diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,15010 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.18433058885567705,
+  "global_step": 1249500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998148198352554e-05,
+      "loss": 2.5633,
+      "step": 500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99962963967051e-05,
+      "loss": 2.4025,
+      "step": 1000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999444459505765e-05,
+      "loss": 2.4857,
+      "step": 1500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9992592793410196e-05,
+      "loss": 2.4036,
+      "step": 2000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999074099176275e-05,
+      "loss": 2.4167,
+      "step": 2500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.998888919011529e-05,
+      "loss": 2.4199,
+      "step": 3000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.998703738846784e-05,
+      "loss": 2.3777,
+      "step": 3500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.998518558682039e-05,
+      "loss": 2.3574,
+      "step": 4000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9983333785172935e-05,
+      "loss": 2.3653,
+      "step": 4500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.998148198352549e-05,
+      "loss": 2.3203,
+      "step": 5000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.997963018187803e-05,
+      "loss": 2.3002,
+      "step": 5500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9977778380230584e-05,
+      "loss": 2.2327,
+      "step": 6000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.997592657858313e-05,
+      "loss": 2.3823,
+      "step": 6500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.997407477693568e-05,
+      "loss": 2.3095,
+      "step": 7000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.997222297528823e-05,
+      "loss": 2.3012,
+      "step": 7500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.997037117364078e-05,
+      "loss": 2.2502,
+      "step": 8000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.996851937199332e-05,
+      "loss": 2.3248,
+      "step": 8500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.996666757034587e-05,
+      "loss": 2.3523,
+      "step": 9000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.996481576869842e-05,
+      "loss": 2.2897,
+      "step": 9500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.996296396705097e-05,
+      "loss": 2.2968,
+      "step": 10000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.996111169011484e-05,
+      "loss": 2.2047,
+      "step": 10500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.995925986583459e-05,
+      "loss": 2.3323,
+      "step": 11000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.995740804155435e-05,
+      "loss": 2.2564,
+      "step": 11500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99555562172741e-05,
+      "loss": 2.3009,
+      "step": 12000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.995370439299386e-05,
+      "loss": 2.2419,
+      "step": 12500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.995185256871361e-05,
+      "loss": 2.2346,
+      "step": 13000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.995000074443336e-05,
+      "loss": 2.304,
+      "step": 13500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9948148920153116e-05,
+      "loss": 2.223,
+      "step": 14000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.994629709587287e-05,
+      "loss": 2.2424,
+      "step": 14500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.994444527159263e-05,
+      "loss": 2.2479,
+      "step": 15000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.994259344731238e-05,
+      "loss": 2.2607,
+      "step": 15500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9940741623032137e-05,
+      "loss": 2.2446,
+      "step": 16000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.993888979875189e-05,
+      "loss": 2.1891,
+      "step": 16500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.993703797447164e-05,
+      "loss": 2.2201,
+      "step": 17000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9935186150191394e-05,
+      "loss": 2.2409,
+      "step": 17500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.993333432591115e-05,
+      "loss": 2.2319,
+      "step": 18000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.993148250163091e-05,
+      "loss": 2.2554,
+      "step": 18500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.992963067735066e-05,
+      "loss": 2.2191,
+      "step": 19000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9927778853070415e-05,
+      "loss": 2.1847,
+      "step": 19500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9925927028790165e-05,
+      "loss": 2.1872,
+      "step": 20000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9924075204509915e-05,
+      "loss": 2.2293,
+      "step": 20500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.992222338022968e-05,
+      "loss": 2.2537,
+      "step": 21000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.992037155594943e-05,
+      "loss": 2.2332,
+      "step": 21500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9918519731669186e-05,
+      "loss": 2.2963,
+      "step": 22000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9916667907388936e-05,
+      "loss": 2.2041,
+      "step": 22500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9914816083108686e-05,
+      "loss": 2.1912,
+      "step": 23000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.991296425882844e-05,
+      "loss": 2.1984,
+      "step": 23500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99111124345482e-05,
+      "loss": 2.2336,
+      "step": 24000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.990926061026796e-05,
+      "loss": 2.1302,
+      "step": 24500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.990740878598771e-05,
+      "loss": 2.2171,
+      "step": 25000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9905556961707464e-05,
+      "loss": 2.1807,
+      "step": 25500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9903705137427215e-05,
+      "loss": 2.1588,
+      "step": 26000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9901853313146965e-05,
+      "loss": 2.1912,
+      "step": 26500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.990000148886672e-05,
+      "loss": 2.2455,
+      "step": 27000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.989814966458648e-05,
+      "loss": 2.1608,
+      "step": 27500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9896297840306236e-05,
+      "loss": 2.1814,
+      "step": 28000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9894446016025986e-05,
+      "loss": 2.0651,
+      "step": 28500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.989259419174574e-05,
+      "loss": 2.1137,
+      "step": 29000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.989074236746549e-05,
+      "loss": 2.2112,
+      "step": 29500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.988889054318524e-05,
+      "loss": 2.1663,
+      "step": 30000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.988703871890501e-05,
+      "loss": 2.1839,
+      "step": 30500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.988518689462476e-05,
+      "loss": 2.1793,
+      "step": 31000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9883335070344514e-05,
+      "loss": 2.1851,
+      "step": 31500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9881483246064264e-05,
+      "loss": 2.1889,
+      "step": 32000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.987963142178402e-05,
+      "loss": 2.0989,
+      "step": 32500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.987777959750377e-05,
+      "loss": 2.102,
+      "step": 33000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.987592777322353e-05,
+      "loss": 2.0911,
+      "step": 33500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9874075948943285e-05,
+      "loss": 2.2145,
+      "step": 34000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9872224124663035e-05,
+      "loss": 2.1246,
+      "step": 34500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.987037230038279e-05,
+      "loss": 2.1473,
+      "step": 35000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.986852047610254e-05,
+      "loss": 2.1508,
+      "step": 35500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.986666865182229e-05,
+      "loss": 2.1553,
+      "step": 36000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.986481682754205e-05,
+      "loss": 2.1734,
+      "step": 36500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9862965003261806e-05,
+      "loss": 2.2153,
+      "step": 37000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.986111317898156e-05,
+      "loss": 2.1443,
+      "step": 37500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9859261354701314e-05,
+      "loss": 2.2392,
+      "step": 38000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.985740953042107e-05,
+      "loss": 2.117,
+      "step": 38500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.985555770614082e-05,
+      "loss": 2.1762,
+      "step": 39000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.985370588186057e-05,
+      "loss": 2.1363,
+      "step": 39500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9851854057580335e-05,
+      "loss": 2.1806,
+      "step": 40000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9850002233300085e-05,
+      "loss": 2.1484,
+      "step": 40500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.984815040901984e-05,
+      "loss": 2.1418,
+      "step": 41000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.984629858473959e-05,
+      "loss": 2.1191,
+      "step": 41500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.984444676045935e-05,
+      "loss": 2.2352,
+      "step": 42000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.98425949361791e-05,
+      "loss": 2.1433,
+      "step": 42500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.984074311189885e-05,
+      "loss": 2.1621,
+      "step": 43000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.983889128761861e-05,
+      "loss": 2.0863,
+      "step": 43500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.983703946333836e-05,
+      "loss": 2.201,
+      "step": 44000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.983518763905812e-05,
+      "loss": 2.1628,
+      "step": 44500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.983333581477787e-05,
+      "loss": 2.1656,
+      "step": 45000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.983148399049763e-05,
+      "loss": 2.154,
+      "step": 45500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.982963216621738e-05,
+      "loss": 2.1882,
+      "step": 46000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9827780341937134e-05,
+      "loss": 2.1192,
+      "step": 46500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.982592851765689e-05,
+      "loss": 2.0503,
+      "step": 47000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.982407669337664e-05,
+      "loss": 2.1834,
+      "step": 47500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.98222248690964e-05,
+      "loss": 2.1964,
+      "step": 48000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.982037304481615e-05,
+      "loss": 2.1844,
+      "step": 48500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9818521220535905e-05,
+      "loss": 2.15,
+      "step": 49000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.981666939625566e-05,
+      "loss": 2.1419,
+      "step": 49500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.981481757197541e-05,
+      "loss": 2.1146,
+      "step": 50000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.981296574769517e-05,
+      "loss": 2.1196,
+      "step": 50500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.981111392341492e-05,
+      "loss": 2.0982,
+      "step": 51000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9809262099134677e-05,
+      "loss": 2.1951,
+      "step": 51500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.980741027485443e-05,
+      "loss": 2.0885,
+      "step": 52000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.980555845057418e-05,
+      "loss": 2.0822,
+      "step": 52500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.980370662629394e-05,
+      "loss": 2.107,
+      "step": 53000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.980185480201369e-05,
+      "loss": 2.0801,
+      "step": 53500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.980000297773345e-05,
+      "loss": 2.1561,
+      "step": 54000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.97981511534532e-05,
+      "loss": 2.0842,
+      "step": 54500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9796299329172955e-05,
+      "loss": 2.0803,
+      "step": 55000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9794447504892705e-05,
+      "loss": 2.2168,
+      "step": 55500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.979259568061246e-05,
+      "loss": 2.114,
+      "step": 56000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.979074385633222e-05,
+      "loss": 2.1965,
+      "step": 56500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.978889203205197e-05,
+      "loss": 2.1872,
+      "step": 57000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9787040207771726e-05,
+      "loss": 2.1307,
+      "step": 57500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9785188383491476e-05,
+      "loss": 2.1543,
+      "step": 58000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.978333655921123e-05,
+      "loss": 2.1638,
+      "step": 58500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.978148473493099e-05,
+      "loss": 2.152,
+      "step": 59000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.977963291065074e-05,
+      "loss": 2.185,
+      "step": 59500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.97777810863705e-05,
+      "loss": 2.083,
+      "step": 60000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.977592926209025e-05,
+      "loss": 2.1583,
+      "step": 60500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9774077437810004e-05,
+      "loss": 2.1324,
+      "step": 61000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9772225613529754e-05,
+      "loss": 2.1022,
+      "step": 61500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.977037378924951e-05,
+      "loss": 2.1016,
+      "step": 62000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.976852196496927e-05,
+      "loss": 2.2154,
+      "step": 62500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.976667014068902e-05,
+      "loss": 2.1377,
+      "step": 63000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9764818316408775e-05,
+      "loss": 2.1611,
+      "step": 63500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9762966492128526e-05,
+      "loss": 2.1962,
+      "step": 64000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.976111466784828e-05,
+      "loss": 2.1179,
+      "step": 64500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.975926284356803e-05,
+      "loss": 2.136,
+      "step": 65000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.975741101928779e-05,
+      "loss": 2.0581,
+      "step": 65500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.975555919500755e-05,
+      "loss": 2.1673,
+      "step": 66000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.97537073707273e-05,
+      "loss": 2.1709,
+      "step": 66500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9751855546447054e-05,
+      "loss": 2.1291,
+      "step": 67000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9750003722166804e-05,
+      "loss": 2.1081,
+      "step": 67500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.974815189788656e-05,
+      "loss": 2.0995,
+      "step": 68000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.974630007360631e-05,
+      "loss": 2.1093,
+      "step": 68500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.974444824932607e-05,
+      "loss": 2.0995,
+      "step": 69000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9742596425045825e-05,
+      "loss": 2.3111,
+      "step": 69500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9740744600765575e-05,
+      "loss": 2.2349,
+      "step": 70000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.973889277648533e-05,
+      "loss": 2.0986,
+      "step": 70500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.973704095220508e-05,
+      "loss": 2.1087,
+      "step": 71000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.973518912792484e-05,
+      "loss": 2.1279,
+      "step": 71500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9733337303644596e-05,
+      "loss": 2.1361,
+      "step": 72000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9731485479364346e-05,
+      "loss": 2.1142,
+      "step": 72500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.97296336550841e-05,
+      "loss": 2.1393,
+      "step": 73000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9727781830803853e-05,
+      "loss": 2.0751,
+      "step": 73500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.972593000652361e-05,
+      "loss": 2.1786,
+      "step": 74000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.972407818224336e-05,
+      "loss": 2.1415,
+      "step": 74500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.972222635796312e-05,
+      "loss": 2.094,
+      "step": 75000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9720374533682874e-05,
+      "loss": 2.158,
+      "step": 75500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9718522709402625e-05,
+      "loss": 2.0468,
+      "step": 76000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.971667088512238e-05,
+      "loss": 2.082,
+      "step": 76500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.971481906084213e-05,
+      "loss": 2.1283,
+      "step": 77000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.971296723656189e-05,
+      "loss": 2.1468,
+      "step": 77500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.971111541228164e-05,
+      "loss": 2.1154,
+      "step": 78000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9709263588001396e-05,
+      "loss": 2.1036,
+      "step": 78500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.970741176372115e-05,
+      "loss": 2.1031,
+      "step": 79000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.97055599394409e-05,
+      "loss": 2.0857,
+      "step": 79500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.970370811516066e-05,
+      "loss": 2.1558,
+      "step": 80000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.970185629088041e-05,
+      "loss": 2.0955,
+      "step": 80500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.970000446660017e-05,
+      "loss": 2.0567,
+      "step": 81000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9698152642319924e-05,
+      "loss": 2.1498,
+      "step": 81500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9696300818039674e-05,
+      "loss": 2.1512,
+      "step": 82000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.969444899375943e-05,
+      "loss": 2.098,
+      "step": 82500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.969259716947918e-05,
+      "loss": 2.1491,
+      "step": 83000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.969074534519894e-05,
+      "loss": 2.1296,
+      "step": 83500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.968889352091869e-05,
+      "loss": 2.064,
+      "step": 84000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9687041696638445e-05,
+      "loss": 2.148,
+      "step": 84500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.96851898723582e-05,
+      "loss": 2.1478,
+      "step": 85000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.968333804807795e-05,
+      "loss": 2.1299,
+      "step": 85500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.968148622379771e-05,
+      "loss": 2.1889,
+      "step": 86000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.967963439951746e-05,
+      "loss": 2.0996,
+      "step": 86500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9677782575237216e-05,
+      "loss": 2.1575,
+      "step": 87000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.967593075095697e-05,
+      "loss": 2.12,
+      "step": 87500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9674078926676724e-05,
+      "loss": 2.094,
+      "step": 88000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.967222710239648e-05,
+      "loss": 2.1591,
+      "step": 88500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.967037527811623e-05,
+      "loss": 2.1463,
+      "step": 89000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.966852345383599e-05,
+      "loss": 2.0924,
+      "step": 89500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.966667162955574e-05,
+      "loss": 2.1121,
+      "step": 90000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9664819805275495e-05,
+      "loss": 2.1229,
+      "step": 90500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.966296798099525e-05,
+      "loss": 2.1183,
+      "step": 91000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9661116156715e-05,
+      "loss": 2.0884,
+      "step": 91500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.965926433243476e-05,
+      "loss": 2.1724,
+      "step": 92000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.965741250815451e-05,
+      "loss": 2.1064,
+      "step": 92500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9655560683874266e-05,
+      "loss": 2.1236,
+      "step": 93000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9653708859594016e-05,
+      "loss": 2.159,
+      "step": 93500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.965185703531377e-05,
+      "loss": 2.0801,
+      "step": 94000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.965000521103353e-05,
+      "loss": 2.1024,
+      "step": 94500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.964815338675328e-05,
+      "loss": 2.1097,
+      "step": 95000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.964630156247304e-05,
+      "loss": 2.1249,
+      "step": 95500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.964444973819279e-05,
+      "loss": 2.0557,
+      "step": 96000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9642597913912544e-05,
+      "loss": 2.0868,
+      "step": 96500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9640746089632294e-05,
+      "loss": 2.1668,
+      "step": 97000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.963889426535205e-05,
+      "loss": 2.1173,
+      "step": 97500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.963704244107181e-05,
+      "loss": 2.1173,
+      "step": 98000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.963519061679156e-05,
+      "loss": 2.0945,
+      "step": 98500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9633338792511315e-05,
+      "loss": 2.0548,
+      "step": 99000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9631486968231066e-05,
+      "loss": 2.0306,
+      "step": 99500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.962963514395082e-05,
+      "loss": 2.0759,
+      "step": 100000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.962778331967058e-05,
+      "loss": 2.1732,
+      "step": 100500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.962593149539033e-05,
+      "loss": 2.1157,
+      "step": 101000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9624079671110087e-05,
+      "loss": 2.148,
+      "step": 101500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.962222784682984e-05,
+      "loss": 2.1079,
+      "step": 102000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9620376022549594e-05,
+      "loss": 2.127,
+      "step": 102500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9618524198269344e-05,
+      "loss": 2.1,
+      "step": 103000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.96166723739891e-05,
+      "loss": 2.071,
+      "step": 103500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.961482054970886e-05,
+      "loss": 2.1219,
+      "step": 104000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.961296872542861e-05,
+      "loss": 2.1071,
+      "step": 104500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9611116901148365e-05,
+      "loss": 2.0968,
+      "step": 105000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9609265076868115e-05,
+      "loss": 2.1153,
+      "step": 105500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.960741325258787e-05,
+      "loss": 2.0603,
+      "step": 106000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.960556142830762e-05,
+      "loss": 2.1243,
+      "step": 106500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.960370960402738e-05,
+      "loss": 2.1658,
+      "step": 107000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9601857779747136e-05,
+      "loss": 2.0951,
+      "step": 107500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9600005955466886e-05,
+      "loss": 2.1128,
+      "step": 108000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.959815413118664e-05,
+      "loss": 2.1391,
+      "step": 108500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.959630230690639e-05,
+      "loss": 2.1156,
+      "step": 109000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.959445048262615e-05,
+      "loss": 2.0878,
+      "step": 109500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.959259865834591e-05,
+      "loss": 2.1156,
+      "step": 110000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.959074683406566e-05,
+      "loss": 2.1387,
+      "step": 110500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9588895009785414e-05,
+      "loss": 2.1429,
+      "step": 111000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9587043185505165e-05,
+      "loss": 2.1445,
+      "step": 111500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.958519136122492e-05,
+      "loss": 2.1135,
+      "step": 112000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.958333953694467e-05,
+      "loss": 2.1144,
+      "step": 112500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.958148771266443e-05,
+      "loss": 2.1433,
+      "step": 113000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9579635888384186e-05,
+      "loss": 2.1211,
+      "step": 113500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9577784064103936e-05,
+      "loss": 2.1933,
+      "step": 114000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.957593223982369e-05,
+      "loss": 2.1649,
+      "step": 114500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.957408041554344e-05,
+      "loss": 2.0976,
+      "step": 115000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.95722285912632e-05,
+      "loss": 2.116,
+      "step": 115500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.957037676698295e-05,
+      "loss": 2.0661,
+      "step": 116000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.956852494270271e-05,
+      "loss": 2.1022,
+      "step": 116500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9566673118422464e-05,
+      "loss": 2.1366,
+      "step": 117000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9564821294142214e-05,
+      "loss": 2.1154,
+      "step": 117500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.956296946986197e-05,
+      "loss": 2.0719,
+      "step": 118000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.956111764558172e-05,
+      "loss": 2.1059,
+      "step": 118500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.955926582130148e-05,
+      "loss": 2.0687,
+      "step": 119000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.955741399702123e-05,
+      "loss": 2.1175,
+      "step": 119500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9555562172740985e-05,
+      "loss": 2.1016,
+      "step": 120000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.955371034846074e-05,
+      "loss": 2.091,
+      "step": 120500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.955185852418049e-05,
+      "loss": 2.1525,
+      "step": 121000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.955000669990025e-05,
+      "loss": 2.1353,
+      "step": 121500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.954815487562e-05,
+      "loss": 2.1404,
+      "step": 122000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9546303051339756e-05,
+      "loss": 2.1124,
+      "step": 122500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.954445122705951e-05,
+      "loss": 2.0695,
+      "step": 123000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9542599402779263e-05,
+      "loss": 2.0679,
+      "step": 123500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.954074757849902e-05,
+      "loss": 2.0789,
+      "step": 124000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.953889575421877e-05,
+      "loss": 2.1837,
+      "step": 124500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.953704392993853e-05,
+      "loss": 2.0892,
+      "step": 125000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.953519210565828e-05,
+      "loss": 2.1429,
+      "step": 125500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9533340281378035e-05,
+      "loss": 2.1141,
+      "step": 126000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.953148845709779e-05,
+      "loss": 2.1018,
+      "step": 126500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.952963663281754e-05,
+      "loss": 2.0873,
+      "step": 127000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.95277848085373e-05,
+      "loss": 2.0736,
+      "step": 127500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.952593298425705e-05,
+      "loss": 2.0752,
+      "step": 128000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9524081159976806e-05,
+      "loss": 2.102,
+      "step": 128500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9522229335696556e-05,
+      "loss": 2.132,
+      "step": 129000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.952037751141631e-05,
+      "loss": 2.1201,
+      "step": 129500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.951852568713607e-05,
+      "loss": 2.1781,
+      "step": 130000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.951667386285582e-05,
+      "loss": 2.0607,
+      "step": 130500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.951482203857558e-05,
+      "loss": 2.1332,
+      "step": 131000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.951297021429533e-05,
+      "loss": 2.0999,
+      "step": 131500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9511118390015084e-05,
+      "loss": 2.1018,
+      "step": 132000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.950926656573484e-05,
+      "loss": 2.0752,
+      "step": 132500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.950741474145459e-05,
+      "loss": 2.1125,
+      "step": 133000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.950556291717435e-05,
+      "loss": 2.0988,
+      "step": 133500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.95037110928941e-05,
+      "loss": 2.0444,
+      "step": 134000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9501859268613855e-05,
+      "loss": 2.0619,
+      "step": 134500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9500007444333606e-05,
+      "loss": 2.1532,
+      "step": 135000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.949815562005336e-05,
+      "loss": 2.1091,
+      "step": 135500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.949630379577312e-05,
+      "loss": 2.0799,
+      "step": 136000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.949445197149287e-05,
+      "loss": 2.0619,
+      "step": 136500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9492600147212627e-05,
+      "loss": 2.0687,
+      "step": 137000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.949074832293238e-05,
+      "loss": 2.133,
+      "step": 137500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9488896498652134e-05,
+      "loss": 2.0766,
+      "step": 138000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9487044674371884e-05,
+      "loss": 2.1147,
+      "step": 138500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.948519285009164e-05,
+      "loss": 2.1444,
+      "step": 139000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.94833410258114e-05,
+      "loss": 2.138,
+      "step": 139500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.948148920153115e-05,
+      "loss": 2.0898,
+      "step": 140000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9479637377250905e-05,
+      "loss": 2.0965,
+      "step": 140500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9477785552970655e-05,
+      "loss": 2.0402,
+      "step": 141000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.947593372869041e-05,
+      "loss": 2.0965,
+      "step": 141500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.947408190441017e-05,
+      "loss": 2.1203,
+      "step": 142000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.947223008012992e-05,
+      "loss": 2.0735,
+      "step": 142500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9470378255849676e-05,
+      "loss": 2.0774,
+      "step": 143000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9468526431569426e-05,
+      "loss": 2.0433,
+      "step": 143500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.946667460728918e-05,
+      "loss": 2.0941,
+      "step": 144000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.946482278300893e-05,
+      "loss": 2.0883,
+      "step": 144500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.946297095872869e-05,
+      "loss": 2.0986,
+      "step": 145000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.946111913444845e-05,
+      "loss": 2.1054,
+      "step": 145500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.94592673101682e-05,
+      "loss": 2.095,
+      "step": 146000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9457415485887954e-05,
+      "loss": 2.048,
+      "step": 146500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9455563661607704e-05,
+      "loss": 2.1582,
+      "step": 147000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.945371183732746e-05,
+      "loss": 2.0933,
+      "step": 147500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.945186001304721e-05,
+      "loss": 2.0986,
+      "step": 148000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9450008188766975e-05,
+      "loss": 2.0879,
+      "step": 148500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9448156364486725e-05,
+      "loss": 2.1448,
+      "step": 149000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9446304540206476e-05,
+      "loss": 2.0606,
+      "step": 149500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.944445271592623e-05,
+      "loss": 2.1036,
+      "step": 150000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.944260089164598e-05,
+      "loss": 2.0317,
+      "step": 150500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.944074906736574e-05,
+      "loss": 2.1222,
+      "step": 151000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.94388972430855e-05,
+      "loss": 2.0757,
+      "step": 151500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.943704541880525e-05,
+      "loss": 2.066,
+      "step": 152000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9435193594525004e-05,
+      "loss": 2.1251,
+      "step": 152500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9433341770244754e-05,
+      "loss": 2.0925,
+      "step": 153000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.943148994596451e-05,
+      "loss": 2.0619,
+      "step": 153500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.942963812168426e-05,
+      "loss": 2.0371,
+      "step": 154000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.942778629740402e-05,
+      "loss": 2.1089,
+      "step": 154500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9425934473123775e-05,
+      "loss": 2.0807,
+      "step": 155000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9424082648843525e-05,
+      "loss": 2.0697,
+      "step": 155500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.942223082456328e-05,
+      "loss": 2.0802,
+      "step": 156000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.942037900028303e-05,
+      "loss": 2.0421,
+      "step": 156500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.941852717600279e-05,
+      "loss": 2.0534,
