530608498025e-05,
+      "loss": 1.2328,
+      "step": 3612000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.087543484286858e-05,
+      "loss": 1.2313,
+      "step": 3612500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.087333487730801e-05,
+      "loss": 1.2286,
+      "step": 3613000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.0871234911747446e-05,
+      "loss": 1.2097,
+      "step": 3613500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.0869134946186886e-05,
+      "loss": 1.2304,
+      "step": 3614000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.086703498062632e-05,
+      "loss": 1.2479,
+      "step": 3614500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.086493501506575e-05,
+      "loss": 1.24,
+      "step": 3615000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.0862835049505186e-05,
+      "loss": 1.1988,
+      "step": 3615500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.086073508394462e-05,
+      "loss": 1.2198,
+      "step": 3616000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.085863511838405e-05,
+      "loss": 1.2425,
+      "step": 3616500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.0856535152823494e-05,
+      "loss": 1.2301,
+      "step": 3617000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.085443518726293e-05,
+      "loss": 1.2253,
+      "step": 3617500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.085233522170236e-05,
+      "loss": 1.228,
+      "step": 3618000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.085023945607292e-05,
+      "loss": 1.215,
+      "step": 3618500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.0848139490512354e-05,
+      "loss": 1.205,
+      "step": 3619000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.084603952495179e-05,
+      "loss": 1.229,
+      "step": 3619500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.084394375932235e-05,
+      "loss": 1.2065,
+      "step": 3620000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.08418479936929e-05,
+      "loss": 1.2249,
+      "step": 3620500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.083974802813234e-05,
+      "loss": 1.2448,
+      "step": 3621000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.0837648062571775e-05,
+      "loss": 1.2269,
+      "step": 3621500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.083554809701121e-05,
+      "loss": 1.2452,
+      "step": 3622000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.083344813145064e-05,
+      "loss": 1.2182,
+      "step": 3622500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.0831348165890076e-05,
+      "loss": 1.2113,
+      "step": 3623000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.082924820032951e-05,
+      "loss": 1.2451,
+      "step": 3623500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.082714823476895e-05,
+      "loss": 1.2117,
+      "step": 3624000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.082505246913951e-05,
+      "loss": 1.2248,
+      "step": 3624500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.0822952503578936e-05,
+      "loss": 1.2089,
+      "step": 3625000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.082085253801838e-05,
+      "loss": 1.2073,
+      "step": 3625500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.081875257245781e-05,
+      "loss": 1.2196,
+      "step": 3626000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.0816652606897244e-05,
+      "loss": 1.2505,
+      "step": 3626500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.0814552641336684e-05,
+      "loss": 1.2404,
+      "step": 3627000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.081245267577612e-05,
+      "loss": 1.2212,
+      "step": 3627500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.081035271021555e-05,
+      "loss": 1.2246,
+      "step": 3628000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.080825274465499e-05,
+      "loss": 1.2115,
+      "step": 3628500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.0806156979025545e-05,
+      "loss": 1.232,
+      "step": 3629000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.080405701346498e-05,
+      "loss": 1.2346,
+      "step": 3629500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.080195704790441e-05,
+      "loss": 1.2611,
+      "step": 3630000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.079985708234385e-05,
+      "loss": 1.2433,
+      "step": 3630500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.0797757116783285e-05,
+      "loss": 1.2363,
+      "step": 3631000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.079565715122272e-05,
+      "loss": 1.2324,
+      "step": 3631500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.079355718566216e-05,
+      "loss": 1.2324,
+      "step": 3632000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.079145722010159e-05,
+      "loss": 1.2142,
+      "step": 3632500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.0789361454472146e-05,
+      "loss": 1.2238,
+      "step": 3633000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.078726148891158e-05,
+      "loss": 1.2173,
+      "step": 3633500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.078516152335102e-05,
+      "loss": 1.2172,
+      "step": 3634000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.078306155779045e-05,
+      "loss": 1.2255,
+      "step": 3634500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.078096579216101e-05,
+      "loss": 1.2241,
+      "step": 3635000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.077887002653156e-05,
+      "loss": 1.2359,
+      "step": 3635500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.0776770060971e-05,
+      "loss": 1.1963,
+      "step": 3636000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.0774670095410434e-05,
+      "loss": 1.2616,
+      "step": 3636500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.077257012984987e-05,
+      "loss": 1.2506,
+      "step": 3637000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.077047016428931e-05,
+      "loss": 1.2387,
+      "step": 3637500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.076837019872874e-05,
+      "loss": 1.2228,
+      "step": 3638000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.0766270233168175e-05,
+      "loss": 1.2354,
+      "step": 3638500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.0764170267607615e-05,
+      "loss": 1.2344,
+      "step": 3639000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.076207030204705e-05,
+      "loss": 1.2388,
+      "step": 3639500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.07599745364176e-05,
+      "loss": 1.2291,
+      "step": 3640000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.0757874570857035e-05,
+      "loss": 1.21,
+      "step": 3640500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.0755774605296476e-05,
+      "loss": 1.2431,
+      "step": 3641000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.075367463973591e-05,
+      "loss": 1.2163,
+      "step": 3641500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.075158307403758e-05,
+      "loss": 1.2121,
+      "step": 3642000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.0749483108477016e-05,
+      "loss": 1.2476,
+      "step": 3642500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.0747383142916456e-05,
+      "loss": 1.268,
+      "step": 3643000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.074528737728702e-05,
+      "loss": 1.2429,
+      "step": 3643500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.0743187411726443e-05,
+      "loss": 1.2271,
+      "step": 3644000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.074108744616588e-05,
+      "loss": 1.2201,
+      "step": 3644500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.073898748060532e-05,
+      "loss": 1.2605,
+      "step": 3645000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.073688751504475e-05,
+      "loss": 1.2273,
+      "step": 3645500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.0734787549484184e-05,
+      "loss": 1.2194,
+      "step": 3646000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.0732687583923624e-05,
+      "loss": 1.2482,
+      "step": 3646500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.073058761836306e-05,
+      "loss": 1.2196,
+      "step": 3647000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.072848765280249e-05,
+      "loss": 1.2508,
+      "step": 3647500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.072638768724193e-05,
+      "loss": 1.2245,
+      "step": 3648000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.0724287721681365e-05,
+      "loss": 1.2326,
+      "step": 3648500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.072219195605192e-05,
+      "loss": 1.2206,
+      "step": 3649000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.072009199049136e-05,
+      "loss": 1.234,
+      "step": 3649500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.071799202493079e-05,
+      "loss": 1.2378,
+      "step": 3650000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.0715892059370226e-05,
+      "loss": 1.2241,
+      "step": 3650500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.0713792093809666e-05,
+      "loss": 1.2061,
+      "step": 3651000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.07116921282491e-05,
+      "loss": 1.2387,
+      "step": 3651500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.070959636261965e-05,
+      "loss": 1.2423,
+      "step": 3652000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.0707496397059086e-05,
+      "loss": 1.2047,
+      "step": 3652500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.0705396431498527e-05,
+      "loss": 1.2417,
+      "step": 3653000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.070329646593796e-05,
+      "loss": 1.2263,
+      "step": 3653500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.0701196500377393e-05,
+      "loss": 1.2074,
+      "step": 3654000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.069910073474795e-05,
+      "loss": 1.2336,
+      "step": 3654500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.069700076918739e-05,
+      "loss": 1.2786,
+      "step": 3655000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.069490080362682e-05,
+      "loss": 1.2572,
+      "step": 3655500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.069280083806626e-05,
+      "loss": 1.2065,
+      "step": 3656000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.0690700872505694e-05,
+      "loss": 1.2142,
+      "step": 3656500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.068860090694512e-05,
+      "loss": 1.2379,
+      "step": 3657000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.068650094138456e-05,
+      "loss": 1.2094,
+      "step": 3657500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.0684400975823995e-05,
+      "loss": 1.2192,
+      "step": 3658000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.068230101026343e-05,
+      "loss": 1.2333,
+      "step": 3658500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.068020944456511e-05,
+      "loss": 1.1983,
+      "step": 3659000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.067810947900454e-05,
+      "loss": 1.2629,
+      "step": 3659500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.067600951344398e-05,
+      "loss": 1.2069,
+      "step": 3660000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.0673909547883416e-05,
+      "loss": 1.2524,
+      "step": 3660500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.067180958232285e-05,
+      "loss": 1.246,
+      "step": 3661000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.06697138166934e-05,
+      "loss": 1.2532,
+      "step": 3661500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.066761385113284e-05,
+      "loss": 1.212,
+      "step": 3662000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.066551388557228e-05,
+      "loss": 1.2123,
+      "step": 3662500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.066341392001172e-05,
+      "loss": 1.2382,
+      "step": 3663000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.066131395445115e-05,
+      "loss": 1.2281,
+      "step": 3663500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.0659222388752824e-05,
+      "loss": 1.2386,
+      "step": 3664000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.065712242319226e-05,
+      "loss": 1.2523,
+      "step": 3664500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.065502665756282e-05,
+      "loss": 1.2435,
+      "step": 3665000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.0652926692002244e-05,
+      "loss": 1.2438,
+      "step": 3665500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.0650826726441685e-05,
+      "loss": 1.2506,
+      "step": 3666000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.064872676088112e-05,
+      "loss": 1.2182,
+      "step": 3666500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.064662679532055e-05,
+      "loss": 1.2312,
+      "step": 3667000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.064452682975999e-05,
+      "loss": 1.2317,
+      "step": 3667500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.0642426864199425e-05,
+      "loss": 1.2195,
+      "step": 3668000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.0640326898638866e-05,
+      "loss": 1.1908,
+      "step": 3668500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.06382269330783e-05,
+      "loss": 1.2278,
+      "step": 3669000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.063612696751773e-05,
+      "loss": 1.2565,
+      "step": 3669500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.063402700195717e-05,
+      "loss": 1.242,
+      "step": 3670000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.0631927036396606e-05,
+      "loss": 1.2207,
+      "step": 3670500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.062983127076716e-05,
+      "loss": 1.2385,
+      "step": 3671000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.062773130520659e-05,
+      "loss": 1.2234,
+      "step": 3671500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.0625631339646033e-05,
+      "loss": 1.2211,
+      "step": 3672000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.062353137408547e-05,
+      "loss": 1.2357,
+      "step": 3672500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.062143560845602e-05,
+      "loss": 1.2332,
+      "step": 3673000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.0619339842826574e-05,
+      "loss": 1.2096,
+      "step": 3673500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.061723987726601e-05,
+      "loss": 1.2311,
+      "step": 3674000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.061513991170545e-05,
+      "loss": 1.2103,
+      "step": 3674500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.061303994614488e-05,
+      "loss": 1.2281,
+      "step": 3675000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.061093998058432e-05,
+      "loss": 1.2381,
+      "step": 3675500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.0608840015023755e-05,
+      "loss": 1.2298,
+      "step": 3676000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.060674004946319e-05,
+      "loss": 1.2287,
+      "step": 3676500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.060464008390263e-05,
+      "loss": 1.2298,
+      "step": 3677000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.060254011834206e-05,
+      "loss": 1.2294,
+      "step": 3677500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.060044015278149e-05,
+      "loss": 1.1901,
+      "step": 3678000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.059834018722093e-05,
+      "loss": 1.2617,
+      "step": 3678500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.059624442159149e-05,
+      "loss": 1.2458,
+      "step": 3679000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.059414445603092e-05,
+      "loss": 1.2122,
+      "step": 3679500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.0592044490470356e-05,
+      "loss": 1.2364,
+      "step": 3680000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.058994872484091e-05,
+      "loss": 1.2302,
+      "step": 3680500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.058784875928035e-05,
+      "loss": 1.2124,
+      "step": 3681000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.0585748793719784e-05,
+      "loss": 1.2392,
+      "step": 3681500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.058364882815922e-05,
+      "loss": 1.2562,
+      "step": 3682000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.058154886259866e-05,
+      "loss": 1.2149,
+      "step": 3682500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.0579448897038084e-05,
+      "loss": 1.2255,
+      "step": 3683000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.0577348931477524e-05,
+      "loss": 1.209,
+      "step": 3683500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.057524896591696e-05,
+      "loss": 1.225,
+      "step": 3684000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.057314900035639e-05,
+      "loss": 1.2127,
+      "step": 3684500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.057104903479583e-05,
+      "loss": 1.2378,
+      "step": 3685000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.0568953269166385e-05,
+      "loss": 1.2231,
+      "step": 3685500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.056685330360582e-05,
+      "loss": 1.1944,
+      "step": 3686000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.056475333804525e-05,
+      "loss": 1.2109,
+      "step": 3686500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.056265757241581e-05,
+      "loss": 1.2287,
+      "step": 3687000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.0560557606855246e-05,
+      "loss": 1.2089,
+      "step": 3687500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.055845764129468e-05,
+      "loss": 1.2439,
+      "step": 3688000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.055635767573411e-05,
+      "loss": 1.2123,
+      "step": 3688500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.055425771017355e-05,
+      "loss": 1.2368,
+      "step": 3689000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.055216194454411e-05,
+      "loss": 1.2622,
+      "step": 3689500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.055006197898354e-05,
+      "loss": 1.2529,
+      "step": 3690000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.054796201342298e-05,
+      "loss": 1.2587,
+      "step": 3690500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.0545862047862413e-05,
+      "loss": 1.2493,
+      "step": 3691000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.054376208230185e-05,
+      "loss": 1.1992,
+      "step": 3691500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.054166211674129e-05,
+      "loss": 1.2125,
+      "step": 3692000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.053956215118072e-05,
+      "loss": 1.2267,
+      "step": 3692500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.0537462185620154e-05,
+      "loss": 1.2221,
+      "step": 3693000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.0535362220059594e-05,
+      "loss": 1.2125,
+      "step": 3693500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.053326645443015e-05,
+      "loss": 1.2395,
+      "step": 3694000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.053116648886958e-05,
+      "loss": 1.2272,
+      "step": 3694500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.0529066523309015e-05,
+      "loss": 1.2237,
+      "step": 3695000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.0526966557748455e-05,
+      "loss": 1.2141,
+      "step": 3695500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.052486659218789e-05,
+      "loss": 1.2262,
+      "step": 3696000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.052276662662732e-05,
+      "loss": 1.2308,
+      "step": 3696500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.052066666106676e-05,
+      "loss": 1.2152,
+      "step": 3697000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.0518566695506196e-05,
+      "loss": 1.2367,
+      "step": 3697500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.051646672994562e-05,
+      "loss": 1.2155,
+      "step": 3698000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.051437096431618e-05,
+      "loss": 1.2354,
+      "step": 3698500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.051227519868674e-05,
+      "loss": 1.2028,
+      "step": 3699000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.0510175233126177e-05,
+      "loss": 1.2355,
+      "step": 3699500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.050807526756561e-05,
+      "loss": 1.2172,
+      "step": 3700000
+    },
+    {
+      "epoch": 2.22,
+      "eval_loss": 1.192676305770874,
+      "eval_runtime": 1108.3091,
+      "eval_samples_per_second": 475.246,
+      "eval_steps_per_second": 79.208,
+      "step": 3700000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.050597530200505e-05,
+      "loss": 1.2434,
+      "step": 3700500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.0503875336444484e-05,
+      "loss": 1.2336,
+      "step": 3701000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.050177537088392e-05,
+      "loss": 1.2125,
+      "step": 3701500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.049967540532336e-05,
+      "loss": 1.2513,
+      "step": 3702000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.049757543976279e-05,
+      "loss": 1.2471,
+      "step": 3702500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.0495479674133344e-05,
+      "loss": 1.2444,
+      "step": 3703000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.049337970857278e-05,
+      "loss": 1.2351,
+      "step": 3703500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.049128394294333e-05,
+      "loss": 1.2162,
+      "step": 3704000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.048918397738277e-05,
+      "loss": 1.2194,
+      "step": 3704500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.0487084011822205e-05,
+      "loss": 1.2149,
+      "step": 3705000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.048498404626164e-05,
+      "loss": 1.2341,
+      "step": 3705500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.04828882806322e-05,
+      "loss": 1.2217,
+      "step": 3706000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.048078831507163e-05,
+      "loss": 1.2377,
+      "step": 3706500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.0478688349511066e-05,
+      "loss": 1.1999,
+      "step": 3707000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.0476588383950506e-05,
+      "loss": 1.2312,
+      "step": 3707500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.047448841838994e-05,
+      "loss": 1.2233,
+      "step": 3708000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.047238845282937e-05,
+      "loss": 1.2217,
+      "step": 3708500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.047028848726881e-05,
+      "loss": 1.2382,
+      "step": 3709000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.046818852170825e-05,
+      "loss": 1.2415,
+      "step": 3709500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.04660927560788e-05,
+      "loss": 1.209,
+      "step": 3710000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.0463992790518234e-05,
+      "loss": 1.2346,
+      "step": 3710500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.0461892824957674e-05,
+      "loss": 1.2342,
+      "step": 3711000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.045979285939711e-05,
+      "loss": 1.236,
+      "step": 3711500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.045769289383654e-05,
+      "loss": 1.2323,
+      "step": 3712000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.0455597128207094e-05,
+      "loss": 1.2306,
+      "step": 3712500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.0453497162646535e-05,
+      "loss": 1.2228,
+      "step": 3713000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.045139719708597e-05,
+      "loss": 1.2014,
+      "step": 3713500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.04492972315254e-05,
+      "loss": 1.2309,
+      "step": 3714000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.044720146589596e-05,
+      "loss": 1.2202,
+      "step": 3714500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.0445101500335395e-05,
+      "loss": 1.2246,
+      "step": 3715000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.044300573470595e-05,
+      "loss": 1.234,
+      "step": 3715500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.044090576914538e-05,
+      "loss": 1.2274,
+      "step": 3716000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.043880580358482e-05,
+      "loss": 1.2261,
+      "step": 3716500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.0436705838024256e-05,
+      "loss": 1.2294,
+      "step": 3717000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.043460587246369e-05,
+      "loss": 1.2285,
+      "step": 3717500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.043250590690313e-05,
+      "loss": 1.2242,
+      "step": 3718000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.043040594134256e-05,
+      "loss": 1.2183,
+      "step": 3718500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.0428305975782e-05,
+      "loss": 1.2699,
+      "step": 3719000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.042620601022143e-05,
+      "loss": 1.2293,
+      "step": 3719500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.0424106044660864e-05,
+      "loss": 1.2611,
+      "step": 3720000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.04220060791003e-05,
+      "loss": 1.2013,
+      "step": 3720500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.041990611353974e-05,
+      "loss": 1.2404,
+      "step": 3721000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.04178103479103e-05,
+      "loss": 1.234,
+      "step": 3721500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.0415710382349724e-05,
+      "loss": 1.2356,
+      "step": 3722000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.0413610416789165e-05,
+      "loss": 1.239,
+      "step": 3722500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.04115104512286e-05,
+      "loss": 1.2559,
+      "step": 3723000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.040941048566803e-05,
+      "loss": 1.2246,
+      "step": 3723500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.040731052010747e-05,
+      "loss": 1.2373,
+      "step": 3724000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.0405210554546905e-05,
+      "loss": 1.2451,
+      "step": 3724500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.040311058898634e-05,
+      "loss": 1.2407,
+      "step": 3725000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.040101482335689e-05,
+      "loss": 1.2441,
+      "step": 3725500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.039891485779633e-05,
+      "loss": 1.209,
+      "step": 3726000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.0396814892235766e-05,
+      "loss": 1.2323,
+      "step": 3726500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.03947149266752e-05,
+      "loss": 1.2416,
+      "step": 3727000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.039261496111464e-05,
+      "loss": 1.2182,
+      "step": 3727500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.039051919548519e-05,
+      "loss": 1.2359,
+      "step": 3728000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.038841922992463e-05,
+      "loss": 1.2215,
+      "step": 3728500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.038631926436407e-05,
+      "loss": 1.2037,
+      "step": 3729000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.03842192988035e-05,
+      "loss": 1.2159,
+      "step": 3729500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.0382119333242934e-05,
+      "loss": 1.2106,
+      "step": 3730000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.038002356761349e-05,
+      "loss": 1.2295,
+      "step": 3730500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.037792360205293e-05,
+      "loss": 1.2282,
+      "step": 3731000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.037582363649236e-05,
+      "loss": 1.2091,
+      "step": 3731500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.0373723670931795e-05,
+      "loss": 1.2213,
+      "step": 3732000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.037162790530235e-05,
+      "loss": 1.2201,
+      "step": 3732500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.036952793974179e-05,
+      "loss": 1.2254,
+      "step": 3733000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.036742797418122e-05,
+      "loss": 1.2457,
+      "step": 3733500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.0365332208551775e-05,
+      "loss": 1.1988,
+      "step": 3734000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.0363232242991216e-05,
+      "loss": 1.2321,
+      "step": 3734500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.036113227743065e-05,
+      "loss": 1.2142,
+      "step": 3735000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.035903231187008e-05,
+      "loss": 1.2166,
+      "step": 3735500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.035693234630952e-05,
+      "loss": 1.2453,
+      "step": 3736000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.0354832380748956e-05,
+      "loss": 1.252,
+      "step": 3736500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.035273661511951e-05,
+      "loss": 1.2193,
+      "step": 3737000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.035063664955894e-05,
+      "loss": 1.2318,
+      "step": 3737500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.0348536683998384e-05,
+      "loss": 1.1904,
+      "step": 3738000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.034643671843782e-05,
+      "loss": 1.2344,
+      "step": 3738500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.034433675287725e-05,
+      "loss": 1.2372,
+      "step": 3739000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.034223678731669e-05,
+      "loss": 1.251,
+      "step": 3739500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.0340136821756124e-05,
+      "loss": 1.1963,
+      "step": 3740000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.033804105612668e-05,
+      "loss": 1.1995,
+      "step": 3740500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.033594109056611e-05,
+      "loss": 1.2382,
+      "step": 3741000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.033384112500555e-05,
+      "loss": 1.2309,
+      "step": 3741500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.0331741159444985e-05,
+      "loss": 1.2407,
+      "step": 3742000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.032964119388442e-05,
+      "loss": 1.2217,
+      "step": 3742500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.032754122832386e-05,
+      "loss": 1.2138,
+      "step": 3743000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.032544126276329e-05,
+      "loss": 1.2027,
+      "step": 3743500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.0323341297202725e-05,
+      "loss": 1.2205,
+      "step": 3744000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.032124133164216e-05,
+      "loss": 1.2238,
+      "step": 3744500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.031914136608159e-05,
+      "loss": 1.2236,
+      "step": 3745000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.031704140052103e-05,
+      "loss": 1.2203,
+      "step": 3745500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.0314941434960466e-05,
+      "loss": 1.2014,
+      "step": 3746000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.031284566933102e-05,
+      "loss": 1.2485,
+      "step": 3746500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.031074990370158e-05,
+      "loss": 1.2402,
+      "step": 3747000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.0308649938141013e-05,
+      "loss": 1.2122,
+      "step": 3747500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.0306549972580454e-05,
+      "loss": 1.2152,
+      "step": 3748000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.030445000701989e-05,
+      "loss": 1.2247,
+      "step": 3748500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.0302350041459314e-05,
+      "loss": 1.2051,
+      "step": 3749000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.0300254275829874e-05,
+      "loss": 1.2276,
+      "step": 3749500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.0298154310269314e-05,
+      "loss": 1.2253,
+      "step": 3750000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.029605434470875e-05,
+      "loss": 1.2133,
+      "step": 3750500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.029395437914818e-05,
+      "loss": 1.2134,
+      "step": 3751000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.0291854413587615e-05,
+      "loss": 1.221,
+      "step": 3751500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.0289758647958175e-05,
+      "loss": 1.2105,
+      "step": 3752000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.028765868239761e-05,
+      "loss": 1.2212,
+      "step": 3752500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.028555871683705e-05,
+      "loss": 1.2855,
+      "step": 3753000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.0283458751276476e-05,
+      "loss": 1.2226,
+      "step": 3753500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.0281362985647036e-05,
+      "loss": 1.1972,
+      "step": 3754000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.027926302008647e-05,
+      "loss": 1.2402,
+      "step": 3754500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.027716305452591e-05,
+      "loss": 1.254,
+      "step": 3755000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.027506308896534e-05,
+      "loss": 1.2116,
+      "step": 3755500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.02729673233359e-05,
+      "loss": 1.2319,
+      "step": 3756000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.027086735777533e-05,
+      "loss": 1.2095,
+      "step": 3756500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.026876739221477e-05,
+      "loss": 1.2341,
+      "step": 3757000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.0266671626585324e-05,
+      "loss": 1.2214,
+      "step": 3757500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.026457166102476e-05,
+      "loss": 1.2548,
+      "step": 3758000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.02624716954642e-05,
+      "loss": 1.2287,
+      "step": 3758500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.026037172990363e-05,
+      "loss": 1.2418,
+      "step": 3759000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.0258271764343065e-05,
+      "loss": 1.2027,
+      "step": 3759500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.025617599871362e-05,
+      "loss": 1.2405,
+      "step": 3760000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.025407603315306e-05,
+      "loss": 1.1929,
+      "step": 3760500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.025197606759249e-05,
+      "loss": 1.2297,
+      "step": 3761000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.0249876102031925e-05,
+      "loss": 1.229,
+      "step": 3761500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.024778033640248e-05,
+      "loss": 1.2265,
+      "step": 3762000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.024568037084192e-05,
+      "loss": 1.2253,
+      "step": 3762500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.024358040528135e-05,
+      "loss": 1.1993,
+      "step": 3763000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.0241480439720786e-05,
+      "loss": 1.2261,
+      "step": 3763500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.0239380474160226e-05,
+      "loss": 1.2344,
+      "step": 3764000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.023728050859966e-05,
+      "loss": 1.2172,
+      "step": 3764500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.023518054303909e-05,
+      "loss": 1.2214,
+      "step": 3765000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.0233080577478527e-05,
+      "loss": 1.208,
+      "step": 3765500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.023098061191796e-05,
+      "loss": 1.206,
+      "step": 3766000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.022888484628852e-05,
+      "loss": 1.2395,
+      "step": 3766500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.0226789080659074e-05,
+      "loss": 1.2404,
+      "step": 3767000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.0224689115098514e-05,
+      "loss": 1.2381,
+      "step": 3767500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.022258914953795e-05,
+      "loss": 1.224,
+      "step": 3768000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.022048918397738e-05,
+      "loss": 1.2424,
+      "step": 3768500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.021838921841682e-05,
+      "loss": 1.2153,
+      "step": 3769000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.0216289252856255e-05,
+      "loss": 1.225,
+      "step": 3769500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.021419348722681e-05,
+      "loss": 1.2436,
+      "step": 3770000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.021209352166624e-05,
+      "loss": 1.2217,
+      "step": 3770500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.020999355610568e-05,
+      "loss": 1.2131,
+      "step": 3771000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.0207893590545116e-05,
+      "loss": 1.2384,
+      "step": 3771500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.020579362498455e-05,
+      "loss": 1.2014,
+      "step": 3772000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.020369365942398e-05,
+      "loss": 1.232,
+      "step": 3772500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.0201593693863416e-05,
+      "loss": 1.2192,
+      "step": 3773000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.0199493728302856e-05,
+      "loss": 1.2221,
+      "step": 3773500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.0197397962673417e-05,
+      "loss": 1.2128,
+      "step": 3774000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.019529799711285e-05,
+      "loss": 1.2326,
+      "step": 3774500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.019319803155228e-05,
+      "loss": 1.2197,
+      "step": 3775000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.019110226592284e-05,
+      "loss": 1.2171,
+      "step": 3775500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.018900230036228e-05,
+      "loss": 1.1986,
+      "step": 3776000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.018690233480171e-05,
+      "loss": 1.2416,
+      "step": 3776500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.0184802369241144e-05,
+      "loss": 1.2002,
+      "step": 3777000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.018270240368058e-05,
+      "loss": 1.2206,
+      "step": 3777500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.018060243812001e-05,
+      "loss": 1.2283,
+      "step": 3778000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.0178502472559445e-05,
+      "loss": 1.2574,
+      "step": 3778500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.0176406706930005e-05,
+      "loss": 1.2058,
+      "step": 3779000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.0174306741369445e-05,
+      "loss": 1.2127,
+      "step": 3779500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.017220677580887e-05,
+      "loss": 1.2163,
+      "step": 3780000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.017010681024831e-05,
+      "loss": 1.2169,
+      "step": 3780500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.0168006844687745e-05,
+      "loss": 1.1947,
+      "step": 3781000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.016590687912718e-05,
+      "loss": 1.2692,
+      "step": 3781500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.016380691356662e-05,
+      "loss": 1.2323,
+      "step": 3782000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.016170694800605e-05,
+      "loss": 1.1993,
+      "step": 3782500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.0159606982445486e-05,
+      "loss": 1.2416,
+      "step": 3783000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.015751121681604e-05,
+      "loss": 1.2126,
+      "step": 3783500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.015541125125548e-05,
+      "loss": 1.2343,
+      "step": 3784000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.015331128569491e-05,
+      "loss": 1.2284,
+      "step": 3784500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.015121132013435e-05,
+      "loss": 1.2446,
+      "step": 3785000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.014911135457379e-05,
+      "loss": 1.2092,
+      "step": 3785500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.014701558894434e-05,
+      "loss": 1.2086,
+      "step": 3786000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.0144915623383774e-05,
+      "loss": 1.2135,
+      "step": 3786500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.0142815657823214e-05,
+      "loss": 1.2134,
+      "step": 3787000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.014071569226265e-05,
+      "loss": 1.2327,
+      "step": 3787500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.013861572670208e-05,
+      "loss": 1.2734,
+      "step": 3788000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.0136519961072635e-05,
+      "loss": 1.2331,
+      "step": 3788500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.0134419995512075e-05,
+      "loss": 1.2391,
+      "step": 3789000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.013232002995151e-05,
+      "loss": 1.2187,
+      "step": 3789500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.013022006439094e-05,
+      "loss": 1.2273,
+      "step": 3790000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.012812009883038e-05,
+      "loss": 1.2226,
+      "step": 3790500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.0126020133269816e-05,
+      "loss": 1.2266,
+      "step": 3791000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.012392436764037e-05,
+      "loss": 1.2393,
+      "step": 3791500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.012182860201092e-05,
+      "loss": 1.2179,
+      "step": 3792000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.0119728636450356e-05,
+      "loss": 1.2371,
+      "step": 3792500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.0117628670889797e-05,
+      "loss": 1.2009,
+      "step": 3793000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.011552870532923e-05,
+      "loss": 1.2375,
+      "step": 3793500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.011342873976867e-05,
+      "loss": 1.2117,
+      "step": 3794000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.0111328774208104e-05,
+      "loss": 1.2382,
+      "step": 3794500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.010922880864754e-05,
+      "loss": 1.225,
+      "step": 3795000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.010712884308698e-05,
+      "loss": 1.2032,
+      "step": 3795500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.010502887752641e-05,
+      "loss": 1.2264,
+      "step": 3796000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.0102933111896964e-05,
+      "loss": 1.2476,
+      "step": 3796500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.010083734626752e-05,
+      "loss": 1.2524,
+      "step": 3797000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.009873738070695e-05,
+      "loss": 1.2193,
+      "step": 3797500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.009663741514639e-05,
+      "loss": 1.2413,
+      "step": 3798000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.0094537449585825e-05,
+      "loss": 1.2038,
+      "step": 3798500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.009243748402526e-05,
+      "loss": 1.2231,
+      "step": 3799000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.00903375184647e-05,
+      "loss": 1.228,
+      "step": 3799500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.008824175283525e-05,
+      "loss": 1.2295,
+      "step": 3800000
+    },
+    {
+      "epoch": 2.28,
+      "eval_loss": 1.1859376430511475,
+      "eval_runtime": 1106.7769,
+      "eval_samples_per_second": 475.904,
+      "eval_steps_per_second": 79.318,
+      "step": 3800000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.0086141787274686e-05,
+      "loss": 1.2295,
+      "step": 3800500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.0084041821714126e-05,
+      "loss": 1.2621,
+      "step": 3801000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.008194185615356e-05,
+      "loss": 1.235,
+      "step": 3801500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.007984189059299e-05,
+      "loss": 1.2179,
+      "step": 3802000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.007774192503243e-05,
+      "loss": 1.2533,
+      "step": 3802500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.007564195947187e-05,
+      "loss": 1.2081,
+      "step": 3803000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.00735419939113e-05,
+      "loss": 1.2337,
+      "step": 3803500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.007144202835074e-05,
+      "loss": 1.2429,
+      "step": 3804000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.0069346262721294e-05,
+      "loss": 1.2493,
+      "step": 3804500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.006724629716073e-05,
+      "loss": 1.2188,
+      "step": 3805000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.006514633160016e-05,
+      "loss": 1.2603,
+      "step": 3805500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.00630463660396e-05,
+      "loss": 1.2455,
+      "step": 3806000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.0060946400479035e-05,
+      "loss": 1.2194,
+      "step": 3806500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.005885063484959e-05,
+      "loss": 1.2155,
+      "step": 3807000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.005675066928902e-05,
+      "loss": 1.2382,
+      "step": 3807500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.005465490365958e-05,
+      "loss": 1.2212,
+      "step": 3808000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.0052554938099015e-05,
+      "loss": 1.2376,
+      "step": 3808500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.005045497253845e-05,
+      "loss": 1.239,
+      "step": 3809000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.004835500697789e-05,
+      "loss": 1.2201,
+      "step": 3809500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.004625504141732e-05,
+      "loss": 1.2301,
+      "step": 3810000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.0044155075856756e-05,
+      "loss": 1.2337,
+      "step": 3810500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.004205931022731e-05,
+      "loss": 1.2127,
+      "step": 3811000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.003995934466675e-05,
+      "loss": 1.2537,
+      "step": 3811500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.003785937910618e-05,
+      "loss": 1.218,
+      "step": 3812000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.003575941354562e-05,
+      "loss": 1.2413,
+      "step": 3812500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.003365944798506e-05,
+      "loss": 1.2389,
+      "step": 3813000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.003155948242449e-05,
+      "loss": 1.2313,
+      "step": 3813500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.002945951686392e-05,
+      "loss": 1.2005,
+      "step": 3814000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.002735955130336e-05,
+      "loss": 1.2185,
+      "step": 3814500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.002525958574279e-05,
+      "loss": 1.1975,
+      "step": 3815000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.002316382011335e-05,
+      "loss": 1.2022,
+      "step": 3815500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.0021063854552785e-05,
+      "loss": 1.2372,
+      "step": 3816000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.0018968088923345e-05,
+      "loss": 1.2136,
+      "step": 3816500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.001686812336278e-05,
+      "loss": 1.2496,
+      "step": 3817000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.001476815780221e-05,
+      "loss": 1.2184,
+      "step": 3817500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.001266819224165e-05,
+      "loss": 1.2038,
+      "step": 3818000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.001056822668108e-05,
+      "loss": 1.2349,
+      "step": 3818500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.000847246105164e-05,
+      "loss": 1.2321,
+      "step": 3819000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.000637249549107e-05,
+      "loss": 1.2216,
+      "step": 3819500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.000427252993051e-05,
+      "loss": 1.2294,
+      "step": 3820000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.0002176764301066e-05,
+      "loss": 1.218,
+      "step": 3820500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.00000767987405e-05,
+      "loss": 1.2304,
+      "step": 3821000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.999797683317993e-05,
+      "loss": 1.229,
+      "step": 3821500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.9995876867619374e-05,
+      "loss": 1.2447,
+      "step": 3822000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.999377690205881e-05,
+      "loss": 1.2044,
+      "step": 3822500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.999167693649824e-05,
+      "loss": 1.2295,
+      "step": 3823000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.9989576970937674e-05,
+      "loss": 1.2095,
+      "step": 3823500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.998747700537711e-05,
+      "loss": 1.2296,
+      "step": 3824000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.998537703981655e-05,
+      "loss": 1.2479,
+      "step": 3824500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.998327707425598e-05,
+      "loss": 1.202,
+      "step": 3825000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.9981177108695415e-05,
+      "loss": 1.2597,
+      "step": 3825500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.9979077143134855e-05,
+      "loss": 1.2141,
+      "step": 3826000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.997697717757429e-05,
+      "loss": 1.2214,
+      "step": 3826500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.997487721201372e-05,
+      "loss": 1.2346,
+      "step": 3827000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.9972781446384275e-05,
+      "loss": 1.2272,
+      "step": 3827500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.9970681480823716e-05,
+      "loss": 1.2402,
+      "step": 3828000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.996858151526315e-05,
+      "loss": 1.2146,
+      "step": 3828500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.996648154970258e-05,
+      "loss": 1.2089,
+      "step": 3829000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.996438158414202e-05,
+      "loss": 1.2236,
+      "step": 3829500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.9962281618581456e-05,
+      "loss": 1.2246,
+      "step": 3830000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.996018585295201e-05,
+      "loss": 1.2155,
+      "step": 3830500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.995808588739144e-05,
+      "loss": 1.2294,
+      "step": 3831000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.9955985921830883e-05,
+      "loss": 1.244,
+      "step": 3831500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.995388595627032e-05,
+      "loss": 1.203,
+      "step": 3832000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.995178599070975e-05,
+      "loss": 1.2403,
+      "step": 3832500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.994968602514919e-05,
+      "loss": 1.2284,
+      "step": 3833000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.9947590259519744e-05,
+      "loss": 1.1975,
+      "step": 3833500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.994549029395918e-05,
+      "loss": 1.2493,
+      "step": 3834000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.994339032839862e-05,
+      "loss": 1.2343,
+      "step": 3834500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.994129036283805e-05,
+      "loss": 1.2659,
+      "step": 3835000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.9939190397277485e-05,
+      "loss": 1.2187,
+      "step": 3835500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.993709463164804e-05,
+      "loss": 1.1905,
+      "step": 3836000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.993499466608748e-05,
+      "loss": 1.2247,
+      "step": 3836500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.993289470052691e-05,
+      "loss": 1.217,
+      "step": 3837000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.9930794734966346e-05,
+      "loss": 1.2256,
+      "step": 3837500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.9928694769405786e-05,
+      "loss": 1.2343,
+      "step": 3838000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.992659480384521e-05,
+      "loss": 1.2218,
+      "step": 3838500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.9924494838284646e-05,
+      "loss": 1.2373,
+      "step": 3839000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.992239907265521e-05,
+      "loss": 1.2347,
+      "step": 3839500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.9920299107094646e-05,
+      "loss": 1.2353,
+      "step": 3840000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.991819914153408e-05,
+      "loss": 1.2096,
+      "step": 3840500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.9916099175973513e-05,
+      "loss": 1.2193,
+      "step": 3841000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.991399921041295e-05,
+      "loss": 1.2244,
+      "step": 3841500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.991189924485238e-05,
+      "loss": 1.2308,
+      "step": 3842000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.990980347922294e-05,
+      "loss": 1.2142,
+      "step": 3842500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.990770351366238e-05,
+      "loss": 1.2318,
+      "step": 3843000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.990560354810181e-05,
+      "loss": 1.2424,
+      "step": 3843500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.990350358254124e-05,
+      "loss": 1.2326,
+      "step": 3844000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.990140361698068e-05,
+      "loss": 1.2217,
+      "step": 3844500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.9899312051282355e-05,
+      "loss": 1.2055,
+      "step": 3845000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.9897212085721795e-05,
+      "loss": 1.1714,
+      "step": 3845500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.989511212016123e-05,
+      "loss": 1.2076,
+      "step": 3846000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.989301215460067e-05,
+      "loss": 1.2242,
+      "step": 3846500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.98909121890401e-05,
+      "loss": 1.2223,
+      "step": 3847000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.9888812223479536e-05,
+      "loss": 1.2323,
+      "step": 3847500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.988671225791897e-05,
+      "loss": 1.2468,
+      "step": 3848000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.98846122923584e-05,
+      "loss": 1.227,
+      "step": 3848500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.9882512326797836e-05,
+      "loss": 1.2182,
+      "step": 3849000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.9880416561168397e-05,
+      "loss": 1.244,
+      "step": 3849500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.987831659560784e-05,
+      "loss": 1.2073,
+      "step": 3850000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.9876216630047263e-05,
+      "loss": 1.2531,
+      "step": 3850500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.98741166644867e-05,
+      "loss": 1.2744,
+      "step": 3851000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.987201669892614e-05,
+      "loss": 1.2052,
+      "step": 3851500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.98699209332967e-05,
+      "loss": 1.2213,
+      "step": 3852000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.986782096773613e-05,
+      "loss": 1.2197,
+      "step": 3852500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.9865725202106685e-05,
+      "loss": 1.2312,
+      "step": 3853000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.9863625236546125e-05,
+      "loss": 1.2026,
+      "step": 3853500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.986152527098556e-05,
+      "loss": 1.2241,
+      "step": 3854000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.985942530542499e-05,
+      "loss": 1.221,
+      "step": 3854500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.9857325339864425e-05,
+      "loss": 1.2316,
+      "step": 3855000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.985522537430386e-05,
+      "loss": 1.2236,
+      "step": 3855500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.985312540874329e-05,
+      "loss": 1.2518,
+      "step": 3856000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.985102544318273e-05,
+      "loss": 1.2274,
+      "step": 3856500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.984892967755329e-05,
+      "loss": 1.2127,
+      "step": 3857000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.9846833911923846e-05,
+      "loss": 1.2219,
+      "step": 3857500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.984473394636328e-05,
+      "loss": 1.2206,
+      "step": 3858000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.984263398080271e-05,
+      "loss": 1.22,
+      "step": 3858500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.9840534015242153e-05,
+      "loss": 1.2221,
+      "step": 3859000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.983843824961271e-05,
+      "loss": 1.207,
+      "step": 3859500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.983634248398326e-05,
+      "loss": 1.2375,
+      "step": 3860000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.9834242518422694e-05,
+      "loss": 1.2418,
+      "step": 3860500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.9832142552862134e-05,
+      "loss": 1.2286,
+      "step": 3861000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.983004258730157e-05,
+      "loss": 1.2387,
+      "step": 3861500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.9827942621741e-05,
+      "loss": 1.2128,
+      "step": 3862000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.982584265618044e-05,
+      "loss": 1.2518,
+      "step": 3862500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.9823742690619875e-05,
+      "loss": 1.2373,
+      "step": 3863000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.982164272505931e-05,
+      "loss": 1.2124,
+      "step": 3863500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.981954275949875e-05,
+      "loss": 1.2217,
+      "step": 3864000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.9817442793938175e-05,
+      "loss": 1.2218,
+      "step": 3864500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.981534282837761e-05,
+      "loss": 1.1981,
+      "step": 3865000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.981324706274817e-05,
+      "loss": 1.2045,
+      "step": 3865500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.981114709718761e-05,
+      "loss": 1.2512,
+      "step": 3866000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.980904713162704e-05,
+      "loss": 1.2418,
+      "step": 3866500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.9806947166066476e-05,
+      "loss": 1.2223,
+      "step": 3867000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.980484720050591e-05,
+      "loss": 1.2658,
+      "step": 3867500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.980274723494534e-05,
+      "loss": 1.2289,
+      "step": 3868000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.980064726938478e-05,
+      "loss": 1.2233,
+      "step": 3868500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.979854730382422e-05,
+      "loss": 1.205,
+      "step": 3869000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.979645153819477e-05,
+      "loss": 1.2314,
+      "step": 3869500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.9794351572634204e-05,
+      "loss": 1.2227,
+      "step": 3870000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.9792251607073644e-05,
+      "loss": 1.232,
+      "step": 3870500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.979015164151308e-05,
+      "loss": 1.2337,
+      "step": 3871000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.978805167595251e-05,
+      "loss": 1.2419,
+      "step": 3871500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.978595171039195e-05,
+      "loss": 1.207,
+      "step": 3872000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.9783851744831385e-05,
+      "loss": 1.2061,
+      "step": 3872500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.978175177927082e-05,
+      "loss": 1.2766,
+      "step": 3873000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.977965181371026e-05,
+      "loss": 1.2395,
+      "step": 3873500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.977755184814969e-05,
+      "loss": 1.2211,
+      "step": 3874000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.9775456082520245e-05,
+      "loss": 1.2168,
+      "step": 3874500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.977335611695968e-05,
+      "loss": 1.2173,
+      "step": 3875000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.977125615139912e-05,
+      "loss": 1.2133,
+      "step": 3875500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.976915618583855e-05,
+      "loss": 1.1967,
+      "step": 3876000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.9767056220277986e-05,
+      "loss": 1.2088,
+      "step": 3876500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.9764956254717426e-05,
+      "loss": 1.2168,
+      "step": 3877000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.976285628915685e-05,
+      "loss": 1.2124,
+      "step": 3877500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.976075632359629e-05,
+      "loss": 1.2249,
+      "step": 3878000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.9758660557966854e-05,
+      "loss": 1.2312,
+      "step": 3878500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.975656059240629e-05,
+      "loss": 1.2233,
+      "step": 3879000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.975446062684572e-05,
+      "loss": 1.2363,
+      "step": 3879500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.9752364861216274e-05,
+      "loss": 1.2148,
+      "step": 3880000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.9750264895655714e-05,
+      "loss": 1.2229,
+      "step": 3880500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.974816493009515e-05,
+      "loss": 1.216,
+      "step": 3881000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.974606496453458e-05,
+      "loss": 1.2214,
+      "step": 3881500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.9743964998974015e-05,
+      "loss": 1.2161,
+      "step": 3882000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.974186503341345e-05,
+      "loss": 1.2188,
+      "step": 3882500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.973976926778401e-05,
+      "loss": 1.2454,
+      "step": 3883000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.973766930222344e-05,
+      "loss": 1.2071,
+      "step": 3883500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.973556933666288e-05,
+      "loss": 1.2426,
+      "step": 3884000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.973346937110231e-05,
+      "loss": 1.2363,
+      "step": 3884500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.973136940554175e-05,
+      "loss": 1.2436,
+      "step": 3885000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.972926943998118e-05,
+      "loss": 1.2437,
+      "step": 3885500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.9727169474420616e-05,
+      "loss": 1.2279,
+      "step": 3886000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.9725069508860056e-05,
+      "loss": 1.2067,
+      "step": 3886500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.972297374323061e-05,
+      "loss": 1.2206,
+      "step": 3887000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.972087797760117e-05,
+      "loss": 1.2149,
+      "step": 3887500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.9718778012040604e-05,
+      "loss": 1.2389,
+      "step": 3888000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.971667804648004e-05,
+      "loss": 1.2087,
+      "step": 3888500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.971457808091948e-05,
+      "loss": 1.2387,
+      "step": 3889000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.9712478115358904e-05,
+      "loss": 1.2262,
+      "step": 3889500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.971037814979834e-05,
+      "loss": 1.2241,
+      "step": 3890000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.970827818423778e-05,
+      "loss": 1.2458,
+      "step": 3890500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.970617821867721e-05,
+      "loss": 1.2198,
+      "step": 3891000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.9704078253116645e-05,
+      "loss": 1.2494,
+      "step": 3891500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.9701982487487205e-05,
+      "loss": 1.2222,
+      "step": 3892000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.969988252192664e-05,
+      "loss": 1.2218,
+      "step": 3892500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.969778255636607e-05,
+      "loss": 1.226,
+      "step": 3893000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.969568259080551e-05,
+      "loss": 1.2203,
+      "step": 3893500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.9693582625244946e-05,
+      "loss": 1.2118,
+      "step": 3894000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.969148265968438e-05,
+      "loss": 1.1966,
+      "step": 3894500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.968938269412382e-05,
+      "loss": 1.2097,
+      "step": 3895000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.968728272856325e-05,
+      "loss": 1.2611,
+      "step": 3895500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.9685186962933806e-05,
+      "loss": 1.2342,
+      "step": 3896000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.968308699737324e-05,
+      "loss": 1.2161,
+      "step": 3896500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.968098703181268e-05,
+      "loss": 1.2041,
+      "step": 3897000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.9678887066252113e-05,
+      "loss": 1.211,
+      "step": 3897500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.967678710069155e-05,
+      "loss": 1.2134,
+      "step": 3898000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.967468713513099e-05,
+      "loss": 1.2278,
+      "step": 3898500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.967258716957042e-05,
+      "loss": 1.1962,
+      "step": 3899000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.9670487204009854e-05,
+      "loss": 1.223,
+      "step": 3899500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.9668391438380414e-05,
+      "loss": 1.2325,
+      "step": 3900000
+    },
+    {
+      "epoch": 2.34,
+      "eval_loss": 1.182997465133667,
+      "eval_runtime": 1103.215,
+      "eval_samples_per_second": 477.441,
+      "eval_steps_per_second": 79.574,
+      "step": 3900000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.966629147281985e-05,
+      "loss": 1.2135,
+      "step": 3900500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.96641957071904e-05,
+      "loss": 1.2202,
+      "step": 3901000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.9662095741629835e-05,
+      "loss": 1.2039,
+      "step": 3901500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.9659995776069275e-05,
+      "loss": 1.2438,
+      "step": 3902000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.965789581050871e-05,
+      "loss": 1.2209,
+      "step": 3902500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.965580004487926e-05,
+      "loss": 1.2199,
+      "step": 3903000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.9653700079318696e-05,
+      "loss": 1.2234,
+      "step": 3903500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.9651600113758136e-05,
+      "loss": 1.2066,
+      "step": 3904000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.964950014819757e-05,
+      "loss": 1.2392,
+      "step": 3904500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.964740438256812e-05,
+      "loss": 1.1814,
+      "step": 3905000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.964530441700756e-05,
+      "loss": 1.2365,
+      "step": 3905500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.9643204451446997e-05,
+      "loss": 1.227,
+      "step": 3906000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.964110448588643e-05,
+      "loss": 1.2452,
+      "step": 3906500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.963900452032587e-05,
+      "loss": 1.2149,
+      "step": 3907000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.9636904554765304e-05,
+      "loss": 1.2287,
+      "step": 3907500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.963480458920474e-05,
+      "loss": 1.2221,
+      "step": 3908000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.963270462364418e-05,
+      "loss": 1.2155,
+      "step": 3908500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.9630604658083604e-05,
+      "loss": 1.2551,
+      "step": 3909000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.962850469252304e-05,
+      "loss": 1.2434,
+      "step": 3909500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.96264089268936e-05,
+      "loss": 1.2046,
+      "step": 3910000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.962430896133304e-05,
+      "loss": 1.2169,
+      "step": 3910500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.962220899577247e-05,
+      "loss": 1.2146,
+      "step": 3911000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.9620113230143025e-05,
+      "loss": 1.2568,
+      "step": 3911500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.961801326458246e-05,
+      "loss": 1.2267,
+      "step": 3912000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.96159132990219e-05,
+      "loss": 1.218,
+      "step": 3912500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.961381333346133e-05,
+      "loss": 1.2001,
+      "step": 3913000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.9611713367900766e-05,
+      "loss": 1.2187,
+      "step": 3913500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.9609617602271326e-05,
+      "loss": 1.2243,
+      "step": 3914000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.960751763671076e-05,
+      "loss": 1.2377,
+      "step": 3914500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.960541767115019e-05,
+      "loss": 1.214,
+      "step": 3915000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.960331770558963e-05,
+      "loss": 1.2322,
+      "step": 3915500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.960121774002907e-05,
+      "loss": 1.2044,
+      "step": 3916000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.9599117774468493e-05,
+      "loss": 1.228,
+      "step": 3916500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.9597017808907934e-05,
+      "loss": 1.2246,
+      "step": 3917000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.959491784334737e-05,
+      "loss": 1.2071,
+      "step": 3917500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.95928178777868e-05,
+      "loss": 1.2354,
+      "step": 3918000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.9590722112157354e-05,
+      "loss": 1.2243,
+      "step": 3918500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.9588626346527915e-05,
+      "loss": 1.2141,
+      "step": 3919000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.9586526380967355e-05,
+      "loss": 1.2364,
+      "step": 3919500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.958442641540679e-05,
+      "loss": 1.2201,
+      "step": 3920000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.958232644984622e-05,
+      "loss": 1.221,
+      "step": 3920500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.9580226484285655e-05,
+      "loss": 1.2547,
+      "step": 3921000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.9578130718656215e-05,
+      "loss": 1.2474,
+      "step": 3921500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.957603075309565e-05,
+      "loss": 1.2294,
+      "step": 3922000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.957393078753509e-05,
+      "loss": 1.263,
+      "step": 3922500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.957183082197452e-05,
+      "loss": 1.2571,
+      "step": 3923000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.956973085641395e-05,
+      "loss": 1.2371,
+      "step": 3923500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.956763089085339e-05,
+      "loss": 1.2215,
+      "step": 3924000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.956553092529282e-05,
+      "loss": 1.2212,
+      "step": 3924500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.9563430959732256e-05,
+      "loss": 1.2221,
+      "step": 3925000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.956133519410282e-05,
+      "loss": 1.2038,
+      "step": 3925500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.955923522854225e-05,
+      "loss": 1.2145,
+      "step": 3926000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.9557135262981684e-05,
+      "loss": 1.2251,
+      "step": 3926500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.955503529742112e-05,
+      "loss": 1.2476,
+      "step": 3927000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.955293533186056e-05,
+      "loss": 1.2225,
+      "step": 3927500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.955083956623111e-05,
+      "loss": 1.213,
+      "step": 3928000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.9548739600670544e-05,
+      "loss": 1.1864,
+      "step": 3928500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.9546639635109985e-05,
+      "loss": 1.2299,
+      "step": 3929000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.954453966954942e-05,
+      "loss": 1.2347,
+      "step": 3929500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.954244390391998e-05,
+      "loss": 1.2048,
+      "step": 3930000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.9540343938359405e-05,
+      "loss": 1.2036,
+      "step": 3930500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.9538243972798845e-05,
+      "loss": 1.2505,
+      "step": 3931000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.953614400723828e-05,
+      "loss": 1.2062,
+      "step": 3931500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.953404404167771e-05,
+      "loss": 1.2233,
+      "step": 3932000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.953194407611715e-05,
+      "loss": 1.2271,
+      "step": 3932500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.9529844110556586e-05,
+      "loss": 1.208,
+      "step": 3933000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.952774414499602e-05,
+      "loss": 1.1932,
+      "step": 3933500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.952564417943546e-05,
+      "loss": 1.2209,
+      "step": 3934000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.952354841380601e-05,
+      "loss": 1.2182,
+      "step": 3934500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.952144844824545e-05,
+      "loss": 1.2195,
+      "step": 3935000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.951934848268488e-05,
+      "loss": 1.2237,
+      "step": 3935500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.951724851712432e-05,
+      "loss": 1.2259,
+      "step": 3936000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.9515148551563754e-05,
+      "loss": 1.2238,
+      "step": 3936500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.951305278593431e-05,
+      "loss": 1.2428,
+      "step": 3937000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.951095282037375e-05,
+      "loss": 1.21,
+      "step": 3937500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.950885285481318e-05,
+      "loss": 1.2413,
+      "step": 3938000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.9506752889252615e-05,
+      "loss": 1.2355,
+      "step": 3938500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.950465712362317e-05,
+      "loss": 1.2127,
+      "step": 3939000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.950255715806261e-05,
+      "loss": 1.2229,
+      "step": 3939500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.950045719250204e-05,
+      "loss": 1.2357,
+      "step": 3940000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.9498357226941475e-05,
+      "loss": 1.2254,
+      "step": 3940500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.9496257261380916e-05,
+      "loss": 1.2408,
+      "step": 3941000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.949416149575147e-05,
+      "loss": 1.2336,
+      "step": 3941500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.94920615301909e-05,
+      "loss": 1.2255,
+      "step": 3942000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.9489961564630336e-05,
+      "loss": 1.2097,
+      "step": 3942500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.9487861599069776e-05,
+      "loss": 1.2258,
+      "step": 3943000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.948576163350921e-05,
+      "loss": 1.2371,
+      "step": 3943500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.948366586787976e-05,
+      "loss": 1.2487,
+      "step": 3944000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.9481565902319204e-05,
+      "loss": 1.2133,
+      "step": 3944500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.947946593675864e-05,
+      "loss": 1.2195,
+      "step": 3945000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.947736597119807e-05,
+      "loss": 1.2327,
+      "step": 3945500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.947526600563751e-05,
+      "loss": 1.2133,
+      "step": 3946000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.9473166040076944e-05,
+      "loss": 1.2001,
+      "step": 3946500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.947106607451638e-05,
+      "loss": 1.2304,
+      "step": 3947000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.946896610895582e-05,
+      "loss": 1.2162,
+      "step": 3947500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.946687034332637e-05,
+      "loss": 1.2176,
+      "step": 3948000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.9464770377765805e-05,
+      "loss": 1.1931,
+      "step": 3948500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.946267041220524e-05,
+      "loss": 1.2594,
+      "step": 3949000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.946057044664468e-05,
+      "loss": 1.2181,
+      "step": 3949500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.945847048108411e-05,
+      "loss": 1.2022,
+      "step": 3950000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.945637051552354e-05,
+      "loss": 1.2217,
+      "step": 3950500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.94542747498941e-05,
+      "loss": 1.2317,
+      "step": 3951000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.945217478433354e-05,
+      "loss": 1.22,
+      "step": 3951500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.945007481877297e-05,
+      "loss": 1.2306,
+      "step": 3952000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.944797485321241e-05,
+      "loss": 1.2177,
+      "step": 3952500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.944587488765184e-05,
+      "loss": 1.1972,
+      "step": 3953000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.94437791220224e-05,
+      "loss": 1.2126,
+      "step": 3953500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.9441679156461834e-05,
+      "loss": 1.227,
+      "step": 3954000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.9439579190901274e-05,
+      "loss": 1.2464,
+      "step": 3954500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.943748342527183e-05,
+      "loss": 1.2256,
+      "step": 3955000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.943538345971126e-05,
+      "loss": 1.2448,
+      "step": 3955500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.9433287694081814e-05,
+      "loss": 1.2113,
+      "step": 3956000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.943118772852125e-05,
+      "loss": 1.2208,
+      "step": 3956500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.942908776296069e-05,
+      "loss": 1.2053,
+      "step": 3957000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.942698779740012e-05,
+      "loss": 1.2084,
+      "step": 3957500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.942488783183956e-05,
+      "loss": 1.2354,
+      "step": 3958000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.9422787866278995e-05,
+      "loss": 1.235,
+      "step": 3958500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.942068790071843e-05,
+      "loss": 1.242,
+      "step": 3959000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.941858793515787e-05,
+      "loss": 1.2198,
+      "step": 3959500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.9416487969597296e-05,
+      "loss": 1.204,
+      "step": 3960000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.941438800403673e-05,
+      "loss": 1.1948,
+      "step": 3960500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.941229223840729e-05,
+      "loss": 1.2152,
+      "step": 3961000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.941019227284673e-05,
+      "loss": 1.2248,
+      "step": 3961500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.940809230728616e-05,
+      "loss": 1.2,
+      "step": 3962000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.940599234172559e-05,
+      "loss": 1.2139,
+      "step": 3962500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.940389237616503e-05,
+      "loss": 1.2378,
+      "step": 3963000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.9401792410604464e-05,
+      "loss": 1.2209,
+      "step": 3963500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.93996924450439e-05,
+      "loss": 1.2103,
+      "step": 3964000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.939759667941446e-05,
+      "loss": 1.2015,
+      "step": 3964500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.939549671385389e-05,
+      "loss": 1.2247,
+      "step": 3965000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.9393396748293324e-05,
+      "loss": 1.2298,
+      "step": 3965500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.9391296782732764e-05,
+      "loss": 1.2251,
+      "step": 3966000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.93891968171722e-05,
+      "loss": 1.2227,
+      "step": 3966500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.938709685161163e-05,
+      "loss": 1.2028,
+      "step": 3967000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.938499688605107e-05,
+      "loss": 1.2052,
+      "step": 3967500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.9382896920490505e-05,
+      "loss": 1.2249,
+      "step": 3968000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.938080115486106e-05,
+      "loss": 1.222,
+      "step": 3968500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.937870118930049e-05,
+      "loss": 1.2299,
+      "step": 3969000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.937660122373993e-05,
+      "loss": 1.2218,
+      "step": 3969500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.9374501258179366e-05,
+      "loss": 1.225,
+      "step": 3970000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.937240549254992e-05,
+      "loss": 1.2338,
+      "step": 3970500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.937030972692048e-05,
+      "loss": 1.1964,
+      "step": 3971000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.936820976135991e-05,
+      "loss": 1.1903,
+      "step": 3971500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.936610979579935e-05,
+      "loss": 1.236,
+      "step": 3972000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.936400983023878e-05,
+      "loss": 1.194,
+      "step": 3972500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.936190986467822e-05,
+      "loss": 1.2338,
+      "step": 3973000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.9359809899117654e-05,
+      "loss": 1.2481,
+      "step": 3973500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.935770993355709e-05,
+      "loss": 1.2367,
+      "step": 3974000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.935560996799653e-05,
+      "loss": 1.2048,
+      "step": 3974500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.935351000243596e-05,
