diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,9541 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "global_step": 793047,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9968476017184356e-05,
+      "loss": 2.2908,
+      "step": 500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.993695203436871e-05,
+      "loss": 1.9499,
+      "step": 1000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.990542805155306e-05,
+      "loss": 1.8515,
+      "step": 1500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9873904068737416e-05,
+      "loss": 1.7764,
+      "step": 2000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.984238008592177e-05,
+      "loss": 1.7312,
+      "step": 2500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.981085610310612e-05,
+      "loss": 1.7093,
+      "step": 3000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9779332120290476e-05,
+      "loss": 1.6757,
+      "step": 3500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.974780813747483e-05,
+      "loss": 1.6496,
+      "step": 4000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.971628415465918e-05,
+      "loss": 1.6336,
+      "step": 4500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.968476017184354e-05,
+      "loss": 1.6159,
+      "step": 5000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.965323618902789e-05,
+      "loss": 1.6152,
+      "step": 5500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9621712206212244e-05,
+      "loss": 1.5905,
+      "step": 6000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.95901882233966e-05,
+      "loss": 1.5704,
+      "step": 6500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.955866424058096e-05,
+      "loss": 1.5667,
+      "step": 7000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9527140257765304e-05,
+      "loss": 1.5538,
+      "step": 7500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.949561627494966e-05,
+      "loss": 1.5581,
+      "step": 8000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.946409229213401e-05,
+      "loss": 1.5497,
+      "step": 8500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9432568309318364e-05,
+      "loss": 1.5373,
+      "step": 9000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9401044326502724e-05,
+      "loss": 1.5261,
+      "step": 9500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.936952034368707e-05,
+      "loss": 1.5222,
+      "step": 10000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9337996360871425e-05,
+      "loss": 1.5082,
+      "step": 10500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.930647237805578e-05,
+      "loss": 1.5114,
+      "step": 11000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.927494839524013e-05,
+      "loss": 1.5152,
+      "step": 11500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9243424412424485e-05,
+      "loss": 1.5012,
+      "step": 12000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9211900429608845e-05,
+      "loss": 1.4883,
+      "step": 12500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.918037644679319e-05,
+      "loss": 1.5023,
+      "step": 13000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9148852463977545e-05,
+      "loss": 1.4903,
+      "step": 13500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.91173284811619e-05,
+      "loss": 1.4818,
+      "step": 14000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.908580449834625e-05,
+      "loss": 1.4761,
+      "step": 14500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.905428051553061e-05,
+      "loss": 1.4685,
+      "step": 15000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.902275653271496e-05,
+      "loss": 1.4781,
+      "step": 15500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.899123254989931e-05,
+      "loss": 1.474,
+      "step": 16000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8959708567083666e-05,
+      "loss": 1.4616,
+      "step": 16500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.892818458426802e-05,
+      "loss": 1.4584,
+      "step": 17000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.889666060145238e-05,
+      "loss": 1.4573,
+      "step": 17500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.886513661863673e-05,
+      "loss": 1.4563,
+      "step": 18000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.883361263582108e-05,
+      "loss": 1.4433,
+      "step": 18500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.880208865300543e-05,
+      "loss": 1.4416,
+      "step": 19000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8770564670189786e-05,
+      "loss": 1.4432,
+      "step": 19500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8739040687374147e-05,
+      "loss": 1.4404,
+      "step": 20000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.87075167045585e-05,
+      "loss": 1.4457,
+      "step": 20500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8675992721742853e-05,
+      "loss": 1.4489,
+      "step": 21000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.86444687389272e-05,
+      "loss": 1.4411,
+      "step": 21500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8612944756111554e-05,
+      "loss": 1.4365,
+      "step": 22000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.858142077329591e-05,
+      "loss": 1.4291,
+      "step": 22500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.854989679048027e-05,
+      "loss": 1.4225,
+      "step": 23000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.851837280766462e-05,
+      "loss": 1.4206,
+      "step": 23500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.848684882484897e-05,
+      "loss": 1.4259,
+      "step": 24000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.845532484203332e-05,
+      "loss": 1.4254,
+      "step": 24500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8423800859217674e-05,
+      "loss": 1.4368,
+      "step": 25000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8392276876402034e-05,
+      "loss": 1.4213,
+      "step": 25500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.836075289358639e-05,
+      "loss": 1.4223,
+      "step": 26000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.832922891077074e-05,
+      "loss": 1.4233,
+      "step": 26500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.829770492795509e-05,
+      "loss": 1.4174,
+      "step": 27000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.826618094513944e-05,
+      "loss": 1.4244,
+      "step": 27500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.82346569623238e-05,
+      "loss": 1.4063,
+      "step": 28000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8203132979508155e-05,
+      "loss": 1.4058,
+      "step": 28500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.817160899669251e-05,
+      "loss": 1.4124,
+      "step": 29000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.814008501387686e-05,
+      "loss": 1.3993,
+      "step": 29500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.810856103106121e-05,
+      "loss": 1.4086,
+      "step": 30000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.807703704824557e-05,
+      "loss": 1.4062,
+      "step": 30500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.804551306542992e-05,
+      "loss": 1.3945,
+      "step": 31000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8013989082614276e-05,
+      "loss": 1.4016,
+      "step": 31500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.798246509979863e-05,
+      "loss": 1.4041,
+      "step": 32000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7950941116982976e-05,
+      "loss": 1.4049,
+      "step": 32500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7919417134167336e-05,
+      "loss": 1.4018,
+      "step": 33000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.788789315135169e-05,
+      "loss": 1.3937,
+      "step": 33500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.785636916853604e-05,
+      "loss": 1.3996,
+      "step": 34000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7824845185720396e-05,
+      "loss": 1.3855,
+      "step": 34500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.779332120290475e-05,
+      "loss": 1.392,
+      "step": 35000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7761797220089096e-05,
+      "loss": 1.3924,
+      "step": 35500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7730273237273457e-05,
+      "loss": 1.3935,
+      "step": 36000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.769874925445781e-05,
+      "loss": 1.3912,
+      "step": 36500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7667225271642163e-05,
+      "loss": 1.3808,
+      "step": 37000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.763570128882652e-05,
+      "loss": 1.3806,
+      "step": 37500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.760417730601087e-05,
+      "loss": 1.3829,
+      "step": 38000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7572653323195224e-05,
+      "loss": 1.3838,
+      "step": 38500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.754112934037958e-05,
+      "loss": 1.3819,
+      "step": 39000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.750960535756393e-05,
+      "loss": 1.3824,
+      "step": 39500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7478081374748284e-05,
+      "loss": 1.3787,
+      "step": 40000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.744655739193264e-05,
+      "loss": 1.3675,
+      "step": 40500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.741503340911699e-05,
+      "loss": 1.3719,
+      "step": 41000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7383509426301344e-05,
+      "loss": 1.3732,
+      "step": 41500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.73519854434857e-05,
+      "loss": 1.3658,
+      "step": 42000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.732046146067005e-05,
+      "loss": 1.3841,
+      "step": 42500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7288937477854405e-05,
+      "loss": 1.3722,
+      "step": 43000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.725741349503876e-05,
+      "loss": 1.3592,
+      "step": 43500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.722588951222311e-05,
+      "loss": 1.3792,
+      "step": 44000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7194365529407465e-05,
+      "loss": 1.3712,
+      "step": 44500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.716284154659182e-05,
+      "loss": 1.374,
+      "step": 45000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.713131756377617e-05,
+      "loss": 1.3716,
+      "step": 45500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7099793580960525e-05,
+      "loss": 1.367,
+      "step": 46000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.706826959814488e-05,
+      "loss": 1.364,
+      "step": 46500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.703674561532923e-05,
+      "loss": 1.3562,
+      "step": 47000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7005221632513586e-05,
+      "loss": 1.3589,
+      "step": 47500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.697369764969794e-05,
+      "loss": 1.3608,
+      "step": 48000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.694217366688229e-05,
+      "loss": 1.3528,
+      "step": 48500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6910649684066646e-05,
+      "loss": 1.3687,
+      "step": 49000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6879125701251e-05,
+      "loss": 1.3676,
+      "step": 49500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.684760171843535e-05,
+      "loss": 1.3578,
+      "step": 50000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6816077735619706e-05,
+      "loss": 1.3606,
+      "step": 50500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.678455375280406e-05,
+      "loss": 1.3575,
+      "step": 51000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.675302976998841e-05,
+      "loss": 1.3589,
+      "step": 51500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.672150578717277e-05,
+      "loss": 1.3624,
+      "step": 52000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.668998180435712e-05,
+      "loss": 1.353,
+      "step": 52500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.665845782154147e-05,