+      "step": 157000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.941667535172254e-05,
+      "loss": 2.0713,
+      "step": 157500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.94148235274423e-05,
+      "loss": 2.0668,
+      "step": 158000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.941297170316205e-05,
+      "loss": 2.0462,
+      "step": 158500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9411119878881803e-05,
+      "loss": 2.0347,
+      "step": 159000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.940926805460156e-05,
+      "loss": 2.1034,
+      "step": 159500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.940741623032131e-05,
+      "loss": 2.0372,
+      "step": 160000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.940556440604107e-05,
+      "loss": 2.0649,
+      "step": 160500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9403712581760824e-05,
+      "loss": 2.104,
+      "step": 161000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.940186075748058e-05,
+      "loss": 2.1171,
+      "step": 161500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.940000893320033e-05,
+      "loss": 2.0524,
+      "step": 162000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.939815710892008e-05,
+      "loss": 2.0812,
+      "step": 162500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.939630528463984e-05,
+      "loss": 2.0633,
+      "step": 163000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.939445346035959e-05,
+      "loss": 2.0875,
+      "step": 163500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9392601636079346e-05,
+      "loss": 2.0515,
+      "step": 164000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.93907498117991e-05,
+      "loss": 2.0684,
+      "step": 164500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.938889798751885e-05,
+      "loss": 2.1043,
+      "step": 165000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.938704616323861e-05,
+      "loss": 2.0895,
+      "step": 165500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.938519433895836e-05,
+      "loss": 2.1472,
+      "step": 166000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.938334251467812e-05,
+      "loss": 2.1056,
+      "step": 166500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.938149069039787e-05,
+      "loss": 2.1839,
+      "step": 167000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.937963886611763e-05,
+      "loss": 2.1721,
+      "step": 167500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.937778704183738e-05,
+      "loss": 2.0804,
+      "step": 168000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.937593521755713e-05,
+      "loss": 2.1136,
+      "step": 168500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.937408339327689e-05,
+      "loss": 2.1256,
+      "step": 169000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.937223156899664e-05,
+      "loss": 2.1167,
+      "step": 169500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9370379744716395e-05,
+      "loss": 2.0798,
+      "step": 170000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9368527920436145e-05,
+      "loss": 2.0936,
+      "step": 170500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.936667609615591e-05,
+      "loss": 2.0949,
+      "step": 171000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.936482427187566e-05,
+      "loss": 2.0507,
+      "step": 171500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.936297244759541e-05,
+      "loss": 2.1179,
+      "step": 172000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9361120623315166e-05,
+      "loss": 2.122,
+      "step": 172500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9359268799034917e-05,
+      "loss": 2.0264,
+      "step": 173000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9357416974754674e-05,
+      "loss": 2.0497,
+      "step": 173500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.935556515047443e-05,
+      "loss": 2.0554,
+      "step": 174000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.935371332619419e-05,
+      "loss": 2.1613,
+      "step": 174500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.935186150191394e-05,
+      "loss": 2.0881,
+      "step": 175000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.935000967763369e-05,
+      "loss": 2.037,
+      "step": 175500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9348157853353445e-05,
+      "loss": 2.0376,
+      "step": 176000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9346306029073195e-05,
+      "loss": 2.044,
+      "step": 176500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.934445420479296e-05,
+      "loss": 2.0583,
+      "step": 177000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.934260238051271e-05,
+      "loss": 2.1067,
+      "step": 177500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.934075055623246e-05,
+      "loss": 2.0444,
+      "step": 178000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9338898731952216e-05,
+      "loss": 2.0586,
+      "step": 178500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9337046907671966e-05,
+      "loss": 2.0813,
+      "step": 179000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.933519508339172e-05,
+      "loss": 2.077,
+      "step": 179500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.933334325911147e-05,
+      "loss": 2.0301,
+      "step": 180000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.933149143483124e-05,
+      "loss": 2.0865,
+      "step": 180500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.932963961055099e-05,
+      "loss": 2.1051,
+      "step": 181000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.932778778627074e-05,
+      "loss": 2.1441,
+      "step": 181500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9325935961990494e-05,
+      "loss": 2.0732,
+      "step": 182000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9324084137710244e-05,
+      "loss": 2.1316,
+      "step": 182500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.932223231343e-05,
+      "loss": 2.1063,
+      "step": 183000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.932038048914976e-05,
+      "loss": 2.1785,
+      "step": 183500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9318528664869515e-05,
+      "loss": 2.1108,
+      "step": 184000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9316676840589265e-05,
+      "loss": 2.0622,
+      "step": 184500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9314825016309016e-05,
+      "loss": 2.0539,
+      "step": 185000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.931297319202877e-05,
+      "loss": 2.0998,
+      "step": 185500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.931112136774852e-05,
+      "loss": 2.0329,
+      "step": 186000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9309269543468286e-05,
+      "loss": 2.1002,
+      "step": 186500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9307417719188037e-05,
+      "loss": 2.1119,
+      "step": 187000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9305565894907794e-05,
+      "loss": 2.1018,
+      "step": 187500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9303714070627544e-05,
+      "loss": 2.0742,
+      "step": 188000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9301862246347294e-05,
+      "loss": 2.0729,
+      "step": 188500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.930001042206705e-05,
+      "loss": 2.0995,
+      "step": 189000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.92981585977868e-05,
+      "loss": 2.1262,
+      "step": 189500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9296306773506565e-05,
+      "loss": 2.1104,
+      "step": 190000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9294454949226315e-05,
+      "loss": 2.0692,
+      "step": 190500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.929260312494607e-05,
+      "loss": 2.0884,
+      "step": 191000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.929075130066582e-05,
+      "loss": 2.047,
+      "step": 191500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.928889947638557e-05,
+      "loss": 2.0389,
+      "step": 192000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.928704765210533e-05,
+      "loss": 2.091,
+      "step": 192500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9285195827825086e-05,
+      "loss": 2.0903,
+      "step": 193000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.928334400354484e-05,
+      "loss": 2.0842,
+      "step": 193500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.928149217926459e-05,
+      "loss": 2.11,
+      "step": 194000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.927964035498434e-05,
+      "loss": 2.1346,
+      "step": 194500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.92777885307041e-05,
+      "loss": 2.0544,
+      "step": 195000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.927593670642385e-05,
+      "loss": 2.1018,
+      "step": 195500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.927408488214361e-05,
+      "loss": 2.1567,
+      "step": 196000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9272233057863364e-05,
+      "loss": 2.038,
+      "step": 196500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.927038123358312e-05,
+      "loss": 2.1098,
+      "step": 197000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.926852940930287e-05,
+      "loss": 2.0882,
+      "step": 197500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.926667758502262e-05,
+      "loss": 2.1377,
+      "step": 198000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.926482576074238e-05,
+      "loss": 2.0804,
+      "step": 198500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.926297393646213e-05,
+      "loss": 2.0966,
+      "step": 199000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.926112211218189e-05,
+      "loss": 2.0618,
+      "step": 199500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.925927028790164e-05,
+      "loss": 2.0894,
+      "step": 200000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.92574184636214e-05,
+      "loss": 2.1082,
+      "step": 200500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.925556663934115e-05,
+      "loss": 2.0714,
+      "step": 201000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.92537148150609e-05,
+      "loss": 2.0899,
+      "step": 201500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.925186299078066e-05,
+      "loss": 2.1127,
+      "step": 202000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9250011166500414e-05,
+      "loss": 2.0379,
+      "step": 202500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.924815934222017e-05,
+      "loss": 2.1386,
+      "step": 203000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.924630751793992e-05,
+      "loss": 2.1004,
+      "step": 203500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.924445569365968e-05,
+      "loss": 2.1193,
+      "step": 204000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.924260386937943e-05,
+      "loss": 2.0352,
+      "step": 204500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.924075204509918e-05,
+      "loss": 1.9945,
+      "step": 205000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9238900220818935e-05,
+      "loss": 2.048,
+      "step": 205500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.923704839653869e-05,
+      "loss": 2.1062,
+      "step": 206000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.923519657225845e-05,
+      "loss": 2.1527,
+      "step": 206500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.92333447479782e-05,
+      "loss": 2.0849,
+      "step": 207000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.923149292369795e-05,
+      "loss": 2.1391,
+      "step": 207500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9229641099417706e-05,
+      "loss": 2.1056,
+      "step": 208000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9227789275137457e-05,
+      "loss": 2.0838,
+      "step": 208500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.922593745085722e-05,
+      "loss": 2.0591,
+      "step": 209000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.922408562657697e-05,
+      "loss": 2.0663,
+      "step": 209500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.922223380229673e-05,
+      "loss": 2.0937,
+      "step": 210000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.922038197801648e-05,
+      "loss": 2.069,
+      "step": 210500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.921853015373623e-05,
+      "loss": 2.1053,
+      "step": 211000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9216678329455985e-05,
+      "loss": 2.1391,
+      "step": 211500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9214826505175735e-05,
+      "loss": 2.1385,
+      "step": 212000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.92129746808955e-05,
+      "loss": 2.0814,
+      "step": 212500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.921112285661525e-05,
+      "loss": 2.0214,
+      "step": 213000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9209271032335006e-05,
+      "loss": 2.0682,
+      "step": 213500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9207419208054756e-05,
+      "loss": 2.0033,
+      "step": 214000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9205567383774506e-05,
+      "loss": 2.1075,
+      "step": 214500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.920371555949426e-05,
+      "loss": 2.1303,
+      "step": 215000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.920186373521402e-05,
+      "loss": 2.0722,
+      "step": 215500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.920001191093378e-05,
+      "loss": 2.0589,
+      "step": 216000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.919816008665353e-05,
+      "loss": 2.1009,
+      "step": 216500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9196308262373284e-05,
+      "loss": 2.0362,
+      "step": 217000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9194456438093034e-05,
+      "loss": 2.0887,
+      "step": 217500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9192604613812784e-05,
+      "loss": 2.0315,
+      "step": 218000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.919075278953255e-05,
+      "loss": 2.0048,
+      "step": 218500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.91889009652523e-05,
+      "loss": 2.0886,
+      "step": 219000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9187049140972055e-05,
+      "loss": 2.1068,
+      "step": 219500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9185197316691805e-05,
+      "loss": 2.0436,
+      "step": 220000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9183345492411555e-05,
+      "loss": 2.1176,
+      "step": 220500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.918149366813131e-05,
+      "loss": 2.0312,
+      "step": 221000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.917964184385106e-05,
+      "loss": 2.0521,
+      "step": 221500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9177790019570826e-05,
+      "loss": 2.1194,
+      "step": 222000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9175938195290577e-05,
+      "loss": 2.1554,
+      "step": 222500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9174086371010333e-05,
+      "loss": 2.0491,
+      "step": 223000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9172234546730084e-05,
+      "loss": 2.1222,
+      "step": 223500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9170382722449834e-05,
+      "loss": 2.1292,
+      "step": 224000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.916853089816959e-05,
+      "loss": 2.1109,
+      "step": 224500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.916667907388935e-05,
+      "loss": 2.1183,
+      "step": 225000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9164827249609105e-05,
+      "loss": 2.0613,
+      "step": 225500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9162975425328855e-05,
+      "loss": 2.1023,
+      "step": 226000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.916112360104861e-05,
+      "loss": 2.1123,
+      "step": 226500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.915927177676836e-05,
+      "loss": 2.0664,
+      "step": 227000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.915741995248811e-05,
+      "loss": 2.0816,
+      "step": 227500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9155568128207876e-05,
+      "loss": 2.0749,
+      "step": 228000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9153716303927626e-05,
+      "loss": 2.0996,
+      "step": 228500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.915186447964738e-05,
+      "loss": 2.0893,
+      "step": 229000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.915001265536713e-05,
+      "loss": 2.1056,
+      "step": 229500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.914816083108689e-05,
+      "loss": 2.0729,
+      "step": 230000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.914630900680664e-05,
+      "loss": 2.0735,
+      "step": 230500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.914445718252639e-05,
+      "loss": 2.1005,
+      "step": 231000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9142605358246154e-05,
+      "loss": 2.144,
+      "step": 231500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9140753533965904e-05,
+      "loss": 2.1338,
+      "step": 232000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.913890170968566e-05,
+      "loss": 2.0836,
+      "step": 232500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.913704988540541e-05,
+      "loss": 2.1221,
+      "step": 233000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.913519806112517e-05,
+      "loss": 2.0788,
+      "step": 233500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.913334623684492e-05,
+      "loss": 2.0521,
+      "step": 234000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9131494412564675e-05,
+      "loss": 2.1534,
+      "step": 234500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.912964258828443e-05,
+      "loss": 2.1214,
+      "step": 235000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.912779076400418e-05,
+      "loss": 2.1256,
+      "step": 235500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.912593893972394e-05,
+      "loss": 2.1058,
+      "step": 236000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.912408711544369e-05,
+      "loss": 2.023,
+      "step": 236500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.912223529116344e-05,
+      "loss": 2.092,
+      "step": 237000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.91203834668832e-05,
+      "loss": 2.0712,
+      "step": 237500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9118531642602954e-05,
+      "loss": 2.0955,
+      "step": 238000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.911667981832271e-05,
+      "loss": 2.088,
+      "step": 238500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.911482799404246e-05,
+      "loss": 2.0495,
+      "step": 239000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.911297616976222e-05,
+      "loss": 2.0736,
+      "step": 239500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.911112434548197e-05,
+      "loss": 2.0218,
+      "step": 240000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.910927252120172e-05,
+      "loss": 2.1898,
+      "step": 240500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.910742069692148e-05,
+      "loss": 2.1209,
+      "step": 241000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.910556887264123e-05,
+      "loss": 2.1103,
+      "step": 241500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.910371704836099e-05,
+      "loss": 2.1283,
+      "step": 242000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.910186522408074e-05,
+      "loss": 2.1047,
+      "step": 242500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9100013399800496e-05,
+      "loss": 2.0582,
+      "step": 243000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9098161575520246e-05,
+      "loss": 2.051,
+      "step": 243500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.909630975124e-05,
+      "loss": 2.1267,
+      "step": 244000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.909445792695976e-05,
+      "loss": 2.0628,
+      "step": 244500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.909260610267951e-05,
+      "loss": 2.0915,
+      "step": 245000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.909075427839927e-05,
+      "loss": 2.0771,
+      "step": 245500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.908890245411902e-05,
+      "loss": 2.0893,
+      "step": 246000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9087050629838774e-05,
+      "loss": 2.0784,
+      "step": 246500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9085198805558525e-05,
+      "loss": 2.0892,
+      "step": 247000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.908334698127828e-05,
+      "loss": 2.0572,
+      "step": 247500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.908149515699804e-05,
+      "loss": 2.0948,
+      "step": 248000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.907964333271779e-05,
+      "loss": 2.1058,
+      "step": 248500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9077791508437546e-05,
+      "loss": 2.0597,
+      "step": 249000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9075939684157296e-05,
+      "loss": 2.0909,
+      "step": 249500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9074087859877046e-05,
+      "loss": 2.0595,
+      "step": 250000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.907223603559681e-05,
+      "loss": 2.0492,
+      "step": 250500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.907038421131656e-05,
+      "loss": 2.0546,
+      "step": 251000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.906853238703632e-05,
+      "loss": 2.1159,
+      "step": 251500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.906668056275607e-05,
+      "loss": 2.1385,
+      "step": 252000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9064828738475824e-05,
+      "loss": 2.0528,
+      "step": 252500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9062976914195574e-05,
+      "loss": 2.0894,
+      "step": 253000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.906112508991533e-05,
+      "loss": 2.0223,
+      "step": 253500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.905927326563509e-05,
+      "loss": 2.0927,
+      "step": 254000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.905742144135484e-05,
+      "loss": 2.0777,
+      "step": 254500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9055569617074595e-05,
+      "loss": 2.0878,
+      "step": 255000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9053717792794345e-05,
+      "loss": 2.1174,
+      "step": 255500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.90518659685141e-05,
+      "loss": 2.0705,
+      "step": 256000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.905001414423385e-05,
+      "loss": 2.0456,
+      "step": 256500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.904816231995361e-05,
+      "loss": 2.0628,
+      "step": 257000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9046310495673366e-05,
+      "loss": 2.1184,
+      "step": 257500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9044458671393116e-05,
+      "loss": 2.1294,
+      "step": 258000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.904260684711287e-05,
+      "loss": 2.078,
+      "step": 258500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9040755022832624e-05,
+      "loss": 2.0868,
+      "step": 259000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.903890319855238e-05,
+      "loss": 2.0613,
+      "step": 259500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.903705137427214e-05,
+      "loss": 2.0518,
+      "step": 260000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.903519954999189e-05,
+      "loss": 2.0432,
+      "step": 260500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9033347725711645e-05,
+      "loss": 2.0214,
+      "step": 261000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9031495901431395e-05,
+      "loss": 2.0357,
+      "step": 261500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.902964407715115e-05,
+      "loss": 2.0896,
+      "step": 262000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.90277922528709e-05,
+      "loss": 2.0427,
+      "step": 262500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.902594042859065e-05,
+      "loss": 2.0874,
+      "step": 263000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9024088604310416e-05,
+      "loss": 2.0956,
+      "step": 263500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9022236780030166e-05,
+      "loss": 2.089,
+      "step": 264000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.902038495574992e-05,
+      "loss": 2.1201,
+      "step": 264500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.901853313146967e-05,
+      "loss": 2.0422,
+      "step": 265000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.901668130718943e-05,
+      "loss": 2.0893,
+      "step": 265500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.901482948290918e-05,
+      "loss": 2.0884,
+      "step": 266000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.901297765862894e-05,
+      "loss": 2.0242,
+      "step": 266500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9011125834348694e-05,
+      "loss": 2.0888,
+      "step": 267000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9009274010068444e-05,
+      "loss": 2.108,
+      "step": 267500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.90074221857882e-05,
+      "loss": 2.0993,
+      "step": 268000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.900557036150795e-05,
+      "loss": 2.0372,
+      "step": 268500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.900371853722771e-05,
+      "loss": 2.0594,
+      "step": 269000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9001866712947465e-05,
+      "loss": 1.9995,
+      "step": 269500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9000014888667215e-05,
+      "loss": 2.1141,
+      "step": 270000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.899816306438697e-05,
+      "loss": 2.0906,
+      "step": 270500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.899631124010672e-05,
+      "loss": 2.0663,
+      "step": 271000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.899445941582648e-05,
+      "loss": 2.0266,
+      "step": 271500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.899260759154623e-05,
+      "loss": 2.0186,
+      "step": 272000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8990755767265987e-05,
+      "loss": 2.0313,
+      "step": 272500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8988903942985744e-05,
+      "loss": 2.0928,
+      "step": 273000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8987052118705494e-05,
+      "loss": 2.0998,
+      "step": 273500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.898520029442525e-05,
+      "loss": 2.0823,
+      "step": 274000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8983348470145e-05,
+      "loss": 2.1264,
+      "step": 274500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.898149664586476e-05,
+      "loss": 2.1021,
+      "step": 275000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.897964482158451e-05,
+      "loss": 2.1273,
+      "step": 275500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8977792997304265e-05,
+      "loss": 2.0928,
+      "step": 276000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.897594117302402e-05,
+      "loss": 2.0685,
+      "step": 276500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.897408934874377e-05,
+      "loss": 2.0811,
+      "step": 277000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.897223752446353e-05,
+      "loss": 2.081,
+      "step": 277500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.897038570018328e-05,
+      "loss": 2.0601,
+      "step": 278000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8968533875903036e-05,
+      "loss": 2.1053,
+      "step": 278500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.896668205162279e-05,
+      "loss": 2.1081,
+      "step": 279000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.896483022734254e-05,
+      "loss": 2.098,
+      "step": 279500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.89629784030623e-05,
+      "loss": 2.1769,
+      "step": 280000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.896112657878205e-05,
+      "loss": 2.083,
+      "step": 280500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.895927475450181e-05,
+      "loss": 2.0933,
+      "step": 281000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.895742293022156e-05,
+      "loss": 2.1113,
+      "step": 281500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8955571105941314e-05,
+      "loss": 2.1,
+      "step": 282000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.895371928166107e-05,
+      "loss": 2.062,
+      "step": 282500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.895186745738082e-05,
+      "loss": 2.1055,
+      "step": 283000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.895001563310058e-05,
+      "loss": 2.018,
+      "step": 283500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.894816380882033e-05,
+      "loss": 2.0741,
+      "step": 284000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8946311984540086e-05,
+      "loss": 2.1163,
+      "step": 284500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8944460160259836e-05,
+      "loss": 2.0718,
+      "step": 285000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.894260833597959e-05,
+      "loss": 2.0572,
+      "step": 285500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.894075651169935e-05,
+      "loss": 2.0786,
+      "step": 286000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.89389046874191e-05,
+      "loss": 2.1539,
+      "step": 286500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.893705286313886e-05,
+      "loss": 2.0523,
+      "step": 287000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.893520103885861e-05,
+      "loss": 2.0635,
+      "step": 287500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8933349214578364e-05,
+      "loss": 2.0962,
+      "step": 288000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8931497390298114e-05,
+      "loss": 2.1501,
+      "step": 288500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.892964556601787e-05,
+      "loss": 2.0864,
+      "step": 289000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.892779374173763e-05,
+      "loss": 2.0495,
+      "step": 289500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.892594191745738e-05,
+      "loss": 2.0418,
+      "step": 290000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8924090093177135e-05,
+      "loss": 2.1161,
+      "step": 290500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8922238268896885e-05,
+      "loss": 2.0743,
+      "step": 291000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.892038644461664e-05,
+      "loss": 2.1056,
+      "step": 291500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.89185346203364e-05,
+      "loss": 2.0449,
+      "step": 292000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.891668279605615e-05,
+      "loss": 2.1395,
+      "step": 292500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8914830971775906e-05,
+      "loss": 2.1422,
+      "step": 293000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8912979147495656e-05,
+      "loss": 2.0949,
+      "step": 293500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.891112732321541e-05,
+      "loss": 2.0853,
+      "step": 294000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8909275498935163e-05,
+      "loss": 2.0809,
+      "step": 294500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.890742367465492e-05,
+      "loss": 2.0908,
+      "step": 295000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.890557185037468e-05,
+      "loss": 1.9894,
+      "step": 295500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.890372002609443e-05,
+      "loss": 2.1431,
+      "step": 296000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8901868201814184e-05,
+      "loss": 2.0502,
+      "step": 296500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8900016377533935e-05,
+      "loss": 2.0828,
+      "step": 297000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.889816455325369e-05,
+      "loss": 1.9749,
+      "step": 297500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.889631272897344e-05,
+      "loss": 2.091,
+      "step": 298000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.88944609046932e-05,
+      "loss": 2.0509,
+      "step": 298500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8892609080412956e-05,
+      "loss": 2.0141,
+      "step": 299000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8890757256132706e-05,
+      "loss": 2.1212,
+      "step": 299500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.888890543185246e-05,
+      "loss": 2.1005,
+      "step": 300000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.888705360757221e-05,
+      "loss": 2.1135,
+      "step": 300500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.888520178329197e-05,
+      "loss": 2.0971,
+      "step": 301000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.888334995901173e-05,
+      "loss": 2.0434,
+      "step": 301500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.888149813473148e-05,
+      "loss": 2.0563,
+      "step": 302000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8879646310451234e-05,
+      "loss": 2.0624,
+      "step": 302500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8877794486170984e-05,
+      "loss": 2.0478,
+      "step": 303000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.887594266189074e-05,
+      "loss": 2.058,
+      "step": 303500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.887409083761049e-05,
+      "loss": 2.0822,
+      "step": 304000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.887223901333025e-05,
+      "loss": 2.0673,
+      "step": 304500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8870387189050005e-05,
+      "loss": 2.0657,
+      "step": 305000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8868535364769755e-05,
+      "loss": 2.0843,
+      "step": 305500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.886668354048951e-05,
+      "loss": 2.068,
+      "step": 306000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.886483171620926e-05,
+      "loss": 2.0716,
+      "step": 306500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.886297989192902e-05,
+      "loss": 2.0293,
+      "step": 307000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.886112806764877e-05,
+      "loss": 2.0616,
+      "step": 307500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8859276243368526e-05,
+      "loss": 2.0941,
+      "step": 308000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8857424419088283e-05,
+      "loss": 1.9884,
+      "step": 308500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8855572594808034e-05,
+      "loss": 2.0897,
+      "step": 309000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.885372077052779e-05,
+      "loss": 2.0541,
+      "step": 309500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.885186894624754e-05,
+      "loss": 1.9981,
+      "step": 310000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.88500171219673e-05,
+      "loss": 2.0618,
+      "step": 310500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8848165297687055e-05,
+      "loss": 2.0483,
+      "step": 311000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8846313473406805e-05,
+      "loss": 2.1106,
+      "step": 311500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.884446164912656e-05,
+      "loss": 2.0483,
+      "step": 312000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.884260982484631e-05,
+      "loss": 2.0277,
+      "step": 312500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.884075800056607e-05,
+      "loss": 1.9978,
+      "step": 313000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.883890617628582e-05,
+      "loss": 2.0311,
