diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,11353 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.3852502500985238,
+  "global_step": 944187,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9973522194226354e-05,
+      "loss": 1.5613,
+      "step": 500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.994704438845271e-05,
+      "loss": 1.4492,
+      "step": 1000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9920566582679065e-05,
+      "loss": 1.3946,
+      "step": 1500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9894088776905424e-05,
+      "loss": 1.3436,
+      "step": 2000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9867610971131776e-05,
+      "loss": 1.3033,
+      "step": 2500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9841133165358134e-05,
+      "loss": 1.2823,
+      "step": 3000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9814655359584486e-05,
+      "loss": 1.2504,
+      "step": 3500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.978817755381085e-05,
+      "loss": 1.2289,
+      "step": 4000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9761699748037204e-05,
+      "loss": 1.2179,
+      "step": 4500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.973522194226356e-05,
+      "loss": 1.1966,
+      "step": 5000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9708744136489914e-05,
+      "loss": 1.1812,
+      "step": 5500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.968226633071627e-05,
+      "loss": 1.1723,
+      "step": 6000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9655788524942625e-05,
+      "loss": 1.1612,
+      "step": 6500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9629310719168984e-05,
+      "loss": 1.1464,
+      "step": 7000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9602832913395336e-05,
+      "loss": 1.1321,
+      "step": 7500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9576355107621694e-05,
+      "loss": 1.1153,
+      "step": 8000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9549877301848046e-05,
+      "loss": 1.113,
+      "step": 8500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9523399496074405e-05,
+      "loss": 1.1049,
+      "step": 9000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.949692169030076e-05,
+      "loss": 1.0948,
+      "step": 9500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9470443884527116e-05,
+      "loss": 1.0879,
+      "step": 10000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.944396607875347e-05,
+      "loss": 1.0756,
+      "step": 10500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.941748827297982e-05,
+      "loss": 1.0727,
+      "step": 11000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9391010467206185e-05,
+      "loss": 1.0657,
+      "step": 11500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.936453266143254e-05,
+      "loss": 1.0599,
+      "step": 12000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9338054855658896e-05,
+      "loss": 1.0534,
+      "step": 12500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.931157704988525e-05,
+      "loss": 1.0443,
+      "step": 13000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9285099244111606e-05,
+      "loss": 1.0386,
+      "step": 13500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.925862143833796e-05,
+      "loss": 1.0342,
+      "step": 14000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.923214363256432e-05,
+      "loss": 1.0284,
+      "step": 14500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.920566582679067e-05,
+      "loss": 1.026,
+      "step": 15000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.917918802101703e-05,
+      "loss": 1.0239,
+      "step": 15500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.915271021524338e-05,
+      "loss": 1.0114,
+      "step": 16000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.912623240946974e-05,
+      "loss": 1.0092,
+      "step": 16500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.909975460369609e-05,
+      "loss": 1.0023,
+      "step": 17000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.907327679792245e-05,
+      "loss": 1.0013,
+      "step": 17500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.90467989921488e-05,
+      "loss": 0.9939,
+      "step": 18000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.902032118637516e-05,
+      "loss": 0.9892,
+      "step": 18500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.899384338060151e-05,
+      "loss": 0.9848,
+      "step": 19000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.896736557482787e-05,
+      "loss": 0.9805,
+      "step": 19500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.894088776905423e-05,
+      "loss": 0.9771,
+      "step": 20000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.891440996328058e-05,
+      "loss": 0.975,
+      "step": 20500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.888793215750694e-05,
+      "loss": 0.9714,
+      "step": 21000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.886145435173329e-05,
+      "loss": 0.9691,
+      "step": 21500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.883497654595965e-05,
+      "loss": 0.966,
+      "step": 22000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8808498740186e-05,
+      "loss": 0.9589,
+      "step": 22500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.878202093441236e-05,
+      "loss": 0.9565,
+      "step": 23000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.875554312863871e-05,
+      "loss": 0.9557,
+      "step": 23500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.872906532286507e-05,
+      "loss": 0.9497,
+      "step": 24000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.870258751709142e-05,
+      "loss": 0.9474,
+      "step": 24500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.867610971131778e-05,
+      "loss": 0.9499,
+      "step": 25000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8649631905544134e-05,
+      "loss": 0.9435,
+      "step": 25500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.862315409977049e-05,
+      "loss": 0.9402,
+      "step": 26000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8596676293996845e-05,
+      "loss": 0.9373,
+      "step": 26500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.85701984882232e-05,
+      "loss": 0.9322,
+      "step": 27000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.854372068244956e-05,
+      "loss": 0.9354,
+      "step": 27500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8517242876675914e-05,
+      "loss": 0.9274,
+      "step": 28000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.849076507090227e-05,
+      "loss": 0.9258,
+      "step": 28500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8464287265128625e-05,
+      "loss": 0.9212,
+      "step": 29000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.843780945935498e-05,
+      "loss": 0.9163,
+      "step": 29500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8411331653581335e-05,
+      "loss": 0.9171,
+      "step": 30000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8384853847807694e-05,
+      "loss": 0.9116,
+      "step": 30500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8358376042034046e-05,
+      "loss": 0.9136,
+      "step": 31000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8331898236260405e-05,
+      "loss": 0.912,
+      "step": 31500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8305420430486757e-05,
+      "loss": 0.9071,
+      "step": 32000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8278942624713115e-05,
+      "loss": 0.9079,
+      "step": 32500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.825246481893947e-05,
+      "loss": 0.9026,
+      "step": 33000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8225987013165826e-05,
+      "loss": 0.9038,
+      "step": 33500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.819950920739218e-05,
+      "loss": 0.8992,
+      "step": 34000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.817303140161854e-05,
+      "loss": 0.9037,
+      "step": 34500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8146553595844895e-05,
+      "loss": 0.9005,
+      "step": 35000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8120075790071254e-05,
+      "loss": 0.8974,
+      "step": 35500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8093597984297606e-05,
+      "loss": 0.8969,
+      "step": 36000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.806712017852396e-05,
+      "loss": 0.8885,
+      "step": 36500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8040642372750317e-05,
+      "loss": 0.8903,
+      "step": 37000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.801416456697667e-05,
+      "loss": 0.8825,
+      "step": 37500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.798768676120303e-05,
+      "loss": 0.8896,
+      "step": 38000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.796120895542938e-05,
+      "loss": 0.8818,
+      "step": 38500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.793473114965574e-05,
+      "loss": 0.8829,
+      "step": 39000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.790825334388209e-05,
+      "loss": 0.8741,
+      "step": 39500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.788177553810845e-05,
+      "loss": 0.8772,
+      "step": 40000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.78552977323348e-05,
+      "loss": 0.8818,
+      "step": 40500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.782881992656116e-05,
+      "loss": 0.8697,
+      "step": 41000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.780234212078751e-05,
+      "loss": 0.8705,
+      "step": 41500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.7775864315013877e-05,
+      "loss": 0.8708,
+      "step": 42000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.774938650924023e-05,
+      "loss": 0.8636,
+      "step": 42500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.772290870346659e-05,
+      "loss": 0.8616,
+      "step": 43000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.769643089769294e-05,
+      "loss": 0.8609,
+      "step": 43500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.76699530919193e-05,
+      "loss": 0.8722,
+      "step": 44000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.764347528614565e-05,
+      "loss": 0.8605,
+      "step": 44500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.761699748037201e-05,
+      "loss": 0.8605,
+      "step": 45000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.759051967459836e-05,
+      "loss": 0.8557,
+      "step": 45500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.756404186882471e-05,
+      "loss": 0.8568,
+      "step": 46000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.753756406305107e-05,
+      "loss": 0.8608,
+      "step": 46500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.751108625727742e-05,
+      "loss": 0.8514,
+      "step": 47000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.748460845150378e-05,
+      "loss": 0.8536,
+      "step": 47500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7458130645730134e-05,
+      "loss": 0.8505,
+      "step": 48000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.743165283995649e-05,
+      "loss": 0.8505,
+      "step": 48500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7405175034182844e-05,
+      "loss": 0.8528,
+      "step": 49000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.737869722840921e-05,
+      "loss": 0.8524,
+      "step": 49500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.735221942263556e-05,
+      "loss": 0.8457,
+      "step": 50000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.732574161686192e-05,
+      "loss": 0.8511,
+      "step": 50500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.729926381108827e-05,
+      "loss": 0.8439,
+      "step": 51000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.727278600531463e-05,
+      "loss": 0.8418,
+      "step": 51500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.724630819954098e-05,
+      "loss": 0.8436,
+      "step": 52000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.721983039376734e-05,
+      "loss": 0.8379,
+      "step": 52500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7193352587993694e-05,
+      "loss": 0.8408,
+      "step": 53000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.716687478222005e-05,
+      "loss": 0.8397,
+      "step": 53500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7140396976446404e-05,
+      "loss": 0.8347,
+      "step": 54000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.711391917067276e-05,
+      "loss": 0.8355,
+      "step": 54500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7087441364899115e-05,
+      "loss": 0.8323,
+      "step": 55000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7060963559125474e-05,
+      "loss": 0.8268,
+      "step": 55500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7034485753351826e-05,
+      "loss": 0.8309,
+      "step": 56000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.700800794757818e-05,
+      "loss": 0.829,
+      "step": 56500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6981530141804536e-05,
+      "loss": 0.8281,
+      "step": 57000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.6955052336030895e-05,
+      "loss": 0.8321,
+      "step": 57500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.6928574530257254e-05,
+      "loss": 0.826,
+      "step": 58000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.6902096724483606e-05,
+      "loss": 0.8275,
+      "step": 58500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.6875618918709964e-05,
+      "loss": 0.8241,
+      "step": 59000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.6849141112936316e-05,
+      "loss": 0.8237,
+      "step": 59500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.6822663307162675e-05,
+      "loss": 0.8237,
+      "step": 60000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.679618550138903e-05,
+      "loss": 0.8212,
+      "step": 60500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.6769707695615386e-05,
+      "loss": 0.8168,
+      "step": 61000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.674322988984174e-05,
+      "loss": 0.8216,
+      "step": 61500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6716752084068096e-05,
+      "loss": 0.8214,
+      "step": 62000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.669027427829445e-05,
+      "loss": 0.8148,
+      "step": 62500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.666379647252081e-05,
+      "loss": 0.8115,
+      "step": 63000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.663731866674716e-05,
+      "loss": 0.8154,
+      "step": 63500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.661084086097352e-05,
+      "loss": 0.821,
+      "step": 64000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.658436305519987e-05,
+      "loss": 0.8143,
+      "step": 64500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.655788524942623e-05,
+      "loss": 0.8134,
+      "step": 65000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.653140744365259e-05,
+      "loss": 0.8152,
+      "step": 65500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.650492963787894e-05,
+      "loss": 0.8131,
+      "step": 66000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.64784518321053e-05,
+      "loss": 0.8077,
+      "step": 66500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.645197402633165e-05,
+      "loss": 0.812,
+      "step": 67000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.642549622055801e-05,
+      "loss": 0.8074,
+      "step": 67500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.639901841478436e-05,
+      "loss": 0.8065,
+      "step": 68000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.637254060901072e-05,
+      "loss": 0.8081,
+      "step": 68500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.634606280323707e-05,
+      "loss": 0.8055,
+      "step": 69000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.631958499746343e-05,
+      "loss": 0.8022,
+      "step": 69500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.629310719168978e-05,
+      "loss": 0.8049,
+      "step": 70000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.626662938591614e-05,
+      "loss": 0.7996,
+      "step": 70500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.624015158014249e-05,
+      "loss": 0.8031,
+      "step": 71000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.621367377436885e-05,
+      "loss": 0.8045,
+      "step": 71500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.61871959685952e-05,
+      "loss": 0.8032,
+      "step": 72000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.616071816282156e-05,
+      "loss": 0.8005,
+      "step": 72500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.613424035704792e-05,
+      "loss": 0.799,
+      "step": 73000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.610776255127427e-05,
+      "loss": 0.7956,
+      "step": 73500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.608128474550063e-05,
+      "loss": 0.7955,
+      "step": 74000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.605480693972698e-05,
+      "loss": 0.7941,
+      "step": 74500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.602832913395334e-05,
+      "loss": 0.796,
+      "step": 75000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6001851328179693e-05,
+      "loss": 0.7914,
+      "step": 75500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.597537352240605e-05,
+      "loss": 0.7878,
+      "step": 76000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5948895716632404e-05,
+      "loss": 0.7877,
+      "step": 76500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.592241791085876e-05,
+      "loss": 0.7877,
+      "step": 77000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5895940105085115e-05,
+      "loss": 0.7898,
+      "step": 77500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5869462299311473e-05,
+      "loss": 0.7861,
+      "step": 78000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5842984493537825e-05,
+      "loss": 0.7952,
+      "step": 78500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5816506687764184e-05,
+      "loss": 0.7836,
+      "step": 79000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5790028881990536e-05,
+      "loss": 0.7911,
+      "step": 79500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.57635510762169e-05,
+      "loss": 0.7919,
+      "step": 80000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5737073270443253e-05,
+      "loss": 0.7861,
+      "step": 80500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.571059546466961e-05,
+      "loss": 0.7786,
+      "step": 81000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5684117658895964e-05,
+      "loss": 0.7834,
+      "step": 81500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5657639853122316e-05,
+      "loss": 0.7799,
+      "step": 82000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5631162047348675e-05,
+      "loss": 0.7845,
+      "step": 82500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.560468424157503e-05,
+      "loss": 0.7788,
+      "step": 83000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5578206435801385e-05,
+      "loss": 0.7782,
+      "step": 83500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.555172863002774e-05,
+      "loss": 0.7783,
+      "step": 84000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5525250824254096e-05,
+      "loss": 0.7754,
+      "step": 84500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.549877301848045e-05,
+      "loss": 0.7779,
+      "step": 85000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.547229521270681e-05,
+      "loss": 0.7782,
+      "step": 85500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.544581740693316e-05,
+      "loss": 0.7783,
+      "step": 86000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.541933960115952e-05,
+      "loss": 0.7792,
+      "step": 86500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.539286179538587e-05,
+      "loss": 0.7744,
+      "step": 87000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.536638398961223e-05,
+      "loss": 0.777,
+      "step": 87500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.533990618383859e-05,
+      "loss": 0.7714,
+      "step": 88000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5313428378064945e-05,
+      "loss": 0.7735,
+      "step": 88500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.52869505722913e-05,
+      "loss": 0.778,
+      "step": 89000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.5260472766517656e-05,
+      "loss": 0.7746,
+      "step": 89500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.523399496074401e-05,
+      "loss": 0.7673,
+      "step": 90000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.520751715497037e-05,
+      "loss": 0.7761,
+      "step": 90500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.518103934919672e-05,
+      "loss": 0.7713,
+      "step": 91000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.515456154342308e-05,
+      "loss": 0.7754,
+      "step": 91500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.512808373764943e-05,
+      "loss": 0.7736,
+      "step": 92000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.510160593187578e-05,
+      "loss": 0.7704,
+      "step": 92500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.507512812610214e-05,
+      "loss": 0.772,
+      "step": 93000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.504865032032849e-05,
+      "loss": 0.7697,
+      "step": 93500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.502217251455485e-05,
+      "loss": 0.7672,
+      "step": 94000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.49956947087812e-05,
+      "loss": 0.7625,
+      "step": 94500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.496921690300756e-05,
+      "loss": 0.7607,
+      "step": 95000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.494273909723392e-05,
+      "loss": 0.7696,
+      "step": 95500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.491626129146028e-05,
+      "loss": 0.7667,
+      "step": 96000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.488978348568663e-05,
+      "loss": 0.7635,
+      "step": 96500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.486330567991299e-05,
+      "loss": 0.7677,
+      "step": 97000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.483682787413934e-05,
+      "loss": 0.7661,
+      "step": 97500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.48103500683657e-05,
+      "loss": 0.7646,
+      "step": 98000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.478387226259205e-05,
+      "loss": 0.7609,
+      "step": 98500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.475739445681841e-05,
+      "loss": 0.7573,
+      "step": 99000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.473091665104476e-05,
+      "loss": 0.7591,
+      "step": 99500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.470443884527112e-05,
+      "loss": 0.7611,
+      "step": 100000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.467796103949747e-05,
+      "loss": 0.759,
+      "step": 100500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.465148323372383e-05,
+      "loss": 0.7588,
+      "step": 101000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.4625005427950184e-05,
+      "loss": 0.7623,
+      "step": 101500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.459852762217654e-05,
+      "loss": 0.7534,
+      "step": 102000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.4572049816402894e-05,
+      "loss": 0.7567,
+      "step": 102500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.454557201062925e-05,
+      "loss": 0.7572,
+      "step": 103000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.451909420485561e-05,
+      "loss": 0.7594,
+      "step": 103500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.4492616399081964e-05,
+      "loss": 0.7563,
+      "step": 104000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.446613859330832e-05,
+      "loss": 0.7543,
+      "step": 104500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.4439660787534674e-05,
+      "loss": 0.7544,
+      "step": 105000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.441318298176103e-05,
+      "loss": 0.751,
+      "step": 105500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.4386705175987385e-05,
+      "loss": 0.7468,
+      "step": 106000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.4360227370213744e-05,
+      "loss": 0.7523,
+      "step": 106500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.4333749564440096e-05,
+      "loss": 0.7504,
+      "step": 107000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.4307271758666454e-05,
+      "loss": 0.7521,
+      "step": 107500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.4280793952892806e-05,
+      "loss": 0.7492,
+      "step": 108000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.4254316147119165e-05,
+      "loss": 0.7509,
+      "step": 108500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.422783834134552e-05,
+      "loss": 0.7493,
+      "step": 109000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.4201360535571876e-05,
+      "loss": 0.7501,
+      "step": 109500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.417488272979823e-05,
+      "loss": 0.7436,
+      "step": 110000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.4148404924024586e-05,
+      "loss": 0.7475,
+      "step": 110500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.4121927118250945e-05,
+      "loss": 0.7529,
+      "step": 111000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.40954493124773e-05,
+      "loss": 0.7447,
+      "step": 111500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.4068971506703656e-05,
+      "loss": 0.7478,
+      "step": 112000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.404249370093001e-05,
+      "loss": 0.7503,
+      "step": 112500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.4016015895156366e-05,
+      "loss": 0.74,
+      "step": 113000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.398953808938272e-05,
+      "loss": 0.7462,
+      "step": 113500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.396306028360908e-05,
+      "loss": 0.7462,
+      "step": 114000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.393658247783543e-05,
+      "loss": 0.7377,
+      "step": 114500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.391010467206179e-05,
+      "loss": 0.7421,
+      "step": 115000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.388362686628814e-05,
+      "loss": 0.7369,
+      "step": 115500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.38571490605145e-05,
+      "loss": 0.7377,
+      "step": 116000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.383067125474085e-05,
+      "loss": 0.7496,
+      "step": 116500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.380419344896721e-05,
+      "loss": 0.7403,
+      "step": 117000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.377771564319356e-05,
+      "loss": 0.7437,
+      "step": 117500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.375123783741992e-05,
+      "loss": 0.7372,
+      "step": 118000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.372476003164628e-05,
+      "loss": 0.7421,
+      "step": 118500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.369828222587263e-05,