+      "loss": 1.3658,
+      "step": 53000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.662693383872583e-05,
+      "loss": 1.3542,
+      "step": 53500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.659540985591018e-05,
+      "loss": 1.3519,
+      "step": 54000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.656388587309454e-05,
+      "loss": 1.3473,
+      "step": 54500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.653236189027889e-05,
+      "loss": 1.3512,
+      "step": 55000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.650083790746324e-05,
+      "loss": 1.3439,
+      "step": 55500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6469313924647594e-05,
+      "loss": 1.3446,
+      "step": 56000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.643778994183195e-05,
+      "loss": 1.3454,
+      "step": 56500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.64062659590163e-05,
+      "loss": 1.35,
+      "step": 57000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.637474197620066e-05,
+      "loss": 1.3424,
+      "step": 57500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.634321799338501e-05,
+      "loss": 1.3411,
+      "step": 58000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.631169401056936e-05,
+      "loss": 1.3423,
+      "step": 58500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6280170027753715e-05,
+      "loss": 1.337,
+      "step": 59000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.624864604493807e-05,
+      "loss": 1.3488,
+      "step": 59500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.621712206212243e-05,
+      "loss": 1.3368,
+      "step": 60000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.618559807930678e-05,
+      "loss": 1.3392,
+      "step": 60500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.615407409649113e-05,
+      "loss": 1.3441,
+      "step": 61000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.612255011367548e-05,
+      "loss": 1.3338,
+      "step": 61500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6091026130859835e-05,
+      "loss": 1.3369,
+      "step": 62000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6059502148044195e-05,
+      "loss": 1.3319,
+      "step": 62500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.602797816522855e-05,
+      "loss": 1.3293,
+      "step": 63000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.5996454182412896e-05,
+      "loss": 1.3421,
+      "step": 63500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.596493019959725e-05,
+      "loss": 1.3262,
+      "step": 64000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.59334062167816e-05,
+      "loss": 1.3422,
+      "step": 64500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.590188223396596e-05,
+      "loss": 1.3312,
+      "step": 65000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5870358251150316e-05,
+      "loss": 1.3497,
+      "step": 65500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.583883426833467e-05,
+      "loss": 1.3355,
+      "step": 66000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5807310285519016e-05,
+      "loss": 1.3249,
+      "step": 66500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.577578630270337e-05,
+      "loss": 1.3319,
+      "step": 67000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.574426231988772e-05,
+      "loss": 1.326,
+      "step": 67500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.571273833707208e-05,
+      "loss": 1.3303,
+      "step": 68000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.568121435425644e-05,
+      "loss": 1.3296,
+      "step": 68500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.564969037144078e-05,
+      "loss": 1.3301,
+      "step": 69000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.561816638862514e-05,
+      "loss": 1.3399,
+      "step": 69500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.558664240580949e-05,
+      "loss": 1.3273,
+      "step": 70000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.555511842299385e-05,
+      "loss": 1.3289,
+      "step": 70500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5523594440178204e-05,
+      "loss": 1.3256,
+      "step": 71000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.549207045736256e-05,
+      "loss": 1.3309,
+      "step": 71500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5460546474546904e-05,
+      "loss": 1.3286,
+      "step": 72000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.542902249173126e-05,
+      "loss": 1.3211,
+      "step": 72500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.539749850891562e-05,
+      "loss": 1.3264,
+      "step": 73000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.536597452609997e-05,
+      "loss": 1.3232,
+      "step": 73500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5334450543284324e-05,
+      "loss": 1.3264,
+      "step": 74000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.530292656046868e-05,
+      "loss": 1.3244,
+      "step": 74500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5271402577653025e-05,
+      "loss": 1.3139,
+      "step": 75000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5239878594837385e-05,
+      "loss": 1.3191,
+      "step": 75500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.520835461202174e-05,
+      "loss": 1.3205,
+      "step": 76000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.517683062920609e-05,
+      "loss": 1.3061,
+      "step": 76500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5145306646390445e-05,
+      "loss": 1.3169,
+      "step": 77000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.511378266357479e-05,
+      "loss": 1.3283,
+      "step": 77500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5082258680759145e-05,
+      "loss": 1.3166,
+      "step": 78000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5050734697943505e-05,
+      "loss": 1.3215,
+      "step": 78500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.501921071512786e-05,
+      "loss": 1.327,
+      "step": 79000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.498768673231221e-05,
+      "loss": 1.3162,
+      "step": 79500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4956162749496566e-05,
+      "loss": 1.3175,
+      "step": 80000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.492463876668091e-05,
+      "loss": 1.3113,
+      "step": 80500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.489311478386527e-05,
+      "loss": 1.3117,
+      "step": 81000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4861590801049626e-05,
+      "loss": 1.3138,
+      "step": 81500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.483006681823398e-05,
+      "loss": 1.3192,
+      "step": 82000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.479854283541833e-05,
+      "loss": 1.3126,
+      "step": 82500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4767018852602686e-05,
+      "loss": 1.3262,
+      "step": 83000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.473549486978704e-05,
+      "loss": 1.313,
+      "step": 83500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.470397088697139e-05,
+      "loss": 1.3137,
+      "step": 84000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.467244690415575e-05,
+      "loss": 1.3055,
+      "step": 84500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.46409229213401e-05,
+      "loss": 1.3179,
+      "step": 85000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4609398938524453e-05,
+      "loss": 1.3109,
+      "step": 85500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.457787495570881e-05,
+      "loss": 1.3089,
+      "step": 86000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.454635097289316e-05,
+      "loss": 1.318,
+      "step": 86500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4514826990077514e-05,
+      "loss": 1.3125,
+      "step": 87000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.448330300726187e-05,
+      "loss": 1.3045,
+      "step": 87500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.445177902444622e-05,
+      "loss": 1.3032,
+      "step": 88000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4420255041630574e-05,
+      "loss": 1.3108,
+      "step": 88500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.438873105881493e-05,
+      "loss": 1.3126,
+      "step": 89000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.435720707599928e-05,
+      "loss": 1.3027,
+      "step": 89500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4325683093183634e-05,
+      "loss": 1.3019,
+      "step": 90000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.429415911036799e-05,
+      "loss": 1.3121,
+      "step": 90500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.426263512755234e-05,
+      "loss": 1.3049,
+      "step": 91000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4231111144736695e-05,
+      "loss": 1.3035,
+      "step": 91500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.419958716192105e-05,
+      "loss": 1.3013,
+      "step": 92000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.41680631791054e-05,
+      "loss": 1.3063,
+      "step": 92500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4136539196289755e-05,
+      "loss": 1.3069,
+      "step": 93000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.410501521347411e-05,
+      "loss": 1.2973,
+      "step": 93500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.407349123065846e-05,
+      "loss": 1.3056,
+      "step": 94000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4041967247842815e-05,
+      "loss": 1.3061,
+      "step": 94500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.401044326502717e-05,
+      "loss": 1.3061,
+      "step": 95000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.397891928221152e-05,
+      "loss": 1.2963,
+      "step": 95500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.3947395299395876e-05,
+      "loss": 1.2948,
+      "step": 96000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.391587131658023e-05,
+      "loss": 1.3008,
+      "step": 96500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.388434733376459e-05,
+      "loss": 1.2946,
+      "step": 97000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3852823350948936e-05,
+      "loss": 1.2885,
+      "step": 97500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.382129936813329e-05,
+      "loss": 1.2894,
+      "step": 98000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.378977538531764e-05,
+      "loss": 1.2977,
+      "step": 98500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3758251402501996e-05,
+      "loss": 1.2929,
+      "step": 99000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.372672741968635e-05,
+      "loss": 1.2893,
+      "step": 99500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.36952034368707e-05,
+      "loss": 1.2903,
+      "step": 100000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3663679454055057e-05,
+      "loss": 1.294,
+      "step": 100500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.363215547123941e-05,
+      "loss": 1.296,
+      "step": 101000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3600631488423763e-05,
+      "loss": 1.2844,
+      "step": 101500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.356910750560812e-05,
+      "loss": 1.2975,
+      "step": 102000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.353758352279248e-05,
+      "loss": 1.2952,
+      "step": 102500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3506059539976824e-05,
+      "loss": 1.2894,
+      "step": 103000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.347453555716118e-05,
+      "loss": 1.2931,
+      "step": 103500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.344301157434553e-05,
+      "loss": 1.2907,
+      "step": 104000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3411487591529884e-05,
+      "loss": 1.3006,
+      "step": 104500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3379963608714244e-05,
+      "loss": 1.2961,
+      "step": 105000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.33484396258986e-05,
+      "loss": 1.2926,
+      "step": 105500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3316915643082944e-05,