+      "step": 313500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8837054352005576e-05,
+      "loss": 2.059,
+      "step": 314000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.883520252772533e-05,
+      "loss": 2.0999,
+      "step": 314500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.883335070344508e-05,
+      "loss": 2.0588,
+      "step": 315000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.883149887916484e-05,
+      "loss": 2.0898,
+      "step": 315500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.882964705488459e-05,
+      "loss": 2.1084,
+      "step": 316000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.882779523060435e-05,
+      "loss": 2.1465,
+      "step": 316500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.88259434063241e-05,
+      "loss": 2.105,
+      "step": 317000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8824091582043854e-05,
+      "loss": 2.0882,
+      "step": 317500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.882223975776361e-05,
+      "loss": 2.0737,
+      "step": 318000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.882038793348336e-05,
+      "loss": 2.1274,
+      "step": 318500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.881853610920312e-05,
+      "loss": 2.0419,
+      "step": 319000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.881668428492287e-05,
+      "loss": 2.0014,
+      "step": 319500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8814832460642625e-05,
+      "loss": 2.0236,
+      "step": 320000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.881298063636238e-05,
+      "loss": 2.1488,
+      "step": 320500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.881112881208213e-05,
+      "loss": 2.0795,
+      "step": 321000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.880927698780189e-05,
+      "loss": 2.0669,
+      "step": 321500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.880742516352164e-05,
+      "loss": 2.1564,
+      "step": 322000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.88055733392414e-05,
+      "loss": 2.0505,
+      "step": 322500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.880372151496115e-05,
+      "loss": 2.0582,
+      "step": 323000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8801869690680904e-05,
+      "loss": 2.1198,
+      "step": 323500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.880001786640066e-05,
+      "loss": 2.0779,
+      "step": 324000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.879816604212041e-05,
+      "loss": 2.098,
+      "step": 324500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.879631421784017e-05,
+      "loss": 2.068,
+      "step": 325000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.879446239355992e-05,
+      "loss": 2.0426,
+      "step": 325500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8792610569279675e-05,
+      "loss": 2.0374,
+      "step": 326000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8790758744999425e-05,
+      "loss": 2.0615,
+      "step": 326500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.878890692071918e-05,
+      "loss": 2.1002,
+      "step": 327000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.878705509643894e-05,
+      "loss": 2.1018,
+      "step": 327500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.878520327215869e-05,
+      "loss": 2.0349,
+      "step": 328000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8783351447878446e-05,
+      "loss": 2.0651,
+      "step": 328500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8781499623598196e-05,
+      "loss": 1.9743,
+      "step": 329000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.877964779931795e-05,
+      "loss": 2.0576,
+      "step": 329500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.877779597503771e-05,
+      "loss": 2.0568,
+      "step": 330000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.877594415075746e-05,
+      "loss": 2.0538,
+      "step": 330500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.877409232647722e-05,
+      "loss": 2.0727,
+      "step": 331000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.877224050219697e-05,
+      "loss": 2.1174,
+      "step": 331500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8770388677916724e-05,
+      "loss": 2.1466,
+      "step": 332000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8768536853636475e-05,
+      "loss": 2.1125,
+      "step": 332500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.876668502935623e-05,
+      "loss": 2.0779,
+      "step": 333000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.876483320507599e-05,
+      "loss": 2.065,
+      "step": 333500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.876298138079574e-05,
+      "loss": 2.0613,
+      "step": 334000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8761129556515496e-05,
+      "loss": 2.1109,
+      "step": 334500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8759277732235246e-05,
+      "loss": 2.0144,
+      "step": 335000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8757425907955e-05,
+      "loss": 2.1222,
+      "step": 335500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.875557408367475e-05,
+      "loss": 2.0914,
+      "step": 336000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.875372225939451e-05,
+      "loss": 2.0208,
+      "step": 336500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.875187043511427e-05,
+      "loss": 2.0179,
+      "step": 337000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.875001861083402e-05,
+      "loss": 2.0414,
+      "step": 337500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8748166786553774e-05,
+      "loss": 2.1039,
+      "step": 338000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8746314962273524e-05,
+      "loss": 2.0596,
+      "step": 338500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.874446313799328e-05,
+      "loss": 2.0743,
+      "step": 339000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.874261131371303e-05,
+      "loss": 2.0327,
+      "step": 339500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.874075948943279e-05,
+      "loss": 2.0576,
+      "step": 340000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8738907665152545e-05,
+      "loss": 2.0559,
+      "step": 340500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8737055840872295e-05,
+      "loss": 2.042,
+      "step": 341000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.873520401659205e-05,
+      "loss": 2.0752,
+      "step": 341500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.87333521923118e-05,
+      "loss": 2.0516,
+      "step": 342000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.873150036803156e-05,
+      "loss": 2.1037,
+      "step": 342500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8729648543751316e-05,
+      "loss": 2.1216,
+      "step": 343000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8727796719471066e-05,
+      "loss": 2.0398,
+      "step": 343500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.872594489519082e-05,
+      "loss": 2.0446,
+      "step": 344000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8724093070910574e-05,
+      "loss": 2.0327,
+      "step": 344500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.872224124663033e-05,
+      "loss": 2.0941,
+      "step": 345000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.872038942235008e-05,
+      "loss": 1.995,
+      "step": 345500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8718537598069844e-05,
+      "loss": 2.0341,
+      "step": 346000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8716685773789595e-05,
+      "loss": 1.9983,
+      "step": 346500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8714833949509345e-05,
+      "loss": 2.0519,
+      "step": 347000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.87129821252291e-05,
+      "loss": 2.0699,
+      "step": 347500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.871113030094885e-05,
+      "loss": 2.0541,
+      "step": 348000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.870927847666861e-05,
+      "loss": 2.1049,
+      "step": 348500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.870742665238836e-05,
+      "loss": 2.0521,
+      "step": 349000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8705574828108116e-05,
+      "loss": 2.0555,
+      "step": 349500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.870372300382787e-05,
+      "loss": 2.0386,
+      "step": 350000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.870187117954762e-05,
+      "loss": 2.0703,
+      "step": 350500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.870001935526738e-05,
+      "loss": 2.1147,
+      "step": 351000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.869816753098713e-05,
+      "loss": 2.025,
+      "step": 351500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.869631570670689e-05,
+      "loss": 2.1052,
+      "step": 352000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8694463882426644e-05,
+      "loss": 2.1232,
+      "step": 352500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8692612058146394e-05,
+      "loss": 2.0589,
+      "step": 353000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.869076023386615e-05,
+      "loss": 2.0957,
+      "step": 353500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.86889084095859e-05,
+      "loss": 2.1189,
+      "step": 354000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.868705658530566e-05,
+      "loss": 2.0901,
+      "step": 354500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.868520476102541e-05,
+      "loss": 2.0272,
+      "step": 355000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.868335293674517e-05,
+      "loss": 2.0451,
+      "step": 355500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.868150111246492e-05,
+      "loss": 2.0311,
+      "step": 356000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.867964928818467e-05,
+      "loss": 2.0617,
+      "step": 356500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.867779746390443e-05,
+      "loss": 2.0693,
+      "step": 357000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.867594563962418e-05,
+      "loss": 2.1253,
+      "step": 357500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8674093815343937e-05,
+      "loss": 2.0701,
+      "step": 358000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.867224199106369e-05,
+      "loss": 2.0521,
+      "step": 358500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.867039016678345e-05,
+      "loss": 2.0298,
+      "step": 359000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.86685383425032e-05,
+      "loss": 2.0321,
+      "step": 359500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.866668651822295e-05,
+      "loss": 2.0373,
+      "step": 360000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.866483469394271e-05,
+      "loss": 2.0453,
+      "step": 360500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.866298286966246e-05,
+      "loss": 2.0423,
+      "step": 361000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8661131045382215e-05,
+      "loss": 2.0792,
+      "step": 361500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.865927922110197e-05,
+      "loss": 2.1337,
+      "step": 362000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.865742739682172e-05,
+      "loss": 1.9964,
+      "step": 362500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.865557557254148e-05,
+      "loss": 2.0639,
+      "step": 363000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.865372374826123e-05,
+      "loss": 2.0159,
+      "step": 363500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8651871923980986e-05,
+      "loss": 2.0956,
+      "step": 364000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8650020099700736e-05,
+      "loss": 2.0309,
+      "step": 364500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.864816827542049e-05,
+      "loss": 2.0339,
+      "step": 365000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.864631645114025e-05,
+      "loss": 2.0868,
+      "step": 365500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.864446462686e-05,
+      "loss": 1.9837,
+      "step": 366000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.864261280257976e-05,
+      "loss": 2.0658,
+      "step": 366500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.864076097829951e-05,
+      "loss": 1.9948,
+      "step": 367000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8638909154019264e-05,
+      "loss": 2.0789,
+      "step": 367500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8637057329739015e-05,
+      "loss": 2.0611,
+      "step": 368000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.863520550545878e-05,
+      "loss": 2.0874,
+      "step": 368500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.863335368117853e-05,
+      "loss": 2.0583,
+      "step": 369000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.863150185689828e-05,
+      "loss": 2.0309,
+      "step": 369500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8629650032618036e-05,
+      "loss": 2.0677,
+      "step": 370000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8627798208337786e-05,
+      "loss": 1.9741,
+      "step": 370500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.862594638405754e-05,
+      "loss": 2.0508,
+      "step": 371000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.86240945597773e-05,
+      "loss": 2.0375,
+      "step": 371500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8622242735497057e-05,
+      "loss": 2.0502,
+      "step": 372000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.862039091121681e-05,
+      "loss": 2.0519,
+      "step": 372500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.861853908693656e-05,
+      "loss": 1.9897,
+      "step": 373000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8616687262656314e-05,
+      "loss": 2.0186,
+      "step": 373500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8614835438376064e-05,
+      "loss": 2.031,
+      "step": 374000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.861298361409582e-05,
+      "loss": 2.0009,
+      "step": 374500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.861113178981558e-05,
+      "loss": 2.0465,
+      "step": 375000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8609279965535335e-05,
+      "loss": 2.0583,
+      "step": 375500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8607428141255085e-05,
+      "loss": 1.9984,
+      "step": 376000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8605576316974835e-05,
+      "loss": 2.0159,
+      "step": 376500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.860372449269459e-05,
+      "loss": 2.0457,
+      "step": 377000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.860187266841434e-05,
+      "loss": 2.0298,
+      "step": 377500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8600020844134106e-05,
+      "loss": 2.027,
+      "step": 378000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8598169019853856e-05,
+      "loss": 2.1008,
+      "step": 378500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8596317195573606e-05,
+      "loss": 2.0358,
+      "step": 379000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.859446537129336e-05,
+      "loss": 2.0131,
+      "step": 379500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8592613547013113e-05,
+      "loss": 2.0221,
+      "step": 380000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.859076172273287e-05,
+      "loss": 2.0612,
+      "step": 380500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.858890989845263e-05,
+      "loss": 1.9445,
+      "step": 381000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8587058074172384e-05,
+      "loss": 2.0688,
+      "step": 381500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8585206249892134e-05,
+      "loss": 2.0329,
+      "step": 382000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8583354425611885e-05,
+      "loss": 1.9824,
+      "step": 382500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.858150260133164e-05,
+      "loss": 2.0359,
+      "step": 383000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.857965077705139e-05,
+      "loss": 2.0619,
+      "step": 383500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.857779895277115e-05,
+      "loss": 1.9951,
+      "step": 384000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8575947128490906e-05,
+      "loss": 2.0312,
+      "step": 384500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.857409530421066e-05,
+      "loss": 1.8873,
+      "step": 385000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.857224347993041e-05,
+      "loss": 1.9459,
+      "step": 385500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.857039165565016e-05,
+      "loss": 2.0782,
+      "step": 386000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.856853983136992e-05,
+      "loss": 1.974,
+      "step": 386500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.856668800708967e-05,
+      "loss": 2.0018,
+      "step": 387000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8564836182809434e-05,
+      "loss": 2.0126,
+      "step": 387500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8562984358529184e-05,
+      "loss": 1.999,
+      "step": 388000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.856113253424894e-05,
+      "loss": 1.5514,
+      "step": 388500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.855928070996869e-05,
+      "loss": 1.5872,
+      "step": 389000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.855742888568844e-05,
+      "loss": 1.5182,
+      "step": 389500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.85555770614082e-05,
+      "loss": 1.4703,
+      "step": 390000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.855372523712795e-05,
+      "loss": 1.454,
+      "step": 390500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.855187341284771e-05,
+      "loss": 1.4345,
+      "step": 391000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.855002158856746e-05,
+      "loss": 1.433,
+      "step": 391500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.854816976428721e-05,
+      "loss": 1.4309,
+      "step": 392000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.854631794000697e-05,
+      "loss": 1.3785,
+      "step": 392500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.854446611572672e-05,
+      "loss": 1.4492,
+      "step": 393000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8542614291446476e-05,
+      "loss": 1.4585,
+      "step": 393500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8540762467166233e-05,
+      "loss": 1.3263,
+      "step": 394000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.853891064288599e-05,
+      "loss": 1.411,
+      "step": 394500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.853705881860574e-05,
+      "loss": 1.3492,
+      "step": 395000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.853520699432549e-05,
+      "loss": 1.4024,
+      "step": 395500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.853335517004525e-05,
+      "loss": 1.3816,
+      "step": 396000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8531503345765e-05,
+      "loss": 1.3385,
+      "step": 396500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.852965152148476e-05,
+      "loss": 1.4043,
+      "step": 397000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.852779969720451e-05,
+      "loss": 1.3078,
+      "step": 397500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.852594787292427e-05,
+      "loss": 1.3505,
+      "step": 398000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.852409604864402e-05,
+      "loss": 1.3205,
+      "step": 398500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.852224422436377e-05,
+      "loss": 1.3744,
+      "step": 399000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8520392400083526e-05,
+      "loss": 1.3697,
+      "step": 399500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8518540575803276e-05,
+      "loss": 1.3913,
+      "step": 400000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.851668875152304e-05,
+      "loss": 1.3664,
+      "step": 400500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.851483692724279e-05,
+      "loss": 1.3093,
+      "step": 401000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.851298510296255e-05,
+      "loss": 1.3701,
+      "step": 401500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.85111332786823e-05,
+      "loss": 1.2745,
+      "step": 402000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.850928145440205e-05,
+      "loss": 1.3443,
+      "step": 402500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8507429630121804e-05,
+      "loss": 1.3313,
+      "step": 403000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.850557780584156e-05,
+      "loss": 1.3441,
+      "step": 403500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.850372598156132e-05,
+      "loss": 1.3171,
+      "step": 404000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.850187415728107e-05,
+      "loss": 1.2825,
+      "step": 404500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.850002233300082e-05,
+      "loss": 1.317,
+      "step": 405000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8498170508720575e-05,
+      "loss": 1.313,
+      "step": 405500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8496318684440326e-05,
+      "loss": 1.2906,
+      "step": 406000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.849446686016009e-05,
+      "loss": 1.3497,
+      "step": 406500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.849261503587984e-05,
+      "loss": 1.352,
+      "step": 407000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8490763211599596e-05,
+      "loss": 1.3175,
+      "step": 407500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.848891138731935e-05,
+      "loss": 1.336,
+      "step": 408000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.84870595630391e-05,
+      "loss": 1.3383,
+      "step": 408500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8485207738758854e-05,
+      "loss": 1.3373,
+      "step": 409000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8483355914478604e-05,
+      "loss": 1.3319,
+      "step": 409500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.848150409019837e-05,
+      "loss": 1.386,
+      "step": 410000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.847965226591812e-05,
+      "loss": 1.3308,
+      "step": 410500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8477800441637875e-05,
+      "loss": 1.306,
+      "step": 411000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8475948617357625e-05,
+      "loss": 1.3129,
+      "step": 411500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8474096793077375e-05,
+      "loss": 1.3699,
+      "step": 412000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.847224496879713e-05,
+      "loss": 1.2743,
+      "step": 412500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.847039314451689e-05,
+      "loss": 1.3443,
+      "step": 413000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8468541320236646e-05,
+      "loss": 1.355,
+      "step": 413500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8466689495956396e-05,
+      "loss": 1.2891,
+      "step": 414000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.846483767167615e-05,
+      "loss": 1.3233,
+      "step": 414500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.84629858473959e-05,
+      "loss": 1.3748,
+      "step": 415000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.846113402311565e-05,
+      "loss": 1.3159,
+      "step": 415500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.845928219883541e-05,
+      "loss": 1.3326,
+      "step": 416000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.845743037455517e-05,
+      "loss": 1.2145,
+      "step": 416500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8455578550274924e-05,
+      "loss": 1.2815,
+      "step": 417000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8453726725994674e-05,
+      "loss": 1.4068,
+      "step": 417500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.845187490171443e-05,
+      "loss": 1.3169,
+      "step": 418000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.845002307743418e-05,
+      "loss": 1.32,
+      "step": 418500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.844817125315393e-05,
+      "loss": 1.3116,
+      "step": 419000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8446319428873695e-05,
+      "loss": 1.3248,
+      "step": 419500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8444467604593446e-05,
+      "loss": 0.9912,
+      "step": 420000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.84426157803132e-05,
+      "loss": 0.2414,
+      "step": 420500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.844076395603295e-05,
+      "loss": 1.1361,
+      "step": 421000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.84389121317527e-05,
+      "loss": 1.0437,
+      "step": 421500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.843706030747246e-05,
+      "loss": 1.0156,
+      "step": 422000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.843520848319222e-05,
+      "loss": 0.9771,
+      "step": 422500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8433356658911974e-05,
+      "loss": 0.947,
+      "step": 423000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8431504834631724e-05,
+      "loss": 0.9558,
+      "step": 423500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.842965301035148e-05,
+      "loss": 0.9735,
+      "step": 424000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.842780118607123e-05,
+      "loss": 0.9152,
+      "step": 424500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.842594936179098e-05,
+      "loss": 0.9708,
+      "step": 425000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.842409753751074e-05,
+      "loss": 0.9948,
+      "step": 425500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8422245713230495e-05,
+      "loss": 0.9122,
+      "step": 426000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.842039388895025e-05,
+      "loss": 0.9536,
+      "step": 426500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.841854206467e-05,
+      "loss": 0.8994,
+      "step": 427000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.841669024038976e-05,
+      "loss": 0.9621,
+      "step": 427500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.841483841610951e-05,
+      "loss": 0.9441,
+      "step": 428000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.841298659182926e-05,
+      "loss": 0.885,
+      "step": 428500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.841113476754902e-05,
+      "loss": 0.9768,
+      "step": 429000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.840928294326877e-05,
+      "loss": 0.8844,
+      "step": 429500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.840743111898853e-05,
+      "loss": 0.9265,
+      "step": 430000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.840557929470828e-05,
+      "loss": 0.8965,
+      "step": 430500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.840372747042804e-05,
+      "loss": 0.9333,
+      "step": 431000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.840187564614779e-05,
+      "loss": 0.9553,
+      "step": 431500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8400023821867545e-05,
+      "loss": 0.9454,
+      "step": 432000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.83981719975873e-05,
+      "loss": 0.9002,
+      "step": 432500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.839632017330705e-05,
+      "loss": 0.869,
+      "step": 433000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.839446834902681e-05,
+      "loss": 0.9202,
+      "step": 433500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.839261652474656e-05,
+      "loss": 0.881,
+      "step": 434000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.839076470046631e-05,
+      "loss": 0.906,
+      "step": 434500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8388912876186066e-05,
+      "loss": 0.9354,
+      "step": 435000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.838706105190582e-05,
+      "loss": 0.9224,
+      "step": 435500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.838520922762558e-05,
+      "loss": 0.9339,
+      "step": 436000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.838335740334533e-05,
+      "loss": 0.8621,
+      "step": 436500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.838150557906509e-05,
+      "loss": 0.9027,
+      "step": 437000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.837965375478484e-05,
+      "loss": 0.8946,
+      "step": 437500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.837780193050459e-05,
+      "loss": 0.8645,
+      "step": 438000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.837595010622435e-05,
+      "loss": 0.9124,
+      "step": 438500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.83740982819441e-05,
+      "loss": 0.9027,
+      "step": 439000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.837224645766386e-05,
+      "loss": 0.9064,
+      "step": 439500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.837039463338361e-05,
+      "loss": 0.9091,
+      "step": 440000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8368542809103365e-05,
+      "loss": 0.9159,
+      "step": 440500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8366690984823115e-05,
+      "loss": 0.907,
+      "step": 441000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8364839160542866e-05,
+      "loss": 0.8951,
+      "step": 441500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.836298733626263e-05,
+      "loss": 0.9397,
+      "step": 442000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.836113551198238e-05,
+      "loss": 0.8823,
+      "step": 442500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8359283687702136e-05,
+      "loss": 0.8911,
+      "step": 443000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8357431863421887e-05,
+      "loss": 0.8936,
+      "step": 443500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8355580039141643e-05,
+      "loss": 0.9403,
+      "step": 444000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8353728214861394e-05,
+      "loss": 0.8754,
+      "step": 444500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.835187639058115e-05,
+      "loss": 0.9071,
+      "step": 445000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.835002456630091e-05,
+      "loss": 0.9121,
+      "step": 445500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.834817274202066e-05,
+      "loss": 0.8832,
+      "step": 446000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8346320917740415e-05,
+      "loss": 0.8992,
+      "step": 446500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8344469093460165e-05,
+      "loss": 0.8978,
+      "step": 447000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.834261726917992e-05,
+      "loss": 0.8992,
+      "step": 447500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.834076544489968e-05,
+      "loss": 0.8995,
+      "step": 448000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.833891362061943e-05,
+      "loss": 0.8194,
+      "step": 448500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8337061796339186e-05,
+      "loss": 0.8614,
+      "step": 449000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8335209972058936e-05,
+      "loss": 0.9629,
+      "step": 449500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.833335814777869e-05,
+      "loss": 0.8858,
+      "step": 450000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.833150632349844e-05,
+      "loss": 0.8869,
+      "step": 450500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.832965449921819e-05,
+      "loss": 0.9027,
+      "step": 451000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.832780267493796e-05,
+      "loss": 0.9113,
+      "step": 451500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.832595085065771e-05,
+      "loss": 2.3254,
+      "step": 452000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8324099026377464e-05,
+      "loss": 2.1624,
+      "step": 452500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8322247202097214e-05,
+      "loss": 2.1429,
+      "step": 453000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.832039537781697e-05,
+      "loss": 2.1301,
+      "step": 453500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.831854355353672e-05,
+      "loss": 2.2401,
+      "step": 454000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.831669172925648e-05,
+      "loss": 2.1386,
+      "step": 454500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8314839904976235e-05,
+      "loss": 2.178,
+      "step": 455000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8312988080695985e-05,
+      "loss": 2.1548,
+      "step": 455500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.831113625641574e-05,
+      "loss": 2.1627,
+      "step": 456000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.830928443213549e-05,
+      "loss": 2.2119,
+      "step": 456500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.830743260785525e-05,
+      "loss": 2.212,
+      "step": 457000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8305580783575007e-05,
+      "loss": 2.1224,
+      "step": 457500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.830372895929476e-05,
+      "loss": 2.2152,
+      "step": 458000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8301877135014514e-05,
+      "loss": 2.1278,
+      "step": 458500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8300025310734264e-05,
+      "loss": 2.1791,
+      "step": 459000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.829817348645402e-05,
+      "loss": 2.137,
+      "step": 459500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.829632166217377e-05,
+      "loss": 2.177,
+      "step": 460000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.829446983789353e-05,
+      "loss": 2.1272,
+      "step": 460500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8292618013613285e-05,
+      "loss": 2.1177,
+      "step": 461000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8290766189333035e-05,
+      "loss": 2.106,
+      "step": 461500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.828891436505279e-05,
+      "loss": 2.2041,
+      "step": 462000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.828706254077254e-05,
+      "loss": 2.1161,
+      "step": 462500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.82852107164923e-05,
+      "loss": 2.1244,
+      "step": 463000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.828335889221205e-05,
+      "loss": 2.0704,
+      "step": 463500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8281507067931806e-05,
+      "loss": 2.1964,
+      "step": 464000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.827965524365156e-05,
+      "loss": 2.1661,
+      "step": 464500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.827780341937131e-05,
+      "loss": 2.1416,
+      "step": 465000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.827595159509107e-05,
+      "loss": 2.1467,
+      "step": 465500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.827409977081082e-05,
+      "loss": 2.1394,
+      "step": 466000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.827224794653058e-05,
+      "loss": 2.079,
+      "step": 466500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.827039612225033e-05,
+      "loss": 2.0361,
+      "step": 467000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8268544297970084e-05,
+      "loss": 2.15,
+      "step": 467500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.826669247368984e-05,
+      "loss": 2.1647,
+      "step": 468000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.826484064940959e-05,
+      "loss": 2.1629,
+      "step": 468500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.826298882512935e-05,
+      "loss": 2.1334,
+      "step": 469000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.82611370008491e-05,
+      "loss": 2.1717,
+      "step": 469500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8259285176568856e-05,
+      "loss": 2.1106,