+      "loss": 1.2289,
+      "step": 3975000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.9351410036875394e-05,
+      "loss": 1.2092,
+      "step": 3975500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.934931427124595e-05,
+      "loss": 1.2271,
+      "step": 3976000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.934721430568539e-05,
+      "loss": 1.23,
+      "step": 3976500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.934511434012482e-05,
+      "loss": 1.2616,
+      "step": 3977000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.9343014374564255e-05,
+      "loss": 1.2253,
+      "step": 3977500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.9340914409003695e-05,
+      "loss": 1.2187,
+      "step": 3978000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.933881444344313e-05,
+      "loss": 1.2422,
+      "step": 3978500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.933671447788256e-05,
+      "loss": 1.2065,
+      "step": 3979000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.9334614512322e-05,
+      "loss": 1.1699,
+      "step": 3979500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.9332518746692556e-05,
+      "loss": 1.1928,
+      "step": 3980000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.933042298106311e-05,
+      "loss": 1.2199,
+      "step": 3980500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.932832301550254e-05,
+      "loss": 1.2142,
+      "step": 3981000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.932622304994198e-05,
+      "loss": 1.2195,
+      "step": 3981500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.932412308438142e-05,
+      "loss": 1.2127,
+      "step": 3982000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.932202731875197e-05,
+      "loss": 1.2224,
+      "step": 3982500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.9319927353191404e-05,
+      "loss": 1.2491,
+      "step": 3983000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.9317827387630844e-05,
+      "loss": 1.2436,
+      "step": 3983500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.931572742207028e-05,
+      "loss": 1.2479,
+      "step": 3984000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.931362745650971e-05,
+      "loss": 1.1887,
+      "step": 3984500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.931152749094915e-05,
+      "loss": 1.2188,
+      "step": 3985000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.9309427525388585e-05,
+      "loss": 1.1963,
+      "step": 3985500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.930732755982802e-05,
+      "loss": 1.1988,
+      "step": 3986000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.930523179419857e-05,
+      "loss": 1.2259,
+      "step": 3986500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.930313182863801e-05,
+      "loss": 1.2126,
+      "step": 3987000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.9301036063008566e-05,
+      "loss": 1.1968,
+      "step": 3987500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.9298936097448e-05,
+      "loss": 1.2178,
+      "step": 3988000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.929683613188744e-05,
+      "loss": 1.204,
+      "step": 3988500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.929473616632687e-05,
+      "loss": 1.2282,
+      "step": 3989000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.9292640400697426e-05,
+      "loss": 1.1988,
+      "step": 3989500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.929054043513686e-05,
+      "loss": 1.2136,
+      "step": 3990000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.92884404695763e-05,
+      "loss": 1.2268,
+      "step": 3990500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.9286340504015733e-05,
+      "loss": 1.2081,
+      "step": 3991000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.928424053845517e-05,
+      "loss": 1.2573,
+      "step": 3991500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.928214057289461e-05,
+      "loss": 1.2153,
+      "step": 3992000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.928004060733404e-05,
+      "loss": 1.2559,
+      "step": 3992500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.9277940641773474e-05,
+      "loss": 1.2327,
+      "step": 3993000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.9275840676212914e-05,
+      "loss": 1.2099,
+      "step": 3993500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.927374071065234e-05,
+      "loss": 1.2278,
+      "step": 3994000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.92716449450229e-05,
+      "loss": 1.25,
+      "step": 3994500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.9269544979462335e-05,
+      "loss": 1.2103,
+      "step": 3995000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.9267449213832895e-05,
+      "loss": 1.2158,
+      "step": 3995500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.926534924827233e-05,
+      "loss": 1.214,
+      "step": 3996000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.926324928271176e-05,
+      "loss": 1.2043,
+      "step": 3996500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.92611493171512e-05,
+      "loss": 1.1996,
+      "step": 3997000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.9259049351590636e-05,
+      "loss": 1.1874,
+      "step": 3997500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.925694938603007e-05,
+      "loss": 1.2523,
+      "step": 3998000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.925484942046951e-05,
+      "loss": 1.2168,
+      "step": 3998500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.9252749454908936e-05,
+      "loss": 1.212,
+      "step": 3999000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.9250653689279496e-05,
+      "loss": 1.2271,
+      "step": 3999500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.924855372371893e-05,
+      "loss": 1.2031,
+      "step": 4000000
+    },
+    {
+      "epoch": 2.4,
+      "eval_loss": 1.1803804636001587,
+      "eval_runtime": 1110.222,
+      "eval_samples_per_second": 474.428,
+      "eval_steps_per_second": 79.072,
+      "step": 4000000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.924645375815837e-05,
+      "loss": 1.2029,
+      "step": 4000500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.92443537925978e-05,
+      "loss": 1.2338,
+      "step": 4001000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.924225382703723e-05,
+      "loss": 1.2061,
+      "step": 4001500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.924015806140779e-05,
+      "loss": 1.2055,
+      "step": 4002000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.923805809584723e-05,
+      "loss": 1.2464,
+      "step": 4002500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.9235958130286664e-05,
+      "loss": 1.2344,
+      "step": 4003000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.92338581647261e-05,
+      "loss": 1.2164,
+      "step": 4003500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.923175819916553e-05,
+      "loss": 1.2256,
+      "step": 4004000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.9229658233604965e-05,
+      "loss": 1.2295,
+      "step": 4004500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.9227558268044405e-05,
+      "loss": 1.2117,
+      "step": 4005000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.922545830248384e-05,
+      "loss": 1.2196,
+      "step": 4005500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.922336253685439e-05,
+      "loss": 1.2075,
+      "step": 4006000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.9221262571293825e-05,
+      "loss": 1.2173,
+      "step": 4006500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.9219162605733266e-05,
+      "loss": 1.2265,
+      "step": 4007000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.92170626401727e-05,
+      "loss": 1.2209,
+      "step": 4007500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.921496687454326e-05,
+      "loss": 1.2143,
+      "step": 4008000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.9212866908982686e-05,
+      "loss": 1.1951,
+      "step": 4008500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.9210766943422126e-05,
+      "loss": 1.203,
+      "step": 4009000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.920866697786156e-05,
+      "loss": 1.1978,
+      "step": 4009500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.920656701230099e-05,
+      "loss": 1.2072,
+      "step": 4010000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.9204475446602674e-05,
+      "loss": 1.2454,
+      "step": 4010500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.9202375481042114e-05,
+      "loss": 1.1936,
+      "step": 4011000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.920027551548155e-05,
+      "loss": 1.2084,
+      "step": 4011500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.919817554992098e-05,
+      "loss": 1.1898,
+      "step": 4012000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.919607558436042e-05,
+      "loss": 1.2102,
+      "step": 4012500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.919397561879985e-05,
+      "loss": 1.2079,
+      "step": 4013000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.919187985317041e-05,
+      "loss": 1.2116,
+      "step": 4013500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.918977988760984e-05,
+      "loss": 1.1974,
+      "step": 4014000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.918767992204928e-05,
+      "loss": 1.2262,
+      "step": 4014500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9185579956488715e-05,
+      "loss": 1.2148,
+      "step": 4015000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.918348419085927e-05,
+      "loss": 1.2092,
+      "step": 4015500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.91813842252987e-05,
+      "loss": 1.2172,
+      "step": 4016000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.917928425973814e-05,
+      "loss": 1.1798,
+      "step": 4016500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9177184294177576e-05,
+      "loss": 1.227,
+      "step": 4017000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9175084328617016e-05,
+      "loss": 1.2369,
+      "step": 4017500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.917298856298757e-05,
+      "loss": 1.2196,
+      "step": 4018000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9170888597427e-05,
+      "loss": 1.2308,
+      "step": 4018500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.916878863186644e-05,
+      "loss": 1.2334,
+      "step": 4019000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.916668866630588e-05,
+      "loss": 1.2158,
+      "step": 4019500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.916458870074531e-05,
+      "loss": 1.2299,
+      "step": 4020000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.916248873518474e-05,
+      "loss": 1.2024,
+      "step": 4020500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.916038876962418e-05,
+      "loss": 1.247,
+      "step": 4021000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.915828880406361e-05,
+      "loss": 1.1977,
+      "step": 4021500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.915619303843417e-05,
+      "loss": 1.2243,
+      "step": 4022000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.91540930728736e-05,
+      "loss": 1.214,
+      "step": 4022500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.915199310731304e-05,
+      "loss": 1.2222,
+      "step": 4023000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.91498973416836e-05,
+      "loss": 1.2033,
+      "step": 4023500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.914779737612303e-05,
+      "loss": 1.2374,
+      "step": 4024000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.914569741056247e-05,
+      "loss": 1.2157,
+      "step": 4024500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.91435974450019e-05,
+      "loss": 1.2258,
+      "step": 4025000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.914149747944133e-05,
+      "loss": 1.2522,
+      "step": 4025500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.913939751388077e-05,
+      "loss": 1.2239,
+      "step": 4026000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9137297548320206e-05,
+      "loss": 1.2323,
+      "step": 4026500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9135201782690766e-05,
+      "loss": 1.2101,
+      "step": 4027000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.913310181713019e-05,
+      "loss": 1.2292,
+      "step": 4027500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.913100185156963e-05,
+      "loss": 1.2309,
+      "step": 4028000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.912890188600907e-05,
+      "loss": 1.2354,
+      "step": 4028500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.91268019204485e-05,
+      "loss": 1.21,
+      "step": 4029000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.912470615481906e-05,
+      "loss": 1.2372,
+      "step": 4029500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.9122606189258494e-05,
+      "loss": 1.2124,
+      "step": 4030000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.912050622369793e-05,
+      "loss": 1.2474,
+      "step": 4030500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.911840625813737e-05,
+      "loss": 1.2053,
+      "step": 4031000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.91163062925768e-05,
+      "loss": 1.217,
+      "step": 4031500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.911421472687848e-05,
+      "loss": 1.2627,
+      "step": 4032000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.9112114761317915e-05,
+      "loss": 1.2186,
+      "step": 4032500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.911001479575735e-05,
+      "loss": 1.2202,
+      "step": 4033000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.910791483019679e-05,
+      "loss": 1.2264,
+      "step": 4033500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.910581486463622e-05,
+      "loss": 1.2483,
+      "step": 4034000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.910371489907565e-05,
+      "loss": 1.2219,
+      "step": 4034500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.910161913344621e-05,
+      "loss": 1.2246,
+      "step": 4035000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.909951916788565e-05,
+      "loss": 1.2301,
+      "step": 4035500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.90974234022562e-05,
+      "loss": 1.2103,
+      "step": 4036000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.9095323436695637e-05,
+      "loss": 1.2154,
+      "step": 4036500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.909322347113508e-05,
+      "loss": 1.2242,
+      "step": 4037000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.909112350557451e-05,
+      "loss": 1.2288,
+      "step": 4037500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.9089023540013944e-05,
+      "loss": 1.2116,
+      "step": 4038000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.9086923574453384e-05,
+      "loss": 1.2543,
+      "step": 4038500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.908483200875505e-05,
+      "loss": 1.2354,
+      "step": 4039000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.908273204319449e-05,
+      "loss": 1.2052,
+      "step": 4039500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.9080632077633925e-05,
+      "loss": 1.2095,
+      "step": 4040000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.907853211207336e-05,
+      "loss": 1.248,
+      "step": 4040500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.90764321465128e-05,
+      "loss": 1.2232,
+      "step": 4041000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.907433218095223e-05,
+      "loss": 1.2078,
+      "step": 4041500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.9072232215391665e-05,
+      "loss": 1.1842,
+      "step": 4042000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.9070132249831105e-05,
+      "loss": 1.2017,
+      "step": 4042500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.906803228427054e-05,
+      "loss": 1.2267,
+      "step": 4043000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.906593231870997e-05,
+      "loss": 1.2406,
+      "step": 4043500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.9063832353149406e-05,
+      "loss": 1.2154,
+      "step": 4044000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.906173238758884e-05,
+      "loss": 1.2191,
+      "step": 4044500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.905963242202828e-05,
+      "loss": 1.2114,
+      "step": 4045000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.905753245646771e-05,
+      "loss": 1.2223,
+      "step": 4045500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.9055432490907146e-05,
+      "loss": 1.25,
+      "step": 4046000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.90533367252777e-05,
+      "loss": 1.2221,
+      "step": 4046500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.905123675971714e-05,
+      "loss": 1.2487,
+      "step": 4047000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.9049136794156574e-05,
+      "loss": 1.2401,
+      "step": 4047500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.904703682859601e-05,
+      "loss": 1.2349,
+      "step": 4048000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.904493686303545e-05,
+      "loss": 1.2031,
+      "step": 4048500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.904283689747488e-05,
+      "loss": 1.2308,
+      "step": 4049000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.9040736931914314e-05,
+      "loss": 1.2282,
+      "step": 4049500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.9038636966353755e-05,
+      "loss": 1.2161,
+      "step": 4050000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.903654120072431e-05,
+      "loss": 1.1919,
+      "step": 4050500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.903444123516374e-05,
+      "loss": 1.2266,
+      "step": 4051000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.9032341269603175e-05,
+      "loss": 1.2017,
+      "step": 4051500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.9030241304042615e-05,
+      "loss": 1.2076,
+      "step": 4052000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.902814133848205e-05,
+      "loss": 1.1977,
+      "step": 4052500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.902604137292148e-05,
+      "loss": 1.2084,
+      "step": 4053000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.902394140736092e-05,
+      "loss": 1.2279,
+      "step": 4053500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.9021841441800356e-05,
+      "loss": 1.2395,
+      "step": 4054000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.901974147623979e-05,
+      "loss": 1.2089,
+      "step": 4054500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.901764151067922e-05,
+      "loss": 1.2309,
+      "step": 4055000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.901554574504978e-05,
+      "loss": 1.222,
+      "step": 4055500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.9013445779489217e-05,
+      "loss": 1.204,
+      "step": 4056000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.901134581392866e-05,
+      "loss": 1.2047,
+      "step": 4056500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.9009245848368084e-05,
+      "loss": 1.2174,
+      "step": 4057000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.900714588280752e-05,
+      "loss": 1.2433,
+      "step": 4057500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.900505011717808e-05,
+      "loss": 1.2193,
+      "step": 4058000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.900295015161752e-05,
+      "loss": 1.2104,
+      "step": 4058500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.9000850186056944e-05,
+      "loss": 1.2314,
+      "step": 4059000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.899875022049638e-05,
+      "loss": 1.2266,
+      "step": 4059500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.899665445486694e-05,
+      "loss": 1.2373,
+      "step": 4060000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.899455448930638e-05,
+      "loss": 1.2424,
+      "step": 4060500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.899245872367693e-05,
+      "loss": 1.2053,
+      "step": 4061000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.8990358758116365e-05,
+      "loss": 1.1964,
+      "step": 4061500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.8988258792555806e-05,
+      "loss": 1.2406,
+      "step": 4062000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.898615882699524e-05,
+      "loss": 1.208,
+      "step": 4062500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.898405886143467e-05,
+      "loss": 1.2105,
+      "step": 4063000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.898195889587411e-05,
+      "loss": 1.2196,
+      "step": 4063500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.897985893031354e-05,
+      "loss": 1.2089,
+      "step": 4064000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.897775896475297e-05,
+      "loss": 1.232,
+      "step": 4064500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.897565899919241e-05,
+      "loss": 1.2286,
+      "step": 4065000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.8973559033631847e-05,
+      "loss": 1.2042,
+      "step": 4065500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.897146326800241e-05,
+      "loss": 1.2156,
+      "step": 4066000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.8969363302441834e-05,
+      "loss": 1.2196,
+      "step": 4066500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.8967263336881274e-05,
+      "loss": 1.1966,
+      "step": 4067000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.896516337132071e-05,
+      "loss": 1.2194,
+      "step": 4067500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.896306340576014e-05,
+      "loss": 1.2137,
+      "step": 4068000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.896096344019958e-05,
+      "loss": 1.209,
+      "step": 4068500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.8958863474639014e-05,
+      "loss": 1.2157,
+      "step": 4069000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.895676350907845e-05,
+      "loss": 1.2316,
+      "step": 4069500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.895466774344901e-05,
+      "loss": 1.23,
+      "step": 4070000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.895256777788844e-05,
+      "loss": 1.2541,
+      "step": 4070500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.8950467812327875e-05,
+      "loss": 1.2285,
+      "step": 4071000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.894837204669843e-05,
+      "loss": 1.2466,
+      "step": 4071500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.894627208113787e-05,
+      "loss": 1.2337,
+      "step": 4072000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.89441721155773e-05,
+      "loss": 1.2221,
+      "step": 4072500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.8942072150016736e-05,
+      "loss": 1.2102,
+      "step": 4073000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.8939972184456176e-05,
+      "loss": 1.2122,
+      "step": 4073500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.893787221889561e-05,
+      "loss": 1.2219,
+      "step": 4074000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.893577225333504e-05,
+      "loss": 1.2086,
+      "step": 4074500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.893367228777448e-05,
+      "loss": 1.2127,
+      "step": 4075000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.893157652214504e-05,
+      "loss": 1.1915,
+      "step": 4075500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.892947655658447e-05,
+      "loss": 1.2305,
+      "step": 4076000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.892737659102391e-05,
+      "loss": 1.2378,
+      "step": 4076500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.8925276625463344e-05,
+      "loss": 1.2216,
+      "step": 4077000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.89231808598339e-05,
+      "loss": 1.1979,
+      "step": 4077500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.892108089427333e-05,
+      "loss": 1.2064,
+      "step": 4078000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.891898092871277e-05,
+      "loss": 1.2332,
+      "step": 4078500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.8916885163083325e-05,
+      "loss": 1.2236,
+      "step": 4079000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.891478519752276e-05,
+      "loss": 1.2301,
+      "step": 4079500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.891268523196219e-05,
+      "loss": 1.2126,
+      "step": 4080000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.891058526640163e-05,
+      "loss": 1.2311,
+      "step": 4080500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.8908485300841065e-05,
+      "loss": 1.2463,
+      "step": 4081000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.89063853352805e-05,
+      "loss": 1.2165,
+      "step": 4081500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.890428536971994e-05,
+      "loss": 1.2168,
+      "step": 4082000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.890218540415937e-05,
+      "loss": 1.2189,
+      "step": 4082500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.8900089638529926e-05,
+      "loss": 1.2384,
+      "step": 4083000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.8897989672969366e-05,
+      "loss": 1.2113,
+      "step": 4083500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.88958897074088e-05,
+      "loss": 1.2245,
+      "step": 4084000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.889378974184823e-05,
+      "loss": 1.1906,
+      "step": 4084500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.8891689776287674e-05,
+      "loss": 1.2169,
+      "step": 4085000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.888958981072711e-05,
+      "loss": 1.2146,
+      "step": 4085500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.8887489845166534e-05,
+      "loss": 1.2385,
+      "step": 4086000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.8885389879605974e-05,
+      "loss": 1.2286,
+      "step": 4086500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.8883294113976534e-05,
+      "loss": 1.2097,
+      "step": 4087000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.888119834834709e-05,
+      "loss": 1.2188,
+      "step": 4087500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.887909838278652e-05,
+      "loss": 1.2236,
+      "step": 4088000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.8876998417225955e-05,
+      "loss": 1.2404,
+      "step": 4088500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.8874898451665395e-05,
+      "loss": 1.1954,
+      "step": 4089000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.887279848610483e-05,
+      "loss": 1.2449,
+      "step": 4089500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.887069852054426e-05,
+      "loss": 1.2372,
+      "step": 4090000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.88685985549837e-05,
+      "loss": 1.2379,
+      "step": 4090500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.8866502789354256e-05,
+      "loss": 1.2335,
+      "step": 4091000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.886440282379369e-05,
+      "loss": 1.244,
+      "step": 4091500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.886230285823313e-05,
+      "loss": 1.2322,
+      "step": 4092000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.886020289267256e-05,
+      "loss": 1.2258,
+      "step": 4092500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.8858102927111996e-05,
+      "loss": 1.2326,
+      "step": 4093000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.885600296155143e-05,
+      "loss": 1.2122,
+      "step": 4093500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.885390719592199e-05,
+      "loss": 1.22,
+      "step": 4094000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.8851807230361424e-05,
+      "loss": 1.2125,
+      "step": 4094500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.884970726480086e-05,
+      "loss": 1.2113,
+      "step": 4095000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.884760729924029e-05,
+      "loss": 1.1873,
+      "step": 4095500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.8845507333679724e-05,
+      "loss": 1.2142,
+      "step": 4096000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.8843411568050284e-05,
+      "loss": 1.211,
+      "step": 4096500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.884131160248972e-05,
+      "loss": 1.2355,
+      "step": 4097000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.883921163692916e-05,
+      "loss": 1.2273,
+      "step": 4097500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.8837111671368585e-05,
+      "loss": 1.218,
+      "step": 4098000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.8835015905739145e-05,
+      "loss": 1.2227,
+      "step": 4098500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.8832915940178585e-05,
+      "loss": 1.1832,
+      "step": 4099000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.883081597461802e-05,
+      "loss": 1.2112,
+      "step": 4099500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.882871600905745e-05,
+      "loss": 1.2406,
+      "step": 4100000
+    },
+    {
+      "epoch": 2.46,
+      "eval_loss": 1.1778119802474976,
+      "eval_runtime": 1105.3868,
+      "eval_samples_per_second": 476.503,
+      "eval_steps_per_second": 79.417,
+      "step": 4100000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.8826616043496886e-05,
+      "loss": 1.2149,
+      "step": 4100500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.8824520277867446e-05,
+      "loss": 1.2216,
+      "step": 4101000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.882242031230688e-05,
+      "loss": 1.2,
+      "step": 4101500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.882032034674631e-05,
+      "loss": 1.2241,
+      "step": 4102000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.881822038118575e-05,
+      "loss": 1.2425,
+      "step": 4102500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.881612041562518e-05,
+      "loss": 1.2225,
+      "step": 4103000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.881402464999574e-05,
+      "loss": 1.1942,
+      "step": 4103500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.8811924684435174e-05,
+      "loss": 1.2254,
+      "step": 4104000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.8809824718874614e-05,
+      "loss": 1.2022,
+      "step": 4104500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.880772475331404e-05,
+      "loss": 1.2268,
+      "step": 4105000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.880562478775348e-05,
+      "loss": 1.2162,
+      "step": 4105500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.8803524822192914e-05,
+      "loss": 1.207,
+      "step": 4106000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.8801429056563475e-05,
+      "loss": 1.216,
+      "step": 4106500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.879932909100291e-05,
+      "loss": 1.2357,
+      "step": 4107000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.879722912544234e-05,
+      "loss": 1.1981,
+      "step": 4107500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.8795129159881775e-05,
+      "loss": 1.2308,
+      "step": 4108000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.879302919432121e-05,
+      "loss": 1.2108,
+      "step": 4108500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.879092922876065e-05,
+      "loss": 1.205,
+      "step": 4109000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.878882926320008e-05,
+      "loss": 1.2105,
+      "step": 4109500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.8786729297639516e-05,
+      "loss": 1.2176,
+      "step": 4110000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.878463353201007e-05,
+      "loss": 1.2359,
+      "step": 4110500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.878253356644951e-05,
+      "loss": 1.2142,
+      "step": 4111000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.878043360088894e-05,
+      "loss": 1.2104,
+      "step": 4111500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.8778333635328376e-05,
+      "loss": 1.2116,
+      "step": 4112000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.877623786969894e-05,
+      "loss": 1.2446,
+      "step": 4112500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.877413790413837e-05,
+      "loss": 1.2482,
+      "step": 4113000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.8772037938577804e-05,
+      "loss": 1.2035,
+      "step": 4113500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.8769937973017244e-05,
+      "loss": 1.195,
+      "step": 4114000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.876783800745668e-05,
+      "loss": 1.2472,
+      "step": 4114500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.876574224182723e-05,
+      "loss": 1.2413,
+      "step": 4115000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.8763642276266664e-05,
+      "loss": 1.2114,
+      "step": 4115500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.8761542310706105e-05,
+      "loss": 1.2233,
+      "step": 4116000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.875944234514554e-05,
+      "loss": 1.1913,
+      "step": 4116500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.875734657951609e-05,
+      "loss": 1.2259,
+      "step": 4117000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.8755246613955525e-05,
+      "loss": 1.2384,
+      "step": 4117500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.8753146648394965e-05,
+      "loss": 1.2053,
+      "step": 4118000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.87510466828344e-05,
+      "loss": 1.2397,
+      "step": 4118500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.874895091720496e-05,
+      "loss": 1.2195,
+      "step": 4119000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.874685095164439e-05,
+      "loss": 1.2145,
+      "step": 4119500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.8744750986083826e-05,
+      "loss": 1.2297,
+      "step": 4120000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.8742655220454386e-05,
+      "loss": 1.1953,
+      "step": 4120500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.874055525489382e-05,
+      "loss": 1.1861,
+      "step": 4121000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.873845528933326e-05,
+      "loss": 1.225,
+      "step": 4121500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.873635532377269e-05,
+      "loss": 1.2175,
+      "step": 4122000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.873425955814325e-05,
+      "loss": 1.2259,
+      "step": 4122500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.873215959258268e-05,
+      "loss": 1.2244,
+      "step": 4123000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.873005962702212e-05,
+      "loss": 1.1998,
+      "step": 4123500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.8727959661461554e-05,
+      "loss": 1.232,
+      "step": 4124000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.872585969590098e-05,
+      "loss": 1.2588,
+      "step": 4124500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.872375973034042e-05,
+      "loss": 1.2015,
+      "step": 4125000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.8721659764779855e-05,
+      "loss": 1.2123,
+      "step": 4125500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.871955979921929e-05,
+      "loss": 1.1864,
+      "step": 4126000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.871745983365873e-05,
+      "loss": 1.2309,
+      "step": 4126500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.871535986809816e-05,
+      "loss": 1.2245,
+      "step": 4127000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.8713264102468715e-05,
+      "loss": 1.2114,
+      "step": 4127500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.8711164136908156e-05,
+      "loss": 1.2328,
+      "step": 4128000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.870906417134759e-05,
+      "loss": 1.2302,
+      "step": 4128500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.870696420578702e-05,
+      "loss": 1.2362,
+      "step": 4129000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.870486424022646e-05,
+      "loss": 1.2266,
+      "step": 4129500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.8702768474597016e-05,
+      "loss": 1.2398,
+      "step": 4130000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.870066850903645e-05,
+      "loss": 1.2278,
+      "step": 4130500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.869856854347588e-05,
+      "loss": 1.202,
+      "step": 4131000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.8696468577915324e-05,
+      "loss": 1.1953,
+      "step": 4131500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.869436861235476e-05,
+      "loss": 1.2555,
+      "step": 4132000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.869226864679419e-05,
+      "loss": 1.2192,
+      "step": 4132500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.869016868123363e-05,
+      "loss": 1.1829,
+      "step": 4133000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.8688072915604184e-05,
+      "loss": 1.2021,
+      "step": 4133500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.868597295004362e-05,
+      "loss": 1.1873,
+      "step": 4134000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.868387298448305e-05,
+      "loss": 1.2125,
+      "step": 4134500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.868177301892249e-05,
+      "loss": 1.2164,
+      "step": 4135000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.8679673053361925e-05,
+      "loss": 1.2382,
+      "step": 4135500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.867757728773248e-05,
+      "loss": 1.2176,
+      "step": 4136000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.867547732217192e-05,
+      "loss": 1.24,
+      "step": 4136500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.867337735661135e-05,
+      "loss": 1.2383,
+      "step": 4137000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.8671277391050786e-05,
+      "loss": 1.2185,
+      "step": 4137500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.8669177425490226e-05,
+      "loss": 1.2171,
+      "step": 4138000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.866708165986078e-05,
+      "loss": 1.2397,
+      "step": 4138500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.866498169430021e-05,
+      "loss": 1.1875,
+      "step": 4139000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.8662881728739646e-05,
+      "loss": 1.2151,
+      "step": 4139500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.8660781763179087e-05,
+      "loss": 1.2395,
+      "step": 4140000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.865868179761852e-05,
+      "loss": 1.2268,
+      "step": 4140500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.8656586031989074e-05,
+      "loss": 1.2022,
+      "step": 4141000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.8654486066428514e-05,
+      "loss": 1.2351,
+      "step": 4141500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.865238610086795e-05,
+      "loss": 1.266,
+      "step": 4142000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.865028613530738e-05,
+      "loss": 1.2249,
+      "step": 4142500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.864818616974682e-05,
+      "loss": 1.252,
+      "step": 4143000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.8646090404117375e-05,
+      "loss": 1.2412,
+      "step": 4143500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.864399043855681e-05,
+      "loss": 1.2271,
+      "step": 4144000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.864189047299624e-05,
+      "loss": 1.2278,
+      "step": 4144500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.863979050743568e-05,
+      "loss": 1.2204,
+      "step": 4145000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.8637694741806235e-05,
+      "loss": 1.2233,
+      "step": 4145500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.863559477624567e-05,
+      "loss": 1.2159,
+      "step": 4146000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.86334948106851e-05,
+      "loss": 1.2006,
+      "step": 4146500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.863139484512454e-05,
+      "loss": 1.2104,
+      "step": 4147000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.8629294879563976e-05,
+      "loss": 1.2443,
+      "step": 4147500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.862719491400341e-05,
+      "loss": 1.1978,
+      "step": 4148000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.862509914837397e-05,
+      "loss": 1.217,
+      "step": 4148500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.86229991828134e-05,
+      "loss": 1.2021,
+      "step": 4149000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.862089921725284e-05,
+      "loss": 1.1947,
+      "step": 4149500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.861879925169228e-05,
+      "loss": 1.2162,
+      "step": 4150000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.861669928613171e-05,
+      "loss": 1.2673,
+      "step": 4150500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.8614603520502264e-05,
+      "loss": 1.2133,
+      "step": 4151000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.86125035549417e-05,
+      "loss": 1.2416,
+      "step": 4151500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.861040358938114e-05,
+      "loss": 1.2186,
+      "step": 4152000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.860830362382057e-05,
+      "loss": 1.2434,
+      "step": 4152500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.8606207858191125e-05,
+      "loss": 1.2165,
+      "step": 4153000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.860410789263056e-05,
+      "loss": 1.2181,
+      "step": 4153500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.860201212700111e-05,
+      "loss": 1.2138,
+      "step": 4154000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.859991216144055e-05,
+      "loss": 1.2212,
+      "step": 4154500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.8597812195879985e-05,
+      "loss": 1.1922,
+      "step": 4155000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.8595712230319426e-05,
+      "loss": 1.2299,
+      "step": 4155500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.859361226475886e-05,
+      "loss": 1.2106,
+      "step": 4156000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.859151229919829e-05,
+      "loss": 1.2559,
+      "step": 4156500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.858941233363773e-05,
+      "loss": 1.2088,
+      "step": 4157000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.8587312368077166e-05,
+      "loss": 1.221,
+      "step": 4157500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.85852124025166e-05,
+      "loss": 1.2206,
+      "step": 4158000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.858311243695603e-05,
+      "loss": 1.2181,
+      "step": 4158500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.8581012471395467e-05,
+      "loss": 1.2389,
+      "step": 4159000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.85789125058349e-05,
+      "loss": 1.2372,
+      "step": 4159500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.857681674020546e-05,
+      "loss": 1.2197,
+      "step": 4160000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.85747167746449e-05,
+      "loss": 1.2104,
+      "step": 4160500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.857261680908433e-05,
+      "loss": 1.2269,
+      "step": 4161000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.857052104345489e-05,
+      "loss": 1.2266,
+      "step": 4161500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.856842107789432e-05,
+      "loss": 1.2125,
+      "step": 4162000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.856632111233376e-05,
+      "loss": 1.2356,
+      "step": 4162500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.856422114677319e-05,
+      "loss": 1.2055,
+      "step": 4163000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.856212118121263e-05,
+      "loss": 1.2046,
+      "step": 4163500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.856002121565206e-05,
+      "loss": 1.2099,
+      "step": 4164000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.855792545002262e-05,
+      "loss": 1.2275,
+      "step": 4164500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.8555825484462056e-05,
+      "loss": 1.212,
+      "step": 4165000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.855372551890149e-05,
+      "loss": 1.2234,
+      "step": 4165500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.855162555334092e-05,
+      "loss": 1.2288,
+      "step": 4166000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.8549525587780356e-05,
+      "loss": 1.2324,
+      "step": 4166500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.8547429822150916e-05,
+      "loss": 1.2201,
+      "step": 4167000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.8545329856590357e-05,
+      "loss": 1.2302,
+      "step": 4167500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.854322989102978e-05,
+      "loss": 1.2271,
+      "step": 4168000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.854112992546922e-05,
+      "loss": 1.2321,
+      "step": 4168500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.853902995990866e-05,
+      "loss": 1.2301,
+      "step": 4169000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.853693419427922e-05,
+      "loss": 1.2289,
+      "step": 4169500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.853483842864977e-05,
+      "loss": 1.211,
+      "step": 4170000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.8532738463089204e-05,
+      "loss": 1.2278,
+      "step": 4170500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.8530638497528645e-05,
+      "loss": 1.2107,
+      "step": 4171000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.852853853196808e-05,
+      "loss": 1.229,
+      "step": 4171500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.852643856640751e-05,
+      "loss": 1.2183,
+      "step": 4172000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.8524338600846945e-05,
+      "loss": 1.1871,
+      "step": 4172500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.852223863528638e-05,
+      "loss": 1.2202,
+      "step": 4173000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.852013866972581e-05,
+      "loss": 1.2457,
+      "step": 4173500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.851803870416525e-05,
+      "loss": 1.2156,
+      "step": 4174000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.851594293853581e-05,
+      "loss": 1.1957,
+      "step": 4174500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.851384297297524e-05,
+      "loss": 1.2108,
+      "step": 4175000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.851174300741467e-05,
+      "loss": 1.206,
+      "step": 4175500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.850964304185411e-05,
+      "loss": 1.1935,
+      "step": 4176000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.8507543076293546e-05,
+      "loss": 1.2108,
+      "step": 4176500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.850544311073298e-05,
+      "loss": 1.2164,
+      "step": 4177000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.850334734510354e-05,
+      "loss": 1.2322,
+      "step": 4177500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.8501247379542973e-05,
+      "loss": 1.2391,
+      "step": 4178000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.849914741398241e-05,
+      "loss": 1.2102,
+      "step": 4178500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.849704744842185e-05,
+      "loss": 1.2259,
+      "step": 4179000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.849494748286128e-05,
+      "loss": 1.2195,
+      "step": 4179500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.8492847517300714e-05,
+      "loss": 1.2088,
+      "step": 4180000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.8490747551740154e-05,
+      "loss": 1.2214,
+      "step": 4180500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.848864758617959e-05,
+      "loss": 1.2155,
+      "step": 4181000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.848655182055014e-05,
+      "loss": 1.2015,
+      "step": 4181500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.8484451854989575e-05,
+      "loss": 1.1976,
+      "step": 4182000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.8482351889429015e-05,
+      "loss": 1.2003,
+      "step": 4182500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.848025192386845e-05,
+      "loss": 1.2114,
+      "step": 4183000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.8478156158239e-05,
+      "loss": 1.2435,
+      "step": 4183500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.8476056192678436e-05,
+      "loss": 1.1696,
+      "step": 4184000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.8473956227117876e-05,
+      "loss": 1.2459,
+      "step": 4184500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.847185626155731e-05,
+      "loss": 1.2216,
+      "step": 4185000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.846976049592786e-05,
+      "loss": 1.2045,
+      "step": 4185500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.84676605303673e-05,
+      "loss": 1.2306,
+      "step": 4186000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.8465560564806737e-05,
+      "loss": 1.2552,
+      "step": 4186500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.846346479917729e-05,
+      "loss": 1.2404,
+      "step": 4187000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.8461364833616724e-05,
+      "loss": 1.2195,
+      "step": 4187500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.8459264868056164e-05,
+      "loss": 1.2228,
+      "step": 4188000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.84571649024956e-05,
+      "loss": 1.2271,
+      "step": 4188500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.845506493693503e-05,
+      "loss": 1.2224,
+      "step": 4189000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.845296497137447e-05,
+      "loss": 1.2429,
+      "step": 4189500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.8450865005813904e-05,
+      "loss": 1.2272,
+      "step": 4190000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.844876504025334e-05,
+      "loss": 1.2059,
+      "step": 4190500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.844666927462389e-05,
+      "loss": 1.2168,
+      "step": 4191000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.844456930906333e-05,
+      "loss": 1.2083,
+      "step": 4191500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.8442469343502765e-05,
+      "loss": 1.1967,
+      "step": 4192000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.84403693779422e-05,
+      "loss": 1.2133,
+      "step": 4192500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.843827781224388e-05,
+      "loss": 1.2022,
+      "step": 4193000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.843617784668332e-05,
+      "loss": 1.2183,
+      "step": 4193500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.8434077881122746e-05,
+      "loss": 1.2195,
+      "step": 4194000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.843197791556218e-05,
+      "loss": 1.2081,
+      "step": 4194500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.842987795000162e-05,
+      "loss": 1.2311,
+      "step": 4195000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.842777798444105e-05,
+      "loss": 1.1919,
+      "step": 4195500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.8425678018880487e-05,
+      "loss": 1.2323,
+      "step": 4196000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.842357805331993e-05,
+      "loss": 1.184,
+      "step": 4196500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.842147808775936e-05,
+      "loss": 1.196,
+      "step": 4197000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.8419378122198794e-05,
+      "loss": 1.2,
+      "step": 4197500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.8417278156638234e-05,
+      "loss": 1.2245,
+      "step": 4198000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.841517819107767e-05,
+      "loss": 1.2268,
+      "step": 4198500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.84130782255171e-05,
+      "loss": 1.1826,
+      "step": 4199000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.8410982459887654e-05,
+      "loss": 1.218,
+      "step": 4199500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.8408882494327095e-05,
+      "loss": 1.1937,
+      "step": 4200000
+    },
+    {
+      "epoch": 2.52,
+      "eval_loss": 1.1766777038574219,
+      "eval_runtime": 1098.718,
+      "eval_samples_per_second": 479.395,
+      "eval_steps_per_second": 79.899,
+      "step": 4200000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.840678252876653e-05,
+      "loss": 1.2039,
+      "step": 4200500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.840468256320597e-05,
+      "loss": 1.2128,
+      "step": 4201000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.84025825976454e-05,
+      "loss": 1.1999,
+      "step": 4201500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.8400486832015955e-05,
+      "loss": 1.1987,
+      "step": 4202000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.839838686645539e-05,
+      "loss": 1.2,
+      "step": 4202500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.839628690089483e-05,
+      "loss": 1.22,
+      "step": 4203000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.839418693533426e-05,
+      "loss": 1.2402,
+      "step": 4203500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.8392086969773696e-05,
+      "loss": 1.2128,
+      "step": 4204000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.838999120414425e-05,
+      "loss": 1.1967,
+      "step": 4204500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.838789123858369e-05,
+      "loss": 1.2259,
+      "step": 4205000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.838579127302312e-05,
+      "loss": 1.205,
+      "step": 4205500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.838369130746256e-05,
+      "loss": 1.2285,
+      "step": 4206000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.8381591341902e-05,
+      "loss": 1.1938,
+      "step": 4206500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.8379491376341424e-05,
+      "loss": 1.2086,
+      "step": 4207000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.8377395610711984e-05,
+      "loss": 1.2319,
+      "step": 4207500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.8375295645151424e-05,
+      "loss": 1.2402,
+      "step": 4208000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.837319567959086e-05,
+      "loss": 1.2184,
+      "step": 4208500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.8371095714030284e-05,
+      "loss": 1.2045,
+      "step": 4209000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.8368995748469725e-05,
+      "loss": 1.2128,
+      "step": 4209500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.836689578290916e-05,
+      "loss": 1.2232,
+      "step": 4210000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.836480001727972e-05,
+      "loss": 1.2138,
+      "step": 4210500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.836270005171915e-05,
+      "loss": 1.2317,
+      "step": 4211000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.8360604286089705e-05,
+      "loss": 1.1962,
+      "step": 4211500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.8358504320529146e-05,
+      "loss": 1.2018,
+      "step": 4212000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.835640435496858e-05,
+      "loss": 1.23,
+      "step": 4212500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.835430438940801e-05,
+      "loss": 1.2023,
+      "step": 4213000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.835220862377857e-05,
+      "loss": 1.196,
+      "step": 4213500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.8350108658218006e-05,
+      "loss": 1.2055,
+      "step": 4214000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.834800869265744e-05,
+      "loss": 1.2194,
+      "step": 4214500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.834590872709688e-05,
+      "loss": 1.2162,
+      "step": 4215000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.8343808761536314e-05,
+      "loss": 1.2344,
+      "step": 4215500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.834170879597575e-05,
+      "loss": 1.229,
+      "step": 4216000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.833960883041518e-05,
+      "loss": 1.2174,
+      "step": 4216500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.8337508864854614e-05,
+      "loss": 1.225,
+      "step": 4217000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.833540889929405e-05,
+      "loss": 1.1931,
+      "step": 4217500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.833331313366461e-05,
+      "loss": 1.2053,
+      "step": 4218000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.833121316810404e-05,
+      "loss": 1.2225,
+      "step": 4218500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.8329113202543475e-05,
+      "loss": 1.2201,
+      "step": 4219000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.832701323698291e-05,
+      "loss": 1.2359,
+      "step": 4219500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.832491327142235e-05,
+      "loss": 1.1967,
+      "step": 4220000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.832281330586178e-05,
+      "loss": 1.1962,
+      "step": 4220500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.8320717540232335e-05,
+      "loss": 1.1961,
+      "step": 4221000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.8318617574671776e-05,
+      "loss": 1.2292,
+      "step": 4221500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.831651760911121e-05,
+      "loss": 1.2014,
+      "step": 4222000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.831441764355064e-05,
+      "loss": 1.1928,
+      "step": 4222500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.831231767799008e-05,
+      "loss": 1.2392,
+      "step": 4223000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.8310217712429516e-05,
+      "loss": 1.201,
+      "step": 4223500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.830811774686895e-05,
+      "loss": 1.2009,
+      "step": 4224000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.83060219812395e-05,
+      "loss": 1.2182,
+      "step": 4224500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.8303922015678944e-05,
+      "loss": 1.231,
+      "step": 4225000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.830182205011838e-05,
+      "loss": 1.2026,
+      "step": 4225500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.829972208455781e-05,
+      "loss": 1.2129,
+      "step": 4226000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.829762211899725e-05,
+      "loss": 1.2392,
+      "step": 4226500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.8295526353367804e-05,
+      "loss": 1.2036,
+      "step": 4227000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.829342638780724e-05,
+      "loss": 1.1836,
+      "step": 4227500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.829132642224667e-05,
+      "loss": 1.2339,
+      "step": 4228000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.828922645668611e-05,
+      "loss": 1.1989,
+      "step": 4228500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.8287126491125545e-05,
+      "loss": 1.2226,
+      "step": 4229000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.828502652556498e-05,
+      "loss": 1.2049,
+      "step": 4229500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.828292656000442e-05,
+      "loss": 1.2163,
+      "step": 4230000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.828083079437497e-05,
+      "loss": 1.2093,
+      "step": 4230500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.8278730828814406e-05,
+      "loss": 1.2233,
+      "step": 4231000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.8276630863253846e-05,
+      "loss": 1.1856,
+      "step": 4231500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.827453089769328e-05,
+      "loss": 1.222,
+      "step": 4232000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.827243093213271e-05,
+      "loss": 1.2037,
+      "step": 4232500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.827033096657215e-05,
+      "loss": 1.2209,
+      "step": 4233000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.8268235200942707e-05,
+      "loss": 1.2315,
+      "step": 4233500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.826613523538214e-05,
+      "loss": 1.2248,
+      "step": 4234000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.8264035269821574e-05,
+      "loss": 1.2221,
+      "step": 4234500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.8261935304261014e-05,
+      "loss": 1.1987,
+      "step": 4235000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.825983533870045e-05,
+      "loss": 1.1886,
+      "step": 4235500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.8257735373139874e-05,
+      "loss": 1.2181,
+      "step": 4236000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.8255635407579314e-05,
+      "loss": 1.2278,
+      "step": 4236500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.825353544201875e-05,
+      "loss": 1.2162,
+      "step": 4237000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.825143967638931e-05,
+      "loss": 1.1972,
+      "step": 4237500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.824933971082874e-05,
+      "loss": 1.2334,
+      "step": 4238000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.8247239745268175e-05,
+      "loss": 1.1983,
+      "step": 4238500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.824513977970761e-05,
+      "loss": 1.1974,
+      "step": 4239000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.824304401407817e-05,
+      "loss": 1.2372,
+      "step": 4239500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.824094824844872e-05,
+      "loss": 1.1997,
+      "step": 4240000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.823884828288816e-05,
+      "loss": 1.2289,
+      "step": 4240500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.8236748317327596e-05,
+      "loss": 1.2322,
+      "step": 4241000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.823464835176703e-05,
+      "loss": 1.2088,
+      "step": 4241500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.823254838620647e-05,
+      "loss": 1.1991,
+      "step": 4242000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.82304484206459e-05,
+      "loss": 1.2232,
+      "step": 4242500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.8228348455085337e-05,
+      "loss": 1.23,
+      "step": 4243000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.822625268945589e-05,
+      "loss": 1.2124,
+      "step": 4243500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.822415272389533e-05,
+      "loss": 1.2083,
+      "step": 4244000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.8222052758334764e-05,
+      "loss": 1.2382,
+      "step": 4244500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.82199527927742e-05,
+      "loss": 1.2191,
+      "step": 4245000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.821785282721363e-05,
+      "loss": 1.2124,
+      "step": 4245500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.8215752861653064e-05,
+      "loss": 1.2305,
+      "step": 4246000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.8213657096023625e-05,
+      "loss": 1.209,
+      "step": 4246500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.8211557130463065e-05,
+      "loss": 1.1887,
+      "step": 4247000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.82094571649025e-05,
+      "loss": 1.22,
+      "step": 4247500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.820736139927305e-05,
+      "loss": 1.2098,
+      "step": 4248000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.8205261433712485e-05,
+      "loss": 1.2315,
+      "step": 4248500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.8203161468151926e-05,
+      "loss": 1.199,
+      "step": 4249000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.820106150259136e-05,
+      "loss": 1.2178,
+      "step": 4249500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.819896153703079e-05,
+      "loss": 1.1936,
+      "step": 4250000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.8196861571470226e-05,
+      "loss": 1.1868,
+      "step": 4250500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.819476160590966e-05,
+      "loss": 1.2008,
+      "step": 4251000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.819266164034909e-05,
+      "loss": 1.2402,
+      "step": 4251500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.819056167478853e-05,
+      "loss": 1.2207,
+      "step": 4252000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.8188461709227966e-05,
+      "loss": 1.1873,
+      "step": 4252500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.81863617436674e-05,
+      "loss": 1.2003,
+      "step": 4253000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.818426177810684e-05,
+      "loss": 1.2167,
+      "step": 4253500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.8182166012477394e-05,
+      "loss": 1.1948,
+      "step": 4254000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.818006604691683e-05,
+      "loss": 1.2249,
+      "step": 4254500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.817796608135627e-05,
+      "loss": 1.2331,
+      "step": 4255000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.81758661157957e-05,
+      "loss": 1.2065,
+      "step": 4255500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.8173766150235134e-05,
+      "loss": 1.2026,
+      "step": 4256000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.817167038460569e-05,
+      "loss": 1.1534,
+      "step": 4256500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.816957041904513e-05,
+      "loss": 1.2397,
+      "step": 4257000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.816747045348456e-05,
+      "loss": 1.1747,
+      "step": 4257500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.8165370487923995e-05,
+      "loss": 1.2296,
+      "step": 4258000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.8163270522363435e-05,
+      "loss": 1.2031,
+      "step": 4258500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.816117475673399e-05,
+      "loss": 1.2304,
+      "step": 4259000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.815907479117342e-05,
+      "loss": 1.2106,
+      "step": 4259500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.815697482561286e-05,
+      "loss": 1.25,
+      "step": 4260000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.8154874860052296e-05,
+      "loss": 1.1865,
+      "step": 4260500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.815277909442285e-05,
+      "loss": 1.2227,
+      "step": 4261000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.815067912886228e-05,
+      "loss": 1.2178,
+      "step": 4261500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.814857916330172e-05,
+      "loss": 1.1893,
+      "step": 4262000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.814647919774116e-05,
+      "loss": 1.211,
+      "step": 4262500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.814437923218059e-05,
+      "loss": 1.2369,
+      "step": 4263000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.814227926662003e-05,
+      "loss": 1.2116,
+      "step": 4263500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.8140183500990584e-05,
+      "loss": 1.1834,
+      "step": 4264000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.813808353543002e-05,
+      "loss": 1.2044,
+      "step": 4264500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.813598356986945e-05,
+      "loss": 1.2249,
+      "step": 4265000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.8133887804240005e-05,
+      "loss": 1.2106,
+      "step": 4265500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.8131787838679445e-05,
+      "loss": 1.2204,
+      "step": 4266000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.812968787311888e-05,
+      "loss": 1.2171,
+      "step": 4266500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.812758790755832e-05,
+      "loss": 1.2397,
+      "step": 4267000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.812548794199775e-05,
+      "loss": 1.2259,
+      "step": 4267500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.8123387976437185e-05,
+      "loss": 1.1981,
+      "step": 4268000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.8121288010876626e-05,
+      "loss": 1.2315,
+      "step": 4268500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.811919224524718e-05,
+      "loss": 1.2061,
+      "step": 4269000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.811709227968661e-05,
+      "loss": 1.1928,
+      "step": 4269500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.8114992314126046e-05,
+      "loss": 1.2096,
+      "step": 4270000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.8112892348565486e-05,
+      "loss": 1.219,
+      "step": 4270500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.811079238300492e-05,
+      "loss": 1.2362,
+      "step": 4271000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.810869241744435e-05,
+      "loss": 1.2177,
+      "step": 4271500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.8106592451883794e-05,
+      "loss": 1.2053,
+      "step": 4272000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.810449248632322e-05,
+      "loss": 1.2097,
+      "step": 4272500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.8102392520762654e-05,
+      "loss": 1.2168,
+      "step": 4273000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.8100296755133214e-05,
+      "loss": 1.2175,
+      "step": 4273500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.8098196789572654e-05,
+      "loss": 1.2465,
+      "step": 4274000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.809609682401209e-05,
+      "loss": 1.1992,
+      "step": 4274500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.809399685845152e-05,
+      "loss": 1.2254,
+      "step": 4275000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.809190109282208e-05,
+      "loss": 1.2336,
+      "step": 4275500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.8089801127261515e-05,
+      "loss": 1.2255,
+      "step": 4276000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.808770116170095e-05,
+      "loss": 1.2097,
+      "step": 4276500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.808560119614039e-05,
+      "loss": 1.1769,
+      "step": 4277000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.8083501230579815e-05,
+      "loss": 1.1974,
+      "step": 4277500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.8081405464950376e-05,
+      "loss": 1.2469,
+      "step": 4278000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.807930969932093e-05,
+      "loss": 1.2071,
+      "step": 4278500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.807720973376036e-05,
+      "loss": 1.2128,
+      "step": 4279000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.80751097681998e-05,
+      "loss": 1.2164,
+      "step": 4279500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.8073009802639236e-05,
+      "loss": 1.1953,
+      "step": 4280000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.807090983707867e-05,
+      "loss": 1.2057,
+      "step": 4280500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.806880987151811e-05,
+      "loss": 1.2418,
+      "step": 4281000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.8066709905957544e-05,
+      "loss": 1.1931,
+      "step": 4281500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.806460994039698e-05,
+      "loss": 1.2296,
+      "step": 4282000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.806250997483641e-05,
+      "loss": 1.2235,
+      "step": 4282500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.8060410009275844e-05,
+      "loss": 1.2435,
+      "step": 4283000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.8058314243646404e-05,
+      "loss": 1.1818,
+      "step": 4283500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.8056214278085845e-05,
+      "loss": 1.2283,
+      "step": 4284000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.805411431252527e-05,
+      "loss": 1.1878,
+      "step": 4284500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.8052014346964705e-05,
+      "loss": 1.2149,
+      "step": 4285000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.8049918581335265e-05,
+      "loss": 1.2161,
+      "step": 4285500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.804782701563694e-05,
+      "loss": 1.2047,
+      "step": 4286000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.804572705007638e-05,
+      "loss": 1.248,
+      "step": 4286500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.804362708451581e-05,
+      "loss": 1.2102,
+      "step": 4287000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.8041527118955246e-05,
+      "loss": 1.2157,
+      "step": 4287500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.8039427153394686e-05,
+      "loss": 1.1973,
+      "step": 4288000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.803732718783412e-05,
+      "loss": 1.2231,
+      "step": 4288500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.803522722227355e-05,
+      "loss": 1.2103,
+      "step": 4289000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.803312725671299e-05,
+      "loss": 1.2212,
+      "step": 4289500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.803102729115243e-05,
+      "loss": 1.2224,
+      "step": 4290000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.802892732559186e-05,
+      "loss": 1.2023,
+      "step": 4290500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.80268273600313e-05,
+      "loss": 1.2175,
+      "step": 4291000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.802472739447073e-05,
+      "loss": 1.1829,
+      "step": 4291500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.802262742891016e-05,
+      "loss": 1.2137,
+      "step": 4292000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.802053166328072e-05,
+      "loss": 1.2476,
+      "step": 4292500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.801843169772016e-05,
+      "loss": 1.2172,
+      "step": 4293000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.8016331732159595e-05,
+      "loss": 1.2304,
+      "step": 4293500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.801423596653015e-05,
+      "loss": 1.209,
+      "step": 4294000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.801213600096958e-05,
+      "loss": 1.2139,
+      "step": 4294500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.801003603540902e-05,
+      "loss": 1.2214,
+      "step": 4295000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.8007936069848455e-05,
+      "loss": 1.191,
+      "step": 4295500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.800583610428789e-05,
+      "loss": 1.197,
+      "step": 4296000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.800373613872732e-05,
+      "loss": 1.2281,
+      "step": 4296500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.8001636173166756e-05,
+      "loss": 1.212,
+      "step": 4297000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7999536207606196e-05,
+      "loss": 1.2066,
+      "step": 4297500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.799743624204563e-05,
+      "loss": 1.232,
+      "step": 4298000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.799534047641619e-05,
+      "loss": 1.2079,
+      "step": 4298500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7993240510855616e-05,
+      "loss": 1.1922,
+      "step": 4299000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.799114054529506e-05,
+      "loss": 1.2116,
+      "step": 4299500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.798904057973449e-05,
+      "loss": 1.1956,
+      "step": 4300000
+    },
+    {
+      "epoch": 2.58,
+      "eval_loss": 1.1711117029190063,
+      "eval_runtime": 1097.9922,
+      "eval_samples_per_second": 479.712,
+      "eval_steps_per_second": 79.952,
+      "step": 4300000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7986940614173924e-05,
+      "loss": 1.2074,
+      "step": 4300500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7984844848544484e-05,
+      "loss": 1.239,
+      "step": 4301000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.798274488298392e-05,
+      "loss": 1.1959,
+      "step": 4301500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.798064491742335e-05,
+      "loss": 1.2282,
+      "step": 4302000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7978544951862784e-05,
+      "loss": 1.215,
+      "step": 4302500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7976444986302225e-05,
+      "loss": 1.2057,
+      "step": 4303000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.797434502074166e-05,
+      "loss": 1.217,
+      "step": 4303500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.797224505518109e-05,
+      "loss": 1.2254,
+      "step": 4304000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.797014508962053e-05,
+      "loss": 1.2297,
+      "step": 4304500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7968049323991085e-05,
+      "loss": 1.1851,
+      "step": 4305000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.796594935843052e-05,
+      "loss": 1.2299,
+      "step": 4305500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.796384939286996e-05,
+      "loss": 1.2062,
+      "step": 4306000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.796174942730939e-05,
+      "loss": 1.2471,
+      "step": 4306500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7959653661679946e-05,
+      "loss": 1.2162,
+      "step": 4307000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7957557896050506e-05,
+      "loss": 1.2207,
+      "step": 4307500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.795545793048994e-05,
+      "loss": 1.2126,
+      "step": 4308000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.795335796492937e-05,
+      "loss": 1.1924,
+      "step": 4308500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7951262199299934e-05,
+      "loss": 1.2228,
+      "step": 4309000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.794916223373937e-05,
+      "loss": 1.201,
+      "step": 4309500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.79470622681788e-05,
+      "loss": 1.209,
+      "step": 4310000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.794496230261824e-05,
+      "loss": 1.2004,
+      "step": 4310500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.794286233705767e-05,
+      "loss": 1.2303,
+      "step": 4311000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.794076237149711e-05,
+      "loss": 1.2122,
+      "step": 4311500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.793866240593654e-05,
+      "loss": 1.1962,
+      "step": 4312000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7936562440375975e-05,
+      "loss": 1.2311,
+      "step": 4312500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7934462474815415e-05,
+      "loss": 1.2239,
+      "step": 4313000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.793236670918597e-05,
+      "loss": 1.223,
+      "step": 4313500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.79302667436254e-05,
+      "loss": 1.1937,
+      "step": 4314000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7928166778064835e-05,
+      "loss": 1.2077,
+      "step": 4314500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7926066812504276e-05,
+      "loss": 1.2371,
+      "step": 4315000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.792397104687483e-05,
+      "loss": 1.248,
+      "step": 4315500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.792187108131426e-05,
+      "loss": 1.2262,
+      "step": 4316000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7919771115753696e-05,
+      "loss": 1.2082,
+      "step": 4316500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7917671150193136e-05,
+      "loss": 1.2298,
+      "step": 4317000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.791557118463257e-05,
+      "loss": 1.1988,
+      "step": 4317500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.791347541900312e-05,
+      "loss": 1.2064,
+      "step": 4318000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7911375453442564e-05,
+      "loss": 1.2163,
+      "step": 4318500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7909275487882e-05,
+      "loss": 1.1809,
+      "step": 4319000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.790717552232143e-05,
+      "loss": 1.2064,
+      "step": 4319500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.790507555676087e-05,
+      "loss": 1.2026,
+      "step": 4320000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7902975591200304e-05,
+      "loss": 1.2201,
+      "step": 4320500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.790087982557086e-05,
+      "loss": 1.2326,
+      "step": 4321000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.789877986001029e-05,
+      "loss": 1.2159,
+      "step": 4321500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.789667989444973e-05,
+      "loss": 1.2023,
+      "step": 4322000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7894579928889165e-05,
+      "loss": 1.2163,
+      "step": 4322500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.789248416325972e-05,
+      "loss": 1.2117,
+      "step": 4323000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.789038419769915e-05,
+      "loss": 1.2017,
+      "step": 4323500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.788828423213859e-05,
+      "loss": 1.2275,
+      "step": 4324000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7886184266578026e-05,
+      "loss": 1.217,
+      "step": 4324500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.788408430101746e-05,
+      "loss": 1.1756,
+      "step": 4325000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.788198853538802e-05,
+      "loss": 1.1987,
+      "step": 4325500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.787988856982745e-05,
+      "loss": 1.2193,
+      "step": 4326000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7877788604266886e-05,
+      "loss": 1.2398,
+      "step": 4326500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.787568863870633e-05,
+      "loss": 1.2238,
+      "step": 4327000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.787359287307688e-05,
+      "loss": 1.2035,
+      "step": 4327500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7871492907516314e-05,
+      "loss": 1.2041,
+      "step": 4328000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.786939294195575e-05,
+      "loss": 1.2587,
+      "step": 4328500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.786729297639519e-05,
+      "loss": 1.217,
+      "step": 4329000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.786519301083462e-05,
+      "loss": 1.2283,
+      "step": 4329500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7863093045274054e-05,
+      "loss": 1.2012,
+      "step": 4330000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.786099727964461e-05,
+      "loss": 1.1985,
+      "step": 4330500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.785889731408405e-05,
+      "loss": 1.2035,
+      "step": 4331000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.785679734852348e-05,
+      "loss": 1.2381,
+      "step": 4331500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.785469738296292e-05,
+      "loss": 1.2051,
+      "step": 4332000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7852597417402355e-05,
+      "loss": 1.2232,
+      "step": 4332500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.785049745184179e-05,
+      "loss": 1.238,
+      "step": 4333000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.784839748628123e-05,
+      "loss": 1.223,
+      "step": 4333500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.784629752072066e-05,
+      "loss": 1.1851,
+      "step": 4334000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7844201755091216e-05,
+      "loss": 1.2077,
+      "step": 4334500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.784210598946177e-05,
+      "loss": 1.2062,
+      "step": 4335000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.784001022383233e-05,
+      "loss": 1.2089,
+      "step": 4335500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.783791025827176e-05,
+      "loss": 1.2074,
+      "step": 4336000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7835810292711204e-05,
+      "loss": 1.2054,
+      "step": 4336500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.783371032715063e-05,
+      "loss": 1.2181,