+      "loss": 0.7404,
+      "step": 119000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.367180442009899e-05,
+      "loss": 0.7363,
+      "step": 119500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.364532661432534e-05,
+      "loss": 0.7413,
+      "step": 120000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.36188488085517e-05,
+      "loss": 0.7382,
+      "step": 120500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.359237100277805e-05,
+      "loss": 0.7383,
+      "step": 121000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.356589319700441e-05,
+      "loss": 0.7362,
+      "step": 121500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.353941539123076e-05,
+      "loss": 0.738,
+      "step": 122000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.351293758545712e-05,
+      "loss": 0.735,
+      "step": 122500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.348645977968347e-05,
+      "loss": 0.7411,
+      "step": 123000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.345998197390983e-05,
+      "loss": 0.7347,
+      "step": 123500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.3433504168136183e-05,
+      "loss": 0.7334,
+      "step": 124000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.340702636236254e-05,
+      "loss": 0.7376,
+      "step": 124500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.3380548556588894e-05,
+      "loss": 0.733,
+      "step": 125000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.335407075081525e-05,
+      "loss": 0.7299,
+      "step": 125500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.332759294504161e-05,
+      "loss": 0.7312,
+      "step": 126000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.330111513926797e-05,
+      "loss": 0.7382,
+      "step": 126500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.327463733349432e-05,
+      "loss": 0.7322,
+      "step": 127000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.324815952772068e-05,
+      "loss": 0.7336,
+      "step": 127500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.322168172194703e-05,
+      "loss": 0.734,
+      "step": 128000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.3195203916173385e-05,
+      "loss": 0.7317,
+      "step": 128500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.3168726110399743e-05,
+      "loss": 0.7299,
+      "step": 129000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.3142248304626095e-05,
+      "loss": 0.7284,
+      "step": 129500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.3115770498852454e-05,
+      "loss": 0.728,
+      "step": 130000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.3089292693078806e-05,
+      "loss": 0.7302,
+      "step": 130500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.3062814887305165e-05,
+      "loss": 0.7237,
+      "step": 131000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.303633708153152e-05,
+      "loss": 0.7277,
+      "step": 131500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.3009859275757875e-05,
+      "loss": 0.7312,
+      "step": 132000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.298338146998423e-05,
+      "loss": 0.728,
+      "step": 132500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.2956903664210586e-05,
+      "loss": 0.7214,
+      "step": 133000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.2930425858436945e-05,
+      "loss": 0.7316,
+      "step": 133500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.2903948052663303e-05,
+      "loss": 0.7207,
+      "step": 134000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.2877470246889655e-05,
+      "loss": 0.7275,
+      "step": 134500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.2850992441116014e-05,
+      "loss": 0.7262,
+      "step": 135000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.2824514635342366e-05,
+      "loss": 0.726,
+      "step": 135500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.2798036829568725e-05,
+      "loss": 0.7259,
+      "step": 136000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.277155902379508e-05,
+      "loss": 0.7253,
+      "step": 136500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.2745081218021435e-05,
+      "loss": 0.7272,
+      "step": 137000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.271860341224779e-05,
+      "loss": 0.7213,
+      "step": 137500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.269212560647414e-05,
+      "loss": 0.7217,
+      "step": 138000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.26656478007005e-05,
+      "loss": 0.7207,
+      "step": 138500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.263916999492685e-05,
+      "loss": 0.7229,
+      "step": 139000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.261269218915321e-05,
+      "loss": 0.7226,
+      "step": 139500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.258621438337956e-05,
+      "loss": 0.7214,
+      "step": 140000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.255973657760592e-05,
+      "loss": 0.7215,
+      "step": 140500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.253325877183228e-05,
+      "loss": 0.7194,
+      "step": 141000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.250678096605864e-05,
+      "loss": 0.7158,
+      "step": 141500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.248030316028499e-05,
+      "loss": 0.7244,
+      "step": 142000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.245382535451135e-05,
+      "loss": 0.7237,
+      "step": 142500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.24273475487377e-05,
+      "loss": 0.7193,
+      "step": 143000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.240086974296406e-05,
+      "loss": 0.7188,
+      "step": 143500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.237439193719041e-05,
+      "loss": 0.7218,
+      "step": 144000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.234791413141677e-05,
+      "loss": 0.7189,
+      "step": 144500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.232143632564312e-05,
+      "loss": 0.7147,
+      "step": 145000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.229495851986948e-05,
+      "loss": 0.7149,
+      "step": 145500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.226848071409583e-05,
+      "loss": 0.7128,
+      "step": 146000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.224200290832219e-05,
+      "loss": 0.7194,
+      "step": 146500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.221552510254854e-05,
+      "loss": 0.7153,
+      "step": 147000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.21890472967749e-05,
+      "loss": 0.7172,
+      "step": 147500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.216256949100125e-05,
+      "loss": 0.7108,
+      "step": 148000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.213609168522761e-05,
+      "loss": 0.7188,
+      "step": 148500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.210961387945397e-05,
+      "loss": 0.714,
+      "step": 149000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.208313607368032e-05,
+      "loss": 0.7138,
+      "step": 149500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.205665826790668e-05,
+      "loss": 0.7153,
+      "step": 150000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.203018046213303e-05,
+      "loss": 0.7167,
+      "step": 150500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.200370265635939e-05,
+      "loss": 0.7145,
+      "step": 151000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.197722485058574e-05,
+      "loss": 0.7173,
+      "step": 151500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.19507470448121e-05,
+      "loss": 0.7087,
+      "step": 152000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.1924269239038454e-05,
+      "loss": 0.7164,
+      "step": 152500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.189779143326481e-05,
+      "loss": 0.7086,
+      "step": 153000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.1871313627491164e-05,
+      "loss": 0.7116,
+      "step": 153500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.184483582171752e-05,
+      "loss": 0.7073,
+      "step": 154000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.1818358015943875e-05,
+      "loss": 0.7081,
+      "step": 154500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.1791880210170234e-05,
+      "loss": 0.7111,
+      "step": 155000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.1765402404396586e-05,
+      "loss": 0.711,
+      "step": 155500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.1738924598622944e-05,
+      "loss": 0.7133,
+      "step": 156000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.17124467928493e-05,
+      "loss": 0.7138,
+      "step": 156500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.1685968987075655e-05,
+      "loss": 0.713,
+      "step": 157000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.1659491181302014e-05,
+      "loss": 0.7082,
+      "step": 157500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.1633013375528366e-05,
+      "loss": 0.7103,
+      "step": 158000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.1606535569754724e-05,
+      "loss": 0.7111,
+      "step": 158500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.1580057763981076e-05,
+      "loss": 0.7078,
+      "step": 159000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.1553579958207435e-05,
+      "loss": 0.7029,
+      "step": 159500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.152710215243379e-05,
+      "loss": 0.7089,
+      "step": 160000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.1500624346660146e-05,
+      "loss": 0.7093,
+      "step": 160500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.14741465408865e-05,
+      "loss": 0.7079,
+      "step": 161000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.1447668735112856e-05,
+      "loss": 0.7076,
+      "step": 161500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.142119092933921e-05,
+      "loss": 0.7083,
+      "step": 162000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.139471312356557e-05,
+      "loss": 0.7057,
+      "step": 162500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.136823531779192e-05,
+      "loss": 0.6995,
+      "step": 163000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.134175751201828e-05,
+      "loss": 0.709,
+      "step": 163500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.1315279706244636e-05,
+      "loss": 0.701,
+      "step": 164000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.128880190047099e-05,
+      "loss": 0.7098,
+      "step": 164500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.126232409469735e-05,
+      "loss": 0.7056,
+      "step": 165000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.12358462889237e-05,
+      "loss": 0.7036,
+      "step": 165500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.120936848315006e-05,
+      "loss": 0.6969,
+      "step": 166000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.118289067737641e-05,
+      "loss": 0.7036,
+      "step": 166500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.115641287160277e-05,
+      "loss": 0.7025,
+      "step": 167000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.112993506582912e-05,
+      "loss": 0.7058,
+      "step": 167500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.110345726005548e-05,
+      "loss": 0.7021,
+      "step": 168000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.107697945428183e-05,
+      "loss": 0.7027,
+      "step": 168500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.105050164850819e-05,
+      "loss": 0.7013,
+      "step": 169000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.102402384273454e-05,
+      "loss": 0.6999,
+      "step": 169500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.09975460369609e-05,
+      "loss": 0.7052,
+      "step": 170000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.097106823118725e-05,
+      "loss": 0.7026,
+      "step": 170500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.094459042541361e-05,
+      "loss": 0.6993,
+      "step": 171000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.091811261963997e-05,
+      "loss": 0.7016,
+      "step": 171500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.089163481386633e-05,
+      "loss": 0.6992,
+      "step": 172000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.086515700809268e-05,
+      "loss": 0.6954,
+      "step": 172500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.083867920231904e-05,
+      "loss": 0.7,
+      "step": 173000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.081220139654539e-05,
+      "loss": 0.698,
+      "step": 173500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.078572359077174e-05,
+      "loss": 0.7005,
+      "step": 174000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.07592457849981e-05,
+      "loss": 0.7035,
+      "step": 174500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.0732767979224453e-05,
+      "loss": 0.6945,
+      "step": 175000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.070629017345081e-05,
+      "loss": 0.6994,
+      "step": 175500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.0679812367677164e-05,
+      "loss": 0.6907,
+      "step": 176000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.065333456190352e-05,
+      "loss": 0.6926,
+      "step": 176500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.0626856756129875e-05,
+      "loss": 0.6953,
+      "step": 177000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.0600378950356233e-05,
+      "loss": 0.6941,
+      "step": 177500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.0573901144582585e-05,
+      "loss": 0.6961,
+      "step": 178000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.0547423338808944e-05,
+      "loss": 0.6959,
+      "step": 178500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.05209455330353e-05,
+      "loss": 0.6917,
+      "step": 179000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.049446772726166e-05,
+      "loss": 0.6976,
+      "step": 179500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.0467989921488013e-05,
+      "loss": 0.7003,
+      "step": 180000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.044151211571437e-05,
+      "loss": 0.6969,
+      "step": 180500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.0415034309940724e-05,
+      "loss": 0.6943,
+      "step": 181000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.038855650416708e-05,
+      "loss": 0.6892,
+      "step": 181500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.0362078698393435e-05,
+      "loss": 0.6923,
+      "step": 182000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.0335600892619793e-05,
+      "loss": 0.6938,
+      "step": 182500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.0309123086846145e-05,
+      "loss": 0.6954,
+      "step": 183000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.0282645281072504e-05,
+      "loss": 0.6936,
+      "step": 183500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.0256167475298856e-05,
+      "loss": 0.6885,
+      "step": 184000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.022968966952521e-05,
+      "loss": 0.6981,
+      "step": 184500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.020321186375157e-05,
+      "loss": 0.6915,
+      "step": 185000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.017673405797792e-05,
+      "loss": 0.6966,
+      "step": 185500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.015025625220428e-05,
+      "loss": 0.6962,
+      "step": 186000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.012377844643063e-05,
+      "loss": 0.6944,
+      "step": 186500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.0097300640656995e-05,
+      "loss": 0.6967,
+      "step": 187000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.007082283488335e-05,
+      "loss": 0.6902,
+      "step": 187500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.0044345029109705e-05,
+      "loss": 0.6914,
+      "step": 188000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.001786722333606e-05,
+      "loss": 0.6925,
+      "step": 188500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.9991389417562416e-05,
+      "loss": 0.6867,
+      "step": 189000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.996491161178877e-05,
+      "loss": 0.6862,
+      "step": 189500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.993843380601513e-05,
+      "loss": 0.6912,
+      "step": 190000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.991195600024148e-05,
+      "loss": 0.685,
+      "step": 190500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.988547819446784e-05,
+      "loss": 0.6885,
+      "step": 191000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.985900038869419e-05,
+      "loss": 0.6867,
+      "step": 191500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.983252258292055e-05,
+      "loss": 0.6952,
+      "step": 192000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.98060447771469e-05,
+      "loss": 0.6912,
+      "step": 192500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.977956697137326e-05,
+      "loss": 0.6835,
+      "step": 193000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.975308916559961e-05,
+      "loss": 0.692,
+      "step": 193500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.972661135982597e-05,
+      "loss": 0.6914,
+      "step": 194000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.970013355405233e-05,
+      "loss": 0.6888,
+      "step": 194500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.967365574827868e-05,
+      "loss": 0.6865,
+      "step": 195000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.964717794250504e-05,
+      "loss": 0.681,
+      "step": 195500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.962070013673139e-05,
+      "loss": 0.6906,
+      "step": 196000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.959422233095775e-05,
+      "loss": 0.6831,
+      "step": 196500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.95677445251841e-05,
+      "loss": 0.6859,
+      "step": 197000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.954126671941046e-05,
+      "loss": 0.6807,
+      "step": 197500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.951478891363681e-05,
+      "loss": 0.691,
+      "step": 198000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.948831110786317e-05,
+      "loss": 0.6868,
+      "step": 198500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.946183330208952e-05,
+      "loss": 0.6896,
+      "step": 199000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.943535549631588e-05,
+      "loss": 0.6883,
+      "step": 199500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.940887769054223e-05,
+      "loss": 0.6808,
+      "step": 200000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.938239988476859e-05,
+      "loss": 0.6841,
+      "step": 200500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.9355922078994944e-05,
+      "loss": 0.6923,
+      "step": 201000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.93294442732213e-05,
+      "loss": 0.6827,
+      "step": 201500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.930296646744766e-05,
+      "loss": 0.6836,
+      "step": 202000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.927648866167401e-05,
+      "loss": 0.6862,
+      "step": 202500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.925001085590037e-05,
+      "loss": 0.6866,
+      "step": 203000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.9223533050126724e-05,
+      "loss": 0.6792,
+      "step": 203500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.919705524435308e-05,
+      "loss": 0.6779,
+      "step": 204000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.9170577438579434e-05,
+      "loss": 0.6828,
+      "step": 204500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.914409963280579e-05,
+      "loss": 0.6816,
+      "step": 205000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.9117621827032145e-05,
+      "loss": 0.6771,
+      "step": 205500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.9091144021258504e-05,
+      "loss": 0.6762,
+      "step": 206000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.9064666215484856e-05,
+      "loss": 0.6833,
+      "step": 206500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.9038188409711214e-05,
+      "loss": 0.6803,
+      "step": 207000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.9011710603937566e-05,
+      "loss": 0.6786,
+      "step": 207500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.8985232798163925e-05,
+      "loss": 0.6817,
+      "step": 208000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.895875499239028e-05,
+      "loss": 0.6825,
+      "step": 208500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.8932277186616636e-05,
+      "loss": 0.6771,
+      "step": 209000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.8905799380842994e-05,
+      "loss": 0.6813,
+      "step": 209500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.8879321575069346e-05,
+      "loss": 0.677,
+      "step": 210000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.8852843769295705e-05,
+      "loss": 0.6758,
+      "step": 210500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.882636596352206e-05,
+      "loss": 0.6777,
+      "step": 211000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.8799888157748416e-05,
+      "loss": 0.6762,
+      "step": 211500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.877341035197477e-05,
+      "loss": 0.6762,
+      "step": 212000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.8746932546201126e-05,
+      "loss": 0.6727,
+      "step": 212500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.872045474042748e-05,
+      "loss": 0.6781,
+      "step": 213000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.869397693465384e-05,
+      "loss": 0.6818,
+      "step": 213500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.866749912888019e-05,
+      "loss": 0.6759,
+      "step": 214000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.864102132310655e-05,
+      "loss": 0.6812,
+      "step": 214500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.86145435173329e-05,
+      "loss": 0.6798,
+      "step": 215000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.858806571155926e-05,
+      "loss": 0.6729,
+      "step": 215500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.856158790578561e-05,
+      "loss": 0.6784,
+      "step": 216000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.853511010001197e-05,
+      "loss": 0.672,
+      "step": 216500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.850863229423832e-05,
+      "loss": 0.6746,
+      "step": 217000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.8482154488464686e-05,
+      "loss": 0.6705,
+      "step": 217500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.845567668269104e-05,
+      "loss": 0.6749,
+      "step": 218000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.84291988769174e-05,
+      "loss": 0.6753,
+      "step": 218500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.840272107114375e-05,
+      "loss": 0.6714,
+      "step": 219000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.837624326537011e-05,
+      "loss": 0.6713,
+      "step": 219500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.834976545959646e-05,
+      "loss": 0.6733,
+      "step": 220000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.832328765382281e-05,
+      "loss": 0.6699,
+      "step": 220500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.829680984804917e-05,
+      "loss": 0.6716,
+      "step": 221000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.827033204227552e-05,
+      "loss": 0.6733,
+      "step": 221500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.824385423650188e-05,
+      "loss": 0.6723,
+      "step": 222000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.821737643072823e-05,
+      "loss": 0.6676,
+      "step": 222500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.819089862495459e-05,
+      "loss": 0.6735,
+      "step": 223000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.8164420819180943e-05,
+      "loss": 0.6809,
+      "step": 223500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.81379430134073e-05,
+      "loss": 0.6669,
+      "step": 224000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8111465207633654e-05,
+      "loss": 0.674,
+      "step": 224500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.808498740186002e-05,
+      "loss": 0.6718,
+      "step": 225000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.805850959608637e-05,
+      "loss": 0.6681,
+      "step": 225500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.803203179031273e-05,
+      "loss": 0.668,
+      "step": 226000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.800555398453908e-05,
+      "loss": 0.6679,
+      "step": 226500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.797907617876544e-05,
+      "loss": 0.6695,
+      "step": 227000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.795259837299179e-05,
+      "loss": 0.6729,
+      "step": 227500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.792612056721815e-05,
+      "loss": 0.6697,
+      "step": 228000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7899642761444503e-05,
+      "loss": 0.6706,
+      "step": 228500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.787316495567086e-05,
+      "loss": 0.6642,
+      "step": 229000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7846687149897214e-05,
+      "loss": 0.6682,
+      "step": 229500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7820209344123566e-05,
+      "loss": 0.6683,
+      "step": 230000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7793731538349925e-05,
+      "loss": 0.6693,
+      "step": 230500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.776725373257628e-05,
+      "loss": 0.6729,
+      "step": 231000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7740775926802635e-05,
+      "loss": 0.6633,
+      "step": 231500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.771429812102899e-05,
+      "loss": 0.6644,
+      "step": 232000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.768782031525535e-05,
+      "loss": 0.6716,
+      "step": 232500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7661342509481705e-05,
+      "loss": 0.6682,
+      "step": 233000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7634864703708063e-05,
+      "loss": 0.6678,
+      "step": 233500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7608386897934415e-05,
+      "loss": 0.6689,
+      "step": 234000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7581909092160774e-05,
+      "loss": 0.666,
+      "step": 234500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7555431286387126e-05,
+      "loss": 0.6672,
+      "step": 235000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7528953480613485e-05,
+      "loss": 0.6612,
+      "step": 235500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.750247567483984e-05,
+      "loss": 0.6624,
+      "step": 236000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7475997869066195e-05,
+      "loss": 0.6631,
+      "step": 236500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.744952006329255e-05,
+      "loss": 0.6633,