+      "step": 470000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.825743335228861e-05,
+      "loss": 2.0979,
+      "step": 470500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.825558152800836e-05,
+      "loss": 2.0478,
+      "step": 471000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.825372970372812e-05,
+      "loss": 2.1778,
+      "step": 471500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.825187787944787e-05,
+      "loss": 2.1486,
+      "step": 472000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.825002605516763e-05,
+      "loss": 2.0521,
+      "step": 472500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.824817423088738e-05,
+      "loss": 2.0744,
+      "step": 473000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8246322406607134e-05,
+      "loss": 2.027,
+      "step": 473500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.824447058232689e-05,
+      "loss": 2.1347,
+      "step": 474000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.824261875804664e-05,
+      "loss": 2.0729,
+      "step": 474500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.82407669337664e-05,
+      "loss": 2.0448,
+      "step": 475000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.823891510948615e-05,
+      "loss": 2.1762,
+      "step": 475500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8237063285205905e-05,
+      "loss": 2.0928,
+      "step": 476000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8235211460925655e-05,
+      "loss": 2.1397,
+      "step": 476500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.823335963664541e-05,
+      "loss": 2.1395,
+      "step": 477000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.823150781236517e-05,
+      "loss": 2.0817,
+      "step": 477500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.822965598808492e-05,
+      "loss": 2.1291,
+      "step": 478000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8227804163804676e-05,
+      "loss": 2.1392,
+      "step": 478500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8225952339524426e-05,
+      "loss": 2.1123,
+      "step": 479000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8224100515244183e-05,
+      "loss": 2.1467,
+      "step": 479500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.822224869096394e-05,
+      "loss": 2.0869,
+      "step": 480000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.822039686668369e-05,
+      "loss": 2.1907,
+      "step": 480500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.821854504240345e-05,
+      "loss": 2.1319,
+      "step": 481000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.82166932181232e-05,
+      "loss": 2.0684,
+      "step": 481500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8214841393842955e-05,
+      "loss": 2.0638,
+      "step": 482000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8212989569562705e-05,
+      "loss": 2.17,
+      "step": 482500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.821113774528246e-05,
+      "loss": 2.0686,
+      "step": 483000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.820928592100222e-05,
+      "loss": 2.1202,
+      "step": 483500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.820743409672197e-05,
+      "loss": 2.1897,
+      "step": 484000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8205582272441726e-05,
+      "loss": 2.0982,
+      "step": 484500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8203730448161476e-05,
+      "loss": 2.1002,
+      "step": 485000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.820187862388123e-05,
+      "loss": 2.0226,
+      "step": 485500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.820002679960098e-05,
+      "loss": 2.1291,
+      "step": 486000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.819817497532074e-05,
+      "loss": 2.1365,
+      "step": 486500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.81963231510405e-05,
+      "loss": 2.0817,
+      "step": 487000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.819447132676025e-05,
+      "loss": 2.0787,
+      "step": 487500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8192619502480004e-05,
+      "loss": 2.0578,
+      "step": 488000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8190767678199754e-05,
+      "loss": 2.0518,
+      "step": 488500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.818891585391951e-05,
+      "loss": 2.0799,
+      "step": 489000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.818706402963927e-05,
+      "loss": 2.1135,
+      "step": 489500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.818521220535902e-05,
+      "loss": 2.1702,
+      "step": 490000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8183360381078775e-05,
+      "loss": 2.0636,
+      "step": 490500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8181508556798525e-05,
+      "loss": 2.0613,
+      "step": 491000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.817965673251828e-05,
+      "loss": 2.0866,
+      "step": 491500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.817780490823803e-05,
+      "loss": 2.0757,
+      "step": 492000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.817595308395779e-05,
+      "loss": 2.071,
+      "step": 492500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8174101259677546e-05,
+      "loss": 2.0989,
+      "step": 493000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8172249435397297e-05,
+      "loss": 2.0395,
+      "step": 493500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8170397611117054e-05,
+      "loss": 2.121,
+      "step": 494000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8168545786836804e-05,
+      "loss": 2.1112,
+      "step": 494500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.816669396255656e-05,
+      "loss": 2.0351,
+      "step": 495000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.816484213827631e-05,
+      "loss": 2.139,
+      "step": 495500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.816299031399607e-05,
+      "loss": 2.0029,
+      "step": 496000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8161138489715825e-05,
+      "loss": 2.05,
+      "step": 496500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8159286665435575e-05,
+      "loss": 2.0657,
+      "step": 497000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.815743484115533e-05,
+      "loss": 2.106,
+      "step": 497500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.815558301687508e-05,
+      "loss": 2.0543,
+      "step": 498000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.815373119259484e-05,
+      "loss": 2.0514,
+      "step": 498500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8151879368314596e-05,
+      "loss": 2.0788,
+      "step": 499000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8150027544034346e-05,
+      "loss": 2.0402,
+      "step": 499500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.81481757197541e-05,
+      "loss": 2.1085,
+      "step": 500000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.814632389547385e-05,
+      "loss": 2.0492,
+      "step": 500500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.814447207119361e-05,
+      "loss": 2.0417,
+      "step": 501000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.814262024691336e-05,
+      "loss": 2.1058,
+      "step": 501500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.814076842263312e-05,
+      "loss": 2.0954,
+      "step": 502000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8138916598352874e-05,
+      "loss": 2.0358,
+      "step": 502500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8137064774072624e-05,
+      "loss": 2.1121,
+      "step": 503000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.813521294979238e-05,
+      "loss": 2.064,
+      "step": 503500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.813336112551213e-05,
+      "loss": 2.0141,
+      "step": 504000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.813150930123189e-05,
+      "loss": 2.0963,
+      "step": 504500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.812965747695164e-05,
+      "loss": 2.092,
+      "step": 505000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8127805652671396e-05,
+      "loss": 2.0702,
+      "step": 505500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.812595382839115e-05,
+      "loss": 2.132,
+      "step": 506000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.81241020041109e-05,
+      "loss": 2.0766,
+      "step": 506500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.812225017983066e-05,
+      "loss": 2.1155,
+      "step": 507000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.812039835555041e-05,
+      "loss": 2.0564,
+      "step": 507500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.811854653127017e-05,
+      "loss": 2.0596,
+      "step": 508000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8116694706989924e-05,
+      "loss": 2.1147,
+      "step": 508500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8114842882709674e-05,
+      "loss": 2.1054,
+      "step": 509000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.811299105842943e-05,
+      "loss": 2.0307,
+      "step": 509500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.811113923414918e-05,
+      "loss": 2.0596,
+      "step": 510000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.810928740986894e-05,
+      "loss": 2.0605,
+      "step": 510500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.810743558558869e-05,
+      "loss": 2.056,
+      "step": 511000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8105583761308445e-05,
+      "loss": 2.0342,
+      "step": 511500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.81037319370282e-05,
+      "loss": 2.1145,
+      "step": 512000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.810188011274795e-05,
+      "loss": 2.0648,
+      "step": 512500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.810002828846771e-05,
+      "loss": 2.1499,
+      "step": 513000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.809817646418746e-05,
+      "loss": 2.1408,
+      "step": 513500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8096324639907216e-05,
+      "loss": 2.0036,
+      "step": 514000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8094472815626966e-05,
+      "loss": 2.0102,
+      "step": 514500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.809262099134672e-05,
+      "loss": 2.0586,
+      "step": 515000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.809076916706648e-05,
+      "loss": 2.094,
+      "step": 515500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.808891734278623e-05,
+      "loss": 2.0283,
+      "step": 516000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.808706551850599e-05,
+      "loss": 2.0327,
+      "step": 516500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.808521369422574e-05,
+      "loss": 2.1024,
+      "step": 517000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8083361869945495e-05,
+      "loss": 2.0714,
+      "step": 517500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8081510045665245e-05,
+      "loss": 2.0896,
+      "step": 518000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8079658221385e-05,
+      "loss": 2.0503,
+      "step": 518500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.807780639710476e-05,
+      "loss": 2.0088,
+      "step": 519000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.807595457282451e-05,
+      "loss": 1.9966,
+      "step": 519500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8074102748544266e-05,
+      "loss": 2.0258,
+      "step": 520000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8072250924264016e-05,
+      "loss": 2.1098,
+      "step": 520500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.807039909998377e-05,
+      "loss": 2.0957,
+      "step": 521000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.806854727570353e-05,
+      "loss": 2.2702,
+      "step": 521500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.806669545142328e-05,
+      "loss": 2.0737,
+      "step": 522000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.806484362714304e-05,
+      "loss": 2.0648,
+      "step": 522500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.806299180286279e-05,
+      "loss": 2.0431,
+      "step": 523000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8061139978582544e-05,
+      "loss": 2.0472,
+      "step": 523500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8059288154302294e-05,
+      "loss": 2.0653,
+      "step": 524000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.805743633002205e-05,
+      "loss": 2.0622,
+      "step": 524500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.805558450574181e-05,
+      "loss": 2.0607,
+      "step": 525000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.805373268146156e-05,
+      "loss": 2.0656,
+      "step": 525500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8051880857181315e-05,
+      "loss": 2.0377,
+      "step": 526000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8050029032901065e-05,
+      "loss": 2.0661,
+      "step": 526500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.804817720862082e-05,
+      "loss": 2.086,
+      "step": 527000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.804632538434057e-05,
+      "loss": 2.029,
+      "step": 527500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.804447356006033e-05,
+      "loss": 2.0559,
+      "step": 528000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8042621735780086e-05,
+      "loss": 2.0894,
+      "step": 528500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8040769911499837e-05,
+      "loss": 2.0668,
+      "step": 529000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8038918087219593e-05,
+      "loss": 2.0502,
+      "step": 529500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8037066262939344e-05,
+      "loss": 2.0705,
+      "step": 530000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.80352144386591e-05,
+      "loss": 2.0961,
+      "step": 530500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.803336261437886e-05,
+      "loss": 2.0796,
+      "step": 531000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.803151079009861e-05,
+      "loss": 2.0869,
+      "step": 531500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8029658965818365e-05,
+      "loss": 2.0623,
+      "step": 532000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8027807141538115e-05,
+      "loss": 2.0733,
+      "step": 532500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.802595531725787e-05,
+      "loss": 2.1174,
+      "step": 533000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.802410349297762e-05,
+      "loss": 2.0586,
+      "step": 533500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.802225166869738e-05,
+      "loss": 2.1289,
+      "step": 534000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8020399844417136e-05,
+      "loss": 2.1147,
+      "step": 534500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8018548020136886e-05,
+      "loss": 2.0457,
+      "step": 535000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.801669619585664e-05,
+      "loss": 2.0674,
+      "step": 535500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.801484437157639e-05,
+      "loss": 2.011,
+      "step": 536000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.801299254729615e-05,
+      "loss": 2.0423,
+      "step": 536500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.80111407230159e-05,
+      "loss": 2.0702,
+      "step": 537000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.800928889873566e-05,
+      "loss": 2.0714,
+      "step": 537500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8007437074455414e-05,
+      "loss": 2.0333,
+      "step": 538000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8005585250175164e-05,
+      "loss": 2.0615,
+      "step": 538500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.800373342589492e-05,
+      "loss": 2.0282,
+      "step": 539000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.800188160161467e-05,
+      "loss": 2.0782,
+      "step": 539500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.800002977733443e-05,
+      "loss": 2.0383,
+      "step": 540000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.7998177953054185e-05,
+      "loss": 2.0591,
+      "step": 540500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.7996326128773935e-05,
+      "loss": 2.0856,
+      "step": 541000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.799447430449369e-05,
+      "loss": 2.0869,
+      "step": 541500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.799262248021344e-05,
+      "loss": 2.0858,
+      "step": 542000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.79907706559332e-05,
+      "loss": 2.0886,
+      "step": 542500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.798891883165295e-05,
+      "loss": 2.0282,
+      "step": 543000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.798706700737271e-05,
+      "loss": 2.0458,
+      "step": 543500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.7985215183092464e-05,
+      "loss": 1.9995,
+      "step": 544000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.7983363358812214e-05,
+      "loss": 2.1449,
+      "step": 544500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.798151153453197e-05,
+      "loss": 2.0537,
+      "step": 545000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.797965971025172e-05,
+      "loss": 2.089,
+      "step": 545500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.797780788597148e-05,
+      "loss": 2.0663,
+      "step": 546000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.797595606169123e-05,
+      "loss": 2.0505,
+      "step": 546500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.797410423741099e-05,
+      "loss": 2.0557,
+      "step": 547000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.797225241313074e-05,
+      "loss": 2.0147,
+      "step": 547500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.797040058885049e-05,
+      "loss": 2.0047,
+      "step": 548000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.796854876457025e-05,
+      "loss": 2.0574,
+      "step": 548500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.796669694029e-05,
+      "loss": 2.0718,
+      "step": 549000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.7964845116009756e-05,
+      "loss": 2.0654,
+      "step": 549500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.796299329172951e-05,
+      "loss": 2.1151,
+      "step": 550000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.796114146744926e-05,
+      "loss": 2.0348,
+      "step": 550500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.795928964316902e-05,
+      "loss": 2.0774,
+      "step": 551000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.795743781888877e-05,
+      "loss": 2.0468,
+      "step": 551500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.795558599460853e-05,
+      "loss": 2.0486,
+      "step": 552000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.795373417032828e-05,
+      "loss": 2.0349,
+      "step": 552500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.7951882346048034e-05,
+      "loss": 2.0806,
+      "step": 553000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.795003052176779e-05,
+      "loss": 2.0708,
+      "step": 553500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.794817869748754e-05,
+      "loss": 1.9739,
+      "step": 554000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.79463268732073e-05,
+      "loss": 2.0079,
+      "step": 554500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.794447504892705e-05,
+      "loss": 2.0891,
+      "step": 555000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.7942623224646806e-05,
+      "loss": 2.0833,
+      "step": 555500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.7940771400366556e-05,
+      "loss": 2.0084,
+      "step": 556000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.793891957608632e-05,
+      "loss": 2.0073,
+      "step": 556500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.793706775180607e-05,
+      "loss": 2.0094,
+      "step": 557000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.793521592752582e-05,
+      "loss": 2.0845,
+      "step": 557500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.793336410324558e-05,
+      "loss": 2.019,
+      "step": 558000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.793151227896533e-05,
+      "loss": 2.0751,
+      "step": 558500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.7929660454685084e-05,
+      "loss": 2.0965,
+      "step": 559000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.792780863040484e-05,
+      "loss": 2.0792,
+      "step": 559500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.79259568061246e-05,
+      "loss": 2.0228,
+      "step": 560000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.792410498184435e-05,
+      "loss": 2.0399,
+      "step": 560500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.79222531575641e-05,
+      "loss": 2.003,
+      "step": 561000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.7920401333283855e-05,
+      "loss": 2.0385,
+      "step": 561500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.7918549509003605e-05,
+      "loss": 2.0809,
+      "step": 562000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.791669768472336e-05,
+      "loss": 2.0491,
+      "step": 562500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.791484586044312e-05,
+      "loss": 2.0152,
+      "step": 563000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.791299403616287e-05,
+      "loss": 1.9715,
+      "step": 563500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.7911142211882626e-05,
+      "loss": 2.0302,
+      "step": 564000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.7909290387602376e-05,
+      "loss": 2.0566,
+      "step": 564500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.7907438563322133e-05,
+      "loss": 2.0745,
+      "step": 565000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.7905586739041884e-05,
+      "loss": 2.045,
+      "step": 565500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.790373491476165e-05,
+      "loss": 2.0369,
+      "step": 566000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.79018830904814e-05,
+      "loss": 1.9952,
+      "step": 566500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.790003126620115e-05,
+      "loss": 2.1119,
+      "step": 567000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.7898179441920905e-05,
+      "loss": 2.046,
+      "step": 567500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.7896327617640655e-05,
+      "loss": 2.0293,
+      "step": 568000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.789447579336041e-05,
+      "loss": 2.0215,
+      "step": 568500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.789262396908016e-05,
+      "loss": 2.0929,
+      "step": 569000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.7890772144799926e-05,
+      "loss": 2.012,
+      "step": 569500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.7888920320519676e-05,
+      "loss": 2.0962,
+      "step": 570000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.7887068496239426e-05,
+      "loss": 1.9749,
+      "step": 570500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.788521667195918e-05,
+      "loss": 2.0525,
+      "step": 571000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.788336484767893e-05,
+      "loss": 2.0129,
+      "step": 571500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.788151302339869e-05,
+      "loss": 2.025,
+      "step": 572000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.787966119911845e-05,
+      "loss": 2.0897,
+      "step": 572500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.7877809374838204e-05,
+      "loss": 2.0451,
+      "step": 573000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.7875957550557954e-05,
+      "loss": 1.9899,
+      "step": 573500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.7874105726277704e-05,
+      "loss": 2.0091,
+      "step": 574000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.787225390199746e-05,
+      "loss": 2.0697,
+      "step": 574500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.787040207771721e-05,
+      "loss": 2.0429,
+      "step": 575000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.7868550253436975e-05,
+      "loss": 2.0134,
+      "step": 575500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.7866698429156725e-05,
+      "loss": 2.0047,
+      "step": 576000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.7864846604876475e-05,
+      "loss": 1.9744,
+      "step": 576500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.786299478059623e-05,
+      "loss": 2.0071,
+      "step": 577000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.786114295631598e-05,
+      "loss": 2.0124,
+      "step": 577500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.785929113203574e-05,
+      "loss": 2.0256,
+      "step": 578000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.785743930775549e-05,
+      "loss": 1.9891,
+      "step": 578500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.785558748347525e-05,
+      "loss": 2.004,
+      "step": 579000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7853735659195004e-05,
+      "loss": 2.0468,
+      "step": 579500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7851883834914754e-05,
+      "loss": 1.9861,
+      "step": 580000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.785003201063451e-05,
+      "loss": 2.022,
+      "step": 580500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.784818018635426e-05,
+      "loss": 2.0494,
+      "step": 581000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.784632836207402e-05,
+      "loss": 2.0752,
+      "step": 581500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7844476537793775e-05,
+      "loss": 2.0183,
+      "step": 582000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.784262471351353e-05,
+      "loss": 2.0207,
+      "step": 582500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.784077288923328e-05,
+      "loss": 2.0097,
+      "step": 583000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.783892106495303e-05,
+      "loss": 2.0166,
+      "step": 583500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.783706924067279e-05,
+      "loss": 1.9941,
+      "step": 584000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.783521741639254e-05,
+      "loss": 2.0063,
+      "step": 584500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7833365592112296e-05,
+      "loss": 2.0428,
+      "step": 585000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.783151376783205e-05,
+      "loss": 2.0325,
+      "step": 585500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.782966194355181e-05,
+      "loss": 2.0902,
+      "step": 586000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.782781011927156e-05,
+      "loss": 2.0336,
+      "step": 586500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.782595829499131e-05,
+      "loss": 2.0924,
+      "step": 587000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.782410647071107e-05,
+      "loss": 2.1075,
+      "step": 587500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.782225464643082e-05,
+      "loss": 2.033,
+      "step": 588000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.782040282215058e-05,
+      "loss": 2.0679,
+      "step": 588500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.781855099787033e-05,
+      "loss": 2.0452,
+      "step": 589000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.781669917359009e-05,
+      "loss": 2.017,
+      "step": 589500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.781484734930984e-05,
+      "loss": 2.0043,
+      "step": 590000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.781299552502959e-05,
+      "loss": 2.0344,
+      "step": 590500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7811143700749346e-05,
+      "loss": 2.0212,
+      "step": 591000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.78092918764691e-05,
+      "loss": 1.9791,
+      "step": 591500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.780744005218886e-05,
+      "loss": 2.0494,
+      "step": 592000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.780558822790861e-05,
+      "loss": 2.0608,
+      "step": 592500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.780373640362836e-05,
+      "loss": 1.9911,
+      "step": 593000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.780188457934812e-05,
+      "loss": 1.9922,
+      "step": 593500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.780003275506787e-05,
+      "loss": 2.0017,
+      "step": 594000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7798180930787624e-05,
+      "loss": 2.1381,
+      "step": 594500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.779632910650738e-05,
+      "loss": 2.0281,
+      "step": 595000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.779447728222714e-05,
+      "loss": 1.9434,
+      "step": 595500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.779262545794689e-05,
+      "loss": 1.9834,
+      "step": 596000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.779077363366664e-05,
+      "loss": 1.9982,
+      "step": 596500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7788921809386395e-05,
+      "loss": 2.035,
+      "step": 597000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7787069985106145e-05,
+      "loss": 2.0292,
+      "step": 597500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.778521816082591e-05,
+      "loss": 2.019,
+      "step": 598000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.778336633654566e-05,
+      "loss": 2.0064,
+      "step": 598500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7781514512265416e-05,
+      "loss": 2.041,
+      "step": 599000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7779662687985166e-05,
+      "loss": 2.0466,
+      "step": 599500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7777810863704916e-05,
+      "loss": 1.9773,
+      "step": 600000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.777595903942467e-05,
+      "loss": 2.0342,
+      "step": 600500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.777410721514443e-05,
+      "loss": 2.0407,
+      "step": 601000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.777225539086419e-05,
+      "loss": 2.0732,
+      "step": 601500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.777040356658394e-05,
+      "loss": 2.0047,
+      "step": 602000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7768551742303694e-05,
+      "loss": 2.0766,
+      "step": 602500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7766699918023445e-05,
+      "loss": 2.0555,
+      "step": 603000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7764848093743195e-05,
+      "loss": 2.1317,
+      "step": 603500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.776299626946295e-05,
+      "loss": 2.0506,
+      "step": 604000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.776114444518271e-05,
+      "loss": 2.0278,
+      "step": 604500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7759292620902466e-05,
+      "loss": 2.0139,
+      "step": 605000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7757440796622216e-05,
+      "loss": 2.0501,
+      "step": 605500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7755588972341966e-05,
+      "loss": 2.0084,
+      "step": 606000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.775373714806172e-05,
+      "loss": 2.0657,
+      "step": 606500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.775188532378147e-05,
+      "loss": 2.0552,
+      "step": 607000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.775003349950124e-05,
+      "loss": 2.0163,
+      "step": 607500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.774818167522099e-05,
+      "loss": 2.0418,
+      "step": 608000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7746329850940744e-05,
+      "loss": 2.0153,
+      "step": 608500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7744478026660494e-05,
+      "loss": 2.0354,
+      "step": 609000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7742626202380244e-05,
+      "loss": 2.0825,
+      "step": 609500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.77407743781e-05,
+      "loss": 2.0415,
+      "step": 610000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.773892255381975e-05,
+      "loss": 2.0278,
+      "step": 610500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7737070729539515e-05,
+      "loss": 2.0174,
+      "step": 611000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7735218905259265e-05,
+      "loss": 1.9517,
+      "step": 611500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.773336708097902e-05,
+      "loss": 1.9906,
+      "step": 612000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.773151525669877e-05,
+      "loss": 2.0513,
+      "step": 612500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.772966343241852e-05,
+      "loss": 2.0269,
+      "step": 613000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.772781160813828e-05,
+      "loss": 2.0175,
+      "step": 613500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7725959783858036e-05,
+      "loss": 2.0251,
+      "step": 614000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.772410795957779e-05,
+      "loss": 2.0621,
+      "step": 614500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7722256135297543e-05,
+      "loss": 1.9919,
+      "step": 615000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.77204043110173e-05,
+      "loss": 2.043,
+      "step": 615500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.771855248673705e-05,
+      "loss": 2.081,
+      "step": 616000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.77167006624568e-05,
+      "loss": 1.9483,
+      "step": 616500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7714848838176564e-05,
+      "loss": 2.0136,
+      "step": 617000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7712997013896315e-05,
+      "loss": 2.0351,
+      "step": 617500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.771114518961607e-05,
+      "loss": 2.0785,
+      "step": 618000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.770929336533582e-05,
+      "loss": 2.0317,
+      "step": 618500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.770744154105557e-05,
+      "loss": 2.0219,
+      "step": 619000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.770558971677533e-05,
+      "loss": 1.9949,
+      "step": 619500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.770373789249508e-05,
+      "loss": 2.0794,
+      "step": 620000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.770188606821484e-05,
+      "loss": 2.0506,
+      "step": 620500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.770003424393459e-05,
+      "loss": 2.0202,
+      "step": 621000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.769818241965435e-05,
+      "loss": 2.0131,
+      "step": 621500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.76963305953741e-05,
+      "loss": 2.0382,
+      "step": 622000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.769447877109385e-05,
+      "loss": 1.9874,
+      "step": 622500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.769262694681361e-05,
+      "loss": 2.0775,
+      "step": 623000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7690775122533364e-05,
+      "loss": 2.0465,
+      "step": 623500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.768892329825312e-05,
+      "loss": 2.0227,
+      "step": 624000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.768707147397287e-05,
+      "loss": 1.9616,
+      "step": 624500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.768521964969263e-05,
+      "loss": 1.9414,
+      "step": 625000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.768336782541238e-05,
+      "loss": 2.0177,
+      "step": 625500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.768151600113213e-05,
+      "loss": 2.1022,
+      "step": 626000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.767966417685189e-05,
+      "loss": 2.0752,