+      "loss": 1.2992,
+      "step": 106000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.32853916602673e-05,
+      "loss": 1.2856,
+      "step": 106500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.325386767745165e-05,
+      "loss": 1.2947,
+      "step": 107000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.322234369463601e-05,
+      "loss": 1.2989,
+      "step": 107500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3190819711820365e-05,
+      "loss": 1.2981,
+      "step": 108000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.315929572900471e-05,
+      "loss": 1.284,
+      "step": 108500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3127771746189065e-05,
+      "loss": 1.282,
+      "step": 109000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.309624776337342e-05,
+      "loss": 1.281,
+      "step": 109500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.306472378055777e-05,
+      "loss": 1.2902,
+      "step": 110000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.303319979774213e-05,
+      "loss": 1.2881,
+      "step": 110500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3001675814926486e-05,
+      "loss": 1.2849,
+      "step": 111000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.297015183211083e-05,
+      "loss": 1.2867,
+      "step": 111500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2938627849295186e-05,
+      "loss": 1.2795,
+      "step": 112000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.290710386647954e-05,
+      "loss": 1.2809,
+      "step": 112500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.28755798836639e-05,
+      "loss": 1.2786,
+      "step": 113000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.284405590084825e-05,
+      "loss": 1.2806,
+      "step": 113500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2812531918032606e-05,
+      "loss": 1.2939,
+      "step": 114000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.278100793521695e-05,
+      "loss": 1.2797,
+      "step": 114500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2749483952401306e-05,
+      "loss": 1.2836,
+      "step": 115000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2717959969585666e-05,
+      "loss": 1.274,
+      "step": 115500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.268643598677002e-05,
+      "loss": 1.2819,
+      "step": 116000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.265491200395437e-05,
+      "loss": 1.2758,
+      "step": 116500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.262338802113872e-05,
+      "loss": 1.2819,
+      "step": 117000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2591864038323073e-05,
+      "loss": 1.2888,
+      "step": 117500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2560340055507434e-05,
+      "loss": 1.2723,
+      "step": 118000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.252881607269179e-05,
+      "loss": 1.2848,
+      "step": 118500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.249729208987614e-05,
+      "loss": 1.2675,
+      "step": 119000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2465768107060494e-05,
+      "loss": 1.2842,
+      "step": 119500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.243424412424484e-05,
+      "loss": 1.2751,
+      "step": 120000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.24027201414292e-05,
+      "loss": 1.278,
+      "step": 120500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2371196158613554e-05,
+      "loss": 1.2786,
+      "step": 121000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.233967217579791e-05,
+      "loss": 1.2816,
+      "step": 121500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.230814819298226e-05,
+      "loss": 1.2707,
+      "step": 122000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.227662421016661e-05,
+      "loss": 1.2715,
+      "step": 122500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.224510022735096e-05,
+      "loss": 1.2801,
+      "step": 123000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.221357624453532e-05,
+      "loss": 1.2818,
+      "step": 123500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2182052261719675e-05,
+      "loss": 1.2813,
+      "step": 124000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.215052827890403e-05,
+      "loss": 1.2805,
+      "step": 124500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.211900429608838e-05,
+      "loss": 1.2766,
+      "step": 125000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.208748031327273e-05,
+      "loss": 1.2736,
+      "step": 125500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.205595633045709e-05,
+      "loss": 1.2794,
+      "step": 126000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.202443234764144e-05,
+      "loss": 1.2829,
+      "step": 126500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1992908364825795e-05,
+      "loss": 1.274,
+      "step": 127000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.196138438201015e-05,
+      "loss": 1.2724,
+      "step": 127500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.19298603991945e-05,
+      "loss": 1.2762,
+      "step": 128000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1898336416378856e-05,
+      "loss": 1.2685,
+      "step": 128500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.186681243356321e-05,
+      "loss": 1.2634,
+      "step": 129000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.183528845074756e-05,
+      "loss": 1.2662,
+      "step": 129500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1803764467931916e-05,
+      "loss": 1.2812,
+      "step": 130000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.177224048511627e-05,
+      "loss": 1.2663,
+      "step": 130500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.174071650230062e-05,
+      "loss": 1.279,
+      "step": 131000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1709192519484976e-05,
+      "loss": 1.2679,
+      "step": 131500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.167766853666933e-05,
+      "loss": 1.2692,
+      "step": 132000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.164614455385368e-05,
+      "loss": 1.2751,
+      "step": 132500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.161462057103804e-05,
+      "loss": 1.272,
+      "step": 133000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.158309658822239e-05,
+      "loss": 1.2676,
+      "step": 133500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1551572605406744e-05,
+      "loss": 1.2581,
+      "step": 134000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.15200486225911e-05,
+      "loss": 1.2685,
+      "step": 134500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.148852463977545e-05,
+      "loss": 1.2698,
+      "step": 135000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1457000656959804e-05,
+      "loss": 1.2633,
+      "step": 135500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.142547667414416e-05,
+      "loss": 1.267,
+      "step": 136000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.139395269132851e-05,
+      "loss": 1.2745,
+      "step": 136500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1362428708512864e-05,
+      "loss": 1.2688,
+      "step": 137000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.133090472569722e-05,
+      "loss": 1.2691,
+      "step": 137500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.129938074288157e-05,
+      "loss": 1.2614,
+      "step": 138000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1267856760065924e-05,
+      "loss": 1.2656,
+      "step": 138500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.123633277725028e-05,
+      "loss": 1.2741,
+      "step": 139000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.120480879443463e-05,
+      "loss": 1.2673,
+      "step": 139500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1173284811618985e-05,
+      "loss": 1.2641,
+      "step": 140000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.114176082880334e-05,
+      "loss": 1.2654,
+      "step": 140500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.111023684598769e-05,
+      "loss": 1.2634,
+      "step": 141000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1078712863172045e-05,
+      "loss": 1.2698,
+      "step": 141500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1047188880356405e-05,
+      "loss": 1.2551,
+      "step": 142000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.101566489754075e-05,
+      "loss": 1.2571,
+      "step": 142500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.0984140914725105e-05,
+      "loss": 1.2685,
+      "step": 143000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.095261693190946e-05,
+      "loss": 1.2608,
+      "step": 143500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.092109294909381e-05,
+      "loss": 1.2628,
+      "step": 144000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0889568966278166e-05,
+      "loss": 1.2666,
+      "step": 144500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.085804498346252e-05,
+      "loss": 1.2568,
+      "step": 145000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.082652100064687e-05,
+      "loss": 1.2561,
+      "step": 145500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0794997017831226e-05,
+      "loss": 1.2623,
+      "step": 146000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.076347303501558e-05,
+      "loss": 1.2662,
+      "step": 146500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.073194905219993e-05,
+      "loss": 1.2621,
+      "step": 147000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.070042506938429e-05,
+      "loss": 1.2644,
+      "step": 147500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.066890108656864e-05,
+      "loss": 1.2604,
+      "step": 148000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.063737710375299e-05,
+      "loss": 1.2581,
+      "step": 148500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.060585312093735e-05,
+      "loss": 1.2607,
+      "step": 149000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.05743291381217e-05,
+      "loss": 1.2517,
+      "step": 149500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.054280515530606e-05,
+      "loss": 1.2597,
+      "step": 150000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0511281172490414e-05,
+      "loss": 1.2582,
+      "step": 150500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.047975718967476e-05,
+      "loss": 1.249,
+      "step": 151000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0448233206859114e-05,
+      "loss": 1.2691,
+      "step": 151500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.041670922404347e-05,
+      "loss": 1.2587,
+      "step": 152000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.038518524122783e-05,
+      "loss": 1.2517,
+      "step": 152500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.035366125841218e-05,
+      "loss": 1.2535,
+      "step": 153000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.032213727559653e-05,
+      "loss": 1.2532,
+      "step": 153500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.029061329278088e-05,
+      "loss": 1.2524,
+      "step": 154000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0259089309965234e-05,
+      "loss": 1.2587,
+      "step": 154500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.022756532714959e-05,
+      "loss": 1.2537,
+      "step": 155000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.019604134433395e-05,
+      "loss": 1.2535,
+      "step": 155500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.01645173615183e-05,
+      "loss": 1.2587,
+      "step": 156000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.013299337870265e-05,
+      "loss": 1.2454,
+      "step": 156500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0101469395887e-05,
+      "loss": 1.245,
+      "step": 157000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0069945413071355e-05,
+      "loss": 1.2485,
+      "step": 157500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0038421430255715e-05,
+      "loss": 1.2613,
+      "step": 158000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.000689744744007e-05,
+      "loss": 1.2438,
