diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -2,312 +2,7146 @@
   "best_metric": null,
   "best_model_checkpoint": null,
   "epoch": 3.0,
-  "global_step": 24495,
+  "global_step": 593715,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.06,
-      "learning_rate": 4.898550724637682e-05,
-      "loss": 5.4391,
+      "epoch": 0.0,
+      "learning_rate": 6.994199237007655e-05,
+      "loss": 6.859,
       "step": 500
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.79648907940396e-05,
-      "loss": 3.2254,
+      "epoch": 0.01,
+      "learning_rate": 6.988304152665841e-05,
+      "loss": 5.2814,
       "step": 1000
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.694427434170239e-05,
-      "loss": 2.7119,
+      "epoch": 0.01,
+      "learning_rate": 6.982409068324027e-05,
+      "loss": 4.4487,
       "step": 1500
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 4.592365788936518e-05,
-      "loss": 2.4259,
+      "epoch": 0.01,
+      "learning_rate": 6.976513983982213e-05,
+      "loss": 4.2191,
       "step": 2000
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 4.490304143702796e-05,
-      "loss": 2.206,
+      "epoch": 0.01,
+      "learning_rate": 6.970618899640399e-05,
+      "loss": 3.8599,
       "step": 2500
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 4.3884466217595436e-05,
-      "loss": 2.1015,
+      "epoch": 0.02,
+      "learning_rate": 6.964723815298586e-05,
+      "loss": 3.6392,
       "step": 3000
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 4.286384976525822e-05,
-      "loss": 1.99,
+      "epoch": 0.02,
+      "learning_rate": 6.958828730956772e-05,
+      "loss": 3.4509,
       "step": 3500
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 4.184527454582568e-05,
-      "loss": 1.9223,
+      "epoch": 0.02,
+      "learning_rate": 6.952933646614957e-05,
+      "loss": 3.3074,
       "step": 4000
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 4.082465809348847e-05,
-      "loss": 1.8566,
+      "epoch": 0.02,
+      "learning_rate": 6.947050352441827e-05,
+      "loss": 3.3332,
       "step": 4500
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 3.9804041641151255e-05,
-      "loss": 1.7873,
+      "epoch": 0.03,
+      "learning_rate": 6.941167058268697e-05,
+      "loss": 3.2866,
       "step": 5000
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 3.8783425188814046e-05,
-      "loss": 1.7276,
+      "epoch": 0.03,
+      "learning_rate": 6.935271973926883e-05,
+      "loss": 3.0393,
       "step": 5500
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 3.776280873647684e-05,
-      "loss": 1.6822,
+      "epoch": 0.03,
+      "learning_rate": 6.92937688958507e-05,
+      "loss": 3.0049,
       "step": 6000
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 3.674219228413962e-05,
-      "loss": 1.6448,
+      "epoch": 0.03,
+      "learning_rate": 6.923481805243257e-05,
+      "loss": 2.9204,
       "step": 6500
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 3.572157583180241e-05,
-      "loss": 1.5829,
+      "epoch": 0.04,
+      "learning_rate": 6.917586720901443e-05,
+      "loss": 2.8666,
       "step": 7000
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 3.47009593794652e-05,
-      "loss": 1.5763,
+      "epoch": 0.04,
+      "learning_rate": 6.911691636559629e-05,
+      "loss": 2.7032,
       "step": 7500
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 3.368034292712799e-05,
-      "loss": 1.5781,
+      "epoch": 0.04,
+      "learning_rate": 6.905796552217815e-05,
+      "loss": 2.7154,
       "step": 8000
     },
     {
-      "epoch": 1.04,
-      "learning_rate": 3.265972647479077e-05,
-      "loss": 1.5427,
+      "epoch": 0.04,
+      "learning_rate": 6.899913258044685e-05,
+      "loss": 2.7145,
       "step": 8500
     },
     {
-      "epoch": 1.1,
-      "learning_rate": 3.1639110022453563e-05,
-      "loss": 1.5025,
+      "epoch": 0.05,
+      "learning_rate": 6.89401817370287e-05,
+      "loss": 2.7003,
       "step": 9000
     },
     {
-      "epoch": 1.16,
-      "learning_rate": 3.0618493570116355e-05,
-      "loss": 1.4859,
+      "epoch": 0.05,
+      "learning_rate": 6.888123089361057e-05,
+      "loss": 2.699,
       "step": 9500
     },
     {
-      "epoch": 1.22,
-      "learning_rate": 2.959787711777914e-05,
-      "loss": 1.4483,
+      "epoch": 0.05,
+      "learning_rate": 6.882228005019243e-05,
+      "loss": 2.6719,
       "step": 10000
     },
     {
-      "epoch": 1.29,
-      "learning_rate": 2.857726066544193e-05,
-      "loss": 1.4282,
+      "epoch": 0.05,
+      "learning_rate": 6.876332920677429e-05,
+      "loss": 2.6672,
       "step": 10500
     },
     {
-      "epoch": 1.35,
-      "learning_rate": 2.7558685446009393e-05,
-      "loss": 1.4401,
+      "epoch": 0.06,
+      "learning_rate": 6.870437836335615e-05,
+      "loss": 2.616,
       "step": 11000
     },
     {
-      "epoch": 1.41,
-      "learning_rate": 2.6538068993672177e-05,
-      "loss": 1.3864,
+      "epoch": 0.06,
+      "learning_rate": 6.864542751993801e-05,
+      "loss": 2.6025,
       "step": 11500
     },
     {
-      "epoch": 1.47,
-      "learning_rate": 2.5517452541334968e-05,
-      "loss": 1.4,
+      "epoch": 0.06,
+      "learning_rate": 6.858647667651987e-05,
+      "loss": 2.4387,
       "step": 12000
     },
     {
-      "epoch": 1.53,
-      "learning_rate": 2.4496836088997756e-05,
-      "loss": 1.3831,
+      "epoch": 0.06,
+      "learning_rate": 6.852752583310173e-05,
+      "loss": 2.3816,
       "step": 12500
     },
     {
-      "epoch": 1.59,
-      "learning_rate": 2.3476219636660544e-05,
-      "loss": 1.3708,
+      "epoch": 0.07,
+      "learning_rate": 6.84685749896836e-05,
+      "loss": 2.4366,
       "step": 13000
     },
     {
-      "epoch": 1.65,
-      "learning_rate": 2.245560318432333e-05,
-      "loss": 1.3737,
+      "epoch": 0.07,
+      "learning_rate": 6.840962414626546e-05,
+      "loss": 2.4259,
       "step": 13500
     },
     {
-      "epoch": 1.71,
-      "learning_rate": 2.143498673198612e-05,
-      "loss": 1.3142,
+      "epoch": 0.07,
+      "learning_rate": 6.835067330284732e-05,
+      "loss": 2.4172,
       "step": 14000
     },
     {
-      "epoch": 1.78,
-      "learning_rate": 2.041437027964891e-05,
-      "loss": 1.333,
+      "epoch": 0.07,
+      "learning_rate": 6.829172245942918e-05,
+      "loss": 2.3022,
       "step": 14500
     },
     {
-      "epoch": 1.84,
-      "learning_rate": 1.939579506021637e-05,
-      "loss": 1.3242,
+      "epoch": 0.08,
+      "learning_rate": 6.823277161601104e-05,
+      "loss": 2.2821,
       "step": 15000
     },
     {
-      "epoch": 1.9,
-      "learning_rate": 1.837517860787916e-05,
-      "loss": 1.3056,
+      "epoch": 0.08,
+      "learning_rate": 6.817393867427974e-05,
+      "loss": 2.2826,
       "step": 15500
     },
     {
-      "epoch": 1.96,
-      "learning_rate": 1.7356603388446624e-05,
-      "loss": 1.2832,
+      "epoch": 0.08,
+      "learning_rate": 6.81149878308616e-05,
+      "loss": 2.2232,
       "step": 16000
     },
     {
-      "epoch": 2.02,
-      "learning_rate": 1.633598693610941e-05,
-      "loss": 1.2872,
+      "epoch": 0.08,
+      "learning_rate": 6.805603698744346e-05,
+      "loss": 2.192,
       "step": 16500
     },
     {
-      "epoch": 2.08,
-      "learning_rate": 1.53153704837722e-05,
-      "loss": 1.2703,
+      "epoch": 0.09,
+      "learning_rate": 6.799708614402532e-05,
+      "loss": 2.2405,
       "step": 17000
     },
     {
-      "epoch": 2.14,
-      "learning_rate": 1.4294754031434987e-05,
-      "loss": 1.261,
+      "epoch": 0.09,
+      "learning_rate": 6.793813530060718e-05,
+      "loss": 2.1968,
       "step": 17500
     },
     {
-      "epoch": 2.2,
-      "learning_rate": 1.3274137579097776e-05,
-      "loss": 1.2512,
+      "epoch": 0.09,
+      "learning_rate": 6.787918445718906e-05,
+      "loss": 2.2894,
       "step": 18000
     },
     {
-      "epoch": 2.27,
-      "learning_rate": 1.2253521126760564e-05,
-      "loss": 1.2212,
+      "epoch": 0.09,
+      "learning_rate": 6.782023361377092e-05,
+      "loss": 2.2866,
       "step": 18500
     },
     {
-      "epoch": 2.33,
-      "learning_rate": 1.1232904674423353e-05,
-      "loss": 1.2351,
+      "epoch": 0.1,
+      "learning_rate": 6.776128277035276e-05,
+      "loss": 2.1333,
       "step": 19000
     },
     {
-      "epoch": 2.39,
-      "learning_rate": 1.0212288222086141e-05,
-      "loss": 1.2337,
+      "epoch": 0.1,
+      "learning_rate": 6.770233192693464e-05,
+      "loss": 2.0818,
       "step": 19500
     },
     {
-      "epoch": 2.45,
-      "learning_rate": 9.191671769748929e-06,
-      "loss": 1.2258,
+      "epoch": 0.1,
+      "learning_rate": 6.76433810835165e-05,
+      "loss": 2.0744,
       "step": 20000
     },
     {
-      "epoch": 2.51,
-      "learning_rate": 8.171055317411718e-06,
-      "loss": 1.2216,
+      "epoch": 0.1,
+      "learning_rate": 6.758443024009836e-05,
+      "loss": 2.2706,
       "step": 20500
     },
     {
-      "epoch": 2.57,
-      "learning_rate": 7.150438865074505e-06,
-      "loss": 1.2057,
+      "epoch": 0.11,
+      "learning_rate": 6.752547939668022e-05,
+      "loss": 2.1421,
       "step": 21000
     },
     {
-      "epoch": 2.63,
-      "learning_rate": 6.129822412737294e-06,
-      "loss": 1.2135,
+      "epoch": 0.11,
+      "learning_rate": 6.746652855326208e-05,
+      "loss": 2.1903,
       "step": 21500
     },
     {
-      "epoch": 2.69,
-      "learning_rate": 5.111247193304756e-06,
-      "loss": 1.2054,
+      "epoch": 0.11,
+      "learning_rate": 6.740769561153078e-05,
+      "loss": 2.0703,
       "step": 22000
     },
     {
-      "epoch": 2.76,
-      "learning_rate": 4.090630740967544e-06,
-      "loss": 1.2021,
+      "epoch": 0.11,
+      "learning_rate": 6.734874476811264e-05,
+      "loss": 2.0504,
       "step": 22500
     },
     {
-      "epoch": 2.82,
-      "learning_rate": 3.0700142886303327e-06,
-      "loss": 1.1868,
+      "epoch": 0.12,
+      "learning_rate": 6.728991182638134e-05,
+      "loss": 2.144,
       "step": 23000
     },
     {
-      "epoch": 2.88,
-      "learning_rate": 2.049397836293121e-06,
-      "loss": 1.1735,
+      "epoch": 0.12,
+      "learning_rate": 6.72309609829632e-05,
+      "loss": 2.0276,
       "step": 23500
     },
     {
-      "epoch": 2.94,
-      "learning_rate": 1.0308226168605838e-06,
-      "loss": 1.2107,
+      "epoch": 0.12,
+      "learning_rate": 6.717201013954506e-05,
+      "loss": 1.9879,
       "step": 24000
     },
+    {
+      "epoch": 0.12,
+      "learning_rate": 6.711305929612692e-05,
+      "loss": 2.0549,
+      "step": 24500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 6.705410845270878e-05,
+      "loss": 2.0771,
+      "step": 25000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 6.699515760929065e-05,
+      "loss": 1.9698,
+      "step": 25500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 6.693644256924618e-05,
+      "loss": 1.9929,
+      "step": 26000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 6.687749172582804e-05,
+      "loss": 2.0061,
+      "step": 26500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 6.68185408824099e-05,
+      "loss": 1.955,
+      "step": 27000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 6.675959003899176e-05,
+      "loss": 2.0245,
+      "step": 27500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 6.670063919557362e-05,
+      "loss": 1.9558,
+      "step": 28000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 6.66416883521555e-05,
+      "loss": 1.996,
+      "step": 28500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.65828554104242e-05,
+      "loss": 1.9741,
+      "step": 29000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.652390456700604e-05,
+      "loss": 1.9904,
+      "step": 29500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.646507162527474e-05,
+      "loss": 1.9764,
+      "step": 30000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.64061207818566e-05,
+      "loss": 1.9839,
+      "step": 30500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.634716993843846e-05,
+      "loss": 1.8744,
+      "step": 31000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.628821909502034e-05,
+      "loss": 1.9974,
+      "step": 31500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.62292682516022e-05,
+      "loss": 1.9997,
+      "step": 32000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.617031740818406e-05,
+      "loss": 1.9917,
+      "step": 32500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.611148446645274e-05,
+      "loss": 1.9973,
+      "step": 33000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.605253362303462e-05,
+      "loss": 1.8879,
+      "step": 33500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.599358277961648e-05,
+      "loss": 1.9089,
+      "step": 34000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.593463193619834e-05,
+      "loss": 2.0488,
+      "step": 34500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.58756810927802e-05,
+      "loss": 1.7955,
+      "step": 35000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.581673024936206e-05,
+      "loss": 1.8981,
+      "step": 35500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.575777940594392e-05,
+      "loss": 1.9835,
+      "step": 36000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.56988285625258e-05,
+      "loss": 1.8466,
+      "step": 36500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.563987771910764e-05,
+      "loss": 1.8713,
+      "step": 37000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.55809268756895e-05,
+      "loss": 1.8304,
+      "step": 37500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.552197603227137e-05,
+      "loss": 1.8735,
+      "step": 38000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.546302518885323e-05,
+      "loss": 1.8013,
+      "step": 38500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.54040743454351e-05,
+      "loss": 1.7514,
+      "step": 39000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.53452414037038e-05,
+      "loss": 1.8538,
+      "step": 39500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.528629056028566e-05,
+      "loss": 1.8968,
+      "step": 40000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.522745761855434e-05,
+      "loss": 1.8126,
+      "step": 40500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.516850677513622e-05,
+      "loss": 1.7436,
+      "step": 41000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.510955593171808e-05,
+      "loss": 1.8164,
+      "step": 41500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.505060508829994e-05,
+      "loss": 1.8852,
+      "step": 42000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.49916542448818e-05,
+      "loss": 1.7415,
+      "step": 42500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.493270340146366e-05,
+      "loss": 1.7809,
+      "step": 43000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.487375255804552e-05,
+      "loss": 1.8898,
+      "step": 43500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.481480171462739e-05,
+      "loss": 1.7151,
+      "step": 44000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.475585087120925e-05,
+      "loss": 1.821,
+      "step": 44500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.469713583116478e-05,
+      "loss": 1.872,
+      "step": 45000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.463818498774664e-05,
+      "loss": 1.8684,
+      "step": 45500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.45792341443285e-05,
+      "loss": 1.8248,
+      "step": 46000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.452028330091036e-05,
+      "loss": 1.7672,
+      "step": 46500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.446133245749223e-05,
+      "loss": 1.9444,
+      "step": 47000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.440238161407409e-05,
+      "loss": 1.8666,
+      "step": 47500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.434343077065595e-05,
+      "loss": 1.7939,
+      "step": 48000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.428459782892464e-05,
+      "loss": 1.7606,
+      "step": 48500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.422564698550651e-05,
+      "loss": 1.7548,
+      "step": 49000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.416669614208837e-05,
+      "loss": 1.8124,
+      "step": 49500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.410774529867023e-05,
+      "loss": 1.7206,
+      "step": 50000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.40487944552521e-05,
+      "loss": 1.7948,
+      "step": 50500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.398984361183395e-05,
+      "loss": 1.6771,
+      "step": 51000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.393089276841581e-05,
+      "loss": 1.6329,
+      "step": 51500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.387194192499767e-05,
+      "loss": 1.7638,
+      "step": 52000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.381299108157953e-05,
+      "loss": 1.7988,
+      "step": 52500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.37540402381614e-05,
+      "loss": 1.7728,
+      "step": 53000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.369508939474327e-05,
+      "loss": 1.7844,
+      "step": 53500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.363613855132513e-05,
+      "loss": 1.8386,
+      "step": 54000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.357718770790699e-05,
+      "loss": 1.8122,
+      "step": 54500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.351823686448885e-05,
+      "loss": 1.741,
+      "step": 55000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.345928602107071e-05,
+      "loss": 1.6201,
+      "step": 55500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.340033517765257e-05,
+      "loss": 1.6551,
+      "step": 56000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.334138433423443e-05,
+      "loss": 1.6896,
+      "step": 56500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.328243349081629e-05,