+      "step": 626500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.767781235257164e-05,
+      "loss": 2.0178,
+      "step": 627000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.76759605282914e-05,
+      "loss": 2.0748,
+      "step": 627500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.767410870401115e-05,
+      "loss": 2.0444,
+      "step": 628000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7672256879730906e-05,
+      "loss": 2.0182,
+      "step": 628500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.767040505545066e-05,
+      "loss": 1.9869,
+      "step": 629000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.766855323117041e-05,
+      "loss": 2.0076,
+      "step": 629500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.766670140689017e-05,
+      "loss": 2.0308,
+      "step": 630000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.766484958260992e-05,
+      "loss": 2.0345,
+      "step": 630500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.766299775832968e-05,
+      "loss": 2.0501,
+      "step": 631000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.766114593404943e-05,
+      "loss": 2.0931,
+      "step": 631500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7659294109769185e-05,
+      "loss": 2.0988,
+      "step": 632000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7657442285488935e-05,
+      "loss": 2.0448,
+      "step": 632500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.765559046120869e-05,
+      "loss": 1.9611,
+      "step": 633000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.765373863692845e-05,
+      "loss": 2.0259,
+      "step": 633500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.76518868126482e-05,
+      "loss": 1.9305,
+      "step": 634000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7650034988367956e-05,
+      "loss": 2.0294,
+      "step": 634500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7648183164087706e-05,
+      "loss": 2.0582,
+      "step": 635000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7646331339807456e-05,
+      "loss": 2.0443,
+      "step": 635500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.764447951552721e-05,
+      "loss": 2.0119,
+      "step": 636000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.764262769124697e-05,
+      "loss": 2.0091,
+      "step": 636500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.764077586696673e-05,
+      "loss": 1.9696,
+      "step": 637000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.763892404268648e-05,
+      "loss": 2.0121,
+      "step": 637500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7637072218406234e-05,
+      "loss": 1.9961,
+      "step": 638000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7635220394125984e-05,
+      "loss": 1.9558,
+      "step": 638500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7633368569845735e-05,
+      "loss": 2.021,
+      "step": 639000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.76315167455655e-05,
+      "loss": 2.0348,
+      "step": 639500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.762966492128525e-05,
+      "loss": 2.019,
+      "step": 640000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7627813097005005e-05,
+      "loss": 2.0637,
+      "step": 640500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7625961272724756e-05,
+      "loss": 1.9666,
+      "step": 641000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.762410944844451e-05,
+      "loss": 2.0059,
+      "step": 641500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.762225762416426e-05,
+      "loss": 2.0479,
+      "step": 642000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.762040579988402e-05,
+      "loss": 2.0932,
+      "step": 642500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.761855397560378e-05,
+      "loss": 1.9762,
+      "step": 643000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.761670215132353e-05,
+      "loss": 2.0326,
+      "step": 643500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7614850327043284e-05,
+      "loss": 2.0609,
+      "step": 644000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7612998502763034e-05,
+      "loss": 2.0542,
+      "step": 644500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.761114667848279e-05,
+      "loss": 2.0496,
+      "step": 645000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.760929485420254e-05,
+      "loss": 2.0059,
+      "step": 645500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.76074430299223e-05,
+      "loss": 2.0539,
+      "step": 646000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7605591205642055e-05,
+      "loss": 2.0313,
+      "step": 646500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7603739381361805e-05,
+      "loss": 2.0165,
+      "step": 647000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.760188755708156e-05,
+      "loss": 2.014,
+      "step": 647500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.760003573280131e-05,
+      "loss": 2.0117,
+      "step": 648000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.759818390852106e-05,
+      "loss": 2.0378,
+      "step": 648500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7596332084240826e-05,
+      "loss": 2.0072,
+      "step": 649000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7594480259960576e-05,
+      "loss": 2.0517,
+      "step": 649500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.759262843568033e-05,
+      "loss": 2.0016,
+      "step": 650000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7590776611400083e-05,
+      "loss": 2.018,
+      "step": 650500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.758892478711984e-05,
+      "loss": 2.0559,
+      "step": 651000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.758707296283959e-05,
+      "loss": 2.0775,
+      "step": 651500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.758522113855935e-05,
+      "loss": 2.0608,
+      "step": 652000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7583369314279104e-05,
+      "loss": 2.0145,
+      "step": 652500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7581517489998855e-05,
+      "loss": 2.0347,
+      "step": 653000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.757966566571861e-05,
+      "loss": 2.0073,
+      "step": 653500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.757781384143836e-05,
+      "loss": 1.9822,
+      "step": 654000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.757596201715812e-05,
+      "loss": 2.0566,
+      "step": 654500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.757411019287787e-05,
+      "loss": 2.0311,
+      "step": 655000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7572258368597626e-05,
+      "loss": 2.0328,
+      "step": 655500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.757040654431738e-05,
+      "loss": 2.0276,
+      "step": 656000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.756855472003713e-05,
+      "loss": 1.9355,
+      "step": 656500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.756670289575689e-05,
+      "loss": 2.0245,
+      "step": 657000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.756485107147664e-05,
+      "loss": 2.0109,
+      "step": 657500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.75629992471964e-05,
+      "loss": 2.0427,
+      "step": 658000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7561147422916154e-05,
+      "loss": 2.0446,
+      "step": 658500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7559295598635904e-05,
+      "loss": 1.9809,
+      "step": 659000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.755744377435566e-05,
+      "loss": 2.0091,
+      "step": 659500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.755559195007541e-05,
+      "loss": 1.9967,
+      "step": 660000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.755374012579517e-05,
+      "loss": 2.1323,
+      "step": 660500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.755188830151492e-05,
+      "loss": 2.0253,
+      "step": 661000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.755003647723467e-05,
+      "loss": 2.0328,
+      "step": 661500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.754818465295443e-05,
+      "loss": 2.0563,
+      "step": 662000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.754633282867418e-05,
+      "loss": 2.0115,
+      "step": 662500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.754448100439394e-05,
+      "loss": 1.9977,
+      "step": 663000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.754262918011369e-05,
+      "loss": 1.9745,
+      "step": 663500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7540777355833446e-05,
+      "loss": 2.043,
+      "step": 664000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7538925531553197e-05,
+      "loss": 2.0064,
+      "step": 664500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7537073707272954e-05,
+      "loss": 2.0269,
+      "step": 665000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.753522188299271e-05,
+      "loss": 1.9855,
+      "step": 665500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.753337005871246e-05,
+      "loss": 2.0638,
+      "step": 666000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.753151823443222e-05,
+      "loss": 2.0233,
+      "step": 666500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.752966641015197e-05,
+      "loss": 2.0178,
+      "step": 667000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7527814585871725e-05,
+      "loss": 2.0099,
+      "step": 667500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.752596276159148e-05,
+      "loss": 2.0214,
+      "step": 668000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.752411093731123e-05,
+      "loss": 2.0505,
+      "step": 668500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.752225911303099e-05,
+      "loss": 2.0089,
+      "step": 669000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.752040728875074e-05,
+      "loss": 2.0446,
+      "step": 669500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7518555464470496e-05,
+      "loss": 1.991,
+      "step": 670000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7516703640190246e-05,
+      "loss": 1.974,
+      "step": 670500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.751485181591e-05,
+      "loss": 1.9898,
+      "step": 671000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.751299999162976e-05,
+      "loss": 2.0428,
+      "step": 671500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.751114816734951e-05,
+      "loss": 2.0752,
+      "step": 672000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.750929634306927e-05,
+      "loss": 1.9674,
+      "step": 672500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.750744451878902e-05,
+      "loss": 2.0534,
+      "step": 673000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7505592694508774e-05,
+      "loss": 1.9736,
+      "step": 673500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7503740870228524e-05,
+      "loss": 2.0287,
+      "step": 674000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.750188904594828e-05,
+      "loss": 2.0177,
+      "step": 674500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.750003722166804e-05,
+      "loss": 2.0582,
+      "step": 675000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.749818539738779e-05,
+      "loss": 2.0747,
+      "step": 675500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7496333573107545e-05,
+      "loss": 2.0076,
+      "step": 676000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7494481748827296e-05,
+      "loss": 1.9876,
+      "step": 676500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.749262992454705e-05,
+      "loss": 1.9917,
+      "step": 677000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.749077810026681e-05,
+      "loss": 2.0341,
+      "step": 677500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.748892627598656e-05,
+      "loss": 2.0301,
+      "step": 678000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7487074451706317e-05,
+      "loss": 2.0224,
+      "step": 678500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.748522262742607e-05,
+      "loss": 2.03,
+      "step": 679000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7483370803145824e-05,
+      "loss": 1.9859,
+      "step": 679500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7481518978865574e-05,
+      "loss": 1.9941,
+      "step": 680000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.747966715458533e-05,
+      "loss": 1.9824,
+      "step": 680500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.747781533030509e-05,
+      "loss": 1.98,
+      "step": 681000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.747596350602484e-05,
+      "loss": 1.9798,
+      "step": 681500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7474111681744595e-05,
+      "loss": 2.0378,
+      "step": 682000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7472259857464345e-05,
+      "loss": 1.9793,
+      "step": 682500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.74704080331841e-05,
+      "loss": 2.0431,
+      "step": 683000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.746855620890385e-05,
+      "loss": 2.0299,
+      "step": 683500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.746670438462361e-05,
+      "loss": 2.0331,
+      "step": 684000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7464852560343366e-05,
+      "loss": 2.0741,
+      "step": 684500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7463000736063116e-05,
+      "loss": 1.9826,
+      "step": 685000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.746114891178287e-05,
+      "loss": 2.0529,
+      "step": 685500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.745929708750262e-05,
+      "loss": 2.0303,
+      "step": 686000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.745744526322238e-05,
+      "loss": 1.9617,
+      "step": 686500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.745559343894213e-05,
+      "loss": 2.0406,
+      "step": 687000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.745374161466189e-05,
+      "loss": 2.0287,
+      "step": 687500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7451889790381644e-05,
+      "loss": 2.0281,
+      "step": 688000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7450037966101394e-05,
+      "loss": 1.9655,
+      "step": 688500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.744818614182115e-05,
+      "loss": 1.9935,
+      "step": 689000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.74463343175409e-05,
+      "loss": 1.9347,
+      "step": 689500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.744448249326066e-05,
+      "loss": 2.0455,
+      "step": 690000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7442630668980416e-05,
+      "loss": 2.0223,
+      "step": 690500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7440778844700166e-05,
+      "loss": 1.9986,
+      "step": 691000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.743892702041992e-05,
+      "loss": 1.9804,
+      "step": 691500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.743707519613967e-05,
+      "loss": 1.9666,
+      "step": 692000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.743522337185943e-05,
+      "loss": 2.0033,
+      "step": 692500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.743337154757918e-05,
+      "loss": 2.0306,
+      "step": 693000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.743151972329894e-05,
+      "loss": 2.0487,
+      "step": 693500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7429667899018694e-05,
+      "loss": 2.014,
+      "step": 694000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7427816074738444e-05,
+      "loss": 2.1064,
+      "step": 694500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.74259642504582e-05,
+      "loss": 2.0571,
+      "step": 695000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.742411242617795e-05,
+      "loss": 2.0911,
+      "step": 695500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.742226060189771e-05,
+      "loss": 2.04,
+      "step": 696000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.742040877761746e-05,
+      "loss": 2.0139,
+      "step": 696500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7418556953337215e-05,
+      "loss": 1.9906,
+      "step": 697000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.741670512905697e-05,
+      "loss": 2.0458,
+      "step": 697500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.741485330477672e-05,
+      "loss": 2.0134,
+      "step": 698000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.741300148049648e-05,
+      "loss": 2.0296,
+      "step": 698500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.741114965621623e-05,
+      "loss": 2.0438,
+      "step": 699000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7409297831935986e-05,
+      "loss": 2.0439,
+      "step": 699500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.740744600765574e-05,
+      "loss": 2.12,
+      "step": 700000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7405594183375493e-05,
+      "loss": 2.0337,
+      "step": 700500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.740374235909525e-05,
+      "loss": 2.0302,
+      "step": 701000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7401890534815e-05,
+      "loss": 2.0418,
+      "step": 701500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.740003871053476e-05,
+      "loss": 2.0212,
+      "step": 702000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.739818688625451e-05,
+      "loss": 1.9915,
+      "step": 702500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7396335061974265e-05,
+      "loss": 2.0892,
+      "step": 703000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.739448323769402e-05,
+      "loss": 1.9904,
+      "step": 703500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.739263141341377e-05,
+      "loss": 2.1306,
+      "step": 704000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.739077958913353e-05,
+      "loss": 2.1016,
+      "step": 704500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.738892776485328e-05,
+      "loss": 2.0034,
+      "step": 705000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7387075940573036e-05,
+      "loss": 1.9966,
+      "step": 705500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7385224116292786e-05,
+      "loss": 2.0254,
+      "step": 706000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.738337229201254e-05,
+      "loss": 2.0833,
+      "step": 706500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.73815204677323e-05,
+      "loss": 2.0133,
+      "step": 707000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.737966864345205e-05,
+      "loss": 1.9995,
+      "step": 707500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.737781681917181e-05,
+      "loss": 2.0442,
+      "step": 708000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.737596499489156e-05,
+      "loss": 2.071,
+      "step": 708500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7374113170611314e-05,
+      "loss": 2.0073,
+      "step": 709000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.737226134633107e-05,
+      "loss": 1.9698,
+      "step": 709500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.737040952205082e-05,
+      "loss": 1.9787,
+      "step": 710000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.736855769777058e-05,
+      "loss": 2.0202,
+      "step": 710500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.736670587349033e-05,
+      "loss": 2.0006,
+      "step": 711000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7364854049210085e-05,
+      "loss": 2.0398,
+      "step": 711500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7363002224929835e-05,
+      "loss": 1.9721,
+      "step": 712000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.736115040064959e-05,
+      "loss": 2.0857,
+      "step": 712500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.735929857636935e-05,
+      "loss": 2.0813,
+      "step": 713000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.73574467520891e-05,
+      "loss": 2.0385,
+      "step": 713500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7355594927808856e-05,
+      "loss": 2.0053,
+      "step": 714000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.735374310352861e-05,
+      "loss": 2.0042,
+      "step": 714500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7351891279248364e-05,
+      "loss": 2.027,
+      "step": 715000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7350039454968114e-05,
+      "loss": 1.933,
+      "step": 715500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.734818763068787e-05,
+      "loss": 2.0986,
+      "step": 716000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.734633580640763e-05,
+      "loss": 1.9804,
+      "step": 716500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.734448398212738e-05,
+      "loss": 2.0332,
+      "step": 717000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7342632157847135e-05,
+      "loss": 1.9209,
+      "step": 717500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7340780333566885e-05,
+      "loss": 2.0463,
+      "step": 718000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.733892850928664e-05,
+      "loss": 2.003,
+      "step": 718500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.73370766850064e-05,
+      "loss": 1.9589,
+      "step": 719000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.733522486072615e-05,
+      "loss": 2.066,
+      "step": 719500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7333373036445906e-05,
+      "loss": 2.0514,
+      "step": 720000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7331521212165656e-05,
+      "loss": 2.042,
+      "step": 720500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.732966938788541e-05,
+      "loss": 2.0349,
+      "step": 721000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.732781756360516e-05,
+      "loss": 1.9522,
+      "step": 721500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.732596573932492e-05,
+      "loss": 2.0051,
+      "step": 722000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.732411391504468e-05,
+      "loss": 1.9622,
+      "step": 722500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.732226209076443e-05,
+      "loss": 1.9827,
+      "step": 723000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7320410266484184e-05,
+      "loss": 2.0155,
+      "step": 723500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7318558442203934e-05,
+      "loss": 2.016,
+      "step": 724000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.731670661792369e-05,
+      "loss": 1.9934,
+      "step": 724500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.731485479364344e-05,
+      "loss": 2.0207,
+      "step": 725000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.73130029693632e-05,
+      "loss": 2.0133,
+      "step": 725500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7311151145082955e-05,
+      "loss": 2.0223,
+      "step": 726000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7309299320802706e-05,
+      "loss": 2.0113,
+      "step": 726500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.730744749652246e-05,
+      "loss": 1.9765,
+      "step": 727000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.730559567224221e-05,
+      "loss": 2.0156,
+      "step": 727500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.730374384796197e-05,
+      "loss": 2.0305,
+      "step": 728000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.730189202368173e-05,
+      "loss": 1.9378,
+      "step": 728500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.730004019940148e-05,
+      "loss": 2.0449,
+      "step": 729000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7298188375121234e-05,
+      "loss": 1.9851,
+      "step": 729500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7296336550840984e-05,
+      "loss": 1.9278,
+      "step": 730000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.729448472656074e-05,
+      "loss": 1.9945,
+      "step": 730500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.729263290228049e-05,
+      "loss": 1.9841,
+      "step": 731000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.729078107800025e-05,
+      "loss": 2.0696,
+      "step": 731500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7288929253720005e-05,
+      "loss": 1.9924,
+      "step": 732000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7287077429439755e-05,
+      "loss": 1.954,
+      "step": 732500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.728522560515951e-05,
+      "loss": 1.9555,
+      "step": 733000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.728337378087926e-05,
+      "loss": 1.9455,
+      "step": 733500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.728152195659902e-05,
+      "loss": 2.0125,
+      "step": 734000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.727967013231877e-05,
+      "loss": 2.0331,
+      "step": 734500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7277818308038526e-05,
+      "loss": 1.9452,
+      "step": 735000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.727596648375828e-05,
+      "loss": 2.0267,
+      "step": 735500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.727411465947803e-05,
+      "loss": 2.0673,
+      "step": 736000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.727226283519779e-05,
+      "loss": 2.0823,
+      "step": 736500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.727041101091754e-05,
+      "loss": 2.0305,
+      "step": 737000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.72685591866373e-05,
+      "loss": 2.025,
+      "step": 737500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.726670736235705e-05,
+      "loss": 2.0237,
+      "step": 738000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7264855538076805e-05,
+      "loss": 2.0442,
+      "step": 738500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.726300371379656e-05,
+      "loss": 1.998,
+      "step": 739000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.726115188951631e-05,
+      "loss": 1.9642,
+      "step": 739500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.725930006523607e-05,
+      "loss": 1.9426,
+      "step": 740000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.725744824095582e-05,
+      "loss": 2.0887,
+      "step": 740500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7255596416675576e-05,
+      "loss": 2.0098,
+      "step": 741000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.725374459239533e-05,
+      "loss": 1.9978,
+      "step": 741500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.725189276811508e-05,
+      "loss": 2.0871,
+      "step": 742000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.725004094383484e-05,
+      "loss": 1.9859,
+      "step": 742500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.724818911955459e-05,
+      "loss": 2.0032,
+      "step": 743000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.724633729527435e-05,
+      "loss": 2.057,
+      "step": 743500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.72444854709941e-05,
+      "loss": 2.0079,
+      "step": 744000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.724263364671386e-05,
+      "loss": 2.0698,
+      "step": 744500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.724078182243361e-05,
+      "loss": 1.9859,
+      "step": 745000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.723892999815336e-05,
+      "loss": 1.9798,
+      "step": 745500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.723707817387312e-05,
+      "loss": 1.9868,
+      "step": 746000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.723522634959287e-05,
+      "loss": 1.9978,
+      "step": 746500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7233374525312625e-05,
+      "loss": 2.0636,
+      "step": 747000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7231522701032375e-05,
+      "loss": 2.0346,
+      "step": 747500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.722967087675213e-05,
+      "loss": 1.9638,
+      "step": 748000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.722781905247189e-05,
+      "loss": 1.9988,
+      "step": 748500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.722596722819164e-05,
+      "loss": 1.9167,
+      "step": 749000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7224115403911396e-05,
+      "loss": 1.9686,
+      "step": 749500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7222263579631147e-05,
+      "loss": 2.0086,
+      "step": 750000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7220411755350904e-05,
+      "loss": 1.9991,
+      "step": 750500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.721855993107066e-05,
+      "loss": 2.0061,
+      "step": 751000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.721670810679041e-05,
+      "loss": 2.0566,
+      "step": 751500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.721485628251017e-05,
+      "loss": 2.0546,
+      "step": 752000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.721300445822992e-05,
+      "loss": 2.0482,
+      "step": 752500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7211152633949675e-05,
+      "loss": 2.0236,
+      "step": 753000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7209300809669425e-05,
+      "loss": 1.9986,
+      "step": 753500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.720744898538919e-05,
+      "loss": 2.0081,
+      "step": 754000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.720559716110894e-05,
+      "loss": 2.0366,
+      "step": 754500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.720374533682869e-05,
+      "loss": 1.9745,
+      "step": 755000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7201893512548446e-05,
+      "loss": 2.0764,
+      "step": 755500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7200041688268196e-05,
+      "loss": 2.0012,
+      "step": 756000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.719818986398795e-05,
+      "loss": 1.9298,
+      "step": 756500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.71963380397077e-05,
+      "loss": 1.9376,
+      "step": 757000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.719448621542747e-05,
+      "loss": 1.9849,
+      "step": 757500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.719263439114722e-05,
+      "loss": 2.0167,
+      "step": 758000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.719078256686697e-05,
+      "loss": 1.9952,
+      "step": 758500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7188930742586724e-05,
+      "loss": 2.0323,
+      "step": 759000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7187078918306474e-05,
+      "loss": 1.9649,
+      "step": 759500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.718522709402623e-05,
+      "loss": 1.9798,
+      "step": 760000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.718337526974599e-05,
+      "loss": 1.9652,
+      "step": 760500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.718152344546574e-05,
+      "loss": 1.9687,
+      "step": 761000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7179671621185495e-05,
+      "loss": 2.0004,
+      "step": 761500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7177819796905246e-05,
+      "loss": 1.9736,
+      "step": 762000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7175967972625e-05,
+      "loss": 2.0267,
+      "step": 762500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.717411614834475e-05,
+      "loss": 2.0713,
+      "step": 763000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.717226432406451e-05,
+      "loss": 1.9763,
+      "step": 763500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7170412499784267e-05,
+      "loss": 1.9805,
+      "step": 764000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.716856067550402e-05,
+      "loss": 1.9565,
+      "step": 764500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7166708851223774e-05,
+      "loss": 2.0273,
+      "step": 765000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7164857026943524e-05,
+      "loss": 1.9555,
+      "step": 765500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.716300520266328e-05,
+      "loss": 1.9901,
+      "step": 766000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.716115337838303e-05,
+      "loss": 1.9322,
+      "step": 766500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7159301554102795e-05,
+      "loss": 1.9818,
+      "step": 767000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7157449729822545e-05,
+      "loss": 1.991,
+      "step": 767500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7155597905542295e-05,
+      "loss": 1.9674,
+      "step": 768000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.715374608126205e-05,
+      "loss": 2.051,
+      "step": 768500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.71518942569818e-05,
+      "loss": 1.978,
+      "step": 769000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.715004243270156e-05,
+      "loss": 1.9921,
+      "step": 769500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7148190608421316e-05,
+      "loss": 1.9828,
+      "step": 770000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.714633878414107e-05,
+      "loss": 1.9944,
+      "step": 770500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.714448695986082e-05,
+      "loss": 2.0489,
+      "step": 771000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.714263513558057e-05,
+      "loss": 1.9693,
+      "step": 771500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.714078331130033e-05,
+      "loss": 2.0434,
+      "step": 772000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.713893148702008e-05,
+      "loss": 2.0486,
+      "step": 772500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.713707966273984e-05,
+      "loss": 1.9808,
+      "step": 773000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7135227838459594e-05,
+      "loss": 2.0136,
+      "step": 773500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.713337601417935e-05,
+      "loss": 2.0469,
+      "step": 774000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.71315241898991e-05,
+      "loss": 2.0264,
+      "step": 774500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.712967236561885e-05,
+      "loss": 1.9639,
+      "step": 775000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.712782054133861e-05,
+      "loss": 1.9759,
+      "step": 775500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.712596871705836e-05,
+      "loss": 2.0106,
+      "step": 776000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.712411689277812e-05,
+      "loss": 2.008,
+      "step": 776500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.712226506849787e-05,
+      "loss": 2.0776,
+      "step": 777000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.712041324421762e-05,
+      "loss": 2.1293,
+      "step": 777500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.711856141993738e-05,
+      "loss": 2.0102,
+      "step": 778000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.711670959565713e-05,
+      "loss": 2.0072,
+      "step": 778500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.711485777137689e-05,
+      "loss": 2.1087,
+      "step": 779000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7113005947096644e-05,
+      "loss": 1.9805,
+      "step": 779500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.71111541228164e-05,
+      "loss": 2.0418,
+      "step": 780000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.710930229853615e-05,
+      "loss": 2.0132,
+      "step": 780500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.71074504742559e-05,
+      "loss": 2.0929,
+      "step": 781000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.710559864997566e-05,
+      "loss": 2.0832,
+      "step": 781500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.710374682569541e-05,
+      "loss": 2.0905,
+      "step": 782000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7101895001415165e-05,
+      "loss": 1.9694,
+      "step": 782500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.710004317713492e-05,
+      "loss": 2.0284,