+      "step": 237000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7423042257518906e-05,
+      "loss": 0.664,
+      "step": 237500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.739656445174526e-05,
+      "loss": 0.667,
+      "step": 238000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.737008664597162e-05,
+      "loss": 0.6677,
+      "step": 238500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.734360884019797e-05,
+      "loss": 0.6607,
+      "step": 239000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.731713103442433e-05,
+      "loss": 0.667,
+      "step": 239500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7290653228650686e-05,
+      "loss": 0.6628,
+      "step": 240000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.726417542287704e-05,
+      "loss": 0.6624,
+      "step": 240500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.72376976171034e-05,
+      "loss": 0.6644,
+      "step": 241000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.721121981132975e-05,
+      "loss": 0.6624,
+      "step": 241500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.718474200555611e-05,
+      "loss": 0.6623,
+      "step": 242000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.715826419978246e-05,
+      "loss": 0.6625,
+      "step": 242500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.713178639400882e-05,
+      "loss": 0.6621,
+      "step": 243000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.710530858823517e-05,
+      "loss": 0.6638,
+      "step": 243500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.707883078246153e-05,
+      "loss": 0.6632,
+      "step": 244000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.705235297668788e-05,
+      "loss": 0.6606,
+      "step": 244500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.702587517091424e-05,
+      "loss": 0.6627,
+      "step": 245000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.699939736514059e-05,
+      "loss": 0.6651,
+      "step": 245500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.697291955936695e-05,
+      "loss": 0.659,
+      "step": 246000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.69464417535933e-05,
+      "loss": 0.6661,
+      "step": 246500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.691996394781966e-05,
+      "loss": 0.6609,
+      "step": 247000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.689348614204602e-05,
+      "loss": 0.665,
+      "step": 247500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.686700833627237e-05,
+      "loss": 0.6575,
+      "step": 248000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.684053053049873e-05,
+      "loss": 0.6592,
+      "step": 248500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.681405272472508e-05,
+      "loss": 0.6576,
+      "step": 249000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.678757491895144e-05,
+      "loss": 0.66,
+      "step": 249500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.676109711317779e-05,
+      "loss": 0.6617,
+      "step": 250000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.673461930740415e-05,
+      "loss": 0.6626,
+      "step": 250500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.67081415016305e-05,
+      "loss": 0.6551,
+      "step": 251000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.668166369585686e-05,
+      "loss": 0.6619,
+      "step": 251500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.6655185890083214e-05,
+      "loss": 0.6563,
+      "step": 252000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.662870808430957e-05,
+      "loss": 0.6573,
+      "step": 252500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.6602230278535924e-05,
+      "loss": 0.6632,
+      "step": 253000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.657575247276228e-05,
+      "loss": 0.6533,
+      "step": 253500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.6549274666988635e-05,
+      "loss": 0.6571,
+      "step": 254000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.6522796861214994e-05,
+      "loss": 0.6555,
+      "step": 254500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.6496319055441346e-05,
+      "loss": 0.6573,
+      "step": 255000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.6469841249667704e-05,
+      "loss": 0.6567,
+      "step": 255500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.644336344389406e-05,
+      "loss": 0.6582,
+      "step": 256000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.6416885638120415e-05,
+      "loss": 0.6583,
+      "step": 256500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.6390407832346774e-05,
+      "loss": 0.6536,
+      "step": 257000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.6363930026573126e-05,
+      "loss": 0.6553,
+      "step": 257500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.6337452220799484e-05,
+      "loss": 0.654,
+      "step": 258000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.6310974415025836e-05,
+      "loss": 0.6556,
+      "step": 258500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.6284496609252195e-05,
+      "loss": 0.6546,
+      "step": 259000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.625801880347855e-05,
+      "loss": 0.6506,
+      "step": 259500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.6231540997704906e-05,
+      "loss": 0.6512,
+      "step": 260000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.620506319193126e-05,
+      "loss": 0.6571,
+      "step": 260500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.6178585386157616e-05,
+      "loss": 0.6576,
+      "step": 261000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.615210758038397e-05,
+      "loss": 0.6563,
+      "step": 261500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.612562977461033e-05,
+      "loss": 0.6572,
+      "step": 262000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.609915196883668e-05,
+      "loss": 0.6532,
+      "step": 262500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.6072674163063044e-05,
+      "loss": 0.6566,
+      "step": 263000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.6046196357289396e-05,
+      "loss": 0.6573,
+      "step": 263500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.6019718551515755e-05,
+      "loss": 0.6513,
+      "step": 264000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.599324074574211e-05,
+      "loss": 0.6572,
+      "step": 264500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.5966762939968466e-05,
+      "loss": 0.6626,
+      "step": 265000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.594028513419482e-05,
+      "loss": 0.6502,
+      "step": 265500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.591380732842117e-05,
+      "loss": 0.6545,
+      "step": 266000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.588732952264753e-05,
+      "loss": 0.6543,
+      "step": 266500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.586085171687388e-05,
+      "loss": 0.6482,
+      "step": 267000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.583437391110024e-05,
+      "loss": 0.6517,
+      "step": 267500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.580789610532659e-05,
+      "loss": 0.6532,
+      "step": 268000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.578141829955295e-05,
+      "loss": 0.6519,
+      "step": 268500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.57549404937793e-05,
+      "loss": 0.6527,
+      "step": 269000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.572846268800566e-05,
+      "loss": 0.6505,
+      "step": 269500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.570198488223201e-05,
+      "loss": 0.6575,
+      "step": 270000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.567550707645838e-05,
+      "loss": 0.6522,
+      "step": 270500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.564902927068473e-05,
+      "loss": 0.6535,
+      "step": 271000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.562255146491109e-05,
+      "loss": 0.6472,
+      "step": 271500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.559607365913744e-05,
+      "loss": 0.6515,
+      "step": 272000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.55695958533638e-05,
+      "loss": 0.6493,
+      "step": 272500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.554311804759015e-05,
+      "loss": 0.649,
+      "step": 273000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.551664024181651e-05,
+      "loss": 0.6535,
+      "step": 273500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.549016243604286e-05,
+      "loss": 0.6499,
+      "step": 274000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.546368463026922e-05,
+      "loss": 0.6472,
+      "step": 274500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.543720682449557e-05,
+      "loss": 0.6503,
+      "step": 275000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.541072901872193e-05,
+      "loss": 0.6494,
+      "step": 275500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.538425121294828e-05,
+      "loss": 0.6498,
+      "step": 276000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.5357773407174635e-05,
+      "loss": 0.6516,
+      "step": 276500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.5331295601400993e-05,
+      "loss": 0.6455,
+      "step": 277000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.5304817795627345e-05,
+      "loss": 0.6484,
+      "step": 277500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.527833998985371e-05,
+      "loss": 0.654,
+      "step": 278000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.525186218408006e-05,
+      "loss": 0.6492,
+      "step": 278500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.522538437830642e-05,
+      "loss": 0.6458,
+      "step": 279000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.5198906572532773e-05,
+      "loss": 0.6462,
+      "step": 279500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.517242876675913e-05,
+      "loss": 0.6488,
+      "step": 280000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.5145950960985484e-05,
+      "loss": 0.6457,
+      "step": 280500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.511947315521184e-05,
+      "loss": 0.6505,
+      "step": 281000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.5092995349438195e-05,
+      "loss": 0.6479,
+      "step": 281500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.5066517543664553e-05,
+      "loss": 0.6458,
+      "step": 282000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.5040039737890905e-05,
+      "loss": 0.6473,
+      "step": 282500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.5013561932117264e-05,
+      "loss": 0.6451,
+      "step": 283000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.4987084126343616e-05,
+      "loss": 0.6517,
+      "step": 283500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.4960606320569975e-05,
+      "loss": 0.6471,
+      "step": 284000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.493412851479633e-05,
+      "loss": 0.6405,
+      "step": 284500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.4907650709022685e-05,
+      "loss": 0.6461,
+      "step": 285000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.488117290324904e-05,
+      "loss": 0.6462,
+      "step": 285500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.4854695097475396e-05,
+      "loss": 0.648,
+      "step": 286000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.4828217291701755e-05,
+      "loss": 0.6424,
+      "step": 286500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.480173948592811e-05,
+      "loss": 0.6446,
+      "step": 287000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4775261680154465e-05,
+      "loss": 0.6406,
+      "step": 287500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.474878387438082e-05,
+      "loss": 0.6417,
+      "step": 288000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4722306068607176e-05,
+      "loss": 0.6477,
+      "step": 288500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.469582826283353e-05,
+      "loss": 0.6456,
+      "step": 289000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.466935045705989e-05,
+      "loss": 0.6446,
+      "step": 289500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.464287265128624e-05,
+      "loss": 0.646,
+      "step": 290000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.46163948455126e-05,
+      "loss": 0.6449,
+      "step": 290500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.458991703973895e-05,
+      "loss": 0.6468,
+      "step": 291000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.456343923396531e-05,
+      "loss": 0.6382,
+      "step": 291500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.453696142819166e-05,
+      "loss": 0.6443,
+      "step": 292000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.451048362241802e-05,
+      "loss": 0.6438,
+      "step": 292500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.448400581664437e-05,
+      "loss": 0.6392,
+      "step": 293000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.445752801087073e-05,
+      "loss": 0.6396,
+      "step": 293500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.443105020509709e-05,
+      "loss": 0.6399,
+      "step": 294000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.440457239932344e-05,
+      "loss": 0.6436,
+      "step": 294500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.43780945935498e-05,
+      "loss": 0.641,
+      "step": 295000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.435161678777615e-05,
+      "loss": 0.6433,
+      "step": 295500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.432513898200251e-05,
+      "loss": 0.6351,
+      "step": 296000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.429866117622886e-05,
+      "loss": 0.6442,
+      "step": 296500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.427218337045522e-05,
+      "loss": 0.6457,
+      "step": 297000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.424570556468157e-05,
+      "loss": 0.6443,
+      "step": 297500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.421922775890793e-05,
+      "loss": 0.6426,
+      "step": 298000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.419274995313428e-05,
+      "loss": 0.6419,
+      "step": 298500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.416627214736064e-05,
+      "loss": 0.6422,
+      "step": 299000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.413979434158699e-05,
+      "loss": 0.6418,
+      "step": 299500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.411331653581335e-05,
+      "loss": 0.643,
+      "step": 300000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.4086838730039704e-05,
+      "loss": 0.6437,
+      "step": 300500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.406036092426607e-05,
+      "loss": 0.6392,
+      "step": 301000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.403388311849242e-05,
+      "loss": 0.635,
+      "step": 301500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.400740531271877e-05,
+      "loss": 0.6384,
+      "step": 302000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.398092750694513e-05,
+      "loss": 0.6444,
+      "step": 302500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.3954449701171484e-05,
+      "loss": 0.6387,
+      "step": 303000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.392797189539784e-05,
+      "loss": 0.6447,
+      "step": 303500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.3901494089624194e-05,
+      "loss": 0.6412,
+      "step": 304000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.387501628385055e-05,
+      "loss": 0.636,
+      "step": 304500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.3848538478076905e-05,
+      "loss": 0.6405,
+      "step": 305000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.3822060672303264e-05,
+      "loss": 0.6451,
+      "step": 305500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.3795582866529616e-05,
+      "loss": 0.6381,
+      "step": 306000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.3769105060755974e-05,
+      "loss": 0.6407,
+      "step": 306500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.3742627254982326e-05,
+      "loss": 0.6357,
+      "step": 307000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.3716149449208685e-05,
+      "loss": 0.6451,
+      "step": 307500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.368967164343504e-05,
+      "loss": 0.6372,
+      "step": 308000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.36631938376614e-05,
+      "loss": 0.6352,
+      "step": 308500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.3636716031887754e-05,
+      "loss": 0.6383,
+      "step": 309000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.361023822611411e-05,
+      "loss": 0.6344,
+      "step": 309500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.3583760420340465e-05,
+      "loss": 0.6385,
+      "step": 310000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.3557282614566824e-05,
+      "loss": 0.6331,
+      "step": 310500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.3530804808793176e-05,
+      "loss": 0.6398,
+      "step": 311000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.3504327003019534e-05,
+      "loss": 0.6387,
+      "step": 311500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.3477849197245886e-05,
+      "loss": 0.6377,
+      "step": 312000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.345137139147224e-05,
+      "loss": 0.6363,
+      "step": 312500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.34248935856986e-05,
+      "loss": 0.6354,
+      "step": 313000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.339841577992495e-05,
+      "loss": 0.6374,
+      "step": 313500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.337193797415131e-05,
+      "loss": 0.6366,
+      "step": 314000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.334546016837766e-05,
+      "loss": 0.6389,
+      "step": 314500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.331898236260402e-05,
+      "loss": 0.637,
+      "step": 315000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.329250455683037e-05,
+      "loss": 0.6339,
+      "step": 315500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.326602675105673e-05,
+      "loss": 0.6368,
+      "step": 316000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.323954894528309e-05,
+      "loss": 0.6369,
+      "step": 316500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.3213071139509446e-05,
+      "loss": 0.6386,
+      "step": 317000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.31865933337358e-05,
+      "loss": 0.6356,
+      "step": 317500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.316011552796216e-05,
+      "loss": 0.6323,
+      "step": 318000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.313363772218851e-05,
+      "loss": 0.6399,
+      "step": 318500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.310715991641487e-05,
+      "loss": 0.6367,
+      "step": 319000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.308068211064122e-05,
+      "loss": 0.6316,
+      "step": 319500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.305420430486758e-05,
+      "loss": 0.63,
+      "step": 320000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.302772649909393e-05,
+      "loss": 0.6303,
+      "step": 320500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.300124869332029e-05,
+      "loss": 0.6345,
+      "step": 321000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.297477088754664e-05,
+      "loss": 0.6303,
+      "step": 321500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.294829308177299e-05,
+      "loss": 0.6329,
+      "step": 322000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.292181527599935e-05,
+      "loss": 0.6289,
+      "step": 322500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2895337470225704e-05,
+      "loss": 0.6342,
+      "step": 323000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.286885966445206e-05,
+      "loss": 0.6359,
+      "step": 323500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.284238185867842e-05,
+      "loss": 0.6355,
+      "step": 324000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.281590405290478e-05,
+      "loss": 0.6314,
+      "step": 324500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.278942624713113e-05,
+      "loss": 0.6349,
+      "step": 325000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.276294844135749e-05,
+      "loss": 0.6372,
+      "step": 325500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.273647063558384e-05,
+      "loss": 0.6297,
+      "step": 326000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.27099928298102e-05,
+      "loss": 0.633,
+      "step": 326500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.268351502403655e-05,
+      "loss": 0.6287,
+      "step": 327000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.265703721826291e-05,
+      "loss": 0.6299,
+      "step": 327500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.2630559412489264e-05,
+      "loss": 0.6317,
+      "step": 328000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.260408160671562e-05,
+      "loss": 0.6361,
+      "step": 328500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.2577603800941974e-05,
+      "loss": 0.6314,
+      "step": 329000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.255112599516833e-05,
+      "loss": 0.6316,
+      "step": 329500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.2524648189394685e-05,
+      "loss": 0.6294,
+      "step": 330000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.2498170383621044e-05,
+      "loss": 0.634,
+      "step": 330500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2471692577847395e-05,
+      "loss": 0.6298,
+      "step": 331000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2445214772073754e-05,
+      "loss": 0.6284,
+      "step": 331500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.241873696630011e-05,
+      "loss": 0.6324,
+      "step": 332000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2392259160526465e-05,
+      "loss": 0.6292,
+      "step": 332500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2365781354752824e-05,
+      "loss": 0.6262,
+      "step": 333000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2339303548979175e-05,
+      "loss": 0.6319,
+      "step": 333500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2312825743205534e-05,
+      "loss": 0.6317,
+      "step": 334000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2286347937431886e-05,
+      "loss": 0.6333,
+      "step": 334500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2259870131658245e-05,
+      "loss": 0.6313,
+      "step": 335000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.22333923258846e-05,
+      "loss": 0.6293,
+      "step": 335500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2206914520110955e-05,
+      "loss": 0.6315,
+      "step": 336000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.218043671433731e-05,
+      "loss": 0.6341,
+      "step": 336500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2153958908563666e-05,
+      "loss": 0.6317,
+      "step": 337000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.212748110279002e-05,
+      "loss": 0.626,
+      "step": 337500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.210100329701638e-05,
+      "loss": 0.6295,
+      "step": 338000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.207452549124273e-05,
+      "loss": 0.6319,
+      "step": 338500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.204804768546909e-05,
+      "loss": 0.6238,
+      "step": 339000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.2021569879695446e-05,
+      "loss": 0.6304,
+      "step": 339500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.19950920739218e-05,
+      "loss": 0.6243,
+      "step": 340000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.196861426814816e-05,
+      "loss": 0.6253,
+      "step": 340500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.194213646237451e-05,
+      "loss": 0.6294,
+      "step": 341000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.191565865660087e-05,
+      "loss": 0.6222,
+      "step": 341500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.188918085082722e-05,
+      "loss": 0.6298,
+      "step": 342000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.186270304505358e-05,
+      "loss": 0.6301,
+      "step": 342500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.183622523927993e-05,
+      "loss": 0.6256,
+      "step": 343000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.180974743350629e-05,
+      "loss": 0.6282,
+      "step": 343500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.178326962773264e-05,
+      "loss": 0.6293,
+      "step": 344000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.1756791821959e-05,
+      "loss": 0.6321,
+      "step": 344500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.173031401618535e-05,
+      "loss": 0.6244,
+      "step": 345000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.170383621041171e-05,
+      "loss": 0.6284,
+      "step": 345500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.167735840463806e-05,
+      "loss": 0.6268,
+      "step": 346000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.165088059886443e-05,
+      "loss": 0.6197,
+      "step": 346500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.162440279309078e-05,
+      "loss": 0.6298,
+      "step": 347000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.159792498731713e-05,
+      "loss": 0.6275,
+      "step": 347500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.157144718154349e-05,
+      "loss": 0.6229,
+      "step": 348000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.154496937576984e-05,
+      "loss": 0.6223,
+      "step": 348500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.15184915699962e-05,
+      "loss": 0.6255,
+      "step": 349000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.149201376422255e-05,
+      "loss": 0.6278,
+      "step": 349500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.146553595844891e-05,
+      "loss": 0.6231,
+      "step": 350000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.143905815267526e-05,
+      "loss": 0.6195,
+      "step": 350500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.141258034690162e-05,
+      "loss": 0.6228,
+      "step": 351000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1386102541127974e-05,
+      "loss": 0.6243,
+      "step": 351500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.135962473535433e-05,
+      "loss": 0.6219,
+      "step": 352000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1333146929580684e-05,
+      "loss": 0.6215,
+      "step": 352500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.130666912380704e-05,
+      "loss": 0.6262,
+      "step": 353000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1280191318033395e-05,
+      "loss": 0.6233,
+      "step": 353500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1253713512259754e-05,
+      "loss": 0.6235,
+      "step": 354000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.122723570648611e-05,
+      "loss": 0.6228,
+      "step": 354500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.120075790071247e-05,
+      "loss": 0.6194,
+      "step": 355000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.117428009493882e-05,