+      "loss": 1.6738,
+      "step": 57000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.322348264739815e-05,
+      "loss": 1.7253,
+      "step": 57500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.316453180398002e-05,
+      "loss": 1.6887,
+      "step": 58000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.310581676393555e-05,
+      "loss": 1.7441,
+      "step": 58500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.304698382220425e-05,
+      "loss": 1.6825,
+      "step": 59000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.298803297878611e-05,
+      "loss": 1.7008,
+      "step": 59500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.292908213536797e-05,
+      "loss": 1.7882,
+      "step": 60000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.287013129194983e-05,
+      "loss": 1.7428,
+      "step": 60500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.281118044853169e-05,
+      "loss": 1.6972,
+      "step": 61000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.275234750680039e-05,
+      "loss": 1.7379,
+      "step": 61500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.269339666338225e-05,
+      "loss": 1.5953,
+      "step": 62000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.263444581996413e-05,
+      "loss": 1.6816,
+      "step": 62500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.257549497654597e-05,
+      "loss": 1.7044,
+      "step": 63000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.251666203481467e-05,
+      "loss": 1.6617,
+      "step": 63500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.245771119139653e-05,
+      "loss": 1.7058,
+      "step": 64000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.239876034797841e-05,
+      "loss": 1.6949,
+      "step": 64500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.233980950456027e-05,
+      "loss": 1.6324,
+      "step": 65000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.228085866114213e-05,
+      "loss": 1.6314,
+      "step": 65500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.222190781772399e-05,
+      "loss": 1.7001,
+      "step": 66000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.216295697430585e-05,
+      "loss": 1.6162,
+      "step": 66500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.210400613088771e-05,
+      "loss": 1.6579,
+      "step": 67000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.204505528746957e-05,
+      "loss": 1.6837,
+      "step": 67500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.198622234573827e-05,
+      "loss": 1.6529,
+      "step": 68000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.192727150232013e-05,
+      "loss": 1.6706,
+      "step": 68500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.186832065890199e-05,
+      "loss": 1.6849,
+      "step": 69000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.180936981548386e-05,
+      "loss": 1.5789,
+      "step": 69500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.175053687375255e-05,
+      "loss": 1.6463,
+      "step": 70000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.169158603033441e-05,
+      "loss": 1.6165,
+      "step": 70500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.163263518691627e-05,
+      "loss": 1.6316,
+      "step": 71000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.157368434349813e-05,
+      "loss": 1.6751,
+      "step": 71500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.151473350008e-05,
+      "loss": 1.6096,
+      "step": 72000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.145578265666186e-05,
+      "loss": 1.6837,
+      "step": 72500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.139683181324373e-05,
+      "loss": 1.5968,
+      "step": 73000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.133799887151243e-05,
+      "loss": 1.7028,
+      "step": 73500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.127904802809427e-05,
+      "loss": 1.6685,
+      "step": 74000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.122009718467615e-05,
+      "loss": 1.669,
+      "step": 74500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.1161146341258e-05,
+      "loss": 1.557,
+      "step": 75000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.11023133995267e-05,
+      "loss": 1.592,
+      "step": 75500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.104336255610857e-05,
+      "loss": 1.6474,
+      "step": 76000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.098441171269043e-05,
+      "loss": 1.6673,
+      "step": 76500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.0925460869272294e-05,
+      "loss": 1.5011,
+      "step": 77000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.0866510025854154e-05,
+      "loss": 1.6295,
+      "step": 77500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.080755918243601e-05,
+      "loss": 1.5999,
+      "step": 78000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.0748608339017875e-05,
+      "loss": 1.607,
+      "step": 78500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.0689657495599735e-05,
+      "loss": 1.5911,
+      "step": 79000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.0630824553868435e-05,
+      "loss": 1.6111,
+      "step": 79500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.0571873710450295e-05,
+      "loss": 1.6162,
+      "step": 80000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.0512922867032156e-05,
+      "loss": 1.6114,
+      "step": 80500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.0454089925300856e-05,
+      "loss": 1.5766,
+      "step": 81000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.0395139081882716e-05,
+      "loss": 1.656,
+      "step": 81500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.0336188238464576e-05,
+      "loss": 1.544,
+      "step": 82000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.0277237395046437e-05,
+      "loss": 1.6107,
+      "step": 82500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.021840445331514e-05,
+      "loss": 1.6119,
+      "step": 83000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.0159453609897e-05,
+      "loss": 1.6393,
+      "step": 83500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.01006206681657e-05,
+      "loss": 1.6269,
+      "step": 84000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0041669824747565e-05,
+      "loss": 1.594,
+      "step": 84500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.998271898132942e-05,
+      "loss": 1.5672,
+      "step": 85000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.992376813791128e-05,
+      "loss": 1.595,
+      "step": 85500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.9864817294493145e-05,
+      "loss": 1.4975,
+      "step": 86000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.980598435276184e-05,
+      "loss": 1.6532,
+      "step": 86500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9747033509343706e-05,
+      "loss": 1.5118,
+      "step": 87000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9688082665925566e-05,
+      "loss": 1.504,
+      "step": 87500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.962913182250743e-05,
+      "loss": 1.5813,
+      "step": 88000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9570180979089286e-05,
+      "loss": 1.5667,
+      "step": 88500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.951123013567115e-05,
+      "loss": 1.5856,
+      "step": 89000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.945227929225301e-05,
+      "loss": 1.5393,
+      "step": 89500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9393328448834874e-05,
+      "loss": 1.5786,
+      "step": 90000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9334495507103574e-05,
+      "loss": 1.6104,
+      "step": 90500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9275544663685434e-05,
+      "loss": 1.5735,
+      "step": 91000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9216711721954135e-05,
+      "loss": 1.6001,
+      "step": 91500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.915776087853599e-05,
+      "loss": 1.6341,
+      "step": 92000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.909881003511785e-05,
+      "loss": 1.551,
+      "step": 92500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9039859191699715e-05,
+      "loss": 1.6586,
+      "step": 93000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8980908348281576e-05,
+      "loss": 1.594,
+      "step": 93500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.892195750486344e-05,
+      "loss": 1.5156,
+      "step": 94000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.88630066614453e-05,
+      "loss": 1.5512,
+      "step": 94500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.880405581802716e-05,
+      "loss": 1.5811,
+      "step": 95000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.874510497460903e-05,
+      "loss": 1.6201,
+      "step": 95500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.868627203287772e-05,
+      "loss": 1.5383,
+      "step": 96000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.862743909114642e-05,
+      "loss": 1.5245,
+      "step": 96500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.8568488247728284e-05,
+      "loss": 1.5338,
+      "step": 97000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.850965530599698e-05,
+      "loss": 1.5225,
+      "step": 97500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.8450704462578845e-05,
+      "loss": 1.5189,
+      "step": 98000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.8391871520847545e-05,
+      "loss": 1.5908,
+      "step": 98500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.83329206774294e-05,
+      "loss": 1.6402,
+      "step": 99000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.827396983401126e-05,
+      "loss": 1.5652,
+      "step": 99500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.8215018990593126e-05,
+      "loss": 1.517,
+      "step": 100000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.8156068147174986e-05,
+      "loss": 1.5143,
+      "step": 100500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.8097117303756846e-05,
+      "loss": 1.5966,
+      "step": 101000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.803816646033871e-05,
+      "loss": 1.5101,
+      "step": 101500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7979215616920574e-05,
+      "loss": 1.5415,
+      "step": 102000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7920264773502434e-05,
+      "loss": 1.5355,
+      "step": 102500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7861313930084294e-05,
+      "loss": 1.5454,
+      "step": 103000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7802363086666154e-05,
+      "loss": 1.5042,
+      "step": 103500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7743412243248014e-05,
+      "loss": 1.562,
+      "step": 104000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.768446139982988e-05,
+      "loss": 1.5272,
+      "step": 104500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7625628458098575e-05,
+      "loss": 1.5825,
+      "step": 105000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.756667761468044e-05,
+      "loss": 1.4985,
+      "step": 105500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.75077267712623e-05,
+      "loss": 1.4954,
+      "step": 106000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.744877592784417e-05,
+      "loss": 1.5467,
+      "step": 106500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.738982508442602e-05,
+      "loss": 1.5303,
+      "step": 107000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.733087424100788e-05,
+      "loss": 1.4862,
+      "step": 107500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.727192339758974e-05,
+      "loss": 1.5106,
+      "step": 108000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.721297255417161e-05,
+      "loss": 1.5593,
+      "step": 108500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.715402171075347e-05,
+      "loss": 1.5959,
+      "step": 109000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.709507086733534e-05,
+      "loss": 1.5636,
+      "step": 109500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.703623792560403e-05,
+      "loss": 1.5684,
+      "step": 110000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.697728708218589e-05,
+      "loss": 1.4454,
+      "step": 110500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.691833623876775e-05,
+      "loss": 1.5496,
+      "step": 111000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.685938539534961e-05,
+      "loss": 1.4763,
+      "step": 111500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.680055245361831e-05,
+      "loss": 1.5981,
+      "step": 112000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.674160161020018e-05,
+      "loss": 1.4676,
+      "step": 112500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.668288657015571e-05,
+      "loss": 1.6023,
+      "step": 113000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.662393572673757e-05,
+      "loss": 1.5173,
+      "step": 113500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.656498488331943e-05,
+      "loss": 1.5024,
+      "step": 114000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.650615194158813e-05,
+      "loss": 1.4905,
+      "step": 114500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6447201098169994e-05,
+      "loss": 1.5263,
+      "step": 115000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6388250254751854e-05,
+      "loss": 1.4874,
+      "step": 115500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6329299411333714e-05,
+      "loss": 1.4981,
+      "step": 116000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.627034856791558e-05,
+      "loss": 1.524,
+      "step": 116500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.621139772449744e-05,
+      "loss": 1.6569,
+      "step": 117000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6152446881079295e-05,
+      "loss": 1.4486,
+      "step": 117500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.609349603766116e-05,
+      "loss": 1.4926,
+      "step": 118000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.603454519424302e-05,
+      "loss": 1.5704,
+      "step": 118500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.597559435082488e-05,
+      "loss": 1.565,
+      "step": 119000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.591664350740675e-05,
+      "loss": 1.5233,
+      "step": 119500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.585769266398861e-05,
+      "loss": 1.4826,
+      "step": 120000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.579874182057047e-05,
+      "loss": 1.5609,
+      "step": 120500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.5739790977152336e-05,
+      "loss": 1.5396,
+      "step": 121000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.568095803542103e-05,
+      "loss": 1.4773,
+      "step": 121500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.562200719200289e-05,
+      "loss": 1.5906,
+      "step": 122000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.556305634858475e-05,
+      "loss": 1.619,
+      "step": 122500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.550410550516662e-05,
+      "loss": 1.4244,
+      "step": 123000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.544515466174848e-05,
+      "loss": 1.5547,
+      "step": 123500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.538620381833034e-05,
+      "loss": 1.3988,
+      "step": 124000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.5327252974912205e-05,
+      "loss": 1.4837,
+      "step": 124500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.5268302131494065e-05,
+      "loss": 1.5698,
+      "step": 125000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.520935128807592e-05,
+      "loss": 1.4586,
+      "step": 125500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.5150400444657786e-05,
+      "loss": 1.4794,
+      "step": 126000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.5091449601239646e-05,
+      "loss": 1.5044,
+      "step": 126500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.5032498757821506e-05,
+      "loss": 1.4507,
+      "step": 127000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.497354791440337e-05,
+      "loss": 1.348,
+      "step": 127500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.491459707098523e-05,
+      "loss": 1.5163,
+      "step": 128000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.485564622756709e-05,
+      "loss": 1.53,
+      "step": 128500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.479669538414896e-05,
+      "loss": 1.515,
+      "step": 129000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.473786244241765e-05,
+      "loss": 1.459,
+      "step": 129500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4678911598999514e-05,
+      "loss": 1.4868,
+      "step": 130000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4620078657268215e-05,
+      "loss": 1.4395,
+      "step": 130500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4561127813850075e-05,
+      "loss": 1.4272,
+      "step": 131000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4502176970431935e-05,
+      "loss": 1.4605,
+      "step": 131500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.44432261270138e-05,
+      "loss": 1.607,
+      "step": 132000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.438427528359566e-05,
+      "loss": 1.4993,
+      "step": 132500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.4325324440177516e-05,
+      "loss": 1.461,
+      "step": 133000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.426637359675938e-05,
+      "loss": 1.5128,
+      "step": 133500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.420742275334124e-05,
+      "loss": 1.3986,
+      "step": 134000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.414858981160994e-05,