+      "step": 783000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.709819135285468e-05,
+      "loss": 2.0593,
+      "step": 783500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.709633952857443e-05,
+      "loss": 2.0327,
+      "step": 784000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.709448770429418e-05,
+      "loss": 2.0788,
+      "step": 784500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7092635880013936e-05,
+      "loss": 2.0705,
+      "step": 785000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7090784055733686e-05,
+      "loss": 2.0316,
+      "step": 785500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.708893223145345e-05,
+      "loss": 2.0512,
+      "step": 786000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.70870804071732e-05,
+      "loss": 2.1205,
+      "step": 786500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.708522858289296e-05,
+      "loss": 2.0344,
+      "step": 787000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.708337675861271e-05,
+      "loss": 2.0452,
+      "step": 787500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.708152493433246e-05,
+      "loss": 2.0642,
+      "step": 788000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7079673110052215e-05,
+      "loss": 2.0969,
+      "step": 788500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7077821285771965e-05,
+      "loss": 2.0998,
+      "step": 789000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.707596946149173e-05,
+      "loss": 2.0191,
+      "step": 789500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.707411763721148e-05,
+      "loss": 2.0373,
+      "step": 790000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.707226581293123e-05,
+      "loss": 2.0556,
+      "step": 790500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7070413988650986e-05,
+      "loss": 2.0802,
+      "step": 791000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7068562164370736e-05,
+      "loss": 2.0375,
+      "step": 791500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.706671034009049e-05,
+      "loss": 2.0806,
+      "step": 792000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.706485851581025e-05,
+      "loss": 2.1036,
+      "step": 792500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.706300669153001e-05,
+      "loss": 2.0088,
+      "step": 793000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.706115486724976e-05,
+      "loss": 2.0511,
+      "step": 793500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.705930304296951e-05,
+      "loss": 2.0661,
+      "step": 794000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7057451218689264e-05,
+      "loss": 2.0636,
+      "step": 794500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7055599394409014e-05,
+      "loss": 2.0155,
+      "step": 795000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.705374757012878e-05,
+      "loss": 2.0469,
+      "step": 795500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.705189574584853e-05,
+      "loss": 2.0243,
+      "step": 796000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7050043921568285e-05,
+      "loss": 1.9999,
+      "step": 796500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7048192097288035e-05,
+      "loss": 2.0995,
+      "step": 797000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7046340273007785e-05,
+      "loss": 1.9903,
+      "step": 797500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.704448844872754e-05,
+      "loss": 2.0851,
+      "step": 798000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.704263662444729e-05,
+      "loss": 2.0155,
+      "step": 798500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7040784800167056e-05,
+      "loss": 2.0831,
+      "step": 799000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7038932975886806e-05,
+      "loss": 2.0779,
+      "step": 799500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7037081151606563e-05,
+      "loss": 2.0864,
+      "step": 800000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7035229327326314e-05,
+      "loss": 2.0051,
+      "step": 800500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7033377503046064e-05,
+      "loss": 2.0414,
+      "step": 801000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.703152567876582e-05,
+      "loss": 2.027,
+      "step": 801500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.702967385448558e-05,
+      "loss": 2.0134,
+      "step": 802000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7027822030205335e-05,
+      "loss": 2.0272,
+      "step": 802500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.7025970205925085e-05,
+      "loss": 2.0365,
+      "step": 803000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.7024118381644835e-05,
+      "loss": 2.086,
+      "step": 803500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.702226655736459e-05,
+      "loss": 2.0551,
+      "step": 804000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.702041473308434e-05,
+      "loss": 2.1284,
+      "step": 804500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.7018562908804106e-05,
+      "loss": 2.0559,
+      "step": 805000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.7016711084523856e-05,
+      "loss": 2.0499,
+      "step": 805500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.701485926024361e-05,
+      "loss": 2.0513,
+      "step": 806000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.701300743596336e-05,
+      "loss": 2.0434,
+      "step": 806500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.701115561168311e-05,
+      "loss": 2.1082,
+      "step": 807000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.700930378740287e-05,
+      "loss": 2.0744,
+      "step": 807500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.700745196312262e-05,
+      "loss": 2.0394,
+      "step": 808000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.7005600138842384e-05,
+      "loss": 2.0527,
+      "step": 808500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.7003748314562134e-05,
+      "loss": 2.0863,
+      "step": 809000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.700189649028189e-05,
+      "loss": 2.0699,
+      "step": 809500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.700004466600164e-05,
+      "loss": 2.017,
+      "step": 810000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.699819284172139e-05,
+      "loss": 2.06,
+      "step": 810500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.699634101744115e-05,
+      "loss": 2.1166,
+      "step": 811000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6994489193160905e-05,
+      "loss": 2.1231,
+      "step": 811500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.699263736888066e-05,
+      "loss": 2.0806,
+      "step": 812000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.699078554460041e-05,
+      "loss": 2.0696,
+      "step": 812500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.698893372032017e-05,
+      "loss": 2.054,
+      "step": 813000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.698708189603992e-05,
+      "loss": 1.9506,
+      "step": 813500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.698523007175967e-05,
+      "loss": 2.1082,
+      "step": 814000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.698337824747943e-05,
+      "loss": 2.0141,
+      "step": 814500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6981526423199184e-05,
+      "loss": 2.1588,
+      "step": 815000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.697967459891894e-05,
+      "loss": 2.0799,
+      "step": 815500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.697782277463869e-05,
+      "loss": 2.0682,
+      "step": 816000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.697597095035845e-05,
+      "loss": 2.0533,
+      "step": 816500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.69741191260782e-05,
+      "loss": 2.0551,
+      "step": 817000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.697226730179795e-05,
+      "loss": 2.094,
+      "step": 817500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.697041547751771e-05,
+      "loss": 2.0754,
+      "step": 818000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.696856365323746e-05,
+      "loss": 2.0422,
+      "step": 818500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.696671182895722e-05,
+      "loss": 2.0126,
+      "step": 819000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.696486000467697e-05,
+      "loss": 2.1076,
+      "step": 819500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.696300818039672e-05,
+      "loss": 2.0516,
+      "step": 820000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6961156356116476e-05,
+      "loss": 2.0432,
+      "step": 820500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.695930453183623e-05,
+      "loss": 2.0257,
+      "step": 821000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.695745270755599e-05,
+      "loss": 2.0407,
+      "step": 821500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.695560088327574e-05,
+      "loss": 2.074,
+      "step": 822000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.69537490589955e-05,
+      "loss": 2.1457,
+      "step": 822500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.695189723471525e-05,
+      "loss": 2.0851,
+      "step": 823000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6950045410435e-05,
+      "loss": 2.1232,
+      "step": 823500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6948193586154755e-05,
+      "loss": 2.0155,
+      "step": 824000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.694634176187451e-05,
+      "loss": 2.0658,
+      "step": 824500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.694448993759427e-05,
+      "loss": 2.0657,
+      "step": 825000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.694263811331402e-05,
+      "loss": 2.0418,
+      "step": 825500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6940786289033776e-05,
+      "loss": 2.0449,
+      "step": 826000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6938934464753526e-05,
+      "loss": 2.0938,
+      "step": 826500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6937082640473276e-05,
+      "loss": 2.0486,
+      "step": 827000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.693523081619304e-05,
+      "loss": 2.0815,
+      "step": 827500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.693337899191279e-05,
+      "loss": 2.0487,
+      "step": 828000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.693152716763255e-05,
+      "loss": 2.0896,
+      "step": 828500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.69296753433523e-05,
+      "loss": 1.9657,
+      "step": 829000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6927823519072054e-05,
+      "loss": 2.0729,
+      "step": 829500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6925971694791804e-05,
+      "loss": 2.0866,
+      "step": 830000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.692411987051156e-05,
+      "loss": 2.055,
+      "step": 830500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.692226804623132e-05,
+      "loss": 1.9861,
+      "step": 831000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.692041622195107e-05,
+      "loss": 2.0604,
+      "step": 831500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6918564397670825e-05,
+      "loss": 2.0937,
+      "step": 832000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6916712573390575e-05,
+      "loss": 2.0143,
+      "step": 832500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6914860749110325e-05,
+      "loss": 2.0175,
+      "step": 833000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.691300892483008e-05,
+      "loss": 2.0605,
+      "step": 833500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.691115710054984e-05,
+      "loss": 2.093,
+      "step": 834000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6909305276269596e-05,
+      "loss": 2.103,
+      "step": 834500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6907453451989346e-05,
+      "loss": 2.0349,
+      "step": 835000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.69056016277091e-05,
+      "loss": 2.0606,
+      "step": 835500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6903749803428854e-05,
+      "loss": 2.062,
+      "step": 836000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6901897979148604e-05,
+      "loss": 2.008,
+      "step": 836500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.690004615486837e-05,
+      "loss": 2.0876,
+      "step": 837000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.689819433058812e-05,
+      "loss": 2.051,
+      "step": 837500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6896342506307875e-05,
+      "loss": 2.0967,
+      "step": 838000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6894490682027625e-05,
+      "loss": 2.0893,
+      "step": 838500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.689263885774738e-05,
+      "loss": 2.0889,
+      "step": 839000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.689078703346713e-05,
+      "loss": 2.0079,
+      "step": 839500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.688893520918688e-05,
+      "loss": 2.0384,
+      "step": 840000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6887083384906646e-05,
+      "loss": 2.1172,
+      "step": 840500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6885231560626396e-05,
+      "loss": 2.0889,
+      "step": 841000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.688337973634615e-05,
+      "loss": 1.9858,
+      "step": 841500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.68815279120659e-05,
+      "loss": 1.9447,
+      "step": 842000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.687967608778566e-05,
+      "loss": 2.0305,
+      "step": 842500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.687782426350541e-05,
+      "loss": 2.0814,
+      "step": 843000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.687597243922517e-05,
+      "loss": 2.055,
+      "step": 843500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6874120614944924e-05,
+      "loss": 2.0451,
+      "step": 844000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6872268790664674e-05,
+      "loss": 2.0967,
+      "step": 844500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.687041696638443e-05,
+      "loss": 2.0854,
+      "step": 845000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.686856514210418e-05,
+      "loss": 2.0563,
+      "step": 845500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.686671331782393e-05,
+      "loss": 2.0294,
+      "step": 846000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6864861493543695e-05,
+      "loss": 2.0123,
+      "step": 846500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6863009669263445e-05,
+      "loss": 2.0806,
+      "step": 847000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.68611578449832e-05,
+      "loss": 2.0407,
+      "step": 847500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.685930602070295e-05,
+      "loss": 2.0924,
+      "step": 848000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.685745419642271e-05,
+      "loss": 2.1218,
+      "step": 848500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.685560237214246e-05,
+      "loss": 2.0243,
+      "step": 849000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.685375054786221e-05,
+      "loss": 2.1099,
+      "step": 849500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6851898723581973e-05,
+      "loss": 2.0467,
+      "step": 850000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6850046899301724e-05,
+      "loss": 2.0296,
+      "step": 850500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.684819507502148e-05,
+      "loss": 2.0885,
+      "step": 851000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.684634325074123e-05,
+      "loss": 2.0313,
+      "step": 851500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.684449142646099e-05,
+      "loss": 2.0465,
+      "step": 852000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.684263960218074e-05,
+      "loss": 2.0468,
+      "step": 852500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6840787777900495e-05,
+      "loss": 2.0439,
+      "step": 853000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.683893595362025e-05,
+      "loss": 2.0157,
+      "step": 853500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.683708412934e-05,
+      "loss": 2.028,
+      "step": 854000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.683523230505976e-05,
+      "loss": 2.0658,
+      "step": 854500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.683338048077951e-05,
+      "loss": 2.083,
+      "step": 855000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6831528656499266e-05,
+      "loss": 2.1119,
+      "step": 855500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.682967683221902e-05,
+      "loss": 2.1112,
+      "step": 856000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.682782500793877e-05,
+      "loss": 2.052,
+      "step": 856500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.682597318365853e-05,
+      "loss": 1.993,
+      "step": 857000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.682412135937828e-05,
+      "loss": 2.0158,
+      "step": 857500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.682226953509804e-05,
+      "loss": 1.9846,
+      "step": 858000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.682041771081779e-05,
+      "loss": 2.0556,
+      "step": 858500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6818565886537544e-05,
+      "loss": 2.0791,
+      "step": 859000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.68167140622573e-05,
+      "loss": 2.0545,
+      "step": 859500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.681486223797705e-05,
+      "loss": 2.0347,
+      "step": 860000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.681301041369681e-05,
+      "loss": 2.0501,
+      "step": 860500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.681115858941656e-05,
+      "loss": 2.0396,
+      "step": 861000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6809306765136315e-05,
+      "loss": 2.04,
+      "step": 861500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6807454940856066e-05,
+      "loss": 2.0529,
+      "step": 862000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.680560311657582e-05,
+      "loss": 2.0652,
+      "step": 862500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.680375129229558e-05,
+      "loss": 2.1191,
+      "step": 863000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.680189946801533e-05,
+      "loss": 2.0805,
+      "step": 863500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.680004764373509e-05,
+      "loss": 2.0712,
+      "step": 864000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.679819581945484e-05,
+      "loss": 2.0654,
+      "step": 864500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6796343995174594e-05,
+      "loss": 2.027,
+      "step": 865000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6794492170894344e-05,
+      "loss": 2.0291,
+      "step": 865500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.67926403466141e-05,
+      "loss": 2.007,
+      "step": 866000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.679078852233386e-05,
+      "loss": 2.0595,
+      "step": 866500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.678893669805361e-05,
+      "loss": 2.0951,
+      "step": 867000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6787084873773365e-05,
+      "loss": 2.0557,
+      "step": 867500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6785233049493115e-05,
+      "loss": 2.0327,
+      "step": 868000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.678338122521287e-05,
+      "loss": 2.0578,
+      "step": 868500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.678152940093263e-05,
+      "loss": 2.0355,
+      "step": 869000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.677967757665238e-05,
+      "loss": 2.0443,
+      "step": 869500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6777825752372136e-05,
+      "loss": 2.1054,
+      "step": 870000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6775973928091886e-05,
+      "loss": 2.0842,
+      "step": 870500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.677412210381164e-05,
+      "loss": 2.0282,
+      "step": 871000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6772270279531393e-05,
+      "loss": 2.1102,
+      "step": 871500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.677041845525115e-05,
+      "loss": 2.1004,
+      "step": 872000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.676856663097091e-05,
+      "loss": 2.043,
+      "step": 872500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.676671480669066e-05,
+      "loss": 2.0572,
+      "step": 873000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6764862982410414e-05,
+      "loss": 2.0051,
+      "step": 873500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6763011158130165e-05,
+      "loss": 2.0985,
+      "step": 874000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.676115933384992e-05,
+      "loss": 2.0952,
+      "step": 874500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.675930750956967e-05,
+      "loss": 2.0283,
+      "step": 875000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.675745568528943e-05,
+      "loss": 2.0579,
+      "step": 875500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6755603861009186e-05,
+      "loss": 2.0514,
+      "step": 876000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6753752036728936e-05,
+      "loss": 2.1338,
+      "step": 876500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.675190021244869e-05,
+      "loss": 2.1004,
+      "step": 877000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.675004838816844e-05,
+      "loss": 2.0008,
+      "step": 877500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.67481965638882e-05,
+      "loss": 2.0826,
+      "step": 878000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.674634473960796e-05,
+      "loss": 2.0655,
+      "step": 878500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.674449291532771e-05,
+      "loss": 2.0472,
+      "step": 879000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6742641091047464e-05,
+      "loss": 1.9664,
+      "step": 879500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6740789266767214e-05,
+      "loss": 2.0571,
+      "step": 880000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.673893744248697e-05,
+      "loss": 2.0605,
+      "step": 880500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.673708561820672e-05,
+      "loss": 2.0526,
+      "step": 881000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.673523379392648e-05,
+      "loss": 2.0348,
+      "step": 881500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6733381969646235e-05,
+      "loss": 2.0136,
+      "step": 882000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6731530145365985e-05,
+      "loss": 2.1102,
+      "step": 882500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.672967832108574e-05,
+      "loss": 2.0325,
+      "step": 883000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.672782649680549e-05,
+      "loss": 2.0209,
+      "step": 883500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.672597467252525e-05,
+      "loss": 2.078,
+      "step": 884000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6724122848245e-05,
+      "loss": 2.0908,
+      "step": 884500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6722271023964756e-05,
+      "loss": 1.9744,
+      "step": 885000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6720419199684513e-05,
+      "loss": 2.0464,
+      "step": 885500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6718567375404264e-05,
+      "loss": 2.0477,
+      "step": 886000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.671671555112402e-05,
+      "loss": 2.0451,
+      "step": 886500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.671486372684377e-05,
+      "loss": 1.9843,
+      "step": 887000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.671301190256353e-05,
+      "loss": 2.012,
+      "step": 887500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6711160078283285e-05,
+      "loss": 2.0533,
+      "step": 888000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6709308254003035e-05,
+      "loss": 2.0422,
+      "step": 888500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.670745642972279e-05,
+      "loss": 2.0168,
+      "step": 889000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.670560460544254e-05,
+      "loss": 2.0118,
+      "step": 889500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.67037527811623e-05,
+      "loss": 2.0773,
+      "step": 890000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.670190095688205e-05,
+      "loss": 2.133,
+      "step": 890500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6700049132601806e-05,
+      "loss": 2.0453,
+      "step": 891000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.669819730832156e-05,
+      "loss": 2.0374,
+      "step": 891500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.669634548404131e-05,
+      "loss": 2.0671,
+      "step": 892000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.669449365976107e-05,
+      "loss": 2.0917,
+      "step": 892500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.669264183548082e-05,
+      "loss": 2.0369,
+      "step": 893000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.669079001120058e-05,
+      "loss": 2.0395,
+      "step": 893500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.668893818692033e-05,
+      "loss": 2.056,
+      "step": 894000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.6687086362640084e-05,
+      "loss": 2.047,
+      "step": 894500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.668523453835984e-05,
+      "loss": 2.0453,
+      "step": 895000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.668338271407959e-05,
+      "loss": 2.0838,
+      "step": 895500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.668153088979935e-05,
+      "loss": 2.0307,
+      "step": 896000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.66796790655191e-05,
+      "loss": 2.01,
+      "step": 896500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.6677827241238855e-05,
+      "loss": 2.0041,
+      "step": 897000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.667597541695861e-05,
+      "loss": 2.0933,
+      "step": 897500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.667412359267836e-05,
+      "loss": 2.1238,
+      "step": 898000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.667227176839812e-05,
+      "loss": 2.0307,
+      "step": 898500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.667041994411787e-05,
+      "loss": 2.0666,
+      "step": 899000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.6668568119837627e-05,
+      "loss": 2.1073,
+      "step": 899500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.666671629555738e-05,
+      "loss": 2.044,
+      "step": 900000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.6664864471277134e-05,
+      "loss": 2.1487,
+      "step": 900500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.666301264699689e-05,
+      "loss": 2.008,
+      "step": 901000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.666116082271664e-05,
+      "loss": 2.0978,
+      "step": 901500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.66593089984364e-05,
+      "loss": 2.0229,
+      "step": 902000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.665745717415615e-05,
+      "loss": 2.0439,
+      "step": 902500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.6655605349875905e-05,
+      "loss": 1.985,
+      "step": 903000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.6653753525595655e-05,
+      "loss": 2.0497,
+      "step": 903500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.665190170131541e-05,
+      "loss": 2.0531,
+      "step": 904000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.665004987703517e-05,
+      "loss": 2.059,
+      "step": 904500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.664819805275492e-05,
+      "loss": 2.056,
+      "step": 905000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.6646346228474676e-05,
+      "loss": 1.9899,
+      "step": 905500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.6644494404194426e-05,
+      "loss": 1.9882,
+      "step": 906000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.664264257991418e-05,
+      "loss": 2.0522,
+      "step": 906500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.664079075563394e-05,
+      "loss": 2.0068,
+      "step": 907000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.663893893135369e-05,
+      "loss": 2.0886,
+      "step": 907500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.663708710707345e-05,
+      "loss": 2.1016,
+      "step": 908000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.66352352827932e-05,
+      "loss": 2.0114,
+      "step": 908500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.6633383458512954e-05,
+      "loss": 1.9802,
+      "step": 909000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.6631531634232705e-05,
+      "loss": 2.0844,
+      "step": 909500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.662967980995246e-05,
+      "loss": 2.0211,
+      "step": 910000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.662782798567222e-05,
+      "loss": 2.0286,
+      "step": 910500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.662597616139197e-05,
+      "loss": 2.0299,
+      "step": 911000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.6624124337111726e-05,
+      "loss": 2.0454,
+      "step": 911500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.6622272512831476e-05,
+      "loss": 2.0932,
+      "step": 912000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.662042068855123e-05,
+      "loss": 2.0509,
+      "step": 912500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.661856886427098e-05,
+      "loss": 2.0061,
+      "step": 913000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.661671703999074e-05,
+      "loss": 1.9975,
+      "step": 913500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.66148652157105e-05,
+      "loss": 1.991,
+      "step": 914000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.661301339143025e-05,
+      "loss": 2.1047,
+      "step": 914500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.6611161567150004e-05,
+      "loss": 2.0214,
+      "step": 915000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.6609309742869754e-05,
+      "loss": 2.0069,
+      "step": 915500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.660745791858951e-05,
+      "loss": 2.0557,
+      "step": 916000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.660560609430926e-05,
+      "loss": 2.0242,
+      "step": 916500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.660375427002902e-05,
+      "loss": 2.0493,
+      "step": 917000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.6601902445748775e-05,
+      "loss": 1.9199,
+      "step": 917500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.6600050621468525e-05,
+      "loss": 2.0903,
+      "step": 918000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.659819879718828e-05,
+      "loss": 2.0927,
+      "step": 918500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.659634697290803e-05,
+      "loss": 2.022,
+      "step": 919000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.659449514862779e-05,
+      "loss": 2.0474,
+      "step": 919500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.6592643324347546e-05,
+      "loss": 2.0682,
+      "step": 920000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.6590791500067296e-05,
+      "loss": 1.9685,
+      "step": 920500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.658893967578705e-05,
+      "loss": 2.0483,
+      "step": 921000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.6587087851506803e-05,
+      "loss": 2.0165,
+      "step": 921500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.658523602722656e-05,
+      "loss": 2.0368,
+      "step": 922000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.658338420294631e-05,
+      "loss": 2.0493,
+      "step": 922500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.658153237866607e-05,
+      "loss": 1.9734,
+      "step": 923000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.6579680554385825e-05,
+      "loss": 2.058,
+      "step": 923500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.6577828730105575e-05,
+      "loss": 2.0363,
+      "step": 924000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.657597690582533e-05,
+      "loss": 2.0874,
+      "step": 924500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.657412508154508e-05,
+      "loss": 2.0214,
+      "step": 925000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.657227325726484e-05,
+      "loss": 2.0202,
+      "step": 925500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.657042143298459e-05,
+      "loss": 2.0941,
+      "step": 926000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.6568569608704346e-05,
+      "loss": 2.1147,
+      "step": 926500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.65667177844241e-05,
+      "loss": 2.013,
+      "step": 927000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.656486596014385e-05,
+      "loss": 2.0582,
+      "step": 927500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.656301413586361e-05,
+      "loss": 1.9939,
+      "step": 928000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.656116231158336e-05,
+      "loss": 2.0514,
+      "step": 928500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.655931048730312e-05,
+      "loss": 2.0541,
+      "step": 929000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.6557458663022874e-05,
+      "loss": 2.0357,
+      "step": 929500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.6555606838742624e-05,
+      "loss": 2.0714,
+      "step": 930000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.655375501446238e-05,
+      "loss": 2.0783,
+      "step": 930500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.655190319018213e-05,
+      "loss": 1.9661,
+      "step": 931000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.655005136590189e-05,
+      "loss": 2.0796,
+      "step": 931500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.654819954162164e-05,
+      "loss": 2.115,
+      "step": 932000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.6546347717341395e-05,
+      "loss": 2.0506,
+      "step": 932500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.654449589306115e-05,
+      "loss": 2.1206,
+      "step": 933000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.65426440687809e-05,
+      "loss": 2.0115,
+      "step": 933500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.654079224450066e-05,
+      "loss": 1.9922,
+      "step": 934000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.653894042022041e-05,
+      "loss": 1.9956,
+      "step": 934500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.6537088595940167e-05,
+      "loss": 2.0415,
+      "step": 935000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.653523677165992e-05,
+      "loss": 2.0342,
+      "step": 935500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.6533384947379674e-05,
+      "loss": 2.0324,
+      "step": 936000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.653153312309943e-05,
+      "loss": 2.0401,
+      "step": 936500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.652968129881918e-05,
+      "loss": 2.0224,
+      "step": 937000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.652782947453894e-05,
+      "loss": 2.0999,
+      "step": 937500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.652597765025869e-05,
+      "loss": 2.0523,
+      "step": 938000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6524125825978445e-05,
+      "loss": 2.0315,
+      "step": 938500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.65222740016982e-05,
+      "loss": 1.9957,
+      "step": 939000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.652042217741795e-05,
+      "loss": 2.073,