+      "loss": 0.6233,
+      "step": 355500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.114780228916518e-05,
+      "loss": 0.6213,
+      "step": 356000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.1121324483391534e-05,
+      "loss": 0.6274,
+      "step": 356500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.109484667761789e-05,
+      "loss": 0.6203,
+      "step": 357000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.1068368871844244e-05,
+      "loss": 0.6166,
+      "step": 357500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.1041891066070596e-05,
+      "loss": 0.6228,
+      "step": 358000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.1015413260296955e-05,
+      "loss": 0.6243,
+      "step": 358500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.098893545452331e-05,
+      "loss": 0.6234,
+      "step": 359000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.0962457648749666e-05,
+      "loss": 0.6251,
+      "step": 359500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.093597984297602e-05,
+      "loss": 0.6198,
+      "step": 360000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.0909502037202376e-05,
+      "loss": 0.6241,
+      "step": 360500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.088302423142873e-05,
+      "loss": 0.6218,
+      "step": 361000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.085654642565509e-05,
+      "loss": 0.6226,
+      "step": 361500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.0830068619881446e-05,
+      "loss": 0.6181,
+      "step": 362000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0803590814107804e-05,
+      "loss": 0.6188,
+      "step": 362500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0777113008334156e-05,
+      "loss": 0.6256,
+      "step": 363000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0750635202560515e-05,
+      "loss": 0.6223,
+      "step": 363500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.072415739678687e-05,
+      "loss": 0.6228,
+      "step": 364000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0697679591013226e-05,
+      "loss": 0.6206,
+      "step": 364500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.067120178523958e-05,
+      "loss": 0.6245,
+      "step": 365000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0644723979465936e-05,
+      "loss": 0.6208,
+      "step": 365500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.061824617369229e-05,
+      "loss": 0.6219,
+      "step": 366000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.059176836791865e-05,
+      "loss": 0.6143,
+      "step": 366500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0565290562145e-05,
+      "loss": 0.6223,
+      "step": 367000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.053881275637136e-05,
+      "loss": 0.6274,
+      "step": 367500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.051233495059771e-05,
+      "loss": 0.6226,
+      "step": 368000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0485857144824065e-05,
+      "loss": 0.6221,
+      "step": 368500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.045937933905042e-05,
+      "loss": 0.6207,
+      "step": 369000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0432901533276782e-05,
+      "loss": 0.6215,
+      "step": 369500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0406423727503138e-05,
+      "loss": 0.6201,
+      "step": 370000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.037994592172949e-05,
+      "loss": 0.6174,
+      "step": 370500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0353468115955845e-05,
+      "loss": 0.6154,
+      "step": 371000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.03269903101822e-05,
+      "loss": 0.6158,
+      "step": 371500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0300512504408556e-05,
+      "loss": 0.6176,
+      "step": 372000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.027403469863491e-05,
+      "loss": 0.6213,
+      "step": 372500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0247556892861266e-05,
+      "loss": 0.6169,
+      "step": 373000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.022107908708762e-05,
+      "loss": 0.615,
+      "step": 373500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0194601281313977e-05,
+      "loss": 0.6233,
+      "step": 374000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0168123475540332e-05,
+      "loss": 0.6206,
+      "step": 374500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0141645669766688e-05,
+      "loss": 0.6201,
+      "step": 375000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0115167863993043e-05,
+      "loss": 0.6168,
+      "step": 375500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0088690058219398e-05,
+      "loss": 0.6154,
+      "step": 376000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0062212252445754e-05,
+      "loss": 0.618,
+      "step": 376500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0035734446672116e-05,
+      "loss": 0.6164,
+      "step": 377000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.000925664089847e-05,
+      "loss": 0.6211,
+      "step": 377500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9982778835124826e-05,
+      "loss": 0.622,
+      "step": 378000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.995630102935118e-05,
+      "loss": 0.6186,
+      "step": 378500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.9929823223577537e-05,
+      "loss": 0.6156,
+      "step": 379000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.9903345417803892e-05,
+      "loss": 0.6119,
+      "step": 379500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.9876867612030248e-05,
+      "loss": 0.6159,
+      "step": 380000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.9850389806256603e-05,
+      "loss": 0.6159,
+      "step": 380500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.9823912000482955e-05,
+      "loss": 0.6215,
+      "step": 381000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.979743419470931e-05,
+      "loss": 0.6145,
+      "step": 381500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9770956388935665e-05,
+      "loss": 0.6208,
+      "step": 382000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.974447858316202e-05,
+      "loss": 0.6141,
+      "step": 382500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9718000777388376e-05,
+      "loss": 0.6142,
+      "step": 383000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.969152297161473e-05,
+      "loss": 0.6165,
+      "step": 383500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9665045165841087e-05,
+      "loss": 0.6127,
+      "step": 384000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9638567360067442e-05,
+      "loss": 0.615,
+      "step": 384500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9612089554293804e-05,
+      "loss": 0.6173,
+      "step": 385000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.958561174852016e-05,
+      "loss": 0.6126,
+      "step": 385500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9559133942746515e-05,
+      "loss": 0.6118,
+      "step": 386000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.953265613697287e-05,
+      "loss": 0.613,
+      "step": 386500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9506178331199225e-05,
+      "loss": 0.6098,
+      "step": 387000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.947970052542558e-05,
+      "loss": 0.6177,
+      "step": 387500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9453222719651936e-05,
+      "loss": 0.6104,
+      "step": 388000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.942674491387829e-05,
+      "loss": 0.6086,
+      "step": 388500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9400267108104647e-05,
+      "loss": 0.6113,
+      "step": 389000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9373789302331002e-05,
+      "loss": 0.6173,
+      "step": 389500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9347311496557357e-05,
+      "loss": 0.6148,
+      "step": 390000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9320833690783713e-05,
+      "loss": 0.6118,
+      "step": 390500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9294355885010065e-05,
+      "loss": 0.6088,
+      "step": 391000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.926787807923642e-05,
+      "loss": 0.6099,
+      "step": 391500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9241400273462775e-05,
+      "loss": 0.6069,
+      "step": 392000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9214922467689137e-05,
+      "loss": 0.6129,
+      "step": 392500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9188444661915493e-05,
+      "loss": 0.6115,
+      "step": 393000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9161966856141848e-05,
+      "loss": 0.6156,
+      "step": 393500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9135489050368203e-05,
+      "loss": 0.6095,
+      "step": 394000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.910901124459456e-05,
+      "loss": 0.6134,
+      "step": 394500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9082533438820914e-05,
+      "loss": 0.6091,
+      "step": 395000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.905605563304727e-05,
+      "loss": 0.6101,
+      "step": 395500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9029577827273625e-05,
+      "loss": 0.6125,
+      "step": 396000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.900310002149998e-05,
+      "loss": 0.6049,
+      "step": 396500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8976622215726335e-05,
+      "loss": 0.6145,
+      "step": 397000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.895014440995269e-05,
+      "loss": 0.6185,
+      "step": 397500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8923666604179046e-05,
+      "loss": 0.6135,
+      "step": 398000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.88971887984054e-05,
+      "loss": 0.6101,
+      "step": 398500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8870710992631757e-05,
+      "loss": 0.613,
+      "step": 399000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8844233186858112e-05,
+      "loss": 0.6118,
+      "step": 399500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.881775538108447e-05,
+      "loss": 0.6111,
+      "step": 400000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8791277575310826e-05,
+      "loss": 0.6134,
+      "step": 400500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.876479976953718e-05,
+      "loss": 0.614,
+      "step": 401000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8738321963763537e-05,
+      "loss": 0.6042,
+      "step": 401500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8711844157989892e-05,
+      "loss": 0.6085,
+      "step": 402000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8685366352216247e-05,
+      "loss": 0.6092,
+      "step": 402500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8658888546442603e-05,
+      "loss": 0.6076,
+      "step": 403000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8632410740668958e-05,
+      "loss": 0.6074,
+      "step": 403500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8605932934895313e-05,
+      "loss": 0.6084,
+      "step": 404000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.857945512912167e-05,
+      "loss": 0.6071,
+      "step": 404500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8552977323348024e-05,
+      "loss": 0.609,
+      "step": 405000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.852649951757438e-05,
+      "loss": 0.6099,
+      "step": 405500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8500021711800735e-05,
+      "loss": 0.6101,
+      "step": 406000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.847354390602709e-05,
+      "loss": 0.6005,
+      "step": 406500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8447066100253445e-05,
+      "loss": 0.6058,
+      "step": 407000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8420588294479804e-05,
+      "loss": 0.6116,
+      "step": 407500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.839411048870616e-05,
+      "loss": 0.6069,
+      "step": 408000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8367632682932515e-05,
+      "loss": 0.6041,
+      "step": 408500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.834115487715887e-05,
+      "loss": 0.6058,
+      "step": 409000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8314677071385225e-05,
+      "loss": 0.6083,
+      "step": 409500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.828819926561158e-05,
+      "loss": 0.6033,
+      "step": 410000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8261721459837936e-05,
+      "loss": 0.6105,
+      "step": 410500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.823524365406429e-05,
+      "loss": 0.6061,
+      "step": 411000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8208765848290646e-05,
+      "loss": 0.605,
+      "step": 411500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8182288042517002e-05,
+      "loss": 0.61,
+      "step": 412000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8155810236743357e-05,
+      "loss": 0.602,
+      "step": 412500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8129332430969712e-05,
+      "loss": 0.6048,
+      "step": 413000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8102854625196068e-05,
+      "loss": 0.6079,
+      "step": 413500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.8076376819422423e-05,
+      "loss": 0.6105,
+      "step": 414000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.804989901364878e-05,
+      "loss": 0.6081,
+      "step": 414500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.8023421207875134e-05,
+      "loss": 0.6075,
+      "step": 415000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7996943402101496e-05,
+      "loss": 0.609,
+      "step": 415500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.797046559632785e-05,
+      "loss": 0.6046,
+      "step": 416000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7943987790554203e-05,
+      "loss": 0.6072,
+      "step": 416500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.791750998478056e-05,
+      "loss": 0.6071,
+      "step": 417000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7891032179006914e-05,
+      "loss": 0.6039,
+      "step": 417500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.786455437323327e-05,
+      "loss": 0.6036,
+      "step": 418000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7838076567459624e-05,
+      "loss": 0.6051,
+      "step": 418500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.781159876168598e-05,
+      "loss": 0.6015,
+      "step": 419000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7785120955912335e-05,
+      "loss": 0.6073,
+      "step": 419500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.775864315013869e-05,
+      "loss": 0.6024,
+      "step": 420000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7732165344365046e-05,
+      "loss": 0.6108,
+      "step": 420500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.77056875385914e-05,
+      "loss": 0.6054,
+      "step": 421000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7679209732817756e-05,
+      "loss": 0.6016,
+      "step": 421500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.765273192704411e-05,
+      "loss": 0.6036,
+      "step": 422000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7626254121270467e-05,
+      "loss": 0.604,
+      "step": 422500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.759977631549683e-05,
+      "loss": 0.6023,
+      "step": 423000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7573298509723184e-05,
+      "loss": 0.6107,
+      "step": 423500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.754682070394954e-05,
+      "loss": 0.6052,
+      "step": 424000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7520342898175895e-05,
+      "loss": 0.6061,
+      "step": 424500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.749386509240225e-05,
+      "loss": 0.6025,
+      "step": 425000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7467387286628606e-05,
+      "loss": 0.604,
+      "step": 425500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.744090948085496e-05,
+      "loss": 0.6043,
+      "step": 426000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7414431675081313e-05,
+      "loss": 0.6043,
+      "step": 426500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7387953869307668e-05,
+      "loss": 0.6052,
+      "step": 427000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7361476063534024e-05,
+      "loss": 0.6025,
+      "step": 427500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.733499825776038e-05,
+      "loss": 0.6032,
+      "step": 428000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7308520451986734e-05,
+      "loss": 0.6035,
+      "step": 428500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.728204264621309e-05,
+      "loss": 0.6035,
+      "step": 429000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7255564840439445e-05,
+      "loss": 0.6036,
+      "step": 429500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.72290870346658e-05,
+      "loss": 0.6018,
+      "step": 430000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7202609228892162e-05,
+      "loss": 0.6001,
+      "step": 430500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7176131423118518e-05,
+      "loss": 0.6036,
+      "step": 431000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7149653617344873e-05,
+      "loss": 0.5989,
+      "step": 431500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7123175811571228e-05,
+      "loss": 0.5996,
+      "step": 432000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7096698005797584e-05,
+      "loss": 0.5998,
+      "step": 432500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.707022020002394e-05,
+      "loss": 0.6035,
+      "step": 433000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7043742394250294e-05,
+      "loss": 0.5994,
+      "step": 433500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.701726458847665e-05,
+      "loss": 0.6047,
+      "step": 434000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.6990786782703005e-05,
+      "loss": 0.602,
+      "step": 434500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.696430897692936e-05,
+      "loss": 0.6006,
+      "step": 435000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.6937831171155715e-05,
+      "loss": 0.6006,
+      "step": 435500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.691135336538207e-05,
+      "loss": 0.6006,
+      "step": 436000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.6884875559608426e-05,
+      "loss": 0.5997,
+      "step": 436500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.6858397753834778e-05,
+      "loss": 0.6002,
+      "step": 437000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.6831919948061133e-05,
+      "loss": 0.5982,
+      "step": 437500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.6805442142287495e-05,
+      "loss": 0.5986,
+      "step": 438000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.677896433651385e-05,
+      "loss": 0.603,
+      "step": 438500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.6752486530740206e-05,
+      "loss": 0.6032,
+      "step": 439000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.672600872496656e-05,
+      "loss": 0.6048,
+      "step": 439500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.6699530919192917e-05,
+      "loss": 0.5979,
+      "step": 440000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.6673053113419272e-05,
+      "loss": 0.6085,
+      "step": 440500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.6646575307645627e-05,
+      "loss": 0.5976,
+      "step": 441000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.6620097501871983e-05,
+      "loss": 0.5967,
+      "step": 441500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.6593619696098338e-05,
+      "loss": 0.597,
+      "step": 442000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.6567141890324693e-05,
+      "loss": 0.5997,
+      "step": 442500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.654066408455105e-05,
+      "loss": 0.6015,
+      "step": 443000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.6514186278777404e-05,
+      "loss": 0.5997,
+      "step": 443500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.648770847300376e-05,
+      "loss": 0.5977,
+      "step": 444000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.6461230667230115e-05,
+      "loss": 0.5953,
+      "step": 444500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.643475286145647e-05,
+      "loss": 0.6038,
+      "step": 445000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.640827505568283e-05,
+      "loss": 0.602,
+      "step": 445500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.6381797249909184e-05,
+      "loss": 0.5974,
+      "step": 446000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.635531944413554e-05,
+      "loss": 0.593,
+      "step": 446500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.6328841638361895e-05,
+      "loss": 0.6019,
+      "step": 447000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.630236383258825e-05,
+      "loss": 0.6006,
+      "step": 447500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.6275886026814605e-05,
+      "loss": 0.5966,
+      "step": 448000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.624940822104096e-05,
+      "loss": 0.5981,
+      "step": 448500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.6222930415267316e-05,
+      "loss": 0.5962,
+      "step": 449000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.619645260949367e-05,
+      "loss": 0.6012,
+      "step": 449500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.6169974803720027e-05,
+      "loss": 0.5944,
+      "step": 450000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.6143496997946382e-05,
+      "loss": 0.5964,
+      "step": 450500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.6117019192172737e-05,
+      "loss": 0.5973,
+      "step": 451000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.6090541386399093e-05,
+      "loss": 0.597,
+      "step": 451500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.6064063580625448e-05,
+      "loss": 0.5939,
+      "step": 452000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.6037585774851803e-05,
+      "loss": 0.5934,
+      "step": 452500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.601110796907816e-05,
+      "loss": 0.5988,
+      "step": 453000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.5984630163304517e-05,
+      "loss": 0.5912,
+      "step": 453500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.5958152357530873e-05,
+      "loss": 0.5939,
+      "step": 454000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.5931674551757228e-05,
+      "loss": 0.6025,
+      "step": 454500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.5905196745983583e-05,
+      "loss": 0.5967,
+      "step": 455000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.587871894020994e-05,
+      "loss": 0.5968,
+      "step": 455500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.5852241134436294e-05,
+      "loss": 0.5958,
+      "step": 456000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.582576332866265e-05,
+      "loss": 0.5946,
+      "step": 456500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.5799285522889005e-05,
+      "loss": 0.5971,
+      "step": 457000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.577280771711536e-05,
+      "loss": 0.5957,
+      "step": 457500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.5746329911341715e-05,
+      "loss": 0.5961,
+      "step": 458000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.571985210556807e-05,
+      "loss": 0.5969,
+      "step": 458500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.5693374299794426e-05,
+      "loss": 0.5987,
+      "step": 459000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.566689649402078e-05,
+      "loss": 0.5959,
+      "step": 459500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.5640418688247136e-05,
+      "loss": 0.6012,
+      "step": 460000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.5613940882473492e-05,
+      "loss": 0.5954,
+      "step": 460500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.5587463076699854e-05,
+      "loss": 0.5913,
+      "step": 461000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.556098527092621e-05,
+      "loss": 0.595,
+      "step": 461500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.5534507465152565e-05,
+      "loss": 0.5886,
+      "step": 462000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.5508029659378916e-05,
+      "loss": 0.5908,
+      "step": 462500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.5481551853605272e-05,
+      "loss": 0.5915,
+      "step": 463000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.5455074047831627e-05,
+      "loss": 0.594,
+      "step": 463500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.5428596242057982e-05,
+      "loss": 0.5894,
+      "step": 464000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.5402118436284338e-05,
+      "loss": 0.5969,
+      "step": 464500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.5375640630510693e-05,
+      "loss": 0.5965,
+      "step": 465000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.534916282473705e-05,
+      "loss": 0.5921,
+      "step": 465500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.5322685018963404e-05,
+      "loss": 0.5959,
+      "step": 466000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.529620721318976e-05,
+      "loss": 0.5926,
+      "step": 466500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.5269729407416114e-05,
+      "loss": 0.593,
+      "step": 467000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.524325160164247e-05,
+      "loss": 0.5947,
+      "step": 467500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.5216773795868825e-05,
+      "loss": 0.5991,
+      "step": 468000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.5190295990095187e-05,
+      "loss": 0.5909,
+      "step": 468500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.5163818184321542e-05,
+      "loss": 0.5925,
+      "step": 469000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.5137340378547898e-05,
+      "loss": 0.5974,
+      "step": 469500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.5110862572774253e-05,
+      "loss": 0.5944,
+      "step": 470000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.508438476700061e-05,
+      "loss": 0.5948,
+      "step": 470500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.5057906961226964e-05,
+      "loss": 0.5934,
+      "step": 471000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.503142915545332e-05,
+      "loss": 0.5932,
+      "step": 471500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.5004951349679674e-05,
+      "loss": 0.5952,
+      "step": 472000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.4978473543906026e-05,
+      "loss": 0.5927,
+      "step": 472500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.495199573813238e-05,
+      "loss": 0.5927,
+      "step": 473000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.4925517932358737e-05,