+      "loss": 1.5693,
+      "step": 134500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.40896389681918e-05,
+      "loss": 1.4682,
+      "step": 135000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.403068812477367e-05,
+      "loss": 1.5161,
+      "step": 135500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.397173728135553e-05,
+      "loss": 1.4837,
+      "step": 136000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.3912904339624224e-05,
+      "loss": 1.4552,
+      "step": 136500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.3853953496206084e-05,
+      "loss": 1.4588,
+      "step": 137000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.3795002652787945e-05,
+      "loss": 1.462,
+      "step": 137500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.373605180936981e-05,
+      "loss": 1.511,
+      "step": 138000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.367710096595167e-05,
+      "loss": 1.3909,
+      "step": 138500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.361826802422037e-05,
+      "loss": 1.456,
+      "step": 139000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.355931718080223e-05,
+      "loss": 1.4701,
+      "step": 139500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.35003663373841e-05,
+      "loss": 1.427,
+      "step": 140000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.344141549396595e-05,
+      "loss": 1.4451,
+      "step": 140500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.338246465054781e-05,
+      "loss": 1.5295,
+      "step": 141000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.332351380712968e-05,
+      "loss": 1.4976,
+      "step": 141500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.326456296371154e-05,
+      "loss": 1.4277,
+      "step": 142000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.32056121202934e-05,
+      "loss": 1.3922,
+      "step": 142500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.31467791785621e-05,
+      "loss": 1.5119,
+      "step": 143000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.30879462368308e-05,
+      "loss": 1.4537,
+      "step": 143500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.3028995393412655e-05,
+      "loss": 1.4404,
+      "step": 144000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.297004454999452e-05,
+      "loss": 1.4981,
+      "step": 144500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.291121160826322e-05,
+      "loss": 1.5137,
+      "step": 145000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.285226076484508e-05,
+      "loss": 1.449,
+      "step": 145500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.279330992142694e-05,
+      "loss": 1.5425,
+      "step": 146000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.273435907800881e-05,
+      "loss": 1.4137,
+      "step": 146500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.267540823459067e-05,
+      "loss": 1.3714,
+      "step": 147000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.261645739117252e-05,
+      "loss": 1.5328,
+      "step": 147500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.2557624449441223e-05,
+      "loss": 1.4122,
+      "step": 148000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.2498673606023084e-05,
+      "loss": 1.4551,
+      "step": 148500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.243972276260495e-05,
+      "loss": 1.4528,
+      "step": 149000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.238077191918681e-05,
+      "loss": 1.5808,
+      "step": 149500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.232182107576867e-05,
+      "loss": 1.4571,
+      "step": 150000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.226287023235054e-05,
+      "loss": 1.4758,
+      "step": 150500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.22039193889324e-05,
+      "loss": 1.5003,
+      "step": 151000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.214496854551425e-05,
+      "loss": 1.4685,
+      "step": 151500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.208601770209612e-05,
+      "loss": 1.4834,
+      "step": 152000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.202706685867798e-05,
+      "loss": 1.3873,
+      "step": 152500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.196811601525984e-05,
+      "loss": 1.409,
+      "step": 153000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.1909165171841706e-05,
+      "loss": 1.4671,
+      "step": 153500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.1850332230110406e-05,
+      "loss": 1.4043,
+      "step": 154000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.179138138669227e-05,
+      "loss": 1.4753,
+      "step": 154500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.173243054327412e-05,
+      "loss": 1.481,
+      "step": 155000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.167347969985599e-05,
+      "loss": 1.5065,
+      "step": 155500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.161452885643785e-05,
+      "loss": 1.4066,
+      "step": 156000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.155557801301971e-05,
+      "loss": 1.3987,
+      "step": 156500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.1496627169601574e-05,
+      "loss": 1.4025,
+      "step": 157000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.143779422787027e-05,
+      "loss": 1.4775,
+      "step": 157500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.137896128613897e-05,
+      "loss": 1.3659,
+      "step": 158000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.132001044272083e-05,
+      "loss": 1.3547,
+      "step": 158500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.126105959930269e-05,
+      "loss": 1.3893,
+      "step": 159000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.120210875588455e-05,
+      "loss": 1.4349,
+      "step": 159500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.1143157912466416e-05,
+      "loss": 1.3757,
+      "step": 160000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.1084207069048276e-05,
+      "loss": 1.4795,
+      "step": 160500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.1025256225630137e-05,
+      "loss": 1.431,
+      "step": 161000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.096642328389884e-05,
+      "loss": 1.383,
+      "step": 161500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.0907472440480704e-05,
+      "loss": 1.4149,
+      "step": 162000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.084852159706256e-05,
+      "loss": 1.4028,
+      "step": 162500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.078968865533126e-05,
+      "loss": 1.397,
+      "step": 163000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.073073781191312e-05,
+      "loss": 1.4919,
+      "step": 163500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.067178696849498e-05,
+      "loss": 1.347,
+      "step": 164000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.0612836125076845e-05,
+      "loss": 1.379,
+      "step": 164500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.0553885281658705e-05,
+      "loss": 1.4026,
+      "step": 165000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.0494934438240566e-05,
+      "loss": 1.3372,
+      "step": 165500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.0435983594822426e-05,
+      "loss": 1.4564,
+      "step": 166000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.0377032751404286e-05,
+      "loss": 1.3968,
+      "step": 166500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.0318081907986146e-05,
+      "loss": 1.4019,
+      "step": 167000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.0259248966254847e-05,
+      "loss": 1.4203,
+      "step": 167500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.0200298122836714e-05,
+      "loss": 1.4195,
+      "step": 168000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.0141347279418574e-05,
+      "loss": 1.4314,
+      "step": 168500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.0082396436000434e-05,
+      "loss": 1.4599,
+      "step": 169000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.00234455925823e-05,
+      "loss": 1.3961,
+      "step": 169500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9964494749164154e-05,
+      "loss": 1.49,
+      "step": 170000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9905543905746015e-05,
+      "loss": 1.388,
+      "step": 170500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.984659306232788e-05,
+      "loss": 1.3399,
+      "step": 171000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.9787760120596575e-05,
+      "loss": 1.3559,
+      "step": 171500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.972880927717844e-05,
+      "loss": 1.3839,
+      "step": 172000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.96698584337603e-05,
+      "loss": 1.3597,
+      "step": 172500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.961090759034216e-05,
+      "loss": 1.4653,
+      "step": 173000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.955195674692403e-05,
+      "loss": 1.412,
+      "step": 173500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.949300590350588e-05,
+      "loss": 1.3362,
+      "step": 174000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.943405506008774e-05,
+      "loss": 1.3354,
+      "step": 174500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.9375222118356444e-05,
+      "loss": 1.3451,
+      "step": 175000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.931627127493831e-05,
+      "loss": 1.4159,
+      "step": 175500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.925732043152017e-05,
+      "loss": 1.442,
+      "step": 176000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.919848748978887e-05,
+      "loss": 1.3793,
+      "step": 176500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.913953664637073e-05,
+      "loss": 1.4447,
+      "step": 177000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9080585802952585e-05,
+      "loss": 1.4068,
+      "step": 177500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.902163495953445e-05,
+      "loss": 1.3833,
+      "step": 178000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.896268411611631e-05,
+      "loss": 1.3802,
+      "step": 178500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.890373327269817e-05,
+      "loss": 1.3929,
+      "step": 179000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.884478242928004e-05,
+      "loss": 1.4342,
+      "step": 179500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.87858315858619e-05,
+      "loss": 1.3436,
+      "step": 180000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8726880742443766e-05,
+      "loss": 1.386,
+      "step": 180500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.866804780071245e-05,
+      "loss": 1.5735,
+      "step": 181000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.860909695729432e-05,
+      "loss": 1.3181,
+      "step": 181500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.855014611387618e-05,
+      "loss": 1.4031,
+      "step": 182000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.849131317214488e-05,
+      "loss": 1.442,
+      "step": 182500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.843236232872674e-05,
+      "loss": 1.4167,
+      "step": 183000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.837341148530861e-05,
+      "loss": 1.3656,
+      "step": 183500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.831446064189047e-05,
+      "loss": 1.3526,
+      "step": 184000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.825550979847233e-05,
+      "loss": 1.3862,
+      "step": 184500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.819655895505418e-05,
+      "loss": 1.3868,
+      "step": 185000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.813772601332288e-05,
+      "loss": 1.4435,
+      "step": 185500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.807877516990475e-05,
+      "loss": 1.4068,
+      "step": 186000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.801982432648661e-05,
+      "loss": 1.3566,
+      "step": 186500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.796099138475531e-05,
+      "loss": 1.3954,
+      "step": 187000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.790204054133717e-05,
+      "loss": 1.4382,
+      "step": 187500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.784308969791904e-05,
+      "loss": 1.4366,
+      "step": 188000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.778413885450089e-05,
+      "loss": 1.3365,
+      "step": 188500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.772518801108275e-05,
+      "loss": 1.323,
+      "step": 189000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.766623716766462e-05,
+      "loss": 1.3379,
+      "step": 189500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.760728632424648e-05,
+      "loss": 1.356,
+      "step": 190000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.754833548082834e-05,
+      "loss": 1.4302,
+      "step": 190500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7489384637410205e-05,
+      "loss": 1.4302,
+      "step": 191000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7430433793992065e-05,
+      "loss": 1.4299,
+      "step": 191500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7371482950573926e-05,
+      "loss": 1.3922,
+      "step": 192000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7312532107155786e-05,
+      "loss": 1.3954,
+      "step": 192500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.7253581263737646e-05,
+      "loss": 1.363,
+      "step": 193000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.719486622369318e-05,
+      "loss": 1.4173,
+      "step": 193500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.713591538027505e-05,
+      "loss": 1.3826,
+      "step": 194000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.707696453685691e-05,
+      "loss": 1.4274,
+      "step": 194500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.701801369343877e-05,
+      "loss": 1.4059,
+      "step": 195000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6959062850020634e-05,
+      "loss": 1.3678,
+      "step": 195500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.690011200660249e-05,
+      "loss": 1.4029,
+      "step": 196000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.684116116318435e-05,
+      "loss": 1.433,
+      "step": 196500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.678232822145305e-05,
+      "loss": 1.3245,
+      "step": 197000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.672349527972175e-05,
+      "loss": 1.3849,
+      "step": 197500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.666454443630361e-05,
+      "loss": 1.2898,
+      "step": 198000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6605593592885476e-05,
+      "loss": 1.4067,
+      "step": 198500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6546642749467336e-05,
+      "loss": 1.3511,
+      "step": 199000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.648769190604919e-05,
+      "loss": 1.3504,
+      "step": 199500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.642885896431789e-05,
+      "loss": 1.3707,
+      "step": 200000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.636990812089976e-05,
+      "loss": 1.3405,
+      "step": 200500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.631095727748162e-05,
+      "loss": 1.4657,
+      "step": 201000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.625200643406348e-05,
+      "loss": 1.3594,
+      "step": 201500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6193055590645344e-05,
+      "loss": 1.3988,
+      "step": 202000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6134104747227204e-05,
+      "loss": 1.4282,
+      "step": 202500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.607515390380906e-05,
+      "loss": 1.3753,
+      "step": 203000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6016203060390925e-05,
+      "loss": 1.3933,
+      "step": 203500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.5957252216972785e-05,
+      "loss": 1.3739,
+      "step": 204000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.5898301373554645e-05,
+      "loss": 1.4223,
+      "step": 204500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.583935053013651e-05,
+      "loss": 1.3495,
+      "step": 205000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.5780517588405206e-05,
+      "loss": 1.2631,
+      "step": 205500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.572156674498707e-05,
+      "loss": 1.3649,
+      "step": 206000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.566261590156893e-05,
+      "loss": 1.4423,
+      "step": 206500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.5603665058150786e-05,
+      "loss": 1.3753,
+      "step": 207000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.5544714214732653e-05,
+      "loss": 1.3592,
+      "step": 207500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.5485763371314514e-05,
+      "loss": 1.4111,
+      "step": 208000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.5426812527896374e-05,
+      "loss": 1.4252,
+      "step": 208500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.536786168447824e-05,
+      "loss": 1.3718,
+      "step": 209000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.53089108410601e-05,
+      "loss": 1.3504,
+      "step": 209500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.524995999764197e-05,
+      "loss": 1.3337,
+      "step": 210000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.519100915422383e-05,
+      "loss": 1.318,
+      "step": 210500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.513217621249252e-05,
+      "loss": 1.4068,
+      "step": 211000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.507322536907438e-05,
+      "loss": 1.2916,