+      "step": 4337000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7831610361590064e-05,
+      "loss": 1.2379,
+      "step": 4337500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7829510396029504e-05,
+      "loss": 1.2081,
+      "step": 4338000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.782741043046894e-05,
+      "loss": 1.1883,
+      "step": 4338500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.782531046490838e-05,
+      "loss": 1.1868,
+      "step": 4339000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.782321049934781e-05,
+      "loss": 1.2058,
+      "step": 4339500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7821110533787245e-05,
+      "loss": 1.1934,
+      "step": 4340000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.78190147681578e-05,
+      "loss": 1.2248,
+      "step": 4340500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.781691480259724e-05,
+      "loss": 1.2268,
+      "step": 4341000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.781481483703667e-05,
+      "loss": 1.2115,
+      "step": 4341500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7812714871476105e-05,
+      "loss": 1.2071,
+      "step": 4342000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7810614905915546e-05,
+      "loss": 1.1998,
+      "step": 4342500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.78085191402861e-05,
+      "loss": 1.232,
+      "step": 4343000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.780641917472553e-05,
+      "loss": 1.2164,
+      "step": 4343500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7804319209164966e-05,
+      "loss": 1.2002,
+      "step": 4344000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7802219243604406e-05,
+      "loss": 1.2062,
+      "step": 4344500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.780012347797496e-05,
+      "loss": 1.2022,
+      "step": 4345000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.779802351241439e-05,
+      "loss": 1.2143,
+      "step": 4345500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.7795923546853834e-05,
+      "loss": 1.2271,
+      "step": 4346000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.779382358129327e-05,
+      "loss": 1.2418,
+      "step": 4346500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.77917236157327e-05,
+      "loss": 1.2214,
+      "step": 4347000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.7789627850103254e-05,
+      "loss": 1.2291,
+      "step": 4347500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.7787532084473814e-05,
+      "loss": 1.2229,
+      "step": 4348000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.7785432118913255e-05,
+      "loss": 1.2125,
+      "step": 4348500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.778333215335268e-05,
+      "loss": 1.202,
+      "step": 4349000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.7781232187792115e-05,
+      "loss": 1.2179,
+      "step": 4349500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.7779132222231555e-05,
+      "loss": 1.1836,
+      "step": 4350000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.777703225667099e-05,
+      "loss": 1.1948,
+      "step": 4350500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.777493229111042e-05,
+      "loss": 1.2334,
+      "step": 4351000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.777283232554986e-05,
+      "loss": 1.2038,
+      "step": 4351500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.7770732359989296e-05,
+      "loss": 1.1914,
+      "step": 4352000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.776863659435985e-05,
+      "loss": 1.2288,
+      "step": 4352500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.776653662879929e-05,
+      "loss": 1.2249,
+      "step": 4353000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.776443666323872e-05,
+      "loss": 1.2435,
+      "step": 4353500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.7762340897609276e-05,
+      "loss": 1.1967,
+      "step": 4354000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.776024093204871e-05,
+      "loss": 1.2522,
+      "step": 4354500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.775814096648815e-05,
+      "loss": 1.2142,
+      "step": 4355000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.7756041000927584e-05,
+      "loss": 1.2276,
+      "step": 4355500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.775394103536702e-05,
+      "loss": 1.2209,
+      "step": 4356000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.775184106980646e-05,
+      "loss": 1.2056,
+      "step": 4356500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.774974110424589e-05,
+      "loss": 1.231,
+      "step": 4357000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.7747641138685324e-05,
+      "loss": 1.2285,
+      "step": 4357500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.7745541173124764e-05,
+      "loss": 1.2244,
+      "step": 4358000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.77434412075642e-05,
+      "loss": 1.2046,
+      "step": 4358500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.7741341242003625e-05,
+      "loss": 1.203,
+      "step": 4359000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.7739241276443065e-05,
+      "loss": 1.1867,
+      "step": 4359500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.7737145510813625e-05,
+      "loss": 1.2079,
+      "step": 4360000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.773504974518418e-05,
+      "loss": 1.2033,
+      "step": 4360500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.773294977962361e-05,
+      "loss": 1.1957,
+      "step": 4361000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.773084981406305e-05,
+      "loss": 1.2122,
+      "step": 4361500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.7728749848502486e-05,
+      "loss": 1.1986,
+      "step": 4362000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.772664988294192e-05,
+      "loss": 1.1876,
+      "step": 4362500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.772455411731247e-05,
+      "loss": 1.1838,
+      "step": 4363000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.772245415175191e-05,
+      "loss": 1.2147,
+      "step": 4363500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.772035418619135e-05,
+      "loss": 1.2008,
+      "step": 4364000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.771825422063078e-05,
+      "loss": 1.2216,
+      "step": 4364500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.771615425507022e-05,
+      "loss": 1.1932,
+      "step": 4365000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.7714054289509654e-05,
+      "loss": 1.2096,
+      "step": 4365500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.771195432394909e-05,
+      "loss": 1.2179,
+      "step": 4366000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.770985435838852e-05,
+      "loss": 1.2099,
+      "step": 4366500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.770775859275908e-05,
+      "loss": 1.2148,
+      "step": 4367000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.7705662827129635e-05,
+      "loss": 1.2221,
+      "step": 4367500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.770356286156907e-05,
+      "loss": 1.2291,
+      "step": 4368000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.770146289600851e-05,
+      "loss": 1.239,
+      "step": 4368500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.769936293044794e-05,
+      "loss": 1.1933,
+      "step": 4369000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.7697262964887375e-05,
+      "loss": 1.2128,
+      "step": 4369500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.7695162999326815e-05,
+      "loss": 1.1933,
+      "step": 4370000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.769306303376625e-05,
+      "loss": 1.2263,
+      "step": 4370500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.76909672681368e-05,
+      "loss": 1.2228,
+      "step": 4371000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.7688867302576236e-05,
+      "loss": 1.2077,
+      "step": 4371500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.7686767337015676e-05,
+      "loss": 1.2177,
+      "step": 4372000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.768466737145511e-05,
+      "loss": 1.2111,
+      "step": 4372500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.768256740589454e-05,
+      "loss": 1.2148,
+      "step": 4373000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.76804716402651e-05,
+      "loss": 1.2335,
+      "step": 4373500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.767837167470454e-05,
+      "loss": 1.2015,
+      "step": 4374000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.767627170914397e-05,
+      "loss": 1.2122,
+      "step": 4374500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.7674171743583404e-05,
+      "loss": 1.2042,
+      "step": 4375000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.7672071778022844e-05,
+      "loss": 1.2212,
+      "step": 4375500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.76699760123934e-05,
+      "loss": 1.1926,
+      "step": 4376000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.766787604683283e-05,
+      "loss": 1.2014,
+      "step": 4376500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.766577608127227e-05,
+      "loss": 1.2006,
+      "step": 4377000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.7663676115711705e-05,
+      "loss": 1.2074,
+      "step": 4377500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.766158035008226e-05,
+      "loss": 1.1944,
+      "step": 4378000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.765948038452169e-05,
+      "loss": 1.205,
+      "step": 4378500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.765738041896113e-05,
+      "loss": 1.2292,
+      "step": 4379000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.7655280453400566e-05,
+      "loss": 1.245,
+      "step": 4379500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.765318048784e-05,
+      "loss": 1.2431,
+      "step": 4380000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.765108472221055e-05,
+      "loss": 1.2105,
+      "step": 4380500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.764898475664999e-05,
+      "loss": 1.215,
+      "step": 4381000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.7646884791089426e-05,
+      "loss": 1.2113,
+      "step": 4381500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.764478482552886e-05,
+      "loss": 1.2354,
+      "step": 4382000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.76426848599683e-05,
+      "loss": 1.1899,
+      "step": 4382500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.764058489440773e-05,
+      "loss": 1.2096,
+      "step": 4383000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.763848492884717e-05,
+      "loss": 1.2048,
+      "step": 4383500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.76363849632866e-05,
+      "loss": 1.2087,
+      "step": 4384000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.763428919765716e-05,
+      "loss": 1.1912,
+      "step": 4384500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.7632189232096594e-05,
+      "loss": 1.2125,
+      "step": 4385000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.763008926653603e-05,
+      "loss": 1.2182,
+      "step": 4385500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.762798930097546e-05,
+      "loss": 1.2146,
+      "step": 4386000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.762589353534602e-05,
+      "loss": 1.2512,
+      "step": 4386500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.7623793569785455e-05,
+      "loss": 1.1869,
+      "step": 4387000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.7621693604224895e-05,
+      "loss": 1.1998,
+      "step": 4387500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.761959363866432e-05,
+      "loss": 1.2329,
+      "step": 4388000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.7617502072966e-05,
+      "loss": 1.2008,
+      "step": 4388500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.7615402107405436e-05,
+      "loss": 1.2257,
+      "step": 4389000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.7613302141844876e-05,
+      "loss": 1.2047,
+      "step": 4389500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.761120217628431e-05,
+      "loss": 1.24,
+      "step": 4390000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.760910221072374e-05,
+      "loss": 1.2129,
+      "step": 4390500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.7607006445094296e-05,
+      "loss": 1.2061,
+      "step": 4391000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.760490647953374e-05,
+      "loss": 1.1795,
+      "step": 4391500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.760280651397317e-05,
+      "loss": 1.2074,
+      "step": 4392000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.7600706548412604e-05,
+      "loss": 1.2085,
+      "step": 4392500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.759861078278316e-05,
+      "loss": 1.1778,
+      "step": 4393000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.759651501715372e-05,
+      "loss": 1.2119,
+      "step": 4393500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.759441505159315e-05,
+      "loss": 1.1892,
+      "step": 4394000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.7592315086032584e-05,
+      "loss": 1.2175,
+      "step": 4394500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.7590215120472025e-05,
+      "loss": 1.2312,
+      "step": 4395000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.758811515491146e-05,
+      "loss": 1.213,
+      "step": 4395500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.758601938928202e-05,
+      "loss": 1.1981,
+      "step": 4396000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.7583919423721445e-05,
+      "loss": 1.2325,
+      "step": 4396500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.7581819458160885e-05,
+      "loss": 1.1897,
+      "step": 4397000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.757971949260032e-05,
+      "loss": 1.2175,
+      "step": 4397500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.757761952703975e-05,
+      "loss": 1.1963,
+      "step": 4398000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.757551956147919e-05,
+      "loss": 1.233,
+      "step": 4398500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.7573419595918626e-05,
+      "loss": 1.1935,
+      "step": 4399000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.757131963035806e-05,
+      "loss": 1.2238,
+      "step": 4399500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.75692196647975e-05,
+      "loss": 1.2251,
+      "step": 4400000
+    },
+    {
+      "epoch": 2.64,
+      "eval_loss": 1.1657322645187378,
+      "eval_runtime": 1099.1385,
+      "eval_samples_per_second": 479.212,
+      "eval_steps_per_second": 79.869,
+      "step": 4400000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.756711969923693e-05,
+      "loss": 1.2357,
+      "step": 4400500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.756501973367637e-05,
+      "loss": 1.2106,
+      "step": 4401000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.756291976811581e-05,
+      "loss": 1.2053,
+      "step": 4401500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.756082820241748e-05,
+      "loss": 1.2313,
+      "step": 4402000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.7558728236856914e-05,
+      "loss": 1.1984,
+      "step": 4402500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.755662827129635e-05,
+      "loss": 1.1957,
+      "step": 4403000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.755452830573579e-05,
+      "loss": 1.2084,
+      "step": 4403500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.755242834017522e-05,
+      "loss": 1.1872,
+      "step": 4404000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.7550328374614655e-05,
+      "loss": 1.2052,
+      "step": 4404500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.7548228409054095e-05,
+      "loss": 1.2295,
+      "step": 4405000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.754612844349353e-05,
+      "loss": 1.2171,
+      "step": 4405500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.754403267786408e-05,
+      "loss": 1.2185,
+      "step": 4406000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.7541932712303515e-05,
+      "loss": 1.2056,
+      "step": 4406500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.7539832746742956e-05,
+      "loss": 1.1981,
+      "step": 4407000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.753773278118239e-05,
+      "loss": 1.245,
+      "step": 4407500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.753563701555294e-05,
+      "loss": 1.1932,
+      "step": 4408000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.7533541249923496e-05,
+      "loss": 1.1845,
+      "step": 4408500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.7531441284362936e-05,
+      "loss": 1.1873,
+      "step": 4409000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.752934131880237e-05,
+      "loss": 1.2121,
+      "step": 4409500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.75272413532418e-05,
+      "loss": 1.2368,
+      "step": 4410000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.7525141387681244e-05,
+      "loss": 1.2022,
+      "step": 4410500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.752304982198292e-05,
+      "loss": 1.2097,
+      "step": 4411000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.752094985642236e-05,
+      "loss": 1.2103,
+      "step": 4411500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.751884989086179e-05,
+      "loss": 1.2152,
+      "step": 4412000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.7516749925301224e-05,
+      "loss": 1.2081,
+      "step": 4412500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.7514649959740665e-05,
+      "loss": 1.192,
+      "step": 4413000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.751254999418009e-05,
+      "loss": 1.2016,
+      "step": 4413500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.7510450028619525e-05,
+      "loss": 1.2182,
+      "step": 4414000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.7508350063058965e-05,
+      "loss": 1.1954,
+      "step": 4414500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.75062500974984e-05,
+      "loss": 1.2268,
+      "step": 4415000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.750415433186895e-05,
+      "loss": 1.2096,
+      "step": 4415500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.750205436630839e-05,
+      "loss": 1.1921,
+      "step": 4416000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.7499954400747826e-05,
+      "loss": 1.2256,
+      "step": 4416500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.749785443518726e-05,
+      "loss": 1.1651,
+      "step": 4417000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.749575866955782e-05,
+      "loss": 1.2143,
+      "step": 4417500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.749365870399725e-05,
+      "loss": 1.1977,
+      "step": 4418000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.7491558738436686e-05,
+      "loss": 1.2257,
+      "step": 4418500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.748945877287612e-05,
+      "loss": 1.2149,
+      "step": 4419000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.748736300724668e-05,
+      "loss": 1.2131,
+      "step": 4419500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.748526304168612e-05,
+      "loss": 1.214,
+      "step": 4420000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.748316307612555e-05,
+      "loss": 1.2003,
+      "step": 4420500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.748106311056498e-05,
+      "loss": 1.199,
+      "step": 4421000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.747896314500442e-05,
+      "loss": 1.2201,
+      "step": 4421500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.7476863179443854e-05,
+      "loss": 1.1687,
+      "step": 4422000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.7474767413814415e-05,
+      "loss": 1.2092,
+      "step": 4422500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.747266744825385e-05,
+      "loss": 1.1789,
+      "step": 4423000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.747056748269328e-05,
+      "loss": 1.2124,
+      "step": 4423500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.7468467517132715e-05,
+      "loss": 1.2179,
+      "step": 4424000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.7466367551572155e-05,
+      "loss": 1.2112,
+      "step": 4424500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.746426758601159e-05,
+      "loss": 1.224,
+      "step": 4425000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.746216762045102e-05,
+      "loss": 1.1806,
+      "step": 4425500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.7460071854821576e-05,
+      "loss": 1.1955,
+      "step": 4426000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.7457971889261016e-05,
+      "loss": 1.1677,
+      "step": 4426500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.745587192370045e-05,
+      "loss": 1.1853,
+      "step": 4427000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.745377195813988e-05,
+      "loss": 1.2243,
+      "step": 4427500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.745167199257932e-05,
+      "loss": 1.2365,
+      "step": 4428000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.744957202701876e-05,
+      "loss": 1.1986,
+      "step": 4428500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.744747206145819e-05,
+      "loss": 1.2228,
+      "step": 4429000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.744537209589763e-05,
+      "loss": 1.1789,
+      "step": 4429500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.7443272130337064e-05,
+      "loss": 1.2196,
+      "step": 4430000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.74411721647765e-05,
+      "loss": 1.1968,
+      "step": 4430500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.743907219921593e-05,
+      "loss": 1.2057,
+      "step": 4431000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.7436972233655364e-05,
+      "loss": 1.2077,
+      "step": 4431500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.7434876468025925e-05,
+      "loss": 1.1919,
+      "step": 4432000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.7432776502465365e-05,
+      "loss": 1.2149,
+      "step": 4432500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.743068073683592e-05,
+      "loss": 1.2114,
+      "step": 4433000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.742858077127535e-05,
+      "loss": 1.2244,
+      "step": 4433500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.7426480805714785e-05,
+      "loss": 1.1967,
+      "step": 4434000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.7424380840154226e-05,
+      "loss": 1.2102,
+      "step": 4434500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.742228087459366e-05,
+      "loss": 1.2152,
+      "step": 4435000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.7420180909033086e-05,
+      "loss": 1.2235,
+      "step": 4435500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.7418080943472526e-05,
+      "loss": 1.2207,
+      "step": 4436000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.741598097791196e-05,
+      "loss": 1.2132,
+      "step": 4436500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.741388521228252e-05,
+      "loss": 1.221,
+      "step": 4437000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.741178524672195e-05,
+      "loss": 1.2138,
+      "step": 4437500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.740968528116139e-05,
+      "loss": 1.221,
+      "step": 4438000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.740758951553195e-05,
+      "loss": 1.2102,
+      "step": 4438500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.740548954997138e-05,
+      "loss": 1.2038,
+      "step": 4439000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.740338958441082e-05,
+      "loss": 1.2192,
+      "step": 4439500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.7401289618850254e-05,
+      "loss": 1.2068,
+      "step": 4440000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.739918965328968e-05,
+      "loss": 1.2141,
+      "step": 4440500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.739708968772912e-05,
+      "loss": 1.1971,
+      "step": 4441000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.7394989722168555e-05,
+      "loss": 1.2089,
+      "step": 4441500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.739288975660799e-05,
+      "loss": 1.1627,
+      "step": 4442000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.739079399097854e-05,
+      "loss": 1.2189,
+      "step": 4442500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.738869402541798e-05,
+      "loss": 1.1798,
+      "step": 4443000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.7386594059857415e-05,
+      "loss": 1.2182,
+      "step": 4443500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.738449409429685e-05,
+      "loss": 1.2382,
+      "step": 4444000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.738240252859853e-05,
+      "loss": 1.2126,
+      "step": 4444500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.738030256303797e-05,
+      "loss": 1.1834,
+      "step": 4445000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.73782025974774e-05,
+      "loss": 1.1854,
+      "step": 4445500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.7376102631916836e-05,
+      "loss": 1.192,
+      "step": 4446000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.7374002666356277e-05,
+      "loss": 1.2063,
+      "step": 4446500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.737190690072683e-05,
+      "loss": 1.2056,
+      "step": 4447000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.7369806935166264e-05,
+      "loss": 1.1957,
+      "step": 4447500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.73677069696057e-05,
+      "loss": 1.2347,
+      "step": 4448000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.736560700404514e-05,
+      "loss": 1.2208,
+      "step": 4448500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.736350703848457e-05,
+      "loss": 1.2232,
+      "step": 4449000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.7361407072924004e-05,
+      "loss": 1.1973,
+      "step": 4449500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.735930710736344e-05,
+      "loss": 1.1996,
+      "step": 4450000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.735720714180287e-05,
+      "loss": 1.2181,
+      "step": 4450500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.735511137617343e-05,
+      "loss": 1.2032,
+      "step": 4451000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.7353011410612865e-05,
+      "loss": 1.2197,
+      "step": 4451500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.73509114450523e-05,
+      "loss": 1.2307,
+      "step": 4452000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.734881147949173e-05,
+      "loss": 1.2136,
+      "step": 4452500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.734671571386229e-05,
+      "loss": 1.1857,
+      "step": 4453000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.7344619948232846e-05,
+      "loss": 1.2111,
+      "step": 4453500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.7342519982672286e-05,
+      "loss": 1.2055,
+      "step": 4454000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.734042001711172e-05,
+      "loss": 1.1874,
+      "step": 4454500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.733832005155115e-05,
+      "loss": 1.2045,
+      "step": 4455000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.733622008599059e-05,
+      "loss": 1.2073,
+      "step": 4455500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.733412012043003e-05,
+      "loss": 1.2253,
+      "step": 4456000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.733202435480058e-05,
+      "loss": 1.1832,
+      "step": 4456500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.7329924389240014e-05,
+      "loss": 1.2181,
+      "step": 4457000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.7327824423679454e-05,
+      "loss": 1.181,
+      "step": 4457500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.732572445811889e-05,
+      "loss": 1.2172,
+      "step": 4458000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.732362449255832e-05,
+      "loss": 1.2273,
+      "step": 4458500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.732152872692888e-05,
+      "loss": 1.2111,
+      "step": 4459000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.7319428761368315e-05,
+      "loss": 1.2224,
+      "step": 4459500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.731732879580775e-05,
+      "loss": 1.1904,
+      "step": 4460000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.731522883024719e-05,
+      "loss": 1.2094,
+      "step": 4460500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.731312886468662e-05,
+      "loss": 1.1945,
+      "step": 4461000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.731102889912605e-05,
+      "loss": 1.2206,
+      "step": 4461500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.730893313349661e-05,
+      "loss": 1.1879,
+      "step": 4462000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.730683316793605e-05,
+      "loss": 1.2016,
+      "step": 4462500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.730473320237548e-05,
+      "loss": 1.2392,
+      "step": 4463000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.7302637436746036e-05,
+      "loss": 1.1793,
+      "step": 4463500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.730053747118547e-05,
+      "loss": 1.1998,
+      "step": 4464000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.729843750562491e-05,
+      "loss": 1.2018,
+      "step": 4464500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.729633754006434e-05,
+      "loss": 1.2131,
+      "step": 4465000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.729423757450378e-05,
+      "loss": 1.1859,
+      "step": 4465500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.729213760894322e-05,
+      "loss": 1.2183,
+      "step": 4466000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.7290037643382644e-05,
+      "loss": 1.213,
+      "step": 4466500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.7287937677822084e-05,
+      "loss": 1.2292,
+      "step": 4467000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.728583771226152e-05,
+      "loss": 1.19,
+      "step": 4467500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.728373774670095e-05,
+      "loss": 1.2123,
+      "step": 4468000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.728164198107151e-05,
+      "loss": 1.2185,
+      "step": 4468500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.7279546215442065e-05,
+      "loss": 1.2111,
+      "step": 4469000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.7277446249881505e-05,
+      "loss": 1.2133,
+      "step": 4469500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.727534628432094e-05,
+      "loss": 1.2369,
+      "step": 4470000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.727324631876037e-05,
+      "loss": 1.2076,
+      "step": 4470500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.727114635319981e-05,
+      "loss": 1.1963,
+      "step": 4471000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.726904638763924e-05,
+      "loss": 1.2016,
+      "step": 4471500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.726694642207867e-05,
+      "loss": 1.2082,
+      "step": 4472000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.726484645651811e-05,
+      "loss": 1.2014,
+      "step": 4472500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.726275069088867e-05,
+      "loss": 1.1942,
+      "step": 4473000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.72606507253281e-05,
+      "loss": 1.1928,
+      "step": 4473500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.725855075976754e-05,
+      "loss": 1.2291,
+      "step": 4474000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.725645079420697e-05,
+      "loss": 1.1859,
+      "step": 4474500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.725435082864641e-05,
+      "loss": 1.2203,
+      "step": 4475000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.725225086308585e-05,
+      "loss": 1.2171,
+      "step": 4475500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.72501550974564e-05,
+      "loss": 1.1794,
+      "step": 4476000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.7248055131895834e-05,
+      "loss": 1.2169,
+      "step": 4476500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.724595516633527e-05,
+      "loss": 1.2109,
+      "step": 4477000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.724385520077471e-05,
+      "loss": 1.197,
+      "step": 4477500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.724175523521414e-05,
+      "loss": 1.2231,
+      "step": 4478000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.7239659469584695e-05,
+      "loss": 1.2241,
+      "step": 4478500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.723755950402413e-05,
+      "loss": 1.2119,
+      "step": 4479000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.723545953846357e-05,
+      "loss": 1.2237,
+      "step": 4479500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.7233359572903e-05,
+      "loss": 1.1747,
+      "step": 4480000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.7231259607342435e-05,
+      "loss": 1.2032,
+      "step": 4480500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.7229159641781875e-05,
+      "loss": 1.1848,
+      "step": 4481000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.722705967622131e-05,
+      "loss": 1.1924,
+      "step": 4481500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.722495971066075e-05,
+      "loss": 1.2005,
+      "step": 4482000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.72228639450313e-05,
+      "loss": 1.2102,
+      "step": 4482500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.7220763979470736e-05,
+      "loss": 1.1917,
+      "step": 4483000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.721866401391017e-05,
+      "loss": 1.1979,
+      "step": 4483500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.721656824828072e-05,
+      "loss": 1.2182,
+      "step": 4484000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.7214472482651284e-05,
+      "loss": 1.2239,
+      "step": 4484500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.7212372517090724e-05,
+      "loss": 1.2424,
+      "step": 4485000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.721027255153015e-05,
+      "loss": 1.1784,
+      "step": 4485500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.7208172585969584e-05,
+      "loss": 1.2004,
+      "step": 4486000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.7206072620409024e-05,
+      "loss": 1.2478,
+      "step": 4486500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.720397265484846e-05,
+      "loss": 1.234,
+      "step": 4487000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.720187268928789e-05,
+      "loss": 1.1881,
+      "step": 4487500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.719977272372733e-05,
+      "loss": 1.1966,
+      "step": 4488000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.7197672758166765e-05,
+      "loss": 1.1886,
+      "step": 4488500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.7195572792606205e-05,
+      "loss": 1.2231,
+      "step": 4489000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.719347282704564e-05,
+      "loss": 1.2004,
+      "step": 4489500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.719137706141619e-05,
+      "loss": 1.2025,
+      "step": 4490000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.7189277095855626e-05,
+      "loss": 1.1911,
+      "step": 4490500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.7187177130295066e-05,
+      "loss": 1.2208,
+      "step": 4491000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.71850771647345e-05,
+      "loss": 1.1989,
+      "step": 4491500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.718297719917393e-05,
+      "loss": 1.2078,
+      "step": 4492000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.7180881433544486e-05,
+      "loss": 1.2043,
+      "step": 4492500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.7178781467983927e-05,
+      "loss": 1.1965,
+      "step": 4493000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.717668150242336e-05,
+      "loss": 1.189,
+      "step": 4493500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.7174581536862793e-05,
+      "loss": 1.2165,
+      "step": 4494000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.7172481571302234e-05,
+      "loss": 1.2124,
+      "step": 4494500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.717038580567279e-05,
+      "loss": 1.2135,
+      "step": 4495000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.716828584011222e-05,
+      "loss": 1.2205,
+      "step": 4495500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.716618587455166e-05,
+      "loss": 1.2408,
+      "step": 4496000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.7164085908991094e-05,
+      "loss": 1.2011,
+      "step": 4496500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.716198594343053e-05,
+      "loss": 1.2201,
+      "step": 4497000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.715988597786997e-05,
+      "loss": 1.1845,
+      "step": 4497500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.715779021224052e-05,
+      "loss": 1.2008,
+      "step": 4498000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.7155690246679955e-05,
+      "loss": 1.204,
+      "step": 4498500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.715359448105051e-05,
+      "loss": 1.2202,
+      "step": 4499000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.715149451548994e-05,
+      "loss": 1.2081,
+      "step": 4499500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.714939454992938e-05,
+      "loss": 1.2204,
+      "step": 4500000
+    },
+    {
+      "epoch": 2.7,
+      "eval_loss": 1.1660878658294678,
+      "eval_runtime": 1104.5668,
+      "eval_samples_per_second": 476.857,
+      "eval_steps_per_second": 79.476,
+      "step": 4500000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.7147294584368816e-05,
+      "loss": 1.2168,
+      "step": 4500500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.714519461880825e-05,
+      "loss": 1.1995,
+      "step": 4501000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.714309465324769e-05,
+      "loss": 1.1965,
+      "step": 4501500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.714099468768712e-05,
+      "loss": 1.198,
+      "step": 4502000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.7138894722126556e-05,
+      "loss": 1.2243,
+      "step": 4502500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.713679895649712e-05,
+      "loss": 1.2141,
+      "step": 4503000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.713470319086767e-05,
+      "loss": 1.2205,
+      "step": 4503500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.713260742523823e-05,
+      "loss": 1.2158,
+      "step": 4504000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.713050745967766e-05,
+      "loss": 1.2192,
+      "step": 4504500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.712840749411709e-05,
+      "loss": 1.2056,
+      "step": 4505000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.712630752855653e-05,
+      "loss": 1.1917,
+      "step": 4505500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.7124207562995965e-05,
+      "loss": 1.2154,
+      "step": 4506000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.71221075974354e-05,
+      "loss": 1.2152,
+      "step": 4506500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.712000763187484e-05,
+      "loss": 1.215,
+      "step": 4507000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.711790766631427e-05,
+      "loss": 1.2267,
+      "step": 4507500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.7115807700753705e-05,
+      "loss": 1.1986,
+      "step": 4508000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.7113707735193145e-05,
+      "loss": 1.1956,
+      "step": 4508500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.711160776963258e-05,
+      "loss": 1.1994,
+      "step": 4509000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.710950780407201e-05,
+      "loss": 1.2137,
+      "step": 4509500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.7107407838511446e-05,
+      "loss": 1.2,
+      "step": 4510000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.7105312072882006e-05,
+      "loss": 1.2109,
+      "step": 4510500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.710321210732144e-05,
+      "loss": 1.2096,
+      "step": 4511000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.710111214176088e-05,
+      "loss": 1.2084,
+      "step": 4511500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.709901217620031e-05,
+      "loss": 1.2131,
+      "step": 4512000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.709691221063974e-05,
+      "loss": 1.2246,
+      "step": 4512500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.709481224507918e-05,
+      "loss": 1.1942,
+      "step": 4513000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.7092712279518614e-05,
+      "loss": 1.2029,
+      "step": 4513500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.709061231395805e-05,
+      "loss": 1.2152,
+      "step": 4514000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.708851654832861e-05,
+      "loss": 1.2066,
+      "step": 4514500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.708641658276804e-05,
+      "loss": 1.1868,
+      "step": 4515000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.7084316617207474e-05,
+      "loss": 1.2129,
+      "step": 4515500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.708221665164691e-05,
+      "loss": 1.1843,
+      "step": 4516000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.708011668608635e-05,
+      "loss": 1.2288,
+      "step": 4516500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.707801672052578e-05,
+      "loss": 1.1749,
+      "step": 4517000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.7075916754965215e-05,
+      "loss": 1.202,
+      "step": 4517500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.7073820989335775e-05,
+      "loss": 1.2309,
+      "step": 4518000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.707172102377521e-05,
+      "loss": 1.1955,
+      "step": 4518500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.706962105821464e-05,
+      "loss": 1.2199,
+      "step": 4519000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.7067525292585196e-05,
+      "loss": 1.218,
+      "step": 4519500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.7065425327024636e-05,
+      "loss": 1.2075,
+      "step": 4520000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.706332536146407e-05,
+      "loss": 1.2114,
+      "step": 4520500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.70612253959035e-05,
+      "loss": 1.1896,
+      "step": 4521000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.705912543034294e-05,
+      "loss": 1.2396,
+      "step": 4521500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.705702546478238e-05,
+      "loss": 1.2188,
+      "step": 4522000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.705492549922181e-05,
+      "loss": 1.1916,
+      "step": 4522500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.705282553366125e-05,
+      "loss": 1.2313,
+      "step": 4523000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.7050725568100684e-05,
+      "loss": 1.2217,
+      "step": 4523500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.704862560254012e-05,
+      "loss": 1.2234,
+      "step": 4524000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.704652563697956e-05,
+      "loss": 1.2273,
+      "step": 4524500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.7044425671418984e-05,
+      "loss": 1.2013,
+      "step": 4525000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.7042329905789545e-05,
+      "loss": 1.228,
+      "step": 4525500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.704022994022898e-05,
+      "loss": 1.2217,
+      "step": 4526000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.703812997466842e-05,
+      "loss": 1.2027,
+      "step": 4526500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.703603000910785e-05,
+      "loss": 1.1982,
+      "step": 4527000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.7033930043547285e-05,
+      "loss": 1.1679,
+      "step": 4527500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.703183847784896e-05,
+      "loss": 1.2017,
+      "step": 4528000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.702974271221952e-05,
+      "loss": 1.2312,
+      "step": 4528500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.702764274665895e-05,
+      "loss": 1.2109,
+      "step": 4529000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.7025542781098386e-05,
+      "loss": 1.2252,
+      "step": 4529500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.702344281553782e-05,
+      "loss": 1.2068,
+      "step": 4530000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.702134284997726e-05,
+      "loss": 1.1873,
+      "step": 4530500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.701924288441669e-05,
+      "loss": 1.2113,
+      "step": 4531000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.701714291885613e-05,
+      "loss": 1.2069,
+      "step": 4531500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.701504295329557e-05,
+      "loss": 1.1964,
+      "step": 4532000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.7012942987735e-05,
+      "loss": 1.2276,
+      "step": 4532500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.7010843022174434e-05,
+      "loss": 1.2111,
+      "step": 4533000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.7008743056613874e-05,
+      "loss": 1.1891,
+      "step": 4533500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.700664309105331e-05,
+      "loss": 1.221,
+      "step": 4534000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.700454732542386e-05,
+      "loss": 1.2013,
+      "step": 4534500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.7002451559794415e-05,
+      "loss": 1.2063,
+      "step": 4535000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.7000351594233855e-05,
+      "loss": 1.214,
+      "step": 4535500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.699825162867329e-05,
+      "loss": 1.191,
+      "step": 4536000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.699615166311272e-05,
+      "loss": 1.2092,
+      "step": 4536500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.699405169755216e-05,
+      "loss": 1.2009,
+      "step": 4537000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6991955931922716e-05,
+      "loss": 1.1854,
+      "step": 4537500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.698985596636215e-05,
+      "loss": 1.2093,
+      "step": 4538000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.698775600080158e-05,
+      "loss": 1.2152,
+      "step": 4538500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.698566023517214e-05,
+      "loss": 1.1986,
+      "step": 4539000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6983560269611576e-05,
+      "loss": 1.1942,
+      "step": 4539500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.698146030405101e-05,
+      "loss": 1.2155,
+      "step": 4540000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.697936033849045e-05,
+      "loss": 1.2098,
+      "step": 4540500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6977260372929884e-05,
+      "loss": 1.1877,
+      "step": 4541000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.697516040736932e-05,
+      "loss": 1.1985,
+      "step": 4541500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.697306044180876e-05,
+      "loss": 1.2029,
+      "step": 4542000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.697096047624819e-05,
+      "loss": 1.2008,
+      "step": 4542500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6968860510687624e-05,
+      "loss": 1.193,
+      "step": 4543000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6966760545127064e-05,
+      "loss": 1.2343,
+      "step": 4543500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.69646605795665e-05,
+      "loss": 1.2487,
+      "step": 4544000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.696256481393705e-05,
+      "loss": 1.193,
+      "step": 4544500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6960464848376485e-05,
+      "loss": 1.2261,
+      "step": 4545000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6958364882815925e-05,
+      "loss": 1.1965,
+      "step": 4545500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.695626491725536e-05,
+      "loss": 1.2251,
+      "step": 4546000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6954164951694785e-05,
+      "loss": 1.2047,
+      "step": 4546500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6952064986134226e-05,
+      "loss": 1.2077,
+      "step": 4547000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.694996502057366e-05,
+      "loss": 1.2017,
+      "step": 4547500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.694786925494422e-05,
+      "loss": 1.1888,
+      "step": 4548000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.694576928938366e-05,
+      "loss": 1.2027,
+      "step": 4548500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6943669323823086e-05,
+      "loss": 1.1987,
+      "step": 4549000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.694156935826252e-05,
+      "loss": 1.2177,
+      "step": 4549500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.693947359263308e-05,
+      "loss": 1.1962,
+      "step": 4550000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.693737362707252e-05,
+      "loss": 1.17,
+      "step": 4550500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6935273661511954e-05,
+      "loss": 1.1927,
+      "step": 4551000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.693317369595138e-05,
+      "loss": 1.2085,
+      "step": 4551500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.693107373039082e-05,
+      "loss": 1.2002,
+      "step": 4552000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.692897796476138e-05,
+      "loss": 1.2013,
+      "step": 4552500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6926877999200815e-05,
+      "loss": 1.203,
+      "step": 4553000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.692477803364025e-05,
+      "loss": 1.223,
+      "step": 4553500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.692267806807968e-05,
+      "loss": 1.1881,
+      "step": 4554000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6920578102519115e-05,
+      "loss": 1.2013,
+      "step": 4554500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6918482336889675e-05,
+      "loss": 1.2175,
+      "step": 4555000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6916382371329115e-05,
+      "loss": 1.1975,
+      "step": 4555500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.691428240576854e-05,
+      "loss": 1.2009,
+      "step": 4556000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6912182440207976e-05,
+      "loss": 1.2237,
+      "step": 4556500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6910082474647416e-05,
+      "loss": 1.2099,
+      "step": 4557000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.690798250908685e-05,
+      "loss": 1.1489,
+      "step": 4557500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.690588254352628e-05,
+      "loss": 1.2198,
+      "step": 4558000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.690378257796572e-05,
+      "loss": 1.1984,
+      "step": 4558500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6901686812336277e-05,
+      "loss": 1.2234,
+      "step": 4559000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.689958684677571e-05,
+      "loss": 1.1995,
+      "step": 4559500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.689749108114627e-05,
+      "loss": 1.2181,
+      "step": 4560000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6895391115585704e-05,
+      "loss": 1.209,
+      "step": 4560500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.689329115002514e-05,
+      "loss": 1.208,
+      "step": 4561000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.689119118446457e-05,
+      "loss": 1.1806,
+      "step": 4561500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.688909541883513e-05,
+      "loss": 1.2011,
+      "step": 4562000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.688699545327457e-05,
+      "loss": 1.2142,
+      "step": 4562500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6884895487714005e-05,
+      "loss": 1.1844,
+      "step": 4563000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.688279552215343e-05,
+      "loss": 1.2255,
+      "step": 4563500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.688069975652399e-05,
+      "loss": 1.2144,
+      "step": 4564000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.687859979096343e-05,
+      "loss": 1.2006,
+      "step": 4564500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6876499825402866e-05,
+      "loss": 1.1816,
+      "step": 4565000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.687439985984229e-05,
+      "loss": 1.2074,
+      "step": 4565500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.687229989428173e-05,
+      "loss": 1.217,
+      "step": 4566000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.687020412865229e-05,
+      "loss": 1.174,
+      "step": 4566500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6868104163091726e-05,
+      "loss": 1.2064,
+      "step": 4567000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.686600419753116e-05,
+      "loss": 1.2314,
+      "step": 4567500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.686390423197059e-05,
+      "loss": 1.1926,
+      "step": 4568000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.686180426641003e-05,
+      "loss": 1.2192,
+      "step": 4568500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.685970850078059e-05,
+      "loss": 1.2161,
+      "step": 4569000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.685760853522003e-05,
+      "loss": 1.1912,
+      "step": 4569500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.685550856965946e-05,
+      "loss": 1.2363,
+      "step": 4570000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.685340860409889e-05,
+      "loss": 1.2171,
+      "step": 4570500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.685131283846945e-05,
+      "loss": 1.1836,
+      "step": 4571000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.684921707284e-05,
+      "loss": 1.2071,
+      "step": 4571500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.684711710727944e-05,
+      "loss": 1.1999,
+      "step": 4572000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6845017141718875e-05,
+      "loss": 1.2004,
+      "step": 4572500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.684291717615831e-05,
+      "loss": 1.2007,
+      "step": 4573000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.684081721059775e-05,
+      "loss": 1.1962,
+      "step": 4573500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.683871724503718e-05,
+      "loss": 1.2038,
+      "step": 4574000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6836617279476616e-05,
+      "loss": 1.2125,
+      "step": 4574500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6834521513847176e-05,
+      "loss": 1.1972,
+      "step": 4575000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.683242154828661e-05,
+      "loss": 1.2274,
+      "step": 4575500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.683032158272604e-05,
+      "loss": 1.2047,
+      "step": 4576000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.682822161716548e-05,
+      "loss": 1.2195,
+      "step": 4576500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6826121651604917e-05,
+      "loss": 1.2151,
+      "step": 4577000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.682402168604434e-05,
+      "loss": 1.206,
+      "step": 4577500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6821925920414904e-05,
+      "loss": 1.2157,
+      "step": 4578000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6819825954854344e-05,
+      "loss": 1.1932,
+      "step": 4578500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.681772598929378e-05,
+      "loss": 1.202,
+      "step": 4579000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.681562602373321e-05,
+      "loss": 1.2344,
+      "step": 4579500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6813526058172644e-05,
+      "loss": 1.2045,
+      "step": 4580000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.681142609261208e-05,
+      "loss": 1.2112,
+      "step": 4580500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.680932612705151e-05,
+      "loss": 1.2218,
+      "step": 4581000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.680722616149095e-05,
+      "loss": 1.2245,
+      "step": 4581500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6805126195930385e-05,
+      "loss": 1.1939,
+      "step": 4582000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.680302623036982e-05,
+      "loss": 1.1896,
+      "step": 4582500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.680092626480926e-05,
+      "loss": 1.1903,
+      "step": 4583000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.679882629924869e-05,
+      "loss": 1.1853,
+      "step": 4583500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6796730533619246e-05,
+      "loss": 1.208,
+      "step": 4584000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6794634767989806e-05,
+      "loss": 1.2015,
+      "step": 4584500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.679253480242924e-05,
+      "loss": 1.215,
+      "step": 4585000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.679043483686867e-05,
+      "loss": 1.2204,
+      "step": 4585500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6788334871308106e-05,
+      "loss": 1.2246,
+      "step": 4586000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6786234905747547e-05,
+      "loss": 1.1694,
+      "step": 4586500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.67841391401181e-05,
+      "loss": 1.1844,
+      "step": 4587000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6782039174557534e-05,
+      "loss": 1.203,
+      "step": 4587500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.677993920899697e-05,
+      "loss": 1.1947,
+      "step": 4588000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.677783924343641e-05,
+      "loss": 1.1909,
+      "step": 4588500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.677573927787584e-05,
+      "loss": 1.1864,
+      "step": 4589000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6773643512246394e-05,
+      "loss": 1.1993,
+      "step": 4589500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6771547746616955e-05,
+      "loss": 1.2095,
+      "step": 4590000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6769447781056395e-05,
+      "loss": 1.194,
+      "step": 4590500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.676734781549583e-05,
+      "loss": 1.1733,
+      "step": 4591000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.676524784993526e-05,
+      "loss": 1.2103,
+      "step": 4591500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6763147884374695e-05,
+      "loss": 1.1967,
+      "step": 4592000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.676104791881413e-05,
+      "loss": 1.1865,
+      "step": 4592500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.675894795325356e-05,
+      "loss": 1.2106,
+      "step": 4593000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6756847987693e-05,
+      "loss": 1.1816,
+      "step": 4593500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6754748022132436e-05,
+      "loss": 1.2173,
+      "step": 4594000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.675265225650299e-05,
+      "loss": 1.2146,
+      "step": 4594500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.675055229094242e-05,
+      "loss": 1.2085,
+      "step": 4595000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.674845232538186e-05,
+      "loss": 1.2103,
+      "step": 4595500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6746352359821297e-05,
+      "loss": 1.2127,
+      "step": 4596000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.674425239426073e-05,
+      "loss": 1.2178,
+      "step": 4596500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.674215242870017e-05,
+      "loss": 1.1744,
+      "step": 4597000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6740056663070724e-05,
+      "loss": 1.2118,
+      "step": 4597500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6737960897441284e-05,
+      "loss": 1.193,
+      "step": 4598000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.673586093188072e-05,
+      "loss": 1.1942,
+      "step": 4598500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.673376096632015e-05,
+      "loss": 1.2093,
+      "step": 4599000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6731661000759585e-05,
+      "loss": 1.1747,
+      "step": 4599500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.672956103519902e-05,
+      "loss": 1.2163,
+      "step": 4600000
+    },
+    {
+      "epoch": 2.76,
+      "eval_loss": 1.163139820098877,
+      "eval_runtime": 1103.2733,
+      "eval_samples_per_second": 477.416,
+      "eval_steps_per_second": 79.57,
+      "step": 4600000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.672746106963846e-05,
+      "loss": 1.2211,
+      "step": 4600500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.672536110407789e-05,
+      "loss": 1.204,
+      "step": 4601000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6723261138517325e-05,
+      "loss": 1.2014,
+      "step": 4601500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6721161172956765e-05,
+      "loss": 1.2161,
+      "step": 4602000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.671906540732732e-05,
+      "loss": 1.2148,
+      "step": 4602500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.671696544176675e-05,
+      "loss": 1.1951,
+      "step": 4603000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6714865476206186e-05,
+      "loss": 1.2089,
+      "step": 4603500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6712765510645626e-05,
+      "loss": 1.2359,
+      "step": 4604000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.671066554508506e-05,
+      "loss": 1.1791,
+      "step": 4604500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.670856557952449e-05,
+      "loss": 1.2236,
+      "step": 4605000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.670646561396393e-05,
+      "loss": 1.1952,
+      "step": 4605500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.670436564840337e-05,
+      "loss": 1.1927,
+      "step": 4606000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.670226988277392e-05,
+      "loss": 1.1981,
+      "step": 4606500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.670016991721336e-05,
+      "loss": 1.1933,
+      "step": 4607000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6698069951652794e-05,
+      "loss": 1.215,
+      "step": 4607500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.669596998609223e-05,
+      "loss": 1.2234,
+      "step": 4608000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.66938784203939e-05,
+      "loss": 1.1973,
+      "step": 4608500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6691778454833335e-05,
+      "loss": 1.2188,
+      "step": 4609000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6689678489272775e-05,
+      "loss": 1.1961,
+      "step": 4609500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.668757852371221e-05,
+      "loss": 1.2124,
+      "step": 4610000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.668547855815164e-05,
+      "loss": 1.1954,
+      "step": 4610500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.66833827925222e-05,
+      "loss": 1.217,
+      "step": 4611000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6681282826961636e-05,
+      "loss": 1.2008,
+      "step": 4611500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.667918286140107e-05,
+      "loss": 1.199,
+      "step": 4612000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.667708289584051e-05,
+      "loss": 1.1947,
+      "step": 4612500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.667498293027994e-05,
+      "loss": 1.1845,
+      "step": 4613000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6672882964719376e-05,
+      "loss": 1.2102,
+      "step": 4613500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.667078719908993e-05,
+      "loss": 1.2148,
+      "step": 4614000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.666868723352937e-05,
+      "loss": 1.1992,
+      "step": 4614500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6666587267968804e-05,
+      "loss": 1.1759,
+      "step": 4615000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.666448730240824e-05,
+      "loss": 1.1837,
+      "step": 4615500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.666238733684768e-05,
+      "loss": 1.202,
+      "step": 4616000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.666028737128711e-05,
+      "loss": 1.1811,
+      "step": 4616500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6658187405726544e-05,
+      "loss": 1.1783,
+      "step": 4617000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6656087440165984e-05,
+      "loss": 1.1965,
+      "step": 4617500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.665399167453654e-05,
+      "loss": 1.2006,
+      "step": 4618000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.665189170897597e-05,
+      "loss": 1.2456,
+      "step": 4618500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.664979174341541e-05,
+      "loss": 1.1703,
+      "step": 4619000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6647691777854845e-05,
+      "loss": 1.1583,
+      "step": 4619500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.66455960122254e-05,
+      "loss": 1.2215,
+      "step": 4620000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.664349604666483e-05,
+      "loss": 1.2196,
+      "step": 4620500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6641400281035386e-05,
+      "loss": 1.1872,
+      "step": 4621000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6639300315474826e-05,
+      "loss": 1.2154,
+      "step": 4621500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.663720034991426e-05,
+      "loss": 1.2243,
+      "step": 4622000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.663510038435369e-05,
+      "loss": 1.2142,
+      "step": 4622500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6633004618724246e-05,
+      "loss": 1.193,
+      "step": 4623000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.663090465316369e-05,
+      "loss": 1.2032,
+      "step": 4623500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.662880468760312e-05,
+      "loss": 1.2135,
+      "step": 4624000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6626704722042554e-05,
+      "loss": 1.2345,
+      "step": 4624500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6624604756481994e-05,
+      "loss": 1.2111,
+      "step": 4625000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.662250899085255e-05,
+      "loss": 1.2073,
+      "step": 4625500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.662040902529198e-05,
+      "loss": 1.2208,
+      "step": 4626000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.661830905973142e-05,
+      "loss": 1.1991,
+      "step": 4626500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6616209094170855e-05,
+      "loss": 1.1907,
+      "step": 4627000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.661410912861029e-05,
+      "loss": 1.209,
+      "step": 4627500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.661200916304973e-05,
+      "loss": 1.1826,
+      "step": 4628000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.660990919748916e-05,
+      "loss": 1.2014,
+      "step": 4628500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6607809231928595e-05,
+      "loss": 1.2212,
+      "step": 4629000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6605717666230276e-05,
+      "loss": 1.1959,
+      "step": 4629500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.66036177006697e-05,
+      "loss": 1.2147,
+      "step": 4630000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.660151773510914e-05,
+      "loss": 1.2333,
+      "step": 4630500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6599417769548576e-05,
+      "loss": 1.2151,
+      "step": 4631000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6597322003919136e-05,
+      "loss": 1.2335,
+      "step": 4631500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6595222038358577e-05,
+      "loss": 1.2113,
+      "step": 4632000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6593122072798e-05,
+      "loss": 1.2189,
+      "step": 4632500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.659102210723744e-05,
+      "loss": 1.2152,
+      "step": 4633000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.658892214167688e-05,
+      "loss": 1.1874,
+      "step": 4633500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.658682217611631e-05,
+      "loss": 1.1624,
+      "step": 4634000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6584722210555744e-05,
+      "loss": 1.1837,
+      "step": 4634500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6582622244995184e-05,
+      "loss": 1.2007,
+      "step": 4635000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.658052227943462e-05,
+      "loss": 1.1904,
+      "step": 4635500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.657842231387405e-05,
+      "loss": 1.1769,
+      "step": 4636000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.657633074817573e-05,
+      "loss": 1.2046,
+      "step": 4636500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.657423078261516e-05,
+      "loss": 1.1649,
+      "step": 4637000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.65721308170546e-05,
+      "loss": 1.1793,
+      "step": 4637500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.657003085149403e-05,
+      "loss": 1.2177,
+      "step": 4638000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.656793088593347e-05,
+      "loss": 1.161,
+      "step": 4638500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6565830920372906e-05,
+      "loss": 1.192,
+      "step": 4639000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.656373095481234e-05,
+      "loss": 1.205,
+      "step": 4639500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.656163098925178e-05,
+      "loss": 1.181,
+      "step": 4640000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.655953102369121e-05,
+      "loss": 1.1877,
+      "step": 4640500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6557435258061766e-05,
+      "loss": 1.2009,
+      "step": 4641000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.65553352925012e-05,
+      "loss": 1.2092,
+      "step": 4641500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.655323532694064e-05,
+      "loss": 1.1991,
+      "step": 4642000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6551135361380073e-05,
+      "loss": 1.1978,
+      "step": 4642500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.654903539581951e-05,
+      "loss": 1.1945,
+      "step": 4643000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.654693543025895e-05,
+      "loss": 1.186,
+      "step": 4643500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.654483546469838e-05,
+      "loss": 1.1879,
+      "step": 4644000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6542735499137814e-05,
+      "loss": 1.1983,
+      "step": 4644500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.654063973350837e-05,
+      "loss": 1.2077,
+      "step": 4645000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.653854396787893e-05,
+      "loss": 1.2137,
+      "step": 4645500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.653644400231836e-05,
+      "loss": 1.2147,
+      "step": 4646000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6534344036757795e-05,
+      "loss": 1.2019,
+      "step": 4646500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6532244071197235e-05,
+      "loss": 1.2145,
+      "step": 4647000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.653014410563667e-05,
+      "loss": 1.2013,
+      "step": 4647500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.65280441400761e-05,
+      "loss": 1.1908,
+      "step": 4648000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6525948374446656e-05,
+      "loss": 1.2172,
+      "step": 4648500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6523848408886096e-05,
+      "loss": 1.207,
+      "step": 4649000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.652174844332553e-05,
+      "loss": 1.2157,
+      "step": 4649500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.651965267769608e-05,
+      "loss": 1.2491,
+      "step": 4650000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6517552712135516e-05,
+      "loss": 1.2084,
+      "step": 4650500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6515452746574957e-05,
+      "loss": 1.1814,
+      "step": 4651000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.651335278101439e-05,
+      "loss": 1.1819,
+      "step": 4651500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6511252815453824e-05,
+      "loss": 1.2158,
+      "step": 4652000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6509152849893264e-05,
+      "loss": 1.1715,
+      "step": 4652500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.65070528843327e-05,
+      "loss": 1.1916,
+      "step": 4653000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.650495291877213e-05,
+      "loss": 1.251,
+      "step": 4653500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.650285295321157e-05,
+      "loss": 1.1896,
+      "step": 4654000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6500761387513245e-05,
+      "loss": 1.1931,
+      "step": 4654500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.649866142195268e-05,
+      "loss": 1.199,
+      "step": 4655000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.649656145639211e-05,
+      "loss": 1.2088,
+      "step": 4655500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.649446149083155e-05,
+      "loss": 1.1944,
+      "step": 4656000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6492361525270985e-05,
+      "loss": 1.2331,
+      "step": 4656500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.649026155971042e-05,
+      "loss": 1.1861,
+      "step": 4657000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.648816159414986e-05,
+      "loss": 1.1927,
+      "step": 4657500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.648606582852041e-05,
+      "loss": 1.248,
+      "step": 4658000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6483965862959846e-05,
+      "loss": 1.2187,
+      "step": 4658500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.648186589739928e-05,
+      "loss": 1.1919,
+      "step": 4659000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.647976593183872e-05,
+      "loss": 1.1589,
+      "step": 4659500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.647766596627815e-05,
+      "loss": 1.2003,
+      "step": 4660000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.647557020064871e-05,
+      "loss": 1.1919,
+      "step": 4660500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.647347023508815e-05,
+      "loss": 1.2178,
+      "step": 4661000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.647137026952758e-05,
+      "loss": 1.2156,
+      "step": 4661500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6469270303967014e-05,
+      "loss": 1.2119,
+      "step": 4662000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6467170338406454e-05,
+      "loss": 1.1729,
+      "step": 4662500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.646507037284589e-05,
+      "loss": 1.1975,
+      "step": 4663000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.646297040728532e-05,
+      "loss": 1.2193,
+      "step": 4663500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6460870441724754e-05,
+      "loss": 1.1868,
+      "step": 4664000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.645877047616419e-05,
+      "loss": 1.2178,
+      "step": 4664500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.645667051060362e-05,
+      "loss": 1.1859,
+      "step": 4665000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.645457054504306e-05,
+      "loss": 1.2217,
+      "step": 4665500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.645247477941362e-05,
+      "loss": 1.1866,
+      "step": 4666000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.645037481385305e-05,
+      "loss": 1.2174,
+      "step": 4666500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.644827484829248e-05,
+      "loss": 1.1947,
+      "step": 4667000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.644617488273192e-05,
+      "loss": 1.1833,
+      "step": 4667500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6444074917171356e-05,
+      "loss": 1.182,
+      "step": 4668000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6441979151541916e-05,
+      "loss": 1.2013,
+      "step": 4668500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.643987918598135e-05,
+      "loss": 1.1957,
+      "step": 4669000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.643778342035191e-05,
+      "loss": 1.2004,
+      "step": 4669500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.643568345479134e-05,
+      "loss": 1.2074,
+      "step": 4670000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.643358348923078e-05,
+      "loss": 1.1961,
+      "step": 4670500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.643148352367022e-05,
+      "loss": 1.1886,
+      "step": 4671000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6429383558109644e-05,
+      "loss": 1.1983,
+      "step": 4671500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6427287792480204e-05,
+      "loss": 1.2227,
+      "step": 4672000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.642518782691964e-05,
+      "loss": 1.2107,
+      "step": 4672500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.642308786135908e-05,
+      "loss": 1.2124,
+      "step": 4673000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6420987895798504e-05,
+      "loss": 1.2263,
+      "step": 4673500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6418892130169065e-05,
+      "loss": 1.2343,
+      "step": 4674000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.64167921646085e-05,
+      "loss": 1.189,
+      "step": 4674500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.641469219904794e-05,
+      "loss": 1.22,
+      "step": 4675000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.641259223348737e-05,
+      "loss": 1.2115,
+      "step": 4675500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6410492267926805e-05,
+      "loss": 1.2095,
+      "step": 4676000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.640839230236624e-05,
+      "loss": 1.2128,
+      "step": 4676500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.640629233680567e-05,
+      "loss": 1.2167,
+      "step": 4677000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.640419237124511e-05,
+      "loss": 1.2097,
+      "step": 4677500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6402092405684546e-05,
+      "loss": 1.1971,
+      "step": 4678000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.639999244012398e-05,
+      "loss": 1.1878,
+      "step": 4678500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.639789247456342e-05,
+      "loss": 1.2122,
+      "step": 4679000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.639579250900285e-05,
+      "loss": 1.1921,
+      "step": 4679500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.639369254344229e-05,
+      "loss": 1.2129,
+      "step": 4680000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.639159257788173e-05,
+      "loss": 1.1916,
+      "step": 4680500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.638949261232116e-05,
+      "loss": 1.1941,
+      "step": 4681000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.638739264676059e-05,
+      "loss": 1.202,
+      "step": 4681500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.638529688113115e-05,
+      "loss": 1.2154,
+      "step": 4682000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.638319691557059e-05,
+      "loss": 1.2032,
+      "step": 4682500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.638109695001002e-05,
+      "loss": 1.1855,
+      "step": 4683000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.6378996984449455e-05,
+      "loss": 1.1944,
+      "step": 4683500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.637690121882001e-05,
+      "loss": 1.2142,
+      "step": 4684000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.637480125325945e-05,
+      "loss": 1.204,
+      "step": 4684500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.637270128769888e-05,
+      "loss": 1.1987,
+      "step": 4685000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.637060132213832e-05,
+      "loss": 1.2046,
+      "step": 4685500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.6368501356577756e-05,
+      "loss": 1.202,
+      "step": 4686000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.636640559094831e-05,
+      "loss": 1.2092,
+      "step": 4686500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.636430562538774e-05,
+      "loss": 1.1874,
+      "step": 4687000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.636220565982718e-05,
+      "loss": 1.2251,
+      "step": 4687500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.6360105694266616e-05,
+      "loss": 1.1958,
+      "step": 4688000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.635800992863717e-05,
+      "loss": 1.2091,
+      "step": 4688500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.63559099630766e-05,
+      "loss": 1.1761,
+      "step": 4689000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.6353809997516044e-05,
+      "loss": 1.1953,
+      "step": 4689500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.635171003195548e-05,
+      "loss": 1.2074,
+      "step": 4690000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.634961006639491e-05,
+      "loss": 1.2042,
+      "step": 4690500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.6347510100834344e-05,
+      "loss": 1.2107,
+      "step": 4691000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.634541013527378e-05,
+      "loss": 1.196,
+      "step": 4691500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.634331016971322e-05,
+      "loss": 1.1701,
+      "step": 4692000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.634121020415265e-05,
+      "loss": 1.2213,
+      "step": 4692500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.6339110238592084e-05,
+      "loss": 1.2067,
+      "step": 4693000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.6337010273031525e-05,
+      "loss": 1.2223,
+      "step": 4693500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.633491450740208e-05,
+      "loss": 1.2028,
+      "step": 4694000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.633281454184151e-05,
+      "loss": 1.2097,
+      "step": 4694500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.6330714576280945e-05,
+      "loss": 1.1858,
+      "step": 4695000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.6328614610720385e-05,
+      "loss": 1.2022,
+      "step": 4695500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.632651464515982e-05,
+      "loss": 1.218,
+      "step": 4696000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.632441467959925e-05,
+      "loss": 1.1727,
+      "step": 4696500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.632231471403869e-05,
+      "loss": 1.1869,
+      "step": 4697000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.6320214748478126e-05,
+      "loss": 1.191,
+      "step": 4697500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.631811898284868e-05,
+      "loss": 1.2257,
+      "step": 4698000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.631601901728811e-05,
+      "loss": 1.2039,
+      "step": 4698500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.631391905172755e-05,
+      "loss": 1.1765,
+      "step": 4699000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.631181908616699e-05,
+      "loss": 1.2188,
+      "step": 4699500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.630971912060642e-05,
+      "loss": 1.1825,
+      "step": 4700000
+    },
+    {
+      "epoch": 2.82,
+      "eval_loss": 1.1633455753326416,
+      "eval_runtime": 1101.2926,