+      "loss": 0.5893,
+      "step": 473500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.4899040126585092e-05,
+      "loss": 0.5922,
+      "step": 474000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.487256232081145e-05,
+      "loss": 0.5896,
+      "step": 474500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.4846084515037806e-05,
+      "loss": 0.5914,
+      "step": 475000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.481960670926416e-05,
+      "loss": 0.5909,
+      "step": 475500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.4793128903490517e-05,
+      "loss": 0.5915,
+      "step": 476000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.4766651097716872e-05,
+      "loss": 0.599,
+      "step": 476500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.4740173291943228e-05,
+      "loss": 0.5945,
+      "step": 477000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.4713695486169583e-05,
+      "loss": 0.5887,
+      "step": 477500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.468721768039594e-05,
+      "loss": 0.5881,
+      "step": 478000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.4660739874622297e-05,
+      "loss": 0.5897,
+      "step": 478500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.4634262068848652e-05,
+      "loss": 0.5913,
+      "step": 479000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.4607784263075008e-05,
+      "loss": 0.5909,
+      "step": 479500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.4581306457301363e-05,
+      "loss": 0.589,
+      "step": 480000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.4554828651527718e-05,
+      "loss": 0.5906,
+      "step": 480500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.4528350845754074e-05,
+      "loss": 0.5884,
+      "step": 481000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.450187303998043e-05,
+      "loss": 0.5906,
+      "step": 481500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.4475395234206784e-05,
+      "loss": 0.5922,
+      "step": 482000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.444891742843314e-05,
+      "loss": 0.5936,
+      "step": 482500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.4422439622659495e-05,
+      "loss": 0.59,
+      "step": 483000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.439596181688585e-05,
+      "loss": 0.5907,
+      "step": 483500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.4369484011112206e-05,
+      "loss": 0.5951,
+      "step": 484000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.434300620533856e-05,
+      "loss": 0.5893,
+      "step": 484500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.4316528399564916e-05,
+      "loss": 0.5909,
+      "step": 485000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.4290050593791275e-05,
+      "loss": 0.594,
+      "step": 485500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.426357278801763e-05,
+      "loss": 0.5944,
+      "step": 486000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.4237094982243986e-05,
+      "loss": 0.5903,
+      "step": 486500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.421061717647034e-05,
+      "loss": 0.5907,
+      "step": 487000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.4184139370696696e-05,
+      "loss": 0.5931,
+      "step": 487500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.415766156492305e-05,
+      "loss": 0.588,
+      "step": 488000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.4131183759149407e-05,
+      "loss": 0.5895,
+      "step": 488500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.4104705953375762e-05,
+      "loss": 0.5857,
+      "step": 489000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.407822814760212e-05,
+      "loss": 0.5864,
+      "step": 489500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.4051750341828476e-05,
+      "loss": 0.5892,
+      "step": 490000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.4025272536054828e-05,
+      "loss": 0.5883,
+      "step": 490500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.3998794730281183e-05,
+      "loss": 0.5883,
+      "step": 491000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.397231692450754e-05,
+      "loss": 0.5869,
+      "step": 491500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.3945839118733894e-05,
+      "loss": 0.5862,
+      "step": 492000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.391936131296025e-05,
+      "loss": 0.5851,
+      "step": 492500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.3892883507186605e-05,
+      "loss": 0.5905,
+      "step": 493000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.3866405701412963e-05,
+      "loss": 0.5869,
+      "step": 493500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.383992789563932e-05,
+      "loss": 0.5931,
+      "step": 494000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.3813450089865674e-05,
+      "loss": 0.5871,
+      "step": 494500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.378697228409203e-05,
+      "loss": 0.5949,
+      "step": 495000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.3760494478318385e-05,
+      "loss": 0.5875,
+      "step": 495500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.373401667254474e-05,
+      "loss": 0.5892,
+      "step": 496000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.3707538866771095e-05,
+      "loss": 0.5849,
+      "step": 496500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.3681061060997454e-05,
+      "loss": 0.5862,
+      "step": 497000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.365458325522381e-05,
+      "loss": 0.588,
+      "step": 497500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.3628105449450165e-05,
+      "loss": 0.5876,
+      "step": 498000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.360162764367652e-05,
+      "loss": 0.5886,
+      "step": 498500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.3575149837902875e-05,
+      "loss": 0.5902,
+      "step": 499000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.354867203212923e-05,
+      "loss": 0.5871,
+      "step": 499500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.3522194226355586e-05,
+      "loss": 0.5889,
+      "step": 500000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.349571642058194e-05,
+      "loss": 0.5859,
+      "step": 500500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.3469238614808297e-05,
+      "loss": 0.5826,
+      "step": 501000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.3442760809034652e-05,
+      "loss": 0.5859,
+      "step": 501500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.3416283003261007e-05,
+      "loss": 0.5891,
+      "step": 502000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.3389805197487363e-05,
+      "loss": 0.5839,
+      "step": 502500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.3363327391713718e-05,
+      "loss": 0.5904,
+      "step": 503000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.3336849585940073e-05,
+      "loss": 0.5866,
+      "step": 503500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.331037178016643e-05,
+      "loss": 0.578,
+      "step": 504000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.3283893974392787e-05,
+      "loss": 0.5838,
+      "step": 504500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.3257416168619143e-05,
+      "loss": 0.5816,
+      "step": 505000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.3230938362845498e-05,
+      "loss": 0.5837,
+      "step": 505500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.3204460557071853e-05,
+      "loss": 0.5833,
+      "step": 506000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.317798275129821e-05,
+      "loss": 0.5828,
+      "step": 506500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.3151504945524564e-05,
+      "loss": 0.5856,
+      "step": 507000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.312502713975092e-05,
+      "loss": 0.586,
+      "step": 507500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.3098549333977275e-05,
+      "loss": 0.5883,
+      "step": 508000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.307207152820363e-05,
+      "loss": 0.5856,
+      "step": 508500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.3045593722429985e-05,
+      "loss": 0.5889,
+      "step": 509000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.301911591665634e-05,
+      "loss": 0.5835,
+      "step": 509500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.2992638110882696e-05,
+      "loss": 0.5876,
+      "step": 510000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.296616030510905e-05,
+      "loss": 0.5803,
+      "step": 510500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.2939682499335406e-05,
+      "loss": 0.5893,
+      "step": 511000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.2913204693561762e-05,
+      "loss": 0.578,
+      "step": 511500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.2886726887788117e-05,
+      "loss": 0.585,
+      "step": 512000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.2860249082014476e-05,
+      "loss": 0.5867,
+      "step": 512500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.283377127624083e-05,
+      "loss": 0.5836,
+      "step": 513000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.2807293470467186e-05,
+      "loss": 0.5853,
+      "step": 513500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.2780815664693542e-05,
+      "loss": 0.5846,
+      "step": 514000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.2754337858919897e-05,
+      "loss": 0.5816,
+      "step": 514500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.2727860053146252e-05,
+      "loss": 0.5798,
+      "step": 515000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.2701382247372608e-05,
+      "loss": 0.5871,
+      "step": 515500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.2674904441598966e-05,
+      "loss": 0.5841,
+      "step": 516000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.2648426635825322e-05,
+      "loss": 0.5825,
+      "step": 516500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.2621948830051677e-05,
+      "loss": 0.5799,
+      "step": 517000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.2595471024278032e-05,
+      "loss": 0.5875,
+      "step": 517500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.2568993218504388e-05,
+      "loss": 0.5861,
+      "step": 518000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.254251541273074e-05,
+      "loss": 0.5825,
+      "step": 518500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.2516037606957095e-05,
+      "loss": 0.5794,
+      "step": 519000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.248955980118345e-05,
+      "loss": 0.5805,
+      "step": 519500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.246308199540981e-05,
+      "loss": 0.5836,
+      "step": 520000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.2436604189636164e-05,
+      "loss": 0.5897,
+      "step": 520500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.241012638386252e-05,
+      "loss": 0.581,
+      "step": 521000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.2383648578088875e-05,
+      "loss": 0.579,
+      "step": 521500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.235717077231523e-05,
+      "loss": 0.5821,
+      "step": 522000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.2330692966541586e-05,
+      "loss": 0.581,
+      "step": 522500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.230421516076794e-05,
+      "loss": 0.5831,
+      "step": 523000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.2277737354994296e-05,
+      "loss": 0.5807,
+      "step": 523500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.2251259549220655e-05,
+      "loss": 0.5849,
+      "step": 524000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.222478174344701e-05,
+      "loss": 0.5836,
+      "step": 524500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.2198303937673366e-05,
+      "loss": 0.5847,
+      "step": 525000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.217182613189972e-05,
+      "loss": 0.585,
+      "step": 525500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.2145348326126076e-05,
+      "loss": 0.583,
+      "step": 526000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.211887052035243e-05,
+      "loss": 0.5813,
+      "step": 526500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.2092392714578787e-05,
+      "loss": 0.5837,
+      "step": 527000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.2065914908805142e-05,
+      "loss": 0.5813,
+      "step": 527500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.2039437103031498e-05,
+      "loss": 0.5787,
+      "step": 528000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.2012959297257853e-05,
+      "loss": 0.5816,
+      "step": 528500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.1986481491484208e-05,
+      "loss": 0.5795,
+      "step": 529000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.1960003685710564e-05,
+      "loss": 0.5805,
+      "step": 529500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.193352587993692e-05,
+      "loss": 0.5786,
+      "step": 530000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.1907048074163274e-05,
+      "loss": 0.5814,
+      "step": 530500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.188057026838963e-05,
+      "loss": 0.5865,
+      "step": 531000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.1854092462615988e-05,
+      "loss": 0.5761,
+      "step": 531500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.1827614656842344e-05,
+      "loss": 0.5805,
+      "step": 532000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.18011368510687e-05,
+      "loss": 0.5801,
+      "step": 532500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.1774659045295054e-05,
+      "loss": 0.5805,
+      "step": 533000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.174818123952141e-05,
+      "loss": 0.5795,
+      "step": 533500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.1721703433747765e-05,
+      "loss": 0.5815,
+      "step": 534000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.169522562797412e-05,
+      "loss": 0.583,
+      "step": 534500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.166874782220048e-05,
+      "loss": 0.5818,
+      "step": 535000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.1642270016426834e-05,
+      "loss": 0.5769,
+      "step": 535500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.161579221065319e-05,
+      "loss": 0.5784,
+      "step": 536000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.158931440487954e-05,
+      "loss": 0.581,
+      "step": 536500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.1562836599105897e-05,
+      "loss": 0.5805,
+      "step": 537000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.1536358793332252e-05,
+      "loss": 0.5778,
+      "step": 537500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.1509880987558607e-05,
+      "loss": 0.5819,
+      "step": 538000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.1483403181784963e-05,
+      "loss": 0.5794,
+      "step": 538500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.145692537601132e-05,
+      "loss": 0.582,
+      "step": 539000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.1430447570237677e-05,
+      "loss": 0.5819,
+      "step": 539500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.1403969764464032e-05,
+      "loss": 0.5817,
+      "step": 540000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.1377491958690387e-05,
+      "loss": 0.578,
+      "step": 540500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.1351014152916743e-05,
+      "loss": 0.5801,
+      "step": 541000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.1324536347143098e-05,
+      "loss": 0.5753,
+      "step": 541500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.1298058541369453e-05,
+      "loss": 0.5801,
+      "step": 542000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.127158073559581e-05,
+      "loss": 0.5788,
+      "step": 542500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.1245102929822167e-05,
+      "loss": 0.5759,
+      "step": 543000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.1218625124048523e-05,
+      "loss": 0.5772,
+      "step": 543500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.1192147318274878e-05,
+      "loss": 0.5729,
+      "step": 544000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.1165669512501233e-05,
+      "loss": 0.5782,
+      "step": 544500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.113919170672759e-05,
+      "loss": 0.5785,
+      "step": 545000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.1112713900953944e-05,
+      "loss": 0.5808,
+      "step": 545500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.10862360951803e-05,
+      "loss": 0.5786,
+      "step": 546000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.1059758289406655e-05,
+      "loss": 0.5755,
+      "step": 546500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.103328048363301e-05,
+      "loss": 0.5789,
+      "step": 547000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.1006802677859365e-05,
+      "loss": 0.5748,
+      "step": 547500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.098032487208572e-05,
+      "loss": 0.5809,
+      "step": 548000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.0953847066312076e-05,
+      "loss": 0.5816,
+      "step": 548500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.092736926053843e-05,
+      "loss": 0.5775,
+      "step": 549000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.0900891454764787e-05,
+      "loss": 0.5807,
+      "step": 549500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.0874413648991142e-05,
+      "loss": 0.5781,
+      "step": 550000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.08479358432175e-05,
+      "loss": 0.5793,
+      "step": 550500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.0821458037443856e-05,
+      "loss": 0.5775,
+      "step": 551000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.079498023167021e-05,
+      "loss": 0.5786,
+      "step": 551500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.0768502425896567e-05,
+      "loss": 0.5781,
+      "step": 552000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.0742024620122922e-05,
+      "loss": 0.5768,
+      "step": 552500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.0715546814349277e-05,
+      "loss": 0.5747,
+      "step": 553000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.0689069008575633e-05,
+      "loss": 0.5767,
+      "step": 553500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.066259120280199e-05,
+      "loss": 0.5761,
+      "step": 554000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.0636113397028343e-05,
+      "loss": 0.5774,
+      "step": 554500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.06096355912547e-05,
+      "loss": 0.5756,
+      "step": 555000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.0583157785481054e-05,
+      "loss": 0.5727,
+      "step": 555500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.055667997970741e-05,
+      "loss": 0.5704,
+      "step": 556000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.0530202173933765e-05,
+      "loss": 0.5749,
+      "step": 556500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.050372436816012e-05,
+      "loss": 0.5735,
+      "step": 557000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.0477246562386475e-05,
+      "loss": 0.5775,
+      "step": 557500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.0450768756612834e-05,
+      "loss": 0.5769,
+      "step": 558000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.042429095083919e-05,
+      "loss": 0.5775,
+      "step": 558500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.0397813145065545e-05,
+      "loss": 0.573,
+      "step": 559000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.03713353392919e-05,
+      "loss": 0.5752,
+      "step": 559500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.0344857533518255e-05,
+      "loss": 0.5757,
+      "step": 560000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.031837972774461e-05,
+      "loss": 0.5747,
+      "step": 560500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.0291901921970966e-05,
+      "loss": 0.5785,
+      "step": 561000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.026542411619732e-05,
+      "loss": 0.5763,
+      "step": 561500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.023894631042368e-05,
+      "loss": 0.575,
+      "step": 562000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.0212468504650035e-05,
+      "loss": 0.5777,
+      "step": 562500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.018599069887639e-05,
+      "loss": 0.5763,
+      "step": 563000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.0159512893102746e-05,
+      "loss": 0.5783,
+      "step": 563500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.01330350873291e-05,
+      "loss": 0.5718,
+      "step": 564000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.0106557281555453e-05,
+      "loss": 0.5712,
+      "step": 564500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.008007947578181e-05,
+      "loss": 0.5775,
+      "step": 565000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.0053601670008167e-05,
+      "loss": 0.5743,
+      "step": 565500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.0027123864234522e-05,
+      "loss": 0.5714,
+      "step": 566000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.0000646058460878e-05,
+      "loss": 0.5717,
+      "step": 566500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9974168252687233e-05,
+      "loss": 0.5769,
+      "step": 567000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.994769044691359e-05,
+      "loss": 0.5683,
+      "step": 567500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9921212641139944e-05,
+      "loss": 0.5731,
+      "step": 568000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.98947348353663e-05,
+      "loss": 0.574,
+      "step": 568500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9868257029592654e-05,
+      "loss": 0.5785,
+      "step": 569000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9841779223819013e-05,
+      "loss": 0.5732,
+      "step": 569500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.981530141804537e-05,
+      "loss": 0.5727,
+      "step": 570000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9788823612271724e-05,
+      "loss": 0.5695,
+      "step": 570500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.976234580649808e-05,
+      "loss": 0.5782,
+      "step": 571000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9735868000724434e-05,
+      "loss": 0.5703,
+      "step": 571500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.970939019495079e-05,
+      "loss": 0.5739,
+      "step": 572000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9682912389177145e-05,
+      "loss": 0.5737,
+      "step": 572500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.96564345834035e-05,
+      "loss": 0.5714,
+      "step": 573000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9629956777629856e-05,
+      "loss": 0.57,
+      "step": 573500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.960347897185621e-05,
+      "loss": 0.5755,
+      "step": 574000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9577001166082566e-05,
+      "loss": 0.5706,
+      "step": 574500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.955052336030892e-05,
+      "loss": 0.57,
+      "step": 575000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9524045554535277e-05,
+      "loss": 0.572,
+      "step": 575500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9497567748761632e-05,
+      "loss": 0.5716,
+      "step": 576000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9471089942987988e-05,
+      "loss": 0.5717,
+      "step": 576500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9444612137214346e-05,
+      "loss": 0.573,
+      "step": 577000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.94181343314407e-05,
+      "loss": 0.5715,
+      "step": 577500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9391656525667057e-05,
+      "loss": 0.5766,
+      "step": 578000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9365178719893412e-05,
+      "loss": 0.5711,
+      "step": 578500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9338700914119768e-05,
+      "loss": 0.5716,
+      "step": 579000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9312223108346123e-05,
+      "loss": 0.5777,
+      "step": 579500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9285745302572478e-05,
+      "loss": 0.5709,
+      "step": 580000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9259267496798834e-05,
+      "loss": 0.5714,
+      "step": 580500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9232789691025192e-05,
+      "loss": 0.5722,
+      "step": 581000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9206311885251548e-05,
+      "loss": 0.569,
+      "step": 581500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9179834079477903e-05,
+      "loss": 0.5739,
+      "step": 582000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9153356273704255e-05,
+      "loss": 0.5739,
+      "step": 582500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.912687846793061e-05,
+      "loss": 0.5661,
+      "step": 583000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9100400662156966e-05,
+      "loss": 0.5715,
+      "step": 583500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.907392285638332e-05,
+      "loss": 0.5682,
+      "step": 584000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.904744505060968e-05,
+      "loss": 0.5717,
+      "step": 584500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9020967244836035e-05,
+      "loss": 0.5687,
+      "step": 585000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.899448943906239e-05,
+      "loss": 0.5697,
+      "step": 585500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.8968011633288746e-05,
+      "loss": 0.5679,
+      "step": 586000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.89415338275151e-05,
+      "loss": 0.5668,
+      "step": 586500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.8915056021741456e-05,
+      "loss": 0.5682,
+      "step": 587000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.888857821596781e-05,
+      "loss": 0.573,
+      "step": 587500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.8862100410194167e-05,
+      "loss": 0.5703,
+      "step": 588000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.8835622604420526e-05,
+      "loss": 0.5722,
+      "step": 588500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.880914479864688e-05,
+      "loss": 0.5706,
+      "step": 589000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.8782666992873236e-05,
+      "loss": 0.5679,
+      "step": 589500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.875618918709959e-05,
+      "loss": 0.5719,
+      "step": 590000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.8729711381325947e-05,
+      "loss": 0.5686,
+      "step": 590500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.8703233575552302e-05,
+      "loss": 0.5746,
+      "step": 591000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.8676755769778658e-05,