+      "step": 211500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.501427452565624e-05,
+      "loss": 1.3078,
+      "step": 212000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.495532368223811e-05,
+      "loss": 1.3244,
+      "step": 212500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.489637283881997e-05,
+      "loss": 1.4635,
+      "step": 213000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.483753989708867e-05,
+      "loss": 1.3647,
+      "step": 213500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.477858905367053e-05,
+      "loss": 1.3064,
+      "step": 214000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.4719756111939224e-05,
+      "loss": 1.3345,
+      "step": 214500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.4660805268521084e-05,
+      "loss": 1.3723,
+      "step": 215000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.460185442510295e-05,
+      "loss": 1.3303,
+      "step": 215500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.454290358168481e-05,
+      "loss": 1.3384,
+      "step": 216000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.448395273826667e-05,
+      "loss": 1.3998,
+      "step": 216500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.442511979653537e-05,
+      "loss": 1.4623,
+      "step": 217000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.436616895311724e-05,
+      "loss": 1.3193,
+      "step": 217500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.430721810969909e-05,
+      "loss": 1.3703,
+      "step": 218000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.424826726628095e-05,
+      "loss": 1.3235,
+      "step": 218500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.418931642286282e-05,
+      "loss": 1.252,
+      "step": 219000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.413036557944468e-05,
+      "loss": 1.3262,
+      "step": 219500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.407141473602654e-05,
+      "loss": 1.287,
+      "step": 220000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.401246389260841e-05,
+      "loss": 1.3144,
+      "step": 220500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.395351304919027e-05,
+      "loss": 1.3907,
+      "step": 221000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.389456220577213e-05,
+      "loss": 1.321,
+      "step": 221500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.383572926404082e-05,
+      "loss": 1.2963,
+      "step": 222000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.377677842062268e-05,
+      "loss": 1.3762,
+      "step": 222500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.371794547889138e-05,
+      "loss": 1.4147,
+      "step": 223000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.365899463547325e-05,
+      "loss": 1.317,
+      "step": 223500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.360004379205511e-05,
+      "loss": 1.353,
+      "step": 224000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.354109294863697e-05,
+      "loss": 1.2843,
+      "step": 224500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.3482142105218836e-05,
+      "loss": 1.2709,
+      "step": 225000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.342319126180069e-05,
+      "loss": 1.3417,
+      "step": 225500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.336424041838255e-05,
+      "loss": 1.3168,
+      "step": 226000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.3305289574964416e-05,
+      "loss": 1.3149,
+      "step": 226500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.324657453491995e-05,
+      "loss": 1.4086,
+      "step": 227000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.318762369150181e-05,
+      "loss": 1.4308,
+      "step": 227500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.312867284808368e-05,
+      "loss": 1.3158,
+      "step": 228000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.306972200466554e-05,
+      "loss": 1.3374,
+      "step": 228500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.301077116124739e-05,
+      "loss": 1.2938,
+      "step": 229000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.295182031782926e-05,
+      "loss": 1.3818,
+      "step": 229500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.289286947441112e-05,
+      "loss": 1.3054,
+      "step": 230000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.283391863099298e-05,
+      "loss": 1.4121,
+      "step": 230500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.2774967787574845e-05,
+      "loss": 1.3855,
+      "step": 231000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.271625274753038e-05,
+      "loss": 1.3088,
+      "step": 231500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.265730190411224e-05,
+      "loss": 1.2938,
+      "step": 232000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.25983510606941e-05,
+      "loss": 1.3244,
+      "step": 232500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.253940021727596e-05,
+      "loss": 1.303,
+      "step": 233000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.248044937385782e-05,
+      "loss": 1.2691,
+      "step": 233500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.242149853043969e-05,
+      "loss": 1.2841,
+      "step": 234000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.236254768702155e-05,
+      "loss": 1.4295,
+      "step": 234500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.230359684360341e-05,
+      "loss": 1.2992,
+      "step": 235000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.2244646000185274e-05,
+      "loss": 1.3605,
+      "step": 235500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.2185695156767135e-05,
+      "loss": 1.3952,
+      "step": 236000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.212674431334899e-05,
+      "loss": 1.3236,
+      "step": 236500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2067793469930855e-05,
+      "loss": 1.2393,
+      "step": 237000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2008842626512715e-05,
+      "loss": 1.3406,
+      "step": 237500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.195024548815509e-05,
+      "loss": 1.2906,
+      "step": 238000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.189129464473695e-05,
+      "loss": 1.3258,
+      "step": 238500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.1832343801318816e-05,
+      "loss": 1.3759,
+      "step": 239000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.177339295790067e-05,
+      "loss": 1.375,
+      "step": 239500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.171444211448253e-05,
+      "loss": 1.3272,
+      "step": 240000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.16554912710644e-05,
+      "loss": 1.3385,
+      "step": 240500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.159654042764626e-05,
+      "loss": 1.3197,
+      "step": 241000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.153758958422812e-05,
+      "loss": 1.2662,
+      "step": 241500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.1478638740809985e-05,
+      "loss": 1.3402,
+      "step": 242000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.141980579907868e-05,
+      "loss": 1.3483,
+      "step": 242500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.136097285734737e-05,
+      "loss": 1.2599,
+      "step": 243000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.130202201392924e-05,
+      "loss": 1.274,
+      "step": 243500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.12430711705111e-05,
+      "loss": 1.372,
+      "step": 244000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.118412032709296e-05,
+      "loss": 1.3183,
+      "step": 244500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.1125169483674826e-05,
+      "loss": 1.2656,
+      "step": 245000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.106633654194352e-05,
+      "loss": 1.2663,
+      "step": 245500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.100750360021222e-05,
+      "loss": 1.3134,
+      "step": 246000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.0948670658480914e-05,
+      "loss": 1.3478,
+      "step": 246500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.0889719815062774e-05,
+      "loss": 1.323,
+      "step": 247000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.083076897164464e-05,
+      "loss": 1.301,
+      "step": 247500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.07718181282265e-05,
+      "loss": 1.3122,
+      "step": 248000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.071286728480836e-05,
+      "loss": 1.3037,
+      "step": 248500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.065391644139023e-05,
+      "loss": 1.3082,
+      "step": 249000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.059496559797209e-05,
+      "loss": 1.3267,
+      "step": 249500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.0536014754553956e-05,
+      "loss": 1.3434,
+      "step": 250000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.047706391113581e-05,
+      "loss": 1.3716,
+      "step": 250500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.041811306771767e-05,
+      "loss": 1.2677,
+      "step": 251000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.035916222429953e-05,
+      "loss": 1.3538,
+      "step": 251500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.0300211380881396e-05,
+      "loss": 1.2902,
+      "step": 252000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.024126053746326e-05,
+      "loss": 1.2978,
+      "step": 252500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.0182309694045124e-05,
+      "loss": 1.4214,
+      "step": 253000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.012347675231382e-05,
+      "loss": 1.2749,
+      "step": 253500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.006452590889568e-05,
+      "loss": 1.3774,
+      "step": 254000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.000569296716437e-05,
+      "loss": 1.3001,
+      "step": 254500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.994674212374624e-05,
+      "loss": 1.3206,
+      "step": 255000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.98877912803281e-05,
+      "loss": 1.2425,
+      "step": 255500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.9828840436909965e-05,
+      "loss": 1.2601,
+      "step": 256000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.9769889593491825e-05,
+      "loss": 1.3718,
+      "step": 256500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.9710938750073686e-05,
+      "loss": 1.2449,
+      "step": 257000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.965198790665555e-05,
+      "loss": 1.395,
+      "step": 257500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.9593037063237406e-05,
+      "loss": 1.3455,
+      "step": 258000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.9534086219819266e-05,
+      "loss": 1.2778,
+      "step": 258500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.947513537640113e-05,
+      "loss": 1.3237,
+      "step": 259000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.9416184532982993e-05,
+      "loss": 1.3166,
+      "step": 259500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.9357351591251694e-05,
+      "loss": 1.3272,
+      "step": 260000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.9298518649520394e-05,
+      "loss": 1.3,
+      "step": 260500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.9239567806102254e-05,
+      "loss": 1.3644,
+      "step": 261000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.918061696268411e-05,
+      "loss": 1.2442,
+      "step": 261500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.9121666119265975e-05,
+      "loss": 1.3674,
+      "step": 262000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.9062715275847835e-05,
+      "loss": 1.3582,
+      "step": 262500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.9003764432429695e-05,
+      "loss": 1.2996,
+      "step": 263000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.8944931490698396e-05,
+      "loss": 1.2659,
+      "step": 263500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.8885980647280256e-05,
+      "loss": 1.2974,
+      "step": 264000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.882702980386212e-05,
+      "loss": 1.3355,
+      "step": 264500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.876807896044398e-05,
+      "loss": 1.2746,
+      "step": 265000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.8709128117025837e-05,
+      "loss": 1.3051,
+      "step": 265500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.8650177273607704e-05,
+      "loss": 1.2458,
+      "step": 266000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.8591344331876404e-05,
+      "loss": 1.2668,
+      "step": 266500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.8532393488458264e-05,
+      "loss": 1.3182,
+      "step": 267000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.8473442645040124e-05,
+      "loss": 1.3013,
+      "step": 267500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.841449180162199e-05,
+      "loss": 1.2176,
+      "step": 268000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.835554095820385e-05,
+      "loss": 1.3966,
+      "step": 268500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.8296708016472545e-05,
+      "loss": 1.2697,
+      "step": 269000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.8237757173054405e-05,
+      "loss": 1.3279,
+      "step": 269500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.8178806329636266e-05,
+      "loss": 1.28,
+      "step": 270000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.811985548621813e-05,
+      "loss": 1.2927,
+      "step": 270500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.806090464279999e-05,
+      "loss": 1.3114,
+      "step": 271000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.800195379938186e-05,
+      "loss": 1.3238,
+      "step": 271500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.794300295596372e-05,
+      "loss": 1.4098,
+      "step": 272000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.788405211254558e-05,
+      "loss": 1.2851,
+      "step": 272500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.7825101269127434e-05,
+      "loss": 1.3233,
+      "step": 273000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.77661504257093e-05,
+      "loss": 1.4169,
+      "step": 273500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.770719958229116e-05,
+      "loss": 1.3181,
+      "step": 274000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.764824873887303e-05,
+      "loss": 1.2962,
+      "step": 274500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.758929789545489e-05,
+      "loss": 1.3229,
+      "step": 275000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.753034705203675e-05,
+      "loss": 1.3531,
+      "step": 275500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.747163201199229e-05,
+      "loss": 1.3241,
+      "step": 276000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.741268116857414e-05,
+      "loss": 1.2468,
+      "step": 276500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.7353730325156e-05,
+      "loss": 1.2878,
+      "step": 277000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.729477948173787e-05,
+      "loss": 1.2675,
+      "step": 277500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.723582863831973e-05,
+      "loss": 1.2372,
+      "step": 278000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.717687779490159e-05,
+      "loss": 1.3147,
+      "step": 278500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.711804485317029e-05,
+      "loss": 1.3194,
+      "step": 279000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.705909400975216e-05,
+      "loss": 1.325,
+      "step": 279500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.700014316633401e-05,
+      "loss": 1.3561,
+      "step": 280000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.694119232291587e-05,
+      "loss": 1.2199,
+      "step": 280500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.688224147949773e-05,
+      "loss": 1.3318,
+      "step": 281000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.682340853776643e-05,
+      "loss": 1.3502,
+      "step": 281500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.67644576943483e-05,
+      "loss": 1.3071,
+      "step": 282000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.670550685093016e-05,
+      "loss": 1.2433,
+      "step": 282500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.664655600751202e-05,
+      "loss": 1.2955,
+      "step": 283000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.6587605164093886e-05,
+      "loss": 1.3713,
+      "step": 283500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.652865432067574e-05,
+      "loss": 1.3452,
+      "step": 284000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.64697034772576e-05,
+      "loss": 1.321,
+      "step": 284500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.6410752633839466e-05,
+      "loss": 1.2524,
+      "step": 285000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.635180179042133e-05,
+      "loss": 1.3178,
+      "step": 285500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.629285094700319e-05,
+      "loss": 1.2231,
+      "step": 286000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.623401800527189e-05,
+      "loss": 1.3051,
+      "step": 286500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6175067161853754e-05,
+      "loss": 1.2286,
+      "step": 287000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.611611631843561e-05,
+      "loss": 1.2875,
+      "step": 287500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.605716547501747e-05,
+      "loss": 1.23,
+      "step": 288000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.599821463159933e-05,
+      "loss": 1.2613,