+      "step": 158500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.997537346462442e-05,
+      "loss": 1.2495,
+      "step": 159000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.994384948180877e-05,
+      "loss": 1.2535,
+      "step": 159500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.991232549899312e-05,
+      "loss": 1.2556,
+      "step": 160000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.988080151617748e-05,
+      "loss": 1.2549,
+      "step": 160500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9849277533361836e-05,
+      "loss": 1.2447,
+      "step": 161000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.981775355054619e-05,
+      "loss": 1.2486,
+      "step": 161500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9786229567730536e-05,
+      "loss": 1.253,
+      "step": 162000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.975470558491489e-05,
+      "loss": 1.2464,
+      "step": 162500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.972318160209925e-05,
+      "loss": 1.2555,
+      "step": 163000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.96916576192836e-05,
+      "loss": 1.2475,
+      "step": 163500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9660133636467956e-05,
+      "loss": 1.2646,
+      "step": 164000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.962860965365231e-05,
+      "loss": 1.2571,
+      "step": 164500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9597085670836657e-05,
+      "loss": 1.2415,
+      "step": 165000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.956556168802101e-05,
+      "loss": 1.2467,
+      "step": 165500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.953403770520537e-05,
+      "loss": 1.2416,
+      "step": 166000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9502513722389724e-05,
+      "loss": 1.249,
+      "step": 166500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.947098973957408e-05,
+      "loss": 1.2416,
+      "step": 167000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9439465756758424e-05,
+      "loss": 1.2555,
+      "step": 167500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.940794177394278e-05,
+      "loss": 1.2515,
+      "step": 168000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.937641779112714e-05,
+      "loss": 1.2445,
+      "step": 168500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.934489380831149e-05,
+      "loss": 1.2498,
+      "step": 169000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9313369825495844e-05,
+      "loss": 1.2459,
+      "step": 169500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.92818458426802e-05,
+      "loss": 1.2501,
+      "step": 170000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9250321859864544e-05,
+      "loss": 1.2514,
+      "step": 170500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9218797877048905e-05,
+      "loss": 1.2483,
+      "step": 171000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.918727389423326e-05,
+      "loss": 1.2369,
+      "step": 171500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.915574991141761e-05,
+      "loss": 1.2338,
+      "step": 172000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9124225928601965e-05,
+      "loss": 1.2439,
+      "step": 172500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.909270194578632e-05,
+      "loss": 1.2419,
+      "step": 173000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.906117796297067e-05,
+      "loss": 1.2442,
+      "step": 173500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9029653980155025e-05,
+      "loss": 1.2567,
+      "step": 174000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.899812999733938e-05,
+      "loss": 1.2435,
+      "step": 174500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.896660601452373e-05,
+      "loss": 1.2391,
+      "step": 175000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8935082031708086e-05,
+      "loss": 1.2427,
+      "step": 175500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.890355804889244e-05,
+      "loss": 1.2344,
+      "step": 176000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.887203406607679e-05,
+      "loss": 1.237,
+      "step": 176500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8840510083261146e-05,
+      "loss": 1.2406,
+      "step": 177000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.88089861004455e-05,
+      "loss": 1.2416,
+      "step": 177500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.877746211762985e-05,
+      "loss": 1.2482,
+      "step": 178000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8745938134814206e-05,
+      "loss": 1.2516,
+      "step": 178500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.871441415199856e-05,
+      "loss": 1.2459,
+      "step": 179000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.868289016918291e-05,
+      "loss": 1.2477,
+      "step": 179500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8651366186367266e-05,
+      "loss": 1.2469,
+      "step": 180000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.861984220355162e-05,
+      "loss": 1.2431,
+      "step": 180500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.858831822073597e-05,
+      "loss": 1.2371,
+      "step": 181000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.855679423792033e-05,
+      "loss": 1.2475,
+      "step": 181500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.852527025510468e-05,
+      "loss": 1.2437,
+      "step": 182000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8493746272289034e-05,
+      "loss": 1.246,
+      "step": 182500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.846222228947339e-05,
+      "loss": 1.2453,
+      "step": 183000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.843069830665774e-05,
+      "loss": 1.2343,
+      "step": 183500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8399174323842094e-05,
+      "loss": 1.2378,
+      "step": 184000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.836765034102645e-05,
+      "loss": 1.2409,
+      "step": 184500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.83361263582108e-05,
+      "loss": 1.2416,
+      "step": 185000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8304602375395154e-05,
+      "loss": 1.237,
+      "step": 185500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.827307839257951e-05,
+      "loss": 1.2444,
+      "step": 186000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.824155440976386e-05,
+      "loss": 1.2306,
+      "step": 186500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8210030426948215e-05,
+      "loss": 1.2276,
+      "step": 187000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.817850644413257e-05,
+      "loss": 1.2383,
+      "step": 187500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.814698246131692e-05,
+      "loss": 1.2423,
+      "step": 188000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8115458478501275e-05,
+      "loss": 1.2351,
+      "step": 188500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.808393449568563e-05,
+      "loss": 1.2293,
+      "step": 189000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.805241051286998e-05,
+      "loss": 1.2369,
+      "step": 189500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.802088653005434e-05,
+      "loss": 1.2348,
+      "step": 190000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.798936254723869e-05,
+      "loss": 1.2385,
+      "step": 190500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.795783856442304e-05,
+      "loss": 1.2393,
+      "step": 191000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7926314581607395e-05,
+      "loss": 1.2429,
+      "step": 191500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.789479059879175e-05,
+      "loss": 1.23,
+      "step": 192000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.786326661597611e-05,
+      "loss": 1.2318,
+      "step": 192500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7831742633160456e-05,
+      "loss": 1.2271,
+      "step": 193000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.780021865034481e-05,
+      "loss": 1.2345,
+      "step": 193500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.776869466752916e-05,
+      "loss": 1.2294,
+      "step": 194000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7737170684713516e-05,
+      "loss": 1.2528,
+      "step": 194500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7705646701897876e-05,
+      "loss": 1.23,
+      "step": 195000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.767412271908223e-05,
+      "loss": 1.2373,
+      "step": 195500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7642598736266576e-05,
+      "loss": 1.2306,
+      "step": 196000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.761107475345093e-05,
+      "loss": 1.2324,
+      "step": 196500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.757955077063528e-05,
+      "loss": 1.2337,
+      "step": 197000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7548026787819643e-05,
+      "loss": 1.2266,
+      "step": 197500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7516502805004e-05,
+      "loss": 1.2244,
+      "step": 198000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7484978822188344e-05,
+      "loss": 1.2234,
+      "step": 198500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.74534548393727e-05,
+      "loss": 1.2287,
+      "step": 199000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.742193085655705e-05,
+      "loss": 1.2306,
+      "step": 199500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7390406873741404e-05,
+      "loss": 1.2294,
+      "step": 200000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7358882890925764e-05,
+      "loss": 1.2289,
+      "step": 200500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.732735890811012e-05,
+      "loss": 1.2345,
+      "step": 201000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7295834925294464e-05,
+      "loss": 1.2345,
+      "step": 201500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.726431094247882e-05,
+      "loss": 1.2319,
+      "step": 202000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.723278695966317e-05,
+      "loss": 1.22,
+      "step": 202500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.720126297684753e-05,
+      "loss": 1.2248,
+      "step": 203000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7169738994031885e-05,
+      "loss": 1.2157,
+      "step": 203500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.713821501121624e-05,
+      "loss": 1.2358,
+      "step": 204000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7106691028400585e-05,
+      "loss": 1.2328,
+      "step": 204500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.707516704558494e-05,
+      "loss": 1.2153,
+      "step": 205000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.70436430627693e-05,
+      "loss": 1.2313,
+      "step": 205500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.701211907995365e-05,
+      "loss": 1.2219,
+      "step": 206000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6980595097138005e-05,
+      "loss": 1.2297,
+      "step": 206500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.694907111432235e-05,
+      "loss": 1.2327,
+      "step": 207000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6917547131506705e-05,
+      "loss": 1.2318,
+      "step": 207500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6886023148691066e-05,
+      "loss": 1.2226,
+      "step": 208000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.685449916587542e-05,
+      "loss": 1.2297,
+      "step": 208500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.682297518305977e-05,
+      "loss": 1.224,
+      "step": 209000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6791451200244126e-05,
+      "loss": 1.2257,
+      "step": 209500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.675992721742847e-05,
+      "loss": 1.2258,
+      "step": 210000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6728403234612826e-05,
+      "loss": 1.2274,
+      "step": 210500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6696879251797186e-05,
+      "loss": 1.2217,
+      "step": 211000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.666535526898154e-05,
+      "loss": 1.2228,