+      "loss": 0.5719,
+      "step": 591500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.8650277964005013e-05,
+      "loss": 0.5677,
+      "step": 592000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.8623800158231368e-05,
+      "loss": 0.5696,
+      "step": 592500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.8597322352457723e-05,
+      "loss": 0.5689,
+      "step": 593000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.857084454668408e-05,
+      "loss": 0.5667,
+      "step": 593500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.8544366740910434e-05,
+      "loss": 0.5659,
+      "step": 594000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.851788893513679e-05,
+      "loss": 0.569,
+      "step": 594500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.8491411129363145e-05,
+      "loss": 0.5647,
+      "step": 595000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.84649333235895e-05,
+      "loss": 0.5662,
+      "step": 595500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.843845551781586e-05,
+      "loss": 0.5669,
+      "step": 596000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.8411977712042214e-05,
+      "loss": 0.5704,
+      "step": 596500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.838549990626857e-05,
+      "loss": 0.5686,
+      "step": 597000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.8359022100494925e-05,
+      "loss": 0.5727,
+      "step": 597500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.833254429472128e-05,
+      "loss": 0.5691,
+      "step": 598000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.8306066488947635e-05,
+      "loss": 0.5654,
+      "step": 598500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.827958868317399e-05,
+      "loss": 0.5671,
+      "step": 599000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.8253110877400346e-05,
+      "loss": 0.5701,
+      "step": 599500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.8226633071626705e-05,
+      "loss": 0.5689,
+      "step": 600000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.8200155265853057e-05,
+      "loss": 0.5712,
+      "step": 600500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.8173677460079412e-05,
+      "loss": 0.5655,
+      "step": 601000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.8147199654305767e-05,
+      "loss": 0.573,
+      "step": 601500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.8120721848532123e-05,
+      "loss": 0.5672,
+      "step": 602000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.8094244042758478e-05,
+      "loss": 0.5665,
+      "step": 602500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.8067766236984833e-05,
+      "loss": 0.5681,
+      "step": 603000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.8041288431211192e-05,
+      "loss": 0.5646,
+      "step": 603500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.8014810625437547e-05,
+      "loss": 0.5682,
+      "step": 604000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7988332819663903e-05,
+      "loss": 0.569,
+      "step": 604500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7961855013890258e-05,
+      "loss": 0.5652,
+      "step": 605000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7935377208116613e-05,
+      "loss": 0.5667,
+      "step": 605500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.790889940234297e-05,
+      "loss": 0.5643,
+      "step": 606000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7882421596569324e-05,
+      "loss": 0.5627,
+      "step": 606500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.785594379079568e-05,
+      "loss": 0.5665,
+      "step": 607000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.7829465985022038e-05,
+      "loss": 0.5711,
+      "step": 607500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.7802988179248393e-05,
+      "loss": 0.5639,
+      "step": 608000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.777651037347475e-05,
+      "loss": 0.5642,
+      "step": 608500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.7750032567701104e-05,
+      "loss": 0.564,
+      "step": 609000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.772355476192746e-05,
+      "loss": 0.5708,
+      "step": 609500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.7697076956153815e-05,
+      "loss": 0.5644,
+      "step": 610000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.7670599150380167e-05,
+      "loss": 0.566,
+      "step": 610500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.7644121344606522e-05,
+      "loss": 0.5601,
+      "step": 611000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.761764353883288e-05,
+      "loss": 0.5652,
+      "step": 611500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.7591165733059236e-05,
+      "loss": 0.5684,
+      "step": 612000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.756468792728559e-05,
+      "loss": 0.5669,
+      "step": 612500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.7538210121511947e-05,
+      "loss": 0.5658,
+      "step": 613000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.7511732315738302e-05,
+      "loss": 0.5623,
+      "step": 613500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.7485254509964657e-05,
+      "loss": 0.5668,
+      "step": 614000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.7458776704191013e-05,
+      "loss": 0.5657,
+      "step": 614500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.743229889841737e-05,
+      "loss": 0.5659,
+      "step": 615000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.7405821092643727e-05,
+      "loss": 0.5667,
+      "step": 615500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.7379343286870082e-05,
+      "loss": 0.5657,
+      "step": 616000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.7352865481096437e-05,
+      "loss": 0.5605,
+      "step": 616500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.7326387675322793e-05,
+      "loss": 0.5649,
+      "step": 617000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.7299909869549148e-05,
+      "loss": 0.5654,
+      "step": 617500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.7273432063775503e-05,
+      "loss": 0.5631,
+      "step": 618000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.724695425800186e-05,
+      "loss": 0.5619,
+      "step": 618500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.7220476452228214e-05,
+      "loss": 0.5614,
+      "step": 619000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.719399864645457e-05,
+      "loss": 0.5613,
+      "step": 619500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.7167520840680924e-05,
+      "loss": 0.5606,
+      "step": 620000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.714104303490728e-05,
+      "loss": 0.5588,
+      "step": 620500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.7114565229133635e-05,
+      "loss": 0.5665,
+      "step": 621000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.708808742335999e-05,
+      "loss": 0.5611,
+      "step": 621500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.7061609617586346e-05,
+      "loss": 0.564,
+      "step": 622000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.70351318118127e-05,
+      "loss": 0.5585,
+      "step": 622500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.700865400603906e-05,
+      "loss": 0.5659,
+      "step": 623000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6982176200265415e-05,
+      "loss": 0.5662,
+      "step": 623500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.695569839449177e-05,
+      "loss": 0.5646,
+      "step": 624000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6929220588718126e-05,
+      "loss": 0.5638,
+      "step": 624500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.690274278294448e-05,
+      "loss": 0.5618,
+      "step": 625000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6876264977170836e-05,
+      "loss": 0.5625,
+      "step": 625500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6849787171397192e-05,
+      "loss": 0.5656,
+      "step": 626000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.682330936562355e-05,
+      "loss": 0.5616,
+      "step": 626500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6796831559849906e-05,
+      "loss": 0.5572,
+      "step": 627000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.677035375407626e-05,
+      "loss": 0.5632,
+      "step": 627500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6743875948302616e-05,
+      "loss": 0.564,
+      "step": 628000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.671739814252897e-05,
+      "loss": 0.5632,
+      "step": 628500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6690920336755324e-05,
+      "loss": 0.5658,
+      "step": 629000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.666444253098168e-05,
+      "loss": 0.5656,
+      "step": 629500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6637964725208034e-05,
+      "loss": 0.5622,
+      "step": 630000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6611486919434393e-05,
+      "loss": 0.5622,
+      "step": 630500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.658500911366075e-05,
+      "loss": 0.5633,
+      "step": 631000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6558531307887104e-05,
+      "loss": 0.5629,
+      "step": 631500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.653205350211346e-05,
+      "loss": 0.5577,
+      "step": 632000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6505575696339814e-05,
+      "loss": 0.566,
+      "step": 632500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.647909789056617e-05,
+      "loss": 0.5634,
+      "step": 633000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6452620084792525e-05,
+      "loss": 0.5608,
+      "step": 633500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6426142279018884e-05,
+      "loss": 0.5586,
+      "step": 634000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.639966447324524e-05,
+      "loss": 0.5645,
+      "step": 634500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6373186667471594e-05,
+      "loss": 0.5657,
+      "step": 635000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.634670886169795e-05,
+      "loss": 0.553,
+      "step": 635500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6320231055924305e-05,
+      "loss": 0.5645,
+      "step": 636000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.629375325015066e-05,
+      "loss": 0.5609,
+      "step": 636500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6267275444377016e-05,
+      "loss": 0.5627,
+      "step": 637000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.624079763860337e-05,
+      "loss": 0.563,
+      "step": 637500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6214319832829726e-05,
+      "loss": 0.5653,
+      "step": 638000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.618784202705608e-05,
+      "loss": 0.558,
+      "step": 638500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6161364221282437e-05,
+      "loss": 0.5572,
+      "step": 639000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6134886415508792e-05,
+      "loss": 0.5609,
+      "step": 639500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6108408609735148e-05,
+      "loss": 0.5633,
+      "step": 640000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6081930803961503e-05,
+      "loss": 0.5591,
+      "step": 640500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6055452998187858e-05,
+      "loss": 0.5611,
+      "step": 641000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6028975192414213e-05,
+      "loss": 0.5578,
+      "step": 641500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6002497386640572e-05,
+      "loss": 0.5627,
+      "step": 642000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5976019580866928e-05,
+      "loss": 0.5584,
+      "step": 642500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5949541775093283e-05,
+      "loss": 0.5583,
+      "step": 643000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5923063969319638e-05,
+      "loss": 0.5632,
+      "step": 643500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5896586163545993e-05,
+      "loss": 0.5574,
+      "step": 644000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.587010835777235e-05,
+      "loss": 0.5579,
+      "step": 644500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5843630551998704e-05,
+      "loss": 0.5581,
+      "step": 645000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5817152746225063e-05,
+      "loss": 0.5565,
+      "step": 645500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5790674940451418e-05,
+      "loss": 0.5598,
+      "step": 646000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.576419713467777e-05,
+      "loss": 0.5566,
+      "step": 646500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5737719328904125e-05,
+      "loss": 0.5579,
+      "step": 647000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.571124152313048e-05,
+      "loss": 0.5644,
+      "step": 647500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5684763717356836e-05,
+      "loss": 0.5597,
+      "step": 648000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.565828591158319e-05,
+      "loss": 0.5611,
+      "step": 648500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5631808105809547e-05,
+      "loss": 0.5579,
+      "step": 649000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5605330300035905e-05,
+      "loss": 0.5597,
+      "step": 649500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.557885249426226e-05,
+      "loss": 0.561,
+      "step": 650000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5552374688488616e-05,
+      "loss": 0.5577,
+      "step": 650500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.552589688271497e-05,
+      "loss": 0.5584,
+      "step": 651000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5499419076941327e-05,
+      "loss": 0.5572,
+      "step": 651500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5472941271167682e-05,
+      "loss": 0.5565,
+      "step": 652000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5446463465394037e-05,
+      "loss": 0.5589,
+      "step": 652500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5419985659620396e-05,
+      "loss": 0.5564,
+      "step": 653000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.539350785384675e-05,
+      "loss": 0.5652,
+      "step": 653500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5367030048073107e-05,
+      "loss": 0.5583,
+      "step": 654000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5340552242299462e-05,
+      "loss": 0.5613,
+      "step": 654500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5314074436525817e-05,
+      "loss": 0.5545,
+      "step": 655000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5287596630752173e-05,
+      "loss": 0.5531,
+      "step": 655500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5261118824978528e-05,
+      "loss": 0.5605,
+      "step": 656000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5234641019204882e-05,
+      "loss": 0.553,
+      "step": 656500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.520816321343124e-05,
+      "loss": 0.558,
+      "step": 657000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5181685407657596e-05,
+      "loss": 0.5544,
+      "step": 657500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.515520760188395e-05,
+      "loss": 0.5541,
+      "step": 658000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5128729796110305e-05,
+      "loss": 0.5563,
+      "step": 658500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.510225199033666e-05,
+      "loss": 0.562,
+      "step": 659000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5075774184563015e-05,
+      "loss": 0.5554,
+      "step": 659500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.504929637878937e-05,
+      "loss": 0.5589,
+      "step": 660000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5022818573015726e-05,
+      "loss": 0.5571,
+      "step": 660500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4996340767242085e-05,
+      "loss": 0.5588,
+      "step": 661000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.496986296146844e-05,
+      "loss": 0.5546,
+      "step": 661500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4943385155694795e-05,
+      "loss": 0.56,
+      "step": 662000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.491690734992115e-05,
+      "loss": 0.5577,
+      "step": 662500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4890429544147504e-05,
+      "loss": 0.5526,
+      "step": 663000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.486395173837386e-05,
+      "loss": 0.5612,
+      "step": 663500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.4837473932600215e-05,
+      "loss": 0.5623,
+      "step": 664000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.4810996126826574e-05,
+      "loss": 0.5586,
+      "step": 664500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.4784518321052929e-05,
+      "loss": 0.5524,
+      "step": 665000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.4758040515279284e-05,
+      "loss": 0.5546,
+      "step": 665500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.473156270950564e-05,
+      "loss": 0.5567,
+      "step": 666000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.4705084903731995e-05,
+      "loss": 0.5599,
+      "step": 666500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.467860709795835e-05,
+      "loss": 0.5567,
+      "step": 667000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.4652129292184706e-05,
+      "loss": 0.5533,
+      "step": 667500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.4625651486411059e-05,
+      "loss": 0.5578,
+      "step": 668000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.4599173680637418e-05,
+      "loss": 0.5544,
+      "step": 668500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.4572695874863773e-05,
+      "loss": 0.5513,
+      "step": 669000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.4546218069090129e-05,
+      "loss": 0.5589,
+      "step": 669500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.4519740263316484e-05,
+      "loss": 0.557,
+      "step": 670000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.4493262457542839e-05,
+      "loss": 0.5499,
+      "step": 670500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.4466784651769194e-05,
+      "loss": 0.552,
+      "step": 671000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.444030684599555e-05,
+      "loss": 0.5528,
+      "step": 671500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.4413829040221905e-05,
+      "loss": 0.5528,
+      "step": 672000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.4387351234448262e-05,
+      "loss": 0.5578,
+      "step": 672500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.4360873428674617e-05,
+      "loss": 0.554,
+      "step": 673000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.4334395622900973e-05,
+      "loss": 0.5486,
+      "step": 673500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.4307917817127328e-05,
+      "loss": 0.5574,
+      "step": 674000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.4281440011353683e-05,
+      "loss": 0.5551,
+      "step": 674500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.4254962205580039e-05,
+      "loss": 0.5538,
+      "step": 675000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.4228484399806394e-05,
+      "loss": 0.5582,
+      "step": 675500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.4202006594032751e-05,
+      "loss": 0.5542,
+      "step": 676000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.4175528788259106e-05,
+      "loss": 0.5556,
+      "step": 676500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.4149050982485462e-05,
+      "loss": 0.5616,
+      "step": 677000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.4122573176711817e-05,
+      "loss": 0.561,
+      "step": 677500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.4096095370938172e-05,
+      "loss": 0.5497,
+      "step": 678000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.4069617565164528e-05,
+      "loss": 0.5564,
+      "step": 678500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4043139759390883e-05,
+      "loss": 0.5495,
+      "step": 679000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4016661953617238e-05,
+      "loss": 0.5572,
+      "step": 679500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3990184147843597e-05,
+      "loss": 0.5552,
+      "step": 680000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3963706342069952e-05,
+      "loss": 0.5523,
+      "step": 680500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3937228536296306e-05,
+      "loss": 0.5539,
+      "step": 681000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3910750730522661e-05,
+      "loss": 0.5592,
+      "step": 681500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3884272924749017e-05,
+      "loss": 0.553,
+      "step": 682000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3857795118975372e-05,
+      "loss": 0.5577,
+      "step": 682500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3831317313201727e-05,
+      "loss": 0.5545,
+      "step": 683000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3804839507428086e-05,
+      "loss": 0.5541,
+      "step": 683500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3778361701654441e-05,
+      "loss": 0.5544,
+      "step": 684000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3751883895880797e-05,
+      "loss": 0.5529,
+      "step": 684500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3725406090107152e-05,
+      "loss": 0.5557,
+      "step": 685000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3698928284333507e-05,
+      "loss": 0.5524,
+      "step": 685500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3672450478559861e-05,
+      "loss": 0.5506,
+      "step": 686000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3645972672786216e-05,
+      "loss": 0.5491,
+      "step": 686500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3619494867012572e-05,
+      "loss": 0.561,
+      "step": 687000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.359301706123893e-05,
+      "loss": 0.5497,
+      "step": 687500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3566539255465286e-05,
+      "loss": 0.552,
+      "step": 688000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3540061449691641e-05,
+      "loss": 0.5577,
+      "step": 688500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3513583643917996e-05,
+      "loss": 0.5575,
+      "step": 689000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3487105838144352e-05,
+      "loss": 0.5505,
+      "step": 689500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3460628032370707e-05,
+      "loss": 0.5511,
+      "step": 690000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3434150226597062e-05,
+      "loss": 0.5471,
+      "step": 690500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.3407672420823416e-05,
+      "loss": 0.5508,
+      "step": 691000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.3381194615049775e-05,
+      "loss": 0.5515,
+      "step": 691500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.335471680927613e-05,
+      "loss": 0.5499,
+      "step": 692000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.3328239003502485e-05,
+      "loss": 0.5451,
+      "step": 692500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.330176119772884e-05,
+      "loss": 0.5491,
+      "step": 693000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.3275283391955196e-05,
+      "loss": 0.5528,
+      "step": 693500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.3248805586181551e-05,
+      "loss": 0.5485,
+      "step": 694000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.3222327780407906e-05,
+      "loss": 0.5512,
+      "step": 694500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.3195849974634264e-05,
+      "loss": 0.5496,
+      "step": 695000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.3169372168860619e-05,
+      "loss": 0.5523,
+      "step": 695500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.3142894363086974e-05,
+      "loss": 0.5502,
+      "step": 696000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.311641655731333e-05,
+      "loss": 0.5502,
+      "step": 696500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.3089938751539685e-05,
+      "loss": 0.5528,
+      "step": 697000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.306346094576604e-05,
+      "loss": 0.5526,
+      "step": 697500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.3036983139992395e-05,
+      "loss": 0.5486,
+      "step": 698000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.301050533421875e-05,
+      "loss": 0.5493,
+      "step": 698500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2984027528445108e-05,
+      "loss": 0.5534,
+      "step": 699000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2957549722671463e-05,
+      "loss": 0.5511,
+      "step": 699500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2931071916897818e-05,
+      "loss": 0.5539,
+      "step": 700000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2904594111124174e-05,
+      "loss": 0.5474,
+      "step": 700500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2878116305350529e-05,
+      "loss": 0.5462,
+      "step": 701000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2851638499576884e-05,
+      "loss": 0.5526,
+      "step": 701500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.282516069380324e-05,
+      "loss": 0.5478,
+      "step": 702000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2798682888029598e-05,
+      "loss": 0.5509,
+      "step": 702500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2772205082255954e-05,
+      "loss": 0.5458,
+      "step": 703000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2745727276482309e-05,
+      "loss": 0.5503,
+      "step": 703500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2719249470708663e-05,
+      "loss": 0.55,
+      "step": 704000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2692771664935018e-05,
+      "loss": 0.5545,
+      "step": 704500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2666293859161373e-05,
+      "loss": 0.5509,
+      "step": 705000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2639816053387729e-05,
+      "loss": 0.5508,
+      "step": 705500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2613338247614084e-05,
+      "loss": 0.5497,
+      "step": 706000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2586860441840443e-05,
+      "loss": 0.5545,
+      "step": 706500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2560382636066798e-05,
+      "loss": 0.5498,
+      "step": 707000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2533904830293153e-05,
+      "loss": 0.5493,
+      "step": 707500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2507427024519509e-05,
+      "loss": 0.5467,
+      "step": 708000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2480949218745864e-05,
+      "loss": 0.5498,
+      "step": 708500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2454471412972218e-05,
+      "loss": 0.5496,
+      "step": 709000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2427993607198575e-05,
+      "loss": 0.5517,