+      "eval_samples_per_second": 478.274,
+      "eval_steps_per_second": 79.713,
+      "step": 4700000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.630761915504586e-05,
+      "loss": 1.2054,
+      "step": 4700500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.6305519189485294e-05,
+      "loss": 1.188,
+      "step": 4701000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.630341922392473e-05,
+      "loss": 1.1919,
+      "step": 4701500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.630132345829529e-05,
+      "loss": 1.2154,
+      "step": 4702000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.629922349273472e-05,
+      "loss": 1.2176,
+      "step": 4702500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.6297123527174155e-05,
+      "loss": 1.1711,
+      "step": 4703000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.6295023561613595e-05,
+      "loss": 1.1888,
+      "step": 4703500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.629292779598415e-05,
+      "loss": 1.2028,
+      "step": 4704000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.62908320303547e-05,
+      "loss": 1.1784,
+      "step": 4704500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.6288732064794136e-05,
+      "loss": 1.2093,
+      "step": 4705000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.628663209923357e-05,
+      "loss": 1.2052,
+      "step": 4705500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.628453213367301e-05,
+      "loss": 1.2177,
+      "step": 4706000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.628243216811244e-05,
+      "loss": 1.1742,
+      "step": 4706500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.6280332202551876e-05,
+      "loss": 1.2079,
+      "step": 4707000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.6278232236991316e-05,
+      "loss": 1.2041,
+      "step": 4707500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.627613227143075e-05,
+      "loss": 1.2185,
+      "step": 4708000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.6274036505801303e-05,
+      "loss": 1.2061,
+      "step": 4708500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.6271936540240744e-05,
+      "loss": 1.1878,
+      "step": 4709000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.626983657468018e-05,
+      "loss": 1.2126,
+      "step": 4709500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.626773660911961e-05,
+      "loss": 1.196,
+      "step": 4710000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.6265640843490164e-05,
+      "loss": 1.1912,
+      "step": 4710500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.6263540877929604e-05,
+      "loss": 1.2203,
+      "step": 4711000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.626144091236904e-05,
+      "loss": 1.1876,
+      "step": 4711500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.625934094680847e-05,
+      "loss": 1.2069,
+      "step": 4712000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.625724098124791e-05,
+      "loss": 1.2107,
+      "step": 4712500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.6255141015687345e-05,
+      "loss": 1.1891,
+      "step": 4713000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.625304105012677e-05,
+      "loss": 1.1799,
+      "step": 4713500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.625094108456621e-05,
+      "loss": 1.2006,
+      "step": 4714000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.624884531893677e-05,
+      "loss": 1.1825,
+      "step": 4714500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.6246745353376206e-05,
+      "loss": 1.2411,
+      "step": 4715000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.624464538781564e-05,
+      "loss": 1.1987,
+      "step": 4715500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.624254542225507e-05,
+      "loss": 1.2261,
+      "step": 4716000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.6240445456694506e-05,
+      "loss": 1.1952,
+      "step": 4716500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.6238345491133946e-05,
+      "loss": 1.195,
+      "step": 4717000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.623624552557338e-05,
+      "loss": 1.2201,
+      "step": 4717500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.623414556001281e-05,
+      "loss": 1.2103,
+      "step": 4718000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.6232053994314494e-05,
+      "loss": 1.2044,
+      "step": 4718500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.622995402875393e-05,
+      "loss": 1.2071,
+      "step": 4719000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.622785406319337e-05,
+      "loss": 1.1848,
+      "step": 4719500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.62257540976328e-05,
+      "loss": 1.2019,
+      "step": 4720000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.622365413207223e-05,
+      "loss": 1.1901,
+      "step": 4720500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.622155416651167e-05,
+      "loss": 1.1973,
+      "step": 4721000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.62194542009511e-05,
+      "loss": 1.1691,
+      "step": 4721500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.6217354235390535e-05,
+      "loss": 1.2154,
+      "step": 4722000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.6215258469761095e-05,
+      "loss": 1.2039,
+      "step": 4722500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.6213162704131655e-05,
+      "loss": 1.2102,
+      "step": 4723000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.621106273857109e-05,
+      "loss": 1.1888,
+      "step": 4723500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.620896277301052e-05,
+      "loss": 1.1894,
+      "step": 4724000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.620686280744996e-05,
+      "loss": 1.1901,
+      "step": 4724500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.6204762841889396e-05,
+      "loss": 1.2517,
+      "step": 4725000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.620266707625995e-05,
+      "loss": 1.1845,
+      "step": 4725500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.62005713106305e-05,
+      "loss": 1.1993,
+      "step": 4726000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.619847134506994e-05,
+      "loss": 1.1984,
+      "step": 4726500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.619637137950938e-05,
+      "loss": 1.194,
+      "step": 4727000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.619427141394881e-05,
+      "loss": 1.2029,
+      "step": 4727500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.6192171448388244e-05,
+      "loss": 1.2284,
+      "step": 4728000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.6190071482827684e-05,
+      "loss": 1.2056,
+      "step": 4728500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.618797151726712e-05,
+      "loss": 1.1984,
+      "step": 4729000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.618587155170655e-05,
+      "loss": 1.1771,
+      "step": 4729500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.6183771586145984e-05,
+      "loss": 1.1913,
+      "step": 4730000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.6181675820516545e-05,
+      "loss": 1.2038,
+      "step": 4730500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.617957585495598e-05,
+      "loss": 1.2057,
+      "step": 4731000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.617747588939542e-05,
+      "loss": 1.2085,
+      "step": 4731500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.617537592383485e-05,
+      "loss": 1.2051,
+      "step": 4732000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.6173280158205405e-05,
+      "loss": 1.2227,
+      "step": 4732500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.617118439257596e-05,
+      "loss": 1.203,
+      "step": 4733000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.616908442701539e-05,
+      "loss": 1.2091,
+      "step": 4733500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.616698446145483e-05,
+      "loss": 1.2071,
+      "step": 4734000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.6164884495894266e-05,
+      "loss": 1.2114,
+      "step": 4734500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.61627845303337e-05,
+      "loss": 1.1817,
+      "step": 4735000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.616068456477314e-05,
+      "loss": 1.2095,
+      "step": 4735500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.615858459921257e-05,
+      "loss": 1.1924,
+      "step": 4736000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.615648463365201e-05,
+      "loss": 1.199,
+      "step": 4736500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.615438886802257e-05,
+      "loss": 1.1851,
+      "step": 4737000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.6152288902462e-05,
+      "loss": 1.2288,
+      "step": 4737500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.6150188936901434e-05,
+      "loss": 1.1873,
+      "step": 4738000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.6148088971340874e-05,
+      "loss": 1.2014,
+      "step": 4738500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.614598900578031e-05,
+      "loss": 1.2026,
+      "step": 4739000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.614389324015086e-05,
+      "loss": 1.1921,
+      "step": 4739500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.6141793274590295e-05,
+      "loss": 1.1905,
+      "step": 4740000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.6139693309029735e-05,
+      "loss": 1.2072,
+      "step": 4740500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.613759334346917e-05,
+      "loss": 1.2167,
+      "step": 4741000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.613549757783972e-05,
+      "loss": 1.207,
+      "step": 4741500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.6133397612279156e-05,
+      "loss": 1.2005,
+      "step": 4742000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.6131297646718596e-05,
+      "loss": 1.2083,
+      "step": 4742500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.612919768115803e-05,
+      "loss": 1.2005,
+      "step": 4743000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.612709771559747e-05,
+      "loss": 1.2197,
+      "step": 4743500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.612500194996802e-05,
+      "loss": 1.2311,
+      "step": 4744000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.6122901984407456e-05,
+      "loss": 1.1843,
+      "step": 4744500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.612080201884689e-05,
+      "loss": 1.1958,
+      "step": 4745000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.611870205328633e-05,
+      "loss": 1.219,
+      "step": 4745500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.6116602087725764e-05,
+      "loss": 1.2087,
+      "step": 4746000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.611450632209632e-05,
+      "loss": 1.1806,
+      "step": 4746500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.611240635653575e-05,
+      "loss": 1.2002,
+      "step": 4747000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.611030639097519e-05,
+      "loss": 1.2056,
+      "step": 4747500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.6108206425414624e-05,
+      "loss": 1.199,
+      "step": 4748000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.610610645985406e-05,
+      "loss": 1.1788,
+      "step": 4748500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.610400649429349e-05,
+      "loss": 1.2023,
+      "step": 4749000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.6101906528732925e-05,
+      "loss": 1.2092,
+      "step": 4749500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.609980656317236e-05,
+      "loss": 1.2067,
+      "step": 4750000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.60977065976118e-05,
+      "loss": 1.1919,
+      "step": 4750500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.609561083198236e-05,
+      "loss": 1.218,
+      "step": 4751000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.6093510866421785e-05,
+      "loss": 1.197,
+      "step": 4751500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.6091410900861226e-05,
+      "loss": 1.2201,
+      "step": 4752000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.608931093530066e-05,
+      "loss": 1.1732,
+      "step": 4752500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.608721096974009e-05,
+      "loss": 1.1765,
+      "step": 4753000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.608511100417953e-05,
+      "loss": 1.1958,
+      "step": 4753500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.6083015238550086e-05,
+      "loss": 1.2067,
+      "step": 4754000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.608091527298952e-05,
+      "loss": 1.1921,
+      "step": 4754500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.607881950736008e-05,
+      "loss": 1.1672,
+      "step": 4755000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.6076719541799514e-05,
+      "loss": 1.1818,
+      "step": 4755500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.6074619576238954e-05,
+      "loss": 1.2246,
+      "step": 4756000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.607251961067838e-05,
+      "loss": 1.1816,
+      "step": 4756500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.607041964511782e-05,
+      "loss": 1.2222,
+      "step": 4757000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.606832387948838e-05,
+      "loss": 1.2449,
+      "step": 4757500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.6066223913927815e-05,
+      "loss": 1.1944,
+      "step": 4758000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.606412394836724e-05,
+      "loss": 1.2011,
+      "step": 4758500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.606202398280668e-05,
+      "loss": 1.1835,
+      "step": 4759000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.6059924017246115e-05,
+      "loss": 1.1727,
+      "step": 4759500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.605782405168555e-05,
+      "loss": 1.2028,
+      "step": 4760000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.605572828605611e-05,
+      "loss": 1.2084,
+      "step": 4760500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.605362832049554e-05,
+      "loss": 1.2158,
+      "step": 4761000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.6051528354934976e-05,
+      "loss": 1.1973,
+      "step": 4761500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.604942838937441e-05,
+      "loss": 1.1967,
+      "step": 4762000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.604732842381385e-05,
+      "loss": 1.1897,
+      "step": 4762500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.604522845825328e-05,
+      "loss": 1.2448,
+      "step": 4763000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.6043132692623837e-05,
+      "loss": 1.2157,
+      "step": 4763500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.604103272706328e-05,
+      "loss": 1.1874,
+      "step": 4764000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.603893276150271e-05,
+      "loss": 1.2273,
+      "step": 4764500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.6036832795942144e-05,
+      "loss": 1.228,
+      "step": 4765000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.6034732830381584e-05,
+      "loss": 1.1743,
+      "step": 4765500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.603263706475214e-05,
+      "loss": 1.175,
+      "step": 4766000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.603053709919157e-05,
+      "loss": 1.1941,
+      "step": 4766500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.6028437133631004e-05,
+      "loss": 1.1667,
+      "step": 4767000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.6026337168070445e-05,
+      "loss": 1.2261,
+      "step": 4767500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.602423720250988e-05,
+      "loss": 1.2095,
+      "step": 4768000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.602213723694931e-05,
+      "loss": 1.2033,
+      "step": 4768500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.6020041471319865e-05,
+      "loss": 1.2201,
+      "step": 4769000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.6017941505759305e-05,
+      "loss": 1.1929,
+      "step": 4769500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.601584154019874e-05,
+      "loss": 1.189,
+      "step": 4770000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.601374157463817e-05,
+      "loss": 1.204,
+      "step": 4770500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.601164160907761e-05,
+      "loss": 1.2056,
+      "step": 4771000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.6009541643517046e-05,
+      "loss": 1.2071,
+      "step": 4771500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.600744167795648e-05,
+      "loss": 1.2204,
+      "step": 4772000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.600534171239592e-05,
+      "loss": 1.2162,
+      "step": 4772500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.600324174683535e-05,
+      "loss": 1.1889,
+      "step": 4773000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.6001141781274787e-05,
+      "loss": 1.1886,
+      "step": 4773500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.599904181571422e-05,
+      "loss": 1.2125,
+      "step": 4774000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5996941850153653e-05,
+      "loss": 1.1845,
+      "step": 4774500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5994846084524214e-05,
+      "loss": 1.1981,
+      "step": 4775000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5992746118963654e-05,
+      "loss": 1.2056,
+      "step": 4775500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.599065035333421e-05,
+      "loss": 1.1964,
+      "step": 4776000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.598855038777364e-05,
+      "loss": 1.1839,
+      "step": 4776500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5986450422213075e-05,
+      "loss": 1.2048,
+      "step": 4777000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5984350456652515e-05,
+      "loss": 1.202,
+      "step": 4777500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.598225049109195e-05,
+      "loss": 1.2022,
+      "step": 4778000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.59801547254625e-05,
+      "loss": 1.1963,
+      "step": 4778500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5978054759901935e-05,
+      "loss": 1.2098,
+      "step": 4779000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5975954794341376e-05,
+      "loss": 1.1974,
+      "step": 4779500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.597385482878081e-05,
+      "loss": 1.2121,
+      "step": 4780000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.597175486322024e-05,
+      "loss": 1.2044,
+      "step": 4780500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5969654897659676e-05,
+      "loss": 1.2116,
+      "step": 4781000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5967559132030236e-05,
+      "loss": 1.2026,
+      "step": 4781500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.596545916646967e-05,
+      "loss": 1.2008,
+      "step": 4782000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.596335920090911e-05,
+      "loss": 1.2317,
+      "step": 4782500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.596125923534854e-05,
+      "loss": 1.1985,
+      "step": 4783000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.59591634697191e-05,
+      "loss": 1.1721,
+      "step": 4783500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.595706350415853e-05,
+      "loss": 1.1958,
+      "step": 4784000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.595496353859797e-05,
+      "loss": 1.2187,
+      "step": 4784500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5952863573037404e-05,
+      "loss": 1.1704,
+      "step": 4785000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.595076780740796e-05,
+      "loss": 1.2295,
+      "step": 4785500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.594866784184739e-05,
+      "loss": 1.1823,
+      "step": 4786000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.594656787628683e-05,
+      "loss": 1.2088,
+      "step": 4786500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5944467910726265e-05,
+      "loss": 1.1953,
+      "step": 4787000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.594237214509682e-05,
+      "loss": 1.2237,
+      "step": 4787500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.594027217953626e-05,
+      "loss": 1.2132,
+      "step": 4788000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.593817221397569e-05,
+      "loss": 1.1921,
+      "step": 4788500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5936072248415126e-05,
+      "loss": 1.236,
+      "step": 4789000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5933972282854566e-05,
+      "loss": 1.2194,
+      "step": 4789500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5931872317294e-05,
+      "loss": 1.2034,
+      "step": 4790000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5929772351733426e-05,
+      "loss": 1.2056,
+      "step": 4790500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5927672386172866e-05,
+      "loss": 1.1931,
+      "step": 4791000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.59255724206123e-05,
+      "loss": 1.1875,
+      "step": 4791500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.592347245505173e-05,
+      "loss": 1.2071,
+      "step": 4792000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.592137248949117e-05,
+      "loss": 1.2045,
+      "step": 4792500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.591927252393061e-05,
+      "loss": 1.1994,
+      "step": 4793000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.591717255837004e-05,
+      "loss": 1.1904,
+      "step": 4793500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.591507259280948e-05,
+      "loss": 1.1726,
+      "step": 4794000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5912976827180034e-05,
+      "loss": 1.2033,
+      "step": 4794500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.591087686161947e-05,
+      "loss": 1.212,
+      "step": 4795000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.59087768960589e-05,
+      "loss": 1.2119,
+      "step": 4795500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.590667693049834e-05,
+      "loss": 1.216,
+      "step": 4796000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5904576964937775e-05,
+      "loss": 1.1873,
+      "step": 4796500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5902476999377215e-05,
+      "loss": 1.1738,
+      "step": 4797000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.590037703381665e-05,
+      "loss": 1.1922,
+      "step": 4797500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.58982812681872e-05,
+      "loss": 1.216,
+      "step": 4798000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5896181302626635e-05,
+      "loss": 1.2084,
+      "step": 4798500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5894081337066076e-05,
+      "loss": 1.1966,
+      "step": 4799000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.589198137150551e-05,
+      "loss": 1.2035,
+      "step": 4799500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.588988560587606e-05,
+      "loss": 1.1797,
+      "step": 4800000
+    },
+    {
+      "epoch": 2.88,
+      "eval_loss": 1.1569877862930298,
+      "eval_runtime": 1104.5998,
+      "eval_samples_per_second": 476.842,
+      "eval_steps_per_second": 79.474,
+      "step": 4800000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5887785640315496e-05,
+      "loss": 1.2247,
+      "step": 4800500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5885685674754936e-05,
+      "loss": 1.205,
+      "step": 4801000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.588358570919437e-05,
+      "loss": 1.1895,
+      "step": 4801500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.58814857436338e-05,
+      "loss": 1.1901,
+      "step": 4802000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.587938997800436e-05,
+      "loss": 1.2086,
+      "step": 4802500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.58772900124438e-05,
+      "loss": 1.1996,
+      "step": 4803000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.587519424681435e-05,
+      "loss": 1.1872,
+      "step": 4803500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5873094281253784e-05,
+      "loss": 1.2036,
+      "step": 4804000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5870994315693224e-05,
+      "loss": 1.203,
+      "step": 4804500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.586889855006378e-05,
+      "loss": 1.1776,
+      "step": 4805000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.586679858450321e-05,
+      "loss": 1.218,
+      "step": 4805500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5864698618942645e-05,
+      "loss": 1.1744,
+      "step": 4806000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5862598653382085e-05,
+      "loss": 1.2032,
+      "step": 4806500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.586049868782152e-05,
+      "loss": 1.2093,
+      "step": 4807000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.585839872226095e-05,
+      "loss": 1.1724,
+      "step": 4807500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.585629875670039e-05,
+      "loss": 1.2006,
+      "step": 4808000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5854198791139826e-05,
+      "loss": 1.1943,
+      "step": 4808500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.585209882557926e-05,
+      "loss": 1.2294,
+      "step": 4809000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.58499988600187e-05,
+      "loss": 1.1754,
+      "step": 4809500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5847898894458126e-05,
+      "loss": 1.2126,
+      "step": 4810000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5845798928897566e-05,
+      "loss": 1.2136,
+      "step": 4810500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5843698963337e-05,
+      "loss": 1.1973,
+      "step": 4811000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.584159899777643e-05,
+      "loss": 1.2096,
+      "step": 4811500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5839499032215874e-05,
+      "loss": 1.1865,
+      "step": 4812000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.583739906665531e-05,
+      "loss": 1.1923,
+      "step": 4812500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.583530330102586e-05,
+      "loss": 1.1776,
+      "step": 4813000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5833203335465294e-05,
+      "loss": 1.2029,
+      "step": 4813500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5831103369904734e-05,
+      "loss": 1.1947,
+      "step": 4814000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.582900340434417e-05,
+      "loss": 1.189,
+      "step": 4814500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.582690763871472e-05,
+      "loss": 1.2078,
+      "step": 4815000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5824807673154155e-05,
+      "loss": 1.1927,
+      "step": 4815500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5822707707593595e-05,
+      "loss": 1.2251,
+      "step": 4816000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.582060774203303e-05,
+      "loss": 1.2161,
+      "step": 4816500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.581850777647246e-05,
+      "loss": 1.19,
+      "step": 4817000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.581641201084302e-05,
+      "loss": 1.2019,
+      "step": 4817500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5814312045282456e-05,
+      "loss": 1.1848,
+      "step": 4818000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.581221207972189e-05,
+      "loss": 1.1792,
+      "step": 4818500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.581011631409245e-05,
+      "loss": 1.1943,
+      "step": 4819000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.580801634853188e-05,
+      "loss": 1.2053,
+      "step": 4819500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5805916382971316e-05,
+      "loss": 1.1727,
+      "step": 4820000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.580381641741075e-05,
+      "loss": 1.1908,
+      "step": 4820500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.580171645185019e-05,
+      "loss": 1.1763,
+      "step": 4821000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5799616486289624e-05,
+      "loss": 1.1968,
+      "step": 4821500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.579751652072906e-05,
+      "loss": 1.2089,
+      "step": 4822000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.57954165551685e-05,
+      "loss": 1.1851,
+      "step": 4822500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.579331658960793e-05,
+      "loss": 1.2242,
+      "step": 4823000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5791216624047364e-05,
+      "loss": 1.1793,
+      "step": 4823500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5789116658486804e-05,
+      "loss": 1.1819,
+      "step": 4824000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.578701669292624e-05,
+      "loss": 1.174,
+      "step": 4824500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.578492092729679e-05,
+      "loss": 1.1942,
+      "step": 4825000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5782820961736225e-05,
+      "loss": 1.2153,
+      "step": 4825500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5780720996175665e-05,
+      "loss": 1.195,
+      "step": 4826000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.57786210306151e-05,
+      "loss": 1.2042,
+      "step": 4826500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.577652526498565e-05,
+      "loss": 1.2042,
+      "step": 4827000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.577442529942509e-05,
+      "loss": 1.1934,
+      "step": 4827500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5772325333864526e-05,
+      "loss": 1.1855,
+      "step": 4828000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.577022536830396e-05,
+      "loss": 1.2214,
+      "step": 4828500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.576812960267451e-05,
+      "loss": 1.175,
+      "step": 4829000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5766033837045066e-05,
+      "loss": 1.1773,
+      "step": 4829500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.576393387148451e-05,
+      "loss": 1.2022,
+      "step": 4830000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.576183390592394e-05,
+      "loss": 1.1988,
+      "step": 4830500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5759733940363374e-05,
+      "loss": 1.2069,
+      "step": 4831000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5757633974802814e-05,
+      "loss": 1.1969,
+      "step": 4831500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.575553820917337e-05,
+      "loss": 1.2188,
+      "step": 4832000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.57534382436128e-05,
+      "loss": 1.2256,
+      "step": 4832500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.575133827805224e-05,
+      "loss": 1.2124,
+      "step": 4833000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5749238312491675e-05,
+      "loss": 1.1965,
+      "step": 4833500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.574713834693111e-05,
+      "loss": 1.2095,
+      "step": 4834000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.574503838137055e-05,
+      "loss": 1.1893,
+      "step": 4834500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.574293841580998e-05,
+      "loss": 1.1957,
+      "step": 4835000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5740838450249415e-05,
+      "loss": 1.206,
+      "step": 4835500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.573874268461997e-05,
+      "loss": 1.185,
+      "step": 4836000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.573664271905941e-05,
+      "loss": 1.1827,
+      "step": 4836500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.573454275349884e-05,
+      "loss": 1.1985,
+      "step": 4837000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5732442787938276e-05,
+      "loss": 1.1878,
+      "step": 4837500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5730342822377716e-05,
+      "loss": 1.2003,
+      "step": 4838000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.572824285681715e-05,
+      "loss": 1.1847,
+      "step": 4838500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.572614289125658e-05,
+      "loss": 1.1819,
+      "step": 4839000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.572404712562714e-05,
+      "loss": 1.194,
+      "step": 4839500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.572194716006658e-05,
+      "loss": 1.1941,
+      "step": 4840000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.571984719450601e-05,
+      "loss": 1.2302,
+      "step": 4840500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5717747228945444e-05,
+      "loss": 1.1904,
+      "step": 4841000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5715647263384884e-05,
+      "loss": 1.2105,
+      "step": 4841500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.571354729782431e-05,
+      "loss": 1.2244,
+      "step": 4842000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.571145573212599e-05,
+      "loss": 1.212,
+      "step": 4842500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5709355766565425e-05,
+      "loss": 1.2059,
+      "step": 4843000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5707255801004865e-05,
+      "loss": 1.1899,
+      "step": 4843500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.57051558354443e-05,
+      "loss": 1.1849,
+      "step": 4844000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.570305586988373e-05,
+      "loss": 1.1731,
+      "step": 4844500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.570095590432317e-05,
+      "loss": 1.2052,
+      "step": 4845000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5698855938762606e-05,
+      "loss": 1.2027,
+      "step": 4845500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.569675597320204e-05,
+      "loss": 1.2156,
+      "step": 4846000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.569466020757259e-05,
+      "loss": 1.1996,
+      "step": 4846500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.569256024201203e-05,
+      "loss": 1.1731,
+      "step": 4847000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5690460276451466e-05,
+      "loss": 1.1668,
+      "step": 4847500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.56883603108909e-05,
+      "loss": 1.1802,
+      "step": 4848000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.568626034533034e-05,
+      "loss": 1.2018,
+      "step": 4848500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5684160379769767e-05,
+      "loss": 1.1728,
+      "step": 4849000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.568206041420921e-05,
+      "loss": 1.1795,
+      "step": 4849500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.567996044864864e-05,
+      "loss": 1.2172,
+      "step": 4850000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.56778646830192e-05,
+      "loss": 1.2268,
+      "step": 4850500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5675764717458634e-05,
+      "loss": 1.2204,
+      "step": 4851000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.567366475189807e-05,
+      "loss": 1.1781,
+      "step": 4851500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.56715647863375e-05,
+      "loss": 1.1988,
+      "step": 4852000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5669464820776934e-05,
+      "loss": 1.2125,
+      "step": 4852500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5667364855216375e-05,
+      "loss": 1.2356,
+      "step": 4853000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.566526488965581e-05,
+      "loss": 1.1736,
+      "step": 4853500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.566316492409524e-05,
+      "loss": 1.1886,
+      "step": 4854000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5661069158465795e-05,
+      "loss": 1.2477,
+      "step": 4854500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5658969192905235e-05,
+      "loss": 1.165,
+      "step": 4855000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5656873427275796e-05,
+      "loss": 1.1912,
+      "step": 4855500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.565477346171522e-05,
+      "loss": 1.2064,
+      "step": 4856000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.565267349615466e-05,
+      "loss": 1.17,
+      "step": 4856500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5650573530594096e-05,
+      "loss": 1.2205,
+      "step": 4857000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.564847356503353e-05,
+      "loss": 1.2147,
+      "step": 4857500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.564637359947297e-05,
+      "loss": 1.1853,
+      "step": 4858000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.56442736339124e-05,
+      "loss": 1.2118,
+      "step": 4858500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.564217786828296e-05,
+      "loss": 1.2127,
+      "step": 4859000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.564007790272239e-05,
+      "loss": 1.2163,
+      "step": 4859500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.563797793716183e-05,
+      "loss": 1.2155,
+      "step": 4860000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5635877971601264e-05,
+      "loss": 1.2094,
+      "step": 4860500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.56337780060407e-05,
+      "loss": 1.192,
+      "step": 4861000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.563167804048014e-05,
+      "loss": 1.2084,
+      "step": 4861500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.562957807491957e-05,
+      "loss": 1.1808,
+      "step": 4862000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5627478109359005e-05,
+      "loss": 1.1785,
+      "step": 4862500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5625382343729565e-05,
+      "loss": 1.1689,
+      "step": 4863000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5623282378169e-05,
+      "loss": 1.198,
+      "step": 4863500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.562118661253955e-05,
+      "loss": 1.2205,
+      "step": 4864000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5619086646978986e-05,
+      "loss": 1.1966,
+      "step": 4864500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5616986681418426e-05,
+      "loss": 1.21,
+      "step": 4865000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.561488671585786e-05,
+      "loss": 1.2158,
+      "step": 4865500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.561278675029729e-05,
+      "loss": 1.2127,
+      "step": 4866000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5610690984667846e-05,
+      "loss": 1.2204,
+      "step": 4866500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5608591019107286e-05,
+      "loss": 1.1964,
+      "step": 4867000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.560649105354672e-05,
+      "loss": 1.1677,
+      "step": 4867500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5604391087986153e-05,
+      "loss": 1.1937,
+      "step": 4868000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5602291122425594e-05,
+      "loss": 1.2094,
+      "step": 4868500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.560019115686503e-05,
+      "loss": 1.2223,
+      "step": 4869000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.559809539123558e-05,
+      "loss": 1.2166,
+      "step": 4869500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.559599542567502e-05,
+      "loss": 1.2035,
+      "step": 4870000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5593895460114454e-05,
+      "loss": 1.2044,
+      "step": 4870500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.559179549455389e-05,
+      "loss": 1.1768,
+      "step": 4871000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.558969552899333e-05,
+      "loss": 1.2329,
+      "step": 4871500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.558759556343276e-05,
+      "loss": 1.1919,
+      "step": 4872000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5585495597872195e-05,
+      "loss": 1.2393,
+      "step": 4872500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5583395632311635e-05,
+      "loss": 1.1783,
+      "step": 4873000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.558129986668219e-05,
+      "loss": 1.2029,
+      "step": 4873500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.557920410105274e-05,
+      "loss": 1.197,
+      "step": 4874000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5577104135492176e-05,
+      "loss": 1.1909,
+      "step": 4874500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.557500416993161e-05,
+      "loss": 1.2012,
+      "step": 4875000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.557290420437105e-05,
+      "loss": 1.2581,
+      "step": 4875500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.557080423881048e-05,
+      "loss": 1.1802,
+      "step": 4876000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5568704273249916e-05,
+      "loss": 1.2095,
+      "step": 4876500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.556660850762048e-05,
+      "loss": 1.1637,
+      "step": 4877000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.556450854205991e-05,
+      "loss": 1.1899,
+      "step": 4877500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5562408576499344e-05,
+      "loss": 1.1925,
+      "step": 4878000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5560308610938784e-05,
+      "loss": 1.2125,
+      "step": 4878500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.555820864537822e-05,
+      "loss": 1.1718,
+      "step": 4879000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.555610867981765e-05,
+      "loss": 1.2255,
+      "step": 4879500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.555400871425709e-05,
+      "loss": 1.1551,
+      "step": 4880000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5551908748696525e-05,
+      "loss": 1.2055,
+      "step": 4880500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.554981298306708e-05,
+      "loss": 1.1939,
+      "step": 4881000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.554771301750651e-05,
+      "loss": 1.188,
+      "step": 4881500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.554561305194595e-05,
+      "loss": 1.2225,
+      "step": 4882000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5543517286316505e-05,
+      "loss": 1.169,
+      "step": 4882500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.554141732075594e-05,
+      "loss": 1.1823,
+      "step": 4883000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.553931735519537e-05,
+      "loss": 1.1724,
+      "step": 4883500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.553721738963481e-05,
+      "loss": 1.1869,
+      "step": 4884000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5535117424074246e-05,
+      "loss": 1.2344,
+      "step": 4884500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.55330216584448e-05,
+      "loss": 1.1936,
+      "step": 4885000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.553092169288424e-05,
+      "loss": 1.1978,
+      "step": 4885500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.552882172732367e-05,
+      "loss": 1.1808,
+      "step": 4886000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.552672176176311e-05,
+      "loss": 1.2096,
+      "step": 4886500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.552462179620255e-05,
+      "loss": 1.1803,
+      "step": 4887000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.552252183064198e-05,
+      "loss": 1.1873,
+      "step": 4887500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.552042186508141e-05,
+      "loss": 1.1726,
+      "step": 4888000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.551832609945197e-05,
+      "loss": 1.1765,
+      "step": 4888500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.551622613389141e-05,
+      "loss": 1.157,
+      "step": 4889000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.551412616833084e-05,
+      "loss": 1.1899,
+      "step": 4889500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5512026202770275e-05,
+      "loss": 1.2315,
+      "step": 4890000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.550992623720971e-05,
+      "loss": 1.201,
+      "step": 4890500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.550782627164914e-05,
+      "loss": 1.2075,
+      "step": 4891000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.55057305060197e-05,
+      "loss": 1.1848,
+      "step": 4891500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5503634740390255e-05,
+      "loss": 1.1803,
+      "step": 4892000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.550153897476081e-05,
+      "loss": 1.177,
+      "step": 4892500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.549943900920025e-05,
+      "loss": 1.2364,
+      "step": 4893000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.549733904363968e-05,
+      "loss": 1.2193,
+      "step": 4893500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5495239078079116e-05,
+      "loss": 1.1924,
+      "step": 4894000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5493139112518556e-05,
+      "loss": 1.1939,
+      "step": 4894500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.549103914695799e-05,
+      "loss": 1.1952,
+      "step": 4895000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.548893918139742e-05,
+      "loss": 1.2072,
+      "step": 4895500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5486839215836864e-05,
+      "loss": 1.1993,
+      "step": 4896000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.548474345020742e-05,
+      "loss": 1.2296,
+      "step": 4896500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.548264348464685e-05,
+      "loss": 1.2044,
+      "step": 4897000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5480543519086284e-05,
+      "loss": 1.2146,
+      "step": 4897500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5478443553525724e-05,
+      "loss": 1.213,
+      "step": 4898000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.547634358796516e-05,
+      "loss": 1.1903,
+      "step": 4898500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.547424362240459e-05,
+      "loss": 1.1772,
+      "step": 4899000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.547214365684403e-05,
+      "loss": 1.2092,
+      "step": 4899500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.547004369128346e-05,
+      "loss": 1.1931,
+      "step": 4900000
+    },
+    {
+      "epoch": 2.94,
+      "eval_loss": 1.150070071220398,
+      "eval_runtime": 1102.0417,
+      "eval_samples_per_second": 477.949,
+      "eval_steps_per_second": 79.659,
+      "step": 4900000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.546794792565402e-05,
+      "loss": 1.1686,
+      "step": 4900500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.546584796009346e-05,
+      "loss": 1.1786,
+      "step": 4901000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.546374799453289e-05,
+      "loss": 1.1834,
+      "step": 4901500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5461648028972326e-05,
+      "loss": 1.1955,
+      "step": 4902000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.545954806341176e-05,
+      "loss": 1.2104,
+      "step": 4902500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.545744809785119e-05,
+      "loss": 1.1799,
+      "step": 4903000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5455348132290626e-05,
+      "loss": 1.1846,
+      "step": 4903500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5453248166730066e-05,
+      "loss": 1.1922,
+      "step": 4904000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.545115240110062e-05,
+      "loss": 1.1906,
+      "step": 4904500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.544905243554005e-05,
+      "loss": 1.1742,
+      "step": 4905000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.544695246997949e-05,
+      "loss": 1.19,
+      "step": 4905500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.544485250441893e-05,
+      "loss": 1.1771,
+      "step": 4906000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.544275253885836e-05,
+      "loss": 1.2039,
+      "step": 4906500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5440656773228914e-05,
+      "loss": 1.2081,
+      "step": 4907000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5438556807668354e-05,
+      "loss": 1.1953,
+      "step": 4907500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.543645684210779e-05,
+      "loss": 1.1965,
+      "step": 4908000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.543436107647835e-05,
+      "loss": 1.1947,
+      "step": 4908500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.543226111091778e-05,
+      "loss": 1.176,
+      "step": 4909000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5430161145357215e-05,
+      "loss": 1.1858,
+      "step": 4909500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.542806117979665e-05,
+      "loss": 1.2069,
+      "step": 4910000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.542596121423608e-05,
+      "loss": 1.1935,
+      "step": 4910500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.542386124867552e-05,
+      "loss": 1.2113,
+      "step": 4911000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.542176548304608e-05,
+      "loss": 1.1929,
+      "step": 4911500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.541966551748551e-05,
+      "loss": 1.1827,
+      "step": 4912000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.541756555192494e-05,
+      "loss": 1.204,
+      "step": 4912500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.541546558636438e-05,
+      "loss": 1.1612,
+      "step": 4913000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5413365620803816e-05,
+      "loss": 1.1853,
+      "step": 4913500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.541126565524325e-05,
+      "loss": 1.1911,
+      "step": 4914000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.540916568968269e-05,
+      "loss": 1.2116,
+      "step": 4914500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5407065724122123e-05,
+      "loss": 1.1766,
+      "step": 4915000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5404965758561564e-05,
+      "loss": 1.1856,
+      "step": 4915500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5402865793001e-05,
+      "loss": 1.1894,
+      "step": 4916000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.540076582744043e-05,
+      "loss": 1.1785,
+      "step": 4916500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5398670061810984e-05,
+      "loss": 1.1615,
+      "step": 4917000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5396570096250424e-05,
+      "loss": 1.1646,
+      "step": 4917500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.539447013068986e-05,
+      "loss": 1.1889,
+      "step": 4918000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.539237016512929e-05,
+      "loss": 1.1881,
+      "step": 4918500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5390274399499845e-05,
+      "loss": 1.1956,
+      "step": 4919000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5388174433939285e-05,
+      "loss": 1.1853,
+      "step": 4919500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.538607446837872e-05,
+      "loss": 1.2042,
+      "step": 4920000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.538397450281815e-05,
+      "loss": 1.1773,
+      "step": 4920500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.538187453725759e-05,
+      "loss": 1.2138,
+      "step": 4921000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5379774571697026e-05,
+      "loss": 1.1997,
+      "step": 4921500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.537767880606758e-05,
+      "loss": 1.2098,
+      "step": 4922000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.537557884050702e-05,
+      "loss": 1.1951,
+      "step": 4922500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.537347887494645e-05,
+      "loss": 1.1966,
+      "step": 4923000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5371378909385887e-05,
+      "loss": 1.2012,
+      "step": 4923500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.536927894382533e-05,
+      "loss": 1.1754,
+      "step": 4924000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5367178978264753e-05,
+      "loss": 1.1853,
+      "step": 4924500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.536507901270419e-05,
+      "loss": 1.1625,
+      "step": 4925000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.536297904714363e-05,
+      "loss": 1.2095,
+      "step": 4925500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.536087908158306e-05,
+      "loss": 1.1875,
+      "step": 4926000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5358779116022494e-05,
+      "loss": 1.1759,
+      "step": 4926500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5356679150461934e-05,
+      "loss": 1.1836,
+      "step": 4927000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.535457918490137e-05,
+      "loss": 1.1699,
+      "step": 4927500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.535248341927192e-05,
+      "loss": 1.1825,
+      "step": 4928000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5350383453711355e-05,
+      "loss": 1.1832,
+      "step": 4928500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5348283488150795e-05,
+      "loss": 1.1735,
+      "step": 4929000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.534618352259023e-05,
+      "loss": 1.2161,
+      "step": 4929500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.534408775696078e-05,
+      "loss": 1.1926,
+      "step": 4930000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.534198779140022e-05,
+      "loss": 1.1985,
+      "step": 4930500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5339887825839656e-05,
+      "loss": 1.1791,
+      "step": 4931000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.533779206021021e-05,
+      "loss": 1.2029,
+      "step": 4931500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.533569209464964e-05,
+      "loss": 1.2021,
+      "step": 4932000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.533359212908908e-05,
+      "loss": 1.1804,
+      "step": 4932500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5331492163528516e-05,
+      "loss": 1.2075,
+      "step": 4933000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.532939639789908e-05,
+      "loss": 1.2154,
+      "step": 4933500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5327296432338503e-05,
+      "loss": 1.1871,
+      "step": 4934000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5325196466777944e-05,
+      "loss": 1.1961,
+      "step": 4934500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.532309650121738e-05,
+      "loss": 1.1789,
+      "step": 4935000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.532099653565681e-05,
+      "loss": 1.1838,
+      "step": 4935500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.531889657009625e-05,
+      "loss": 1.1975,
+      "step": 4936000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5316796604535684e-05,
+      "loss": 1.1985,
+      "step": 4936500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.531469663897512e-05,
+      "loss": 1.1751,
+      "step": 4937000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.531259667341456e-05,
+      "loss": 1.1921,
+      "step": 4937500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.531049670785399e-05,
+      "loss": 1.1814,
+      "step": 4938000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5308400942224545e-05,
+      "loss": 1.1779,
+      "step": 4938500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5306300976663985e-05,
+      "loss": 1.2068,
+      "step": 4939000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.530420101110342e-05,
+      "loss": 1.1964,
+      "step": 4939500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.530210524547397e-05,
+      "loss": 1.1794,
+      "step": 4940000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5300005279913406e-05,
+      "loss": 1.2067,
+      "step": 4940500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5297905314352846e-05,
+      "loss": 1.203,
+      "step": 4941000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.529580534879228e-05,
+      "loss": 1.1975,
+      "step": 4941500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.529370538323171e-05,
+      "loss": 1.1813,
+      "step": 4942000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.529160541767115e-05,
+      "loss": 1.1987,
+      "step": 4942500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.528950965204171e-05,
+      "loss": 1.1871,
+      "step": 4943000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.528740968648114e-05,
+      "loss": 1.1964,
+      "step": 4943500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5285309720920574e-05,
+      "loss": 1.1983,
+      "step": 4944000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5283209755360014e-05,
+      "loss": 1.1786,
+      "step": 4944500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.528110978979945e-05,
+      "loss": 1.1886,
+      "step": 4945000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.527900982423888e-05,
+      "loss": 1.1916,
+      "step": 4945500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.527690985867832e-05,
+      "loss": 1.2012,
+      "step": 4946000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.527480989311775e-05,
+      "loss": 1.1637,
+      "step": 4946500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.527270992755719e-05,
+      "loss": 1.2104,
+      "step": 4947000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.527061416192775e-05,
+      "loss": 1.1895,
+      "step": 4947500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.526851419636718e-05,
+      "loss": 1.2104,
+      "step": 4948000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5266414230806615e-05,
+      "loss": 1.1835,
+      "step": 4948500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.526431426524605e-05,
+      "loss": 1.1696,
+      "step": 4949000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.526221429968548e-05,
+      "loss": 1.1902,
+      "step": 4949500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5260114334124916e-05,
+      "loss": 1.1867,
+      "step": 4950000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5258014368564356e-05,
+      "loss": 1.1919,
+      "step": 4950500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5255918602934916e-05,
+      "loss": 1.1933,
+      "step": 4951000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.525381863737434e-05,
+      "loss": 1.1714,
+      "step": 4951500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5251718671813776e-05,
+      "loss": 1.2231,
+      "step": 4952000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5249618706253217e-05,
+      "loss": 1.1904,
+      "step": 4952500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.524751874069265e-05,
+      "loss": 1.2286,
+      "step": 4953000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5245418775132084e-05,
+      "loss": 1.2117,
+      "step": 4953500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5243318809571524e-05,
+      "loss": 1.1688,
+      "step": 4954000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.524122304394208e-05,
+      "loss": 1.1621,
+      "step": 4954500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.523912307838151e-05,
+      "loss": 1.2109,
+      "step": 4955000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.523702311282095e-05,
+      "loss": 1.2456,
+      "step": 4955500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5234923147260384e-05,
+      "loss": 1.1945,
+      "step": 4956000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.523282738163094e-05,
+      "loss": 1.1965,
+      "step": 4956500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.523072741607037e-05,
+      "loss": 1.1922,
+      "step": 4957000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.522862745050981e-05,
+      "loss": 1.1867,
+      "step": 4957500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5226527484949245e-05,
+      "loss": 1.2077,
+      "step": 4958000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.522442751938868e-05,
+      "loss": 1.2082,
+      "step": 4958500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.522232755382812e-05,
+      "loss": 1.1866,
+      "step": 4959000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.522022758826755e-05,
+      "loss": 1.1989,
+      "step": 4959500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5218131822638106e-05,
+      "loss": 1.1908,
+      "step": 4960000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.521603185707754e-05,
+      "loss": 1.2024,
+      "step": 4960500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.521393189151698e-05,
+      "loss": 1.2216,
+      "step": 4961000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.521183192595641e-05,
+      "loss": 1.1952,
+      "step": 4961500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.520973196039585e-05,
+      "loss": 1.1668,
+      "step": 4962000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.520763619476641e-05,
+      "loss": 1.2139,
+      "step": 4962500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.520553622920584e-05,
+      "loss": 1.2309,
+      "step": 4963000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5203436263645274e-05,
+      "loss": 1.1917,
+      "step": 4963500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5201336298084714e-05,
+      "loss": 1.1939,
+      "step": 4964000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.519923633252415e-05,
+      "loss": 1.1804,
+      "step": 4964500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.519713636696358e-05,
+      "loss": 1.1879,
+      "step": 4965000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.519503640140302e-05,
+      "loss": 1.1878,
+      "step": 4965500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5192940635773575e-05,
+      "loss": 1.1847,
+      "step": 4966000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.519084067021301e-05,
+      "loss": 1.227,
+      "step": 4966500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.518874070465244e-05,
+      "loss": 1.1889,
+      "step": 4967000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.518664073909188e-05,
+      "loss": 1.1746,
+      "step": 4967500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5184540773531315e-05,
+      "loss": 1.1893,
+      "step": 4968000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.518244500790187e-05,
+      "loss": 1.1907,
+      "step": 4968500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.518034504234131e-05,
+      "loss": 1.1976,
+      "step": 4969000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.517824507678074e-05,
+      "loss": 1.1862,
+      "step": 4969500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5176145111220176e-05,
+      "loss": 1.2249,
+      "step": 4970000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5174045145659616e-05,
+      "loss": 1.1959,
+      "step": 4970500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.517194518009905e-05,
+      "loss": 1.1911,
+      "step": 4971000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5169845214538476e-05,
+      "loss": 1.1862,
+      "step": 4971500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.516774524897792e-05,
+      "loss": 1.1583,
+      "step": 4972000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.516564948334848e-05,
+      "loss": 1.1852,
+      "step": 4972500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.516354951778791e-05,
+      "loss": 1.165,
+      "step": 4973000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.516144955222734e-05,
+      "loss": 1.1931,
+      "step": 4973500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.515934958666678e-05,
+      "loss": 1.1914,
+      "step": 4974000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.515725802096846e-05,
+      "loss": 1.1932,
+      "step": 4974500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.515515805540789e-05,
+      "loss": 1.2081,
+      "step": 4975000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5153058089847325e-05,
+      "loss": 1.172,
+      "step": 4975500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5150958124286765e-05,
+      "loss": 1.1788,
+      "step": 4976000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.51488581587262e-05,
+      "loss": 1.213,
+      "step": 4976500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.514675819316563e-05,
+      "loss": 1.1947,
+      "step": 4977000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.514465822760507e-05,
+      "loss": 1.2347,
+      "step": 4977500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5142558262044506e-05,
+      "loss": 1.2143,
+      "step": 4978000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.514046249641506e-05,
+      "loss": 1.1915,
+      "step": 4978500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.513836253085449e-05,
+      "loss": 1.195,
+      "step": 4979000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.513626256529393e-05,
+      "loss": 1.1789,
+      "step": 4979500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5134162599733366e-05,
+      "loss": 1.2034,
+      "step": 4980000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.51320626341728e-05,
+      "loss": 1.1813,
+      "step": 4980500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5129966868543353e-05,
+      "loss": 1.2187,
+      "step": 4981000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5127866902982794e-05,
+      "loss": 1.1764,
+      "step": 4981500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.512577113735335e-05,
+      "loss": 1.1802,
+      "step": 4982000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.512367117179278e-05,
+      "loss": 1.1749,
+      "step": 4982500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.512157120623222e-05,
+      "loss": 1.1856,
+      "step": 4983000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5119471240671654e-05,
+      "loss": 1.173,
+      "step": 4983500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.511737127511109e-05,
+      "loss": 1.1714,
+      "step": 4984000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.511527130955053e-05,
+      "loss": 1.1896,
+      "step": 4984500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.511317134398996e-05,
+      "loss": 1.2135,
+      "step": 4985000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5111075578360515e-05,
+      "loss": 1.1937,
+      "step": 4985500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.510897561279995e-05,
+      "loss": 1.2188,
+      "step": 4986000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.510687564723939e-05,
+      "loss": 1.235,
+      "step": 4986500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.510477568167882e-05,
+      "loss": 1.2255,
+      "step": 4987000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5102675716118256e-05,
+      "loss": 1.2107,
+      "step": 4987500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.510057575055769e-05,
+      "loss": 1.2078,
+      "step": 4988000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.509847578499712e-05,
+      "loss": 1.2072,
+      "step": 4988500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.509638001936768e-05,
+      "loss": 1.1822,
+      "step": 4989000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5094280053807116e-05,
+      "loss": 1.1902,
+      "step": 4989500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.509218008824656e-05,
+      "loss": 1.2092,
+      "step": 4990000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5090080122685983e-05,
+      "loss": 1.1944,
+      "step": 4990500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5087980157125424e-05,
+      "loss": 1.2213,
+      "step": 4991000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.508588019156486e-05,
+      "loss": 1.2173,
+      "step": 4991500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.508378022600429e-05,
+      "loss": 1.1726,
+      "step": 4992000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.508168026044373e-05,
+      "loss": 1.2283,
+      "step": 4992500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5079580294883164e-05,
+      "loss": 1.2095,
+      "step": 4993000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.507748452925372e-05,
+      "loss": 1.2043,
+      "step": 4993500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.507538456369315e-05,
+      "loss": 1.1754,
+      "step": 4994000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.507328459813259e-05,
+      "loss": 1.2093,
+      "step": 4994500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5071184632572025e-05,
+      "loss": 1.2137,
+      "step": 4995000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.506908886694258e-05,
+      "loss": 1.234,
+      "step": 4995500
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.506699310131314e-05,
+      "loss": 1.1872,
+      "step": 4996000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.506489313575257e-05,
+      "loss": 1.1619,
+      "step": 4996500
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.506279317019201e-05,
+      "loss": 1.2051,
+      "step": 4997000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.506069320463144e-05,
+      "loss": 1.2021,
+      "step": 4997500
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5058597439002e-05,
+      "loss": 1.1838,
+      "step": 4998000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.505649747344144e-05,
+      "loss": 1.1752,
+      "step": 4998500
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.505439750788087e-05,
+      "loss": 1.2044,
+      "step": 4999000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.505229754232031e-05,
+      "loss": 1.1751,
+      "step": 4999500
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.505019757675974e-05,
+      "loss": 1.2131,
+      "step": 5000000
+    },
+    {
+      "epoch": 3.0,
+      "eval_loss": 1.148377776145935,
+      "eval_runtime": 1098.3919,
+      "eval_samples_per_second": 479.537,
+      "eval_steps_per_second": 79.923,
+      "step": 5000000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5048097611199174e-05,
+      "loss": 1.1856,
+      "step": 5000500
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.504599764563861e-05,
+      "loss": 1.2012,
+      "step": 5001000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.504389768007805e-05,
+      "loss": 1.2196,
+      "step": 5001500
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.504180191444861e-05,
+      "loss": 1.194,
+      "step": 5002000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5039701948888034e-05,
+      "loss": 1.1778,
+      "step": 5002500
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.503760198332747e-05,
+      "loss": 1.1772,
+      "step": 5003000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.503550201776691e-05,
+      "loss": 1.185,
+      "step": 5003500
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.503340205220634e-05,
+      "loss": 1.2015,
+      "step": 5004000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5031302086645775e-05,
+      "loss": 1.1376,
+      "step": 5004500
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5029202121085215e-05,
+      "loss": 1.1188,
+      "step": 5005000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.502710215552465e-05,
+      "loss": 1.139,
+      "step": 5005500
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.502500218996408e-05,
+      "loss": 1.1583,
+      "step": 5006000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.502290222440352e-05,
+      "loss": 1.1542,
+      "step": 5006500
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5020802258842956e-05,
+      "loss": 1.1757,
+      "step": 5007000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5018702293282396e-05,
+      "loss": 1.1799,
+      "step": 5007500
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.501660232772182e-05,
+      "loss": 1.1484,
+      "step": 5008000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5014502362161256e-05,
+      "loss": 1.1543,
+      "step": 5008500
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5012402396600696e-05,
+      "loss": 1.125,
+      "step": 5009000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.501030243104013e-05,
+      "loss": 1.147,
+      "step": 5009500
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5008206665410684e-05,
+      "loss": 1.1666,
+      "step": 5010000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.500610669985012e-05,
+      "loss": 1.1843,
+      "step": 5010500
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.500400673428956e-05,
+      "loss": 1.1633,
+      "step": 5011000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.500190676872899e-05,
+      "loss": 1.1559,
+      "step": 5011500
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.499981100309955e-05,
+      "loss": 1.1695,
+      "step": 5012000
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.499771103753898e-05,
+      "loss": 1.1516,
+      "step": 5012500
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.499561107197842e-05,
+      "loss": 1.1696,
+      "step": 5013000
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.499351110641785e-05,
+      "loss": 1.1501,
+      "step": 5013500
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.499141534078841e-05,
+      "loss": 1.152,
+      "step": 5014000
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.498931537522785e-05,
+      "loss": 1.1424,
+      "step": 5014500
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.498721540966728e-05,
+      "loss": 1.1464,
+      "step": 5015000
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.498511544410671e-05,
+      "loss": 1.1582,
+      "step": 5015500
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.498301547854615e-05,
+      "loss": 1.1566,
+      "step": 5016000
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.498091971291671e-05,
+      "loss": 1.1515,
+      "step": 5016500
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4978819747356146e-05,
+      "loss": 1.1634,
+      "step": 5017000
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.497671978179557e-05,
+      "loss": 1.1652,
+      "step": 5017500
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.497462401616613e-05,
+      "loss": 1.1473,
+      "step": 5018000
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4972524050605573e-05,
+      "loss": 1.1453,
+      "step": 5018500
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.497042408504501e-05,
+      "loss": 1.161,
+      "step": 5019000
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4968324119484434e-05,
+      "loss": 1.18,
+      "step": 5019500
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4966224153923874e-05,
+      "loss": 1.1652,
+      "step": 5020000
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.496412418836331e-05,
+      "loss": 1.1181,
+      "step": 5020500
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.496202422280275e-05,
+      "loss": 1.143,
+      "step": 5021000
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.495992425724218e-05,
+      "loss": 1.1664,
+      "step": 5021500
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4957824291681614e-05,
+      "loss": 1.1568,
+      "step": 5022000
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.495572852605217e-05,
+      "loss": 1.1728,
+      "step": 5022500
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.495362856049161e-05,
+      "loss": 1.134,
+      "step": 5023000
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.495152859493104e-05,
+      "loss": 1.1872,
+      "step": 5023500
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4949428629370475e-05,
+      "loss": 1.1649,
+      "step": 5024000
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4947337063672156e-05,
+      "loss": 1.1377,
+      "step": 5024500
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.494523709811159e-05,
+      "loss": 1.1376,
+      "step": 5025000
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.494313713255103e-05,
+      "loss": 1.1705,
+      "step": 5025500
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.494103716699046e-05,
+      "loss": 1.1507,
+      "step": 5026000
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4938937201429896e-05,
+      "loss": 1.1481,
+      "step": 5026500
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4936841435800457e-05,
+      "loss": 1.1766,
+      "step": 5027000
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.493474147023989e-05,
+      "loss": 1.1501,
+      "step": 5027500
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4932641504679324e-05,
+      "loss": 1.1493,
+      "step": 5028000
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4930541539118764e-05,
+      "loss": 1.1341,
+      "step": 5028500
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.492844157355819e-05,
+      "loss": 1.1606,
+      "step": 5029000
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.492634580792875e-05,
+      "loss": 1.1669,
+      "step": 5029500
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4924245842368184e-05,
+      "loss": 1.1753,
+      "step": 5030000
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4922145876807624e-05,
+      "loss": 1.17,
+      "step": 5030500
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.492004591124706e-05,
+      "loss": 1.1479,
+      "step": 5031000
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4917945945686485e-05,
+      "loss": 1.1393,
+      "step": 5031500
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4915845980125925e-05,
+      "loss": 1.1538,
+      "step": 5032000
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.491374601456536e-05,
+      "loss": 1.149,
+      "step": 5032500
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.491164604900479e-05,
+      "loss": 1.1827,
+      "step": 5033000
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.490955028337535e-05,
+      "loss": 1.1527,
+      "step": 5033500
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4907450317814786e-05,
+      "loss": 1.1285,
+      "step": 5034000
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.490535035225422e-05,
+      "loss": 1.1637,
+      "step": 5034500
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.490325038669366e-05,
+      "loss": 1.1503,
+      "step": 5035000
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.490115042113309e-05,
+      "loss": 1.1693,
+      "step": 5035500
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.489905465550365e-05,
+      "loss": 1.1443,
+      "step": 5036000
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.489695468994308e-05,
+      "loss": 1.188,
+      "step": 5036500
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.489485472438252e-05,
+      "loss": 1.1658,
+      "step": 5037000
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4892754758821953e-05,
+      "loss": 1.1719,
+      "step": 5037500
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.489065479326139e-05,
+      "loss": 1.1546,
+      "step": 5038000
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.488855482770083e-05,
+      "loss": 1.158,
+      "step": 5038500
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.488645486214026e-05,
+      "loss": 1.1637,
+      "step": 5039000
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4884354896579694e-05,
+      "loss": 1.1619,
+      "step": 5039500
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.488225913095025e-05,
+      "loss": 1.1549,
+      "step": 5040000
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.488015916538969e-05,
+      "loss": 1.1661,
+      "step": 5040500
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.487806339976024e-05,
+      "loss": 1.1758,
+      "step": 5041000
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4875963434199675e-05,
+      "loss": 1.1814,
+      "step": 5041500
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4873863468639115e-05,
+      "loss": 1.1791,
+      "step": 5042000
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.487176350307855e-05,
+      "loss": 1.1868,
+      "step": 5042500
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.486966353751798e-05,
+      "loss": 1.1497,
+      "step": 5043000
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4867567771888536e-05,
+      "loss": 1.1821,
+      "step": 5043500
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4865467806327976e-05,
+      "loss": 1.1656,
+      "step": 5044000
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.486336784076741e-05,
+      "loss": 1.1687,
+      "step": 5044500
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.486126787520684e-05,
+      "loss": 1.1724,
+      "step": 5045000
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.485916790964628e-05,
+      "loss": 1.1566,
+      "step": 5045500
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4857067944085716e-05,
+      "loss": 1.1816,
+      "step": 5046000
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.485496797852515e-05,
+      "loss": 1.1953,
+      "step": 5046500
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4852872212895704e-05,
+      "loss": 1.1385,
+      "step": 5047000
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4850772247335144e-05,
+      "loss": 1.1679,
+      "step": 5047500
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.484867228177458e-05,
+      "loss": 1.1657,
+      "step": 5048000
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.484657231621401e-05,
+      "loss": 1.143,
+      "step": 5048500
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.484447235065345e-05,
+      "loss": 1.1685,
+      "step": 5049000
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4842372385092884e-05,
+      "loss": 1.1395,
+      "step": 5049500
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.484027241953232e-05,
+      "loss": 1.1605,
+      "step": 5050000
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.483817245397176e-05,
+      "loss": 1.1602,
+      "step": 5050500
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.483607668834231e-05,
+      "loss": 1.1565,
+      "step": 5051000
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4833976722781745e-05,
+      "loss": 1.1485,
+      "step": 5051500
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4831876757221185e-05,
+      "loss": 1.1394,
+      "step": 5052000
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.482977679166062e-05,
+      "loss": 1.1427,
+      "step": 5052500
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.482768102603117e-05,
+      "loss": 1.1716,
+      "step": 5053000
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4825581060470606e-05,
+      "loss": 1.1655,
+      "step": 5053500
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4823481094910046e-05,
+      "loss": 1.1793,
+      "step": 5054000
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.482138112934948e-05,
+      "loss": 1.1777,
+      "step": 5054500
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.481928536372003e-05,
+      "loss": 1.1851,
+      "step": 5055000
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4817185398159467e-05,
+      "loss": 1.1471,
+      "step": 5055500
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.481508963253003e-05,
+      "loss": 1.1726,
+      "step": 5056000
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.481298966696946e-05,
+      "loss": 1.1588,
+      "step": 5056500
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4810889701408894e-05,
+      "loss": 1.1643,
+      "step": 5057000
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4808789735848334e-05,
+      "loss": 1.1906,
+      "step": 5057500
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.480668977028777e-05,
+      "loss": 1.1519,
+      "step": 5058000
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.48045898047272e-05,
+      "loss": 1.1751,
+      "step": 5058500
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.480248983916664e-05,
+      "loss": 1.1854,
+      "step": 5059000
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4800389873606075e-05,
+      "loss": 1.1599,
+      "step": 5059500
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.479829410797663e-05,
+      "loss": 1.1621,
+      "step": 5060000
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.479619834234718e-05,
+      "loss": 1.1936,
+      "step": 5060500
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4794098376786615e-05,
+      "loss": 1.1335,
+      "step": 5061000
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4791998411226056e-05,
+      "loss": 1.1487,
+      "step": 5061500
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.478989844566549e-05,
+      "loss": 1.1465,
+      "step": 5062000
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.478779848010492e-05,
+      "loss": 1.1591,
+      "step": 5062500
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.478569851454436e-05,