+      "step": 211500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.663383128616589e-05,
+      "loss": 1.2263,
+      "step": 212000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6602307303350247e-05,
+      "loss": 1.215,
+      "step": 212500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.657078332053459e-05,
+      "loss": 1.2297,
+      "step": 213000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6539259337718953e-05,
+      "loss": 1.2213,
+      "step": 213500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.650773535490331e-05,
+      "loss": 1.2106,
+      "step": 214000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.647621137208766e-05,
+      "loss": 1.2199,
+      "step": 214500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6444687389272014e-05,
+      "loss": 1.2183,
+      "step": 215000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.641316340645636e-05,
+      "loss": 1.2175,
+      "step": 215500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.638163942364072e-05,
+      "loss": 1.2245,
+      "step": 216000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6350115440825074e-05,
+      "loss": 1.2266,
+      "step": 216500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.631859145800943e-05,
+      "loss": 1.2147,
+      "step": 217000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.628706747519378e-05,
+      "loss": 1.2244,
+      "step": 217500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6255543492378134e-05,
+      "loss": 1.2199,
+      "step": 218000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.622401950956249e-05,
+      "loss": 1.2264,
+      "step": 218500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.619249552674684e-05,
+      "loss": 1.2274,
+      "step": 219000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6160971543931195e-05,
+      "loss": 1.2145,
+      "step": 219500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.612944756111555e-05,
+      "loss": 1.2281,
+      "step": 220000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.60979235782999e-05,
+      "loss": 1.225,
+      "step": 220500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.606639959548425e-05,
+      "loss": 1.2272,
+      "step": 221000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.603487561266861e-05,
+      "loss": 1.2137,
+      "step": 221500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.600335162985296e-05,
+      "loss": 1.2131,
+      "step": 222000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5971827647037315e-05,
+      "loss": 1.2113,
+      "step": 222500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.594030366422167e-05,
+      "loss": 1.2021,
+      "step": 223000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.590877968140602e-05,
+      "loss": 1.2248,
+      "step": 223500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5877255698590376e-05,
+      "loss": 1.2194,
+      "step": 224000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.584573171577473e-05,
+      "loss": 1.2115,
+      "step": 224500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.581420773295908e-05,
+      "loss": 1.2223,
+      "step": 225000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5782683750143436e-05,
+      "loss": 1.2088,
+      "step": 225500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.575115976732779e-05,
+      "loss": 1.218,
+      "step": 226000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.571963578451214e-05,
+      "loss": 1.2144,
+      "step": 226500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5688111801696496e-05,
+      "loss": 1.2152,
+      "step": 227000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.565658781888085e-05,
+      "loss": 1.2081,
+      "step": 227500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.56250638360652e-05,
+      "loss": 1.2153,
+      "step": 228000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5593539853249557e-05,
+      "loss": 1.2123,
+      "step": 228500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.556201587043391e-05,
+      "loss": 1.2165,
+      "step": 229000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.553049188761826e-05,
+      "loss": 1.2143,
+      "step": 229500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.549896790480262e-05,
+      "loss": 1.2039,
+      "step": 230000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.546744392198697e-05,
+      "loss": 1.2169,
+      "step": 230500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5435919939171324e-05,
+      "loss": 1.2179,
+      "step": 231000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.540439595635568e-05,
+      "loss": 1.2178,
+      "step": 231500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.537287197354003e-05,
+      "loss": 1.2184,
+      "step": 232000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5341347990724384e-05,
+      "loss": 1.2141,
+      "step": 232500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.530982400790874e-05,
+      "loss": 1.2104,
+      "step": 233000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.527830002509309e-05,
+      "loss": 1.2114,
+      "step": 233500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5246776042277444e-05,
+      "loss": 1.2092,
+      "step": 234000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.52152520594618e-05,
+      "loss": 1.2129,
+      "step": 234500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.518372807664616e-05,
+      "loss": 1.2038,
+      "step": 235000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5152204093830505e-05,
+      "loss": 1.213,
+      "step": 235500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.512068011101486e-05,
+      "loss": 1.2105,
+      "step": 236000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.508915612819921e-05,
+      "loss": 1.2051,
+      "step": 236500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5057632145383565e-05,
+      "loss": 1.2105,
+      "step": 237000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5026108162567925e-05,
+      "loss": 1.2066,
+      "step": 237500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.499458417975227e-05,
+      "loss": 1.2162,
+      "step": 238000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4963060196936625e-05,
+      "loss": 1.2113,
+      "step": 238500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.493153621412098e-05,
+      "loss": 1.2056,
+      "step": 239000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.490001223130533e-05,
+      "loss": 1.2125,
+      "step": 239500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.486848824848969e-05,
+      "loss": 1.2104,
+      "step": 240000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4836964265674046e-05,
+      "loss": 1.2174,
+      "step": 240500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.480544028285839e-05,
+      "loss": 1.2045,
+      "step": 241000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4773916300042746e-05,
+      "loss": 1.2054,
+      "step": 241500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.47423923172271e-05,
+      "loss": 1.2001,
+      "step": 242000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.471086833441145e-05,
+      "loss": 1.1995,
+      "step": 242500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.467934435159581e-05,
+      "loss": 1.2107,
+      "step": 243000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.464782036878016e-05,
+      "loss": 1.2138,
+      "step": 243500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.461629638596451e-05,
+      "loss": 1.2026,
+      "step": 244000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4584772403148866e-05,
+      "loss": 1.2116,
+      "step": 244500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.455324842033322e-05,
+      "loss": 1.2067,
+      "step": 245000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.452172443751758e-05,
+      "loss": 1.2112,
+      "step": 245500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4490200454701934e-05,
+      "loss": 1.1996,
+      "step": 246000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.445867647188628e-05,
+      "loss": 1.1978,
+      "step": 246500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4427152489070634e-05,
+      "loss": 1.2188,
+      "step": 247000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.439562850625499e-05,
+      "loss": 1.2098,
+      "step": 247500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.436410452343935e-05,
+      "loss": 1.2038,
+      "step": 248000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.43325805406237e-05,
+      "loss": 1.1974,
+      "step": 248500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4301056557808054e-05,
+      "loss": 1.1995,
+      "step": 249000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.42695325749924e-05,
+      "loss": 1.2148,
+      "step": 249500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4238008592176754e-05,
+      "loss": 1.2018,
+      "step": 250000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4206484609361114e-05,
+      "loss": 1.2058,
+      "step": 250500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.417496062654547e-05,
+      "loss": 1.2057,
+      "step": 251000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.414343664372982e-05,
+      "loss": 1.2059,
+      "step": 251500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.411191266091417e-05,
+      "loss": 1.2017,
+      "step": 252000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.408038867809852e-05,
+      "loss": 1.2028,
+      "step": 252500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4048864695282875e-05,
+      "loss": 1.2041,
+      "step": 253000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4017340712467235e-05,
+      "loss": 1.1936,
+      "step": 253500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.398581672965159e-05,
+      "loss": 1.2032,
+      "step": 254000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.395429274683594e-05,
+      "loss": 1.2067,
+      "step": 254500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.392276876402029e-05,
+      "loss": 1.1945,
+      "step": 255000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.389124478120464e-05,
+      "loss": 1.2049,
+      "step": 255500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3859720798389e-05,
+      "loss": 1.1975,
+      "step": 256000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3828196815573356e-05,
+      "loss": 1.1973,
+      "step": 256500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.379667283275771e-05,
+      "loss": 1.2014,
+      "step": 257000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.376514884994206e-05,
+      "loss": 1.1997,
+      "step": 257500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.373362486712641e-05,
+      "loss": 1.2044,
+      "step": 258000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.370210088431077e-05,
+      "loss": 1.2105,
+      "step": 258500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.367057690149512e-05,
+      "loss": 1.1934,
+      "step": 259000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3639052918679476e-05,
+      "loss": 1.1967,
+      "step": 259500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.360752893586383e-05,
+      "loss": 1.201,
+      "step": 260000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3576004953048176e-05,
+      "loss": 1.2017,
+      "step": 260500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.354448097023254e-05,
+      "loss": 1.2053,
+      "step": 261000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.351295698741689e-05,
+      "loss": 1.1973,
+      "step": 261500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3481433004601243e-05,
+      "loss": 1.1972,
+      "step": 262000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.34499090217856e-05,
+      "loss": 1.1902,
+      "step": 262500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.341838503896995e-05,
+      "loss": 1.2027,
+      "step": 263000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3386861056154304e-05,
+      "loss": 1.1981,
+      "step": 263500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.335533707333866e-05,
+      "loss": 1.201,
+      "step": 264000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.332381309052301e-05,