+      "step": 709500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.240151580142493e-05,
+      "loss": 0.5457,
+      "step": 710000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2375037995651285e-05,
+      "loss": 0.5519,
+      "step": 710500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.234856018987764e-05,
+      "loss": 0.5498,
+      "step": 711000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2322082384103998e-05,
+      "loss": 0.543,
+      "step": 711500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2295604578330353e-05,
+      "loss": 0.5444,
+      "step": 712000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2269126772556708e-05,
+      "loss": 0.5496,
+      "step": 712500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2242648966783064e-05,
+      "loss": 0.548,
+      "step": 713000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2216171161009419e-05,
+      "loss": 0.5512,
+      "step": 713500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2189693355235774e-05,
+      "loss": 0.5511,
+      "step": 714000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.216321554946213e-05,
+      "loss": 0.5439,
+      "step": 714500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2136737743688487e-05,
+      "loss": 0.5518,
+      "step": 715000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2110259937914842e-05,
+      "loss": 0.5514,
+      "step": 715500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2083782132141197e-05,
+      "loss": 0.551,
+      "step": 716000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2057304326367553e-05,
+      "loss": 0.5535,
+      "step": 716500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.203082652059391e-05,
+      "loss": 0.5518,
+      "step": 717000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2004348714820265e-05,
+      "loss": 0.5464,
+      "step": 717500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1977870909046619e-05,
+      "loss": 0.5461,
+      "step": 718000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1951393103272974e-05,
+      "loss": 0.5504,
+      "step": 718500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1924915297499331e-05,
+      "loss": 0.5448,
+      "step": 719000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1898437491725686e-05,
+      "loss": 0.5457,
+      "step": 719500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1871959685952042e-05,
+      "loss": 0.5493,
+      "step": 720000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1845481880178397e-05,
+      "loss": 0.5498,
+      "step": 720500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1819004074404754e-05,
+      "loss": 0.5481,
+      "step": 721000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.179252626863111e-05,
+      "loss": 0.5449,
+      "step": 721500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1766048462857464e-05,
+      "loss": 0.5509,
+      "step": 722000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.173957065708382e-05,
+      "loss": 0.5436,
+      "step": 722500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1713092851310175e-05,
+      "loss": 0.5438,
+      "step": 723000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.168661504553653e-05,
+      "loss": 0.5462,
+      "step": 723500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1660137239762886e-05,
+      "loss": 0.5486,
+      "step": 724000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1633659433989243e-05,
+      "loss": 0.5455,
+      "step": 724500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1607181628215598e-05,
+      "loss": 0.5478,
+      "step": 725000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1580703822441953e-05,
+      "loss": 0.5508,
+      "step": 725500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1554226016668309e-05,
+      "loss": 0.5456,
+      "step": 726000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1527748210894666e-05,
+      "loss": 0.542,
+      "step": 726500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.150127040512102e-05,
+      "loss": 0.5475,
+      "step": 727000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1474792599347375e-05,
+      "loss": 0.5489,
+      "step": 727500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.144831479357373e-05,
+      "loss": 0.5475,
+      "step": 728000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1421836987800087e-05,
+      "loss": 0.5472,
+      "step": 728500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1395359182026442e-05,
+      "loss": 0.5441,
+      "step": 729000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1368881376252798e-05,
+      "loss": 0.5465,
+      "step": 729500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1342403570479153e-05,
+      "loss": 0.5483,
+      "step": 730000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.131592576470551e-05,
+      "loss": 0.5486,
+      "step": 730500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1289447958931865e-05,
+      "loss": 0.5408,
+      "step": 731000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.126297015315822e-05,
+      "loss": 0.5465,
+      "step": 731500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1236492347384576e-05,
+      "loss": 0.5458,
+      "step": 732000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1210014541610931e-05,
+      "loss": 0.5419,
+      "step": 732500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1183536735837287e-05,
+      "loss": 0.5447,
+      "step": 733000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1157058930063642e-05,
+      "loss": 0.5427,
+      "step": 733500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1130581124289999e-05,
+      "loss": 0.5423,
+      "step": 734000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1104103318516354e-05,
+      "loss": 0.5465,
+      "step": 734500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.107762551274271e-05,
+      "loss": 0.5433,
+      "step": 735000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1051147706969065e-05,
+      "loss": 0.546,
+      "step": 735500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.102466990119542e-05,
+      "loss": 0.5453,
+      "step": 736000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.0998192095421776e-05,
+      "loss": 0.5457,
+      "step": 736500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.0971714289648131e-05,
+      "loss": 0.5469,
+      "step": 737000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.0945236483874486e-05,
+      "loss": 0.5459,
+      "step": 737500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.0918758678100843e-05,
+      "loss": 0.544,
+      "step": 738000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0892280872327199e-05,
+      "loss": 0.5437,
+      "step": 738500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0865803066553554e-05,
+      "loss": 0.5433,
+      "step": 739000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.083932526077991e-05,
+      "loss": 0.5484,
+      "step": 739500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0812847455006266e-05,
+      "loss": 0.5436,
+      "step": 740000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0786369649232622e-05,
+      "loss": 0.5438,
+      "step": 740500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0759891843458975e-05,
+      "loss": 0.5453,
+      "step": 741000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0733414037685332e-05,
+      "loss": 0.5471,
+      "step": 741500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0706936231911688e-05,
+      "loss": 0.547,
+      "step": 742000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0680458426138043e-05,
+      "loss": 0.5425,
+      "step": 742500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0653980620364398e-05,
+      "loss": 0.5461,
+      "step": 743000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0627502814590755e-05,
+      "loss": 0.5505,
+      "step": 743500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.060102500881711e-05,
+      "loss": 0.5408,
+      "step": 744000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0574547203043466e-05,
+      "loss": 0.5447,
+      "step": 744500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0548069397269821e-05,
+      "loss": 0.5443,
+      "step": 745000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0521591591496177e-05,
+      "loss": 0.5456,
+      "step": 745500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0495113785722532e-05,
+      "loss": 0.5416,
+      "step": 746000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0468635979948887e-05,
+      "loss": 0.5471,
+      "step": 746500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0442158174175242e-05,
+      "loss": 0.5433,
+      "step": 747000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.04156803684016e-05,
+      "loss": 0.542,
+      "step": 747500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0389202562627955e-05,
+      "loss": 0.5436,
+      "step": 748000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.036272475685431e-05,
+      "loss": 0.5407,
+      "step": 748500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0336246951080665e-05,
+      "loss": 0.548,
+      "step": 749000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0309769145307022e-05,
+      "loss": 0.5429,
+      "step": 749500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0283291339533376e-05,
+      "loss": 0.5446,
+      "step": 750000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0256813533759731e-05,
+      "loss": 0.5498,
+      "step": 750500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0230335727986087e-05,
+      "loss": 0.5436,
+      "step": 751000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0203857922212444e-05,
+      "loss": 0.5399,
+      "step": 751500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0177380116438799e-05,
+      "loss": 0.5411,
+      "step": 752000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0150902310665154e-05,
+      "loss": 0.5437,
+      "step": 752500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0124424504891511e-05,
+      "loss": 0.5415,
+      "step": 753000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0097946699117867e-05,
+      "loss": 0.5467,
+      "step": 753500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0071468893344222e-05,
+      "loss": 0.544,
+      "step": 754000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0044991087570577e-05,
+      "loss": 0.5422,
+      "step": 754500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0018513281796933e-05,
+      "loss": 0.5385,
+      "step": 755000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.992035476023288e-06,
+      "loss": 0.5418,
+      "step": 755500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.965557670249643e-06,
+      "loss": 0.5444,
+      "step": 756000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.939079864475999e-06,
+      "loss": 0.5416,
+      "step": 756500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.912602058702356e-06,
+      "loss": 0.54,
+      "step": 757000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.886124252928711e-06,
+      "loss": 0.5406,
+      "step": 757500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.859646447155066e-06,
+      "loss": 0.5452,
+      "step": 758000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.833168641381422e-06,
+      "loss": 0.5398,
+      "step": 758500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.806690835607777e-06,
+      "loss": 0.5415,
+      "step": 759000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.780213029834132e-06,
+      "loss": 0.542,
+      "step": 759500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.753735224060488e-06,
+      "loss": 0.5399,
+      "step": 760000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.727257418286843e-06,
+      "loss": 0.5469,
+      "step": 760500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.7007796125132e-06,
+      "loss": 0.5392,
+      "step": 761000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.674301806739555e-06,
+      "loss": 0.542,
+      "step": 761500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.64782400096591e-06,
+      "loss": 0.541,
+      "step": 762000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.621346195192268e-06,
+      "loss": 0.5416,
+      "step": 762500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.594868389418623e-06,
+      "loss": 0.5419,
+      "step": 763000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.568390583644978e-06,
+      "loss": 0.5395,
+      "step": 763500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.541912777871332e-06,
+      "loss": 0.5359,
+      "step": 764000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.515434972097689e-06,
+      "loss": 0.5432,
+      "step": 764500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.488957166324044e-06,
+      "loss": 0.546,
+      "step": 765000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.4624793605504e-06,
+      "loss": 0.5418,
+      "step": 765500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.436001554776755e-06,
+      "loss": 0.5397,
+      "step": 766000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.409523749003112e-06,
+      "loss": 0.5403,
+      "step": 766500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.383045943229467e-06,
+      "loss": 0.5385,
+      "step": 767000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.356568137455823e-06,
+      "loss": 0.5402,
+      "step": 767500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.330090331682178e-06,
+      "loss": 0.538,
+      "step": 768000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.303612525908533e-06,
+      "loss": 0.5412,
+      "step": 768500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.277134720134889e-06,
+      "loss": 0.5401,
+      "step": 769000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.250656914361244e-06,
+      "loss": 0.5396,
+      "step": 769500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.2241791085876e-06,
+      "loss": 0.5385,
+      "step": 770000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.197701302813956e-06,
+      "loss": 0.5426,
+      "step": 770500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.171223497040312e-06,
+      "loss": 0.5437,
+      "step": 771000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.144745691266667e-06,
+      "loss": 0.5411,
+      "step": 771500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.118267885493024e-06,
+      "loss": 0.5378,
+      "step": 772000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.09179007971938e-06,
+      "loss": 0.5419,
+      "step": 772500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.065312273945733e-06,
+      "loss": 0.5415,
+      "step": 773000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.038834468172088e-06,
+      "loss": 0.537,
+      "step": 773500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.012356662398445e-06,
+      "loss": 0.539,
+      "step": 774000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 8.9858788566248e-06,
+      "loss": 0.5408,
+      "step": 774500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 8.959401050851156e-06,
+      "loss": 0.5422,
+      "step": 775000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 8.932923245077511e-06,
+      "loss": 0.5367,
+      "step": 775500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 8.906445439303868e-06,
+      "loss": 0.5436,
+      "step": 776000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 8.879967633530223e-06,
+      "loss": 0.5392,
+      "step": 776500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 8.853489827756579e-06,
+      "loss": 0.5412,
+      "step": 777000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 8.827012021982934e-06,
+      "loss": 0.5354,
+      "step": 777500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.80053421620929e-06,
+      "loss": 0.5394,
+      "step": 778000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.774056410435645e-06,
+      "loss": 0.542,
+      "step": 778500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.747578604662e-06,
+      "loss": 0.538,
+      "step": 779000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.721100798888355e-06,
+      "loss": 0.5364,
+      "step": 779500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.694622993114712e-06,
+      "loss": 0.5419,
+      "step": 780000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.668145187341068e-06,
+      "loss": 0.5388,
+      "step": 780500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.641667381567423e-06,
+      "loss": 0.5412,
+      "step": 781000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.615189575793778e-06,
+      "loss": 0.5416,
+      "step": 781500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.588711770020134e-06,
+      "loss": 0.5415,
+      "step": 782000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.562233964246489e-06,
+      "loss": 0.5357,
+      "step": 782500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.535756158472844e-06,
+      "loss": 0.5387,
+      "step": 783000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.509278352699201e-06,
+      "loss": 0.5395,
+      "step": 783500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.482800546925557e-06,
+      "loss": 0.5369,
+      "step": 784000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.456322741151912e-06,
+      "loss": 0.54,
+      "step": 784500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.429844935378267e-06,
+      "loss": 0.5347,
+      "step": 785000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.403367129604624e-06,
+      "loss": 0.5396,
+      "step": 785500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.37688932383098e-06,
+      "loss": 0.5392,
+      "step": 786000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.350411518057335e-06,
+      "loss": 0.5407,
+      "step": 786500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.323933712283689e-06,
+      "loss": 0.5349,
+      "step": 787000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.297455906510046e-06,
+      "loss": 0.5358,
+      "step": 787500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.270978100736401e-06,
+      "loss": 0.5424,
+      "step": 788000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.244500294962756e-06,
+      "loss": 0.5378,
+      "step": 788500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.218022489189112e-06,
+      "loss": 0.5406,
+      "step": 789000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.191544683415469e-06,
+      "loss": 0.5438,
+      "step": 789500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 8.165066877641824e-06,
+      "loss": 0.5389,
+      "step": 790000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 8.13858907186818e-06,
+      "loss": 0.5383,
+      "step": 790500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 8.112111266094535e-06,
+      "loss": 0.5362,
+      "step": 791000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 8.08563346032089e-06,
+      "loss": 0.5397,
+      "step": 791500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 8.059155654547245e-06,
+      "loss": 0.5342,
+      "step": 792000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 8.0326778487736e-06,
+      "loss": 0.538,
+      "step": 792500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 8.006200042999958e-06,
+      "loss": 0.5374,
+      "step": 793000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.979722237226313e-06,
+      "loss": 0.5387,
+      "step": 793500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 7.953244431452668e-06,
+      "loss": 0.5363,
+      "step": 794000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 7.926766625679024e-06,
+      "loss": 0.5362,
+      "step": 794500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 7.90028881990538e-06,
+      "loss": 0.5386,
+      "step": 795000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 7.873811014131736e-06,
+      "loss": 0.5355,
+      "step": 795500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 7.84733320835809e-06,
+      "loss": 0.5376,
+      "step": 796000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 7.820855402584445e-06,
+      "loss": 0.535,
+      "step": 796500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 7.794377596810802e-06,
+      "loss": 0.5376,
+      "step": 797000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 7.767899791037157e-06,
+      "loss": 0.5393,
+      "step": 797500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 7.741421985263513e-06,
+      "loss": 0.5398,
+      "step": 798000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 7.714944179489868e-06,
+      "loss": 0.5351,
+      "step": 798500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 7.688466373716225e-06,
+      "loss": 0.5312,
+      "step": 799000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 7.66198856794258e-06,
+      "loss": 0.5378,
+      "step": 799500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 7.635510762168935e-06,
+      "loss": 0.5353,
+      "step": 800000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 7.60903295639529e-06,
+      "loss": 0.5319,
+      "step": 800500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 7.582555150621647e-06,
+      "loss": 0.5363,
+      "step": 801000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 7.5560773448480015e-06,
+      "loss": 0.5334,
+      "step": 801500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 7.529599539074357e-06,
+      "loss": 0.5369,
+      "step": 802000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 7.503121733300714e-06,
+      "loss": 0.5373,
+      "step": 802500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 7.476643927527069e-06,
+      "loss": 0.5404,
+      "step": 803000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 7.4501661217534244e-06,
+      "loss": 0.5346,
+      "step": 803500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 7.423688315979779e-06,
+      "loss": 0.537,
+      "step": 804000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 7.397210510206136e-06,
+      "loss": 0.5378,
+      "step": 804500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 7.370732704432491e-06,
+      "loss": 0.5339,
+      "step": 805000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 7.344254898658847e-06,
+      "loss": 0.5361,
+      "step": 805500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 7.317777092885202e-06,
+      "loss": 0.5341,
+      "step": 806000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 7.291299287111558e-06,
+      "loss": 0.5341,
+      "step": 806500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 7.264821481337913e-06,
+      "loss": 0.5327,
+      "step": 807000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 7.238343675564269e-06,
+      "loss": 0.536,
+      "step": 807500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 7.211865869790624e-06,
+      "loss": 0.5364,
+      "step": 808000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 7.18538806401698e-06,
+      "loss": 0.5332,
+      "step": 808500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 7.1589102582433355e-06,
+      "loss": 0.5342,
+      "step": 809000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 7.132432452469691e-06,
+      "loss": 0.5355,
+      "step": 809500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 7.105954646696046e-06,
+      "loss": 0.5348,
+      "step": 810000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 7.079476840922402e-06,
+      "loss": 0.5369,
+      "step": 810500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 7.052999035148758e-06,
+      "loss": 0.5368,
+      "step": 811000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 7.026521229375113e-06,
+      "loss": 0.5352,
+      "step": 811500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 7.00004342360147e-06,
+      "loss": 0.5329,
+      "step": 812000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.973565617827825e-06,
+      "loss": 0.535,
+      "step": 812500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.94708781205418e-06,
+      "loss": 0.5306,
+      "step": 813000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.920610006280535e-06,
+      "loss": 0.5332,
+      "step": 813500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.894132200506892e-06,
+      "loss": 0.5364,
+      "step": 814000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.8676543947332475e-06,
+      "loss": 0.5366,
+      "step": 814500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.841176588959603e-06,
+      "loss": 0.5364,
+      "step": 815000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.814698783185957e-06,
+      "loss": 0.5319,
+      "step": 815500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.788220977412314e-06,
+      "loss": 0.5386,
+      "step": 816000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.76174317163867e-06,
+      "loss": 0.5388,
+      "step": 816500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.735265365865025e-06,
+      "loss": 0.5316,
+      "step": 817000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.70878756009138e-06,
+      "loss": 0.5298,
+      "step": 817500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.6823097543177364e-06,
+      "loss": 0.5332,
+      "step": 818000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.655831948544092e-06,
+      "loss": 0.5317,
+      "step": 818500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.629354142770447e-06,
+      "loss": 0.5371,
+      "step": 819000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.602876336996802e-06,
+      "loss": 0.5316,
+      "step": 819500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.5763985312231586e-06,
+      "loss": 0.5337,
+      "step": 820000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.549920725449514e-06,
+      "loss": 0.5354,
+      "step": 820500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.523442919675869e-06,
+      "loss": 0.534,
+      "step": 821000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.496965113902226e-06,
+      "loss": 0.5305,
+      "step": 821500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.470487308128581e-06,
+      "loss": 0.5344,
+      "step": 822000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.444009502354936e-06,
+      "loss": 0.5338,
+      "step": 822500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.417531696581291e-06,
+      "loss": 0.5323,
+      "step": 823000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.391053890807648e-06,
+      "loss": 0.5344,
+      "step": 823500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.364576085034004e-06,
+      "loss": 0.5358,
+      "step": 824000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.338098279260358e-06,
+      "loss": 0.5334,
+      "step": 824500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.3116204734867135e-06,
+      "loss": 0.5333,
+      "step": 825000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.2851426677130705e-06,
+      "loss": 0.5296,
+      "step": 825500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.258664861939426e-06,
+      "loss": 0.5341,
+      "step": 826000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.232187056165781e-06,
+      "loss": 0.5318,
+      "step": 826500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.2057092503921365e-06,
+      "loss": 0.5348,
+      "step": 827000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.179231444618492e-06,
+      "loss": 0.5319,