+      "step": 288500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5939263788181195e-05,
+      "loss": 1.2641,
+      "step": 289000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5880312944763055e-05,
+      "loss": 1.2473,
+      "step": 289500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.582136210134492e-05,
+      "loss": 1.2474,
+      "step": 290000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5762529159613616e-05,
+      "loss": 1.2923,
+      "step": 290500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.570369621788231e-05,
+      "loss": 1.3024,
+      "step": 291000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5644745374464176e-05,
+      "loss": 1.2859,
+      "step": 291500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.558579453104604e-05,
+      "loss": 1.1786,
+      "step": 292000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.55268436876279e-05,
+      "loss": 1.2681,
+      "step": 292500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5467892844209764e-05,
+      "loss": 1.3482,
+      "step": 293000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.540905990247846e-05,
+      "loss": 1.3399,
+      "step": 293500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.5350109059060325e-05,
+      "loss": 1.2904,
+      "step": 294000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.529127611732902e-05,
+      "loss": 1.3039,
+      "step": 294500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.523232527391088e-05,
+      "loss": 1.2332,
+      "step": 295000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.517337443049274e-05,
+      "loss": 1.3332,
+      "step": 295500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5114423587074606e-05,
+      "loss": 1.2235,
+      "step": 296000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5055472743656466e-05,
+      "loss": 1.3569,
+      "step": 296500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.4996521900238326e-05,
+      "loss": 1.2554,
+      "step": 297000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.4937571056820186e-05,
+      "loss": 1.2249,
+      "step": 297500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4878620213402046e-05,
+      "loss": 1.3162,
+      "step": 298000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.481978727167075e-05,
+      "loss": 1.2694,
+      "step": 298500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4760836428252614e-05,
+      "loss": 1.2175,
+      "step": 299000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.470188558483447e-05,
+      "loss": 1.2431,
+      "step": 299500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4642934741416334e-05,
+      "loss": 1.245,
+      "step": 300000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4583983897998194e-05,
+      "loss": 1.341,
+      "step": 300500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.452503305458006e-05,
+      "loss": 1.2291,
+      "step": 301000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4466082211161915e-05,
+      "loss": 1.2578,
+      "step": 301500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4407249269430615e-05,
+      "loss": 1.2904,
+      "step": 302000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.434829842601248e-05,
+      "loss": 1.2795,
+      "step": 302500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4289347582594336e-05,
+      "loss": 1.2287,
+      "step": 303000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.42303967391762e-05,
+      "loss": 1.2741,
+      "step": 303500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.41715637974449e-05,
+      "loss": 1.2506,
+      "step": 304000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.411261295402676e-05,
+      "loss": 1.328,
+      "step": 304500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4053662110608623e-05,
+      "loss": 1.2775,
+      "step": 305000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.3994711267190484e-05,
+      "loss": 1.2351,
+      "step": 305500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.3935760423772344e-05,
+      "loss": 1.3361,
+      "step": 306000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.387680958035421e-05,
+      "loss": 1.2548,
+      "step": 306500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.381785873693607e-05,
+      "loss": 1.2169,
+      "step": 307000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.375890789351793e-05,
+      "loss": 1.278,
+      "step": 307500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.370007495178663e-05,
+      "loss": 1.1629,
+      "step": 308000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.364112410836849e-05,
+      "loss": 1.268,
+      "step": 308500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.358217326495035e-05,
+      "loss": 1.2995,
+      "step": 309000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.352322242153221e-05,
+      "loss": 1.2686,
+      "step": 309500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.346427157811408e-05,
+      "loss": 1.3668,
+      "step": 310000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.340532073469594e-05,
+      "loss": 1.2884,
+      "step": 310500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.33463698912778e-05,
+      "loss": 1.3142,
+      "step": 311000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.32875369495465e-05,
+      "loss": 1.287,
+      "step": 311500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.322858610612836e-05,
+      "loss": 1.2372,
+      "step": 312000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.316963526271022e-05,
+      "loss": 1.2591,
+      "step": 312500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.311068441929208e-05,
+      "loss": 1.2634,
+      "step": 313000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.305173357587394e-05,
+      "loss": 1.2233,
+      "step": 313500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.299278273245581e-05,
+      "loss": 1.2714,
+      "step": 314000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.293383188903767e-05,
+      "loss": 1.2526,
+      "step": 314500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.287488104561953e-05,
+      "loss": 1.2608,
+      "step": 315000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.281593020220139e-05,
+      "loss": 1.289,
+      "step": 315500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.2756979358783255e-05,
+      "loss": 1.2376,
+      "step": 316000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.269802851536511e-05,
+      "loss": 1.2847,
+      "step": 316500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.2639077671946976e-05,
+      "loss": 1.264,
+      "step": 317000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.2580244730215676e-05,
+      "loss": 1.2889,
+      "step": 317500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.2521293886797536e-05,
+      "loss": 1.3118,
+      "step": 318000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.24623430433794e-05,
+      "loss": 1.3439,
+      "step": 318500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.24035101016481e-05,
+      "loss": 1.3455,
+      "step": 319000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.234455925822996e-05,
+      "loss": 1.2317,
+      "step": 319500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.228560841481182e-05,
+      "loss": 1.2395,
+      "step": 320000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.222665757139368e-05,
+      "loss": 1.2622,
+      "step": 320500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.2167706727975545e-05,
+      "loss": 1.2654,
+      "step": 321000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.210887378624424e-05,
+      "loss": 1.3169,
+      "step": 321500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.20499229428261e-05,
+      "loss": 1.3072,
+      "step": 322000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.1990972099407965e-05,
+      "loss": 1.2517,
+      "step": 322500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.1932021255989826e-05,
+      "loss": 1.2509,
+      "step": 323000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.1873070412571686e-05,
+      "loss": 1.2573,
+      "step": 323500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.1814237470840386e-05,
+      "loss": 1.276,
+      "step": 324000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.1755286627422247e-05,
+      "loss": 1.2172,
+      "step": 324500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.169633578400411e-05,
+      "loss": 1.2464,
+      "step": 325000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.163738494058597e-05,
+      "loss": 1.2765,
+      "step": 325500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.157855199885467e-05,
+      "loss": 1.2238,
+      "step": 326000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.151960115543653e-05,
+      "loss": 1.2464,
+      "step": 326500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.1460650312018395e-05,
+      "loss": 1.2596,
+      "step": 327000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.140169946860025e-05,
+      "loss": 1.3547,
+      "step": 327500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.1342748625182115e-05,
+      "loss": 1.2478,
+      "step": 328000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.1283797781763975e-05,
+      "loss": 1.2495,
+      "step": 328500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.1224846938345835e-05,
+      "loss": 1.2631,
+      "step": 329000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.1165896094927696e-05,
+      "loss": 1.152,
+      "step": 329500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.1107063153196396e-05,
+      "loss": 1.2581,
+      "step": 330000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.1048112309778256e-05,
+      "loss": 1.2761,
+      "step": 330500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.0989161466360116e-05,
+      "loss": 1.2881,
+      "step": 331000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.093021062294198e-05,
+      "loss": 1.268,
+      "step": 331500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.087137768121068e-05,
+      "loss": 1.1981,
+      "step": 332000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.0812426837792544e-05,
+      "loss": 1.2601,
+      "step": 332500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.0753475994374404e-05,
+      "loss": 1.2364,
+      "step": 333000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.0694525150956264e-05,
+      "loss": 1.2036,
+      "step": 333500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.0635574307538125e-05,
+      "loss": 1.2192,
+      "step": 334000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.0576741365806825e-05,
+      "loss": 1.227,
+      "step": 334500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.051790842407552e-05,
+      "loss": 1.1997,
+      "step": 335000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.0459075482344222e-05,
+      "loss": 1.205,
+      "step": 335500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.040012463892608e-05,
+      "loss": 1.2236,
+      "step": 336000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.0341173795507943e-05,
+      "loss": 1.1391,
+      "step": 336500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.0282222952089806e-05,
+      "loss": 1.2229,
+      "step": 337000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.02233900103585e-05,
+      "loss": 1.2597,
+      "step": 337500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.0164439166940364e-05,
+      "loss": 1.26,
+      "step": 338000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.0105488323522227e-05,
+      "loss": 1.203,
+      "step": 338500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.0046537480104087e-05,
+      "loss": 1.2286,
+      "step": 339000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.998758663668595e-05,
+      "loss": 1.2209,
+      "step": 339500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.992863579326781e-05,
+      "loss": 1.2839,
+      "step": 340000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.986968494984967e-05,
+      "loss": 1.2837,
+      "step": 340500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.9810734106431535e-05,
+      "loss": 1.2309,
+      "step": 341000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.9751901164700232e-05,
+      "loss": 1.2584,
+      "step": 341500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.9692950321282092e-05,
+      "loss": 1.2027,
+      "step": 342000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.9633999477863956e-05,
+      "loss": 1.1769,
+      "step": 342500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.957504863444582e-05,
+      "loss": 1.1513,
+      "step": 343000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.9516215692714513e-05,
+      "loss": 1.3123,
+      "step": 343500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.9457264849296377e-05,
+      "loss": 1.236,
+      "step": 344000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.939831400587824e-05,
+      "loss": 1.2736,
+      "step": 344500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.9339363162460104e-05,
+      "loss": 1.1916,
+      "step": 345000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.928041231904196e-05,
+      "loss": 1.1626,
+      "step": 345500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.9221461475623824e-05,
+      "loss": 1.3372,
+      "step": 346000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.9162510632205684e-05,
+      "loss": 1.2332,
+      "step": 346500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.910367769047438e-05,
+      "loss": 1.1655,
+      "step": 347000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.9044726847056245e-05,
+      "loss": 1.206,
+      "step": 347500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.898577600363811e-05,
+      "loss": 1.2023,
+      "step": 348000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.892682516021997e-05,
+      "loss": 1.2383,
+      "step": 348500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.8867992218488666e-05,
+      "loss": 1.2346,
+      "step": 349000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.880904137507053e-05,
+      "loss": 1.2237,
+      "step": 349500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.875009053165239e-05,
+      "loss": 1.2113,
+      "step": 350000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.8691139688234253e-05,
+      "loss": 1.242,
+      "step": 350500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.863230674650295e-05,
+      "loss": 1.2793,
+      "step": 351000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.857335590308481e-05,
+      "loss": 1.1956,
+      "step": 351500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.8514405059666674e-05,
+      "loss": 1.2009,
+      "step": 352000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.8455454216248534e-05,
+      "loss": 1.3208,
+      "step": 352500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.8396503372830395e-05,
+      "loss": 1.2701,
+      "step": 353000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.8337670431099095e-05,
+      "loss": 1.2164,
+      "step": 353500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.827871958768096e-05,
+      "loss": 1.205,
+      "step": 354000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.8219768744262815e-05,
+      "loss": 1.1913,
+      "step": 354500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.816081790084468e-05,
+      "loss": 1.2342,
+      "step": 355000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.8101867057426543e-05,
+      "loss": 1.277,
+      "step": 355500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.8042916214008403e-05,
+      "loss": 1.2478,
+      "step": 356000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.7983965370590263e-05,
+      "loss": 1.2457,
+      "step": 356500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.7925014527172127e-05,
+      "loss": 1.2791,
+      "step": 357000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.7866181585440824e-05,
+      "loss": 1.1806,
+      "step": 357500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.7807230742022684e-05,
+      "loss": 1.3453,
+      "step": 358000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.7748397800291384e-05,
+      "loss": 1.2719,
+      "step": 358500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.7689446956873244e-05,
+      "loss": 1.1959,
+      "step": 359000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7630496113455108e-05,
+      "loss": 1.2623,
+      "step": 359500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7571545270036968e-05,
+      "loss": 1.2014,
+      "step": 360000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.751259442661883e-05,
+      "loss": 1.1756,
+      "step": 360500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.745376148488753e-05,
+      "loss": 1.2209,
+      "step": 361000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.739481064146939e-05,
+      "loss": 1.255,
+      "step": 361500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.733585979805125e-05,
+      "loss": 1.1713,
+      "step": 362000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.7276908954633113e-05,
+      "loss": 1.2254,
+      "step": 362500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.7217958111214976e-05,
+      "loss": 1.2165,
+      "step": 363000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.715912516948367e-05,
+      "loss": 1.1452,
+      "step": 363500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.7100174326065534e-05,
+      "loss": 1.2771,
+      "step": 364000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.7041223482647397e-05,
+      "loss": 1.2054,
+      "step": 364500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.698227263922926e-05,
+      "loss": 1.2138,
+      "step": 365000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.6923321795811118e-05,
+      "loss": 1.1762,