+      "loss": 1.1885,
+      "step": 264500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3292289107707364e-05,
+      "loss": 1.1934,
+      "step": 265000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.326076512489172e-05,
+      "loss": 1.1933,
+      "step": 265500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.322924114207607e-05,
+      "loss": 1.1872,
+      "step": 266000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3197717159260424e-05,
+      "loss": 1.1967,
+      "step": 266500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.316619317644478e-05,
+      "loss": 1.201,
+      "step": 267000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.313466919362913e-05,
+      "loss": 1.1889,
+      "step": 267500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3103145210813485e-05,
+      "loss": 1.1834,
+      "step": 268000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.307162122799784e-05,
+      "loss": 1.1921,
+      "step": 268500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.304009724518219e-05,
+      "loss": 1.1926,
+      "step": 269000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3008573262366545e-05,
+      "loss": 1.191,
+      "step": 269500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.29770492795509e-05,
+      "loss": 1.1933,
+      "step": 270000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.294552529673525e-05,
+      "loss": 1.1884,
+      "step": 270500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2914001313919605e-05,
+      "loss": 1.1909,
+      "step": 271000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.288247733110396e-05,
+      "loss": 1.1941,
+      "step": 271500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.285095334828831e-05,
+      "loss": 1.194,
+      "step": 272000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2819429365472666e-05,
+      "loss": 1.1948,
+      "step": 272500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.278790538265702e-05,
+      "loss": 1.1893,
+      "step": 273000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.275638139984137e-05,
+      "loss": 1.1865,
+      "step": 273500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2724857417025726e-05,
+      "loss": 1.1893,
+      "step": 274000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.269333343421008e-05,
+      "loss": 1.1983,
+      "step": 274500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.266180945139443e-05,
+      "loss": 1.191,
+      "step": 275000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2630285468578786e-05,
+      "loss": 1.1934,
+      "step": 275500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.259876148576314e-05,
+      "loss": 1.1913,
+      "step": 276000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.256723750294749e-05,
+      "loss": 1.1924,
+      "step": 276500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2535713520131847e-05,
+      "loss": 1.188,
+      "step": 277000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.25041895373162e-05,
+      "loss": 1.1844,
+      "step": 277500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2472665554500553e-05,
+      "loss": 1.196,
+      "step": 278000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.244114157168491e-05,
+      "loss": 1.1861,
+      "step": 278500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.240961758886926e-05,
+      "loss": 1.1922,
+      "step": 279000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2378093606053614e-05,
+      "loss": 1.1899,
+      "step": 279500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2346569623237974e-05,
+      "loss": 1.1898,
+      "step": 280000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.231504564042232e-05,
+      "loss": 1.1885,
+      "step": 280500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2283521657606674e-05,
+      "loss": 1.1911,
+      "step": 281000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.225199767479103e-05,
+      "loss": 1.1901,
+      "step": 281500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.222047369197538e-05,
+      "loss": 1.192,
+      "step": 282000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.218894970915974e-05,
+      "loss": 1.1859,
+      "step": 282500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.215742572634409e-05,
+      "loss": 1.1943,
+      "step": 283000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.212590174352844e-05,
+      "loss": 1.1932,
+      "step": 283500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2094377760712795e-05,
+      "loss": 1.1895,
+      "step": 284000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.206285377789715e-05,
+      "loss": 1.1864,
+      "step": 284500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.203132979508151e-05,
+      "loss": 1.1811,
+      "step": 285000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.199980581226586e-05,
+      "loss": 1.188,
+      "step": 285500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.196828182945021e-05,
+      "loss": 1.1901,
+      "step": 286000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.193675784663456e-05,
+      "loss": 1.1814,
+      "step": 286500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1905233863818915e-05,
+      "loss": 1.1816,
+      "step": 287000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.187370988100327e-05,
+      "loss": 1.1975,
+      "step": 287500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.184218589818763e-05,
+      "loss": 1.1889,
+      "step": 288000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.181066191537198e-05,
+      "loss": 1.1843,
+      "step": 288500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.177913793255633e-05,
+      "loss": 1.1848,
+      "step": 289000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.174761394974068e-05,
+      "loss": 1.1829,
+      "step": 289500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1716089966925036e-05,
+      "loss": 1.1828,
+      "step": 290000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1684565984109396e-05,
+      "loss": 1.1915,
+      "step": 290500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.165304200129375e-05,
+      "loss": 1.1803,
+      "step": 291000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1621518018478096e-05,
+      "loss": 1.1891,
+      "step": 291500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.158999403566245e-05,
+      "loss": 1.18,
+      "step": 292000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.15584700528468e-05,
+      "loss": 1.1817,
+      "step": 292500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.152694607003116e-05,
+      "loss": 1.1788,
+      "step": 293000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.149542208721552e-05,
+      "loss": 1.1777,
+      "step": 293500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.146389810439987e-05,
+      "loss": 1.172,
+      "step": 294000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.143237412158422e-05,
+      "loss": 1.1754,
+      "step": 294500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.140085013876857e-05,
+      "loss": 1.1897,
+      "step": 295000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.136932615595293e-05,
+      "loss": 1.178,
+      "step": 295500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1337802173137284e-05,
+      "loss": 1.1772,
+      "step": 296000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.130627819032164e-05,
+      "loss": 1.1848,
+      "step": 296500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1274754207505984e-05,
+      "loss": 1.1804,
+      "step": 297000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.124323022469034e-05,
+      "loss": 1.179,
+      "step": 297500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.121170624187469e-05,
+      "loss": 1.1881,
+      "step": 298000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.118018225905905e-05,
+      "loss": 1.1759,
+      "step": 298500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1148658276243405e-05,
+      "loss": 1.181,
+      "step": 299000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.111713429342776e-05,
+      "loss": 1.1858,
+      "step": 299500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1085610310612105e-05,
+      "loss": 1.1794,
+      "step": 300000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.105408632779646e-05,
+      "loss": 1.1764,
+      "step": 300500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.102256234498082e-05,
+      "loss": 1.1845,
+      "step": 301000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.099103836216517e-05,
+      "loss": 1.1801,
+      "step": 301500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.0959514379349525e-05,
+      "loss": 1.1797,
+      "step": 302000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.092799039653388e-05,
+      "loss": 1.183,
+      "step": 302500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0896466413718225e-05,
+      "loss": 1.1838,
+      "step": 303000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0864942430902585e-05,
+      "loss": 1.1831,
+      "step": 303500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.083341844808694e-05,
+      "loss": 1.1847,
+      "step": 304000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.080189446527129e-05,
+      "loss": 1.1718,
+      "step": 304500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0770370482455646e-05,
+      "loss": 1.1844,
+      "step": 305000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.073884649963999e-05,
+      "loss": 1.1784,
+      "step": 305500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.070732251682435e-05,
+      "loss": 1.1768,
+      "step": 306000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0675798534008706e-05,
+      "loss": 1.1803,
+      "step": 306500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.064427455119306e-05,
+      "loss": 1.1822,
+      "step": 307000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.061275056837741e-05,
+      "loss": 1.169,
+      "step": 307500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0581226585561766e-05,
+      "loss": 1.176,
+      "step": 308000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.054970260274611e-05,
+      "loss": 1.1807,
+      "step": 308500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.051817861993047e-05,
+      "loss": 1.1791,
+      "step": 309000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0486654637114827e-05,
+      "loss": 1.1832,
+      "step": 309500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.045513065429918e-05,
+      "loss": 1.1734,
+      "step": 310000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0423606671483534e-05,
+      "loss": 1.1731,
+      "step": 310500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.039208268866789e-05,
+      "loss": 1.1725,
+      "step": 311000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0360558705852237e-05,
+      "loss": 1.1691,
+      "step": 311500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0329034723036594e-05,
+      "loss": 1.1698,
+      "step": 312000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0297510740220947e-05,
+      "loss": 1.1782,
+      "step": 312500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.02659867574053e-05,
+      "loss": 1.1785,
+      "step": 313000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0234462774589654e-05,
+      "loss": 1.1708,
+      "step": 313500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0202938791774004e-05,
+      "loss": 1.1863,
+      "step": 314000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0171414808958358e-05,
+      "loss": 1.1716,
+      "step": 314500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0139890826142714e-05,
+      "loss": 1.1781,
+      "step": 315000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0108366843327068e-05,
+      "loss": 1.1697,
+      "step": 315500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.007684286051142e-05,
+      "loss": 1.177,
+      "step": 316000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0045318877695778e-05,
+      "loss": 1.171,
+      "step": 316500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0013794894880125e-05,
+      "loss": 1.1679,