+      "step": 827500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.152753638844848e-06,
+      "loss": 0.5347,
+      "step": 828000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.126275833071203e-06,
+      "loss": 0.5283,
+      "step": 828500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.099798027297559e-06,
+      "loss": 0.5296,
+      "step": 829000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.073320221523914e-06,
+      "loss": 0.5333,
+      "step": 829500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.04684241575027e-06,
+      "loss": 0.5316,
+      "step": 830000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.020364609976626e-06,
+      "loss": 0.5328,
+      "step": 830500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.993886804202982e-06,
+      "loss": 0.5337,
+      "step": 831000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.967408998429337e-06,
+      "loss": 0.5339,
+      "step": 831500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.940931192655692e-06,
+      "loss": 0.5316,
+      "step": 832000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.914453386882048e-06,
+      "loss": 0.5298,
+      "step": 832500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.887975581108404e-06,
+      "loss": 0.5301,
+      "step": 833000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.861497775334759e-06,
+      "loss": 0.5299,
+      "step": 833500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.835019969561114e-06,
+      "loss": 0.5338,
+      "step": 834000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.8085421637874706e-06,
+      "loss": 0.5329,
+      "step": 834500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.782064358013826e-06,
+      "loss": 0.5301,
+      "step": 835000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.755586552240182e-06,
+      "loss": 0.5305,
+      "step": 835500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.7291087464665365e-06,
+      "loss": 0.5338,
+      "step": 836000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.702630940692893e-06,
+      "loss": 0.5278,
+      "step": 836500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.676153134919248e-06,
+      "loss": 0.5297,
+      "step": 837000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.649675329145604e-06,
+      "loss": 0.5318,
+      "step": 837500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.6231975233719595e-06,
+      "loss": 0.5349,
+      "step": 838000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.596719717598315e-06,
+      "loss": 0.5302,
+      "step": 838500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.57024191182467e-06,
+      "loss": 0.532,
+      "step": 839000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.543764106051026e-06,
+      "loss": 0.5302,
+      "step": 839500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.517286300277382e-06,
+      "loss": 0.5297,
+      "step": 840000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.490808494503737e-06,
+      "loss": 0.5306,
+      "step": 840500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.464330688730093e-06,
+      "loss": 0.5298,
+      "step": 841000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.4378528829564485e-06,
+      "loss": 0.5311,
+      "step": 841500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.411375077182805e-06,
+      "loss": 0.5278,
+      "step": 842000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.38489727140916e-06,
+      "loss": 0.5333,
+      "step": 842500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.358419465635515e-06,
+      "loss": 0.5287,
+      "step": 843000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.331941659861871e-06,
+      "loss": 0.5331,
+      "step": 843500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.305463854088227e-06,
+      "loss": 0.5292,
+      "step": 844000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.278986048314582e-06,
+      "loss": 0.531,
+      "step": 844500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.252508242540937e-06,
+      "loss": 0.5256,
+      "step": 845000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.226030436767293e-06,
+      "loss": 0.5318,
+      "step": 845500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.199552630993649e-06,
+      "loss": 0.5339,
+      "step": 846000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.173074825220004e-06,
+      "loss": 0.5295,
+      "step": 846500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.14659701944636e-06,
+      "loss": 0.5283,
+      "step": 847000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.120119213672715e-06,
+      "loss": 0.5308,
+      "step": 847500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.093641407899071e-06,
+      "loss": 0.5325,
+      "step": 848000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.067163602125426e-06,
+      "loss": 0.5287,
+      "step": 848500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.0406857963517825e-06,
+      "loss": 0.5324,
+      "step": 849000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.014207990578138e-06,
+      "loss": 0.5285,
+      "step": 849500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.987730184804493e-06,
+      "loss": 0.5337,
+      "step": 850000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.961252379030849e-06,
+      "loss": 0.5261,
+      "step": 850500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.934774573257205e-06,
+      "loss": 0.5285,
+      "step": 851000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.908296767483561e-06,
+      "loss": 0.5289,
+      "step": 851500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.881818961709915e-06,
+      "loss": 0.5215,
+      "step": 852000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.8553411559362715e-06,
+      "loss": 0.5288,
+      "step": 852500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.828863350162627e-06,
+      "loss": 0.5304,
+      "step": 853000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.802385544388983e-06,
+      "loss": 0.5312,
+      "step": 853500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.775907738615338e-06,
+      "loss": 0.5289,
+      "step": 854000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.749429932841694e-06,
+      "loss": 0.5315,
+      "step": 854500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.722952127068049e-06,
+      "loss": 0.5301,
+      "step": 855000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.696474321294405e-06,
+      "loss": 0.5289,
+      "step": 855500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.6699965155207604e-06,
+      "loss": 0.5297,
+      "step": 856000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.643518709747116e-06,
+      "loss": 0.5294,
+      "step": 856500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.617040903973471e-06,
+      "loss": 0.5273,
+      "step": 857000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.590563098199827e-06,
+      "loss": 0.53,
+      "step": 857500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.564085292426183e-06,
+      "loss": 0.5303,
+      "step": 858000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.537607486652539e-06,
+      "loss": 0.5275,
+      "step": 858500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.511129680878893e-06,
+      "loss": 0.5259,
+      "step": 859000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.484651875105249e-06,
+      "loss": 0.5306,
+      "step": 859500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.4581740693316056e-06,
+      "loss": 0.5284,
+      "step": 860000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.431696263557961e-06,
+      "loss": 0.5297,
+      "step": 860500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.405218457784316e-06,
+      "loss": 0.5275,
+      "step": 861000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.3787406520106715e-06,
+      "loss": 0.524,
+      "step": 861500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.352262846237028e-06,
+      "loss": 0.532,
+      "step": 862000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.325785040463383e-06,
+      "loss": 0.5268,
+      "step": 862500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.299307234689739e-06,
+      "loss": 0.5293,
+      "step": 863000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.272829428916094e-06,
+      "loss": 0.5275,
+      "step": 863500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.24635162314245e-06,
+      "loss": 0.5286,
+      "step": 864000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.219873817368805e-06,
+      "loss": 0.5276,
+      "step": 864500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.193396011595161e-06,
+      "loss": 0.5244,
+      "step": 865000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.166918205821517e-06,
+      "loss": 0.5311,
+      "step": 865500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.140440400047872e-06,
+      "loss": 0.5311,
+      "step": 866000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.113962594274227e-06,
+      "loss": 0.5275,
+      "step": 866500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.0874847885005835e-06,
+      "loss": 0.5301,
+      "step": 867000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.061006982726939e-06,
+      "loss": 0.5317,
+      "step": 867500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.034529176953294e-06,
+      "loss": 0.5317,
+      "step": 868000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.0080513711796494e-06,
+      "loss": 0.527,
+      "step": 868500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.981573565406006e-06,
+      "loss": 0.5273,
+      "step": 869000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.955095759632361e-06,
+      "loss": 0.5317,
+      "step": 869500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.928617953858717e-06,
+      "loss": 0.53,
+      "step": 870000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.902140148085072e-06,
+      "loss": 0.5331,
+      "step": 870500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.875662342311428e-06,
+      "loss": 0.5264,
+      "step": 871000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.849184536537784e-06,
+      "loss": 0.5248,
+      "step": 871500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.822706730764139e-06,
+      "loss": 0.5299,
+      "step": 872000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.796228924990495e-06,
+      "loss": 0.5311,
+      "step": 872500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.7697511192168503e-06,
+      "loss": 0.5259,
+      "step": 873000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.743273313443206e-06,
+      "loss": 0.5307,
+      "step": 873500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.7167955076695614e-06,
+      "loss": 0.531,
+      "step": 874000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.690317701895917e-06,
+      "loss": 0.5289,
+      "step": 874500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.6638398961222724e-06,
+      "loss": 0.5258,
+      "step": 875000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.637362090348628e-06,
+      "loss": 0.5234,
+      "step": 875500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.6108842845749835e-06,
+      "loss": 0.5307,
+      "step": 876000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.5844064788013393e-06,
+      "loss": 0.5306,
+      "step": 876500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.5579286730276946e-06,
+      "loss": 0.5274,
+      "step": 877000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.5314508672540508e-06,
+      "loss": 0.5302,
+      "step": 877500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.5049730614804057e-06,
+      "loss": 0.5353,
+      "step": 878000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.478495255706762e-06,
+      "loss": 0.5291,
+      "step": 878500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.4520174499331167e-06,
+      "loss": 0.5345,
+      "step": 879000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.425539644159473e-06,
+      "loss": 0.5257,
+      "step": 879500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3990618383858286e-06,
+      "loss": 0.5291,
+      "step": 880000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.372584032612184e-06,
+      "loss": 0.5278,
+      "step": 880500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.3461062268385397e-06,
+      "loss": 0.526,
+      "step": 881000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.319628421064895e-06,
+      "loss": 0.5259,
+      "step": 881500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.293150615291251e-06,
+      "loss": 0.5239,
+      "step": 882000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.266672809517606e-06,
+      "loss": 0.527,
+      "step": 882500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2401950037439623e-06,
+      "loss": 0.5272,
+      "step": 883000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.213717197970317e-06,
+      "loss": 0.5271,
+      "step": 883500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.1872393921966733e-06,
+      "loss": 0.524,
+      "step": 884000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.1607615864230287e-06,
+      "loss": 0.5264,
+      "step": 884500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1342837806493844e-06,
+      "loss": 0.5268,
+      "step": 885000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.10780597487574e-06,
+      "loss": 0.5227,
+      "step": 885500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0813281691020955e-06,
+      "loss": 0.5305,
+      "step": 886000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0548503633284512e-06,
+      "loss": 0.529,
+      "step": 886500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0283725575548065e-06,
+      "loss": 0.5294,
+      "step": 887000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0018947517811623e-06,
+      "loss": 0.5251,
+      "step": 887500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.9754169460075176e-06,
+      "loss": 0.5289,
+      "step": 888000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.9489391402338734e-06,
+      "loss": 0.5228,
+      "step": 888500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.922461334460229e-06,
+      "loss": 0.5237,
+      "step": 889000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.8959835286865844e-06,
+      "loss": 0.5238,
+      "step": 889500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.86950572291294e-06,
+      "loss": 0.5252,
+      "step": 890000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.8430279171392955e-06,
+      "loss": 0.5237,
+      "step": 890500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.8165501113656512e-06,
+      "loss": 0.5305,
+      "step": 891000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.7900723055920066e-06,
+      "loss": 0.5237,
+      "step": 891500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.7635944998183623e-06,
+      "loss": 0.5284,
+      "step": 892000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.737116694044718e-06,
+      "loss": 0.5248,
+      "step": 892500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.7106388882710734e-06,
+      "loss": 0.5307,
+      "step": 893000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.684161082497429e-06,
+      "loss": 0.5235,
+      "step": 893500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.6576832767237845e-06,
+      "loss": 0.5262,
+      "step": 894000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.6312054709501406e-06,
+      "loss": 0.5263,
+      "step": 894500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.604727665176496e-06,
+      "loss": 0.5275,
+      "step": 895000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5782498594028517e-06,
+      "loss": 0.521,
+      "step": 895500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.551772053629207e-06,
+      "loss": 0.5218,
+      "step": 896000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5252942478555628e-06,
+      "loss": 0.5237,
+      "step": 896500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4988164420819185e-06,
+      "loss": 0.5211,
+      "step": 897000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.472338636308274e-06,
+      "loss": 0.5294,
+      "step": 897500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4458608305346296e-06,
+      "loss": 0.5284,
+      "step": 898000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.419383024760985e-06,
+      "loss": 0.526,
+      "step": 898500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.3929052189873406e-06,
+      "loss": 0.5201,
+      "step": 899000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.366427413213696e-06,
+      "loss": 0.5281,
+      "step": 899500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.3399496074400517e-06,
+      "loss": 0.5218,
+      "step": 900000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.3134718016664075e-06,
+      "loss": 0.5236,
+      "step": 900500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.2869939958927628e-06,
+      "loss": 0.5235,
+      "step": 901000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.2605161901191185e-06,
+      "loss": 0.5283,
+      "step": 901500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.234038384345474e-06,
+      "loss": 0.5301,
+      "step": 902000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.2075605785718296e-06,
+      "loss": 0.5253,
+      "step": 902500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.181082772798185e-06,
+      "loss": 0.5238,
+      "step": 903000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.1546049670245407e-06,
+      "loss": 0.5297,
+      "step": 903500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.1281271612508964e-06,
+      "loss": 0.5215,
+      "step": 904000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.101649355477252e-06,
+      "loss": 0.5215,
+      "step": 904500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.075171549703608e-06,
+      "loss": 0.5296,
+      "step": 905000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.0486937439299632e-06,
+      "loss": 0.5266,
+      "step": 905500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.022215938156319e-06,
+      "loss": 0.5266,
+      "step": 906000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9957381323826743e-06,
+      "loss": 0.5243,
+      "step": 906500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.96926032660903e-06,
+      "loss": 0.5292,
+      "step": 907000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9427825208353854e-06,
+      "loss": 0.5277,
+      "step": 907500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.916304715061741e-06,
+      "loss": 0.5244,
+      "step": 908000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.8898269092880966e-06,
+      "loss": 0.5246,
+      "step": 908500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.8633491035144522e-06,
+      "loss": 0.5239,
+      "step": 909000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.8368712977408077e-06,
+      "loss": 0.528,
+      "step": 909500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.8103934919671632e-06,
+      "loss": 0.5221,
+      "step": 910000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.783915686193519e-06,
+      "loss": 0.5269,
+      "step": 910500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.7574378804198745e-06,
+      "loss": 0.5294,
+      "step": 911000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.73096007464623e-06,
+      "loss": 0.5273,
+      "step": 911500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.7044822688725856e-06,
+      "loss": 0.5216,
+      "step": 912000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.6780044630989411e-06,
+      "loss": 0.5272,
+      "step": 912500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.6515266573252967e-06,
+      "loss": 0.5192,
+      "step": 913000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.6250488515516522e-06,
+      "loss": 0.5241,
+      "step": 913500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.598571045778008e-06,
+      "loss": 0.5295,
+      "step": 914000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5720932400043637e-06,
+      "loss": 0.5249,
+      "step": 914500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5456154342307192e-06,
+      "loss": 0.524,
+      "step": 915000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5191376284570748e-06,
+      "loss": 0.5244,
+      "step": 915500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.4926598226834303e-06,
+      "loss": 0.5259,
+      "step": 916000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.4661820169097858e-06,
+      "loss": 0.5267,
+      "step": 916500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.4397042111361416e-06,
+      "loss": 0.5299,
+      "step": 917000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.4132264053624971e-06,
+      "loss": 0.5209,
+      "step": 917500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.3867485995888526e-06,
+      "loss": 0.5188,
+      "step": 918000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.3602707938152084e-06,
+      "loss": 0.5218,
+      "step": 918500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.333792988041564e-06,
+      "loss": 0.5257,
+      "step": 919000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.3073151822679195e-06,
+      "loss": 0.5247,
+      "step": 919500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.280837376494275e-06,
+      "loss": 0.5266,
+      "step": 920000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.2543595707206305e-06,
+      "loss": 0.5174,
+      "step": 920500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.227881764946986e-06,
+      "loss": 0.5266,
+      "step": 921000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.2014039591733416e-06,
+      "loss": 0.5224,
+      "step": 921500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1749261533996973e-06,
+      "loss": 0.5277,
+      "step": 922000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.148448347626053e-06,
+      "loss": 0.5233,
+      "step": 922500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1219705418524086e-06,
+      "loss": 0.5268,
+      "step": 923000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0954927360787642e-06,
+      "loss": 0.5271,
+      "step": 923500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0690149303051197e-06,
+      "loss": 0.519,
+      "step": 924000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0425371245314752e-06,
+      "loss": 0.5221,
+      "step": 924500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0160593187578308e-06,
+      "loss": 0.5267,
+      "step": 925000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.895815129841863e-07,
+      "loss": 0.5229,
+      "step": 925500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.63103707210542e-07,
+      "loss": 0.5221,
+      "step": 926000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.366259014368975e-07,
+      "loss": 0.524,
+      "step": 926500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.101480956632532e-07,
+      "loss": 0.522,
+      "step": 927000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 8.836702898896089e-07,
+      "loss": 0.5223,
+      "step": 927500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 8.571924841159644e-07,
+      "loss": 0.5236,
+      "step": 928000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 8.307146783423199e-07,
+      "loss": 0.5259,
+      "step": 928500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 8.042368725686756e-07,
+      "loss": 0.5233,
+      "step": 929000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.777590667950311e-07,
+      "loss": 0.5299,
+      "step": 929500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.512812610213866e-07,
+      "loss": 0.5265,
+      "step": 930000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.248034552477423e-07,
+      "loss": 0.5208,
+      "step": 930500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.983256494740979e-07,
+      "loss": 0.5226,
+      "step": 931000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.718478437004535e-07,
+      "loss": 0.5257,
+      "step": 931500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.45370037926809e-07,
+      "loss": 0.525,
+      "step": 932000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.188922321531645e-07,
+      "loss": 0.5288,
+      "step": 932500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.924144263795203e-07,
+      "loss": 0.5266,
+      "step": 933000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.659366206058758e-07,
+      "loss": 0.5248,
+      "step": 933500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.394588148322313e-07,
+      "loss": 0.5185,
+      "step": 934000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.129810090585869e-07,
+      "loss": 0.5248,
+      "step": 934500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.865032032849425e-07,
+      "loss": 0.5213,
+      "step": 935000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.6002539751129815e-07,
+      "loss": 0.5222,
+      "step": 935500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.335475917376537e-07,
+      "loss": 0.5279,
+      "step": 936000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.070697859640093e-07,
+      "loss": 0.5254,
+      "step": 936500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.805919801903648e-07,
+      "loss": 0.5261,
+      "step": 937000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.5411417441672045e-07,
+      "loss": 0.519,
+      "step": 937500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.27636368643076e-07,
+      "loss": 0.5252,
+      "step": 938000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.011585628694316e-07,
+      "loss": 0.5254,
+      "step": 938500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.7468075709578716e-07,
+      "loss": 0.5232,
+      "step": 939000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.4820295132214275e-07,
+      "loss": 0.523,
+      "step": 939500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2172514554849836e-07,
+      "loss": 0.5235,
+      "step": 940000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9524733977485395e-07,
+      "loss": 0.5209,
+      "step": 940500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.687695340012095e-07,
+      "loss": 0.5203,
+      "step": 941000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.422917282275651e-07,
+      "loss": 0.5236,
+      "step": 941500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1581392245392068e-07,
+      "loss": 0.5236,
+      "step": 942000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.933611668027626e-08,
+      "loss": 0.5225,
+      "step": 942500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.285831090663185e-08,
+      "loss": 0.5259,
+      "step": 943000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.6380505132987426e-08,
+      "loss": 0.5201,
+      "step": 943500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.902699359343012e-09,
+      "loss": 0.5231,
+      "step": 944000
+    },
+    {
+      "epoch": 2.39,
+      "step": 944187,
+      "total_flos": 7.959444107321715e+18,
+      "train_loss": 0.6286116640401175,
+      "train_runtime": 410211.7129,
+      "train_samples_per_second": 73.655,
+      "train_steps_per_second": 2.302
+    }
+  ],
+  "max_steps": 944187,
+  "num_train_epochs": 3,
+  "total_flos": 7.959444107321715e+18,
+  "trial_name": null,
+  "trial_params": null
+}