+      "step": 365500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.686437095239298e-05,
+      "loss": 1.2403,
+      "step": 366000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.6805420108974845e-05,
+      "loss": 1.2552,
+      "step": 366500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.6746469265556705e-05,
+      "loss": 1.2517,
+      "step": 367000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6687636323825402e-05,
+      "loss": 1.1744,
+      "step": 367500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6628685480407266e-05,
+      "loss": 1.2521,
+      "step": 368000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6569734636989126e-05,
+      "loss": 1.1839,
+      "step": 368500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6510783793570986e-05,
+      "loss": 1.1896,
+      "step": 369000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.6451950851839686e-05,
+      "loss": 1.2454,
+      "step": 369500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.6393000008421547e-05,
+      "loss": 1.2666,
+      "step": 370000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.633404916500341e-05,
+      "loss": 1.2005,
+      "step": 370500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.627509832158527e-05,
+      "loss": 1.2315,
+      "step": 371000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.621614747816713e-05,
+      "loss": 1.2595,
+      "step": 371500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6157196634748994e-05,
+      "loss": 1.2481,
+      "step": 372000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.609836369301769e-05,
+      "loss": 1.2914,
+      "step": 372500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.603941284959955e-05,
+      "loss": 1.1879,
+      "step": 373000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.5980462006181415e-05,
+      "loss": 1.1608,
+      "step": 373500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.592151116276328e-05,
+      "loss": 1.163,
+      "step": 374000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.5862560319345136e-05,
+      "loss": 1.2001,
+      "step": 374500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.5803609475927e-05,
+      "loss": 1.1682,
+      "step": 375000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.57447765341957e-05,
+      "loss": 1.2287,
+      "step": 375500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.568582569077756e-05,
+      "loss": 1.1357,
+      "step": 376000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.562687484735942e-05,
+      "loss": 1.1565,
+      "step": 376500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.5567924003941284e-05,
+      "loss": 1.2437,
+      "step": 377000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.550909106220998e-05,
+      "loss": 1.1682,
+      "step": 377500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.545014021879184e-05,
+      "loss": 1.1594,
+      "step": 378000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.5391189375373704e-05,
+      "loss": 1.2731,
+      "step": 378500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.5332238531955565e-05,
+      "loss": 1.228,
+      "step": 379000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.5273287688537428e-05,
+      "loss": 1.2309,
+      "step": 379500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.521433684511929e-05,
+      "loss": 1.1674,
+      "step": 380000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.515538600170115e-05,
+      "loss": 1.2896,
+      "step": 380500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.5096435158283012e-05,
+      "loss": 1.2069,
+      "step": 381000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.5037484314864876e-05,
+      "loss": 1.1584,
+      "step": 381500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.497865137313357e-05,
+      "loss": 1.2353,
+      "step": 382000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.4919700529715433e-05,
+      "loss": 1.1736,
+      "step": 382500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.4860867587984133e-05,
+      "loss": 1.1823,
+      "step": 383000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.480191674456599e-05,
+      "loss": 1.2652,
+      "step": 383500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.4742965901147854e-05,
+      "loss": 1.2173,
+      "step": 384000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.4684015057729717e-05,
+      "loss": 1.2165,
+      "step": 384500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.462506421431158e-05,
+      "loss": 1.1748,
+      "step": 385000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.4566113370893438e-05,
+      "loss": 1.149,
+      "step": 385500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.45071625274753e-05,
+      "loss": 1.1918,
+      "step": 386000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.4448211684057165e-05,
+      "loss": 1.1725,
+      "step": 386500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4389260840639025e-05,
+      "loss": 1.1267,
+      "step": 387000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4330309997220885e-05,
+      "loss": 1.2385,
+      "step": 387500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4271477055489586e-05,
+      "loss": 1.159,
+      "step": 388000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4212526212071446e-05,
+      "loss": 1.2492,
+      "step": 388500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.415357536865331e-05,
+      "loss": 1.1856,
+      "step": 389000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.409462452523517e-05,
+      "loss": 1.1974,
+      "step": 389500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.403567368181703e-05,
+      "loss": 1.1594,
+      "step": 390000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3976722838398894e-05,
+      "loss": 1.1298,
+      "step": 390500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3917771994980757e-05,
+      "loss": 1.2669,
+      "step": 391000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3858821151562614e-05,
+      "loss": 1.1915,
+      "step": 391500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3799870308144478e-05,
+      "loss": 1.1861,
+      "step": 392000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3741037366413178e-05,
+      "loss": 1.1185,
+      "step": 392500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.368220442468187e-05,
+      "loss": 1.1989,
+      "step": 393000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3623253581263735e-05,
+      "loss": 1.1478,
+      "step": 393500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.35643027378456e-05,
+      "loss": 1.2042,
+      "step": 394000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3505469796114292e-05,
+      "loss": 1.2027,
+      "step": 394500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3446518952696156e-05,
+      "loss": 1.1731,
+      "step": 395000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.338756810927802e-05,
+      "loss": 1.1656,
+      "step": 395500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3328735167546717e-05,
+      "loss": 1.2331,
+      "step": 396000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3269784324128577e-05,
+      "loss": 1.2218,
+      "step": 396500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.321083348071044e-05,
+      "loss": 1.2254,
+      "step": 397000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.31518826372923e-05,
+      "loss": 1.2235,
+      "step": 397500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.3092931793874164e-05,
+      "loss": 1.2181,
+      "step": 398000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.3033980950456025e-05,
+      "loss": 1.1699,
+      "step": 398500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2975030107037885e-05,
+      "loss": 1.1735,
+      "step": 399000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2916079263619748e-05,
+      "loss": 1.1364,
+      "step": 399500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2857128420201612e-05,
+      "loss": 1.1461,
+      "step": 400000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.279817757678347e-05,
+      "loss": 1.1873,
+      "step": 400500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2739226733365332e-05,
+      "loss": 1.1424,
+      "step": 401000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2680393791634033e-05,
+      "loss": 1.2363,
+      "step": 401500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.262144294821589e-05,
+      "loss": 1.1695,
+      "step": 402000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2562492104797753e-05,
+      "loss": 1.1056,
+      "step": 402500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2503541261379617e-05,
+      "loss": 1.1959,
+      "step": 403000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.244459041796148e-05,
+      "loss": 1.2244,
+      "step": 403500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.238563957454334e-05,
+      "loss": 1.2254,
+      "step": 404000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.23266887311252e-05,
+      "loss": 1.1684,
+      "step": 404500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2267737887707064e-05,
+      "loss": 1.1584,
+      "step": 405000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2208787044288925e-05,
+      "loss": 1.1854,
+      "step": 405500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.214995410255762e-05,
+      "loss": 1.1488,
+      "step": 406000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2091003259139485e-05,
+      "loss": 1.1655,
+      "step": 406500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2032052415721345e-05,
+      "loss": 1.161,
+      "step": 407000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.197310157230321e-05,
+      "loss": 1.1538,
+      "step": 407500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.191415072888507e-05,
+      "loss": 1.155,
+      "step": 408000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.1855317787153766e-05,
+      "loss": 1.1437,
+      "step": 408500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.179636694373563e-05,
+      "loss": 1.1404,
+      "step": 409000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.1737416100317493e-05,
+      "loss": 1.166,
+      "step": 409500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.167846525689935e-05,
+      "loss": 1.1788,
+      "step": 410000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.1619514413481214e-05,
+      "loss": 1.2474,
+      "step": 410500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.1560563570063077e-05,
+      "loss": 1.1323,
+      "step": 411000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.150173062833177e-05,
+      "loss": 1.1443,
+      "step": 411500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.1442779784913635e-05,
+      "loss": 1.1085,
+      "step": 412000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.1383828941495498e-05,
+      "loss": 1.1595,
+      "step": 412500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.132487809807736e-05,
+      "loss": 1.2336,
+      "step": 413000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.126592725465922e-05,
+      "loss": 1.1398,
+      "step": 413500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.120709431292792e-05,
+      "loss": 1.236,
+      "step": 414000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.114814346950978e-05,
+      "loss": 1.2135,
+      "step": 414500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.1089192626091643e-05,
+      "loss": 1.1443,
+      "step": 415000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.1030241782673503e-05,
+      "loss": 1.0692,
+      "step": 415500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.0971290939255363e-05,
+      "loss": 1.1897,
+      "step": 416000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.0912340095837227e-05,
+      "loss": 1.1329,
+      "step": 416500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.085338925241909e-05,
+      "loss": 1.184,
+      "step": 417000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.0794438409000947e-05,
+      "loss": 1.1203,
+      "step": 417500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.0735605467269648e-05,
+      "loss": 1.1576,
+      "step": 418000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.067665462385151e-05,
+      "loss": 1.1962,
+      "step": 418500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.0617821682120205e-05,
+      "loss": 1.1513,
+      "step": 419000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.055887083870207e-05,
+      "loss": 1.1488,
+      "step": 419500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.050003789697077e-05,
+      "loss": 1.2417,
+      "step": 420000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.0441087053552626e-05,
+      "loss": 1.1547,
+      "step": 420500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.038213621013449e-05,
+      "loss": 1.1786,
+      "step": 421000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.0323185366716353e-05,
+      "loss": 1.1449,
+      "step": 421500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.0264234523298216e-05,
+      "loss": 1.1384,
+      "step": 422000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.0205283679880073e-05,
+      "loss": 1.1989,
+      "step": 422500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.0146332836461937e-05,
+      "loss": 1.2179,
+      "step": 423000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.00873819930438e-05,
+      "loss": 1.2169,
+      "step": 423500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.002843114962566e-05,
+      "loss": 1.1653,
+      "step": 424000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.996948030620752e-05,
+      "loss": 1.2552,
+      "step": 424500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9910529462789384e-05,
+      "loss": 1.0916,
+      "step": 425000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.985169652105808e-05,
+      "loss": 1.1899,
+      "step": 425500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.979286357932678e-05,
+      "loss": 1.1469,
+      "step": 426000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9733912735908642e-05,
+      "loss": 1.1358,
+      "step": 426500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9674961892490502e-05,
+      "loss": 1.1754,
+      "step": 427000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9616011049072366e-05,
+      "loss": 1.1624,
+      "step": 427500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9557060205654226e-05,
+      "loss": 1.1019,
+      "step": 428000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9498109362236086e-05,
+      "loss": 1.1592,
+      "step": 428500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.943915851881795e-05,
+      "loss": 1.1909,
+      "step": 429000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9380207675399813e-05,
+      "loss": 1.1488,
+      "step": 429500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.932125683198167e-05,
+      "loss": 1.2036,
+      "step": 430000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9262305988563534e-05,
+      "loss": 1.2028,
+      "step": 430500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9203355145145397e-05,
+      "loss": 1.1806,
+      "step": 431000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9144522203414095e-05,
+      "loss": 1.1659,
+      "step": 431500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9085571359995955e-05,
+      "loss": 1.151,
+      "step": 432000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.902662051657782e-05,
+      "loss": 1.1106,
+      "step": 432500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.896766967315968e-05,
+      "loss": 1.2034,
+      "step": 433000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.8908718829741542e-05,
+      "loss": 1.239,
+      "step": 433500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.884988588801024e-05,
+      "loss": 1.1395,
+      "step": 434000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.87909350445921e-05,
+      "loss": 1.2253,
+      "step": 434500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.8731984201173963e-05,
+      "loss": 1.1857,
+      "step": 435000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.8673033357755823e-05,
+      "loss": 1.1693,
+      "step": 435500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.861420041602452e-05,
+      "loss": 1.1043,
+      "step": 436000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.8555249572606384e-05,
+      "loss": 1.1789,
+      "step": 436500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.8496298729188247e-05,
+      "loss": 1.1296,
+      "step": 437000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.8437347885770104e-05,
+      "loss": 1.1615,
+      "step": 437500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.8378397042351968e-05,
+      "loss": 1.1349,
+      "step": 438000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.8319564100620668e-05,
+      "loss": 1.1539,
+      "step": 438500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.8260613257202525e-05,
+      "loss": 1.2216,
+      "step": 439000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.820166241378439e-05,
+      "loss": 1.1701,
+      "step": 439500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.8142711570366252e-05,
+      "loss": 1.1634,
+      "step": 440000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.8083760726948116e-05,
+      "loss": 1.1049,
+      "step": 440500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.8024809883529973e-05,
+      "loss": 1.1572,
+      "step": 441000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.7965976941798673e-05,
+      "loss": 1.1914,
+      "step": 441500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.7907026098380537e-05,
+      "loss": 1.1036,
+      "step": 442000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.7848075254962397e-05,
+      "loss": 1.1079,