+      "loss": 1.1636,
+      "step": 5063000
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4783598548983796e-05,
+      "loss": 1.145,
+      "step": 5063500
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.478149858342323e-05,
+      "loss": 1.1814,
+      "step": 5064000
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.477940281779379e-05,
+      "loss": 1.1652,
+      "step": 5064500
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4777307052164344e-05,
+      "loss": 1.1713,
+      "step": 5065000
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.477520708660378e-05,
+      "loss": 1.1264,
+      "step": 5065500
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.477310712104321e-05,
+      "loss": 1.1617,
+      "step": 5066000
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.477100715548265e-05,
+      "loss": 1.1768,
+      "step": 5066500
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4768907189922084e-05,
+      "loss": 1.1809,
+      "step": 5067000
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.476680722436152e-05,
+      "loss": 1.1635,
+      "step": 5067500
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.476470725880096e-05,
+      "loss": 1.1727,
+      "step": 5068000
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.476261149317151e-05,
+      "loss": 1.185,
+      "step": 5068500
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4760511527610945e-05,
+      "loss": 1.1826,
+      "step": 5069000
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.475841156205038e-05,
+      "loss": 1.1794,
+      "step": 5069500
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.475631579642094e-05,
+      "loss": 1.1416,
+      "step": 5070000
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.475421583086037e-05,
+      "loss": 1.1545,
+      "step": 5070500
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4752115865299806e-05,
+      "loss": 1.1677,
+      "step": 5071000
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4750015899739246e-05,
+      "loss": 1.1722,
+      "step": 5071500
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.474791593417868e-05,
+      "loss": 1.1362,
+      "step": 5072000
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.474581596861811e-05,
+      "loss": 1.1738,
+      "step": 5072500
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.474371600305755e-05,
+      "loss": 1.1662,
+      "step": 5073000
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4741616037496986e-05,
+      "loss": 1.1414,
+      "step": 5073500
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.473951607193642e-05,
+      "loss": 1.1692,
+      "step": 5074000
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.473741610637586e-05,
+      "loss": 1.163,
+      "step": 5074500
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4735316140815294e-05,
+      "loss": 1.1498,
+      "step": 5075000
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.473321617525472e-05,
+      "loss": 1.1321,
+      "step": 5075500
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.473112040962528e-05,
+      "loss": 1.1773,
+      "step": 5076000
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4729024643995834e-05,
+      "loss": 1.1429,
+      "step": 5076500
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4726928878366395e-05,
+      "loss": 1.1393,
+      "step": 5077000
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.472482891280583e-05,
+      "loss": 1.1425,
+      "step": 5077500
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.472272894724526e-05,
+      "loss": 1.159,
+      "step": 5078000
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.47206289816847e-05,
+      "loss": 1.1996,
+      "step": 5078500
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4718529016124135e-05,
+      "loss": 1.1572,
+      "step": 5079000
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.471642905056357e-05,
+      "loss": 1.1635,
+      "step": 5079500
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.471432908500301e-05,
+      "loss": 1.1472,
+      "step": 5080000
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.471222911944244e-05,
+      "loss": 1.1643,
+      "step": 5080500
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4710129153881876e-05,
+      "loss": 1.1504,
+      "step": 5081000
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4708029188321316e-05,
+      "loss": 1.1789,
+      "step": 5081500
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.470593342269187e-05,
+      "loss": 1.1777,
+      "step": 5082000
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.47038334571313e-05,
+      "loss": 1.1615,
+      "step": 5082500
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4701733491570736e-05,
+      "loss": 1.178,
+      "step": 5083000
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.469963352601018e-05,
+      "loss": 1.1391,
+      "step": 5083500
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.469753356044961e-05,
+      "loss": 1.178,
+      "step": 5084000
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4695433594889044e-05,
+      "loss": 1.1655,
+      "step": 5084500
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.469333362932848e-05,
+      "loss": 1.1338,
+      "step": 5085000
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.469123366376791e-05,
+      "loss": 1.1757,
+      "step": 5085500
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.468913789813847e-05,
+      "loss": 1.1656,
+      "step": 5086000
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.468703793257791e-05,
+      "loss": 1.1735,
+      "step": 5086500
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.468493796701734e-05,
+      "loss": 1.1662,
+      "step": 5087000
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.468283800145677e-05,
+      "loss": 1.1587,
+      "step": 5087500
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.468074223582733e-05,
+      "loss": 1.1637,
+      "step": 5088000
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4678646470197885e-05,
+      "loss": 1.156,
+      "step": 5088500
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4676546504637325e-05,
+      "loss": 1.1744,
+      "step": 5089000
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.467444653907676e-05,
+      "loss": 1.1842,
+      "step": 5089500
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.467234657351619e-05,
+      "loss": 1.1882,
+      "step": 5090000
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.467024660795563e-05,
+      "loss": 1.1722,
+      "step": 5090500
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4668146642395066e-05,
+      "loss": 1.1785,
+      "step": 5091000
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.466605087676562e-05,
+      "loss": 1.1747,
+      "step": 5091500
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.466395091120505e-05,
+      "loss": 1.1917,
+      "step": 5092000
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.466185094564449e-05,
+      "loss": 1.1643,
+      "step": 5092500
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.465975098008393e-05,
+      "loss": 1.1629,
+      "step": 5093000
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.465765101452337e-05,
+      "loss": 1.1315,
+      "step": 5093500
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.46555510489628e-05,
+      "loss": 1.1755,
+      "step": 5094000
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4653455283333354e-05,
+      "loss": 1.1576,
+      "step": 5094500
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.465135531777279e-05,
+      "loss": 1.1908,
+      "step": 5095000
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.464925535221223e-05,
+      "loss": 1.1754,
+      "step": 5095500
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.464715538665166e-05,
+      "loss": 1.1544,
+      "step": 5096000
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4645059621022215e-05,
+      "loss": 1.1655,
+      "step": 5096500
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.464295965546165e-05,
+      "loss": 1.1512,
+      "step": 5097000
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.464085968990109e-05,
+      "loss": 1.1611,
+      "step": 5097500
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.463875972434052e-05,
+      "loss": 1.1394,
+      "step": 5098000
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4636659758779955e-05,
+      "loss": 1.159,
+      "step": 5098500
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.463455979321939e-05,
+      "loss": 1.1599,
+      "step": 5099000
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.463246402758995e-05,
+      "loss": 1.1619,
+      "step": 5099500
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.463036406202938e-05,
+      "loss": 1.1291,
+      "step": 5100000
+    },
+    {
+      "epoch": 3.06,
+      "eval_loss": 1.1477725505828857,
+      "eval_runtime": 1108.108,
+      "eval_samples_per_second": 475.333,
+      "eval_steps_per_second": 79.222,
+      "step": 5100000
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.462826409646882e-05,
+      "loss": 1.1777,
+      "step": 5100500
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4626164130908256e-05,
+      "loss": 1.151,
+      "step": 5101000
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.462406416534768e-05,
+      "loss": 1.1536,
+      "step": 5101500
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.462196419978712e-05,
+      "loss": 1.1529,
+      "step": 5102000
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.461986423422656e-05,
+      "loss": 1.1782,
+      "step": 5102500
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.461776426866599e-05,
+      "loss": 1.1513,
+      "step": 5103000
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.461566850303655e-05,
+      "loss": 1.1653,
+      "step": 5103500
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4613568537475984e-05,
+      "loss": 1.1866,
+      "step": 5104000
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4611472771846544e-05,
+      "loss": 1.1638,
+      "step": 5104500
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.460937280628598e-05,
+      "loss": 1.1641,
+      "step": 5105000
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.460727284072541e-05,
+      "loss": 1.1569,
+      "step": 5105500
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4605172875164845e-05,
+      "loss": 1.1469,
+      "step": 5106000
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.460307290960428e-05,
+      "loss": 1.1823,
+      "step": 5106500
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.460097294404372e-05,
+      "loss": 1.1686,
+      "step": 5107000
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.459887297848315e-05,
+      "loss": 1.158,
+      "step": 5107500
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.459677721285371e-05,
+      "loss": 1.1699,
+      "step": 5108000
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.459467724729314e-05,
+      "loss": 1.1607,
+      "step": 5108500
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.459257728173258e-05,
+      "loss": 1.1454,
+      "step": 5109000
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.459047731617201e-05,
+      "loss": 1.1684,
+      "step": 5109500
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4588377350611446e-05,
+      "loss": 1.162,
+      "step": 5110000
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4586277385050886e-05,
+      "loss": 1.1588,
+      "step": 5110500
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.458417741949032e-05,
+      "loss": 1.1503,
+      "step": 5111000
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.458207745392975e-05,
+      "loss": 1.1632,
+      "step": 5111500
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4579985888231434e-05,
+      "loss": 1.1492,
+      "step": 5112000
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.457788592267087e-05,
+      "loss": 1.1504,
+      "step": 5112500
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.457578595711031e-05,
+      "loss": 1.1795,
+      "step": 5113000
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4573685991549734e-05,
+      "loss": 1.1777,
+      "step": 5113500
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4571586025989174e-05,
+      "loss": 1.1716,
+      "step": 5114000
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4569490260359735e-05,
+      "loss": 1.1592,
+      "step": 5114500
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.456739029479917e-05,
+      "loss": 1.1643,
+      "step": 5115000
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.45652903292386e-05,
+      "loss": 1.1944,
+      "step": 5115500
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4563190363678035e-05,
+      "loss": 1.1755,
+      "step": 5116000
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.456109039811747e-05,
+      "loss": 1.1651,
+      "step": 5116500
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.45589904325569e-05,
+      "loss": 1.1571,
+      "step": 5117000
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.455689046699634e-05,
+      "loss": 1.1635,
+      "step": 5117500
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4554794701366896e-05,
+      "loss": 1.1843,
+      "step": 5118000
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.455269473580633e-05,
+      "loss": 1.1745,
+      "step": 5118500
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.455059477024576e-05,
+      "loss": 1.1586,
+      "step": 5119000
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.45484948046852e-05,
+      "loss": 1.1568,
+      "step": 5119500
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4546394839124636e-05,
+      "loss": 1.1885,
+      "step": 5120000
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.454429487356407e-05,
+      "loss": 1.1882,
+      "step": 5120500
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.454219490800351e-05,
+      "loss": 1.1538,
+      "step": 5121000
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4540094942442944e-05,
+      "loss": 1.1836,
+      "step": 5121500
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.45379991768135e-05,
+      "loss": 1.1663,
+      "step": 5122000
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.453590341118406e-05,
+      "loss": 1.118,
+      "step": 5122500
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.453380344562349e-05,
+      "loss": 1.1574,
+      "step": 5123000
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4531703480062924e-05,
+      "loss": 1.1688,
+      "step": 5123500
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.452960351450236e-05,
+      "loss": 1.1644,
+      "step": 5124000
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.45275035489418e-05,
+      "loss": 1.1635,
+      "step": 5124500
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.452540358338123e-05,
+      "loss": 1.149,
+      "step": 5125000
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4523307817751785e-05,
+      "loss": 1.1669,
+      "step": 5125500
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.452120785219122e-05,
+      "loss": 1.1293,
+      "step": 5126000
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.451910788663066e-05,
+      "loss": 1.1639,
+      "step": 5126500
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.451700792107009e-05,
+      "loss": 1.1415,
+      "step": 5127000
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4514907955509526e-05,
+      "loss": 1.1787,
+      "step": 5127500
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4512807989948966e-05,
+      "loss": 1.1856,
+      "step": 5128000
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.45107080243884e-05,
+      "loss": 1.1218,
+      "step": 5128500
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.450860805882783e-05,
+      "loss": 1.1869,
+      "step": 5129000
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.450651229319839e-05,
+      "loss": 1.1676,
+      "step": 5129500
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.450441652756895e-05,
+      "loss": 1.1703,
+      "step": 5130000
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.450231656200838e-05,
+      "loss": 1.1621,
+      "step": 5130500
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4500216596447814e-05,
+      "loss": 1.1605,
+      "step": 5131000
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4498116630887254e-05,
+      "loss": 1.1544,
+      "step": 5131500
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4496020865257814e-05,
+      "loss": 1.1684,
+      "step": 5132000
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.449392089969724e-05,
+      "loss": 1.1565,
+      "step": 5132500
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4491820934136674e-05,
+      "loss": 1.1775,
+      "step": 5133000
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4489720968576115e-05,
+      "loss": 1.1334,
+      "step": 5133500
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.448762100301555e-05,
+      "loss": 1.1533,
+      "step": 5134000
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.448552103745498e-05,
+      "loss": 1.1779,
+      "step": 5134500
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.448342107189442e-05,
+      "loss": 1.1855,
+      "step": 5135000
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4481321106333855e-05,
+      "loss": 1.1363,
+      "step": 5135500
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.447922534070441e-05,
+      "loss": 1.1588,
+      "step": 5136000
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.447712957507497e-05,
+      "loss": 1.1472,
+      "step": 5136500
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.44750296095144e-05,
+      "loss": 1.1904,
+      "step": 5137000
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4472929643953836e-05,
+      "loss": 1.1813,
+      "step": 5137500
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.447082967839327e-05,
+      "loss": 1.1553,
+      "step": 5138000
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.446872971283271e-05,
+      "loss": 1.1512,
+      "step": 5138500
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.446662974727214e-05,
+      "loss": 1.186,
+      "step": 5139000
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.446452978171158e-05,
+      "loss": 1.1542,
+      "step": 5139500
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.446242981615102e-05,
+      "loss": 1.1547,
+      "step": 5140000
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.446033405052157e-05,
+      "loss": 1.164,
+      "step": 5140500
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4458234084961004e-05,
+      "loss": 1.1528,
+      "step": 5141000
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4456138319331564e-05,
+      "loss": 1.172,
+      "step": 5141500
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4454038353771e-05,
+      "loss": 1.1634,
+      "step": 5142000
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.445193838821043e-05,
+      "loss": 1.1785,
+      "step": 5142500
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4449838422649865e-05,
+      "loss": 1.1635,
+      "step": 5143000
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4447738457089305e-05,
+      "loss": 1.1579,
+      "step": 5143500
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.444563849152874e-05,
+      "loss": 1.1812,
+      "step": 5144000
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.444353852596817e-05,
+      "loss": 1.2038,
+      "step": 5144500
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.444143856040761e-05,
+      "loss": 1.1697,
+      "step": 5145000
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4439342794778166e-05,
+      "loss": 1.1301,
+      "step": 5145500
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.44372428292176e-05,
+      "loss": 1.1288,
+      "step": 5146000
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.443514286365703e-05,
+      "loss": 1.1832,
+      "step": 5146500
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.443304289809647e-05,
+      "loss": 1.1595,
+      "step": 5147000
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4430947132467026e-05,
+      "loss": 1.178,
+      "step": 5147500
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.442884716690646e-05,
+      "loss": 1.1671,
+      "step": 5148000
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.442674720134589e-05,
+      "loss": 1.1416,
+      "step": 5148500
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4424647235785334e-05,
+      "loss": 1.163,
+      "step": 5149000
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.442255147015589e-05,
+      "loss": 1.1784,
+      "step": 5149500
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.442045150459532e-05,
+      "loss": 1.1777,
+      "step": 5150000
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.441835153903476e-05,
+      "loss": 1.1762,
+      "step": 5150500
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.441625577340532e-05,
+      "loss": 1.1757,
+      "step": 5151000
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.441415580784475e-05,
+      "loss": 1.1626,
+      "step": 5151500
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.441205584228418e-05,
+      "loss": 1.1287,
+      "step": 5152000
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.440995587672362e-05,
+      "loss": 1.1468,
+      "step": 5152500
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4407855911163055e-05,
+      "loss": 1.1755,
+      "step": 5153000
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.440575594560249e-05,
+      "loss": 1.1621,
+      "step": 5153500
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.440365598004193e-05,
+      "loss": 1.1269,
+      "step": 5154000
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.440156021441248e-05,
+      "loss": 1.1528,
+      "step": 5154500
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4399460248851916e-05,
+      "loss": 1.1714,
+      "step": 5155000
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.439736028329135e-05,
+      "loss": 1.2,
+      "step": 5155500
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.439526031773079e-05,
+      "loss": 1.1606,
+      "step": 5156000
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.439316035217022e-05,
+      "loss": 1.1981,
+      "step": 5156500
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4391060386609656e-05,
+      "loss": 1.1692,
+      "step": 5157000
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.43889604210491e-05,
+      "loss": 1.1747,
+      "step": 5157500
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.438686045548853e-05,
+      "loss": 1.1719,
+      "step": 5158000
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4384768889790204e-05,
+      "loss": 1.1672,
+      "step": 5158500
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.438266892422964e-05,
+      "loss": 1.1738,
+      "step": 5159000
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.438056895866908e-05,
+      "loss": 1.1766,
+      "step": 5159500
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.437846899310851e-05,
+      "loss": 1.1598,
+      "step": 5160000
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.437637322747907e-05,
+      "loss": 1.1275,
+      "step": 5160500
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.43742732619185e-05,
+      "loss": 1.1568,
+      "step": 5161000
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.437217329635794e-05,
+      "loss": 1.1494,
+      "step": 5161500
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.437007333079737e-05,
+      "loss": 1.1419,
+      "step": 5162000
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4367973365236805e-05,
+      "loss": 1.1695,
+      "step": 5162500
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4365873399676245e-05,
+      "loss": 1.1782,
+      "step": 5163000
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.436377343411568e-05,
+      "loss": 1.1773,
+      "step": 5163500
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.436167346855512e-05,
+      "loss": 1.1414,
+      "step": 5164000
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.435957350299455e-05,
+      "loss": 1.1794,
+      "step": 5164500
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4357473537433986e-05,
+      "loss": 1.1836,
+      "step": 5165000
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.435537777180454e-05,
+      "loss": 1.1359,
+      "step": 5165500
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.435327780624398e-05,
+      "loss": 1.1512,
+      "step": 5166000
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.435117784068341e-05,
+      "loss": 1.1679,
+      "step": 5166500
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.434907787512285e-05,
+      "loss": 1.1509,
+      "step": 5167000
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.434697790956229e-05,
+      "loss": 1.1544,
+      "step": 5167500
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.434488214393284e-05,
+      "loss": 1.1652,
+      "step": 5168000
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4342782178372274e-05,
+      "loss": 1.1348,
+      "step": 5168500
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.434068221281171e-05,
+      "loss": 1.1794,
+      "step": 5169000
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.433858224725115e-05,
+      "loss": 1.1466,
+      "step": 5169500
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.433648228169058e-05,
+      "loss": 1.1726,
+      "step": 5170000
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4334386516061135e-05,
+      "loss": 1.165,
+      "step": 5170500
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4332286550500575e-05,
+      "loss": 1.1513,
+      "step": 5171000
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.433018658494001e-05,
+      "loss": 1.1536,
+      "step": 5171500
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.432808661937944e-05,
+      "loss": 1.1542,
+      "step": 5172000
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.432598665381888e-05,
+      "loss": 1.1494,
+      "step": 5172500
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.432389508812055e-05,
+      "loss": 1.1518,
+      "step": 5173000
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.432179512255999e-05,
+      "loss": 1.1486,
+      "step": 5173500
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.431969515699942e-05,
+      "loss": 1.1897,
+      "step": 5174000
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4317595191438856e-05,
+      "loss": 1.1523,
+      "step": 5174500
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4315495225878296e-05,
+      "loss": 1.1794,
+      "step": 5175000
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.431339526031773e-05,
+      "loss": 1.1795,
+      "step": 5175500
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.431129529475716e-05,
+      "loss": 1.151,
+      "step": 5176000
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4309195329196604e-05,
+      "loss": 1.1613,
+      "step": 5176500
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.430709536363604e-05,
+      "loss": 1.1612,
+      "step": 5177000
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.430499539807547e-05,
+      "loss": 1.1539,
+      "step": 5177500
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.430289543251491e-05,
+      "loss": 1.1579,
+      "step": 5178000
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.430079546695434e-05,
+      "loss": 1.1498,
+      "step": 5178500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.42986997013249e-05,
+      "loss": 1.136,
+      "step": 5179000
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.429659973576434e-05,
+      "loss": 1.169,
+      "step": 5179500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.429449977020377e-05,
+      "loss": 1.1862,
+      "step": 5180000
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4292399804643205e-05,
+      "loss": 1.1693,
+      "step": 5180500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.429030403901376e-05,
+      "loss": 1.1988,
+      "step": 5181000
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.42882040734532e-05,
+      "loss": 1.1741,
+      "step": 5181500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.428610830782375e-05,
+      "loss": 1.1554,
+      "step": 5182000
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4284008342263186e-05,
+      "loss": 1.1465,
+      "step": 5182500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.428190837670262e-05,
+      "loss": 1.168,
+      "step": 5183000
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.427980841114206e-05,
+      "loss": 1.1677,
+      "step": 5183500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.427770844558149e-05,
+      "loss": 1.1748,
+      "step": 5184000
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4275608480020926e-05,
+      "loss": 1.1878,
+      "step": 5184500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4273508514460367e-05,
+      "loss": 1.1437,
+      "step": 5185000
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.427141274883092e-05,
+      "loss": 1.1731,
+      "step": 5185500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4269312783270354e-05,
+      "loss": 1.1823,
+      "step": 5186000
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4267212817709794e-05,
+      "loss": 1.1693,
+      "step": 5186500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.426511285214923e-05,
+      "loss": 1.1903,
+      "step": 5187000
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.426301288658866e-05,
+      "loss": 1.1629,
+      "step": 5187500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4260912921028094e-05,
+      "loss": 1.1494,
+      "step": 5188000
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.425881295546753e-05,
+      "loss": 1.155,
+      "step": 5188500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.425671298990696e-05,
+      "loss": 1.1768,
+      "step": 5189000
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.425461722427752e-05,
+      "loss": 1.1818,
+      "step": 5189500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4252521458648075e-05,
+      "loss": 1.157,
+      "step": 5190000
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4250421493087515e-05,
+      "loss": 1.1508,
+      "step": 5190500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.424832572745807e-05,
+      "loss": 1.1913,
+      "step": 5191000
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.42462257618975e-05,
+      "loss": 1.1816,
+      "step": 5191500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.424412579633694e-05,
+      "loss": 1.1617,
+      "step": 5192000
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4242025830776376e-05,
+      "loss": 1.154,
+      "step": 5192500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.423993006514693e-05,
+      "loss": 1.1697,
+      "step": 5193000
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.423783009958636e-05,
+      "loss": 1.1489,
+      "step": 5193500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.42357301340258e-05,
+      "loss": 1.1716,
+      "step": 5194000
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.423363436839636e-05,
+      "loss": 1.1734,
+      "step": 5194500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.423153440283579e-05,
+      "loss": 1.1538,
+      "step": 5195000
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4229434437275224e-05,
+      "loss": 1.1546,
+      "step": 5195500
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4227334471714664e-05,
+      "loss": 1.1936,
+      "step": 5196000
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.42252345061541e-05,
+      "loss": 1.1827,
+      "step": 5196500
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.422313454059353e-05,
+      "loss": 1.1716,
+      "step": 5197000
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.422103457503297e-05,
+      "loss": 1.1482,
+      "step": 5197500
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4218934609472405e-05,
+      "loss": 1.1445,
+      "step": 5198000
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.421683464391184e-05,
+      "loss": 1.1744,
+      "step": 5198500
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.421473467835128e-05,
+      "loss": 1.1511,
+      "step": 5199000
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.421263471279071e-05,
+      "loss": 1.1544,
+      "step": 5199500
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4210534747230145e-05,
+      "loss": 1.1808,
+      "step": 5200000
+    },
+    {
+      "epoch": 3.12,
+      "eval_loss": 1.1460847854614258,
+      "eval_runtime": 1102.6522,
+      "eval_samples_per_second": 477.685,
+      "eval_steps_per_second": 79.614,
+      "step": 5200000
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.420843478166958e-05,
+      "loss": 1.1751,
+      "step": 5200500
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.420633481610901e-05,
+      "loss": 1.1673,
+      "step": 5201000
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.420423485054845e-05,
+      "loss": 1.1478,
+      "step": 5201500
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.420213908491901e-05,
+      "loss": 1.1684,
+      "step": 5202000
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.420003911935844e-05,
+      "loss": 1.1746,
+      "step": 5202500
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.419793915379787e-05,
+      "loss": 1.1453,
+      "step": 5203000
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.419583918823731e-05,
+      "loss": 1.1699,
+      "step": 5203500
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4193739222676747e-05,
+      "loss": 1.1739,
+      "step": 5204000
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.41916434570473e-05,
+      "loss": 1.1912,
+      "step": 5204500
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4189543491486734e-05,
+      "loss": 1.1381,
+      "step": 5205000
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4187443525926174e-05,
+      "loss": 1.1931,
+      "step": 5205500
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.418534356036561e-05,
+      "loss": 1.1767,
+      "step": 5206000
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.418324359480504e-05,
+      "loss": 1.169,
+      "step": 5206500
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.418114362924448e-05,
+      "loss": 1.1641,
+      "step": 5207000
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4179043663683914e-05,
+      "loss": 1.1464,
+      "step": 5207500
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.417694369812335e-05,
+      "loss": 1.1557,
+      "step": 5208000
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.417484373256279e-05,
+      "loss": 1.1831,
+      "step": 5208500
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.417274796693334e-05,
+      "loss": 1.1928,
+      "step": 5209000
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4170652201303895e-05,
+      "loss": 1.1676,
+      "step": 5209500
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.416855223574333e-05,
+      "loss": 1.1631,
+      "step": 5210000
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.416645227018277e-05,
+      "loss": 1.1481,
+      "step": 5210500
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.41643523046222e-05,
+      "loss": 1.1631,
+      "step": 5211000
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4162252339061636e-05,
+      "loss": 1.1601,
+      "step": 5211500
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4160152373501076e-05,
+      "loss": 1.1651,
+      "step": 5212000
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.415805240794051e-05,
+      "loss": 1.1779,
+      "step": 5212500
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.415595664231106e-05,
+      "loss": 1.1369,
+      "step": 5213000
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.41538566767505e-05,
+      "loss": 1.1697,
+      "step": 5213500
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.415175671118994e-05,
+      "loss": 1.1403,
+      "step": 5214000
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.414966094556049e-05,
+      "loss": 1.1545,
+      "step": 5214500
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4147560979999924e-05,
+      "loss": 1.1466,
+      "step": 5215000
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4145461014439364e-05,
+      "loss": 1.1723,
+      "step": 5215500
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.41433610488788e-05,
+      "loss": 1.1806,
+      "step": 5216000
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.414126108331823e-05,
+      "loss": 1.1873,
+      "step": 5216500
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.413916111775767e-05,
+      "loss": 1.1938,
+      "step": 5217000
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4137065352128225e-05,
+      "loss": 1.2079,
+      "step": 5217500
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.413496538656766e-05,
+      "loss": 1.1327,
+      "step": 5218000
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.413286542100709e-05,
+      "loss": 1.2024,
+      "step": 5218500
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.413076545544653e-05,
+      "loss": 1.1843,
+      "step": 5219000
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4128665489885965e-05,
+      "loss": 1.146,
+      "step": 5219500
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.41265655243254e-05,
+      "loss": 1.1853,
+      "step": 5220000
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.412446555876484e-05,
+      "loss": 1.1709,
+      "step": 5220500
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.412236559320427e-05,
+      "loss": 1.1425,
+      "step": 5221000
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4120265627643706e-05,
+      "loss": 1.1484,
+      "step": 5221500
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.411816986201426e-05,
+      "loss": 1.1402,
+      "step": 5222000
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.41160698964537e-05,
+      "loss": 1.1471,
+      "step": 5222500
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.411396993089313e-05,
+      "loss": 1.1571,
+      "step": 5223000
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4111869965332574e-05,
+      "loss": 1.1678,
+      "step": 5223500
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.410976999977201e-05,
+      "loss": 1.16,
+      "step": 5224000
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4107670034211434e-05,
+      "loss": 1.1952,
+      "step": 5224500
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4105570068650874e-05,
+      "loss": 1.1712,
+      "step": 5225000
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.410347010309031e-05,
+      "loss": 1.1833,
+      "step": 5225500
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.410137433746087e-05,
+      "loss": 1.1686,
+      "step": 5226000
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.40992743719003e-05,
+      "loss": 1.1937,
+      "step": 5226500
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4097178606270855e-05,
+      "loss": 1.183,
+      "step": 5227000
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4095078640710295e-05,
+      "loss": 1.181,
+      "step": 5227500
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.409297867514973e-05,
+      "loss": 1.1785,
+      "step": 5228000
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.409087870958916e-05,
+      "loss": 1.148,
+      "step": 5228500
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.40887787440286e-05,
+      "loss": 1.2019,
+      "step": 5229000
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.408667877846803e-05,
+      "loss": 1.1529,
+      "step": 5229500
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.408457881290747e-05,
+      "loss": 1.1653,
+      "step": 5230000
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.40824788473469e-05,
+      "loss": 1.1553,
+      "step": 5230500
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.408038308171746e-05,
+      "loss": 1.1844,
+      "step": 5231000
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.407828311615689e-05,
+      "loss": 1.1875,
+      "step": 5231500
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.407618735052745e-05,
+      "loss": 1.1693,
+      "step": 5232000
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.407408738496689e-05,
+      "loss": 1.1507,
+      "step": 5232500
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4071987419406324e-05,
+      "loss": 1.1529,
+      "step": 5233000
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.406988745384576e-05,
+      "loss": 1.1726,
+      "step": 5233500
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.406778748828519e-05,
+      "loss": 1.1601,
+      "step": 5234000
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4065687522724624e-05,
+      "loss": 1.1722,
+      "step": 5234500
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.406358755716406e-05,
+      "loss": 1.14,
+      "step": 5235000
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.40614875916035e-05,
+      "loss": 1.1924,
+      "step": 5235500
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.405939182597406e-05,
+      "loss": 1.1706,
+      "step": 5236000
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4057291860413485e-05,
+      "loss": 1.1802,
+      "step": 5236500
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4055191894852925e-05,
+      "loss": 1.1847,
+      "step": 5237000
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.405309192929236e-05,
+      "loss": 1.1825,
+      "step": 5237500
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.405099616366292e-05,
+      "loss": 1.1495,
+      "step": 5238000
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.404890039803347e-05,
+      "loss": 1.1652,
+      "step": 5238500
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4046800432472906e-05,
+      "loss": 1.1546,
+      "step": 5239000
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4044700466912346e-05,
+      "loss": 1.1809,
+      "step": 5239500
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.404260050135178e-05,
+      "loss": 1.1423,
+      "step": 5240000
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.404050053579121e-05,
+      "loss": 1.1882,
+      "step": 5240500
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4038404770161767e-05,
+      "loss": 1.1722,
+      "step": 5241000
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.403630480460121e-05,
+      "loss": 1.1693,
+      "step": 5241500
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.403420483904064e-05,
+      "loss": 1.1449,
+      "step": 5242000
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4032104873480074e-05,
+      "loss": 1.2002,
+      "step": 5242500
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4030009107850634e-05,
+      "loss": 1.1715,
+      "step": 5243000
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.402790914229007e-05,
+      "loss": 1.1691,
+      "step": 5243500
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.40258091767295e-05,
+      "loss": 1.1815,
+      "step": 5244000
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.402370921116894e-05,
+      "loss": 1.1443,
+      "step": 5244500
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4021609245608375e-05,
+      "loss": 1.1581,
+      "step": 5245000
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.401950928004781e-05,
+      "loss": 1.1545,
+      "step": 5245500
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.401740931448724e-05,
+      "loss": 1.1413,
+      "step": 5246000
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.40153135488578e-05,
+      "loss": 1.1474,
+      "step": 5246500
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4013213583297235e-05,
+      "loss": 1.1267,
+      "step": 5247000
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.401111361773667e-05,
+      "loss": 1.1732,
+      "step": 5247500
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.400901365217611e-05,
+      "loss": 1.1621,
+      "step": 5248000
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4006913686615536e-05,
+      "loss": 1.1766,
+      "step": 5248500
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.400481372105497e-05,
+      "loss": 1.1352,
+      "step": 5249000
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.400271375549441e-05,
+      "loss": 1.1602,
+      "step": 5249500
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.400061378993384e-05,
+      "loss": 1.1443,
+      "step": 5250000
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.39985180243044e-05,
+      "loss": 1.198,
+      "step": 5250500
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.399641805874384e-05,
+      "loss": 1.1858,
+      "step": 5251000
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.399431809318327e-05,
+      "loss": 1.1506,
+      "step": 5251500
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.399222232755383e-05,
+      "loss": 1.1801,
+      "step": 5252000
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.3990122361993264e-05,
+      "loss": 1.1862,
+      "step": 5252500
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.39880223964327e-05,
+      "loss": 1.1771,
+      "step": 5253000
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.398592243087213e-05,
+      "loss": 1.181,
+      "step": 5253500
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.3983822465311564e-05,
+      "loss": 1.1685,
+      "step": 5254000
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.3981726699682125e-05,
+      "loss": 1.1605,
+      "step": 5254500
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.3979626734121565e-05,
+      "loss": 1.166,
+      "step": 5255000
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.397752676856099e-05,
+      "loss": 1.1554,
+      "step": 5255500
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.3975426803000425e-05,
+      "loss": 1.1752,
+      "step": 5256000
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.3973331037370985e-05,
+      "loss": 1.171,
+      "step": 5256500
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.3971231071810426e-05,
+      "loss": 1.1568,
+      "step": 5257000
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.396913110624986e-05,
+      "loss": 1.1652,
+      "step": 5257500
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.396703114068929e-05,
+      "loss": 1.1686,
+      "step": 5258000
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.396493537505985e-05,
+      "loss": 1.1715,
+      "step": 5258500
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.3962835409499286e-05,
+      "loss": 1.1795,
+      "step": 5259000
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.396073544393872e-05,
+      "loss": 1.178,
+      "step": 5259500
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.395863547837816e-05,
+      "loss": 1.148,
+      "step": 5260000
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.395653551281759e-05,
+      "loss": 1.1626,
+      "step": 5260500
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.395443554725702e-05,
+      "loss": 1.176,
+      "step": 5261000
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.395233558169646e-05,
+      "loss": 1.172,
+      "step": 5261500
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.3950235616135894e-05,
+      "loss": 1.1641,
+      "step": 5262000
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.394813985050645e-05,
+      "loss": 1.1709,
+      "step": 5262500
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.394603988494588e-05,
+      "loss": 1.1976,
+      "step": 5263000
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.394393991938532e-05,
+      "loss": 1.179,
+      "step": 5263500
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.3941839953824755e-05,
+      "loss": 1.2022,
+      "step": 5264000
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.393973998826419e-05,
+      "loss": 1.1834,
+      "step": 5264500
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.393764002270363e-05,
+      "loss": 1.1586,
+      "step": 5265000
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.393554425707418e-05,
+      "loss": 1.1665,
+      "step": 5265500
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.3933444291513615e-05,
+      "loss": 1.1882,
+      "step": 5266000
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.3931344325953056e-05,
+      "loss": 1.155,
+      "step": 5266500
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.392924436039249e-05,
+      "loss": 1.1745,
+      "step": 5267000
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.392714439483192e-05,
+      "loss": 1.1598,
+      "step": 5267500
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.392504442927136e-05,
+      "loss": 1.1596,
+      "step": 5268000
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.3922944463710796e-05,
+      "loss": 1.174,
+      "step": 5268500
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.392084449815023e-05,
+      "loss": 1.1552,
+      "step": 5269000
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.391874873252078e-05,
+      "loss": 1.1933,
+      "step": 5269500
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.3916648766960224e-05,
+      "loss": 1.1477,
+      "step": 5270000
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.39145572012619e-05,
+      "loss": 1.1744,
+      "step": 5270500
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.391245723570134e-05,
+      "loss": 1.1997,
+      "step": 5271000
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.391035727014077e-05,
+      "loss": 1.154,
+      "step": 5271500
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.3908257304580204e-05,
+      "loss": 1.1838,
+      "step": 5272000
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.390615733901964e-05,
+      "loss": 1.1571,
+      "step": 5272500
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.390405737345907e-05,
+      "loss": 1.1455,
+      "step": 5273000
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.390195740789851e-05,
+      "loss": 1.1389,
+      "step": 5273500
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.3899857442337945e-05,
+      "loss": 1.1551,
+      "step": 5274000
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.389775747677738e-05,
+      "loss": 1.1707,
+      "step": 5274500
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.389565751121682e-05,
+      "loss": 1.1959,
+      "step": 5275000
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.389355754565625e-05,
+      "loss": 1.19,
+      "step": 5275500
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.3891457580095686e-05,
+      "loss": 1.1814,
+      "step": 5276000
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.388936181446624e-05,
+      "loss": 1.1413,
+      "step": 5276500
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.388726604883679e-05,
+      "loss": 1.171,
+      "step": 5277000
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.388516608327623e-05,
+      "loss": 1.1502,
+      "step": 5277500
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.3883066117715666e-05,
+      "loss": 1.1542,
+      "step": 5278000
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.38809661521551e-05,
+      "loss": 1.1897,
+      "step": 5278500
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.387886618659454e-05,
+      "loss": 1.1769,
+      "step": 5279000
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.3876770420965094e-05,
+      "loss": 1.1686,
+      "step": 5279500
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.387467045540453e-05,
+      "loss": 1.1454,
+      "step": 5280000
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.387257048984397e-05,
+      "loss": 1.1846,
+      "step": 5280500
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.38704705242834e-05,
+      "loss": 1.1737,
+      "step": 5281000
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.3868370558722834e-05,
+      "loss": 1.1612,
+      "step": 5281500
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.3866270593162275e-05,
+      "loss": 1.1786,
+      "step": 5282000
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.386417062760171e-05,
+      "loss": 1.1634,
+      "step": 5282500
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.386207066204114e-05,
+      "loss": 1.1766,
+      "step": 5283000
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.385997069648058e-05,
+      "loss": 1.1729,
+      "step": 5283500
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.3857870730920015e-05,
+      "loss": 1.1702,
+      "step": 5284000
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.385577496529057e-05,
+      "loss": 1.158,
+      "step": 5284500
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.385367499973e-05,
+      "loss": 1.1882,
+      "step": 5285000
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.385157503416944e-05,
+      "loss": 1.1644,
+      "step": 5285500
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.3849475068608876e-05,
+      "loss": 1.1724,
+      "step": 5286000
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.384737510304831e-05,
+      "loss": 1.1693,
+      "step": 5286500
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.384527513748775e-05,
+      "loss": 1.1537,
+      "step": 5287000
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.38431793718583e-05,
+      "loss": 1.1891,
+      "step": 5287500
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.384107940629774e-05,
+      "loss": 1.1749,
+      "step": 5288000
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.383897944073718e-05,
+      "loss": 1.1731,
+      "step": 5288500
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.383687947517661e-05,
+      "loss": 1.152,
+      "step": 5289000
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.383477950961604e-05,
+      "loss": 1.1807,
+      "step": 5289500
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.383267954405548e-05,
+      "loss": 1.1774,
+      "step": 5290000
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.383058377842604e-05,
+      "loss": 1.1496,
+      "step": 5290500
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.382848381286547e-05,
+      "loss": 1.1536,
+      "step": 5291000
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.3826383847304905e-05,
+      "loss": 1.2075,
+      "step": 5291500
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.382428388174434e-05,
+      "loss": 1.1642,
+      "step": 5292000
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.382218391618377e-05,
+      "loss": 1.1535,
+      "step": 5292500
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.382008815055433e-05,
+      "loss": 1.1653,
+      "step": 5293000
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.3817988184993765e-05,
+      "loss": 1.1669,
+      "step": 5293500
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.3815888219433205e-05,
+      "loss": 1.1386,
+      "step": 5294000
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.381378825387263e-05,
+      "loss": 1.1548,
+      "step": 5294500
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.3811688288312066e-05,
+      "loss": 1.1754,
+      "step": 5295000
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.3809588322751506e-05,
+      "loss": 1.1608,
+      "step": 5295500
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.380748835719094e-05,
+      "loss": 1.1903,
+      "step": 5296000
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.380538839163038e-05,
+      "loss": 1.1513,
+      "step": 5296500
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.380329682593205e-05,
+      "loss": 1.1523,
+      "step": 5297000
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.3801196860371493e-05,
+      "loss": 1.1539,
+      "step": 5297500
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.379909689481093e-05,
+      "loss": 1.1538,
+      "step": 5298000
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.379699692925036e-05,
+      "loss": 1.15,
+      "step": 5298500
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.3794896963689794e-05,
+      "loss": 1.1573,
+      "step": 5299000
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.3792801198060354e-05,
+      "loss": 1.1431,
+      "step": 5299500
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.379070123249979e-05,
+      "loss": 1.1464,
+      "step": 5300000
+    },
+    {
+      "epoch": 3.18,
+      "eval_loss": 1.144875407218933,
+      "eval_runtime": 1110.2882,
+      "eval_samples_per_second": 474.399,
+      "eval_steps_per_second": 79.067,
+      "step": 5300000
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.378860546687034e-05,
+      "loss": 1.1742,
+      "step": 5300500
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.3786505501309775e-05,
+      "loss": 1.2002,
+      "step": 5301000
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.3784405535749215e-05,
+      "loss": 1.1369,
+      "step": 5301500
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.378230557018865e-05,
+      "loss": 1.1692,
+      "step": 5302000
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.378020560462809e-05,
+      "loss": 1.1997,
+      "step": 5302500
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.377810563906752e-05,
+      "loss": 1.1565,
+      "step": 5303000
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.3776005673506956e-05,
+      "loss": 1.1792,
+      "step": 5303500
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.377390570794639e-05,
+      "loss": 1.1762,
+      "step": 5304000
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.377180574238582e-05,
+      "loss": 1.179,
+      "step": 5304500
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.376970997675638e-05,
+      "loss": 1.1768,
+      "step": 5305000
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.3767610011195816e-05,
+      "loss": 1.1532,
+      "step": 5305500
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.3765510045635257e-05,
+      "loss": 1.1553,
+      "step": 5306000
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.376341008007468e-05,
+      "loss": 1.1361,
+      "step": 5306500
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.376131011451412e-05,
+      "loss": 1.1716,
+      "step": 5307000
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.375921434888468e-05,
+      "loss": 1.147,
+      "step": 5307500
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.375711438332412e-05,
+      "loss": 1.174,
+      "step": 5308000
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.3755014417763544e-05,
+      "loss": 1.1645,
+      "step": 5308500
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.3752914452202984e-05,
+      "loss": 1.1649,
+      "step": 5309000
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.375081448664242e-05,
+      "loss": 1.1613,
+      "step": 5309500
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.374871452108185e-05,
+      "loss": 1.1791,
+      "step": 5310000
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.374661455552129e-05,
+      "loss": 1.1696,
+      "step": 5310500
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.3744518789891845e-05,
+      "loss": 1.167,
+      "step": 5311000
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.374241882433128e-05,
+      "loss": 1.1543,
+      "step": 5311500
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.374031885877071e-05,
+      "loss": 1.1633,
+      "step": 5312000
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.373821889321015e-05,
+      "loss": 1.1657,
+      "step": 5312500
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.3736118927649586e-05,
+      "loss": 1.1861,
+      "step": 5313000
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.373402316202014e-05,
+      "loss": 1.2088,
+      "step": 5313500
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.373192319645957e-05,
+      "loss": 1.1784,
+      "step": 5314000
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.372982323089901e-05,
+      "loss": 1.1717,
+      "step": 5314500
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.3727723265338446e-05,
+      "loss": 1.1704,
+      "step": 5315000
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.372562329977788e-05,
+      "loss": 1.1454,
+      "step": 5315500
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.372352333421732e-05,
+      "loss": 1.1564,
+      "step": 5316000
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.3721423368656753e-05,
+      "loss": 1.1685,
+      "step": 5316500
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.371932760302731e-05,
+      "loss": 1.1875,
+      "step": 5317000
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.371722763746675e-05,
+      "loss": 1.1784,
+      "step": 5317500
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.37151318718373e-05,
+      "loss": 1.1826,
+      "step": 5318000
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.3713031906276734e-05,
+      "loss": 1.1569,
+      "step": 5318500
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.371093194071617e-05,
+      "loss": 1.1666,
+      "step": 5319000
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.370883197515561e-05,
+      "loss": 1.1509,
+      "step": 5319500
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.370673200959504e-05,
+      "loss": 1.1687,
+      "step": 5320000
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.3704632044034475e-05,
+      "loss": 1.1747,
+      "step": 5320500
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.3702532078473915e-05,
+      "loss": 1.1714,
+      "step": 5321000
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.370043211291335e-05,
+      "loss": 1.1413,
+      "step": 5321500
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.369833214735278e-05,
+      "loss": 1.1586,
+      "step": 5322000
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.3696236381723336e-05,
+      "loss": 1.168,
+      "step": 5322500
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.3694136416162776e-05,
+      "loss": 1.1707,
+      "step": 5323000
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.369203645060221e-05,
+      "loss": 1.2057,
+      "step": 5323500
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.368993648504164e-05,
+      "loss": 1.1536,
+      "step": 5324000
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.368783651948108e-05,
+      "loss": 1.1697,
+      "step": 5324500
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.3685736553920516e-05,
+      "loss": 1.1584,
+      "step": 5325000
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.368363658835995e-05,
+      "loss": 1.1843,
+      "step": 5325500
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.368153662279938e-05,
+      "loss": 1.1776,
+      "step": 5326000
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.3679440857169944e-05,
+      "loss": 1.1261,
+      "step": 5326500
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.367734089160938e-05,
+      "loss": 1.198,
+      "step": 5327000
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.367524512597993e-05,
+      "loss": 1.1932,
+      "step": 5327500
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.367314516041937e-05,
+      "loss": 1.167,
+      "step": 5328000
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.3671049394789925e-05,
+      "loss": 1.1424,
+      "step": 5328500
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.366894942922936e-05,
+      "loss": 1.1492,
+      "step": 5329000
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.366684946366879e-05,
+      "loss": 1.1446,
+      "step": 5329500
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.366474949810823e-05,
+      "loss": 1.1653,
+      "step": 5330000
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.3662649532547665e-05,
+      "loss": 1.1483,
+      "step": 5330500
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.36605495669871e-05,
+      "loss": 1.163,
+      "step": 5331000
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.365844960142654e-05,
+      "loss": 1.1611,
+      "step": 5331500
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.365634963586597e-05,
+      "loss": 1.1659,
+      "step": 5332000
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.3654249670305406e-05,
+      "loss": 1.1664,
+      "step": 5332500
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.3652149704744846e-05,
+      "loss": 1.2027,
+      "step": 5333000
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.365004973918427e-05,
+      "loss": 1.1881,
+      "step": 5333500
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.364794977362371e-05,
+      "loss": 1.1847,
+      "step": 5334000
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.364585400799427e-05,
+      "loss": 1.1891,
+      "step": 5334500
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.364375824236483e-05,
+      "loss": 1.1615,
+      "step": 5335000
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.364165827680426e-05,
+      "loss": 1.1784,
+      "step": 5335500
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.3639558311243694e-05,
+      "loss": 1.1849,
+      "step": 5336000
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.3637458345683134e-05,
+      "loss": 1.1648,
+      "step": 5336500
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.363535838012257e-05,
+      "loss": 1.1437,
+      "step": 5337000
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.3633258414562e-05,
+      "loss": 1.1634,
+      "step": 5337500
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.3631162648932554e-05,
+      "loss": 1.1636,
+      "step": 5338000
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.3629062683371995e-05,
+      "loss": 1.1512,
+      "step": 5338500
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.362696271781143e-05,
+      "loss": 1.1552,
+      "step": 5339000
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.362486275225086e-05,
+      "loss": 1.1736,
+      "step": 5339500
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.362276698662142e-05,
+      "loss": 1.1628,
+      "step": 5340000
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.3620667021060855e-05,
+      "loss": 1.2001,
+      "step": 5340500
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.361856705550029e-05,
+      "loss": 1.1654,
+      "step": 5341000
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.361646708993973e-05,
+      "loss": 1.1589,
+      "step": 5341500
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.361436712437916e-05,
+      "loss": 1.1443,
+      "step": 5342000
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.3612267158818596e-05,
+      "loss": 1.1866,
+      "step": 5342500
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.361016719325803e-05,
+      "loss": 1.1784,
+      "step": 5343000
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.360806722769746e-05,
+      "loss": 1.1663,
+      "step": 5343500
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.360597146206802e-05,
+      "loss": 1.1434,
+      "step": 5344000
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.360387149650746e-05,
+      "loss": 1.1535,
+      "step": 5344500
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.360177153094689e-05,
+      "loss": 1.1903,
+      "step": 5345000
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.359967576531745e-05,
+      "loss": 1.1557,
+      "step": 5345500
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3597575799756884e-05,
+      "loss": 1.1887,
+      "step": 5346000
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.359547583419632e-05,
+      "loss": 1.1675,
+      "step": 5346500
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.359337586863576e-05,
+      "loss": 1.1825,
+      "step": 5347000
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3591275903075184e-05,
+      "loss": 1.181,
+      "step": 5347500
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3589175937514625e-05,
+      "loss": 1.1513,
+      "step": 5348000
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.358707597195406e-05,
+      "loss": 1.1662,
+      "step": 5348500
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.358497600639349e-05,
+      "loss": 1.1656,
+      "step": 5349000
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.358288024076405e-05,
+      "loss": 1.1638,
+      "step": 5349500
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3580780275203485e-05,
+      "loss": 1.1872,
+      "step": 5350000
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.357868030964292e-05,
+      "loss": 1.1764,
+      "step": 5350500
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.357658034408235e-05,
+      "loss": 1.1776,
+      "step": 5351000
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.357448037852179e-05,
+      "loss": 1.146,
+      "step": 5351500
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.357238461289235e-05,
+      "loss": 1.1664,
+      "step": 5352000
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.357028464733178e-05,
+      "loss": 1.183,
+      "step": 5352500
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.356818468177121e-05,
+      "loss": 1.1656,
+      "step": 5353000
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.356608471621065e-05,
+      "loss": 1.1752,
+      "step": 5353500
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.356398475065009e-05,
+      "loss": 1.1802,
+      "step": 5354000
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.356188478508953e-05,
+      "loss": 1.158,
+      "step": 5354500
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.355978481952896e-05,
+      "loss": 1.1878,
+      "step": 5355000
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3557684853968394e-05,
+      "loss": 1.1831,
+      "step": 5355500
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.355558908833895e-05,
+      "loss": 1.1679,
+      "step": 5356000
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.355348912277839e-05,
+      "loss": 1.1638,
+      "step": 5356500
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.355138915721782e-05,
+      "loss": 1.1638,
+      "step": 5357000
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3549289191657255e-05,
+      "loss": 1.1934,
+      "step": 5357500
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3547189226096695e-05,
+      "loss": 1.1778,
+      "step": 5358000
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.354508926053613e-05,
+      "loss": 1.1576,
+      "step": 5358500
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.354298929497556e-05,
+      "loss": 1.1737,
+      "step": 5359000
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3540889329415e-05,
+      "loss": 1.1674,
+      "step": 5359500
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3538793563785556e-05,
+      "loss": 1.17,
+      "step": 5360000
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.353669359822499e-05,
+      "loss": 1.1683,
+      "step": 5360500
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.353459783259554e-05,
+      "loss": 1.1624,
+      "step": 5361000
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.353249786703498e-05,
+      "loss": 1.1736,
+      "step": 5361500
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3530397901474416e-05,
+      "loss": 1.1524,
+      "step": 5362000
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.352829793591385e-05,
+      "loss": 1.1611,
+      "step": 5362500
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.352619797035329e-05,
+      "loss": 1.1708,
+      "step": 5363000
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3524098004792723e-05,
+      "loss": 1.1678,
+      "step": 5363500
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.352199803923216e-05,
+      "loss": 1.1549,
+      "step": 5364000
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.35198980736716e-05,
+      "loss": 1.2398,
+      "step": 5364500
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.351780230804215e-05,
+      "loss": 1.1568,
+      "step": 5365000
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3515702342481584e-05,
+      "loss": 1.1711,
+      "step": 5365500
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.351360237692102e-05,
+      "loss": 1.1801,
+      "step": 5366000
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.351150241136046e-05,
+      "loss": 1.1759,
+      "step": 5366500
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.350940664573101e-05,
+      "loss": 1.157,
+      "step": 5367000
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3507306680170445e-05,
+      "loss": 1.1495,
+      "step": 5367500
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.350520671460988e-05,
+      "loss": 1.1261,
+      "step": 5368000
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.350310674904932e-05,
+      "loss": 1.1927,
+      "step": 5368500
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.350101098341987e-05,
+      "loss": 1.1402,
+      "step": 5369000
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3498911017859306e-05,
+      "loss": 1.159,
+      "step": 5369500
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3496811052298746e-05,
+      "loss": 1.1661,
+      "step": 5370000
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.349471108673818e-05,
+      "loss": 1.178,
+      "step": 5370500
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.349261112117761e-05,
+      "loss": 1.1467,
+      "step": 5371000
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.349051115561705e-05,
+      "loss": 1.1741,
+      "step": 5371500
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.348841119005648e-05,
+      "loss": 1.1815,
+      "step": 5372000
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.348631122449591e-05,
+      "loss": 1.1874,
+      "step": 5372500
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3484215458866474e-05,
+      "loss": 1.1452,
+      "step": 5373000
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3482115493305914e-05,
+      "loss": 1.1737,
+      "step": 5373500
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.348001552774535e-05,
+      "loss": 1.1563,
+      "step": 5374000
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3477915562184774e-05,
+      "loss": 1.169,
+      "step": 5374500
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3475815596624214e-05,
+      "loss": 1.1668,
+      "step": 5375000
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.347371563106365e-05,
+      "loss": 1.1901,
+      "step": 5375500
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.347161566550308e-05,
+      "loss": 1.1438,
+      "step": 5376000
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.346951569994252e-05,
+      "loss": 1.1613,
+      "step": 5376500
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3467419934313075e-05,
+      "loss": 1.1709,
+      "step": 5377000
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3465324168683635e-05,
+      "loss": 1.178,
+      "step": 5377500
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.346322420312307e-05,
+      "loss": 1.167,
+      "step": 5378000
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.346112423756251e-05,
+      "loss": 1.1412,
+      "step": 5378500
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.345902427200194e-05,
+      "loss": 1.1629,
+      "step": 5379000
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.345692430644137e-05,
+      "loss": 1.1608,
+      "step": 5379500
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.345482854081193e-05,
+      "loss": 1.1547,
+      "step": 5380000
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.345272857525137e-05,
+      "loss": 1.1811,
+      "step": 5380500
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.34506286096908e-05,
+      "loss": 1.1707,
+      "step": 5381000
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.344853284406136e-05,
+      "loss": 1.158,
+      "step": 5381500
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.344643287850079e-05,
+      "loss": 1.1686,
+      "step": 5382000
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.344433291294023e-05,
+      "loss": 1.1531,
+      "step": 5382500
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3442232947379664e-05,
+      "loss": 1.1236,
+      "step": 5383000
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.34401329818191e-05,
+      "loss": 1.1588,
+      "step": 5383500
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.343803301625853e-05,
+      "loss": 1.1649,
+      "step": 5384000
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3435933050697964e-05,
+      "loss": 1.1502,
+      "step": 5384500
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3433833085137404e-05,
+      "loss": 1.1575,
+      "step": 5385000
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3431737319507965e-05,
+      "loss": 1.1721,
+      "step": 5385500
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.342964155387852e-05,
+      "loss": 1.1654,
+      "step": 5386000
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.342754158831795e-05,
+      "loss": 1.1513,
+      "step": 5386500
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3425441622757385e-05,
+      "loss": 1.1709,
+      "step": 5387000
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3423341657196826e-05,
+      "loss": 1.1836,
+      "step": 5387500
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.342124169163626e-05,
+      "loss": 1.1345,
+      "step": 5388000
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.341914172607569e-05,
+      "loss": 1.197,
+      "step": 5388500
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3417041760515126e-05,
+      "loss": 1.165,
+      "step": 5389000
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.341494179495456e-05,
+      "loss": 1.14,
+      "step": 5389500
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.341284182939399e-05,
+      "loss": 1.1837,
+      "step": 5390000
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.341074186383343e-05,
+      "loss": 1.1634,
+      "step": 5390500
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3408641898272867e-05,
+      "loss": 1.183,
+      "step": 5391000
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.34065419327123e-05,
+      "loss": 1.1713,
+      "step": 5391500
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.340444616708286e-05,
+      "loss": 1.1559,
+      "step": 5392000
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.340235040145342e-05,
+      "loss": 1.1913,
+      "step": 5392500
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3400250435892854e-05,
+      "loss": 1.1692,
+      "step": 5393000
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.339815047033228e-05,
+      "loss": 1.1881,
+      "step": 5393500
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.339605050477172e-05,
+      "loss": 1.1612,
+      "step": 5394000
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3393950539211155e-05,
+      "loss": 1.1806,
+      "step": 5394500
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3391854773581715e-05,
+      "loss": 1.1496,
+      "step": 5395000
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.338975480802115e-05,
+      "loss": 1.1743,
+      "step": 5395500
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.338765484246058e-05,
+      "loss": 1.1724,
+      "step": 5396000
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3385554876900015e-05,
+      "loss": 1.1768,
+      "step": 5396500
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3383459111270576e-05,
+      "loss": 1.1587,
+      "step": 5397000
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.338135914571001e-05,
+      "loss": 1.1746,
+      "step": 5397500
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.337925918014945e-05,
+      "loss": 1.1671,
+      "step": 5398000
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3377159214588876e-05,
+      "loss": 1.187,
+      "step": 5398500
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3375059249028316e-05,
+      "loss": 1.1635,
+      "step": 5399000
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.337295928346775e-05,
+      "loss": 1.2195,
+      "step": 5399500
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.337085931790718e-05,
+      "loss": 1.1819,
+      "step": 5400000
+    },
+    {
+      "epoch": 3.24,
+      "eval_loss": 1.1395820379257202,
+      "eval_runtime": 1108.0224,
+      "eval_samples_per_second": 475.369,
+      "eval_steps_per_second": 79.229,
+      "step": 5400000
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.336875935234662e-05,
+      "loss": 1.1601,
+      "step": 5400500
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.336665938678606e-05,
+      "loss": 1.1496,
+      "step": 5401000
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.336456362115661e-05,
+      "loss": 1.1535,
+      "step": 5401500
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3362463655596044e-05,
+      "loss": 1.205,
+      "step": 5402000
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3360363690035484e-05,
+      "loss": 1.1644,
+      "step": 5402500
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.335826372447492e-05,
+      "loss": 1.1973,
+      "step": 5403000
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.335616375891435e-05,
+      "loss": 1.1611,
+      "step": 5403500
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.335406379335379e-05,
+      "loss": 1.1469,
+      "step": 5404000
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3351968027724345e-05,
+      "loss": 1.1849,
+      "step": 5404500
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.334986806216378e-05,
+      "loss": 1.1697,
+      "step": 5405000
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.334776809660321e-05,
+      "loss": 1.1596,
+      "step": 5405500
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.334566813104265e-05,
+      "loss": 1.1605,
+      "step": 5406000
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3343572365413206e-05,
+      "loss": 1.1731,
+      "step": 5406500
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.334147239985264e-05,
+      "loss": 1.1637,
+      "step": 5407000
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.333937243429208e-05,
+      "loss": 1.1695,
+      "step": 5407500
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.333727246873151e-05,
+      "loss": 1.1654,
+      "step": 5408000
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3335172503170946e-05,
+      "loss": 1.1595,
+      "step": 5408500
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3333072537610386e-05,
+      "loss": 1.1599,
+      "step": 5409000
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.333097257204982e-05,
+      "loss": 1.1603,
+      "step": 5409500
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.332887260648925e-05,
+      "loss": 1.1606,
+      "step": 5410000
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.332677684085981e-05,
+      "loss": 1.1638,
+      "step": 5410500
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.332467687529925e-05,
+      "loss": 1.161,
+      "step": 5411000
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.332257690973868e-05,
+      "loss": 1.18,
+      "step": 5411500
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3320476944178114e-05,
+      "loss": 1.1594,
+      "step": 5412000
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.331838117854867e-05,
+      "loss": 1.1587,
+      "step": 5412500
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.331628121298811e-05,
+      "loss": 1.1667,
+      "step": 5413000
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.331418544735866e-05,
+      "loss": 1.1327,
+      "step": 5413500
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3312085481798095e-05,
+      "loss": 1.167,
+      "step": 5414000
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3309985516237535e-05,
+      "loss": 1.155,
+      "step": 5414500
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.330788555067697e-05,
+      "loss": 1.1407,
+      "step": 5415000
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.330578978504752e-05,
+      "loss": 1.1831,
+      "step": 5415500
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3303689819486956e-05,
+      "loss": 1.1503,
+      "step": 5416000
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3301589853926396e-05,
+      "loss": 1.1606,
+      "step": 5416500
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.329948988836583e-05,
+      "loss": 1.1233,
+      "step": 5417000
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.329738992280526e-05,
+      "loss": 1.159,
+      "step": 5417500
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.32952899572447e-05,
+      "loss": 1.1773,
+      "step": 5418000
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3293194191615257e-05,
+      "loss": 1.1834,
+      "step": 5418500
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.329109422605469e-05,
+      "loss": 1.1522,
+      "step": 5419000
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3288994260494123e-05,
+      "loss": 1.1777,
+      "step": 5419500
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3286894294933564e-05,
+      "loss": 1.1646,
+      "step": 5420000
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3284794329373e-05,
+      "loss": 1.1515,
+      "step": 5420500
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.328269856374355e-05,
+      "loss": 1.1313,
+      "step": 5421000
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.328059859818299e-05,
+      "loss": 1.1284,
+      "step": 5421500
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3278498632622424e-05,
+      "loss": 1.1759,
+      "step": 5422000
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.327639866706186e-05,
+      "loss": 1.1636,
+      "step": 5422500
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.32742987015013e-05,