+      "step": 939500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.651857035313771e-05,
+      "loss": 1.9975,
+      "step": 940000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.651671852885746e-05,
+      "loss": 1.9802,
+      "step": 940500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6514866704577216e-05,
+      "loss": 2.0256,
+      "step": 941000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6513014880296966e-05,
+      "loss": 2.0479,
+      "step": 941500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.651116305601672e-05,
+      "loss": 2.0764,
+      "step": 942000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.650931123173648e-05,
+      "loss": 1.9871,
+      "step": 942500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.650745940745623e-05,
+      "loss": 2.0974,
+      "step": 943000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.650560758317599e-05,
+      "loss": 1.9827,
+      "step": 943500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.650375575889574e-05,
+      "loss": 2.076,
+      "step": 944000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6501903934615494e-05,
+      "loss": 2.0276,
+      "step": 944500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6500052110335244e-05,
+      "loss": 2.0284,
+      "step": 945000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6498200286055e-05,
+      "loss": 2.0503,
+      "step": 945500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.649634846177476e-05,
+      "loss": 2.117,
+      "step": 946000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.649449663749451e-05,
+      "loss": 2.0124,
+      "step": 946500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6492644813214265e-05,
+      "loss": 1.9898,
+      "step": 947000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6490792988934016e-05,
+      "loss": 2.0763,
+      "step": 947500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.648894116465377e-05,
+      "loss": 2.0705,
+      "step": 948000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.648708934037353e-05,
+      "loss": 2.0323,
+      "step": 948500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.648523751609328e-05,
+      "loss": 2.0466,
+      "step": 949000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.648338569181304e-05,
+      "loss": 2.001,
+      "step": 949500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.648153386753279e-05,
+      "loss": 2.0487,
+      "step": 950000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6479682043252544e-05,
+      "loss": 2.086,
+      "step": 950500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6477830218972294e-05,
+      "loss": 1.9899,
+      "step": 951000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.647597839469205e-05,
+      "loss": 2.0867,
+      "step": 951500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.647412657041181e-05,
+      "loss": 2.0617,
+      "step": 952000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.647227474613156e-05,
+      "loss": 1.9912,
+      "step": 952500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6470422921851315e-05,
+      "loss": 2.0193,
+      "step": 953000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6468571097571065e-05,
+      "loss": 2.0106,
+      "step": 953500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.646671927329082e-05,
+      "loss": 2.0216,
+      "step": 954000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.646486744901057e-05,
+      "loss": 2.0525,
+      "step": 954500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6463015624730336e-05,
+      "loss": 2.0356,
+      "step": 955000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6461163800450086e-05,
+      "loss": 2.1147,
+      "step": 955500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6459311976169836e-05,
+      "loss": 2.0354,
+      "step": 956000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.645746015188959e-05,
+      "loss": 2.0831,
+      "step": 956500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6455608327609343e-05,
+      "loss": 2.0503,
+      "step": 957000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.64537565033291e-05,
+      "loss": 1.9702,
+      "step": 957500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.645190467904886e-05,
+      "loss": 1.9823,
+      "step": 958000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6450052854768614e-05,
+      "loss": 2.0705,
+      "step": 958500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6448201030488364e-05,
+      "loss": 2.0601,
+      "step": 959000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6446349206208115e-05,
+      "loss": 2.0164,
+      "step": 959500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.644449738192787e-05,
+      "loss": 2.1266,
+      "step": 960000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.644264555764762e-05,
+      "loss": 2.0423,
+      "step": 960500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.644079373336738e-05,
+      "loss": 2.0309,
+      "step": 961000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6438941909087136e-05,
+      "loss": 2.0068,
+      "step": 961500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6437090084806886e-05,
+      "loss": 2.0237,
+      "step": 962000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.643523826052664e-05,
+      "loss": 2.0418,
+      "step": 962500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.643338643624639e-05,
+      "loss": 2.0352,
+      "step": 963000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.643153461196615e-05,
+      "loss": 2.0765,
+      "step": 963500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.64296827876859e-05,
+      "loss": 2.092,
+      "step": 964000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6427830963405664e-05,
+      "loss": 2.0081,
+      "step": 964500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6425979139125414e-05,
+      "loss": 2.01,
+      "step": 965000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6424127314845164e-05,
+      "loss": 2.0473,
+      "step": 965500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.642227549056492e-05,
+      "loss": 2.0386,
+      "step": 966000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.642042366628467e-05,
+      "loss": 2.0354,
+      "step": 966500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.641857184200443e-05,
+      "loss": 2.0138,
+      "step": 967000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.641672001772418e-05,
+      "loss": 2.0468,
+      "step": 967500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.641486819344394e-05,
+      "loss": 1.9797,
+      "step": 968000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.641301636916369e-05,
+      "loss": 2.1049,
+      "step": 968500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.641116454488344e-05,
+      "loss": 2.0446,
+      "step": 969000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.64093127206032e-05,
+      "loss": 2.0752,
+      "step": 969500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.640746089632295e-05,
+      "loss": 2.0608,
+      "step": 970000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6405609072042706e-05,
+      "loss": 2.0546,
+      "step": 970500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6403757247762463e-05,
+      "loss": 2.0337,
+      "step": 971000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.640190542348222e-05,
+      "loss": 2.0205,
+      "step": 971500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.640005359920197e-05,
+      "loss": 2.1151,
+      "step": 972000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.639820177492172e-05,
+      "loss": 2.0644,
+      "step": 972500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.639634995064148e-05,
+      "loss": 2.0661,
+      "step": 973000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.639449812636123e-05,
+      "loss": 1.9976,
+      "step": 973500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.639264630208099e-05,
+      "loss": 2.0132,
+      "step": 974000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.639079447780074e-05,
+      "loss": 2.0034,
+      "step": 974500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.638894265352049e-05,
+      "loss": 1.9839,
+      "step": 975000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.638709082924025e-05,
+      "loss": 2.1059,
+      "step": 975500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.638523900496e-05,
+      "loss": 1.9894,
+      "step": 976000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6383387180679756e-05,
+      "loss": 2.047,
+      "step": 976500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6381535356399506e-05,
+      "loss": 1.9589,
+      "step": 977000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.637968353211927e-05,
+      "loss": 2.0083,
+      "step": 977500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.637783170783902e-05,
+      "loss": 2.0254,
+      "step": 978000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.637597988355877e-05,
+      "loss": 2.0772,
+      "step": 978500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.637412805927853e-05,
+      "loss": 2.0772,
+      "step": 979000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.637227623499828e-05,
+      "loss": 2.0938,
+      "step": 979500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6370424410718034e-05,
+      "loss": 2.0982,
+      "step": 980000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.636857258643779e-05,
+      "loss": 2.0493,
+      "step": 980500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.636672076215755e-05,
+      "loss": 2.0379,
+      "step": 981000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.63648689378773e-05,
+      "loss": 2.0638,
+      "step": 981500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.636301711359705e-05,
+      "loss": 2.1105,
+      "step": 982000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6361165289316805e-05,
+      "loss": 2.0282,
+      "step": 982500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.6359313465036556e-05,
+      "loss": 1.988,
+      "step": 983000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.635746164075631e-05,
+      "loss": 2.0554,
+      "step": 983500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.635560981647607e-05,
+      "loss": 2.1075,
+      "step": 984000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.6353757992195826e-05,
+      "loss": 2.0504,
+      "step": 984500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.6351906167915577e-05,
+      "loss": 2.0456,
+      "step": 985000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.635005434363533e-05,
+      "loss": 2.0108,
+      "step": 985500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.6348202519355084e-05,
+      "loss": 2.0372,
+      "step": 986000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.6346350695074834e-05,
+      "loss": 2.0026,
+      "step": 986500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.63444988707946e-05,
+      "loss": 1.9993,
+      "step": 987000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.634264704651435e-05,
+      "loss": 2.0701,
+      "step": 987500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.63407952222341e-05,
+      "loss": 2.0102,
+      "step": 988000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.6338943397953855e-05,
+      "loss": 1.982,
+      "step": 988500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.6337091573673605e-05,
+      "loss": 2.0209,
+      "step": 989000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.633523974939336e-05,
+      "loss": 2.0488,
+      "step": 989500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.633338792511312e-05,
+      "loss": 2.0136,
+      "step": 990000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.6331536100832876e-05,
+      "loss": 2.1251,
+      "step": 990500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.6329684276552626e-05,
+      "loss": 2.0328,
+      "step": 991000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.6327832452272376e-05,
+      "loss": 1.9825,
+      "step": 991500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.632598062799213e-05,
+      "loss": 2.06,
+      "step": 992000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.632412880371188e-05,
+      "loss": 2.0039,
+      "step": 992500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.632227697943164e-05,
+      "loss": 1.9938,
+      "step": 993000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.63204251551514e-05,
+      "loss": 2.0274,
+      "step": 993500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.6318573330871154e-05,
+      "loss": 2.0711,
+      "step": 994000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.6316721506590904e-05,
+      "loss": 1.9769,
+      "step": 994500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.6314869682310655e-05,
+      "loss": 1.9745,
+      "step": 995000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.631301785803041e-05,
+      "loss": 1.992,
+      "step": 995500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.631116603375016e-05,
+      "loss": 2.0302,
+      "step": 996000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.6309314209469925e-05,
+      "loss": 2.0624,
+      "step": 996500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.6307462385189676e-05,
+      "loss": 1.9892,
+      "step": 997000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.630561056090943e-05,
+      "loss": 2.0339,
+      "step": 997500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.630375873662918e-05,
+      "loss": 2.0717,
+      "step": 998000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.630190691234893e-05,
+      "loss": 2.0768,
+      "step": 998500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.630005508806869e-05,
+      "loss": 2.0493,
+      "step": 999000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.629820326378845e-05,
+      "loss": 2.0204,
+      "step": 999500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.6296351439508204e-05,
+      "loss": 2.0824,
+      "step": 1000000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.6294499615227954e-05,
+      "loss": 2.1043,
+      "step": 1000500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.629264779094771e-05,
+      "loss": 2.0029,
+      "step": 1001000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.629079596666746e-05,
+      "loss": 2.0412,
+      "step": 1001500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.628894414238721e-05,
+      "loss": 2.0522,
+      "step": 1002000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.628709231810697e-05,
+      "loss": 2.0814,
+      "step": 1002500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.6285240493826725e-05,
+      "loss": 2.1019,
+      "step": 1003000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.628338866954648e-05,
+      "loss": 2.0045,
+      "step": 1003500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.628153684526623e-05,
+      "loss": 1.9963,
+      "step": 1004000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.627968502098598e-05,
+      "loss": 2.0839,
+      "step": 1004500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.627783319670574e-05,
+      "loss": 2.0057,
+      "step": 1005000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.627598137242549e-05,
+      "loss": 1.9783,
+      "step": 1005500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.627412954814525e-05,
+      "loss": 2.0617,
+      "step": 1006000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.6272277723865e-05,
+      "loss": 2.0306,
+      "step": 1006500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.627042589958476e-05,
+      "loss": 2.0356,
+      "step": 1007000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.626857407530451e-05,
+      "loss": 1.9083,
+      "step": 1007500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.626672225102426e-05,
+      "loss": 2.0926,
+      "step": 1008000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.626487042674402e-05,
+      "loss": 2.1131,
+      "step": 1008500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.626301860246377e-05,
+      "loss": 2.0052,
+      "step": 1009000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.626116677818353e-05,
+      "loss": 2.1092,
+      "step": 1009500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.625931495390328e-05,
+      "loss": 1.9954,
+      "step": 1010000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.625746312962304e-05,
+      "loss": 1.9873,
+      "step": 1010500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.625561130534279e-05,
+      "loss": 2.0889,
+      "step": 1011000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.625375948106254e-05,
+      "loss": 2.0882,
+      "step": 1011500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.6251907656782296e-05,
+      "loss": 2.044,
+      "step": 1012000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.625005583250205e-05,
+      "loss": 1.9891,
+      "step": 1012500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.624820400822181e-05,
+      "loss": 2.0041,
+      "step": 1013000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.624635218394156e-05,
+      "loss": 2.0249,
+      "step": 1013500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.624450035966132e-05,
+      "loss": 2.0084,
+      "step": 1014000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.624264853538107e-05,
+      "loss": 2.0522,
+      "step": 1014500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.624079671110082e-05,
+      "loss": 2.0631,
+      "step": 1015000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.623894488682058e-05,
+      "loss": 2.1273,
+      "step": 1015500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.623709306254033e-05,
+      "loss": 2.0269,
+      "step": 1016000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.623524123826009e-05,
+      "loss": 1.9982,
+      "step": 1016500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.623338941397984e-05,
+      "loss": 2.0475,
+      "step": 1017000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.623153758969959e-05,
+      "loss": 2.0393,
+      "step": 1017500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.6229685765419345e-05,
+      "loss": 2.0585,
+      "step": 1018000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.6227833941139095e-05,
+      "loss": 2.05,
+      "step": 1018500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.622598211685886e-05,
+      "loss": 1.9668,
+      "step": 1019000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.622413029257861e-05,
+      "loss": 2.0707,
+      "step": 1019500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.6222278468298366e-05,
+      "loss": 2.0517,
+      "step": 1020000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.6220426644018117e-05,
+      "loss": 2.1031,
+      "step": 1020500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.621857481973787e-05,
+      "loss": 2.0798,
+      "step": 1021000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.6216722995457624e-05,
+      "loss": 2.0894,
+      "step": 1021500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.621487117117738e-05,
+      "loss": 2.0638,
+      "step": 1022000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.621301934689714e-05,
+      "loss": 2.0821,
+      "step": 1022500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.621116752261689e-05,
+      "loss": 2.0366,
+      "step": 1023000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.6209315698336645e-05,
+      "loss": 2.042,
+      "step": 1023500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.6207463874056395e-05,
+      "loss": 1.9892,
+      "step": 1024000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.6205612049776145e-05,
+      "loss": 2.0839,
+      "step": 1024500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.620376022549591e-05,
+      "loss": 2.0717,
+      "step": 1025000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.620190840121566e-05,
+      "loss": 2.0489,
+      "step": 1025500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.6200056576935416e-05,
+      "loss": 2.1028,
+      "step": 1026000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.6198204752655166e-05,
+      "loss": 2.0154,
+      "step": 1026500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.619635292837492e-05,
+      "loss": 2.1096,
+      "step": 1027000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.619450110409467e-05,
+      "loss": 2.0815,
+      "step": 1027500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.619264927981442e-05,
+      "loss": 2.0281,
+      "step": 1028000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.619079745553419e-05,
+      "loss": 2.1375,
+      "step": 1028500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.618894563125394e-05,
+      "loss": 2.0621,
+      "step": 1029000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6187093806973694e-05,
+      "loss": 1.9994,
+      "step": 1029500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6185241982693444e-05,
+      "loss": 2.0178,
+      "step": 1030000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6183390158413194e-05,
+      "loss": 2.0268,
+      "step": 1030500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.618153833413295e-05,
+      "loss": 2.0808,
+      "step": 1031000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.617968650985271e-05,
+      "loss": 2.0078,
+      "step": 1031500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6177834685572465e-05,
+      "loss": 2.0562,
+      "step": 1032000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6175982861292215e-05,
+      "loss": 2.0047,
+      "step": 1032500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.617413103701197e-05,
+      "loss": 2.0494,
+      "step": 1033000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.617227921273172e-05,
+      "loss": 2.0175,
+      "step": 1033500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.617042738845147e-05,
+      "loss": 2.0532,
+      "step": 1034000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.616857556417123e-05,
+      "loss": 2.085,
+      "step": 1034500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.616672373989099e-05,
+      "loss": 2.0665,
+      "step": 1035000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6164871915610744e-05,
+      "loss": 1.9793,
+      "step": 1035500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6163020091330494e-05,
+      "loss": 2.0211,
+      "step": 1036000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.616116826705025e-05,
+      "loss": 2.0065,
+      "step": 1036500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.615931644277e-05,
+      "loss": 2.0266,
+      "step": 1037000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.615746461848975e-05,
+      "loss": 2.049,
+      "step": 1037500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6155612794209515e-05,
+      "loss": 2.0853,
+      "step": 1038000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6153760969929265e-05,
+      "loss": 2.0657,
+      "step": 1038500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.615190914564902e-05,
+      "loss": 2.0305,
+      "step": 1039000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.615005732136877e-05,
+      "loss": 2.0104,
+      "step": 1039500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.614820549708853e-05,
+      "loss": 2.0452,
+      "step": 1040000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.614635367280828e-05,
+      "loss": 2.0034,
+      "step": 1040500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6144501848528036e-05,
+      "loss": 2.0114,
+      "step": 1041000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.614265002424779e-05,
+      "loss": 2.0384,
+      "step": 1041500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.614079819996754e-05,
+      "loss": 2.0609,
+      "step": 1042000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.61389463756873e-05,
+      "loss": 2.0699,
+      "step": 1042500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.613709455140705e-05,
+      "loss": 2.0296,
+      "step": 1043000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.613524272712681e-05,
+      "loss": 2.0341,
+      "step": 1043500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.613339090284656e-05,
+      "loss": 2.0044,
+      "step": 1044000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6131539078566314e-05,
+      "loss": 2.0148,
+      "step": 1044500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.612968725428607e-05,
+      "loss": 1.9761,
+      "step": 1045000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.612783543000582e-05,
+      "loss": 1.9635,
+      "step": 1045500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.612598360572558e-05,
+      "loss": 2.0579,
+      "step": 1046000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.612413178144533e-05,
+      "loss": 2.0461,
+      "step": 1046500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.612227995716508e-05,
+      "loss": 2.0131,
+      "step": 1047000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.612042813288484e-05,
+      "loss": 2.0641,
+      "step": 1047500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.611857630860459e-05,
+      "loss": 2.0182,
+      "step": 1048000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.611672448432435e-05,
+      "loss": 2.0754,
+      "step": 1048500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.61148726600441e-05,
+      "loss": 2.0843,
+      "step": 1049000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.611302083576386e-05,
+      "loss": 2.0939,
+      "step": 1049500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.611116901148361e-05,
+      "loss": 2.0775,
+      "step": 1050000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6109317187203364e-05,
+      "loss": 2.0064,
+      "step": 1050500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.610746536292312e-05,
+      "loss": 1.9742,
+      "step": 1051000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.610561353864287e-05,
+      "loss": 2.0307,
+      "step": 1051500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.610376171436263e-05,
+      "loss": 2.0393,
+      "step": 1052000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.610190989008238e-05,
+      "loss": 2.0434,
+      "step": 1052500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6100058065802135e-05,
+      "loss": 1.9794,
+      "step": 1053000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6098206241521885e-05,
+      "loss": 2.0144,
+      "step": 1053500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.609635441724164e-05,
+      "loss": 2.1097,
+      "step": 1054000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.60945025929614e-05,
+      "loss": 2.0711,
+      "step": 1054500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.609265076868115e-05,
+      "loss": 2.0481,
+      "step": 1055000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6090798944400906e-05,
+      "loss": 1.982,
+      "step": 1055500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6088947120120656e-05,
+      "loss": 1.9887,
+      "step": 1056000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.608709529584041e-05,
+      "loss": 2.0213,
+      "step": 1056500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.608524347156017e-05,
+      "loss": 2.0595,
+      "step": 1057000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.608339164727992e-05,
+      "loss": 2.0202,
+      "step": 1057500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.608153982299968e-05,
+      "loss": 2.0724,
+      "step": 1058000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.607968799871943e-05,
+      "loss": 2.0174,
+      "step": 1058500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6077836174439185e-05,
+      "loss": 2.0645,
+      "step": 1059000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6075984350158935e-05,
+      "loss": 2.0285,
+      "step": 1059500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6074132525878685e-05,
+      "loss": 2.0194,
+      "step": 1060000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.607228070159845e-05,
+      "loss": 2.0411,
+      "step": 1060500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.60704288773182e-05,
+      "loss": 2.0141,
+      "step": 1061000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6068577053037956e-05,
+      "loss": 2.0023,
+      "step": 1061500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6066725228757706e-05,
+      "loss": 2.0146,
+      "step": 1062000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.606487340447746e-05,
+      "loss": 2.0538,
+      "step": 1062500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.606302158019721e-05,
+      "loss": 2.0315,
+      "step": 1063000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.606116975591697e-05,
+      "loss": 2.0389,
+      "step": 1063500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.605931793163673e-05,
+      "loss": 2.0816,
+      "step": 1064000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.605746610735648e-05,
+      "loss": 2.0201,
+      "step": 1064500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6055614283076234e-05,
+      "loss": 2.0379,
+      "step": 1065000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6053762458795984e-05,
+      "loss": 1.9517,
+      "step": 1065500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.605191063451574e-05,
+      "loss": 2.0394,
+      "step": 1066000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.60500588102355e-05,
+      "loss": 2.0043,
+      "step": 1066500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.604820698595525e-05,
+      "loss": 2.0323,
+      "step": 1067000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6046355161675005e-05,
+      "loss": 2.0322,
+      "step": 1067500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6044503337394755e-05,
+      "loss": 1.9481,
+      "step": 1068000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.604265151311451e-05,
+      "loss": 2.0834,
+      "step": 1068500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.604079968883426e-05,
+      "loss": 2.0398,
+      "step": 1069000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.603894786455402e-05,
+      "loss": 2.0324,
+      "step": 1069500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6037096040273776e-05,
+      "loss": 2.0576,
+      "step": 1070000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6035244215993527e-05,
+      "loss": 1.9908,
+      "step": 1070500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6033392391713284e-05,
+      "loss": 2.004,
+      "step": 1071000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6031540567433034e-05,
+      "loss": 2.0846,
+      "step": 1071500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.602968874315279e-05,
+      "loss": 1.9894,
+      "step": 1072000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.602783691887254e-05,
+      "loss": 2.0524,
+      "step": 1072500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.60259850945923e-05,
+      "loss": 2.0706,
+      "step": 1073000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.6024133270312055e-05,
+      "loss": 1.9906,
+      "step": 1073500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.6022281446031805e-05,
+      "loss": 1.9503,
+      "step": 1074000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.602042962175156e-05,
+      "loss": 2.0304,
+      "step": 1074500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.601857779747131e-05,
+      "loss": 2.0718,
+      "step": 1075000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.601672597319107e-05,
+      "loss": 2.036,
+      "step": 1075500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.6014874148910826e-05,
+      "loss": 2.0283,
+      "step": 1076000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.6013022324630576e-05,
+      "loss": 2.0526,
+      "step": 1076500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.601117050035033e-05,
+      "loss": 2.0656,
+      "step": 1077000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.600931867607008e-05,
+      "loss": 1.9943,
+      "step": 1077500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.600746685178984e-05,
+      "loss": 2.0667,
+      "step": 1078000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.600561502750959e-05,
+      "loss": 2.0688,
+      "step": 1078500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.600376320322935e-05,
+      "loss": 2.0692,
+      "step": 1079000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.6001911378949104e-05,
+      "loss": 2.0125,
+      "step": 1079500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.6000059554668854e-05,
+      "loss": 2.0157,
+      "step": 1080000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.599820773038861e-05,
+      "loss": 2.0318,
+      "step": 1080500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.599635590610836e-05,
+      "loss": 2.0082,
+      "step": 1081000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.599450408182812e-05,
+      "loss": 2.0442,
+      "step": 1081500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.599265225754787e-05,
+      "loss": 1.9515,
+      "step": 1082000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.5990800433267626e-05,
+      "loss": 2.03,
+      "step": 1082500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.598894860898738e-05,
+      "loss": 2.007,
+      "step": 1083000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.598709678470713e-05,
+      "loss": 2.0143,
+      "step": 1083500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.598524496042689e-05,
+      "loss": 2.045,
+      "step": 1084000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.598339313614664e-05,
+      "loss": 2.0811,
+      "step": 1084500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.59815413118664e-05,
+      "loss": 2.0096,
+      "step": 1085000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.597968948758615e-05,
+      "loss": 1.973,
+      "step": 1085500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.5977837663305904e-05,
+      "loss": 1.9465,
+      "step": 1086000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.597598583902566e-05,
+      "loss": 1.9726,
+      "step": 1086500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.597413401474541e-05,
+      "loss": 2.0709,
+      "step": 1087000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.597228219046517e-05,
+      "loss": 2.0036,
+      "step": 1087500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.597043036618492e-05,
+      "loss": 1.9962,
+      "step": 1088000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.5968578541904675e-05,
+      "loss": 1.9791,
+      "step": 1088500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.596672671762443e-05,
+      "loss": 2.0552,
+      "step": 1089000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.596487489334418e-05,
+      "loss": 1.9897,
+      "step": 1089500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.596302306906394e-05,
+      "loss": 2.0258,
+      "step": 1090000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.596117124478369e-05,
+      "loss": 2.01,
+      "step": 1090500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.5959319420503446e-05,
+      "loss": 2.0384,
+      "step": 1091000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.5957467596223196e-05,
+      "loss": 2.0046,
+      "step": 1091500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.595561577194295e-05,
+      "loss": 2.0059,
+      "step": 1092000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.595376394766271e-05,
+      "loss": 2.0942,
+      "step": 1092500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.595191212338246e-05,
+      "loss": 2.0283,
+      "step": 1093000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.595006029910222e-05,
+      "loss": 2.0263,
+      "step": 1093500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.594820847482197e-05,
+      "loss": 2.0925,
+      "step": 1094000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.5946356650541724e-05,
+      "loss": 1.9677,
+      "step": 1094500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.5944504826261475e-05,
+      "loss": 2.0579,
+      "step": 1095000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.594265300198123e-05,