+      "step": 442500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.7789124411544257e-05,
+      "loss": 1.0967,
+      "step": 443000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.773017356812612e-05,
+      "loss": 1.0886,
+      "step": 443500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.767122272470798e-05,
+      "loss": 1.2437,
+      "step": 444000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.7612271881289844e-05,
+      "loss": 1.1384,
+      "step": 444500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.7553321037871705e-05,
+      "loss": 1.2413,
+      "step": 445000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.74944880961404e-05,
+      "loss": 1.2516,
+      "step": 445500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.7435537252722262e-05,
+      "loss": 1.2607,
+      "step": 446000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.7376586409304125e-05,
+      "loss": 1.1788,
+      "step": 446500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.7317635565885986e-05,
+      "loss": 1.1807,
+      "step": 447000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.7258802624154686e-05,
+      "loss": 1.1212,
+      "step": 447500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.7199851780736546e-05,
+      "loss": 1.1052,
+      "step": 448000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.714090093731841e-05,
+      "loss": 1.1732,
+      "step": 448500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.708195009390027e-05,
+      "loss": 1.1986,
+      "step": 449000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.7022999250482134e-05,
+      "loss": 1.1305,
+      "step": 449500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.6964284210437667e-05,
+      "loss": 1.1177,
+      "step": 450000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.6905333367019528e-05,
+      "loss": 1.106,
+      "step": 450500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.684638252360139e-05,
+      "loss": 1.1806,
+      "step": 451000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.678743168018325e-05,
+      "loss": 1.1752,
+      "step": 451500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.6728480836765115e-05,
+      "loss": 1.1708,
+      "step": 452000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.6669529993346975e-05,
+      "loss": 1.1829,
+      "step": 452500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.6610697051615672e-05,
+      "loss": 1.0927,
+      "step": 453000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.6551746208197536e-05,
+      "loss": 1.1679,
+      "step": 453500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.6492795364779396e-05,
+      "loss": 1.1808,
+      "step": 454000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.6433844521361256e-05,
+      "loss": 1.1363,
+      "step": 454500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.637489367794312e-05,
+      "loss": 1.0806,
+      "step": 455000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.6316060736211817e-05,
+      "loss": 1.1608,
+      "step": 455500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.6257109892793677e-05,
+      "loss": 1.204,
+      "step": 456000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.619815904937554e-05,
+      "loss": 1.1361,
+      "step": 456500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.61392082059574e-05,
+      "loss": 1.1163,
+      "step": 457000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.6080257362539265e-05,
+      "loss": 1.1429,
+      "step": 457500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.6021306519121125e-05,
+      "loss": 1.1361,
+      "step": 458000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.596235567570299e-05,
+      "loss": 1.0787,
+      "step": 458500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.5903522733971685e-05,
+      "loss": 1.2084,
+      "step": 459000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.5844571890553546e-05,
+      "loss": 1.2054,
+      "step": 459500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.578562104713541e-05,
+      "loss": 1.16,
+      "step": 460000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.572667020371727e-05,
+      "loss": 1.1465,
+      "step": 460500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5667719360299133e-05,
+      "loss": 1.172,
+      "step": 461000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5608768516880993e-05,
+      "loss": 1.1564,
+      "step": 461500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5549935575149694e-05,
+      "loss": 1.0888,
+      "step": 462000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5490984731731554e-05,
+      "loss": 1.1401,
+      "step": 462500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.543215179000025e-05,
+      "loss": 1.1677,
+      "step": 463000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5373200946582114e-05,
+      "loss": 1.1087,
+      "step": 463500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5314250103163975e-05,
+      "loss": 1.2204,
+      "step": 464000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.5255299259745836e-05,
+      "loss": 1.1369,
+      "step": 464500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.5196348416327697e-05,
+      "loss": 1.111,
+      "step": 465000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.513739757290956e-05,
+      "loss": 1.1991,
+      "step": 465500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.5078564631178257e-05,
+      "loss": 1.0535,
+      "step": 466000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.501961378776012e-05,
+      "loss": 1.1935,
+      "step": 466500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.4960662944341981e-05,
+      "loss": 1.1171,
+      "step": 467000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.4901712100923843e-05,
+      "loss": 1.2015,
+      "step": 467500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.4842761257505703e-05,
+      "loss": 1.1403,
+      "step": 468000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.4783928315774402e-05,
+      "loss": 1.2389,
+      "step": 468500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.4724977472356264e-05,
+      "loss": 1.0835,
+      "step": 469000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.4666026628938124e-05,
+      "loss": 1.1687,
+      "step": 469500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.4607075785519988e-05,
+      "loss": 1.1683,
+      "step": 470000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.4548124942101848e-05,
+      "loss": 1.1314,
+      "step": 470500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.4489174098683711e-05,
+      "loss": 1.0656,
+      "step": 471000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.4430223255265572e-05,
+      "loss": 1.085,
+      "step": 471500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.4371272411847434e-05,
+      "loss": 1.1091,
+      "step": 472000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.4312439470116132e-05,
+      "loss": 1.2264,
+      "step": 472500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.4253488626697994e-05,
+      "loss": 1.0943,
+      "step": 473000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.4194537783279854e-05,
+      "loss": 1.0504,
+      "step": 473500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.4135586939861718e-05,
+      "loss": 1.1083,
+      "step": 474000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.4076636096443578e-05,
+      "loss": 1.1178,
+      "step": 474500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.401768525302544e-05,
+      "loss": 1.1602,
+      "step": 475000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.3958734409607302e-05,
+      "loss": 1.1508,
+      "step": 475500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.3899783566189164e-05,
+      "loss": 1.1951,
+      "step": 476000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.3840832722771024e-05,
+      "loss": 1.1492,
+      "step": 476500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.3781881879352888e-05,
+      "loss": 1.0209,
+      "step": 477000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.3722931035934748e-05,
+      "loss": 1.1813,
+      "step": 477500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.3663980192516611e-05,
+      "loss": 1.1068,
+      "step": 478000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.3605265152472145e-05,
+      "loss": 1.1782,
+      "step": 478500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.3546432210740842e-05,
+      "loss": 1.1442,
+      "step": 479000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.3487481367322703e-05,
+      "loss": 1.0932,
+      "step": 479500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.3428530523904566e-05,
+      "loss": 1.12,
+      "step": 480000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.3369579680486426e-05,
+      "loss": 1.1215,
+      "step": 480500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.331062883706829e-05,
+      "loss": 1.1167,
+      "step": 481000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.325167799365015e-05,
+      "loss": 1.1132,
+      "step": 481500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.3192727150232012e-05,
+      "loss": 1.1323,
+      "step": 482000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.313389420850071e-05,
+      "loss": 1.1822,
+      "step": 482500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.3074943365082573e-05,
+      "loss": 1.1408,
+      "step": 483000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.3015992521664433e-05,
+      "loss": 1.1328,
+      "step": 483500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.2957041678246296e-05,
+      "loss": 1.0896,
+      "step": 484000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.2898090834828157e-05,
+      "loss": 1.1126,
+      "step": 484500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.2839139991410019e-05,
+      "loss": 1.1636,
+      "step": 485000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.278018914799188e-05,
+      "loss": 1.1159,
+      "step": 485500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.2721238304573742e-05,
+      "loss": 1.1699,
+      "step": 486000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.266240536284244e-05,
+      "loss": 1.0885,
+      "step": 486500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.2603454519424301e-05,
+      "loss": 1.102,
+      "step": 487000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.2544503676006163e-05,
+      "loss": 1.1209,
+      "step": 487500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2485552832588027e-05,
+      "loss": 1.1678,
+      "step": 488000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2426601989169887e-05,
+      "loss": 1.1369,
+      "step": 488500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2367651145751749e-05,
+      "loss": 1.0852,
+      "step": 489000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2308818204020448e-05,
+      "loss": 1.1961,
+      "step": 489500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2249867360602308e-05,
+      "loss": 1.0893,
+      "step": 490000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.219091651718417e-05,
+      "loss": 1.1555,
+      "step": 490500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2131965673766032e-05,
+      "loss": 1.1691,
+      "step": 491000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2073014830347893e-05,
+      "loss": 1.1374,
+      "step": 491500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.201418188861659e-05,
+      "loss": 1.1329,
+      "step": 492000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1955231045198454e-05,
+      "loss": 1.0668,
+      "step": 492500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1896280201780314e-05,
+      "loss": 1.1424,
+      "step": 493000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1837329358362176e-05,
+      "loss": 1.1749,
+      "step": 493500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1778378514944038e-05,
+      "loss": 1.1726,
+      "step": 494000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.17194276715259e-05,
+      "loss": 1.1148,
+      "step": 494500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1660594729794599e-05,
+      "loss": 1.1525,
+      "step": 495000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1601643886376459e-05,
+      "loss": 1.1298,
+      "step": 495500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.154269304295832e-05,
+      "loss": 1.0978,
+      "step": 496000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.1483742199540181e-05,
+      "loss": 1.1544,
+      "step": 496500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.1424791356122045e-05,
+      "loss": 1.1432,
+      "step": 497000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.1366076316077578e-05,
+      "loss": 1.1145,
+      "step": 497500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.130712547265944e-05,
+      "loss": 1.0662,
+      "step": 498000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.1248174629241302e-05,
+      "loss": 1.1682,
+      "step": 498500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.1189223785823162e-05,
+      "loss": 1.0792,
+      "step": 499000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.1130272942405026e-05,
+      "loss": 1.1048,
+      "step": 499500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.1071322098986886e-05,
+      "loss": 1.129,
+      "step": 500000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.1012371255568748e-05,
+      "loss": 1.0655,
+      "step": 500500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.095342041215061e-05,
+      "loss": 1.1566,
+      "step": 501000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0894587470419307e-05,
+      "loss": 1.1391,
+      "step": 501500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0835636627001169e-05,
+      "loss": 1.1428,
+      "step": 502000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0776803685269868e-05,
+      "loss": 1.1577,
+      "step": 502500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.071785284185173e-05,
+      "loss": 1.0436,
+      "step": 503000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.065890199843359e-05,
+      "loss": 1.1324,
+      "step": 503500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0599951155015453e-05,
+      "loss": 1.2022,
+      "step": 504000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0541000311597314e-05,
+      "loss": 1.1497,
+      "step": 504500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0482049468179177e-05,
+      "loss": 1.1561,
+      "step": 505000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0423098624761037e-05,
+      "loss": 1.1448,
+      "step": 505500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.03641477813429e-05,
+      "loss": 1.1069,
+      "step": 506000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0305196937924761e-05,
+      "loss": 1.2171,
+      "step": 506500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0246246094506623e-05,
+      "loss": 1.0737,
+      "step": 507000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0187295251088483e-05,
+      "loss": 1.1163,
+      "step": 507500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0128344407670347e-05,
+      "loss": 1.052,
+      "step": 508000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0069393564252207e-05,
+      "loss": 1.167,
+      "step": 508500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0010560622520906e-05,
+      "loss": 1.1351,
+      "step": 509000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.951727680789605e-06,
+      "loss": 1.0973,
+      "step": 509500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.892776837371465e-06,
+      "loss": 1.192,
+      "step": 510000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.833825993953327e-06,
+      "loss": 1.0856,
+      "step": 510500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.774875150535189e-06,
+      "loss": 1.0788,
+      "step": 511000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.71592430711705e-06,
+      "loss": 1.1291,
+      "step": 511500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.65697346369891e-06,
+      "loss": 1.1339,
+      "step": 512000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.598022620280774e-06,
+      "loss": 1.1495,
+      "step": 512500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.539189678549471e-06,
+      "loss": 1.1007,
+      "step": 513000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.480238835131335e-06,
+      "loss": 1.129,
+      "step": 513500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.421287991713195e-06,
+      "loss": 1.1553,
+      "step": 514000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.362337148295057e-06,
+      "loss": 1.0732,
+      "step": 514500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.303386304876919e-06,
+      "loss": 1.0946,
+      "step": 515000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.244553363145616e-06,
+      "loss": 1.0184,
+      "step": 515500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.185602519727478e-06,
+      "loss": 1.1658,
+      "step": 516000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.12665167630934e-06,
+      "loss": 1.1194,
+      "step": 516500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.067700832891202e-06,
+      "loss": 1.132,
+      "step": 517000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.008749989473062e-06,
+      "loss": 1.1043,
+      "step": 517500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.949799146054925e-06,
+      "loss": 1.1069,
+      "step": 518000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.890966204323622e-06,
+      "loss": 1.07,
+      "step": 518500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.832015360905484e-06,
+      "loss": 1.209,
+      "step": 519000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.773064517487346e-06,
+      "loss": 1.1133,