+      "loss": 1.1684,
+      "step": 5423000
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.327219873594073e-05,
+      "loss": 1.1606,
+      "step": 5423500
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3270098770380165e-05,
+      "loss": 1.1744,
+      "step": 5424000
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3267998804819605e-05,
+      "loss": 1.182,
+      "step": 5424500
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.326590303919016e-05,
+      "loss": 1.1715,
+      "step": 5425000
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.326380307362959e-05,
+      "loss": 1.1513,
+      "step": 5425500
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3261703108069026e-05,
+      "loss": 1.147,
+      "step": 5426000
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3259603142508466e-05,
+      "loss": 1.1498,
+      "step": 5426500
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.325750737687902e-05,
+      "loss": 1.1705,
+      "step": 5427000
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.325540741131845e-05,
+      "loss": 1.1551,
+      "step": 5427500
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.325330744575789e-05,
+      "loss": 1.1418,
+      "step": 5428000
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.325120748019733e-05,
+      "loss": 1.1837,
+      "step": 5428500
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.324911171456788e-05,
+      "loss": 1.1873,
+      "step": 5429000
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3247011749007314e-05,
+      "loss": 1.1557,
+      "step": 5429500
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3244911783446754e-05,
+      "loss": 1.1508,
+      "step": 5430000
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.324281181788619e-05,
+      "loss": 1.1694,
+      "step": 5430500
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.324071605225674e-05,
+      "loss": 1.1638,
+      "step": 5431000
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.32386202866273e-05,
+      "loss": 1.161,
+      "step": 5431500
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.323652032106673e-05,
+      "loss": 1.2012,
+      "step": 5432000
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.323442035550617e-05,
+      "loss": 1.1718,
+      "step": 5432500
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.32323203899456e-05,
+      "loss": 1.1588,
+      "step": 5433000
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.323022042438504e-05,
+      "loss": 1.1767,
+      "step": 5433500
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3228124658755596e-05,
+      "loss": 1.1446,
+      "step": 5434000
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.322602469319503e-05,
+      "loss": 1.1679,
+      "step": 5434500
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.322392472763446e-05,
+      "loss": 1.1593,
+      "step": 5435000
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.32218247620739e-05,
+      "loss": 1.1671,
+      "step": 5435500
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3219724796513336e-05,
+      "loss": 1.1591,
+      "step": 5436000
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.321762483095277e-05,
+      "loss": 1.1673,
+      "step": 5436500
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.321552486539221e-05,
+      "loss": 1.1723,
+      "step": 5437000
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.321342489983164e-05,
+      "loss": 1.1452,
+      "step": 5437500
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.321132493427108e-05,
+      "loss": 1.1548,
+      "step": 5438000
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.320922916864163e-05,
+      "loss": 1.1962,
+      "step": 5438500
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.320712920308107e-05,
+      "loss": 1.1682,
+      "step": 5439000
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3205033437451624e-05,
+      "loss": 1.1735,
+      "step": 5439500
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.320293347189106e-05,
+      "loss": 1.1659,
+      "step": 5440000
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.32008335063305e-05,
+      "loss": 1.1713,
+      "step": 5440500
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.319873354076993e-05,
+      "loss": 1.1762,
+      "step": 5441000
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3196633575209365e-05,
+      "loss": 1.1634,
+      "step": 5441500
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3194533609648805e-05,
+      "loss": 1.1536,
+      "step": 5442000
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.319243364408824e-05,
+      "loss": 1.1683,
+      "step": 5442500
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.319033367852767e-05,
+      "loss": 1.1876,
+      "step": 5443000
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3188237912898226e-05,
+      "loss": 1.1797,
+      "step": 5443500
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3186137947337666e-05,
+      "loss": 1.1658,
+      "step": 5444000
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.318404218170822e-05,
+      "loss": 1.1528,
+      "step": 5444500
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.318194221614765e-05,
+      "loss": 1.1678,
+      "step": 5445000
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3179842250587086e-05,
+      "loss": 1.1561,
+      "step": 5445500
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3177742285026526e-05,
+      "loss": 1.1711,
+      "step": 5446000
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.317564231946596e-05,
+      "loss": 1.1713,
+      "step": 5446500
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3173542353905393e-05,
+      "loss": 1.1589,
+      "step": 5447000
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3171446588275954e-05,
+      "loss": 1.1698,
+      "step": 5447500
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.316934662271539e-05,
+      "loss": 1.1669,
+      "step": 5448000
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.316724665715482e-05,
+      "loss": 1.1636,
+      "step": 5448500
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.316514669159426e-05,
+      "loss": 1.1744,
+      "step": 5449000
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3163046726033694e-05,
+      "loss": 1.1552,
+      "step": 5449500
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.316094676047313e-05,
+      "loss": 1.1609,
+      "step": 5450000
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.315885099484368e-05,
+      "loss": 1.1952,
+      "step": 5450500
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.315675102928312e-05,
+      "loss": 1.1671,
+      "step": 5451000
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3154651063722555e-05,
+      "loss": 1.1823,
+      "step": 5451500
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.315255529809311e-05,
+      "loss": 1.172,
+      "step": 5452000
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.315045533253254e-05,
+      "loss": 1.1717,
+      "step": 5452500
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.314835536697198e-05,
+      "loss": 1.1687,
+      "step": 5453000
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3146255401411416e-05,
+      "loss": 1.1762,
+      "step": 5453500
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.314415543585085e-05,
+      "loss": 1.1514,
+      "step": 5454000
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.314205547029029e-05,
+      "loss": 1.1831,
+      "step": 5454500
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.313995550472972e-05,
+      "loss": 1.1613,
+      "step": 5455000
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3137855539169156e-05,
+      "loss": 1.1873,
+      "step": 5455500
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.31357555736086e-05,
+      "loss": 1.1413,
+      "step": 5456000
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.313365980797915e-05,
+      "loss": 1.1679,
+      "step": 5456500
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3131559842418584e-05,
+      "loss": 1.1777,
+      "step": 5457000
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3129459876858024e-05,
+      "loss": 1.1392,
+      "step": 5457500
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.312735991129746e-05,
+      "loss": 1.2143,
+      "step": 5458000
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.312526414566801e-05,
+      "loss": 1.1879,
+      "step": 5458500
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3123164180107444e-05,
+      "loss": 1.1868,
+      "step": 5459000
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3121064214546885e-05,
+      "loss": 1.185,
+      "step": 5459500
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.311896424898632e-05,
+      "loss": 1.1529,
+      "step": 5460000
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.311686428342575e-05,
+      "loss": 1.1518,
+      "step": 5460500
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3114764317865185e-05,
+      "loss": 1.188,
+      "step": 5461000
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.311266435230462e-05,
+      "loss": 1.1941,
+      "step": 5461500
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.311056438674405e-05,
+      "loss": 1.1534,
+      "step": 5462000
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.310847282104573e-05,
+      "loss": 1.1342,
+      "step": 5462500
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.310637285548517e-05,
+      "loss": 1.1565,
+      "step": 5463000
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3104272889924606e-05,
+      "loss": 1.1423,
+      "step": 5463500
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.310217292436404e-05,
+      "loss": 1.1779,
+      "step": 5464000
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.310007295880348e-05,
+      "loss": 1.1334,
+      "step": 5464500
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.309797299324291e-05,
+      "loss": 1.1437,
+      "step": 5465000
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.309587722761347e-05,
+      "loss": 1.1339,
+      "step": 5465500
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.30937772620529e-05,
+      "loss": 1.1452,
+      "step": 5466000
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.309167729649234e-05,
+      "loss": 1.1541,
+      "step": 5466500
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3089577330931774e-05,
+      "loss": 1.1505,
+      "step": 5467000
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.308747736537121e-05,
+      "loss": 1.1438,
+      "step": 5467500
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.308537739981065e-05,
+      "loss": 1.1637,
+      "step": 5468000
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.30832816341812e-05,
+      "loss": 1.1919,
+      "step": 5468500
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3081181668620635e-05,
+      "loss": 1.1632,
+      "step": 5469000
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.307908170306007e-05,
+      "loss": 1.1246,
+      "step": 5469500
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.307698173749951e-05,
+      "loss": 1.1809,
+      "step": 5470000
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3074881771938935e-05,
+      "loss": 1.1424,
+      "step": 5470500
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3072786006309495e-05,
+      "loss": 1.1798,
+      "step": 5471000
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3070686040748936e-05,
+      "loss": 1.155,
+      "step": 5471500
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.306858607518837e-05,
+      "loss": 1.1303,
+      "step": 5472000
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.30664861096278e-05,
+      "loss": 1.1962,
+      "step": 5472500
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3064386144067236e-05,
+      "loss": 1.1701,
+      "step": 5473000
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.306228617850667e-05,
+      "loss": 1.1438,
+      "step": 5473500
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.30601862129461e-05,
+      "loss": 1.1357,
+      "step": 5474000
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.305808624738554e-05,
+      "loss": 1.1799,
+      "step": 5474500
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.305598628182498e-05,
+      "loss": 1.1823,
+      "step": 5475000
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.305389051619553e-05,
+      "loss": 1.1611,
+      "step": 5475500
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.305179895049721e-05,
+      "loss": 1.2026,
+      "step": 5476000
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3049698984936644e-05,
+      "loss": 1.1814,
+      "step": 5476500
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3047599019376084e-05,
+      "loss": 1.1593,
+      "step": 5477000
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.304549905381552e-05,
+      "loss": 1.1712,
+      "step": 5477500
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.304339908825495e-05,
+      "loss": 1.1595,
+      "step": 5478000
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.304129912269439e-05,
+      "loss": 1.1467,
+      "step": 5478500
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3039199157133825e-05,
+      "loss": 1.1679,
+      "step": 5479000
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.303709919157326e-05,
+      "loss": 1.1644,
+      "step": 5479500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.303499922601269e-05,
+      "loss": 1.1744,
+      "step": 5480000
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.303290346038325e-05,
+      "loss": 1.1404,
+      "step": 5480500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3030803494822686e-05,
+      "loss": 1.1551,
+      "step": 5481000
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.302870352926212e-05,
+      "loss": 1.199,
+      "step": 5481500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.302660356370156e-05,
+      "loss": 1.1867,
+      "step": 5482000
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3024503598140986e-05,
+      "loss": 1.1575,
+      "step": 5482500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3022407832511547e-05,
+      "loss": 1.1857,
+      "step": 5483000
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.302030786695098e-05,
+      "loss": 1.1904,
+      "step": 5483500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.301820790139042e-05,
+      "loss": 1.1542,
+      "step": 5484000
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3016107935829854e-05,
+      "loss": 1.1606,
+      "step": 5484500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.301400797026929e-05,
+      "loss": 1.185,
+      "step": 5485000
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.301190800470872e-05,
+      "loss": 1.1759,
+      "step": 5485500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.300981223907928e-05,
+      "loss": 1.1896,
+      "step": 5486000
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3007712273518714e-05,
+      "loss": 1.1442,
+      "step": 5486500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3005612307958155e-05,
+      "loss": 1.1471,
+      "step": 5487000
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.300351234239758e-05,
+      "loss": 1.1923,
+      "step": 5487500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.300141657676814e-05,
+      "loss": 1.2072,
+      "step": 5488000
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.2999316611207575e-05,
+      "loss": 1.1648,
+      "step": 5488500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.2997216645647015e-05,
+      "loss": 1.1729,
+      "step": 5489000
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.299511668008644e-05,
+      "loss": 1.1857,
+      "step": 5489500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.2993016714525875e-05,
+      "loss": 1.1502,
+      "step": 5490000
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.2990916748965316e-05,
+      "loss": 1.1833,
+      "step": 5490500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.298881678340475e-05,
+      "loss": 1.1546,
+      "step": 5491000
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.298671681784418e-05,
+      "loss": 1.1611,
+      "step": 5491500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.298462105221474e-05,
+      "loss": 1.1693,
+      "step": 5492000
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.2982521086654176e-05,
+      "loss": 1.1526,
+      "step": 5492500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.298042112109361e-05,
+      "loss": 1.1717,
+      "step": 5493000
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.297832115553305e-05,
+      "loss": 1.1482,
+      "step": 5493500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.2976221189972484e-05,
+      "loss": 1.1426,
+      "step": 5494000
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.297412542434304e-05,
+      "loss": 1.1362,
+      "step": 5494500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.297202545878247e-05,
+      "loss": 1.1979,
+      "step": 5495000
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.296992549322191e-05,
+      "loss": 1.1556,
+      "step": 5495500
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.2967825527661344e-05,
+      "loss": 1.1764,
+      "step": 5496000
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.296572556210078e-05,
+      "loss": 1.1845,
+      "step": 5496500
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.296362979647133e-05,
+      "loss": 1.1422,
+      "step": 5497000
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.296152983091077e-05,
+      "loss": 1.1756,
+      "step": 5497500
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.2959429865350205e-05,
+      "loss": 1.1572,
+      "step": 5498000
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.2957329899789645e-05,
+      "loss": 1.167,
+      "step": 5498500
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.295522993422908e-05,
+      "loss": 1.1462,
+      "step": 5499000
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.295312996866851e-05,
+      "loss": 1.1485,
+      "step": 5499500
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.295103000310795e-05,
+      "loss": 1.1524,
+      "step": 5500000
+    },
+    {
+      "epoch": 3.3,
+      "eval_loss": 1.1378905773162842,
+      "eval_runtime": 1099.047,
+      "eval_samples_per_second": 479.252,
+      "eval_steps_per_second": 79.876,
+      "step": 5500000
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.2948930037547386e-05,
+      "loss": 1.1625,
+      "step": 5500500
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.2946838471849066e-05,
+      "loss": 1.1708,
+      "step": 5501000
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.294473850628849e-05,
+      "loss": 1.1493,
+      "step": 5501500
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.2942638540727927e-05,
+      "loss": 1.1854,
+      "step": 5502000
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.294053857516737e-05,
+      "loss": 1.1913,
+      "step": 5502500
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.29384386096068e-05,
+      "loss": 1.1578,
+      "step": 5503000
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.2936338644046234e-05,
+      "loss": 1.1537,
+      "step": 5503500
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.2934238678485674e-05,
+      "loss": 1.1585,
+      "step": 5504000
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.293213871292511e-05,
+      "loss": 1.1688,
+      "step": 5504500
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.293003874736454e-05,
+      "loss": 1.1186,
+      "step": 5505000
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.29279429817351e-05,
+      "loss": 1.1879,
+      "step": 5505500
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.2925843016174535e-05,
+      "loss": 1.1895,
+      "step": 5506000
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.292374725054509e-05,
+      "loss": 1.1507,
+      "step": 5506500
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.292164728498452e-05,
+      "loss": 1.1481,
+      "step": 5507000
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.291954731942396e-05,
+      "loss": 1.1713,
+      "step": 5507500
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.2917447353863395e-05,
+      "loss": 1.138,
+      "step": 5508000
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.291534738830283e-05,
+      "loss": 1.1811,
+      "step": 5508500
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.291324742274227e-05,
+      "loss": 1.133,
+      "step": 5509000
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.29111474571817e-05,
+      "loss": 1.155,
+      "step": 5509500
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.2909047491621136e-05,
+      "loss": 1.1687,
+      "step": 5510000
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.290695172599169e-05,
+      "loss": 1.1658,
+      "step": 5510500
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.290485176043113e-05,
+      "loss": 1.1992,
+      "step": 5511000
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.290275179487056e-05,
+      "loss": 1.1878,
+      "step": 5511500
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.290065182931e-05,
+      "loss": 1.1614,
+      "step": 5512000
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.289855606368056e-05,
+      "loss": 1.145,
+      "step": 5512500
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.289646029805112e-05,
+      "loss": 1.1661,
+      "step": 5513000
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.2894360332490544e-05,
+      "loss": 1.17,
+      "step": 5513500
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.289226036692998e-05,
+      "loss": 1.1862,
+      "step": 5514000
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.289016040136942e-05,
+      "loss": 1.186,
+      "step": 5514500
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.288806043580885e-05,
+      "loss": 1.1473,
+      "step": 5515000
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.2885960470248285e-05,
+      "loss": 1.1498,
+      "step": 5515500
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.288386470461884e-05,
+      "loss": 1.1522,
+      "step": 5516000
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.288176473905828e-05,
+      "loss": 1.1552,
+      "step": 5516500
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.287966477349771e-05,
+      "loss": 1.1527,
+      "step": 5517000
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.2877564807937145e-05,
+      "loss": 1.1367,
+      "step": 5517500
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.2875464842376586e-05,
+      "loss": 1.1411,
+      "step": 5518000
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.287336907674714e-05,
+      "loss": 1.1738,
+      "step": 5518500
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.287126911118657e-05,
+      "loss": 1.1583,
+      "step": 5519000
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.286916914562601e-05,
+      "loss": 1.1476,
+      "step": 5519500
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.2867069180065446e-05,
+      "loss": 1.1541,
+      "step": 5520000
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.2864973414436e-05,
+      "loss": 1.1742,
+      "step": 5520500
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.2862873448875433e-05,
+      "loss": 1.1569,
+      "step": 5521000
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.2860773483314874e-05,
+      "loss": 1.1619,
+      "step": 5521500
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.285867351775431e-05,
+      "loss": 1.1665,
+      "step": 5522000
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.285657355219374e-05,
+      "loss": 1.141,
+      "step": 5522500
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.2854477786564294e-05,
+      "loss": 1.14,
+      "step": 5523000
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.2852377821003734e-05,
+      "loss": 1.1496,
+      "step": 5523500
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.285027785544317e-05,
+      "loss": 1.1462,
+      "step": 5524000
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.28481778898826e-05,
+      "loss": 1.2024,
+      "step": 5524500
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.284607792432204e-05,
+      "loss": 1.1859,
+      "step": 5525000
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.2843977958761475e-05,
+      "loss": 1.153,
+      "step": 5525500
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.284187799320091e-05,
+      "loss": 1.1873,
+      "step": 5526000
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.283977802764035e-05,
+      "loss": 1.1425,
+      "step": 5526500
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.283768646194203e-05,
+      "loss": 1.1748,
+      "step": 5527000
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.283558649638146e-05,
+      "loss": 1.1748,
+      "step": 5527500
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.283348653082089e-05,
+      "loss": 1.1708,
+      "step": 5528000
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.283138656526033e-05,
+      "loss": 1.1719,
+      "step": 5528500
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.282928659969976e-05,
+      "loss": 1.1561,
+      "step": 5529000
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.282719083407032e-05,
+      "loss": 1.1466,
+      "step": 5529500
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.282509086850976e-05,
+      "loss": 1.1463,
+      "step": 5530000
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.282299090294919e-05,
+      "loss": 1.1693,
+      "step": 5530500
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.2820890937388624e-05,
+      "loss": 1.1917,
+      "step": 5531000
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.281879097182806e-05,
+      "loss": 1.1586,
+      "step": 5531500
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.28166910062675e-05,
+      "loss": 1.1624,
+      "step": 5532000
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.281459104070693e-05,
+      "loss": 1.1395,
+      "step": 5532500
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.2812491075146364e-05,
+      "loss": 1.1747,
+      "step": 5533000
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.2810395309516925e-05,
+      "loss": 1.1776,
+      "step": 5533500
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.2808299543887485e-05,
+      "loss": 1.1549,
+      "step": 5534000
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.280619957832692e-05,
+      "loss": 1.156,
+      "step": 5534500
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.2804099612766345e-05,
+      "loss": 1.1574,
+      "step": 5535000
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.2801999647205785e-05,
+      "loss": 1.1571,
+      "step": 5535500
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.279989968164522e-05,
+      "loss": 1.1835,
+      "step": 5536000
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.279779971608465e-05,
+      "loss": 1.1816,
+      "step": 5536500
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.279569975052409e-05,
+      "loss": 1.1394,
+      "step": 5537000
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.2793599784963526e-05,
+      "loss": 1.131,
+      "step": 5537500
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.279149981940296e-05,
+      "loss": 1.1544,
+      "step": 5538000
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.27893998538424e-05,
+      "loss": 1.182,
+      "step": 5538500
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.278729988828183e-05,
+      "loss": 1.1708,
+      "step": 5539000
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.278519992272127e-05,
+      "loss": 1.1644,
+      "step": 5539500
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.278310415709182e-05,
+      "loss": 1.1766,
+      "step": 5540000
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.278100419153126e-05,
+      "loss": 1.1715,
+      "step": 5540500
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.2778904225970694e-05,
+      "loss": 1.1449,
+      "step": 5541000
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.277680846034125e-05,
+      "loss": 1.1597,
+      "step": 5541500
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.277470849478069e-05,
+      "loss": 1.1782,
+      "step": 5542000
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.277260852922012e-05,
+      "loss": 1.1908,
+      "step": 5542500
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.2770508563659555e-05,
+      "loss": 1.1709,
+      "step": 5543000
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.2768408598098995e-05,
+      "loss": 1.1774,
+      "step": 5543500
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.276631283246955e-05,
+      "loss": 1.1743,
+      "step": 5544000
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.276421286690898e-05,
+      "loss": 1.1763,
+      "step": 5544500
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.2762117101279535e-05,
+      "loss": 1.1776,
+      "step": 5545000
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.276001713571897e-05,
+      "loss": 1.1648,
+      "step": 5545500
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.275791717015841e-05,
+      "loss": 1.1566,
+      "step": 5546000
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.275581720459784e-05,
+      "loss": 1.172,
+      "step": 5546500
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.2753717239037276e-05,
+      "loss": 1.1489,
+      "step": 5547000
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.2751617273476716e-05,
+      "loss": 1.181,
+      "step": 5547500
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.274951730791615e-05,
+      "loss": 1.1685,
+      "step": 5548000
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.274741734235558e-05,
+      "loss": 1.1771,
+      "step": 5548500
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.2745321576726144e-05,
+      "loss": 1.1511,
+      "step": 5549000
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.274322161116558e-05,
+      "loss": 1.1519,
+      "step": 5549500
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.274112164560501e-05,
+      "loss": 1.1419,
+      "step": 5550000
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.273902168004445e-05,
+      "loss": 1.1502,
+      "step": 5550500
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.2736921714483884e-05,
+      "loss": 1.1489,
+      "step": 5551000
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.273482174892332e-05,
+      "loss": 1.1766,
+      "step": 5551500
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.273272598329387e-05,
+      "loss": 1.1568,
+      "step": 5552000
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.2730630217664425e-05,
+      "loss": 1.1706,
+      "step": 5552500
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.2728530252103865e-05,
+      "loss": 1.1486,
+      "step": 5553000
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.27264302865433e-05,
+      "loss": 1.1537,
+      "step": 5553500
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.272433032098273e-05,
+      "loss": 1.1729,
+      "step": 5554000
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.272223035542217e-05,
+      "loss": 1.1485,
+      "step": 5554500
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.2720130389861606e-05,
+      "loss": 1.1878,
+      "step": 5555000
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.271803042430104e-05,
+      "loss": 1.1427,
+      "step": 5555500
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.271593045874048e-05,
+      "loss": 1.1504,
+      "step": 5556000
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.271383049317991e-05,
+      "loss": 1.1673,
+      "step": 5556500
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.2711730527619346e-05,
+      "loss": 1.1653,
+      "step": 5557000
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.270963476198991e-05,
+      "loss": 1.1439,
+      "step": 5557500
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.270753479642934e-05,
+      "loss": 1.1659,
+      "step": 5558000
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.2705434830868774e-05,
+      "loss": 1.1871,
+      "step": 5558500
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.270333906523933e-05,
+      "loss": 1.1778,
+      "step": 5559000
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.270123909967877e-05,
+      "loss": 1.1815,
+      "step": 5559500
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.26991391341182e-05,
+      "loss": 1.1466,
+      "step": 5560000
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.2697039168557634e-05,
+      "loss": 1.1849,
+      "step": 5560500
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.2694939202997075e-05,
+      "loss": 1.1925,
+      "step": 5561000
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.269283923743651e-05,
+      "loss": 1.1572,
+      "step": 5561500
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.269074347180706e-05,
+      "loss": 1.1872,
+      "step": 5562000
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.2688643506246495e-05,
+      "loss": 1.1712,
+      "step": 5562500
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.2686543540685935e-05,
+      "loss": 1.1812,
+      "step": 5563000
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.268444357512537e-05,
+      "loss": 1.1597,
+      "step": 5563500
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.268234360956481e-05,
+      "loss": 1.1551,
+      "step": 5564000
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.2680243644004236e-05,
+      "loss": 1.184,
+      "step": 5564500
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.267814367844367e-05,
+      "loss": 1.1797,
+      "step": 5565000
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.267604371288311e-05,
+      "loss": 1.1994,
+      "step": 5565500
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.267394374732254e-05,
+      "loss": 1.1719,
+      "step": 5566000
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.26718479816931e-05,
+      "loss": 1.1334,
+      "step": 5566500
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.266974801613253e-05,
+      "loss": 1.1589,
+      "step": 5567000
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.266764805057197e-05,
+      "loss": 1.1703,
+      "step": 5567500
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.2665548085011403e-05,
+      "loss": 1.1446,
+      "step": 5568000
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.266344811945084e-05,
+      "loss": 1.1726,
+      "step": 5568500
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.266135235382139e-05,
+      "loss": 1.1584,
+      "step": 5569000
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.265925238826083e-05,
+      "loss": 1.1766,
+      "step": 5569500
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.2657152422700264e-05,
+      "loss": 1.172,
+      "step": 5570000
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.2655052457139704e-05,
+      "loss": 1.176,
+      "step": 5570500
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.265295249157914e-05,
+      "loss": 1.161,
+      "step": 5571000
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.265085252601857e-05,
+      "loss": 1.1801,
+      "step": 5571500
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.264875256045801e-05,
+      "loss": 1.1811,
+      "step": 5572000
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.2646656794828565e-05,
+      "loss": 1.1385,
+      "step": 5572500
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.2644561029199126e-05,
+      "loss": 1.1526,
+      "step": 5573000
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.264246106363856e-05,
+      "loss": 1.1776,
+      "step": 5573500
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.2640361098077986e-05,
+      "loss": 1.1654,
+      "step": 5574000
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.2638261132517426e-05,
+      "loss": 1.1677,
+      "step": 5574500
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.263616116695686e-05,
+      "loss": 1.1366,
+      "step": 5575000
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.263406120139629e-05,
+      "loss": 1.1738,
+      "step": 5575500
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.263196123583573e-05,
+      "loss": 1.1653,
+      "step": 5576000
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.2629861270275167e-05,
+      "loss": 1.1197,
+      "step": 5576500
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.26277613047146e-05,
+      "loss": 1.1872,
+      "step": 5577000
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.262566133915404e-05,
+      "loss": 1.1891,
+      "step": 5577500
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.2623565573524594e-05,
+      "loss": 1.1761,
+      "step": 5578000
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.262146560796403e-05,
+      "loss": 1.173,
+      "step": 5578500
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.261936564240347e-05,
+      "loss": 1.2092,
+      "step": 5579000
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.26172656768429e-05,
+      "loss": 1.1165,
+      "step": 5579500
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.2615165711282334e-05,
+      "loss": 1.174,
+      "step": 5580000
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.2613065745721775e-05,
+      "loss": 1.1497,
+      "step": 5580500
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.261096578016121e-05,
+      "loss": 1.1227,
+      "step": 5581000
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.260886581460064e-05,
+      "loss": 1.1727,
+      "step": 5581500
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.2606774248902315e-05,
+      "loss": 1.1553,
+      "step": 5582000
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.260467428334175e-05,
+      "loss": 1.1526,
+      "step": 5582500
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.260257431778119e-05,
+      "loss": 1.1357,
+      "step": 5583000
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.260047435222062e-05,
+      "loss": 1.1499,
+      "step": 5583500
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.2598374386660056e-05,
+      "loss": 1.1476,
+      "step": 5584000
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.2596274421099496e-05,
+      "loss": 1.177,
+      "step": 5584500
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.259417445553893e-05,
+      "loss": 1.1596,
+      "step": 5585000
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.259207868990948e-05,
+      "loss": 1.1741,
+      "step": 5585500
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.258997872434892e-05,
+      "loss": 1.1765,
+      "step": 5586000
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.258787875878836e-05,
+      "loss": 1.1748,
+      "step": 5586500
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.258577879322779e-05,
+      "loss": 1.1551,
+      "step": 5587000
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.258367882766723e-05,
+      "loss": 1.1721,
+      "step": 5587500
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.2581578862106664e-05,
+      "loss": 1.1569,
+      "step": 5588000
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.25794788965461e-05,
+      "loss": 1.1801,
+      "step": 5588500
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.257737893098553e-05,
+      "loss": 1.1581,
+      "step": 5589000
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.2575278965424964e-05,
+      "loss": 1.1685,
+      "step": 5589500
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.25731789998644e-05,
+      "loss": 1.1671,
+      "step": 5590000
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.257107903430384e-05,
+      "loss": 1.181,
+      "step": 5590500
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.256897906874327e-05,
+      "loss": 1.1481,
+      "step": 5591000
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.2566883303113825e-05,
+      "loss": 1.1585,
+      "step": 5591500
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.256478333755326e-05,
+      "loss": 1.1532,
+      "step": 5592000
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.256268757192382e-05,
+      "loss": 1.1392,
+      "step": 5592500
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.256058760636326e-05,
+      "loss": 1.1482,
+      "step": 5593000
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.2558487640802686e-05,
+      "loss": 1.1802,
+      "step": 5593500
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.2556387675242126e-05,
+      "loss": 1.1745,
+      "step": 5594000
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.255428770968156e-05,
+      "loss": 1.1544,
+      "step": 5594500
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.255219194405212e-05,
+      "loss": 1.177,
+      "step": 5595000
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.255009197849155e-05,
+      "loss": 1.1666,
+      "step": 5595500
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.254799201293099e-05,
+      "loss": 1.149,
+      "step": 5596000
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.254589204737042e-05,
+      "loss": 1.1711,
+      "step": 5596500
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.2543792081809854e-05,
+      "loss": 1.1619,
+      "step": 5597000
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.2541692116249294e-05,
+      "loss": 1.1795,
+      "step": 5597500
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.253959215068873e-05,
+      "loss": 1.1767,
+      "step": 5598000
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.253749218512816e-05,
+      "loss": 1.156,
+      "step": 5598500
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.2535396419498714e-05,
+      "loss": 1.1811,
+      "step": 5599000
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.2533296453938155e-05,
+      "loss": 1.1904,
+      "step": 5599500
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.253119648837759e-05,
+      "loss": 1.1792,
+      "step": 5600000
+    },
+    {
+      "epoch": 3.36,
+      "eval_loss": 1.133358359336853,
+      "eval_runtime": 1102.5182,
+      "eval_samples_per_second": 477.743,
+      "eval_steps_per_second": 79.624,
+      "step": 5600000
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.252909652281702e-05,
+      "loss": 1.1808,
+      "step": 5600500
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.252699655725646e-05,
+      "loss": 1.1619,
+      "step": 5601000
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.2524896591695895e-05,
+      "loss": 1.1489,
+      "step": 5601500
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.252279662613533e-05,
+      "loss": 1.1419,
+      "step": 5602000
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.252069666057477e-05,
+      "loss": 1.1486,
+      "step": 5602500
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.251860509487645e-05,
+      "loss": 1.1613,
+      "step": 5603000
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.2516505129315876e-05,
+      "loss": 1.1704,
+      "step": 5603500
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.251440516375531e-05,
+      "loss": 1.1719,
+      "step": 5604000
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.251230519819475e-05,
+      "loss": 1.1553,
+      "step": 5604500
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.251020943256531e-05,
+      "loss": 1.1717,
+      "step": 5605000
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.250810946700474e-05,
+      "loss": 1.1541,
+      "step": 5605500
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.250600950144417e-05,
+      "loss": 1.1557,
+      "step": 5606000
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.250390953588361e-05,
+      "loss": 1.1617,
+      "step": 5606500
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.2501809570323044e-05,
+      "loss": 1.1977,
+      "step": 5607000
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.249970960476248e-05,
+      "loss": 1.1225,
+      "step": 5607500
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.249761383913304e-05,
+      "loss": 1.1355,
+      "step": 5608000
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.249551387357247e-05,
+      "loss": 1.2044,
+      "step": 5608500
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.2493413908011905e-05,
+      "loss": 1.1625,
+      "step": 5609000
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.2491313942451345e-05,
+      "loss": 1.218,
+      "step": 5609500
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.248921397689078e-05,
+      "loss": 1.1813,
+      "step": 5610000
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.248711401133021e-05,
+      "loss": 1.1655,
+      "step": 5610500
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.248501404576965e-05,
+      "loss": 1.1774,
+      "step": 5611000
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.2482914080209086e-05,
+      "loss": 1.1804,
+      "step": 5611500
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.248081411464852e-05,
+      "loss": 1.1701,
+      "step": 5612000
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.247871414908796e-05,
+      "loss": 1.1715,
+      "step": 5612500
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.247661418352739e-05,
+      "loss": 1.1438,
+      "step": 5613000
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.247451421796682e-05,
+      "loss": 1.1704,
+      "step": 5613500
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.247241845233738e-05,
+      "loss": 1.1741,
+      "step": 5614000
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.247032268670793e-05,
+      "loss": 1.1767,
+      "step": 5614500
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.2468222721147374e-05,
+      "loss": 1.1701,
+      "step": 5615000
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.246612275558681e-05,
+      "loss": 1.1704,
+      "step": 5615500
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.246402279002624e-05,
+      "loss": 1.1816,
+      "step": 5616000
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.246192282446568e-05,
+      "loss": 1.1832,
+      "step": 5616500
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.2459822858905114e-05,
+      "loss": 1.1316,
+      "step": 5617000
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.2457722893344554e-05,
+      "loss": 1.1756,
+      "step": 5617500
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.245562292778399e-05,
+      "loss": 1.1507,
+      "step": 5618000
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.245352716215454e-05,
+      "loss": 1.1894,
+      "step": 5618500
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.2451427196593975e-05,
+      "loss": 1.1887,
+      "step": 5619000
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.2449327231033415e-05,
+      "loss": 1.1716,
+      "step": 5619500
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.244722726547285e-05,
+      "loss": 1.1675,
+      "step": 5620000
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.24451314998434e-05,
+      "loss": 1.162,
+      "step": 5620500
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.2443035734213956e-05,
+      "loss": 1.1518,
+      "step": 5621000
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.244093576865339e-05,
+      "loss": 1.1753,
+      "step": 5621500
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.243883580309283e-05,
+      "loss": 1.174,
+      "step": 5622000
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.243673583753226e-05,
+      "loss": 1.1792,
+      "step": 5622500
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.24346358719717e-05,
+      "loss": 1.1855,
+      "step": 5623000
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.2432535906411137e-05,
+      "loss": 1.1751,
+      "step": 5623500
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.243043594085057e-05,
+      "loss": 1.1527,
+      "step": 5624000
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.242833597529001e-05,
+      "loss": 1.1513,
+      "step": 5624500
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.2426240209660564e-05,
+      "loss": 1.1684,
+      "step": 5625000
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.242414444403112e-05,
+      "loss": 1.1693,
+      "step": 5625500
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.242204447847055e-05,
+      "loss": 1.1837,
+      "step": 5626000
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.2419944512909984e-05,
+      "loss": 1.1661,
+      "step": 5626500
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.2417844547349425e-05,
+      "loss": 1.1463,
+      "step": 5627000
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.241574878171998e-05,
+      "loss": 1.1676,
+      "step": 5627500
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.241364881615941e-05,
+      "loss": 1.1473,
+      "step": 5628000
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.2411548850598845e-05,
+      "loss": 1.1347,
+      "step": 5628500
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.2409448885038285e-05,
+      "loss": 1.1763,
+      "step": 5629000
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.240734891947772e-05,
+      "loss": 1.1725,
+      "step": 5629500
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.240524895391716e-05,
+      "loss": 1.1698,
+      "step": 5630000
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.240315318828771e-05,
+      "loss": 1.179,
+      "step": 5630500
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.2401053222727146e-05,
+      "loss": 1.1706,
+      "step": 5631000
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.239895325716658e-05,
+      "loss": 1.1759,
+      "step": 5631500
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.239685329160602e-05,
+      "loss": 1.1692,
+      "step": 5632000
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.239475332604545e-05,
+      "loss": 1.1353,
+      "step": 5632500
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.239265336048489e-05,
+      "loss": 1.1444,
+      "step": 5633000
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.239055759485544e-05,
+      "loss": 1.1692,
+      "step": 5633500
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.238845762929488e-05,
+      "loss": 1.1608,
+      "step": 5634000
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.2386357663734314e-05,
+      "loss": 1.1699,
+      "step": 5634500
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.238425769817375e-05,
+      "loss": 1.1614,
+      "step": 5635000
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.238216193254431e-05,
+      "loss": 1.15,
+      "step": 5635500
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.238006196698374e-05,
+      "loss": 1.1875,
+      "step": 5636000
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.2377962001423175e-05,
+      "loss": 1.1752,
+      "step": 5636500
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.2375862035862615e-05,
+      "loss": 1.1718,
+      "step": 5637000
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.237376207030205e-05,
+      "loss": 1.1659,
+      "step": 5637500
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.23716663046726e-05,
+      "loss": 1.1794,
+      "step": 5638000
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.2369566339112035e-05,
+      "loss": 1.1792,
+      "step": 5638500
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.2367466373551476e-05,
+      "loss": 1.1934,
+      "step": 5639000
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.236537060792203e-05,
+      "loss": 1.1655,
+      "step": 5639500
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.236327064236146e-05,
+      "loss": 1.1692,
+      "step": 5640000
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.2361170676800896e-05,
+      "loss": 1.1593,
+      "step": 5640500
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.2359070711240336e-05,
+      "loss": 1.1541,
+      "step": 5641000
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.235697074567977e-05,
+      "loss": 1.1514,
+      "step": 5641500
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.23548707801192e-05,
+      "loss": 1.1655,
+      "step": 5642000
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.2352770814558644e-05,
+      "loss": 1.1697,
+      "step": 5642500
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.235067084899808e-05,
+      "loss": 1.1728,
+      "step": 5643000
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.234857088343751e-05,
+      "loss": 1.1941,
+      "step": 5643500
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.234647091787695e-05,
+      "loss": 1.1786,
+      "step": 5644000
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.234437095231638e-05,
+      "loss": 1.1452,
+      "step": 5644500
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.234227098675582e-05,
+      "loss": 1.167,
+      "step": 5645000
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.234017522112638e-05,
+      "loss": 1.1921,
+      "step": 5645500
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.233807525556581e-05,
+      "loss": 1.174,
+      "step": 5646000
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.2335979489936365e-05,
+      "loss": 1.1794,
+      "step": 5646500
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.23338795243758e-05,
+      "loss": 1.1822,
+      "step": 5647000
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.233177955881524e-05,
+      "loss": 1.1585,
+      "step": 5647500
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.232967959325467e-05,
+      "loss": 1.1533,
+      "step": 5648000
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.2327579627694106e-05,
+      "loss": 1.2131,
+      "step": 5648500
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.2325479662133546e-05,
+      "loss": 1.171,
+      "step": 5649000
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.23233838965041e-05,
+      "loss": 1.1628,
+      "step": 5649500
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.232128813087465e-05,
+      "loss": 1.1329,
+      "step": 5650000
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.2319188165314086e-05,
+      "loss": 1.1574,
+      "step": 5650500
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.231708819975353e-05,
+      "loss": 1.1651,
+      "step": 5651000
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.231498823419296e-05,
+      "loss": 1.1865,
+      "step": 5651500
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.2312888268632394e-05,
+      "loss": 1.185,
+      "step": 5652000
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.2310788303071834e-05,
+      "loss": 1.1538,
+      "step": 5652500
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.230868833751127e-05,
+      "loss": 1.1288,
+      "step": 5653000
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.23065883719507e-05,
+      "loss": 1.1766,
+      "step": 5653500
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.2304492606321254e-05,
+      "loss": 1.1806,
+      "step": 5654000
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.2302392640760695e-05,
+      "loss": 1.126,
+      "step": 5654500
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.230029267520013e-05,
+      "loss": 1.2017,
+      "step": 5655000
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.229819270963956e-05,
+      "loss": 1.1871,
+      "step": 5655500
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.2296092744079e-05,
+      "loss": 1.156,
+      "step": 5656000
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.2293996978449555e-05,
+      "loss": 1.2061,
+      "step": 5656500
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.229189701288899e-05,
+      "loss": 1.1497,
+      "step": 5657000
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.228979704732842e-05,
+      "loss": 1.1775,
+      "step": 5657500
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.228769708176786e-05,
+      "loss": 1.1785,
+      "step": 5658000
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.2285597116207296e-05,
+      "loss": 1.1697,
+      "step": 5658500
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.228349715064673e-05,
+      "loss": 1.144,
+      "step": 5659000
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.228140138501729e-05,
+      "loss": 1.1763,
+      "step": 5659500
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.227930141945672e-05,
+      "loss": 1.1717,
+      "step": 5660000
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.2277201453896157e-05,
+      "loss": 1.1606,
+      "step": 5660500
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.227510148833559e-05,
+      "loss": 1.1757,
+      "step": 5661000
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.2273001522775024e-05,
+      "loss": 1.1664,
+      "step": 5661500
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.227090155721446e-05,
+      "loss": 1.1787,
+      "step": 5662000
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.22688015916539e-05,
+      "loss": 1.1451,
+      "step": 5662500
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.226670582602446e-05,
+      "loss": 1.1424,
+      "step": 5663000
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.2264605860463884e-05,
+      "loss": 1.1706,
+      "step": 5663500
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.226250589490332e-05,
+      "loss": 1.1625,
+      "step": 5664000
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.226040592934276e-05,
+      "loss": 1.1533,
+      "step": 5664500
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.225830596378219e-05,
+      "loss": 1.1547,
+      "step": 5665000
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.2256205998221625e-05,
+      "loss": 1.1701,
+      "step": 5665500
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.2254110232592185e-05,
+      "loss": 1.1818,
+      "step": 5666000
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.225201026703162e-05,
+      "loss": 1.161,
+      "step": 5666500
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.224991030147105e-05,
+      "loss": 1.1827,
+      "step": 5667000
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.224781033591049e-05,
+      "loss": 1.1575,
+      "step": 5667500
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.2245710370349926e-05,
+      "loss": 1.1496,
+      "step": 5668000
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.224361040478936e-05,
+      "loss": 1.1928,
+      "step": 5668500
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.224151463915991e-05,
+      "loss": 1.1721,
+      "step": 5669000
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.223941467359935e-05,
+      "loss": 1.1539,
+      "step": 5669500
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.2237314708038787e-05,
+      "loss": 1.1492,
+      "step": 5670000
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.223521474247822e-05,
+      "loss": 1.1445,
+      "step": 5670500
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.223311477691766e-05,
+      "loss": 1.1622,
+      "step": 5671000
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.2231019011288214e-05,
+      "loss": 1.165,
+      "step": 5671500
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.222891904572765e-05,
+      "loss": 1.1525,
+      "step": 5672000
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.222682328009821e-05,
+      "loss": 1.175,
+      "step": 5672500
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.222472331453764e-05,
+      "loss": 1.1524,
+      "step": 5673000
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.2222623348977075e-05,
+      "loss": 1.1751,
+      "step": 5673500
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.222052338341651e-05,
+      "loss": 1.1471,
+      "step": 5674000
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.221842341785595e-05,
+      "loss": 1.185,
+      "step": 5674500
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.221632345229538e-05,
+      "loss": 1.1564,
+      "step": 5675000
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.2214223486734815e-05,
+      "loss": 1.1444,
+      "step": 5675500
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.2212123521174255e-05,
+      "loss": 1.1681,
+      "step": 5676000
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.221002355561369e-05,
+      "loss": 1.1551,
+      "step": 5676500
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.220792359005312e-05,
+      "loss": 1.1717,
+      "step": 5677000
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.220582362449256e-05,
+      "loss": 1.1558,
+      "step": 5677500
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.2203727858863116e-05,
+      "loss": 1.1918,
+      "step": 5678000
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.220162789330255e-05,
+      "loss": 1.1825,
+      "step": 5678500
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.219952792774198e-05,
+      "loss": 1.1828,
+      "step": 5679000
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.219742796218142e-05,
+      "loss": 1.1944,
+      "step": 5679500
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.219533219655198e-05,
+      "loss": 1.1802,
+      "step": 5680000
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.219323223099141e-05,
+      "loss": 1.1518,
+      "step": 5680500
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.2191132265430844e-05,
+      "loss": 1.1576,
+      "step": 5681000
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.2189032299870284e-05,
+      "loss": 1.1639,
+      "step": 5681500
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.218693233430972e-05,
+      "loss": 1.1385,
+      "step": 5682000
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.218483236874916e-05,
+      "loss": 1.1536,
+      "step": 5682500
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.218273240318859e-05,
+      "loss": 1.1451,
+      "step": 5683000
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.218063243762802e-05,
+      "loss": 1.1563,
+      "step": 5683500
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.217853667199858e-05,
+      "loss": 1.1711,
+      "step": 5684000
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.217643670643802e-05,
+      "loss": 1.168,
+      "step": 5684500
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.217434094080857e-05,
+      "loss": 1.1343,
+      "step": 5685000
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.2172240975248005e-05,
+      "loss": 1.1616,
+      "step": 5685500
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.217014100968744e-05,
+      "loss": 1.1852,
+      "step": 5686000
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.216804104412688e-05,
+      "loss": 1.1697,
+      "step": 5686500
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.216594527849743e-05,
+      "loss": 1.1901,
+      "step": 5687000
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.2163845312936866e-05,
+      "loss": 1.139,
+      "step": 5687500
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.2161745347376306e-05,
+      "loss": 1.1687,
+      "step": 5688000
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.215964538181574e-05,
+      "loss": 1.1308,
+      "step": 5688500
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.215754541625517e-05,
+      "loss": 1.1691,
+      "step": 5689000
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.215544965062573e-05,
+      "loss": 1.1597,
+      "step": 5689500
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.215334968506517e-05,
+      "loss": 1.1738,
+      "step": 5690000
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.21512497195046e-05,
+      "loss": 1.1635,
+      "step": 5690500
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.2149153953875154e-05,
+      "loss": 1.1641,
+      "step": 5691000
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.214705398831459e-05,
+      "loss": 1.1459,
+      "step": 5691500
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.214495402275403e-05,
+      "loss": 1.1648,
+      "step": 5692000
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.214285405719346e-05,
+      "loss": 1.1644,
+      "step": 5692500
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.2140754091632895e-05,
+      "loss": 1.1855,
+      "step": 5693000
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.2138654126072335e-05,
+      "loss": 1.1784,
+      "step": 5693500
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.213655416051177e-05,
+      "loss": 1.1485,
+      "step": 5694000
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.21344541949512e-05,
+      "loss": 1.1618,
+      "step": 5694500
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.213235422939064e-05,
+      "loss": 1.1597,
+      "step": 5695000
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.213025426383007e-05,
+      "loss": 1.131,
+      "step": 5695500
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.212815429826951e-05,
+      "loss": 1.1581,
+      "step": 5696000
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.212605433270894e-05,
+      "loss": 1.1535,
+      "step": 5696500
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.21239585670795e-05,
+      "loss": 1.1674,
+      "step": 5697000
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.212185860151893e-05,
+      "loss": 1.1664,
+      "step": 5697500
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.211975863595837e-05,
+      "loss": 1.1804,
+      "step": 5698000
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.211766287032893e-05,
+      "loss": 1.1701,
+      "step": 5698500
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2115562904768364e-05,
+      "loss": 1.1658,
+      "step": 5699000
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.21134629392078e-05,
+      "loss": 1.1671,
+      "step": 5699500
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.211136297364723e-05,
+      "loss": 1.1312,
+      "step": 5700000
+    },
+    {
+      "epoch": 3.42,
+      "eval_loss": 1.132497787475586,
+      "eval_runtime": 1107.8438,
+      "eval_samples_per_second": 475.446,
+      "eval_steps_per_second": 79.241,
+      "step": 5700000
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.210926720801779e-05,
+      "loss": 1.1698,
+      "step": 5700500
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2107167242457224e-05,
+      "loss": 1.141,
+      "step": 5701000
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.210506727689666e-05,
+      "loss": 1.1997,
+      "step": 5701500
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.21029673113361e-05,
+      "loss": 1.1503,
+      "step": 5702000
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2100867345775525e-05,
+      "loss": 1.1567,
+      "step": 5702500
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2098767380214965e-05,
+      "loss": 1.1632,
+      "step": 5703000
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.20966674146544e-05,
+      "loss": 1.1549,
+      "step": 5703500
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.209456744909383e-05,
+      "loss": 1.1559,
+      "step": 5704000
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.209247168346439e-05,
+      "loss": 1.1703,
+      "step": 5704500
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2090371717903826e-05,
+      "loss": 1.1443,
+      "step": 5705000
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.208827175234326e-05,
+      "loss": 1.1661,
+      "step": 5705500
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.208617178678269e-05,
+      "loss": 1.1596,
+      "step": 5706000
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.208407602115325e-05,
+      "loss": 1.1798,
+      "step": 5706500
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.208197605559269e-05,
+      "loss": 1.1757,
+      "step": 5707000
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.207988028996325e-05,
+      "loss": 1.1434,
+      "step": 5707500
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.207778032440268e-05,
+      "loss": 1.161,
+      "step": 5708000
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2075680358842114e-05,
+      "loss": 1.1387,
+      "step": 5708500
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2073580393281554e-05,
+      "loss": 1.1582,
+      "step": 5709000
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.207148042772098e-05,
+      "loss": 1.1697,
+      "step": 5709500
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.206938046216042e-05,
+      "loss": 1.1606,
+      "step": 5710000
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2067280496599854e-05,
+      "loss": 1.1694,
+      "step": 5710500
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.206518053103929e-05,
+      "loss": 1.1898,
+      "step": 5711000
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.206308056547873e-05,
+      "loss": 1.1401,
+      "step": 5711500
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.206098479984928e-05,
+      "loss": 1.1683,
+      "step": 5712000
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2058884834288715e-05,
+      "loss": 1.1466,
+      "step": 5712500
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.205678486872815e-05,
+      "loss": 1.1438,
+      "step": 5713000
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.205468490316759e-05,
+      "loss": 1.1322,
+      "step": 5713500
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.205259333746926e-05,
+      "loss": 1.1568,
+      "step": 5714000
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.20504933719087e-05,
+      "loss": 1.1858,
+      "step": 5714500
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2048397606279256e-05,
+      "loss": 1.1724,
+      "step": 5715000
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.204629764071869e-05,
+      "loss": 1.1717,
+      "step": 5715500
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.204419767515813e-05,
+      "loss": 1.1658,
+      "step": 5716000
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2042097709597563e-05,
+      "loss": 1.1671,
+      "step": 5716500
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.204000194396812e-05,
+      "loss": 1.1767,
+      "step": 5717000
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.203790197840755e-05,
+      "loss": 1.1672,
+      "step": 5717500
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.203580201284699e-05,
+      "loss": 1.1223,
+      "step": 5718000
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2033702047286424e-05,
+      "loss": 1.1336,
+      "step": 5718500
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.203160208172586e-05,
+      "loss": 1.1758,
+      "step": 5719000
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.20295021161653e-05,
+      "loss": 1.1549,
+      "step": 5719500
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.202740215060473e-05,
+      "loss": 1.1436,
+      "step": 5720000
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2025302185044165e-05,
+      "loss": 1.1579,
+      "step": 5720500
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2023202219483605e-05,
+      "loss": 1.158,
+      "step": 5721000
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.202110225392303e-05,
+      "loss": 1.1989,
+      "step": 5721500
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2019002288362465e-05,
+      "loss": 1.1657,
+      "step": 5722000
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2016902322801905e-05,
+      "loss": 1.1528,
+      "step": 5722500
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.201480235724134e-05,
+      "loss": 1.1252,
+      "step": 5723000
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.20127065916119e-05,
+      "loss": 1.1798,
+      "step": 5723500
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.201060662605133e-05,
+      "loss": 1.1683,
+      "step": 5724000
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2008506660490766e-05,
+      "loss": 1.1565,
+      "step": 5724500
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2006410894861326e-05,
+      "loss": 1.1366,
+      "step": 5725000
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.200431092930076e-05,
+      "loss": 1.1843,
+      "step": 5725500
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.20022109637402e-05,
+      "loss": 1.1941,
+      "step": 5726000
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.200011099817963e-05,
+      "loss": 1.1581,
+      "step": 5726500
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.199801103261906e-05,
+      "loss": 1.1518,
+      "step": 5727000
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.19959110670585e-05,
+      "loss": 1.1554,
+      "step": 5727500
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.1993811101497934e-05,
+      "loss": 1.1484,
+      "step": 5728000
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.199171113593737e-05,
+      "loss": 1.1363,
+      "step": 5728500
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.198961117037681e-05,
+      "loss": 1.1542,
+      "step": 5729000
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.198751120481624e-05,
+      "loss": 1.1512,
+      "step": 5729500
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.1985411239255675e-05,
+      "loss": 1.156,
+      "step": 5730000
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.1983311273695115e-05,
+      "loss": 1.1595,
+      "step": 5730500
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.198121550806567e-05,
+      "loss": 1.1697,
+      "step": 5731000
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.19791155425051e-05,
+      "loss": 1.1599,
+      "step": 5731500
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.1977015576944535e-05,
+      "loss": 1.164,
+      "step": 5732000
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.1974915611383976e-05,
+      "loss": 1.1451,
+      "step": 5732500
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.197281984575453e-05,
+      "loss": 1.1526,
+      "step": 5733000
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.197072408012508e-05,
+      "loss": 1.1653,
+      "step": 5733500
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.1968624114564516e-05,
+      "loss": 1.1628,
+      "step": 5734000
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.1966524149003956e-05,
+      "loss": 1.1765,
+      "step": 5734500
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.196442418344339e-05,
+      "loss": 1.1748,
+      "step": 5735000
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.196232421788282e-05,
+      "loss": 1.1739,
+      "step": 5735500
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.1960224252322264e-05,
+      "loss": 1.1745,
+      "step": 5736000
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.19581242867617e-05,
+      "loss": 1.1649,
+      "step": 5736500
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.195602432120113e-05,
+      "loss": 1.1575,
+      "step": 5737000
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.1953928555571684e-05,
+      "loss": 1.1695,
+      "step": 5737500
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.1951828590011124e-05,
+      "loss": 1.1413,
+      "step": 5738000
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.194972862445056e-05,
+      "loss": 1.173,
+      "step": 5738500
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.194762865888999e-05,
+      "loss": 1.1741,
+      "step": 5739000
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.194552869332943e-05,
+      "loss": 1.1698,
+      "step": 5739500
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.1943428727768865e-05,
+      "loss": 1.1832,
+      "step": 5740000
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.19413287622083e-05,
+      "loss": 1.166,
+      "step": 5740500
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.193922879664774e-05,
+      "loss": 1.1473,
+      "step": 5741000
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.193713723094941e-05,
+      "loss": 1.1404,
+      "step": 5741500
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.1935037265388846e-05,
+      "loss": 1.1378,
+      "step": 5742000
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.193293729982828e-05,
+      "loss": 1.1562,
+      "step": 5742500
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.193083733426772e-05,
+      "loss": 1.1554,
+      "step": 5743000
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.192873736870715e-05,
+      "loss": 1.1599,
+      "step": 5743500
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.1926637403146586e-05,
+      "loss": 1.1747,
+      "step": 5744000
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.192454163751714e-05,
+      "loss": 1.1836,
+      "step": 5744500
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.192244167195658e-05,
+      "loss": 1.1401,
+      "step": 5745000
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.1920341706396014e-05,
+      "loss": 1.185,
+      "step": 5745500
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.191824174083545e-05,
+      "loss": 1.1379,
+      "step": 5746000
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.191614177527489e-05,
+      "loss": 1.16,
+      "step": 5746500
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.191404180971432e-05,
+      "loss": 1.1609,
+      "step": 5747000
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.191194184415376e-05,
+      "loss": 1.1223,
+      "step": 5747500
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.1909841878593194e-05,
+      "loss": 1.1856,
+      "step": 5748000
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.190774611296375e-05,
+      "loss": 1.1863,
+      "step": 5748500
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.190564614740318e-05,
+      "loss": 1.1641,
+      "step": 5749000
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.190354618184262e-05,
+      "loss": 1.159,
+      "step": 5749500
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.1901446216282055e-05,
+      "loss": 1.16,
+      "step": 5750000
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.189935045065261e-05,
+      "loss": 1.1428,
+      "step": 5750500
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.189725048509204e-05,
+      "loss": 1.1544,
+      "step": 5751000
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.189515051953148e-05,
+      "loss": 1.1539,
+      "step": 5751500
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.1893050553970916e-05,
+      "loss": 1.1577,
+      "step": 5752000
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.189095478834147e-05,
+      "loss": 1.1709,
+      "step": 5752500
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.188885902271202e-05,
+      "loss": 1.1955,
+      "step": 5753000
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.188675905715146e-05,
+      "loss": 1.1875,
+      "step": 5753500
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.18846590915909e-05,
+      "loss": 1.1712,
+      "step": 5754000
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.188255912603033e-05,
+      "loss": 1.1749,
+      "step": 5754500
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.188045916046977e-05,
+      "loss": 1.1689,
+      "step": 5755000
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.1878359194909204e-05,
+      "loss": 1.1436,
+      "step": 5755500
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.187625922934864e-05,
+      "loss": 1.1434,
+      "step": 5756000
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.187415926378808e-05,
+      "loss": 1.1472,
+      "step": 5756500
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.187206349815863e-05,
+      "loss": 1.1759,
+      "step": 5757000
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.1869963532598065e-05,
+      "loss": 1.1698,
+      "step": 5757500
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.18678635670375e-05,
+      "loss": 1.1382,
+      "step": 5758000
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.186576360147694e-05,
+      "loss": 1.1299,
+      "step": 5758500
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.186366363591637e-05,
+      "loss": 1.1768,
+      "step": 5759000
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.1861563670355805e-05,
+      "loss": 1.1664,
+      "step": 5759500
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.1859463704795245e-05,
+      "loss": 1.1634,
+      "step": 5760000
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.185736373923467e-05,
+      "loss": 1.1438,
+      "step": 5760500
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.185526797360523e-05,
+      "loss": 1.1761,
+      "step": 5761000
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.1853172207975786e-05,
+      "loss": 1.1601,
+      "step": 5761500
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.1851072242415226e-05,
+      "loss": 1.1417,
+      "step": 5762000
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.184897227685466e-05,
+      "loss": 1.1427,
+      "step": 5762500
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.184687231129409e-05,
+      "loss": 1.1951,
+      "step": 5763000
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.1844772345733533e-05,
+      "loss": 1.2043,
+      "step": 5763500
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.184267238017297e-05,
+      "loss": 1.1861,
+      "step": 5764000
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.18405724146124e-05,
+      "loss": 1.1932,
+      "step": 5764500
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.1838472449051834e-05,
+      "loss": 1.1597,
+      "step": 5765000
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.1836376683422394e-05,
+      "loss": 1.1548,
+      "step": 5765500
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.183427671786183e-05,
+      "loss": 1.1686,
+      "step": 5766000
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.183217675230126e-05,
+      "loss": 1.169,
+      "step": 5766500
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.18300767867407e-05,
+      "loss": 1.1618,
+      "step": 5767000
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.182797682118013e-05,
+      "loss": 1.1584,
+      "step": 5767500
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.182587685561957e-05,
+      "loss": 1.1554,
+      "step": 5768000
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.1823776890059e-05,
+      "loss": 1.1216,
+      "step": 5768500
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.1821676924498435e-05,
+      "loss": 1.1736,
+      "step": 5769000
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.1819581158868996e-05,
+      "loss": 1.1407,
+      "step": 5769500
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.181748119330843e-05,
+      "loss": 1.1424,
+      "step": 5770000
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.181538122774786e-05,
+      "loss": 1.1856,
+      "step": 5770500
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.1813281262187296e-05,
+      "loss": 1.1468,
+      "step": 5771000
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.1811185496557856e-05,
+      "loss": 1.1634,
+      "step": 5771500
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.1809085530997296e-05,
+      "loss": 1.181,
+      "step": 5772000
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.180698556543672e-05,
+      "loss": 1.1913,
+      "step": 5772500
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.180488559987616e-05,
+      "loss": 1.1863,
+      "step": 5773000
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.180278983424672e-05,
+      "loss": 1.1734,
+      "step": 5773500
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.180068986868616e-05,
+      "loss": 1.141,
+      "step": 5774000
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.179859410305671e-05,
+      "loss": 1.1709,
+      "step": 5774500
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.1796494137496144e-05,
+      "loss": 1.1674,
+      "step": 5775000
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.1794394171935584e-05,
+      "loss": 1.1371,
+      "step": 5775500
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.179229420637502e-05,
+      "loss": 1.1782,
+      "step": 5776000
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.179019424081445e-05,
+      "loss": 1.1681,
+      "step": 5776500
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.1788094275253885e-05,
+      "loss": 1.1641,
+      "step": 5777000
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.1785998509624445e-05,
+      "loss": 1.1706,
+      "step": 5777500
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.178389854406388e-05,
+      "loss": 1.1547,
+      "step": 5778000
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.178179857850331e-05,
+      "loss": 1.1442,
+      "step": 5778500
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.177969861294275e-05,
+      "loss": 1.1858,
+      "step": 5779000
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.177759864738218e-05,
+      "loss": 1.183,
+      "step": 5779500
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.177550288175274e-05,
+      "loss": 1.1728,
+      "step": 5780000
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.177340291619217e-05,
+      "loss": 1.1735,
+      "step": 5780500
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.177130295063161e-05,
+      "loss": 1.1416,
+      "step": 5781000
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.1769202985071047e-05,
+      "loss": 1.1453,
+      "step": 5781500
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.176710301951048e-05,
+      "loss": 1.186,
+      "step": 5782000
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.176500725388104e-05,
+      "loss": 1.1518,
+      "step": 5782500
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.1762911488251594e-05,
+      "loss": 1.1559,
+      "step": 5783000