+      "loss": 2.0022,
+      "step": 1095500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.594080117770099e-05,
+      "loss": 2.0642,
+      "step": 1096000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.593894935342074e-05,
+      "loss": 2.0249,
+      "step": 1096500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.5937097529140496e-05,
+      "loss": 2.0108,
+      "step": 1097000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.5935245704860246e-05,
+      "loss": 2.0411,
+      "step": 1097500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.593339388058e-05,
+      "loss": 2.0254,
+      "step": 1098000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.593154205629976e-05,
+      "loss": 2.0359,
+      "step": 1098500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.592969023201951e-05,
+      "loss": 1.9545,
+      "step": 1099000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.592783840773927e-05,
+      "loss": 2.0384,
+      "step": 1099500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.592598658345902e-05,
+      "loss": 2.0318,
+      "step": 1100000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.5924134759178774e-05,
+      "loss": 2.0491,
+      "step": 1100500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.5922282934898524e-05,
+      "loss": 2.0369,
+      "step": 1101000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.592043111061828e-05,
+      "loss": 1.9809,
+      "step": 1101500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.591857928633804e-05,
+      "loss": 2.0405,
+      "step": 1102000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.591672746205779e-05,
+      "loss": 2.0773,
+      "step": 1102500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.5914875637777545e-05,
+      "loss": 1.9896,
+      "step": 1103000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.5913023813497295e-05,
+      "loss": 2.0292,
+      "step": 1103500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.591117198921705e-05,
+      "loss": 2.0313,
+      "step": 1104000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.59093201649368e-05,
+      "loss": 2.051,
+      "step": 1104500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.590746834065656e-05,
+      "loss": 2.0128,
+      "step": 1105000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.5905616516376316e-05,
+      "loss": 2.006,
+      "step": 1105500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.5903764692096066e-05,
+      "loss": 2.0581,
+      "step": 1106000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.5901912867815823e-05,
+      "loss": 1.9873,
+      "step": 1106500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.5900061043535574e-05,
+      "loss": 2.0416,
+      "step": 1107000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.589820921925533e-05,
+      "loss": 2.0782,
+      "step": 1107500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.589635739497509e-05,
+      "loss": 2.0184,
+      "step": 1108000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.589450557069484e-05,
+      "loss": 2.0644,
+      "step": 1108500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.5892653746414595e-05,
+      "loss": 1.9821,
+      "step": 1109000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.5890801922134345e-05,
+      "loss": 2.0233,
+      "step": 1109500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.58889500978541e-05,
+      "loss": 2.0325,
+      "step": 1110000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.588709827357385e-05,
+      "loss": 2.0265,
+      "step": 1110500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.588524644929361e-05,
+      "loss": 2.0334,
+      "step": 1111000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.5883394625013366e-05,
+      "loss": 2.0002,
+      "step": 1111500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.5881542800733116e-05,
+      "loss": 1.9944,
+      "step": 1112000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.587969097645287e-05,
+      "loss": 2.071,
+      "step": 1112500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.587783915217262e-05,
+      "loss": 1.9979,
+      "step": 1113000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.587598732789238e-05,
+      "loss": 2.0582,
+      "step": 1113500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.587413550361213e-05,
+      "loss": 1.9804,
+      "step": 1114000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.587228367933189e-05,
+      "loss": 2.0089,
+      "step": 1114500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.5870431855051644e-05,
+      "loss": 2.0825,
+      "step": 1115000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.5868580030771394e-05,
+      "loss": 2.0297,
+      "step": 1115500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.586672820649115e-05,
+      "loss": 2.0135,
+      "step": 1116000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.58648763822109e-05,
+      "loss": 2.0259,
+      "step": 1116500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.586302455793066e-05,
+      "loss": 2.095,
+      "step": 1117000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.5861172733650415e-05,
+      "loss": 1.9724,
+      "step": 1117500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.5859320909370165e-05,
+      "loss": 1.9995,
+      "step": 1118000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.585746908508992e-05,
+      "loss": 2.0823,
+      "step": 1118500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.585561726080967e-05,
+      "loss": 1.9989,
+      "step": 1119000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.585376543652943e-05,
+      "loss": 1.9797,
+      "step": 1119500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.585191361224918e-05,
+      "loss": 2.0385,
+      "step": 1120000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.585006178796894e-05,
+      "loss": 2.088,
+      "step": 1120500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.5848209963688694e-05,
+      "loss": 2.0721,
+      "step": 1121000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.5846358139408444e-05,
+      "loss": 1.9574,
+      "step": 1121500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.58445063151282e-05,
+      "loss": 2.036,
+      "step": 1122000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.584265449084795e-05,
+      "loss": 1.9909,
+      "step": 1122500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.584080266656771e-05,
+      "loss": 1.9879,
+      "step": 1123000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.583895084228746e-05,
+      "loss": 2.089,
+      "step": 1123500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.5837099018007215e-05,
+      "loss": 1.9891,
+      "step": 1124000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.583524719372697e-05,
+      "loss": 2.0456,
+      "step": 1124500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.583339536944672e-05,
+      "loss": 1.9962,
+      "step": 1125000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.583154354516648e-05,
+      "loss": 2.003,
+      "step": 1125500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.582969172088623e-05,
+      "loss": 2.0341,
+      "step": 1126000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.5827839896605986e-05,
+      "loss": 2.0192,
+      "step": 1126500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.582598807232574e-05,
+      "loss": 1.9658,
+      "step": 1127000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.582413624804549e-05,
+      "loss": 1.9976,
+      "step": 1127500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.582228442376525e-05,
+      "loss": 2.0131,
+      "step": 1128000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.5820432599485e-05,
+      "loss": 2.0462,
+      "step": 1128500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.581858077520476e-05,
+      "loss": 2.0675,
+      "step": 1129000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.581672895092451e-05,
+      "loss": 2.0386,
+      "step": 1129500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.5814877126644264e-05,
+      "loss": 2.0208,
+      "step": 1130000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.581302530236402e-05,
+      "loss": 1.9804,
+      "step": 1130500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.581117347808377e-05,
+      "loss": 1.9951,
+      "step": 1131000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.580932165380353e-05,
+      "loss": 1.9944,
+      "step": 1131500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.580746982952328e-05,
+      "loss": 1.9934,
+      "step": 1132000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.5805618005243036e-05,
+      "loss": 2.0437,
+      "step": 1132500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.5803766180962786e-05,
+      "loss": 2.0452,
+      "step": 1133000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.580191435668254e-05,
+      "loss": 2.0159,
+      "step": 1133500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.58000625324023e-05,
+      "loss": 1.9983,
+      "step": 1134000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.579821070812205e-05,
+      "loss": 2.0513,
+      "step": 1134500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.579635888384181e-05,
+      "loss": 2.0089,
+      "step": 1135000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.579450705956156e-05,
+      "loss": 2.0429,
+      "step": 1135500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.5792655235281314e-05,
+      "loss": 2.0428,
+      "step": 1136000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.5790803411001064e-05,
+      "loss": 2.0082,
+      "step": 1136500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.578895158672082e-05,
+      "loss": 2.0299,
+      "step": 1137000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.578709976244058e-05,
+      "loss": 1.9961,
+      "step": 1137500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.578524793816033e-05,
+      "loss": 2.04,
+      "step": 1138000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.5783396113880085e-05,
+      "loss": 2.0125,
+      "step": 1138500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.5781544289599835e-05,
+      "loss": 2.0359,
+      "step": 1139000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.577969246531959e-05,
+      "loss": 2.0513,
+      "step": 1139500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.577784064103935e-05,
+      "loss": 1.9749,
+      "step": 1140000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.57759888167591e-05,
+      "loss": 1.9861,
+      "step": 1140500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.5774136992478856e-05,
+      "loss": 2.0236,
+      "step": 1141000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.5772285168198606e-05,
+      "loss": 2.0338,
+      "step": 1141500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.577043334391836e-05,
+      "loss": 2.0213,
+      "step": 1142000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.5768581519638114e-05,
+      "loss": 2.0148,
+      "step": 1142500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.576672969535788e-05,
+      "loss": 1.9878,
+      "step": 1143000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.576487787107763e-05,
+      "loss": 1.9863,
+      "step": 1143500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.576302604679738e-05,
+      "loss": 2.0854,
+      "step": 1144000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.5761174222517135e-05,
+      "loss": 2.0522,
+      "step": 1144500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.5759322398236885e-05,
+      "loss": 1.9886,
+      "step": 1145000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.575747057395664e-05,
+      "loss": 2.0324,
+      "step": 1145500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.575561874967639e-05,
+      "loss": 2.0437,
+      "step": 1146000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.575376692539615e-05,
+      "loss": 1.9871,
+      "step": 1146500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.5751915101115906e-05,
+      "loss": 2.0339,
+      "step": 1147000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.5750063276835656e-05,
+      "loss": 1.9999,
+      "step": 1147500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.574821145255541e-05,
+      "loss": 2.0565,
+      "step": 1148000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.574635962827516e-05,
+      "loss": 1.9668,
+      "step": 1148500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.574450780399492e-05,
+      "loss": 1.9387,
+      "step": 1149000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.574265597971468e-05,
+      "loss": 2.0382,
+      "step": 1149500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.574080415543443e-05,
+      "loss": 2.018,
+      "step": 1150000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.5738952331154184e-05,
+      "loss": 2.0086,
+      "step": 1150500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.5737100506873934e-05,
+      "loss": 2.0138,
+      "step": 1151000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.573524868259369e-05,
+      "loss": 2.0495,
+      "step": 1151500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.573339685831344e-05,
+      "loss": 2.0011,
+      "step": 1152000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.5731545034033205e-05,
+      "loss": 2.0153,
+      "step": 1152500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.5729693209752955e-05,
+      "loss": 1.9924,
+      "step": 1153000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.5727841385472705e-05,
+      "loss": 2.0114,
+      "step": 1153500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.572598956119246e-05,
+      "loss": 1.9851,
+      "step": 1154000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.572413773691221e-05,
+      "loss": 2.0088,
+      "step": 1154500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.572228591263197e-05,
+      "loss": 2.0583,
+      "step": 1155000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.572043408835172e-05,
+      "loss": 2.0373,
+      "step": 1155500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.571858226407148e-05,
+      "loss": 2.0252,
+      "step": 1156000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.5716730439791234e-05,
+      "loss": 2.0274,
+      "step": 1156500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.5714878615510984e-05,
+      "loss": 2.0626,
+      "step": 1157000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.571302679123074e-05,
+      "loss": 2.0646,
+      "step": 1157500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.571117496695049e-05,
+      "loss": 2.0683,
+      "step": 1158000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.570932314267025e-05,
+      "loss": 2.0225,
+      "step": 1158500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.5707471318390005e-05,
+      "loss": 1.9791,
+      "step": 1159000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.5705619494109755e-05,
+      "loss": 2.0097,
+      "step": 1159500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.570376766982951e-05,
+      "loss": 2.01,
+      "step": 1160000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.570191584554926e-05,
+      "loss": 1.9769,
+      "step": 1160500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.570006402126902e-05,
+      "loss": 2.0588,
+      "step": 1161000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.569821219698877e-05,
+      "loss": 2.0773,
+      "step": 1161500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.5696360372708526e-05,
+      "loss": 2.0109,
+      "step": 1162000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.569450854842828e-05,
+      "loss": 2.0382,
+      "step": 1162500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.569265672414803e-05,
+      "loss": 2.022,
+      "step": 1163000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.569080489986779e-05,
+      "loss": 1.9878,
+      "step": 1163500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.568895307558754e-05,
+      "loss": 1.9953,
+      "step": 1164000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.56871012513073e-05,
+      "loss": 2.0324,
+      "step": 1164500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.568524942702705e-05,
+      "loss": 1.9929,
+      "step": 1165000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.568339760274681e-05,
+      "loss": 2.0366,
+      "step": 1165500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.568154577846656e-05,
+      "loss": 1.9813,
+      "step": 1166000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.567969395418631e-05,
+      "loss": 1.9828,
+      "step": 1166500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.567784212990607e-05,
+      "loss": 2.0847,
+      "step": 1167000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.567599030562582e-05,
+      "loss": 1.9774,
+      "step": 1167500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.5674138481345576e-05,
+      "loss": 2.0894,
+      "step": 1168000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.567228665706533e-05,
+      "loss": 1.9716,
+      "step": 1168500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.567043483278509e-05,
+      "loss": 2.0528,
+      "step": 1169000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.566858300850484e-05,
+      "loss": 1.9761,
+      "step": 1169500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.566673118422459e-05,
+      "loss": 2.0017,
+      "step": 1170000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.566487935994435e-05,
+      "loss": 2.0189,
+      "step": 1170500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.56630275356641e-05,
+      "loss": 1.9752,
+      "step": 1171000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.5661175711383854e-05,
+      "loss": 1.9397,
+      "step": 1171500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.565932388710361e-05,
+      "loss": 2.0487,
+      "step": 1172000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.565747206282336e-05,
+      "loss": 2.0155,
+      "step": 1172500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.565562023854312e-05,
+      "loss": 1.9776,
+      "step": 1173000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.565376841426287e-05,
+      "loss": 2.0014,
+      "step": 1173500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.5651916589982625e-05,
+      "loss": 1.9584,
+      "step": 1174000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.5650064765702375e-05,
+      "loss": 1.9828,
+      "step": 1174500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.564821294142214e-05,
+      "loss": 2.0071,
+      "step": 1175000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.564636111714189e-05,
+      "loss": 1.9823,
+      "step": 1175500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.564450929286164e-05,
+      "loss": 1.9643,
+      "step": 1176000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.5642657468581396e-05,
+      "loss": 2.0379,
+      "step": 1176500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.5640805644301146e-05,
+      "loss": 2.0049,
+      "step": 1177000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.56389538200209e-05,
+      "loss": 2.0408,
+      "step": 1177500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.563710199574066e-05,
+      "loss": 2.0529,
+      "step": 1178000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.563525017146042e-05,
+      "loss": 2.0234,
+      "step": 1178500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.563339834718017e-05,
+      "loss": 2.046,
+      "step": 1179000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.563154652289992e-05,
+      "loss": 2.0445,
+      "step": 1179500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.5629694698619674e-05,
+      "loss": 2.0227,
+      "step": 1180000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.5627842874339425e-05,
+      "loss": 2.0697,
+      "step": 1180500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.562599105005918e-05,
+      "loss": 2.0044,
+      "step": 1181000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.562413922577894e-05,
+      "loss": 1.9942,
+      "step": 1181500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.5622287401498695e-05,
+      "loss": 1.9858,
+      "step": 1182000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.5620435577218446e-05,
+      "loss": 2.0483,
+      "step": 1182500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.5618583752938196e-05,
+      "loss": 2.1029,
+      "step": 1183000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.561673192865795e-05,
+      "loss": 2.034,
+      "step": 1183500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.56148801043777e-05,
+      "loss": 2.0263,
+      "step": 1184000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.561302828009747e-05,
+      "loss": 2.0165,
+      "step": 1184500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.561117645581722e-05,
+      "loss": 2.1276,
+      "step": 1185000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.5609324631536974e-05,
+      "loss": 2.0112,
+      "step": 1185500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.5607472807256724e-05,
+      "loss": 2.0514,
+      "step": 1186000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.5605620982976474e-05,
+      "loss": 2.0065,
+      "step": 1186500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.560376915869623e-05,
+      "loss": 2.0716,
+      "step": 1187000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.560191733441598e-05,
+      "loss": 2.0466,
+      "step": 1187500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.5600065510135745e-05,
+      "loss": 2.0075,
+      "step": 1188000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.5598213685855495e-05,
+      "loss": 1.9578,
+      "step": 1188500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.5596361861575245e-05,
+      "loss": 2.063,
+      "step": 1189000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.5594510037295e-05,
+      "loss": 2.0231,
+      "step": 1189500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.559265821301475e-05,
+      "loss": 2.0299,
+      "step": 1190000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.559080638873451e-05,
+      "loss": 2.1112,
+      "step": 1190500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.5588954564454266e-05,
+      "loss": 2.0467,
+      "step": 1191000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.558710274017402e-05,
+      "loss": 1.9726,
+      "step": 1191500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.5585250915893773e-05,
+      "loss": 2.0669,
+      "step": 1192000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.5583399091613524e-05,
+      "loss": 1.9855,
+      "step": 1192500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.558154726733328e-05,
+      "loss": 2.0615,
+      "step": 1193000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.557969544305303e-05,
+      "loss": 2.026,
+      "step": 1193500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.5577843618772794e-05,
+      "loss": 1.9829,
+      "step": 1194000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.5575991794492545e-05,
+      "loss": 1.9632,
+      "step": 1194500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.55741399702123e-05,
+      "loss": 2.0398,
+      "step": 1195000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.557228814593205e-05,
+      "loss": 2.0325,
+      "step": 1195500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.55704363216518e-05,
+      "loss": 2.0186,
+      "step": 1196000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.556858449737156e-05,
+      "loss": 1.9944,
+      "step": 1196500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.556673267309131e-05,
+      "loss": 2.0366,
+      "step": 1197000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.556488084881107e-05,
+      "loss": 1.9559,
+      "step": 1197500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.556302902453082e-05,
+      "loss": 2.0433,
+      "step": 1198000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.556117720025058e-05,
+      "loss": 2.0049,
+      "step": 1198500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.555932537597033e-05,
+      "loss": 1.9842,
+      "step": 1199000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.555747355169008e-05,
+      "loss": 1.9653,
+      "step": 1199500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.555562172740984e-05,
+      "loss": 2.0843,
+      "step": 1200000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.5553769903129594e-05,
+      "loss": 2.0587,
+      "step": 1200500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.555191807884935e-05,
+      "loss": 2.0026,
+      "step": 1201000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.55500662545691e-05,
+      "loss": 2.0253,
+      "step": 1201500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.554821443028885e-05,
+      "loss": 1.9794,
+      "step": 1202000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.554636260600861e-05,
+      "loss": 2.0672,
+      "step": 1202500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.554451078172836e-05,
+      "loss": 1.9861,
+      "step": 1203000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.554265895744812e-05,
+      "loss": 2.0267,
+      "step": 1203500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.554080713316787e-05,
+      "loss": 2.0156,
+      "step": 1204000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.553895530888763e-05,
+      "loss": 1.9506,
+      "step": 1204500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.553710348460738e-05,
+      "loss": 2.0643,
+      "step": 1205000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.553525166032713e-05,
+      "loss": 1.9997,
+      "step": 1205500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.553339983604689e-05,
+      "loss": 2.0217,
+      "step": 1206000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.553154801176664e-05,
+      "loss": 2.061,
+      "step": 1206500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.55296961874864e-05,
+      "loss": 1.9712,
+      "step": 1207000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.552784436320615e-05,
+      "loss": 1.9848,
+      "step": 1207500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.552599253892591e-05,
+      "loss": 1.9508,
+      "step": 1208000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.552414071464566e-05,
+      "loss": 2.0733,
+      "step": 1208500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.552228889036541e-05,
+      "loss": 2.0766,
+      "step": 1209000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5520437066085165e-05,
+      "loss": 2.0268,
+      "step": 1209500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.551858524180492e-05,
+      "loss": 2.0696,
+      "step": 1210000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.551673341752468e-05,
+      "loss": 2.0019,
+      "step": 1210500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.551488159324443e-05,
+      "loss": 2.0083,
+      "step": 1211000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5513029768964186e-05,
+      "loss": 2.0026,
+      "step": 1211500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5511177944683936e-05,
+      "loss": 2.0136,
+      "step": 1212000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5509326120403686e-05,
+      "loss": 2.0361,
+      "step": 1212500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.550747429612344e-05,
+      "loss": 2.0245,
+      "step": 1213000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.55056224718432e-05,
+      "loss": 2.0176,
+      "step": 1213500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.550377064756296e-05,
+      "loss": 2.0951,
+      "step": 1214000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.550191882328271e-05,
+      "loss": 2.0029,
+      "step": 1214500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.550006699900246e-05,
+      "loss": 2.0388,
+      "step": 1215000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5498215174722214e-05,
+      "loss": 1.9956,
+      "step": 1215500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5496363350441965e-05,
+      "loss": 2.0941,
+      "step": 1216000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.549451152616173e-05,
+      "loss": 2.0493,
+      "step": 1216500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.549265970188148e-05,
+      "loss": 1.9668,
+      "step": 1217000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5490807877601235e-05,
+      "loss": 2.0112,
+      "step": 1217500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5488956053320986e-05,
+      "loss": 2.01,
+      "step": 1218000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5487104229040736e-05,
+      "loss": 2.0454,
+      "step": 1218500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.548525240476049e-05,
+      "loss": 1.9673,
+      "step": 1219000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.548340058048025e-05,
+      "loss": 2.0913,
+      "step": 1219500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5481548756200007e-05,
+      "loss": 2.0534,
+      "step": 1220000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.547969693191976e-05,
+      "loss": 1.996,
+      "step": 1220500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5477845107639514e-05,
+      "loss": 2.0159,
+      "step": 1221000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5475993283359264e-05,
+      "loss": 2.0308,
+      "step": 1221500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5474141459079014e-05,
+      "loss": 2.0111,
+      "step": 1222000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.547228963479877e-05,
+      "loss": 1.9588,
+      "step": 1222500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.547043781051853e-05,
+      "loss": 1.9907,
+      "step": 1223000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5468585986238285e-05,
+      "loss": 1.9381,
+      "step": 1223500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5466734161958035e-05,
+      "loss": 2.0461,
+      "step": 1224000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.546488233767779e-05,
+      "loss": 2.0119,
+      "step": 1224500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.546303051339754e-05,
+      "loss": 2.0473,
+      "step": 1225000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.546117868911729e-05,
+      "loss": 2.0373,
+      "step": 1225500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5459326864837056e-05,
+      "loss": 1.948,
+      "step": 1226000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5457475040556806e-05,
+      "loss": 1.9465,
+      "step": 1226500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.545562321627656e-05,
+      "loss": 2.0531,
+      "step": 1227000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.545377139199631e-05,
+      "loss": 2.0145,
+      "step": 1227500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.545191956771607e-05,
+      "loss": 1.9744,
+      "step": 1228000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.545006774343582e-05,
+      "loss": 2.0182,
+      "step": 1228500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.544821591915558e-05,
+      "loss": 2.0322,
+      "step": 1229000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5446364094875334e-05,
+      "loss": 2.0257,
+      "step": 1229500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5444512270595085e-05,
+      "loss": 2.0792,
+      "step": 1230000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.544266044631484e-05,
+      "loss": 1.97,
+      "step": 1230500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.544080862203459e-05,
+      "loss": 1.9821,
+      "step": 1231000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.543895679775434e-05,
+      "loss": 2.0477,
+      "step": 1231500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.54371049734741e-05,
+      "loss": 1.9921,
+      "step": 1232000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5435253149193856e-05,
+      "loss": 1.9702,
+      "step": 1232500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.543340132491361e-05,
+      "loss": 2.0535,
+      "step": 1233000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.543154950063336e-05,
+      "loss": 2.0334,
+      "step": 1233500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.542969767635312e-05,
+      "loss": 1.9955,
+      "step": 1234000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.542784585207287e-05,
+      "loss": 2.0129,
+      "step": 1234500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.542599402779262e-05,
+      "loss": 2.0197,
+      "step": 1235000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5424142203512384e-05,
+      "loss": 2.0355,
+      "step": 1235500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5422290379232134e-05,
+      "loss": 1.9437,
+      "step": 1236000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.542043855495189e-05,
+      "loss": 2.0292,
+      "step": 1236500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.541858673067164e-05,
+      "loss": 2.0148,
+      "step": 1237000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.54167349063914e-05,
+      "loss": 1.999,
+      "step": 1237500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.541488308211115e-05,
+      "loss": 1.9867,
+      "step": 1238000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.54130312578309e-05,
+      "loss": 2.0071,
+      "step": 1238500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.541117943355066e-05,
+      "loss": 1.9873,
+      "step": 1239000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.540932760927041e-05,
+      "loss": 2.0363,
+      "step": 1239500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.540747578499017e-05,
+      "loss": 2.0003,
+      "step": 1240000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.540562396070992e-05,
+      "loss": 2.0571,
+      "step": 1240500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5403772136429676e-05,
+      "loss": 2.04,
+      "step": 1241000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5401920312149427e-05,
+      "loss": 2.0291,
+      "step": 1241500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5400068487869183e-05,
+      "loss": 2.0359,
+      "step": 1242000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.539821666358894e-05,
+      "loss": 1.9683,
+      "step": 1242500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.539636483930869e-05,
+      "loss": 2.0222,
+      "step": 1243000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.539451301502845e-05,
+      "loss": 1.9784,
+      "step": 1243500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.53926611907482e-05,
+      "loss": 2.0293,
+      "step": 1244000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.539080936646795e-05,
+      "loss": 2.0801,
+      "step": 1244500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.538895754218771e-05,
+      "loss": 2.0896,
+      "step": 1245000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.538710571790746e-05,
+      "loss": 2.0129,
+      "step": 1245500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.538525389362722e-05,
+      "loss": 1.9908,
+      "step": 1246000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.538340206934697e-05,
+      "loss": 2.0363,
+      "step": 1246500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5381550245066726e-05,
+      "loss": 2.009,
+      "step": 1247000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5379698420786476e-05,
+      "loss": 2.0021,
+      "step": 1247500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5377846596506226e-05,
+      "loss": 2.0504,
+      "step": 1248000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.537599477222599e-05,
+      "loss": 1.942,
+      "step": 1248500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.537414294794574e-05,
+      "loss": 2.0592,
+      "step": 1249000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.53722911236655e-05,
+      "loss": 1.9323,
+      "step": 1249500
+    }
+  ],
+  "max_steps": 13500201,
+  "num_train_epochs": 3,
+  "total_flos": 1.594877582673861e+17,
+  "trial_name": null,
+  "trial_params": null
+}