+      "step": 519500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.714113674069208e-06,
+      "loss": 1.1408,
+      "step": 520000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.655162830651068e-06,
+      "loss": 1.0819,
+      "step": 520500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.59621198723293e-06,
+      "loss": 1.1985,
+      "step": 521000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.537261143814792e-06,
+      "loss": 1.0531,
+      "step": 521500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.478310300396654e-06,
+      "loss": 1.1547,
+      "step": 522000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.419359456978516e-06,
+      "loss": 1.1247,
+      "step": 522500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.36064441693405e-06,
+      "loss": 1.132,
+      "step": 523000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 8.301693573515912e-06,
+      "loss": 1.0849,
+      "step": 523500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 8.24286063178461e-06,
+      "loss": 1.06,
+      "step": 524000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 8.183909788366472e-06,
+      "loss": 1.2021,
+      "step": 524500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 8.124958944948334e-06,
+      "loss": 1.1077,
+      "step": 525000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 8.066008101530194e-06,
+      "loss": 1.0834,
+      "step": 525500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 8.007057258112056e-06,
+      "loss": 1.1898,
+      "step": 526000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.948106414693918e-06,
+      "loss": 1.0924,
+      "step": 526500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.88915557127578e-06,
+      "loss": 1.1666,
+      "step": 527000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.830204727857642e-06,
+      "loss": 1.1483,
+      "step": 527500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.771253884439504e-06,
+      "loss": 1.1006,
+      "step": 528000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.712303041021366e-06,
+      "loss": 1.1411,
+      "step": 528500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.653352197603226e-06,
+      "loss": 1.1328,
+      "step": 529000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.594401354185089e-06,
+      "loss": 1.1238,
+      "step": 529500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.535568412453786e-06,
+      "loss": 1.0613,
+      "step": 530000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.476617569035648e-06,
+      "loss": 1.1304,
+      "step": 530500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.4176667256175095e-06,
+      "loss": 1.0985,
+      "step": 531000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 7.3587158821993706e-06,
+      "loss": 1.0552,
+      "step": 531500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 7.299882940468069e-06,
+      "loss": 1.0772,
+      "step": 532000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 7.240932097049931e-06,
+      "loss": 1.0503,
+      "step": 532500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 7.181981253631792e-06,
+      "loss": 1.112,
+      "step": 533000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 7.12314831190049e-06,
+      "loss": 1.0917,
+      "step": 533500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 7.064197468482352e-06,
+      "loss": 1.0451,
+      "step": 534000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 7.005246625064214e-06,
+      "loss": 1.1209,
+      "step": 534500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.946295781646075e-06,
+      "loss": 1.104,
+      "step": 535000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.887344938227937e-06,
+      "loss": 1.0351,
+      "step": 535500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.828394094809799e-06,
+      "loss": 1.0651,
+      "step": 536000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.769561153078497e-06,
+      "loss": 1.123,
+      "step": 536500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.7106103096603585e-06,
+      "loss": 1.0592,
+      "step": 537000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.65165946624222e-06,
+      "loss": 1.1219,
+      "step": 537500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.592708622824082e-06,
+      "loss": 1.1148,
+      "step": 538000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.533757779405943e-06,
+      "loss": 1.146,
+      "step": 538500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.474806935987805e-06,
+      "loss": 1.0631,
+      "step": 539000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.415856092569667e-06,
+      "loss": 1.0659,
+      "step": 539500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.356905249151528e-06,
+      "loss": 1.0902,
+      "step": 540000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.29795440573339e-06,
+      "loss": 1.1162,
+      "step": 540500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.239121464002088e-06,
+      "loss": 1.0639,
+      "step": 541000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.180170620583949e-06,
+      "loss": 1.104,
+      "step": 541500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.121219777165811e-06,
+      "loss": 1.094,
+      "step": 542000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.062268933747673e-06,
+      "loss": 1.1028,
+      "step": 542500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.0034359920163716e-06,
+      "loss": 1.119,
+      "step": 543000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 5.944485148598233e-06,
+      "loss": 1.1794,
+      "step": 543500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 5.8855343051800945e-06,
+      "loss": 1.026,
+      "step": 544000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 5.826583461761956e-06,
+      "loss": 1.0395,
+      "step": 544500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 5.7676326183438175e-06,
+      "loss": 1.0916,
+      "step": 545000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 5.708799676612515e-06,
+      "loss": 1.0913,
+      "step": 545500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 5.649848833194377e-06,
+      "loss": 0.991,
+      "step": 546000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 5.590897989776238e-06,
+      "loss": 1.0942,
+      "step": 546500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 5.5319471463581e-06,
+      "loss": 1.0675,
+      "step": 547000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.473114204626799e-06,
+      "loss": 1.1203,
+      "step": 547500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.414163361208661e-06,
+      "loss": 1.1392,
+      "step": 548000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.355212517790522e-06,
+      "loss": 1.1748,
+      "step": 548500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.296261674372384e-06,
+      "loss": 1.1011,
+      "step": 549000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.237310830954246e-06,
+      "loss": 1.1237,
+      "step": 549500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.178477889222943e-06,
+      "loss": 1.14,
+      "step": 550000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.1195270458048046e-06,
+      "loss": 1.1082,
+      "step": 550500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.0605762023866665e-06,
+      "loss": 1.0511,
+      "step": 551000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.0016253589685275e-06,
+      "loss": 1.0746,
+      "step": 551500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.942674515550389e-06,
+      "loss": 1.0512,
+      "step": 552000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.883723672132251e-06,
+      "loss": 1.1061,
+      "step": 552500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.82489073040095e-06,
+      "loss": 1.1637,
+      "step": 553000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.765939886982811e-06,
+      "loss": 1.033,
+      "step": 553500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.706989043564673e-06,
+      "loss": 1.1271,
+      "step": 554000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.648038200146535e-06,
+      "loss": 1.115,
+      "step": 554500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.589205258415232e-06,
+      "loss": 1.1516,
+      "step": 555000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.530254414997094e-06,
+      "loss": 1.0802,
+      "step": 555500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.471303571578956e-06,
+      "loss": 1.0404,
+      "step": 556000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.4124706298476544e-06,
+      "loss": 1.0606,
+      "step": 556500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.3535197864295155e-06,
+      "loss": 1.0665,
+      "step": 557000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.294568943011377e-06,
+      "loss": 1.0785,
+      "step": 557500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.235618099593239e-06,
+      "loss": 1.06,
+      "step": 558000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.1766672561751e-06,
+      "loss": 1.1075,
+      "step": 558500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.117716412756962e-06,
+      "loss": 1.0468,
+      "step": 559000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.058765569338824e-06,
+      "loss": 1.0891,
+      "step": 559500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.999814725920685e-06,
+      "loss": 1.1616,
+      "step": 560000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.940863882502547e-06,
+      "loss": 1.0247,
+      "step": 560500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.882030940771245e-06,
+      "loss": 1.0789,
+      "step": 561000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.823080097353107e-06,
+      "loss": 1.0501,
+      "step": 561500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.7641292539349687e-06,
+      "loss": 1.0249,
+      "step": 562000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.7051784105168302e-06,
+      "loss": 1.1371,
+      "step": 562500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.646345468785528e-06,
+      "loss": 1.0623,
+      "step": 563000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.58739462536739e-06,
+      "loss": 1.003,
+      "step": 563500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5284437819492515e-06,
+      "loss": 1.0979,
+      "step": 564000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.4696108402179498e-06,
+      "loss": 1.126,
+      "step": 564500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.4106599967998112e-06,
+      "loss": 1.053,
+      "step": 565000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.3517091533816727e-06,
+      "loss": 1.0729,
+      "step": 565500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.2927583099635346e-06,
+      "loss": 1.0286,
+      "step": 566000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.233807466545396e-06,
+      "loss": 1.1291,
+      "step": 566500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.174856623127258e-06,
+      "loss": 1.075,
+      "step": 567000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.1159057797091195e-06,
+      "loss": 1.1489,
+      "step": 567500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.056954936290981e-06,
+      "loss": 1.0357,
+      "step": 568000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.998004092872843e-06,
+      "loss": 1.0894,
+      "step": 568500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.9390532494547043e-06,
+      "loss": 1.0034,
+      "step": 569000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.880102406036566e-06,
+      "loss": 1.1578,
+      "step": 569500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.821269464305264e-06,
+      "loss": 1.0334,
+      "step": 570000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.762318620887126e-06,
+      "loss": 1.096,
+      "step": 570500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.7033677774689875e-06,
+      "loss": 1.069,
+      "step": 571000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.644416934050849e-06,
+      "loss": 1.0984,
+      "step": 571500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.5855839923195472e-06,
+      "loss": 1.0847,
+      "step": 572000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.5266331489014087e-06,
+      "loss": 1.098,
+      "step": 572500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.4676823054832706e-06,
+      "loss": 1.0521,
+      "step": 573000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.4088493637519685e-06,
+      "loss": 1.1309,
+      "step": 573500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.34989852033383e-06,
+      "loss": 1.0719,
+      "step": 574000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.290947676915692e-06,
+      "loss": 1.0152,
+      "step": 574500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.2319968334975533e-06,
+      "loss": 1.0853,
+      "step": 575000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.1730459900794152e-06,
+      "loss": 1.0781,
+      "step": 575500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.1140951466612767e-06,
+      "loss": 1.1535,
+      "step": 576000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.055144303243138e-06,
+      "loss": 1.0598,
+      "step": 576500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.996193459825e-06,
+      "loss": 1.1467,
+      "step": 577000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9372426164068615e-06,
+      "loss": 1.0705,
+      "step": 577500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.8782917729887232e-06,
+      "loss": 1.1656,
+      "step": 578000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.8193409295705851e-06,
+      "loss": 1.0631,
+      "step": 578500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.7603900861524468e-06,
+      "loss": 1.1149,
+      "step": 579000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.7015571444211447e-06,
+      "loss": 1.1648,
+      "step": 579500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.6426063010030062e-06,
+      "loss": 1.0288,
+      "step": 580000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.5836554575848679e-06,
+      "loss": 1.0671,
+      "step": 580500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.5247046141667297e-06,
+      "loss": 1.048,
+      "step": 581000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4658716724354276e-06,
+      "loss": 1.0739,
+      "step": 581500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4069208290172893e-06,
+      "loss": 1.0812,
+      "step": 582000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.347969985599151e-06,
+      "loss": 1.0971,
+      "step": 582500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2890191421810125e-06,
+      "loss": 1.0779,
+      "step": 583000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2300682987628744e-06,
+      "loss": 1.1095,
+      "step": 583500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.171117455344736e-06,
+      "loss": 1.0638,
+      "step": 584000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.1121666119265977e-06,
+      "loss": 1.1076,
+      "step": 584500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0532157685084594e-06,
+      "loss": 1.1115,
+      "step": 585000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.942649250903211e-07,
+      "loss": 1.0967,
+      "step": 585500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.355498850458553e-07,
+      "loss": 1.0556,
+      "step": 586000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.765990416277169e-07,
+      "loss": 1.0782,
+      "step": 586500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.176481982095787e-07,
+      "loss": 1.0308,
+      "step": 587000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.586973547914403e-07,
+      "loss": 1.047,
+      "step": 587500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.997465113733019e-07,
+      "loss": 1.0906,
+      "step": 588000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.407956679551636e-07,
+      "loss": 1.0376,
+      "step": 588500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.818448245370253e-07,
+      "loss": 1.0616,
+      "step": 589000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.230118828057232e-07,
+      "loss": 1.0688,
+      "step": 589500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.640610393875849e-07,
+      "loss": 1.0337,
+      "step": 590000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.051101959694466e-07,
+      "loss": 1.1031,
+      "step": 590500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.461593525513083e-07,
+      "loss": 1.1921,
+      "step": 591000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.872085091331699e-07,
+      "loss": 1.1171,
+      "step": 591500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.282576657150316e-07,
+      "loss": 1.0845,
+      "step": 592000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.6930682229689326e-07,
+      "loss": 1.0091,
+      "step": 592500
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.1047388056559123e-07,
+      "loss": 1.0612,
+      "step": 593000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.15230371474529e-08,
+      "loss": 1.0389,
+      "step": 593500
+    },
     {
       "epoch": 3.0,
-      "step": 24495,
-      "total_flos": 1.290813800478588e+16,
-      "train_loss": 1.6035245578856974,
-      "train_runtime": 14452.6928,
-      "train_samples_per_second": 1.695,
-      "train_steps_per_second": 1.695
+      "step": 593715,
+      "total_flos": 6.354828841670784e+16,
+      "train_loss": 1.3742914497521532,
+      "train_runtime": 129030.9031,
+      "train_samples_per_second": 18.405,
+      "train_steps_per_second": 4.601
     }
   ],
-  "max_steps": 24495,
+  "max_steps": 593715,
   "num_train_epochs": 3,
-  "total_flos": 1.290813800478588e+16,
+  "total_flos": 6.354828841670784e+16,
   "trial_name": null,
   "trial_params": null
 }