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.176081152269103e-05,
+      "loss": 1.1769,
+      "step": 5783500
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.175871155713046e-05,
+      "loss": 1.1566,
+      "step": 5784000
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.17566115915699e-05,
+      "loss": 1.1915,
+      "step": 5784500
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.1754511626009335e-05,
+      "loss": 1.1506,
+      "step": 5785000
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.175241166044877e-05,
+      "loss": 1.1684,
+      "step": 5785500
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.175031169488821e-05,
+      "loss": 1.1573,
+      "step": 5786000
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.1748211729327635e-05,
+      "loss": 1.1638,
+      "step": 5786500
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.174611176376707e-05,
+      "loss": 1.1645,
+      "step": 5787000
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.174401599813763e-05,
+      "loss": 1.1609,
+      "step": 5787500
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.174191603257707e-05,
+      "loss": 1.1569,
+      "step": 5788000
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.17398160670165e-05,
+      "loss": 1.1569,
+      "step": 5788500
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.1737716101455936e-05,
+      "loss": 1.1638,
+      "step": 5789000
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.1735620335826496e-05,
+      "loss": 1.1536,
+      "step": 5789500
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.173352037026593e-05,
+      "loss": 1.1527,
+      "step": 5790000
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.173142040470536e-05,
+      "loss": 1.1528,
+      "step": 5790500
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.1729320439144803e-05,
+      "loss": 1.1453,
+      "step": 5791000
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.172722047358423e-05,
+      "loss": 1.1826,
+      "step": 5791500
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.1725120508023664e-05,
+      "loss": 1.1846,
+      "step": 5792000
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.1723020542463104e-05,
+      "loss": 1.1534,
+      "step": 5792500
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.1720924776833664e-05,
+      "loss": 1.1646,
+      "step": 5793000
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.17188248112731e-05,
+      "loss": 1.1722,
+      "step": 5793500
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.1716724845712524e-05,
+      "loss": 1.1695,
+      "step": 5794000
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.1714624880151965e-05,
+      "loss": 1.1833,
+      "step": 5794500
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.17125249145914e-05,
+      "loss": 1.1601,
+      "step": 5795000
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.171042494903083e-05,
+      "loss": 1.1613,
+      "step": 5795500
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.170833338333251e-05,
+      "loss": 1.1725,
+      "step": 5796000
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.170623341777195e-05,
+      "loss": 1.1689,
+      "step": 5796500
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.1704133452211386e-05,
+      "loss": 1.152,
+      "step": 5797000
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.170203348665082e-05,
+      "loss": 1.1679,
+      "step": 5797500
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.169993352109026e-05,
+      "loss": 1.1857,
+      "step": 5798000
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.1697833555529686e-05,
+      "loss": 1.1694,
+      "step": 5798500
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.169573358996912e-05,
+      "loss": 1.1209,
+      "step": 5799000
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.169363362440856e-05,
+      "loss": 1.157,
+      "step": 5799500
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.169153365884799e-05,
+      "loss": 1.1589,
+      "step": 5800000
+    },
+    {
+      "epoch": 3.48,
+      "eval_loss": 1.126969337463379,
+      "eval_runtime": 1104.9848,
+      "eval_samples_per_second": 476.676,
+      "eval_steps_per_second": 79.446,
+      "step": 5800000
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.1689437893218553e-05,
+      "loss": 1.1474,
+      "step": 5800500
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.168733792765798e-05,
+      "loss": 1.1733,
+      "step": 5801000
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.168523796209742e-05,
+      "loss": 1.1411,
+      "step": 5801500
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.1683137996536854e-05,
+      "loss": 1.1551,
+      "step": 5802000
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.168103803097629e-05,
+      "loss": 1.1822,
+      "step": 5802500
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.167893806541573e-05,
+      "loss": 1.151,
+      "step": 5803000
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.167683809985516e-05,
+      "loss": 1.1747,
+      "step": 5803500
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.1674738134294594e-05,
+      "loss": 1.136,
+      "step": 5804000
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.1672638168734035e-05,
+      "loss": 1.1617,
+      "step": 5804500
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.167054240310459e-05,
+      "loss": 1.1403,
+      "step": 5805000
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.166844243754402e-05,
+      "loss": 1.1473,
+      "step": 5805500
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.166634247198346e-05,
+      "loss": 1.155,
+      "step": 5806000
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.1664242506422895e-05,
+      "loss": 1.1345,
+      "step": 5806500
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.166214674079345e-05,
+      "loss": 1.176,
+      "step": 5807000
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.166004677523288e-05,
+      "loss": 1.179,
+      "step": 5807500
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.165794680967232e-05,
+      "loss": 1.1442,
+      "step": 5808000
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.1655851044042876e-05,
+      "loss": 1.1426,
+      "step": 5808500
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.165375107848231e-05,
+      "loss": 1.1678,
+      "step": 5809000
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.165165531285287e-05,
+      "loss": 1.1509,
+      "step": 5809500
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.1649555347292304e-05,
+      "loss": 1.1512,
+      "step": 5810000
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.1647459581662864e-05,
+      "loss": 1.1695,
+      "step": 5810500
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.16453596161023e-05,
+      "loss": 1.1764,
+      "step": 5811000
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.164325965054173e-05,
+      "loss": 1.1493,
+      "step": 5811500
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.164115968498117e-05,
+      "loss": 1.1406,
+      "step": 5812000
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.1639059719420604e-05,
+      "loss": 1.1742,
+      "step": 5812500
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.163696395379116e-05,
+      "loss": 1.1853,
+      "step": 5813000
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.163486398823059e-05,
+      "loss": 1.1598,
+      "step": 5813500
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.163276402267003e-05,
+      "loss": 1.1819,
+      "step": 5814000
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.1630664057109465e-05,
+      "loss": 1.1759,
+      "step": 5814500
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.16285640915489e-05,
+      "loss": 1.1665,
+      "step": 5815000
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.162646412598833e-05,
+      "loss": 1.1531,
+      "step": 5815500
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.1624364160427766e-05,
+      "loss": 1.1687,
+      "step": 5816000
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.16222641948672e-05,
+      "loss": 1.1561,
+      "step": 5816500
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.162016422930664e-05,
+      "loss": 1.1491,
+      "step": 5817000
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.161806426374607e-05,
+      "loss": 1.1724,
+      "step": 5817500
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.1615964298185506e-05,
+      "loss": 1.1631,
+      "step": 5818000
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.1613864332624946e-05,
+      "loss": 1.1127,
+      "step": 5818500
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.161176436706438e-05,
+      "loss": 1.163,
+      "step": 5819000
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.1609668601434933e-05,
+      "loss": 1.1589,
+      "step": 5819500
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.1607568635874374e-05,
+      "loss": 1.1583,
+      "step": 5820000
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.160546867031381e-05,
+      "loss": 1.1588,
+      "step": 5820500
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.160336870475324e-05,
+      "loss": 1.1696,
+      "step": 5821000
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.160126873919268e-05,
+      "loss": 1.1645,
+      "step": 5821500
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.1599168773632114e-05,
+      "loss": 1.1631,
+      "step": 5822000
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.159706880807155e-05,
+      "loss": 1.1635,
+      "step": 5822500
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.159496884251098e-05,
+      "loss": 1.1264,
+      "step": 5823000
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.159287307688154e-05,
+      "loss": 1.1745,
+      "step": 5823500
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.1590773111320975e-05,
+      "loss": 1.1581,
+      "step": 5824000
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.158867314576041e-05,
+      "loss": 1.1759,
+      "step": 5824500
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.158657318019985e-05,
+      "loss": 1.1641,
+      "step": 5825000
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.15844774145704e-05,
+      "loss": 1.1687,
+      "step": 5825500
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.1582377449009836e-05,
+      "loss": 1.1481,
+      "step": 5826000
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.1580277483449276e-05,
+      "loss": 1.1726,
+      "step": 5826500
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.157818171781983e-05,
+      "loss": 1.1738,
+      "step": 5827000
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.157608175225926e-05,
+      "loss": 1.1579,
+      "step": 5827500
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.1573981786698697e-05,
+      "loss": 1.1702,
+      "step": 5828000
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.157188182113814e-05,
+      "loss": 1.1387,
+      "step": 5828500
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.156978185557757e-05,
+      "loss": 1.1632,
+      "step": 5829000
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.1567681890017004e-05,
+      "loss": 1.1309,
+      "step": 5829500
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.156559032431868e-05,
+      "loss": 1.178,
+      "step": 5830000
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.156349035875811e-05,
+      "loss": 1.1572,
+      "step": 5830500
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.156139039319755e-05,
+      "loss": 1.1894,
+      "step": 5831000
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.155929462756811e-05,
+      "loss": 1.1866,
+      "step": 5831500
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.155719466200754e-05,
+      "loss": 1.1443,
+      "step": 5832000
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.155509469644698e-05,
+      "loss": 1.1382,
+      "step": 5832500
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.155299473088641e-05,
+      "loss": 1.1823,
+      "step": 5833000
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.155089896525697e-05,
+      "loss": 1.1531,
+      "step": 5833500
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.1548798999696406e-05,
+      "loss": 1.1411,
+      "step": 5834000
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.154669903413584e-05,
+      "loss": 1.1435,
+      "step": 5834500
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.154459906857527e-05,
+      "loss": 1.148,
+      "step": 5835000
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.1542499103014706e-05,
+      "loss": 1.1628,
+      "step": 5835500
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.1540399137454146e-05,
+      "loss": 1.1548,
+      "step": 5836000
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.153829917189358e-05,
+      "loss": 1.1289,
+      "step": 5836500
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.153619920633301e-05,
+      "loss": 1.1408,
+      "step": 5837000
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.153409924077245e-05,
+      "loss": 1.1604,
+      "step": 5837500
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.153199927521189e-05,
+      "loss": 1.1388,
+      "step": 5838000
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.152989930965132e-05,
+      "loss": 1.1554,
+      "step": 5838500
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.152779934409076e-05,
+      "loss": 1.1423,
+      "step": 5839000
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.1525703578461314e-05,
+      "loss": 1.1777,
+      "step": 5839500
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.152360361290075e-05,
+      "loss": 1.148,
+      "step": 5840000
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.152150364734019e-05,
+      "loss": 1.1437,
+      "step": 5840500
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.151940368177962e-05,
+      "loss": 1.1623,
+      "step": 5841000
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.1517303716219055e-05,
+      "loss": 1.1405,
+      "step": 5841500
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.151520795058961e-05,
+      "loss": 1.1685,
+      "step": 5842000
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.151310798502905e-05,
+      "loss": 1.1948,
+      "step": 5842500
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.151100801946848e-05,
+      "loss": 1.1379,
+      "step": 5843000
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.1508908053907915e-05,
+      "loss": 1.1651,
+      "step": 5843500
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.1506808088347356e-05,
+      "loss": 1.1415,
+      "step": 5844000
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.150471232271791e-05,
+      "loss": 1.1794,
+      "step": 5844500
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.150261235715734e-05,
+      "loss": 1.1576,
+      "step": 5845000
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.1500512391596776e-05,
+      "loss": 1.1421,
+      "step": 5845500
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.1498412426036216e-05,
+      "loss": 1.1453,
+      "step": 5846000
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.149631246047565e-05,
+      "loss": 1.1576,
+      "step": 5846500
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.149421249491508e-05,
+      "loss": 1.1578,
+      "step": 5847000
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.149211252935452e-05,
+      "loss": 1.1643,
+      "step": 5847500
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.149001256379395e-05,
+      "loss": 1.1605,
+      "step": 5848000
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.148791679816451e-05,
+      "loss": 1.157,
+      "step": 5848500
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.148581683260395e-05,
+      "loss": 1.1816,
+      "step": 5849000
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.1483721066974504e-05,
+      "loss": 1.1649,
+      "step": 5849500
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.148162110141394e-05,
+      "loss": 1.208,
+      "step": 5850000
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.147952113585337e-05,
+      "loss": 1.1497,
+      "step": 5850500
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.1477425370223925e-05,
+      "loss": 1.1664,
+      "step": 5851000
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.1475325404663365e-05,
+      "loss": 1.141,
+      "step": 5851500
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.14732254391028e-05,
+      "loss": 1.1654,
+      "step": 5852000
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.147112547354223e-05,
+      "loss": 1.1721,
+      "step": 5852500
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.146902550798167e-05,
+      "loss": 1.1481,
+      "step": 5853000
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.1466925542421106e-05,
+      "loss": 1.153,
+      "step": 5853500
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.146482557686054e-05,
+      "loss": 1.185,
+      "step": 5854000
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.14627298112311e-05,
+      "loss": 1.1701,
+      "step": 5854500
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.146062984567053e-05,
+      "loss": 1.1651,
+      "step": 5855000
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.1458529880109966e-05,
+      "loss": 1.1614,
+      "step": 5855500
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.145642991454941e-05,
+      "loss": 1.166,
+      "step": 5856000
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.145432994898883e-05,
+      "loss": 1.1691,
+      "step": 5856500
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.145222998342827e-05,
+      "loss": 1.1458,
+      "step": 5857000
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.145013001786771e-05,
+      "loss": 1.1614,
+      "step": 5857500
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.144803005230714e-05,
+      "loss": 1.1673,
+      "step": 5858000
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.1445930086746574e-05,
+      "loss": 1.1499,
+      "step": 5858500
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.1443830121186014e-05,
+      "loss": 1.1274,
+      "step": 5859000
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.144173015562545e-05,
+      "loss": 1.1846,
+      "step": 5859500
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.143963019006488e-05,
+      "loss": 1.1797,
+      "step": 5860000
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.143753862436656e-05,
+      "loss": 1.1241,
+      "step": 5860500
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.1435438658805995e-05,
+      "loss": 1.1764,
+      "step": 5861000
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.1433342893176555e-05,
+      "loss": 1.1689,
+      "step": 5861500
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.143124292761599e-05,
+      "loss": 1.1577,
+      "step": 5862000
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.142914296205542e-05,
+      "loss": 1.1434,
+      "step": 5862500
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.142704299649486e-05,
+      "loss": 1.1596,
+      "step": 5863000
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.142494303093429e-05,
+      "loss": 1.1496,
+      "step": 5863500
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.142284306537372e-05,
+      "loss": 1.1566,
+      "step": 5864000
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.142074309981316e-05,
+      "loss": 1.1699,
+      "step": 5864500
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.1418643134252596e-05,
+      "loss": 1.1789,
+      "step": 5865000
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.141654736862316e-05,
+      "loss": 1.1931,
+      "step": 5865500
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.1414447403062583e-05,
+      "loss": 1.1471,
+      "step": 5866000
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.1412347437502024e-05,
+      "loss": 1.1312,
+      "step": 5866500
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.141024747194146e-05,
+      "loss": 1.1229,
+      "step": 5867000
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.140814750638089e-05,
+      "loss": 1.1285,
+      "step": 5867500
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.140604754082033e-05,
+      "loss": 1.1751,
+      "step": 5868000
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.1403951775190884e-05,
+      "loss": 1.1562,
+      "step": 5868500
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.140185180963032e-05,
+      "loss": 1.1648,
+      "step": 5869000
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.139975184406976e-05,
+      "loss": 1.1274,
+      "step": 5869500
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.139765187850919e-05,
+      "loss": 1.1679,
+      "step": 5870000
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.1395551912948625e-05,
+      "loss": 1.1176,
+      "step": 5870500
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.1393451947388065e-05,
+      "loss": 1.1843,
+      "step": 5871000
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.13913519818275e-05,
+      "loss": 1.1483,
+      "step": 5871500
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.138925621619805e-05,
+      "loss": 1.1767,
+      "step": 5872000
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.1387156250637486e-05,
+      "loss": 1.1732,
+      "step": 5872500
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.1385056285076926e-05,
+      "loss": 1.1732,
+      "step": 5873000
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.138295631951636e-05,
+      "loss": 1.16,
+      "step": 5873500
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.138085635395579e-05,
+      "loss": 1.1241,
+      "step": 5874000
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.137875638839523e-05,
+      "loss": 1.1589,
+      "step": 5874500
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.1376656422834667e-05,
+      "loss": 1.1451,
+      "step": 5875000
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.13745564572741e-05,
+      "loss": 1.1634,
+      "step": 5875500
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.1372460691644654e-05,
+      "loss": 1.1576,
+      "step": 5876000
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.1370360726084094e-05,
+      "loss": 1.1629,
+      "step": 5876500
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.136826496045465e-05,
+      "loss": 1.1765,
+      "step": 5877000
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.136616499489408e-05,
+      "loss": 1.1559,
+      "step": 5877500
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.136406502933352e-05,
+      "loss": 1.169,
+      "step": 5878000
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.1361965063772955e-05,
+      "loss": 1.1637,
+      "step": 5878500
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.135986509821239e-05,
+      "loss": 1.1497,
+      "step": 5879000
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.135776513265183e-05,
+      "loss": 1.1297,
+      "step": 5879500
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.135566516709126e-05,
+      "loss": 1.1368,
+      "step": 5880000
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.1353565201530695e-05,
+      "loss": 1.1574,
+      "step": 5880500
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.135146523597013e-05,
+      "loss": 1.1638,
+      "step": 5881000
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.134936527040956e-05,
+      "loss": 1.1624,
+      "step": 5881500
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.1347265304848996e-05,
+      "loss": 1.1721,
+      "step": 5882000
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.1345165339288436e-05,
+      "loss": 1.1526,
+      "step": 5882500
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.1343069573658996e-05,
+      "loss": 1.1727,
+      "step": 5883000
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.134096960809842e-05,
+      "loss": 1.1446,
+      "step": 5883500
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.1338869642537856e-05,
+      "loss": 1.1466,
+      "step": 5884000
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.1336769676977297e-05,
+      "loss": 1.1409,
+      "step": 5884500
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.133467391134786e-05,
+      "loss": 1.1721,
+      "step": 5885000
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.133257394578729e-05,
+      "loss": 1.1691,
+      "step": 5885500
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.1330473980226724e-05,
+      "loss": 1.1641,
+      "step": 5886000
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.132837401466616e-05,
+      "loss": 1.1508,
+      "step": 5886500
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.132627404910559e-05,
+      "loss": 1.1638,
+      "step": 5887000
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.132417408354503e-05,
+      "loss": 1.16,
+      "step": 5887500
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.1322074117984464e-05,
+      "loss": 1.1597,
+      "step": 5888000
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.131997835235502e-05,
+      "loss": 1.146,
+      "step": 5888500
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.131787838679445e-05,
+      "loss": 1.1556,
+      "step": 5889000
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.131577842123389e-05,
+      "loss": 1.1483,
+      "step": 5889500
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.1313678455673325e-05,
+      "loss": 1.1471,
+      "step": 5890000
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.131157849011276e-05,
+      "loss": 1.1895,
+      "step": 5890500
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.13094785245522e-05,
+      "loss": 1.1553,
+      "step": 5891000
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.130738275892275e-05,
+      "loss": 1.1956,
+      "step": 5891500
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.1305282793362186e-05,
+      "loss": 1.1658,
+      "step": 5892000
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.1303182827801626e-05,
+      "loss": 1.1574,
+      "step": 5892500
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.130108286224106e-05,
+      "loss": 1.1656,
+      "step": 5893000
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.129898709661161e-05,
+      "loss": 1.148,
+      "step": 5893500
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.1296887131051047e-05,
+      "loss": 1.1602,
+      "step": 5894000
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.129479136542161e-05,
+      "loss": 1.1267,
+      "step": 5894500
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.129269139986105e-05,
+      "loss": 1.1904,
+      "step": 5895000
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.1290591434300474e-05,
+      "loss": 1.137,
+      "step": 5895500
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.128849146873991e-05,
+      "loss": 1.158,
+      "step": 5896000
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.128639150317935e-05,
+      "loss": 1.1577,
+      "step": 5896500
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.128429153761878e-05,
+      "loss": 1.1484,
+      "step": 5897000
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.1282191572058214e-05,
+      "loss": 1.1578,
+      "step": 5897500
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.1280091606497655e-05,
+      "loss": 1.1826,
+      "step": 5898000
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.127799164093709e-05,
+      "loss": 1.1629,
+      "step": 5898500
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.127589167537652e-05,
+      "loss": 1.1477,
+      "step": 5899000
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.127379590974708e-05,
+      "loss": 1.1423,
+      "step": 5899500
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.1271695944186515e-05,
+      "loss": 1.2008,
+      "step": 5900000
+    },
+    {
+      "epoch": 3.54,
+      "eval_loss": 1.1256732940673828,
+      "eval_runtime": 1107.8778,
+      "eval_samples_per_second": 475.431,
+      "eval_steps_per_second": 79.239,
+      "step": 5900000
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.126959597862595e-05,
+      "loss": 1.1658,
+      "step": 5900500
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.126749601306539e-05,
+      "loss": 1.1671,
+      "step": 5901000
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.126539604750482e-05,
+      "loss": 1.1394,
+      "step": 5901500
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.1263296081944256e-05,
+      "loss": 1.1411,
+      "step": 5902000
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.126120031631481e-05,
+      "loss": 1.1527,
+      "step": 5902500
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.125910035075425e-05,
+      "loss": 1.102,
+      "step": 5903000
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.125700038519368e-05,
+      "loss": 1.1446,
+      "step": 5903500
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.125490041963312e-05,
+      "loss": 1.1394,
+      "step": 5904000
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.125280465400367e-05,
+      "loss": 1.1653,
+      "step": 5904500
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.125070468844311e-05,
+      "loss": 1.1468,
+      "step": 5905000
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.1248604722882544e-05,
+      "loss": 1.1894,
+      "step": 5905500
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.124650475732198e-05,
+      "loss": 1.1312,
+      "step": 5906000
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.124440479176142e-05,
+      "loss": 1.1552,
+      "step": 5906500
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.124230482620085e-05,
+      "loss": 1.1743,
+      "step": 5907000
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.1240204860640285e-05,
+      "loss": 1.1693,
+      "step": 5907500
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.123810489507972e-05,
+      "loss": 1.1624,
+      "step": 5908000
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.123600912945028e-05,
+      "loss": 1.1454,
+      "step": 5908500
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.123390916388971e-05,
+      "loss": 1.1572,
+      "step": 5909000
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.123180919832915e-05,
+      "loss": 1.1449,
+      "step": 5909500
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.1229713432699706e-05,
+      "loss": 1.1558,
+      "step": 5910000
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.122761346713914e-05,
+      "loss": 1.1809,
+      "step": 5910500
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.122551350157857e-05,
+      "loss": 1.1526,
+      "step": 5911000
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.122341353601801e-05,
+      "loss": 1.1782,
+      "step": 5911500
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.1221317770388566e-05,
+      "loss": 1.171,
+      "step": 5912000
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.1219217804828e-05,
+      "loss": 1.149,
+      "step": 5912500
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.1217117839267433e-05,
+      "loss": 1.1716,
+      "step": 5913000
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.1215022073637994e-05,
+      "loss": 1.1784,
+      "step": 5913500
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.121292210807743e-05,
+      "loss": 1.1595,
+      "step": 5914000
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.121082214251686e-05,
+      "loss": 1.1569,
+      "step": 5914500
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.12087221769563e-05,
+      "loss": 1.1631,
+      "step": 5915000
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.1206622211395734e-05,
+      "loss": 1.1243,
+      "step": 5915500
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.120452224583517e-05,
+      "loss": 1.153,
+      "step": 5916000
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.120242228027461e-05,
+      "loss": 1.1717,
+      "step": 5916500
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.120032231471404e-05,
+      "loss": 1.1821,
+      "step": 5917000
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.119822234915347e-05,
+      "loss": 1.1428,
+      "step": 5917500
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.119612238359291e-05,
+      "loss": 1.1571,
+      "step": 5918000
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.119402661796347e-05,
+      "loss": 1.1713,
+      "step": 5918500
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.11919266524029e-05,
+      "loss": 1.1632,
+      "step": 5919000
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.1189826686842336e-05,
+      "loss": 1.1444,
+      "step": 5919500
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.118772672128177e-05,
+      "loss": 1.1484,
+      "step": 5920000
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.11856267557212e-05,
+      "loss": 1.1473,
+      "step": 5920500
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.1183526790160636e-05,
+      "loss": 1.1684,
+      "step": 5921000
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.1181426824600076e-05,
+      "loss": 1.1445,
+      "step": 5921500
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.117932685903951e-05,
+      "loss": 1.142,
+      "step": 5922000
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.117723109341006e-05,
+      "loss": 1.1717,
+      "step": 5922500
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.1175135327780624e-05,
+      "loss": 1.1472,
+      "step": 5923000
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.1173035362220064e-05,
+      "loss": 1.1487,
+      "step": 5923500
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.117093959659062e-05,
+      "loss": 1.1424,
+      "step": 5924000
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.116883963103005e-05,
+      "loss": 1.1648,
+      "step": 5924500
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.1166739665469484e-05,
+      "loss": 1.1606,
+      "step": 5925000
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.1164639699908925e-05,
+      "loss": 1.1984,
+      "step": 5925500
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.116253973434836e-05,
+      "loss": 1.1892,
+      "step": 5926000
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.116043976878779e-05,
+      "loss": 1.1595,
+      "step": 5926500
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.1158339803227225e-05,
+      "loss": 1.1622,
+      "step": 5927000
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.115623983766666e-05,
+      "loss": 1.1702,
+      "step": 5927500
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.115413987210609e-05,
+      "loss": 1.1708,
+      "step": 5928000
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.115203990654553e-05,
+      "loss": 1.1384,
+      "step": 5928500
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.1149939940984966e-05,
+      "loss": 1.1754,
+      "step": 5929000
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.11478399754244e-05,
+      "loss": 1.1443,
+      "step": 5929500
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.114574420979496e-05,
+      "loss": 1.174,
+      "step": 5930000
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.114364424423439e-05,
+      "loss": 1.148,
+      "step": 5930500
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.1141544278673826e-05,
+      "loss": 1.1494,
+      "step": 5931000
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.1139444313113267e-05,
+      "loss": 1.1627,
+      "step": 5931500
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.113734854748382e-05,
+      "loss": 1.1867,
+      "step": 5932000
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.113525278185438e-05,
+      "loss": 1.1328,
+      "step": 5932500
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.1133152816293814e-05,
+      "loss": 1.1611,
+      "step": 5933000
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.113105285073325e-05,
+      "loss": 1.1465,
+      "step": 5933500
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.112895288517269e-05,
+      "loss": 1.1771,
+      "step": 5934000
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.112685711954324e-05,
+      "loss": 1.1597,
+      "step": 5934500
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.1124757153982675e-05,
+      "loss": 1.1522,
+      "step": 5935000
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.112265718842211e-05,
+      "loss": 1.1517,
+      "step": 5935500
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.112055722286155e-05,
+      "loss": 1.1615,
+      "step": 5936000
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.1118457257300975e-05,
+      "loss": 1.1654,
+      "step": 5936500
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.1116357291740415e-05,
+      "loss": 1.1515,
+      "step": 5937000
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.1114261526110976e-05,
+      "loss": 1.1499,
+      "step": 5937500
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.111216156055041e-05,
+      "loss": 1.1658,
+      "step": 5938000
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.111006159498984e-05,
+      "loss": 1.1458,
+      "step": 5938500
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.1107961629429276e-05,
+      "loss": 1.1981,
+      "step": 5939000
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.110586166386871e-05,
+      "loss": 1.1786,
+      "step": 5939500
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.110376169830814e-05,
+      "loss": 1.1494,
+      "step": 5940000
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.110166173274758e-05,
+      "loss": 1.1672,
+      "step": 5940500
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.1099565967118144e-05,
+      "loss": 1.1746,
+      "step": 5941000
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.109746600155757e-05,
+      "loss": 1.1339,
+      "step": 5941500
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.1095366035997004e-05,
+      "loss": 1.1534,
+      "step": 5942000
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.1093266070436444e-05,
+      "loss": 1.1508,
+      "step": 5942500
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.1091170304807004e-05,
+      "loss": 1.1381,
+      "step": 5943000
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.108907033924644e-05,
+      "loss": 1.1895,
+      "step": 5943500
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.108697037368587e-05,
+      "loss": 1.1562,
+      "step": 5944000
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.1084870408125305e-05,
+      "loss": 1.1579,
+      "step": 5944500
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.1082774642495865e-05,
+      "loss": 1.1596,
+      "step": 5945000
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.10806746769353e-05,
+      "loss": 1.1483,
+      "step": 5945500
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.107857471137473e-05,
+      "loss": 1.1724,
+      "step": 5946000
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.1076474745814165e-05,
+      "loss": 1.1667,
+      "step": 5946500
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.10743747802536e-05,
+      "loss": 1.1498,
+      "step": 5947000
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.107227481469304e-05,
+      "loss": 1.1598,
+      "step": 5947500
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.107017484913247e-05,
+      "loss": 1.1706,
+      "step": 5948000
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.1068074883571906e-05,
+      "loss": 1.1771,
+      "step": 5948500
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.1065974918011346e-05,
+      "loss": 1.161,
+      "step": 5949000
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.106387495245078e-05,
+      "loss": 1.1309,
+      "step": 5949500
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.106177498689021e-05,
+      "loss": 1.1689,
+      "step": 5950000
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.1059675021329653e-05,
+      "loss": 1.1624,
+      "step": 5950500
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.105757925570021e-05,
+      "loss": 1.1674,
+      "step": 5951000
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.105547929013964e-05,
+      "loss": 1.1829,
+      "step": 5951500
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.105337932457908e-05,
+      "loss": 1.1584,
+      "step": 5952000
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.1051279359018514e-05,
+      "loss": 1.1646,
+      "step": 5952500
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.104917939345795e-05,
+      "loss": 1.1595,
+      "step": 5953000
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.104707942789739e-05,
+      "loss": 1.1445,
+      "step": 5953500
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.1044979462336815e-05,
+      "loss": 1.1636,
+      "step": 5954000
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.1042883696707375e-05,
+      "loss": 1.1679,
+      "step": 5954500
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.104078373114681e-05,
+      "loss": 1.1605,
+      "step": 5955000
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.103868376558625e-05,
+      "loss": 1.1737,
+      "step": 5955500
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.103658380002568e-05,
+      "loss": 1.1143,
+      "step": 5956000
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.103448383446511e-05,
+      "loss": 1.1539,
+      "step": 5956500
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.103238386890455e-05,
+      "loss": 1.1602,
+      "step": 5957000
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.103028390334398e-05,
+      "loss": 1.1507,
+      "step": 5957500
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.1028183937783416e-05,
+      "loss": 1.1976,
+      "step": 5958000
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.1026088172153976e-05,
+      "loss": 1.1228,
+      "step": 5958500
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.1023992406524537e-05,
+      "loss": 1.1585,
+      "step": 5959000
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.102189244096397e-05,
+      "loss": 1.1258,
+      "step": 5959500
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.1019792475403403e-05,
+      "loss": 1.1572,
+      "step": 5960000
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.1017692509842844e-05,
+      "loss": 1.1701,
+      "step": 5960500
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.101559254428228e-05,
+      "loss": 1.1691,
+      "step": 5961000
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.1013492578721704e-05,
+      "loss": 1.1618,
+      "step": 5961500
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.1011396813092264e-05,
+      "loss": 1.1777,
+      "step": 5962000
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.1009296847531704e-05,
+      "loss": 1.1518,
+      "step": 5962500
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.100719688197114e-05,
+      "loss": 1.166,
+      "step": 5963000
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.1005096916410565e-05,
+      "loss": 1.1649,
+      "step": 5963500
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.1002996950850005e-05,
+      "loss": 1.166,
+      "step": 5964000
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.100089698528944e-05,
+      "loss": 1.1495,
+      "step": 5964500
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.099879701972887e-05,
+      "loss": 1.1867,
+      "step": 5965000
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.099670125409943e-05,
+      "loss": 1.1511,
+      "step": 5965500
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.0994601288538866e-05,
+      "loss": 1.1707,
+      "step": 5966000
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.09925013229783e-05,
+      "loss": 1.1873,
+      "step": 5966500
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.099040135741774e-05,
+      "loss": 1.1604,
+      "step": 5967000
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.098830139185717e-05,
+      "loss": 1.1714,
+      "step": 5967500
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.0986201426296606e-05,
+      "loss": 1.1375,
+      "step": 5968000
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.0984101460736046e-05,
+      "loss": 1.1387,
+      "step": 5968500
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.09820056951066e-05,
+      "loss": 1.1694,
+      "step": 5969000
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.0979905729546033e-05,
+      "loss": 1.1235,
+      "step": 5969500
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.097780576398547e-05,
+      "loss": 1.1716,
+      "step": 5970000
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.097570579842491e-05,
+      "loss": 1.153,
+      "step": 5970500
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.097360583286434e-05,
+      "loss": 1.1445,
+      "step": 5971000
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.0971505867303774e-05,
+      "loss": 1.1352,
+      "step": 5971500
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.0969405901743214e-05,
+      "loss": 1.1645,
+      "step": 5972000
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.096730593618265e-05,
+      "loss": 1.1357,
+      "step": 5972500
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.09652101705532e-05,
+      "loss": 1.1388,
+      "step": 5973000
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.0963110204992635e-05,
+      "loss": 1.1792,
+      "step": 5973500
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.0961010239432075e-05,
+      "loss": 1.1273,
+      "step": 5974000
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.095891027387151e-05,
+      "loss": 1.1706,
+      "step": 5974500
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.095681450824206e-05,
+      "loss": 1.1705,
+      "step": 5975000
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.0954718742612616e-05,
+      "loss": 1.1275,
+      "step": 5975500
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.0952618777052056e-05,
+      "loss": 1.1598,
+      "step": 5976000
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.095051881149149e-05,
+      "loss": 1.1587,
+      "step": 5976500
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.094841884593092e-05,
+      "loss": 1.1657,
+      "step": 5977000
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.094632308030148e-05,
+      "loss": 1.1642,
+      "step": 5977500
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.0944223114740917e-05,
+      "loss": 1.1787,
+      "step": 5978000
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.094212314918035e-05,
+      "loss": 1.1465,
+      "step": 5978500
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.0940023183619783e-05,
+      "loss": 1.1526,
+      "step": 5979000
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.0937927417990344e-05,
+      "loss": 1.1358,
+      "step": 5979500
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.0935827452429784e-05,
+      "loss": 1.1675,
+      "step": 5980000
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.093372748686921e-05,
+      "loss": 1.1751,
+      "step": 5980500
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.093162752130865e-05,
+      "loss": 1.1717,
+      "step": 5981000
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.0929527555748084e-05,
+      "loss": 1.154,
+      "step": 5981500
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.092742759018752e-05,
+      "loss": 1.1558,
+      "step": 5982000
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.092532762462696e-05,
+      "loss": 1.1396,
+      "step": 5982500
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.092322765906639e-05,
+      "loss": 1.1186,
+      "step": 5983000
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.0921131893436945e-05,
+      "loss": 1.164,
+      "step": 5983500
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.091903192787638e-05,
+      "loss": 1.162,
+      "step": 5984000
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.091693196231582e-05,
+      "loss": 1.1591,
+      "step": 5984500
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.091483199675525e-05,
+      "loss": 1.1537,
+      "step": 5985000
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.091274043105693e-05,
+      "loss": 1.1672,
+      "step": 5985500
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.0910640465496366e-05,
+      "loss": 1.1763,
+      "step": 5986000
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.09085404999358e-05,
+      "loss": 1.1476,
+      "step": 5986500
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.090644053437524e-05,
+      "loss": 1.1415,
+      "step": 5987000
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.090434056881467e-05,
+      "loss": 1.1441,
+      "step": 5987500
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.090224480318523e-05,
+      "loss": 1.1301,
+      "step": 5988000
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.090014483762467e-05,
+      "loss": 1.1673,
+      "step": 5988500
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.08980448720641e-05,
+      "loss": 1.1652,
+      "step": 5989000
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.0895949106434654e-05,
+      "loss": 1.1336,
+      "step": 5989500
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.089384914087409e-05,
+      "loss": 1.1357,
+      "step": 5990000
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.089174917531353e-05,
+      "loss": 1.172,
+      "step": 5990500
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.088964920975296e-05,
+      "loss": 1.1875,
+      "step": 5991000
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.0887549244192395e-05,
+      "loss": 1.1785,
+      "step": 5991500
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.0885449278631835e-05,
+      "loss": 1.1766,
+      "step": 5992000
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.088334931307126e-05,
+      "loss": 1.1447,
+      "step": 5992500
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.0881249347510695e-05,
+      "loss": 1.1534,
+      "step": 5993000
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.0879149381950135e-05,
+      "loss": 1.1705,
+      "step": 5993500
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.087704941638957e-05,
+      "loss": 1.1489,
+      "step": 5994000
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.0874949450829e-05,
+      "loss": 1.1322,
+      "step": 5994500
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.087284948526844e-05,
+      "loss": 1.1457,
+      "step": 5995000
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.0870753719638996e-05,
+      "loss": 1.156,
+      "step": 5995500
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.0868657954009557e-05,
+      "loss": 1.1486,
+      "step": 5996000
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.086655798844899e-05,
+      "loss": 1.1583,
+      "step": 5996500
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.0864458022888423e-05,
+      "loss": 1.1577,
+      "step": 5997000
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.086235805732786e-05,
+      "loss": 1.1596,
+      "step": 5997500
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.086025809176729e-05,
+      "loss": 1.1564,
+      "step": 5998000
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.085815812620673e-05,
+      "loss": 1.1541,
+      "step": 5998500
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.0856058160646164e-05,
+      "loss": 1.1695,
+      "step": 5999000
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.08539581950856e-05,
+      "loss": 1.1244,
+      "step": 5999500
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.085186242945615e-05,
+      "loss": 1.1495,
+      "step": 6000000
+    },
+    {
+      "epoch": 3.6,
+      "eval_loss": 1.1201045513153076,
+      "eval_runtime": 1104.3731,
+      "eval_samples_per_second": 476.94,
+      "eval_steps_per_second": 79.49,
+      "step": 6000000
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.084976246389559e-05,
+      "loss": 1.1418,
+      "step": 6000500
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.0847662498335025e-05,
+      "loss": 1.1379,
+      "step": 6001000
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.084556253277446e-05,
+      "loss": 1.1534,
+      "step": 6001500
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.08434625672139e-05,
+      "loss": 1.1605,
+      "step": 6002000
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.084136260165333e-05,
+      "loss": 1.162,
+      "step": 6002500
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.083926263609277e-05,
+      "loss": 1.1471,
+      "step": 6003000
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.0837166870463326e-05,
+      "loss": 1.1559,
+      "step": 6003500
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.083506690490276e-05,
+      "loss": 1.1769,
+      "step": 6004000
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.083296693934219e-05,
+      "loss": 1.1522,
+      "step": 6004500
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.083086697378163e-05,
+      "loss": 1.1381,
+      "step": 6005000
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.0828767008221066e-05,
+      "loss": 1.1566,
+      "step": 6005500
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.08266670426605e-05,
+      "loss": 1.1572,
+      "step": 6006000
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.082456707709994e-05,
+      "loss": 1.1549,
+      "step": 6006500
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.0822471311470494e-05,
+      "loss": 1.1552,
+      "step": 6007000
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.082037134590993e-05,
+      "loss": 1.1433,
+      "step": 6007500
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.081827138034936e-05,
+      "loss": 1.1434,
+      "step": 6008000
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.08161714147888e-05,
+      "loss": 1.1637,
+      "step": 6008500
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.0814075649159354e-05,
+      "loss": 1.1689,
+      "step": 6009000
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.081197568359879e-05,
+      "loss": 1.1448,
+      "step": 6009500
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.080987571803823e-05,
+      "loss": 1.1786,
+      "step": 6010000
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.080777575247766e-05,
+      "loss": 1.1791,
+      "step": 6010500
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.0805675786917095e-05,
+      "loss": 1.1565,
+      "step": 6011000
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.0803575821356535e-05,
+      "loss": 1.1375,
+      "step": 6011500
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.080147585579596e-05,
+      "loss": 1.1463,
+      "step": 6012000
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.0799375890235395e-05,
+      "loss": 1.1686,
+      "step": 6012500
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.0797280124605956e-05,
+      "loss": 1.1566,
+      "step": 6013000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.0795180159045396e-05,
+      "loss": 1.165,
+      "step": 6013500
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.079308019348483e-05,
+      "loss": 1.1428,
+      "step": 6014000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.079098442785538e-05,
+      "loss": 1.1633,
+      "step": 6014500
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.0788884462294816e-05,
+      "loss": 1.1404,
+      "step": 6015000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.078678449673426e-05,
+      "loss": 1.1488,
+      "step": 6015500
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.078468453117369e-05,
+      "loss": 1.1348,
+      "step": 6016000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.0782584565613124e-05,
+      "loss": 1.1345,
+      "step": 6016500
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.078048460005256e-05,
+      "loss": 1.1492,
+      "step": 6017000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.077838463449199e-05,
+      "loss": 1.1607,
+      "step": 6017500
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.077628886886255e-05,
+      "loss": 1.1618,
+      "step": 6018000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.077418890330199e-05,
+      "loss": 1.1709,
+      "step": 6018500
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.077208893774142e-05,
+      "loss": 1.1431,
+      "step": 6019000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.076998897218085e-05,
+      "loss": 1.1381,
+      "step": 6019500
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.076788900662029e-05,
+      "loss": 1.129,
+      "step": 6020000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.0765789041059725e-05,
+      "loss": 1.1241,
+      "step": 6020500
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.076368907549916e-05,
+      "loss": 1.1424,
+      "step": 6021000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.07615891099386e-05,
+      "loss": 1.1486,
+      "step": 6021500
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.075949334430915e-05,
+      "loss": 1.1554,
+      "step": 6022000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.0757393378748586e-05,
+      "loss": 1.1303,
+      "step": 6022500
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.0755297613119146e-05,
+      "loss": 1.1941,
+      "step": 6023000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.075319764755858e-05,
+      "loss": 1.1647,
+      "step": 6023500
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.075109768199801e-05,
+      "loss": 1.1657,
+      "step": 6024000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.0748997716437446e-05,
+      "loss": 1.1495,
+      "step": 6024500
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.074689775087689e-05,
+      "loss": 1.1555,
+      "step": 6025000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.074479778531632e-05,
+      "loss": 1.131,
+      "step": 6025500
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.0742697819755754e-05,
+      "loss": 1.1432,
+      "step": 6026000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.0740597854195194e-05,
+      "loss": 1.1818,
+      "step": 6026500
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.073850208856575e-05,
+      "loss": 1.1621,
+      "step": 6027000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.073640212300518e-05,
+      "loss": 1.1507,
+      "step": 6027500
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.073430635737574e-05,
+      "loss": 1.1735,
+      "step": 6028000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.0732206391815175e-05,
+      "loss": 1.1571,
+      "step": 6028500
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.073010642625461e-05,
+      "loss": 1.1825,
+      "step": 6029000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.072800646069404e-05,
+      "loss": 1.1564,
+      "step": 6029500
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.0725906495133475e-05,
+      "loss": 1.1501,
+      "step": 6030000
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.0723806529572915e-05,
+      "loss": 1.1777,
+      "step": 6030500
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.0721714963874596e-05,
+      "loss": 1.1715,
+      "step": 6031000
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.071961499831403e-05,
+      "loss": 1.1574,
+      "step": 6031500
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.071751503275346e-05,
+      "loss": 1.1526,
+      "step": 6032000
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.07154150671929e-05,
+      "loss": 1.1695,
+      "step": 6032500
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.0713315101632336e-05,
+      "loss": 1.1318,
+      "step": 6033000
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.071121513607176e-05,
+      "loss": 1.1538,
+      "step": 6033500
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.07091151705112e-05,
+      "loss": 1.1761,
+      "step": 6034000
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.070701520495064e-05,
+      "loss": 1.1666,
+      "step": 6034500
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.070491523939007e-05,
+      "loss": 1.128,
+      "step": 6035000
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.070281947376063e-05,
+      "loss": 1.1584,
+      "step": 6035500
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.0700719508200064e-05,
+      "loss": 1.1697,
+      "step": 6036000
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.06986195426395e-05,
+      "loss": 1.1391,
+      "step": 6036500
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.069651957707893e-05,
+      "loss": 1.1531,
+      "step": 6037000
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.069442381144949e-05,
+      "loss": 1.1702,
+      "step": 6037500
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.069232384588893e-05,
+      "loss": 1.1404,
+      "step": 6038000
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.069022388032836e-05,
+      "loss": 1.158,
+      "step": 6038500
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.06881239147678e-05,
+      "loss": 1.154,
+      "step": 6039000
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.068602394920723e-05,
+      "loss": 1.1648,
+      "step": 6039500
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.0683923983646665e-05,
+      "loss": 1.141,
+      "step": 6040000
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.0681824018086106e-05,
+      "loss": 1.1544,
+      "step": 6040500
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.067972405252554e-05,
+      "loss": 1.1476,
+      "step": 6041000
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.067762828689609e-05,
+      "loss": 1.1499,
+      "step": 6041500
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.0675528321335526e-05,
+      "loss": 1.1424,
+      "step": 6042000
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.0673428355774966e-05,
+      "loss": 1.166,
+      "step": 6042500
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.06713283902144e-05,
+      "loss": 1.1406,
+      "step": 6043000
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.066923262458495e-05,
+      "loss": 1.1702,
+      "step": 6043500
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.0667136858955514e-05,
+      "loss": 1.1587,
+      "step": 6044000
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.066503689339495e-05,
+      "loss": 1.1714,
+      "step": 6044500
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.066293692783439e-05,
+      "loss": 1.1564,
+      "step": 6045000
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.0660836962273814e-05,
+      "loss": 1.1633,
+      "step": 6045500
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.0658736996713254e-05,
+      "loss": 1.1154,
+      "step": 6046000
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.0656641231083815e-05,
+      "loss": 1.1355,
+      "step": 6046500
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.065454126552325e-05,
+      "loss": 1.1485,
+      "step": 6047000
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.065244129996268e-05,
+      "loss": 1.1724,
+      "step": 6047500
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.0650341334402115e-05,
+      "loss": 1.171,
+      "step": 6048000
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.064824136884155e-05,
+      "loss": 1.1207,
+      "step": 6048500
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.064614560321211e-05,
+      "loss": 1.1697,
+      "step": 6049000
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.064404563765154e-05,
+      "loss": 1.1359,
+      "step": 6049500
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.0641945672090976e-05,
+      "loss": 1.1411,
+      "step": 6050000
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.063984570653041e-05,
+      "loss": 1.1821,
+      "step": 6050500
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.063774574096984e-05,
+      "loss": 1.1201,
+      "step": 6051000
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.063564577540928e-05,
+      "loss": 1.1454,
+      "step": 6051500
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.0633545809848716e-05,
+      "loss": 1.1424,
+      "step": 6052000
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.063145004421927e-05,
+      "loss": 1.1445,
+      "step": 6052500
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.062935007865871e-05,
+      "loss": 1.1573,
+      "step": 6053000
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.0627250113098144e-05,
+      "loss": 1.156,
+      "step": 6053500
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.062515014753758e-05,
+      "loss": 1.1538,
+      "step": 6054000
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.062305018197702e-05,
+      "loss": 1.1468,
+      "step": 6054500
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.062095021641645e-05,
+      "loss": 1.1716,
+      "step": 6055000
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.0618850250855884e-05,
+      "loss": 1.1697,
+      "step": 6055500
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.061675448522644e-05,
+      "loss": 1.1826,
+      "step": 6056000
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.061465451966588e-05,
+      "loss": 1.1528,
+      "step": 6056500
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.061255875403644e-05,
+      "loss": 1.1555,
+      "step": 6057000
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.0610458788475865e-05,
+      "loss": 1.1688,
+      "step": 6057500
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.06083588229153e-05,
+      "loss": 1.1609,
+      "step": 6058000
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.060625885735474e-05,
+      "loss": 1.1878,
+      "step": 6058500
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.060415889179417e-05,
+      "loss": 1.1578,
+      "step": 6059000
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.0602058926233606e-05,
+      "loss": 1.1734,
+      "step": 6059500
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.0599958960673046e-05,
+      "loss": 1.1419,
+      "step": 6060000
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.059785899511248e-05,
+      "loss": 1.18,
+      "step": 6060500
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.059576322948303e-05,
+      "loss": 1.1561,
+      "step": 6061000
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.059366326392247e-05,
+      "loss": 1.159,
+      "step": 6061500
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.059156329836191e-05,
+      "loss": 1.1631,
+      "step": 6062000
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.058946333280134e-05,
+      "loss": 1.1621,
+      "step": 6062500
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.058736336724078e-05,
+      "loss": 1.1111,
+      "step": 6063000
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.0585263401680214e-05,
+      "loss": 1.1738,
+      "step": 6063500
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.058316343611965e-05,
+      "loss": 1.1416,
+      "step": 6064000
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.058106347055909e-05,
+      "loss": 1.1578,
+      "step": 6064500
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.057896770492964e-05,
+      "loss": 1.1813,
+      "step": 6065000
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.0576867739369075e-05,
+      "loss": 1.1445,
+      "step": 6065500
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.057476777380851e-05,
+      "loss": 1.1671,
+      "step": 6066000
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.057266780824795e-05,
+      "loss": 1.1427,
+      "step": 6066500
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.05705720426185e-05,
+      "loss": 1.1394,
+      "step": 6067000
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.0568472077057935e-05,
+      "loss": 1.1664,
+      "step": 6067500
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.056637211149737e-05,
+      "loss": 1.1623,
+      "step": 6068000
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.056427214593681e-05,
+      "loss": 1.1842,
+      "step": 6068500
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.056217638030736e-05,
+      "loss": 1.1929,
+      "step": 6069000
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.0560076414746796e-05,
+      "loss": 1.1508,
+      "step": 6069500
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.0557976449186236e-05,
+      "loss": 1.1145,
+      "step": 6070000
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.055588068355679e-05,
+      "loss": 1.1495,
+      "step": 6070500
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.055378071799622e-05,
+      "loss": 1.1704,
+      "step": 6071000
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.055168075243566e-05,
+      "loss": 1.152,
+      "step": 6071500
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.05495807868751e-05,
+      "loss": 1.1681,
+      "step": 6072000
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.054748082131453e-05,
+      "loss": 1.1432,
+      "step": 6072500
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.0545380855753964e-05,
+      "loss": 1.1594,
+      "step": 6073000
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.0543280890193404e-05,
+      "loss": 1.1433,
+      "step": 6073500
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.054118092463284e-05,
+      "loss": 1.1563,
+      "step": 6074000
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.053908095907227e-05,
+      "loss": 1.1349,
+      "step": 6074500
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.0536980993511704e-05,
+      "loss": 1.1558,
+      "step": 6075000
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.053488102795114e-05,
+      "loss": 1.1547,
+      "step": 6075500
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.053278106239058e-05,
+      "loss": 1.1457,
+      "step": 6076000
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.053068529676114e-05,
+      "loss": 1.1422,
+      "step": 6076500
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.0528585331200565e-05,
+      "loss": 1.1602,
+      "step": 6077000
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.052648536564e-05,
+      "loss": 1.1473,
+      "step": 6077500
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.052438540007944e-05,
+      "loss": 1.1515,
+      "step": 6078000
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.052228543451887e-05,
+      "loss": 1.153,
+      "step": 6078500
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.0520185468958306e-05,
+      "loss": 1.136,
+      "step": 6079000
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.051808970332886e-05,
+      "loss": 1.134,
+      "step": 6079500
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.0515989737768296e-05,
+      "loss": 1.1399,
+      "step": 6080000
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.0513893972138857e-05,
+      "loss": 1.1549,
+      "step": 6080500
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.0511794006578293e-05,
+      "loss": 1.138,
+      "step": 6081000
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.050969404101773e-05,
+      "loss": 1.1546,
+      "step": 6081500
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.050759407545716e-05,
+      "loss": 1.1754,
+      "step": 6082000
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.0505494109896594e-05,
+      "loss": 1.1669,
+      "step": 6082500
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.050339414433603e-05,
+      "loss": 1.1386,
+      "step": 6083000
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.0501294178775467e-05,
+      "loss": 1.1635,
+      "step": 6083500
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.04991942132149e-05,
+      "loss": 1.1371,
+      "step": 6084000
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.0497098447585458e-05,
+      "loss": 1.1633,
+      "step": 6084500
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.049499848202489e-05,
+      "loss": 1.1573,
+      "step": 6085000
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.0492898516464328e-05,
+      "loss": 1.1577,
+      "step": 6085500
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.0490798550903765e-05,
+      "loss": 1.1513,
+      "step": 6086000
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.04886985853432e-05,
+      "loss": 1.1378,
+      "step": 6086500
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.0486602819713752e-05,
+      "loss": 1.1809,
+      "step": 6087000
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.048450285415319e-05,
+      "loss": 1.1462,
+      "step": 6087500
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.0482402888592626e-05,
+      "loss": 1.1566,
+      "step": 6088000
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.0480302923032063e-05,
+      "loss": 1.1596,
+      "step": 6088500
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.0478202957471496e-05,
+      "loss": 1.1632,
+      "step": 6089000
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.0476102991910933e-05,
+      "loss": 1.1371,
+      "step": 6089500
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.047400302635037e-05,
+      "loss": 1.1612,
+      "step": 6090000
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.0471903060789803e-05,
+      "loss": 1.1379,
+      "step": 6090500
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.0469807295160357e-05,
+      "loss": 1.1444,
+      "step": 6091000
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.0467707329599794e-05,
+      "loss": 1.1628,
+      "step": 6091500
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.046560736403923e-05,
+      "loss": 1.1588,
+      "step": 6092000
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.0463507398478664e-05,
+      "loss": 1.1528,
+      "step": 6092500
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.04614074329181e-05,
+      "loss": 1.1547,
+      "step": 6093000
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.0459311667288654e-05,
+      "loss": 1.1631,
+      "step": 6093500
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.045721170172809e-05,
+      "loss": 1.134,
+      "step": 6094000
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.0455111736167528e-05,
+      "loss": 1.1726,
+      "step": 6094500
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.045301177060696e-05,
+      "loss": 1.1727,
+      "step": 6095000
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.04509118050464e-05,
+      "loss": 1.1599,
+      "step": 6095500
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.0448811839485835e-05,
+      "loss": 1.1706,
+      "step": 6096000
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.044671187392527e-05,
+      "loss": 1.1685,
+      "step": 6096500
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.04446119083647e-05,
+      "loss": 1.1744,
+      "step": 6097000
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.044251614273526e-05,
+      "loss": 1.151,
+      "step": 6097500
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.0440416177174696e-05,
+      "loss": 1.1316,
+      "step": 6098000
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.043832041154525e-05,
+      "loss": 1.1459,
+      "step": 6098500
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.0436220445984686e-05,
+      "loss": 1.1428,
+      "step": 6099000
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.043412048042412e-05,
+      "loss": 1.1525,
+      "step": 6099500
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.0432020514863557e-05,
+      "loss": 1.1382,
+      "step": 6100000
+    },
+    {
+      "epoch": 3.66,
+      "eval_loss": 1.120174527168274,
+      "eval_runtime": 1109.1773,
+      "eval_samples_per_second": 474.874,
+      "eval_steps_per_second": 79.146,
+      "step": 6100000
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.0429920549302994e-05,
+      "loss": 1.1469,
+      "step": 6100500
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.0427820583742427e-05,
+      "loss": 1.1104,
+      "step": 6101000
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.0425724818112984e-05,
+      "loss": 1.1603,
+      "step": 6101500
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.0423624852552417e-05,
+      "loss": 1.1347,
+      "step": 6102000
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.0421524886991854e-05,
+      "loss": 1.1423,
+      "step": 6102500
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.0419429121362408e-05,
+      "loss": 1.156,
+      "step": 6103000
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.0417329155801845e-05,
+      "loss": 1.1341,
+      "step": 6103500
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.041522919024128e-05,
+      "loss": 1.1442,
+      "step": 6104000
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.0413129224680715e-05,
+      "loss": 1.1476,
+      "step": 6104500
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.0411029259120152e-05,
+      "loss": 1.1441,
+      "step": 6105000
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.040892929355959e-05,
+      "loss": 1.1622,
+      "step": 6105500
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.0406829327999022e-05,
+      "loss": 1.1632,
+      "step": 6106000
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.0404729362438452e-05,
+      "loss": 1.1527,
+      "step": 6106500
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.0402633596809013e-05,
+      "loss": 1.1411,
+      "step": 6107000
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.040053363124845e-05,
+      "loss": 1.1428,
+      "step": 6107500
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.0398433665687886e-05,
+      "loss": 1.172,
+      "step": 6108000
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.039633370012732e-05,
+      "loss": 1.1856,
+      "step": 6108500
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.039423373456675e-05,
+      "loss": 1.1281,
+      "step": 6109000
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.039213796893731e-05,
+      "loss": 1.1458,
+      "step": 6109500
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.0390038003376747e-05,
+      "loss": 1.1556,
+      "step": 6110000
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.038793803781618e-05,
+      "loss": 1.1614,
+      "step": 6110500
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.0385838072255617e-05,
+      "loss": 1.1637,
+      "step": 6111000
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.038374230662617e-05,
+      "loss": 1.142,
+      "step": 6111500
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.0381642341065608e-05,
+      "loss": 1.1465,
+      "step": 6112000
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.0379542375505045e-05,
+      "loss": 1.1742,
+      "step": 6112500
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.0377442409944478e-05,
+      "loss": 1.1331,
+      "step": 6113000
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.0375342444383908e-05,
+      "loss": 1.1521,
+      "step": 6113500
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.037324667875447e-05,
+      "loss": 1.1347,
+      "step": 6114000
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.0371146713193905e-05,
+      "loss": 1.1877,
+      "step": 6114500
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.0369046747633342e-05,
+      "loss": 1.1685,
+      "step": 6115000
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.0366946782072776e-05,
+      "loss": 1.1454,
+      "step": 6115500
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.0364846816512206e-05,
+      "loss": 1.1869,
+      "step": 6116000
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.0362755250813886e-05,
+      "loss": 1.1657,
+      "step": 6116500
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.036065528525332e-05,
+      "loss": 1.1685,
+      "step": 6117000
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.0358555319692756e-05,
+      "loss": 1.1543,
+      "step": 6117500
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.035645955406331e-05,
+      "loss": 1.1836,
+      "step": 6118000
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.0354359588502747e-05,
+      "loss": 1.1189,
+      "step": 6118500
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.035225962294218e-05,
+      "loss": 1.1723,
+      "step": 6119000
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.0350159657381617e-05,
+      "loss": 1.1785,
+      "step": 6119500
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.0348059691821054e-05,
+      "loss": 1.1609,
+      "step": 6120000
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.034595972626049e-05,
+      "loss": 1.1392,
+      "step": 6120500
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.0343859760699924e-05,
+      "loss": 1.1431,
+      "step": 6121000
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.034175979513936e-05,
+      "loss": 1.1535,
+      "step": 6121500
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.0339659829578798e-05,
+      "loss": 1.1606,
+      "step": 6122000
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.033755986401823e-05,
+      "loss": 1.1676,
+      "step": 6122500
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.033545989845766e-05,
+      "loss": 1.1375,
+      "step": 6123000
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.03333599328971e-05,
+      "loss": 1.1511,
+      "step": 6123500
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.0331259967336532e-05,
+      "loss": 1.1453,
+      "step": 6124000
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.032916000177597e-05,
+      "loss": 1.1658,
+      "step": 6124500
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.0327060036215406e-05,
+      "loss": 1.1464,
+      "step": 6125000
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.032496427058596e-05,
+      "loss": 1.1363,
+      "step": 6125500
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.0322864305025396e-05,
+      "loss": 1.1365,
+      "step": 6126000
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.032076433946483e-05,
+      "loss": 1.1587,
+      "step": 6126500
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.0318664373904266e-05,
+      "loss": 1.1528,
+      "step": 6127000
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.0316564408343703e-05,
+      "loss": 1.1654,
+      "step": 6127500
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.0314464442783137e-05,
+      "loss": 1.1701,
+      "step": 6128000
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.0312364477222573e-05,
+      "loss": 1.1637,
+      "step": 6128500
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.031026451166201e-05,
+      "loss": 1.1302,
+      "step": 6129000
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.0308168746032564e-05,
+      "loss": 1.171,
+      "step": 6129500
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.0306068780472e-05,
+      "loss": 1.1277,
+      "step": 6130000
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.0303968814911434e-05,
+      "loss": 1.1384,
+      "step": 6130500
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.030186884935087e-05,
+      "loss": 1.1523,
+      "step": 6131000
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.0299768883790308e-05,
+      "loss": 1.1489,
+      "step": 6131500
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.029766891822974e-05,
+      "loss": 1.1221,
+      "step": 6132000
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.0295568952669178e-05,
+      "loss": 1.179,
+      "step": 6132500
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.0293473187039732e-05,
+      "loss": 1.141,
+      "step": 6133000
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.0291377421410285e-05,
+      "loss": 1.1342,
+      "step": 6133500
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.0289277455849722e-05,
+      "loss": 1.1506,
+      "step": 6134000
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.028717749028916e-05,
+      "loss": 1.15,
+      "step": 6134500
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.0285077524728592e-05,
+      "loss": 1.1467,
+      "step": 6135000
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.028297755916803e-05,
+      "loss": 1.1683,
+      "step": 6135500
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.0280877593607466e-05,
+      "loss": 1.1494,
+      "step": 6136000
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.02787776280469e-05,
+      "loss": 1.1451,
+      "step": 6136500
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.0276677662486336e-05,
+      "loss": 1.1338,
+      "step": 6137000
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.0274577696925773e-05,
+      "loss": 1.1574,
+      "step": 6137500
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.0272477731365207e-05,
+      "loss": 1.1702,
+      "step": 6138000
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.0270377765804637e-05,
+      "loss": 1.1666,
+      "step": 6138500
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.0268277800244074e-05,
+      "loss": 1.131,
+      "step": 6139000
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.0266182034614634e-05,
+      "loss": 1.142,
+      "step": 6139500
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.0264086268985188e-05,
+      "loss": 1.151,
+      "step": 6140000
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.0261986303424624e-05,
+      "loss": 1.1583,
+      "step": 6140500
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.025988633786406e-05,
+      "loss": 1.155,
+      "step": 6141000
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.0257786372303495e-05,
+      "loss": 1.1406,
+      "step": 6141500
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.025568640674293e-05,
+      "loss": 1.1571,
+      "step": 6142000