+      "step": 317000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.998227091206448e-05,
+      "loss": 1.173,
+      "step": 317500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.9950746929248835e-05,
+      "loss": 1.1747,
+      "step": 318000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.991922294643319e-05,
+      "loss": 1.1665,
+      "step": 318500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9887698963617545e-05,
+      "loss": 1.1717,
+      "step": 319000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9856174980801892e-05,
+      "loss": 1.1799,
+      "step": 319500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.982465099798625e-05,
+      "loss": 1.1683,
+      "step": 320000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9793127015170602e-05,
+      "loss": 1.1765,
+      "step": 320500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9761603032354956e-05,
+      "loss": 1.1657,
+      "step": 321000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9730079049539313e-05,
+      "loss": 1.1745,
+      "step": 321500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9698555066723666e-05,
+      "loss": 1.1861,
+      "step": 322000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9667031083908016e-05,
+      "loss": 1.1799,
+      "step": 322500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.963550710109237e-05,
+      "loss": 1.168,
+      "step": 323000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9603983118276723e-05,
+      "loss": 1.1693,
+      "step": 323500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.957245913546108e-05,
+      "loss": 1.1737,
+      "step": 324000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9540935152645433e-05,
+      "loss": 1.1775,
+      "step": 324500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9509411169829787e-05,
+      "loss": 1.1732,
+      "step": 325000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9477887187014137e-05,
+      "loss": 1.1633,
+      "step": 325500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.944636320419849e-05,
+      "loss": 1.1749,
+      "step": 326000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9414839221382844e-05,
+      "loss": 1.1698,
+      "step": 326500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.93833152385672e-05,
+      "loss": 1.1632,
+      "step": 327000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9351791255751554e-05,
+      "loss": 1.1621,
+      "step": 327500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9320267272935904e-05,
+      "loss": 1.1646,
+      "step": 328000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9288743290120257e-05,
+      "loss": 1.1695,
+      "step": 328500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.925721930730461e-05,
+      "loss": 1.1661,
+      "step": 329000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9225695324488968e-05,
+      "loss": 1.1669,
+      "step": 329500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.919417134167332e-05,
+      "loss": 1.1718,
+      "step": 330000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9162647358857674e-05,
+      "loss": 1.1702,
+      "step": 330500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9131123376042024e-05,
+      "loss": 1.1697,
+      "step": 331000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9099599393226378e-05,
+      "loss": 1.1607,
+      "step": 331500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9068075410410735e-05,
+      "loss": 1.168,
+      "step": 332000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9036551427595088e-05,
+      "loss": 1.1717,
+      "step": 332500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.900502744477944e-05,
+      "loss": 1.1638,
+      "step": 333000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.89735034619638e-05,
+      "loss": 1.1763,
+      "step": 333500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.8941979479148145e-05,
+      "loss": 1.1716,
+      "step": 334000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8910455496332502e-05,
+      "loss": 1.1693,
+      "step": 334500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8878931513516855e-05,
+      "loss": 1.1638,
+      "step": 335000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.884740753070121e-05,
+      "loss": 1.1697,
+      "step": 335500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8815883547885562e-05,
+      "loss": 1.1701,
+      "step": 336000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8784359565069912e-05,
+      "loss": 1.163,
+      "step": 336500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8752835582254266e-05,
+      "loss": 1.1738,
+      "step": 337000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8721311599438622e-05,
+      "loss": 1.1666,
+      "step": 337500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8689787616622976e-05,
+      "loss": 1.167,
+      "step": 338000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.865826363380733e-05,
+      "loss": 1.1655,
+      "step": 338500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8626739650991686e-05,
+      "loss": 1.1576,
+      "step": 339000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8595215668176033e-05,
+      "loss": 1.1653,
+      "step": 339500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.856369168536039e-05,
+      "loss": 1.1618,
+      "step": 340000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8532167702544743e-05,
+      "loss": 1.1714,
+      "step": 340500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8500643719729097e-05,
+      "loss": 1.1704,
+      "step": 341000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8469119736913453e-05,
+      "loss": 1.1672,
+      "step": 341500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8437595754097807e-05,
+      "loss": 1.1655,
+      "step": 342000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8406071771282157e-05,
+      "loss": 1.1657,
+      "step": 342500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.837454778846651e-05,
+      "loss": 1.1631,
+      "step": 343000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8343023805650864e-05,
+      "loss": 1.1749,
+      "step": 343500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.831149982283522e-05,
+      "loss": 1.1672,
+      "step": 344000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8279975840019574e-05,
+      "loss": 1.154,
+      "step": 344500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8248451857203924e-05,
+      "loss": 1.1668,
+      "step": 345000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8216927874388277e-05,
+      "loss": 1.1682,
+      "step": 345500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.818540389157263e-05,
+      "loss": 1.1652,
+      "step": 346000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8153879908756984e-05,
+      "loss": 1.1582,
+      "step": 346500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.812235592594134e-05,
+      "loss": 1.1633,
+      "step": 347000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8090831943125695e-05,
+      "loss": 1.1594,
+      "step": 347500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8059307960310045e-05,
+      "loss": 1.1639,
+      "step": 348000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8027783977494398e-05,
+      "loss": 1.1659,
+      "step": 348500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.799625999467875e-05,
+      "loss": 1.1676,
+      "step": 349000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.796473601186311e-05,
+      "loss": 1.1583,
+      "step": 349500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.7933212029047462e-05,
+      "loss": 1.1621,
+      "step": 350000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7901688046231812e-05,
+      "loss": 1.1594,
+      "step": 350500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7870164063416165e-05,
+      "loss": 1.161,
+      "step": 351000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.783864008060052e-05,
+      "loss": 1.1571,
+      "step": 351500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7807116097784876e-05,
+      "loss": 1.1655,
+      "step": 352000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.777559211496923e-05,
+      "loss": 1.1636,
+      "step": 352500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7744068132153582e-05,
+      "loss": 1.1568,
+      "step": 353000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7712544149337932e-05,
+      "loss": 1.161,
+      "step": 353500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7681020166522286e-05,
+      "loss": 1.1611,
+      "step": 354000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7649496183706643e-05,
+      "loss": 1.156,
+      "step": 354500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7617972200890996e-05,
+      "loss": 1.1614,
+      "step": 355000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.758644821807535e-05,
+      "loss": 1.1648,
+      "step": 355500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7554924235259706e-05,
+      "loss": 1.1559,
+      "step": 356000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7523400252444053e-05,
+      "loss": 1.1606,
+      "step": 356500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.749187626962841e-05,
+      "loss": 1.1581,
+      "step": 357000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7460352286812763e-05,
+      "loss": 1.1583,
+      "step": 357500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7428828303997117e-05,
+      "loss": 1.1593,
+      "step": 358000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.739730432118147e-05,
+      "loss": 1.1539,
+      "step": 358500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.736578033836582e-05,
+      "loss": 1.1579,
+      "step": 359000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7334256355550174e-05,
+      "loss": 1.1578,
+      "step": 359500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.730273237273453e-05,
+      "loss": 1.16,
+      "step": 360000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7271208389918884e-05,
+      "loss": 1.1641,
+      "step": 360500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7239684407103237e-05,
+      "loss": 1.1561,
+      "step": 361000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7208160424287594e-05,
+      "loss": 1.164,
+      "step": 361500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.717663644147194e-05,
+      "loss": 1.153,
+      "step": 362000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7145112458656298e-05,
+      "loss": 1.1539,
+      "step": 362500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.711358847584065e-05,
+      "loss": 1.1532,
+      "step": 363000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7082064493025005e-05,
+      "loss": 1.1611,
+      "step": 363500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.705054051020936e-05,
+      "loss": 1.1558,
+      "step": 364000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7019016527393715e-05,
+      "loss": 1.1565,
+      "step": 364500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6987492544578065e-05,
+      "loss": 1.1502,
+      "step": 365000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6955968561762418e-05,
+      "loss": 1.1487,
+      "step": 365500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6924444578946772e-05,
+      "loss": 1.1519,
+      "step": 366000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.689292059613113e-05,
+      "loss": 1.1592,
+      "step": 366500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6861396613315482e-05,
+      "loss": 1.1537,
+      "step": 367000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6829872630499832e-05,
+      "loss": 1.1574,
+      "step": 367500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6798348647684185e-05,
+      "loss": 1.153,
+      "step": 368000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.676682466486854e-05,
+      "loss": 1.1462,
+      "step": 368500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6735300682052892e-05,
+      "loss": 1.1511,
+      "step": 369000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.670377669923725e-05,
+      "loss": 1.1518,
+      "step": 369500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6672252716421603e-05,