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.025358644118237e-05,
+      "loss": 1.154,
+      "step": 6142500
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.0251486475621795e-05,
+      "loss": 1.1897,
+      "step": 6143000
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.0249386510061232e-05,
+      "loss": 1.1374,
+      "step": 6143500
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.0247290744431792e-05,
+      "loss": 1.1547,
+      "step": 6144000
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.0245194978802346e-05,
+      "loss": 1.1428,
+      "step": 6144500
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.0243095013241783e-05,
+      "loss": 1.1508,
+      "step": 6145000
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.024099504768122e-05,
+      "loss": 1.1623,
+      "step": 6145500
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.0238895082120653e-05,
+      "loss": 1.1682,
+      "step": 6146000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.023679511656009e-05,
+      "loss": 1.1247,
+      "step": 6146500
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.0234699350930644e-05,
+      "loss": 1.1689,
+      "step": 6147000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.023259938537008e-05,
+      "loss": 1.1537,
+      "step": 6147500
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.0230499419809517e-05,
+      "loss": 1.1619,
+      "step": 6148000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.022839945424895e-05,
+      "loss": 1.1738,
+      "step": 6148500
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.0226299488688388e-05,
+      "loss": 1.1228,
+      "step": 6149000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.0224199523127824e-05,
+      "loss": 1.1653,
+      "step": 6149500
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.0222103757498378e-05,
+      "loss": 1.1271,
+      "step": 6150000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.022000379193781e-05,
+      "loss": 1.1697,
+      "step": 6150500
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.0217903826377248e-05,
+      "loss": 1.156,
+      "step": 6151000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.0215803860816685e-05,
+      "loss": 1.1462,
+      "step": 6151500
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.021370389525612e-05,
+      "loss": 1.1442,
+      "step": 6152000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.021160392969555e-05,
+      "loss": 1.1822,
+      "step": 6152500
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.020950816406611e-05,
+      "loss": 1.1674,
+      "step": 6153000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.0207408198505546e-05,
+      "loss": 1.1494,
+      "step": 6153500
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.0205308232944983e-05,
+      "loss": 1.1846,
+      "step": 6154000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.0203208267384416e-05,
+      "loss": 1.1832,
+      "step": 6154500
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.0201108301823846e-05,
+      "loss": 1.1234,
+      "step": 6155000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.0199008336263283e-05,
+      "loss": 1.1712,
+      "step": 6155500
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.0196912570633843e-05,
+      "loss": 1.1637,
+      "step": 6156000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.019481260507328e-05,
+      "loss": 1.1471,
+      "step": 6156500
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.0192712639512714e-05,
+      "loss": 1.1457,
+      "step": 6157000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.0190612673952144e-05,
+      "loss": 1.1688,
+      "step": 6157500
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.018851270839158e-05,
+      "loss": 1.1388,
+      "step": 6158000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.0186412742831014e-05,
+      "loss": 1.1892,
+      "step": 6158500
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.018431277727045e-05,
+      "loss": 1.1355,
+      "step": 6159000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.0182212811709888e-05,
+      "loss": 1.1735,
+      "step": 6159500
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.018011704608044e-05,
+      "loss": 1.1391,
+      "step": 6160000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.0178017080519878e-05,
+      "loss": 1.1477,
+      "step": 6160500
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.017591711495931e-05,
+      "loss": 1.1407,
+      "step": 6161000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.017381714939875e-05,
+      "loss": 1.1555,
+      "step": 6161500
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.0171721383769302e-05,
+      "loss": 1.1495,
+      "step": 6162000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.016962141820874e-05,
+      "loss": 1.1324,
+      "step": 6162500
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.0167521452648176e-05,
+      "loss": 1.1659,
+      "step": 6163000
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.016542148708761e-05,
+      "loss": 1.1679,
+      "step": 6163500
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.0163321521527046e-05,
+      "loss": 1.1372,
+      "step": 6164000
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.0161229955828723e-05,
+      "loss": 1.1401,
+      "step": 6164500
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.015913419019928e-05,
+      "loss": 1.148,
+      "step": 6165000
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.0157034224638714e-05,
+      "loss": 1.1725,
+      "step": 6165500
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.015493425907815e-05,
+      "loss": 1.1322,
+      "step": 6166000
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.0152834293517587e-05,
+      "loss": 1.1643,
+      "step": 6166500
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.015073432795702e-05,
+      "loss": 1.1411,
+      "step": 6167000
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.0148634362396458e-05,
+      "loss": 1.1757,
+      "step": 6167500
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.0146534396835894e-05,
+      "loss": 1.1603,
+      "step": 6168000
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.0144434431275328e-05,
+      "loss": 1.1595,
+      "step": 6168500
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.0142334465714765e-05,
+      "loss": 1.1181,
+      "step": 6169000
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.0140238700085318e-05,
+      "loss": 1.1606,
+      "step": 6169500
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.0138138734524755e-05,
+      "loss": 1.1676,
+      "step": 6170000
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.0136038768964192e-05,
+      "loss": 1.1244,
+      "step": 6170500
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.0133938803403625e-05,
+      "loss": 1.1441,
+      "step": 6171000
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.013184303777418e-05,
+      "loss": 1.1508,
+      "step": 6171500
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.0129743072213616e-05,
+      "loss": 1.1545,
+      "step": 6172000
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.0127643106653053e-05,
+      "loss": 1.139,
+      "step": 6172500
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.0125543141092486e-05,
+      "loss": 1.172,
+      "step": 6173000
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.0123443175531923e-05,
+      "loss": 1.1656,
+      "step": 6173500
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.0121343209971353e-05,
+      "loss": 1.1637,
+      "step": 6174000
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.011924324441079e-05,
+      "loss": 1.1281,
+      "step": 6174500
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.0117143278850223e-05,
+      "loss": 1.1649,
+      "step": 6175000
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.011504331328966e-05,
+      "loss": 1.1623,
+      "step": 6175500
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.011294754766022e-05,
+      "loss": 1.1399,
+      "step": 6176000
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.011084758209965e-05,
+      "loss": 1.1642,
+      "step": 6176500
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.010875181647021e-05,
+      "loss": 1.1463,
+      "step": 6177000
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.0106651850909648e-05,
+      "loss": 1.1625,
+      "step": 6177500
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.010455188534908e-05,
+      "loss": 1.1537,
+      "step": 6178000
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.0102451919788518e-05,
+      "loss": 1.1777,
+      "step": 6178500
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.0100351954227948e-05,
+      "loss": 1.1529,
+      "step": 6179000
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.0098251988667382e-05,
+      "loss": 1.1825,
+      "step": 6179500
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.009615202310682e-05,
+      "loss": 1.1534,
+      "step": 6180000
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.0094052057546255e-05,
+      "loss": 1.161,
+      "step": 6180500
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.0091952091985692e-05,
+      "loss": 1.1582,
+      "step": 6181000
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.0089852126425126e-05,
+      "loss": 1.1355,
+      "step": 6181500
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.0087752160864563e-05,
+      "loss": 1.1742,
+      "step": 6182000
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.0085656395235116e-05,
+      "loss": 1.1637,
+      "step": 6182500
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.0083556429674553e-05,
+      "loss": 1.1849,
+      "step": 6183000
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.0081460664045107e-05,
+      "loss": 1.1681,
+      "step": 6183500
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.0079360698484543e-05,
+      "loss": 1.1403,
+      "step": 6184000
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.0077260732923977e-05,
+      "loss": 1.1678,
+      "step": 6184500
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.0075160767363414e-05,
+      "loss": 1.1464,
+      "step": 6185000
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.007306080180285e-05,
+      "loss": 1.1458,
+      "step": 6185500
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.0070960836242284e-05,
+      "loss": 1.1497,
+      "step": 6186000
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.006886087068172e-05,
+      "loss": 1.1336,
+      "step": 6186500
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.0066760905121158e-05,
+      "loss": 1.1482,
+      "step": 6187000
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.006466093956059e-05,
+      "loss": 1.157,
+      "step": 6187500
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.0062560974000028e-05,
+      "loss": 1.1463,
+      "step": 6188000
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.006046520837058e-05,
+      "loss": 1.1655,
+      "step": 6188500
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.005836524281002e-05,
+      "loss": 1.1561,
+      "step": 6189000
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.0056265277249455e-05,
+      "loss": 1.1351,
+      "step": 6189500
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.005416531168889e-05,
+      "loss": 1.1453,
+      "step": 6190000
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.0052065346128326e-05,
+      "loss": 1.1432,
+      "step": 6190500
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.0049965380567762e-05,
+      "loss": 1.1202,
+      "step": 6191000
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.004786541500719e-05,
+      "loss": 1.1381,
+      "step": 6191500
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.0045765449446626e-05,
+      "loss": 1.1366,
+      "step": 6192000
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.0043673883748306e-05,
+      "loss": 1.1712,
+      "step": 6192500
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.004157391818774e-05,
+      "loss": 1.1298,
+      "step": 6193000
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.0039473952627177e-05,
+      "loss": 1.1207,
+      "step": 6193500
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.0037373987066614e-05,
+      "loss": 1.1794,
+      "step": 6194000
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.0035274021506047e-05,
+      "loss": 1.1426,
+      "step": 6194500
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.0033174055945484e-05,
+      "loss": 1.133,
+      "step": 6195000
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.003107409038492e-05,
+      "loss": 1.1499,
+      "step": 6195500
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.0028978324755474e-05,
+      "loss": 1.1531,
+      "step": 6196000
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.002687835919491e-05,
+      "loss": 1.155,
+      "step": 6196500
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.0024778393634345e-05,
+      "loss": 1.1642,
+      "step": 6197000
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.002267842807378e-05,
+      "loss": 1.1266,
+      "step": 6197500
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.002057846251322e-05,
+      "loss": 1.1724,
+      "step": 6198000
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.0018478496952645e-05,
+      "loss": 1.1978,
+      "step": 6198500
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.0016378531392082e-05,
+      "loss": 1.1399,
+      "step": 6199000
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.001427856583152e-05,
+      "loss": 1.164,
+      "step": 6199500
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.001218280020208e-05,
+      "loss": 1.1524,
+      "step": 6200000
+    },
+    {
+      "epoch": 3.72,
+      "eval_loss": 1.116134762763977,
+      "eval_runtime": 1099.3849,
+      "eval_samples_per_second": 479.104,
+      "eval_steps_per_second": 79.851,
+      "step": 6200000
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.0010082834641516e-05,
+      "loss": 1.1378,
+      "step": 6200500
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.0007982869080943e-05,
+      "loss": 1.1335,
+      "step": 6201000
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.000588290352038e-05,
+      "loss": 1.142,
+      "step": 6201500
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.000378713789094e-05,
+      "loss": 1.1316,
+      "step": 6202000
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.0001691372261493e-05,
+      "loss": 1.1503,
+      "step": 6202500
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.999959140670093e-05,
+      "loss": 1.1535,
+      "step": 6203000
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.9997491441140367e-05,
+      "loss": 1.1509,
+      "step": 6203500
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.99953914755798e-05,
+      "loss": 1.1635,
+      "step": 6204000
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.9993291510019237e-05,
+      "loss": 1.1405,
+      "step": 6204500
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.9991191544458674e-05,
+      "loss": 1.1681,
+      "step": 6205000
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.9989095778829228e-05,
+      "loss": 1.1532,
+      "step": 6205500
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.9986995813268665e-05,
+      "loss": 1.1368,
+      "step": 6206000
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.9984895847708098e-05,
+      "loss": 1.139,
+      "step": 6206500
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.9982795882147535e-05,
+      "loss": 1.143,
+      "step": 6207000
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.9980695916586972e-05,
+      "loss": 1.1573,
+      "step": 6207500
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.99785959510264e-05,
+      "loss": 1.134,
+      "step": 6208000
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.997650018539696e-05,
+      "loss": 1.1822,
+      "step": 6208500
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.9974400219836396e-05,
+      "loss": 1.1359,
+      "step": 6209000
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.9972300254275832e-05,
+      "loss": 1.1359,
+      "step": 6209500
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.9970204488646386e-05,
+      "loss": 1.1415,
+      "step": 6210000
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.9968104523085823e-05,
+      "loss": 1.1546,
+      "step": 6210500
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.9966004557525256e-05,
+      "loss": 1.1757,
+      "step": 6211000
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.9963904591964693e-05,
+      "loss": 1.1412,
+      "step": 6211500
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.996180462640413e-05,
+      "loss": 1.1388,
+      "step": 6212000
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.9959704660843564e-05,
+      "loss": 1.1408,
+      "step": 6212500
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.9957604695282994e-05,
+      "loss": 1.1606,
+      "step": 6213000
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.995550472972243e-05,
+      "loss": 1.154,
+      "step": 6213500
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.9953404764161867e-05,
+      "loss": 1.1552,
+      "step": 6214000
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.99513047986013e-05,
+      "loss": 1.1552,
+      "step": 6214500
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.9949204833040738e-05,
+      "loss": 1.125,
+      "step": 6215000
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.9947104867480174e-05,
+      "loss": 1.1474,
+      "step": 6215500
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.9945009101850728e-05,
+      "loss": 1.1463,
+      "step": 6216000
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.994290913629016e-05,
+      "loss": 1.1347,
+      "step": 6216500
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.99408091707296e-05,
+      "loss": 1.1755,
+      "step": 6217000
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.9938709205169035e-05,
+      "loss": 1.1816,
+      "step": 6217500
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.993661343953959e-05,
+      "loss": 1.1187,
+      "step": 6218000
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.9934513473979026e-05,
+      "loss": 1.1528,
+      "step": 6218500
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.993241350841846e-05,
+      "loss": 1.163,
+      "step": 6219000
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.9930313542857896e-05,
+      "loss": 1.1477,
+      "step": 6219500
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.9928213577297333e-05,
+      "loss": 1.1582,
+      "step": 6220000
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.9926113611736766e-05,
+      "loss": 1.1518,
+      "step": 6220500
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.9924017846107323e-05,
+      "loss": 1.171,
+      "step": 6221000
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.9921917880546757e-05,
+      "loss": 1.1644,
+      "step": 6221500
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.9919817914986193e-05,
+      "loss": 1.1546,
+      "step": 6222000
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.991771794942563e-05,
+      "loss": 1.1343,
+      "step": 6222500
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.9915617983865064e-05,
+      "loss": 1.1176,
+      "step": 6223000
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.9913522218235617e-05,
+      "loss": 1.1206,
+      "step": 6223500
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.9911422252675054e-05,
+      "loss": 1.1348,
+      "step": 6224000
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.990932228711449e-05,
+      "loss": 1.1622,
+      "step": 6224500
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.9907222321553925e-05,
+      "loss": 1.1402,
+      "step": 6225000
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.990512235599336e-05,
+      "loss": 1.1905,
+      "step": 6225500
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.9903022390432798e-05,
+      "loss": 1.1585,
+      "step": 6226000
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.9900922424872235e-05,
+      "loss": 1.1717,
+      "step": 6226500
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.989882245931167e-05,
+      "loss": 1.1645,
+      "step": 6227000
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.9896726693682222e-05,
+      "loss": 1.1655,
+      "step": 6227500
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.989463092805278e-05,
+      "loss": 1.1356,
+      "step": 6228000
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.9892530962492213e-05,
+      "loss": 1.1626,
+      "step": 6228500
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.989043099693165e-05,
+      "loss": 1.1559,
+      "step": 6229000
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.9888331031371086e-05,
+      "loss": 1.1745,
+      "step": 6229500
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.988623106581052e-05,
+      "loss": 1.1542,
+      "step": 6230000
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.9884131100249957e-05,
+      "loss": 1.1397,
+      "step": 6230500
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.9882031134689393e-05,
+      "loss": 1.1583,
+      "step": 6231000
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.9879931169128827e-05,
+      "loss": 1.1385,
+      "step": 6231500
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.987783540349938e-05,
+      "loss": 1.1636,
+      "step": 6232000
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.9875735437938817e-05,
+      "loss": 1.1114,
+      "step": 6232500
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.9873635472378254e-05,
+      "loss": 1.1735,
+      "step": 6233000
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.987153550681769e-05,
+      "loss": 1.1443,
+      "step": 6233500
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.9869439741188245e-05,
+      "loss": 1.1686,
+      "step": 6234000
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.9867339775627678e-05,
+      "loss": 1.1631,
+      "step": 6234500
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.9865244009998235e-05,
+      "loss": 1.161,
+      "step": 6235000
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.986314404443767e-05,
+      "loss": 1.1653,
+      "step": 6235500
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.9861044078877105e-05,
+      "loss": 1.1406,
+      "step": 6236000
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.9858944113316542e-05,
+      "loss": 1.1427,
+      "step": 6236500
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.9856844147755976e-05,
+      "loss": 1.1699,
+      "step": 6237000
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.9854744182195412e-05,
+      "loss": 1.1711,
+      "step": 6237500
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.985264421663485e-05,
+      "loss": 1.1351,
+      "step": 6238000
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.9850544251074283e-05,
+      "loss": 1.1303,
+      "step": 6238500
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.984844848544484e-05,
+      "loss": 1.1704,
+      "step": 6239000
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.9846348519884273e-05,
+      "loss": 1.1441,
+      "step": 6239500
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.984424855432371e-05,
+      "loss": 1.1517,
+      "step": 6240000
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.9842152788694264e-05,
+      "loss": 1.1693,
+      "step": 6240500
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.98400528231337e-05,
+      "loss": 1.1604,
+      "step": 6241000
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.9837952857573134e-05,
+      "loss": 1.1735,
+      "step": 6241500
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.983585289201257e-05,
+      "loss": 1.1373,
+      "step": 6242000
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.9833757126383124e-05,
+      "loss": 1.1246,
+      "step": 6242500
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.983165716082256e-05,
+      "loss": 1.1595,
+      "step": 6243000
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.9829557195261998e-05,
+      "loss": 1.1617,
+      "step": 6243500
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.982745722970143e-05,
+      "loss": 1.1339,
+      "step": 6244000
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.9825357264140868e-05,
+      "loss": 1.1773,
+      "step": 6244500
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.9823257298580305e-05,
+      "loss": 1.1555,
+      "step": 6245000
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.982116153295086e-05,
+      "loss": 1.1475,
+      "step": 6245500
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.9819061567390296e-05,
+      "loss": 1.1174,
+      "step": 6246000
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.981696160182973e-05,
+      "loss": 1.1641,
+      "step": 6246500
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.9814861636269166e-05,
+      "loss": 1.1374,
+      "step": 6247000
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.9812761670708603e-05,
+      "loss": 1.1506,
+      "step": 6247500
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.9810661705148036e-05,
+      "loss": 1.143,
+      "step": 6248000
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.9808561739587473e-05,
+      "loss": 1.1523,
+      "step": 6248500
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.980646177402691e-05,
+      "loss": 1.1386,
+      "step": 6249000
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.9804366008397463e-05,
+      "loss": 1.1433,
+      "step": 6249500
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.9802266042836897e-05,
+      "loss": 1.1422,
+      "step": 6250000
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.9800170277207454e-05,
+      "loss": 1.1313,
+      "step": 6250500
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.9798070311646887e-05,
+      "loss": 1.1299,
+      "step": 6251000
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.9795970346086324e-05,
+      "loss": 1.1819,
+      "step": 6251500
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.979387038052576e-05,
+      "loss": 1.1488,
+      "step": 6252000
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.9791770414965194e-05,
+      "loss": 1.1652,
+      "step": 6252500
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.978967464933575e-05,
+      "loss": 1.1318,
+      "step": 6253000
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.9787574683775185e-05,
+      "loss": 1.1702,
+      "step": 6253500
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.9785474718214622e-05,
+      "loss": 1.1657,
+      "step": 6254000
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.978337475265406e-05,
+      "loss": 1.1684,
+      "step": 6254500
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.9781274787093492e-05,
+      "loss": 1.1595,
+      "step": 6255000
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.9779179021464046e-05,
+      "loss": 1.1631,
+      "step": 6255500
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.9777079055903482e-05,
+      "loss": 1.1437,
+      "step": 6256000
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.977497909034292e-05,
+      "loss": 1.1439,
+      "step": 6256500
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.9772879124782353e-05,
+      "loss": 1.1488,
+      "step": 6257000
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.977077915922179e-05,
+      "loss": 1.1564,
+      "step": 6257500
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.9768679193661226e-05,
+      "loss": 1.1542,
+      "step": 6258000
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.976657922810066e-05,
+      "loss": 1.15,
+      "step": 6258500
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.976447926254009e-05,
+      "loss": 1.1665,
+      "step": 6259000
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.976238349691065e-05,
+      "loss": 1.1457,
+      "step": 6259500
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.9760287731281207e-05,
+      "loss": 1.1529,
+      "step": 6260000
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.975818776572064e-05,
+      "loss": 1.1935,
+      "step": 6260500
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.9756087800160078e-05,
+      "loss": 1.1577,
+      "step": 6261000
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.9753987834599514e-05,
+      "loss": 1.1538,
+      "step": 6261500
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.9751887869038948e-05,
+      "loss": 1.1364,
+      "step": 6262000
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.9749787903478385e-05,
+      "loss": 1.166,
+      "step": 6262500
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.974768793791782e-05,
+      "loss": 1.1569,
+      "step": 6263000
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.9745592172288375e-05,
+      "loss": 1.158,
+      "step": 6263500
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.974349220672781e-05,
+      "loss": 1.15,
+      "step": 6264000
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.9741392241167245e-05,
+      "loss": 1.1489,
+      "step": 6264500
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.9739292275606682e-05,
+      "loss": 1.1415,
+      "step": 6265000
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.9737196509977236e-05,
+      "loss": 1.1564,
+      "step": 6265500
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.9735096544416673e-05,
+      "loss": 1.1131,
+      "step": 6266000
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.9732996578856106e-05,
+      "loss": 1.1563,
+      "step": 6266500
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.9730900813226663e-05,
+      "loss": 1.1357,
+      "step": 6267000
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.9728800847666097e-05,
+      "loss": 1.1619,
+      "step": 6267500
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.9726700882105533e-05,
+      "loss": 1.1224,
+      "step": 6268000
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.972460091654497e-05,
+      "loss": 1.1513,
+      "step": 6268500
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.9722500950984404e-05,
+      "loss": 1.143,
+      "step": 6269000
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.972040098542384e-05,
+      "loss": 1.152,
+      "step": 6269500
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.9718301019863277e-05,
+      "loss": 1.1715,
+      "step": 6270000
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.971620105430271e-05,
+      "loss": 1.1505,
+      "step": 6270500
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.971410108874214e-05,
+      "loss": 1.1466,
+      "step": 6271000
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.9712001123181578e-05,
+      "loss": 1.1332,
+      "step": 6271500
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.9709901157621015e-05,
+      "loss": 1.1262,
+      "step": 6272000
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.9707805391991575e-05,
+      "loss": 1.1396,
+      "step": 6272500
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.9705705426431002e-05,
+      "loss": 1.1239,
+      "step": 6273000
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.970360546087044e-05,
+      "loss": 1.184,
+      "step": 6273500
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.9701505495309875e-05,
+      "loss": 1.1437,
+      "step": 6274000
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.969940552974931e-05,
+      "loss": 1.1741,
+      "step": 6274500
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.9697305564188746e-05,
+      "loss": 1.1421,
+      "step": 6275000
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.96952097985593e-05,
+      "loss": 1.1659,
+      "step": 6275500
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.9693109832998736e-05,
+      "loss": 1.1747,
+      "step": 6276000
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.9691009867438173e-05,
+      "loss": 1.1387,
+      "step": 6276500
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.9688909901877606e-05,
+      "loss": 1.1428,
+      "step": 6277000
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.9686809936317043e-05,
+      "loss": 1.1533,
+      "step": 6277500
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.968470997075648e-05,
+      "loss": 1.1676,
+      "step": 6278000
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.9682610005195914e-05,
+      "loss": 1.1697,
+      "step": 6278500
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.968051003963535e-05,
+      "loss": 1.1545,
+      "step": 6279000
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.9678414274005904e-05,
+      "loss": 1.14,
+      "step": 6279500
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.967631430844534e-05,
+      "loss": 1.1575,
+      "step": 6280000
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9674218542815894e-05,
+      "loss": 1.1682,
+      "step": 6280500
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.967211857725533e-05,
+      "loss": 1.1556,
+      "step": 6281000
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9670018611694765e-05,
+      "loss": 1.1523,
+      "step": 6281500
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.96679186461342e-05,
+      "loss": 1.1714,
+      "step": 6282000
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.966581868057364e-05,
+      "loss": 1.1294,
+      "step": 6282500
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9663718715013072e-05,
+      "loss": 1.1608,
+      "step": 6283000
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.966162294938363e-05,
+      "loss": 1.1272,
+      "step": 6283500
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9659522983823062e-05,
+      "loss": 1.1666,
+      "step": 6284000
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.96574230182625e-05,
+      "loss": 1.1339,
+      "step": 6284500
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9655323052701936e-05,
+      "loss": 1.1549,
+      "step": 6285000
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.965322308714137e-05,
+      "loss": 1.1629,
+      "step": 6285500
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9651123121580806e-05,
+      "loss": 1.149,
+      "step": 6286000
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.964902735595136e-05,
+      "loss": 1.1445,
+      "step": 6286500
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9646927390390797e-05,
+      "loss": 1.1671,
+      "step": 6287000
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9644827424830234e-05,
+      "loss": 1.1227,
+      "step": 6287500
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9642727459269667e-05,
+      "loss": 1.1383,
+      "step": 6288000
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9640627493709104e-05,
+      "loss": 1.1429,
+      "step": 6288500
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9638531728079657e-05,
+      "loss": 1.1414,
+      "step": 6289000
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9636431762519094e-05,
+      "loss": 1.176,
+      "step": 6289500
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9634331796958528e-05,
+      "loss": 1.1837,
+      "step": 6290000
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9632231831397965e-05,
+      "loss": 1.1506,
+      "step": 6290500
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.96301318658374e-05,
+      "loss": 1.13,
+      "step": 6291000
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9628036100207955e-05,
+      "loss": 1.1653,
+      "step": 6291500
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9625936134647392e-05,
+      "loss": 1.163,
+      "step": 6292000
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9623836169086825e-05,
+      "loss": 1.1654,
+      "step": 6292500
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9621736203526262e-05,
+      "loss": 1.1521,
+      "step": 6293000
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.96196362379657e-05,
+      "loss": 1.1532,
+      "step": 6293500
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9617536272405133e-05,
+      "loss": 1.1584,
+      "step": 6294000
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.961543630684457e-05,
+      "loss": 1.1674,
+      "step": 6294500
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9613336341284006e-05,
+      "loss": 1.1768,
+      "step": 6295000
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.961124057565456e-05,
+      "loss": 1.1619,
+      "step": 6295500
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9609140610093997e-05,
+      "loss": 1.1394,
+      "step": 6296000
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.960704064453343e-05,
+      "loss": 1.173,
+      "step": 6296500
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9604940678972867e-05,
+      "loss": 1.1725,
+      "step": 6297000
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9602840713412304e-05,
+      "loss": 1.1771,
+      "step": 6297500
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.960074074785173e-05,
+      "loss": 1.1591,
+      "step": 6298000
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9598644982222294e-05,
+      "loss": 1.1779,
+      "step": 6298500
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9596545016661728e-05,
+      "loss": 1.1493,
+      "step": 6299000
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.959444925103228e-05,
+      "loss": 1.1416,
+      "step": 6299500
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9592349285471718e-05,
+      "loss": 1.1302,
+      "step": 6300000
+    },
+    {
+      "epoch": 3.78,
+      "eval_loss": 1.1142629384994507,
+      "eval_runtime": 1100.8622,
+      "eval_samples_per_second": 478.461,
+      "eval_steps_per_second": 79.744,
+      "step": 6300000
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9590249319911155e-05,
+      "loss": 1.1701,
+      "step": 6300500
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.958814935435059e-05,
+      "loss": 1.1607,
+      "step": 6301000
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9586049388790025e-05,
+      "loss": 1.1362,
+      "step": 6301500
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9583949423229462e-05,
+      "loss": 1.1545,
+      "step": 6302000
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9581849457668892e-05,
+      "loss": 1.1501,
+      "step": 6302500
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9579749492108326e-05,
+      "loss": 1.1484,
+      "step": 6303000
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9577649526547762e-05,
+      "loss": 1.1433,
+      "step": 6303500
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.95755495609872e-05,
+      "loss": 1.1189,
+      "step": 6304000
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.957345379535776e-05,
+      "loss": 1.1372,
+      "step": 6304500
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.957135382979719e-05,
+      "loss": 1.1482,
+      "step": 6305000
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9569253864236623e-05,
+      "loss": 1.1552,
+      "step": 6305500
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.956715389867606e-05,
+      "loss": 1.145,
+      "step": 6306000
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9565053933115497e-05,
+      "loss": 1.1358,
+      "step": 6306500
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.956295396755493e-05,
+      "loss": 1.1507,
+      "step": 6307000
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9560854001994367e-05,
+      "loss": 1.1342,
+      "step": 6307500
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9558754036433804e-05,
+      "loss": 1.1372,
+      "step": 6308000
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9556658270804358e-05,
+      "loss": 1.133,
+      "step": 6308500
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.955455830524379e-05,
+      "loss": 1.1505,
+      "step": 6309000
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9552458339683228e-05,
+      "loss": 1.1621,
+      "step": 6309500
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9550358374122665e-05,
+      "loss": 1.1193,
+      "step": 6310000
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.954826260849322e-05,
+      "loss": 1.1537,
+      "step": 6310500
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9546162642932655e-05,
+      "loss": 1.1231,
+      "step": 6311000
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.954406267737209e-05,
+      "loss": 1.171,
+      "step": 6311500
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9541966911742646e-05,
+      "loss": 1.1666,
+      "step": 6312000
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.953986694618208e-05,
+      "loss": 1.1388,
+      "step": 6312500
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.953777118055264e-05,
+      "loss": 1.1517,
+      "step": 6313000
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9535671214992076e-05,
+      "loss": 1.1569,
+      "step": 6313500
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9533571249431513e-05,
+      "loss": 1.1477,
+      "step": 6314000
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.953147128387094e-05,
+      "loss": 1.1786,
+      "step": 6314500
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9529371318310377e-05,
+      "loss": 1.1513,
+      "step": 6315000
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9527271352749813e-05,
+      "loss": 1.1475,
+      "step": 6315500
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9525171387189247e-05,
+      "loss": 1.154,
+      "step": 6316000
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9523071421628684e-05,
+      "loss": 1.1686,
+      "step": 6316500
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.952097145606812e-05,
+      "loss": 1.1663,
+      "step": 6317000
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9518871490507554e-05,
+      "loss": 1.1444,
+      "step": 6317500
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.951677152494699e-05,
+      "loss": 1.1384,
+      "step": 6318000
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9514671559386428e-05,
+      "loss": 1.1615,
+      "step": 6318500
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.951257579375698e-05,
+      "loss": 1.1554,
+      "step": 6319000
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9510475828196418e-05,
+      "loss": 1.1294,
+      "step": 6319500
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.950837586263585e-05,
+      "loss": 1.1821,
+      "step": 6320000
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.950628009700641e-05,
+      "loss": 1.1441,
+      "step": 6320500
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9504180131445842e-05,
+      "loss": 1.169,
+      "step": 6321000
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.950208016588528e-05,
+      "loss": 1.1573,
+      "step": 6321500
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9499980200324716e-05,
+      "loss": 1.115,
+      "step": 6322000
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.949788023476415e-05,
+      "loss": 1.1358,
+      "step": 6322500
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9495780269203586e-05,
+      "loss": 1.1601,
+      "step": 6323000
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9493680303643023e-05,
+      "loss": 1.1571,
+      "step": 6323500
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9491580338082456e-05,
+      "loss": 1.1412,
+      "step": 6324000
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9489484572453013e-05,
+      "loss": 1.1825,
+      "step": 6324500
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9487384606892447e-05,
+      "loss": 1.1538,
+      "step": 6325000
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9485284641331884e-05,
+      "loss": 1.1491,
+      "step": 6325500
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9483188875702437e-05,
+      "loss": 1.1397,
+      "step": 6326000
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9481088910141874e-05,
+      "loss": 1.1431,
+      "step": 6326500
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9478988944581308e-05,
+      "loss": 1.1394,
+      "step": 6327000
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9476888979020744e-05,
+      "loss": 1.1483,
+      "step": 6327500
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9474793213391298e-05,
+      "loss": 1.1308,
+      "step": 6328000
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9472693247830735e-05,
+      "loss": 1.1452,
+      "step": 6328500
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.947059328227017e-05,
+      "loss": 1.1507,
+      "step": 6329000
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9468493316709605e-05,
+      "loss": 1.1404,
+      "step": 6329500
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.9466393351149042e-05,
+      "loss": 1.1502,
+      "step": 6330000
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.946429338558848e-05,
+      "loss": 1.1437,
+      "step": 6330500
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.9462197619959032e-05,
+      "loss": 1.1592,
+      "step": 6331000
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.946009765439847e-05,
+      "loss": 1.1501,
+      "step": 6331500
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.9457997688837903e-05,
+      "loss": 1.1403,
+      "step": 6332000
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.945589772327734e-05,
+      "loss": 1.1749,
+      "step": 6332500
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.9453801957647893e-05,
+      "loss": 1.1462,
+      "step": 6333000
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.945170199208733e-05,
+      "loss": 1.1396,
+      "step": 6333500
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.9449602026526763e-05,
+      "loss": 1.1589,
+      "step": 6334000
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.94475020609662e-05,
+      "loss": 1.1418,
+      "step": 6334500
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.9445402095405637e-05,
+      "loss": 1.1617,
+      "step": 6335000
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.944330212984507e-05,
+      "loss": 1.1423,
+      "step": 6335500
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.9441202164284507e-05,
+      "loss": 1.1528,
+      "step": 6336000
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.9439102198723938e-05,
+      "loss": 1.1143,
+      "step": 6336500
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.9437010633025615e-05,
+      "loss": 1.1624,
+      "step": 6337000
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.943491066746505e-05,
+      "loss": 1.1583,
+      "step": 6337500
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.9432810701904488e-05,
+      "loss": 1.146,
+      "step": 6338000
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.9430710736343925e-05,
+      "loss": 1.1578,
+      "step": 6338500
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.942861077078336e-05,
+      "loss": 1.1176,
+      "step": 6339000
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.9426510805222795e-05,
+      "loss": 1.1487,
+      "step": 6339500
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.942441503959335e-05,
+      "loss": 1.127,
+      "step": 6340000
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.9422315074032786e-05,
+      "loss": 1.1658,
+      "step": 6340500
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.942021510847222e-05,
+      "loss": 1.1271,
+      "step": 6341000
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.9418115142911656e-05,
+      "loss": 1.1245,
+      "step": 6341500
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.9416015177351093e-05,
+      "loss": 1.1522,
+      "step": 6342000
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.9413919411721647e-05,
+      "loss": 1.1556,
+      "step": 6342500
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.9411819446161083e-05,
+      "loss": 1.1405,
+      "step": 6343000
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.9409719480600517e-05,
+      "loss": 1.1715,
+      "step": 6343500
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.9407619515039954e-05,
+      "loss": 1.1773,
+      "step": 6344000
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.940551954947939e-05,
+      "loss": 1.171,
+      "step": 6344500
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.9403419583918824e-05,
+      "loss": 1.1588,
+      "step": 6345000
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.940132381828938e-05,
+      "loss": 1.166,
+      "step": 6345500
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.9399223852728814e-05,
+      "loss": 1.1576,
+      "step": 6346000
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.939712388716825e-05,
+      "loss": 1.1573,
+      "step": 6346500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.9395023921607688e-05,
+      "loss": 1.163,
+      "step": 6347000
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.939292395604712e-05,
+      "loss": 1.1324,
+      "step": 6347500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.939082399048656e-05,
+      "loss": 1.1718,
+      "step": 6348000
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.938872402492599e-05,
+      "loss": 1.1377,
+      "step": 6348500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.9386624059365422e-05,
+      "loss": 1.1313,
+      "step": 6349000
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.9384532493667102e-05,
+      "loss": 1.139,
+      "step": 6349500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.938243252810654e-05,
+      "loss": 1.1574,
+      "step": 6350000
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.9380332562545973e-05,
+      "loss": 1.153,
+      "step": 6350500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.937823259698541e-05,
+      "loss": 1.1503,
+      "step": 6351000
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.9376132631424846e-05,
+      "loss": 1.1715,
+      "step": 6351500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.93740368657954e-05,
+      "loss": 1.1512,
+      "step": 6352000
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.9371936900234837e-05,
+      "loss": 1.1293,
+      "step": 6352500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.936983693467427e-05,
+      "loss": 1.1209,
+      "step": 6353000
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.9367736969113707e-05,
+      "loss": 1.1816,
+      "step": 6353500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.936564120348426e-05,
+      "loss": 1.1411,
+      "step": 6354000
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.9363541237923698e-05,
+      "loss": 1.1558,
+      "step": 6354500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.936144127236313e-05,
+      "loss": 1.1356,
+      "step": 6355000
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.9359341306802568e-05,
+      "loss": 1.156,
+      "step": 6355500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.935724554117312e-05,
+      "loss": 1.1202,
+      "step": 6356000
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.935514557561256e-05,
+      "loss": 1.1772,
+      "step": 6356500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.9353045610051995e-05,
+      "loss": 1.1753,
+      "step": 6357000
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.935094564449143e-05,
+      "loss": 1.1371,
+      "step": 6357500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.9348845678930865e-05,
+      "loss": 1.1471,
+      "step": 6358000
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.9346745713370302e-05,
+      "loss": 1.1319,
+      "step": 6358500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.9344645747809736e-05,
+      "loss": 1.1418,
+      "step": 6359000
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.9342545782249173e-05,
+      "loss": 1.1395,
+      "step": 6359500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.9340454216550846e-05,
+      "loss": 1.1293,
+      "step": 6360000
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.933835425099028e-05,
+      "loss": 1.1455,
+      "step": 6360500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.9336254285429717e-05,
+      "loss": 1.1404,
+      "step": 6361000
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.9334154319869153e-05,
+      "loss": 1.1588,
+      "step": 6361500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.9332054354308587e-05,
+      "loss": 1.1373,
+      "step": 6362000
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.9329954388748024e-05,
+      "loss": 1.1468,
+      "step": 6362500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.932785442318746e-05,
+      "loss": 1.1653,
+      "step": 6363000
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.9325754457626897e-05,
+      "loss": 1.1602,
+      "step": 6363500
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.932365449206633e-05,
+      "loss": 1.1418,
+      "step": 6364000
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.9321554526505768e-05,
+      "loss": 1.158,
+      "step": 6364500
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.9319454560945205e-05,
+      "loss": 1.121,
+      "step": 6365000
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.931735459538463e-05,
+      "loss": 1.1797,
+      "step": 6365500
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.931525882975519e-05,
+      "loss": 1.1629,
+      "step": 6366000
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.931316306412575e-05,
+      "loss": 1.1188,
+      "step": 6366500
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.9311063098565182e-05,
+      "loss": 1.1251,
+      "step": 6367000
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.930896313300462e-05,
+      "loss": 1.141,
+      "step": 6367500
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.9306863167444056e-05,
+      "loss": 1.1275,
+      "step": 6368000
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.930476320188349e-05,
+      "loss": 1.1674,
+      "step": 6368500
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.9302663236322926e-05,
+      "loss": 1.1019,
+      "step": 6369000
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.93005716706246e-05,
+      "loss": 1.1742,
+      "step": 6369500
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.9298471705064033e-05,
+      "loss": 1.1239,
+      "step": 6370000
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.929637173950347e-05,
+      "loss": 1.1452,
+      "step": 6370500
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.9294271773942907e-05,
+      "loss": 1.1468,
+      "step": 6371000
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.929217180838234e-05,
+      "loss": 1.1493,
+      "step": 6371500
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.9290071842821777e-05,
+      "loss": 1.1628,
+      "step": 6372000
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.9287971877261214e-05,
+      "loss": 1.1375,
+      "step": 6372500
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.9285871911700648e-05,
+      "loss": 1.1646,
+      "step": 6373000
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.9283771946140084e-05,
+      "loss": 1.1526,
+      "step": 6373500
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.928167198057952e-05,
+      "loss": 1.1391,
+      "step": 6374000
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.9279576214950075e-05,
+      "loss": 1.1551,
+      "step": 6374500
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.927747624938951e-05,
+      "loss": 1.158,
+      "step": 6375000
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.9275376283828945e-05,
+      "loss": 1.1802,
+      "step": 6375500
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.9273276318268382e-05,
+      "loss": 1.1535,
+      "step": 6376000
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.927117635270782e-05,
+      "loss": 1.1703,
+      "step": 6376500
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.926907638714725e-05,
+      "loss": 1.137,
+      "step": 6377000
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.926698062151781e-05,
+      "loss": 1.1448,
+      "step": 6377500
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.9264880655957243e-05,
+      "loss": 1.1348,
+      "step": 6378000
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.926278069039668e-05,
+      "loss": 1.1508,
+      "step": 6378500
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.9260680724836116e-05,
+      "loss": 1.1369,
+      "step": 6379000
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.9258580759275543e-05,
+      "loss": 1.1485,
+      "step": 6379500
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.925648079371498e-05,
+      "loss": 1.1537,
+      "step": 6380000
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.9254380828154417e-05,
+      "loss": 1.1395,
+      "step": 6380500
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.925228086259385e-05,
+      "loss": 1.1564,
+      "step": 6381000
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.925018509696441e-05,
+      "loss": 1.1556,
+      "step": 6381500
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.924808513140384e-05,
+      "loss": 1.1532,
+      "step": 6382000
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.9245985165843278e-05,
+      "loss": 1.1614,
+      "step": 6382500
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.9243893600144958e-05,
+      "loss": 1.145,
+      "step": 6383000
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.924179363458439e-05,
+      "loss": 1.1421,
+      "step": 6383500
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.9239693669023828e-05,
+      "loss": 1.1744,
+      "step": 6384000
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.9237593703463265e-05,
+      "loss": 1.1265,
+      "step": 6384500
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.92354937379027e-05,
+      "loss": 1.1366,
+      "step": 6385000
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.9233393772342135e-05,
+      "loss": 1.1851,
+      "step": 6385500
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.9231293806781572e-05,
+      "loss": 1.1504,
+      "step": 6386000
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.9229193841221006e-05,
+      "loss": 1.1536,
+      "step": 6386500
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.9227093875660436e-05,
+      "loss": 1.1324,
+      "step": 6387000
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.9224993910099873e-05,
+      "loss": 1.1473,
+      "step": 6387500
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.9222898144470433e-05,
+      "loss": 1.1204,
+      "step": 6388000
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.9220798178909866e-05,
+      "loss": 1.1727,
+      "step": 6388500
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.9218698213349297e-05,
+      "loss": 1.137,
+      "step": 6389000
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.9216598247788733e-05,
+      "loss": 1.1289,
+      "step": 6389500
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.921449828222817e-05,
+      "loss": 1.1758,
+      "step": 6390000
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.921240251659873e-05,
+      "loss": 1.1721,
+      "step": 6390500
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.9210302551038164e-05,
+      "loss": 1.1607,
+      "step": 6391000
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.9208202585477594e-05,
+      "loss": 1.1545,
+      "step": 6391500
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.920610261991703e-05,
+      "loss": 1.1675,
+      "step": 6392000
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.9204002654356468e-05,
+      "loss": 1.1299,
+      "step": 6392500
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.9201906888727028e-05,
+      "loss": 1.1492,
+      "step": 6393000
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.919980692316646e-05,
+      "loss": 1.1301,
+      "step": 6393500
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.919770695760589e-05,
+      "loss": 1.1461,
+      "step": 6394000
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.919560699204533e-05,
+      "loss": 1.1367,
+      "step": 6394500
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.9193507026484762e-05,
+      "loss": 1.1604,
+      "step": 6395000
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.91914070609242e-05,
+      "loss": 1.1456,
+      "step": 6395500
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.918931549522588e-05,
+      "loss": 1.1273,
+      "step": 6396000
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.9187215529665313e-05,
+      "loss": 1.1729,
+      "step": 6396500
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.918511556410475e-05,
+      "loss": 1.1879,
+      "step": 6397000
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.9183015598544186e-05,
+      "loss": 1.1485,
+      "step": 6397500
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.918091563298362e-05,
+      "loss": 1.1499,
+      "step": 6398000
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.917881566742305e-05,
+      "loss": 1.1123,
+      "step": 6398500
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.9176715701862487e-05,
+      "loss": 1.1355,
+      "step": 6399000
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.9174615736301924e-05,
+      "loss": 1.1486,
+      "step": 6399500
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.9172515770741357e-05,
+      "loss": 1.1244,
+      "step": 6400000
+    },
+    {
+      "epoch": 3.84,
+      "eval_loss": 1.1107254028320312,
+      "eval_runtime": 1100.5378,
+      "eval_samples_per_second": 478.602,
+      "eval_steps_per_second": 79.767,
+      "step": 6400000
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.9170420005111918e-05,
+      "loss": 1.1573,
+      "step": 6400500
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.9168320039551348e-05,
+      "loss": 1.1234,
+      "step": 6401000
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.9166220073990784e-05,
+      "loss": 1.144,
+      "step": 6401500
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.9164124308361345e-05,
+      "loss": 1.1565,
+      "step": 6402000
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.916202434280078e-05,
+      "loss": 1.152,
+      "step": 6402500
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.9159924377240215e-05,
+      "loss": 1.1448,
+      "step": 6403000
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.9157824411679645e-05,
+      "loss": 1.1413,
+      "step": 6403500
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.9155724446119082e-05,
+      "loss": 1.1792,
+      "step": 6404000
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.9153624480558515e-05,
+      "loss": 1.1571,
+      "step": 6404500
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.9151524514997952e-05,
+      "loss": 1.1721,
+      "step": 6405000
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.914942454943739e-05,
+      "loss": 1.1198,
+      "step": 6405500
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.9147324583876823e-05,
+      "loss": 1.1264,
+      "step": 6406000
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.914522461831626e-05,
+      "loss": 1.1476,
+      "step": 6406500
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.9143124652755696e-05,
+      "loss": 1.1388,
+      "step": 6407000
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.914102888712625e-05,
+      "loss": 1.1306,
+      "step": 6407500
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.9138928921565687e-05,
+      "loss": 1.1588,
+      "step": 6408000
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.913682895600512e-05,
+      "loss": 1.1406,
+      "step": 6408500
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.9134728990444557e-05,
+      "loss": 1.1738,
+      "step": 6409000
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.9132629024883994e-05,
+      "loss": 1.1782,
+      "step": 6409500
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.9130529059323427e-05,
+      "loss": 1.119,
+      "step": 6410000
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.9128429093762864e-05,
+      "loss": 1.1334,
+      "step": 6410500
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.91263291282023e-05,
+      "loss": 1.1608,
+      "step": 6411000
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.9124233362572855e-05,
+      "loss": 1.1497,
+      "step": 6411500
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.912213339701229e-05,
+      "loss": 1.1336,
+      "step": 6412000
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.9120033431451725e-05,
+      "loss": 1.1536,
+      "step": 6412500
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.911793766582228e-05,
+      "loss": 1.1429,
+      "step": 6413000
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.9115837700261715e-05,
+      "loss": 1.1292,
+      "step": 6413500
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.9113737734701152e-05,
+      "loss": 1.1498,
+      "step": 6414000
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.9111637769140586e-05,
+      "loss": 1.1494,
+      "step": 6414500
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.9109537803580022e-05,
+      "loss": 1.1556,
+      "step": 6415000
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.9107442037950576e-05,
+      "loss": 1.1526,
+      "step": 6415500
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.9105342072390013e-05,
+      "loss": 1.151,
+      "step": 6416000
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.910324210682945e-05,
+      "loss": 1.135,
+      "step": 6416500
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.9101142141268883e-05,
+      "loss": 1.1464,
+      "step": 6417000
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.909904217570832e-05,
+      "loss": 1.1554,
+      "step": 6417500
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.9096942210147757e-05,
+      "loss": 1.1383,
+      "step": 6418000
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.9094842244587187e-05,
+      "loss": 1.1781,
+      "step": 6418500
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.909274227902662e-05,
+      "loss": 1.1394,
+      "step": 6419000
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.909064651339718e-05,
+      "loss": 1.1346,
+      "step": 6419500
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.9088546547836618e-05,
+      "loss": 1.153,
+      "step": 6420000
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.9086446582276054e-05,
+      "loss": 1.1331,
+      "step": 6420500
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.908434661671548e-05,
+      "loss": 1.1259,
+      "step": 6421000
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.908225085108604e-05,
+      "loss": 1.1542,
+      "step": 6421500
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.908015088552548e-05,
+      "loss": 1.136,
+      "step": 6422000
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.9078050919964915e-05,
+      "loss": 1.1283,
+      "step": 6422500
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.9075950954404352e-05,
+      "loss": 1.1524,
+      "step": 6423000
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.907385098884378e-05,
+      "loss": 1.1465,
+      "step": 6423500
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.9071751023283216e-05,
+      "loss": 1.1425,
+      "step": 6424000
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.9069651057722652e-05,
+      "loss": 1.1341,
+      "step": 6424500
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.9067555292093213e-05,
+      "loss": 1.1486,
+      "step": 6425000
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.9065455326532643e-05,
+      "loss": 1.1451,
+      "step": 6425500
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.9063355360972076e-05,
+      "loss": 1.1687,
+      "step": 6426000
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.9061255395411513e-05,
+      "loss": 1.1163,
+      "step": 6426500
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.905915542985095e-05,
+      "loss": 1.1413,
+      "step": 6427000
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.9057055464290383e-05,
+      "loss": 1.1364,
+      "step": 6427500
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.905495549872982e-05,
+      "loss": 1.1197,
+      "step": 6428000
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.9052859733100374e-05,
+      "loss": 1.1473,
+      "step": 6428500
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.905075976753981e-05,
+      "loss": 1.1556,
+      "step": 6429000
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.9048659801979248e-05,
+      "loss": 1.1462,
+      "step": 6429500
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.904655983641868e-05,
+      "loss": 1.1499,
+      "step": 6430000
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.9044464070789235e-05,
+      "loss": 1.1586,
+      "step": 6430500
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.904236410522867e-05,
+      "loss": 1.147,
+      "step": 6431000
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.904026413966811e-05,
+      "loss": 1.1532,
+      "step": 6431500
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.9038164174107542e-05,
+      "loss": 1.1314,
+      "step": 6432000
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.903606420854698e-05,
+      "loss": 1.1111,
+      "step": 6432500
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.9033968442917532e-05,
+      "loss": 1.1443,
+      "step": 6433000
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.903186847735697e-05,
+      "loss": 1.1481,
+      "step": 6433500
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.9029768511796406e-05,
+      "loss": 1.1732,
+      "step": 6434000
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.902766854623584e-05,
+      "loss": 1.1428,
+      "step": 6434500
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.9025572780606393e-05,
+      "loss": 1.1729,
+      "step": 6435000
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.902347281504583e-05,
+      "loss": 1.165,
+      "step": 6435500
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.9021372849485267e-05,
+      "loss": 1.1532,
+      "step": 6436000
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.9019272883924703e-05,
+      "loss": 1.137,
+      "step": 6436500
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.9017177118295264e-05,
+      "loss": 1.1305,
+      "step": 6437000
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.901507715273469e-05,
+      "loss": 1.178,
+      "step": 6437500
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.9012977187174127e-05,
+      "loss": 1.1525,
+      "step": 6438000
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.9010877221613564e-05,
+      "loss": 1.1507,
+      "step": 6438500
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.9008777256052998e-05,
+      "loss": 1.143,
+      "step": 6439000
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.9006677290492434e-05,
+      "loss": 1.1121,
+      "step": 6439500
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.900457732493187e-05,
+      "loss": 1.1336,
+      "step": 6440000
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.9002477359371305e-05,
+      "loss": 1.1587,
+      "step": 6440500
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.9000385793672985e-05,
+      "loss": 1.1389,
+      "step": 6441000
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.8998285828112422e-05,
+      "loss": 1.1539,
+      "step": 6441500
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.8996185862551856e-05,
+      "loss": 1.1405,
+      "step": 6442000
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.8994085896991286e-05,
+      "loss": 1.1906,
+      "step": 6442500
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.8991985931430722e-05,
+      "loss": 1.1849,
+      "step": 6443000
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.8989890165801283e-05,
+      "loss": 1.1908,
+      "step": 6443500
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.898779020024072e-05,
+      "loss": 1.1568,
+      "step": 6444000
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.8985690234680146e-05,
+      "loss": 1.1257,
+      "step": 6444500
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.8983590269119583e-05,
+      "loss": 1.1728,
+      "step": 6445000
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.8981494503490144e-05,
+      "loss": 1.1386,
+      "step": 6445500
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.897939453792958e-05,
+      "loss": 1.1422,
+      "step": 6446000
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.8977294572369014e-05,
+      "loss": 1.1376,
+      "step": 6446500
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8975194606808444e-05,
+      "loss": 1.1537,
+      "step": 6447000
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.897309464124788e-05,
+      "loss": 1.1788,
+      "step": 6447500
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.897099887561844e-05,
+      "loss": 1.1385,
+      "step": 6448000
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8968903109988995e-05,
+      "loss": 1.1402,
+      "step": 6448500
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.896680314442843e-05,
+      "loss": 1.1382,
+      "step": 6449000
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.896470317886787e-05,
+      "loss": 1.1366,
+      "step": 6449500
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8962603213307302e-05,
+      "loss": 1.1474,
+      "step": 6450000
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8960507447677855e-05,
+      "loss": 1.1267,
+      "step": 6450500
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8958407482117292e-05,
+      "loss": 1.1365,
+      "step": 6451000
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.895630751655673e-05,
+      "loss": 1.155,
+      "step": 6451500
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8954207550996163e-05,
+      "loss": 1.1758,
+      "step": 6452000
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.89521075854356e-05,
+      "loss": 1.1655,
+      "step": 6452500
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8950007619875036e-05,
+      "loss": 1.1277,
+      "step": 6453000
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.894790765431447e-05,
+      "loss": 1.139,
+      "step": 6453500
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8945807688753907e-05,
+      "loss": 1.1529,
+      "step": 6454000
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8943707723193337e-05,
+      "loss": 1.1426,
+      "step": 6454500
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8941616157495017e-05,
+      "loss": 1.157,
+      "step": 6455000
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.893951619193445e-05,
+      "loss": 1.159,
+      "step": 6455500
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8937416226373887e-05,
+      "loss": 1.1571,
+      "step": 6456000
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8935316260813324e-05,
+      "loss": 1.1744,
+      "step": 6456500
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8933216295252758e-05,
+      "loss": 1.1638,
+      "step": 6457000
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8931116329692195e-05,
+      "loss": 1.1291,
+      "step": 6457500
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.892901636413163e-05,
+      "loss": 1.1318,
+      "step": 6458000
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8926916398571065e-05,
+      "loss": 1.1379,
+      "step": 6458500
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8924816433010495e-05,
+      "loss": 1.1387,
+      "step": 6459000
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8922720667381055e-05,
+      "loss": 1.1413,
+      "step": 6459500
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8920620701820492e-05,
+      "loss": 1.127,
+      "step": 6460000
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8918520736259926e-05,
+      "loss": 1.167,
+      "step": 6460500
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8916420770699362e-05,
+      "loss": 1.1246,
+      "step": 6461000
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8914320805138793e-05,
+      "loss": 1.1271,
+      "step": 6461500
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.891222083957823e-05,
+      "loss": 1.1196,
+      "step": 6462000
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.891012507394879e-05,
+      "loss": 1.1589,
+      "step": 6462500
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8908025108388223e-05,
+      "loss": 1.1359,
+      "step": 6463000
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.890592514282766e-05,
+      "loss": 1.1473,
+      "step": 6463500
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.890382517726709e-05,
+      "loss": 1.16,
+      "step": 6464000
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.8901725211706527e-05,
+      "loss": 1.1472,
+      "step": 6464500
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.8899629446077087e-05,
+      "loss": 1.1498,
+      "step": 6465000
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.889752948051652e-05,
+      "loss": 1.1576,
+      "step": 6465500
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.889542951495595e-05,
+      "loss": 1.1406,
+      "step": 6466000
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.8893329549395388e-05,
+      "loss": 1.1619,
+      "step": 6466500
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.889122958383482e-05,
+      "loss": 1.1479,
+      "step": 6467000
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.8889129618274258e-05,
+      "loss": 1.1235,
+      "step": 6467500
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.8887029652713695e-05,
+      "loss": 1.1605,
+      "step": 6468000
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.8884929687153132e-05,
+      "loss": 1.1235,
+      "step": 6468500
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.8882833921523685e-05,
+      "loss": 1.1486,
+      "step": 6469000
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.888073395596312e-05,
+      "loss": 1.1016,
+      "step": 6469500
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.887863819033368e-05,
+      "loss": 1.1495,
+      "step": 6470000
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.8876538224773116e-05,
+      "loss": 1.1418,
+      "step": 6470500
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.8874438259212546e-05,
+      "loss": 1.1416,
+      "step": 6471000
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.8872338293651983e-05,
+      "loss": 1.1443,
+      "step": 6471500
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.8870238328091416e-05,
+      "loss": 1.1518,
+      "step": 6472000
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.8868138362530853e-05,
+      "loss": 1.1398,
+      "step": 6472500
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.886603839697029e-05,
+      "loss": 1.1661,
+      "step": 6473000
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.8863938431409723e-05,
+      "loss": 1.1479,
+      "step": 6473500
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.8861842665780277e-05,
+      "loss": 1.1499,
+      "step": 6474000
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.885974690015084e-05,
+      "loss": 1.1365,
+      "step": 6474500
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.8857646934590274e-05,
+      "loss": 1.1503,
+      "step": 6475000
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.8855546969029704e-05,
+      "loss": 1.1549,
+      "step": 6475500
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.885344700346914e-05,
+      "loss": 1.1305,
+      "step": 6476000
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.8851347037908575e-05,
+      "loss": 1.1436,
+      "step": 6476500
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.8849251272279135e-05,
+      "loss": 1.1596,
+      "step": 6477000
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.8847151306718572e-05,
+      "loss": 1.1337,
+      "step": 6477500
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.8845051341158002e-05,
+      "loss": 1.1407,
+      "step": 6478000
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.884295137559744e-05,
+      "loss": 1.1245,
+      "step": 6478500
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.8840851410036872e-05,
+      "loss": 1.146,
+      "step": 6479000
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.8838755644407433e-05,
+      "loss": 1.1464,
+      "step": 6479500
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.883665567884687e-05,
+      "loss": 1.1535,
+      "step": 6480000
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.88345557132863e-05,
+      "loss": 1.1589,
+      "step": 6480500
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.8832455747725736e-05,
+      "loss": 1.1364,
+      "step": 6481000
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.883035578216517e-05,
+      "loss": 1.1205,
+      "step": 6481500
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.882826001653573e-05,
+      "loss": 1.1261,
+      "step": 6482000
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.8826160050975167e-05,
+      "loss": 1.1695,
+      "step": 6482500
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.8824060085414597e-05,
+      "loss": 1.1242,
+      "step": 6483000
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.882196011985403e-05,
+      "loss": 1.143,
+      "step": 6483500
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.8819860154293467e-05,
+      "loss": 1.1313,
+      "step": 6484000
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.8817764388664028e-05,
+      "loss": 1.1538,
+      "step": 6484500
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.8815664423103465e-05,
+      "loss": 1.1352,
+      "step": 6485000
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.8813564457542895e-05,
+      "loss": 1.1497,
+      "step": 6485500
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.8811464491982328e-05,
+      "loss": 1.1356,
+      "step": 6486000
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.880936872635289e-05,
+      "loss": 1.1486,
+      "step": 6486500
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.8807268760792325e-05,
+      "loss": 1.118,
+      "step": 6487000
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.8805168795231755e-05,
+      "loss": 1.1366,
+      "step": 6487500
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.8803068829671192e-05,
+      "loss": 1.1601,
+      "step": 6488000
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.8800968864110626e-05,
+      "loss": 1.1001,
+      "step": 6488500
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.8798868898550062e-05,
+      "loss": 1.1464,
+      "step": 6489000
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.8796773132920623e-05,
+      "loss": 1.1466,
+      "step": 6489500
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.8794673167360053e-05,
+      "loss": 1.1205,
+      "step": 6490000
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.8792573201799486e-05,
+      "loss": 1.1337,
+      "step": 6490500
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.8790473236238923e-05,
+      "loss": 1.1529,
+      "step": 6491000
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.878837327067836e-05,
+      "loss": 1.1405,
+      "step": 6491500
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.878627750504892e-05,
+      "loss": 1.1446,
+      "step": 6492000
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.878417753948835e-05,
+      "loss": 1.1416,
+      "step": 6492500
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.8782077573927784e-05,
+      "loss": 1.1228,
+      "step": 6493000
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.877997760836722e-05,
+      "loss": 1.1209,
+      "step": 6493500
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.8777877642806658e-05,
+      "loss": 1.169,
+      "step": 6494000
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.8775781877177218e-05,
+      "loss": 1.1356,
+      "step": 6494500
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.877368611154777e-05,
+      "loss": 1.1532,
+      "step": 6495000
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.877158614598721e-05,
+      "loss": 1.147,
+      "step": 6495500
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.8769486180426642e-05,
+      "loss": 1.1345,
+      "step": 6496000
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.876738621486608e-05,
+      "loss": 1.1453,
+      "step": 6496500
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.876528624930551e-05,
+      "loss": 1.1289,
+      "step": 6497000
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.8763186283744942e-05,
+      "loss": 1.1727,
+      "step": 6497500
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.876108631818438e-05,
+      "loss": 1.1603,
+      "step": 6498000
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.8758986352623816e-05,
+      "loss": 1.1164,
+      "step": 6498500
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.875688638706325e-05,
+      "loss": 1.1503,
+      "step": 6499000
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.8754790621433806e-05,
+      "loss": 1.1488,
+      "step": 6499500
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.875269065587324e-05,
+      "loss": 1.1639,
+      "step": 6500000
+    },
+    {
+      "epoch": 3.9,
+      "eval_loss": 1.1070631742477417,
+      "eval_runtime": 1102.123,
+      "eval_samples_per_second": 477.914,
+      "eval_steps_per_second": 79.653,
+      "step": 6500000
+    }
+  ],
+  "max_steps": 13343552,
+  "num_train_epochs": 8,
+  "total_flos": 3.442361330450166e+18,
+  "trial_name": null,
+  "trial_params": null
+}