+      "loss": 1.1539,
+      "step": 370000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6640728733605953e-05,
+      "loss": 1.1613,
+      "step": 370500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6609204750790306e-05,
+      "loss": 1.1626,
+      "step": 371000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.657768076797466e-05,
+      "loss": 1.1557,
+      "step": 371500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6546156785159016e-05,
+      "loss": 1.1559,
+      "step": 372000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.651463280234337e-05,
+      "loss": 1.1476,
+      "step": 372500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.648310881952772e-05,
+      "loss": 1.146,
+      "step": 373000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6451584836712073e-05,
+      "loss": 1.1551,
+      "step": 373500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6420060853896427e-05,
+      "loss": 1.1516,
+      "step": 374000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6388536871080784e-05,
+      "loss": 1.1531,
+      "step": 374500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6357012888265137e-05,
+      "loss": 1.1548,
+      "step": 375000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.632548890544949e-05,
+      "loss": 1.1588,
+      "step": 375500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.629396492263384e-05,
+      "loss": 1.1569,
+      "step": 376000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6262440939818194e-05,
+      "loss": 1.1547,
+      "step": 376500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.623091695700255e-05,
+      "loss": 1.1444,
+      "step": 377000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6199392974186904e-05,
+      "loss": 1.1588,
+      "step": 377500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6167868991371258e-05,
+      "loss": 1.1513,
+      "step": 378000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6136345008555614e-05,
+      "loss": 1.1513,
+      "step": 378500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.610482102573996e-05,
+      "loss": 1.1551,
+      "step": 379000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6073297042924314e-05,
+      "loss": 1.1521,
+      "step": 379500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.604177306010867e-05,
+      "loss": 1.154,
+      "step": 380000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6010249077293025e-05,
+      "loss": 1.157,
+      "step": 380500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.5978725094477378e-05,
+      "loss": 1.1499,
+      "step": 381000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.5947201111661728e-05,
+      "loss": 1.1456,
+      "step": 381500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.591567712884608e-05,
+      "loss": 1.1486,
+      "step": 382000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.588415314603044e-05,
+      "loss": 1.1507,
+      "step": 382500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5852629163214792e-05,
+      "loss": 1.1544,
+      "step": 383000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5821105180399145e-05,
+      "loss": 1.1478,
+      "step": 383500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5789581197583502e-05,
+      "loss": 1.1473,
+      "step": 384000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.575805721476785e-05,
+      "loss": 1.1528,
+      "step": 384500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5726533231952206e-05,
+      "loss": 1.1443,
+      "step": 385000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.569500924913656e-05,
+      "loss": 1.1494,
+      "step": 385500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5663485266320913e-05,
+      "loss": 1.148,
+      "step": 386000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.563196128350527e-05,
+      "loss": 1.1475,
+      "step": 386500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5600437300689623e-05,
+      "loss": 1.1492,
+      "step": 387000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5568913317873973e-05,
+      "loss": 1.1483,
+      "step": 387500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5537389335058326e-05,
+      "loss": 1.1499,
+      "step": 388000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.550586535224268e-05,
+      "loss": 1.1446,
+      "step": 388500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5474341369427037e-05,
+      "loss": 1.142,
+      "step": 389000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.544281738661139e-05,
+      "loss": 1.1489,
+      "step": 389500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.541129340379574e-05,
+      "loss": 1.1574,
+      "step": 390000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5379769420980093e-05,
+      "loss": 1.1388,
+      "step": 390500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5348245438164447e-05,
+      "loss": 1.1489,
+      "step": 391000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.53167214553488e-05,
+      "loss": 1.1436,
+      "step": 391500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5285197472533157e-05,
+      "loss": 1.1476,
+      "step": 392000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.525367348971751e-05,
+      "loss": 1.1431,
+      "step": 392500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.522214950690186e-05,
+      "loss": 1.1512,
+      "step": 393000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5190625524086214e-05,
+      "loss": 1.1446,
+      "step": 393500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5159101541270568e-05,
+      "loss": 1.1529,
+      "step": 394000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5127577558454924e-05,
+      "loss": 1.1408,
+      "step": 394500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5096053575639278e-05,
+      "loss": 1.1436,
+      "step": 395000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.506452959282363e-05,
+      "loss": 1.1522,
+      "step": 395500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.503300561000798e-05,
+      "loss": 1.1495,
+      "step": 396000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5001481627192335e-05,
+      "loss": 1.1469,
+      "step": 396500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.496995764437669e-05,
+      "loss": 1.1473,
+      "step": 397000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.4938433661561045e-05,
+      "loss": 1.1339,
+      "step": 397500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4906909678745395e-05,
+      "loss": 1.1437,
+      "step": 398000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4875385695929752e-05,
+      "loss": 1.1392,
+      "step": 398500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4843861713114105e-05,
+      "loss": 1.1421,
+      "step": 399000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.481233773029846e-05,
+      "loss": 1.1483,
+      "step": 399500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4780813747482812e-05,
+      "loss": 1.137,
+      "step": 400000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4749289764667162e-05,
+      "loss": 1.1475,
+      "step": 400500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.471776578185152e-05,
+      "loss": 1.1365,
+      "step": 401000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4686241799035872e-05,
+      "loss": 1.1503,
+      "step": 401500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4654717816220222e-05,
+      "loss": 1.149,
+      "step": 402000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.462319383340458e-05,
+      "loss": 1.147,
+      "step": 402500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4591669850588933e-05,
+      "loss": 1.1412,
+      "step": 403000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4560145867773286e-05,
+      "loss": 1.1497,
+      "step": 403500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.452862188495764e-05,
+      "loss": 1.1409,
+      "step": 404000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4497097902141993e-05,
+      "loss": 1.1371,
+      "step": 404500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4465573919326347e-05,
+      "loss": 1.1587,
+      "step": 405000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.44340499365107e-05,
+      "loss": 1.144,
+      "step": 405500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4402525953695053e-05,
+      "loss": 1.1456,
+      "step": 406000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4371001970879407e-05,
+      "loss": 1.1413,
+      "step": 406500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.433947798806376e-05,
+      "loss": 1.1506,
+      "step": 407000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4307954005248114e-05,
+      "loss": 1.1397,
+      "step": 407500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4276430022432467e-05,
+      "loss": 1.1385,
+      "step": 408000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.424490603961682e-05,
+      "loss": 1.1399,
+      "step": 408500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4213382056801174e-05,
+      "loss": 1.1373,
+      "step": 409000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4181858073985527e-05,
+      "loss": 1.1437,
+      "step": 409500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.415033409116988e-05,
+      "loss": 1.1429,
+      "step": 410000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4118810108354234e-05,
+      "loss": 1.1454,
+      "step": 410500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4087286125538588e-05,
+      "loss": 1.1479,
+      "step": 411000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4055762142722945e-05,
+      "loss": 1.1334,
+      "step": 411500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4024238159907295e-05,
+      "loss": 1.1325,
+      "step": 412000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3992714177091648e-05,
+      "loss": 1.1382,
+      "step": 412500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3961190194276005e-05,
+      "loss": 1.1407,
+      "step": 413000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3929666211460355e-05,
+      "loss": 1.1441,
+      "step": 413500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.389814222864471e-05,
+      "loss": 1.1395,
+      "step": 414000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3866618245829065e-05,
+      "loss": 1.1428,
+      "step": 414500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3835094263013415e-05,
+      "loss": 1.1471,
+      "step": 415000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3803570280197772e-05,
+      "loss": 1.1455,
+      "step": 415500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3772046297382122e-05,
+      "loss": 1.1428,
+      "step": 416000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3740522314566476e-05,
+      "loss": 1.1418,
+      "step": 416500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3708998331750832e-05,
+      "loss": 1.1344,
+      "step": 417000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3677474348935182e-05,
+      "loss": 1.1401,
+      "step": 417500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3645950366119536e-05,
+      "loss": 1.1418,
+      "step": 418000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3614426383303893e-05,
+      "loss": 1.137,
+      "step": 418500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3582902400488243e-05,
+      "loss": 1.1241,
+      "step": 419000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.35513784176726e-05,
+      "loss": 1.1377,
+      "step": 419500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3519854434856953e-05,
+      "loss": 1.1325,
+      "step": 420000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3488330452041303e-05,
+      "loss": 1.147,
+      "step": 420500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.345680646922566e-05,
+      "loss": 1.135,
+      "step": 421000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3425282486410013e-05,
+      "loss": 1.1351,
+      "step": 421500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3393758503594367e-05,
+      "loss": 1.1491,
+      "step": 422000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.336223452077872e-05,
+      "loss": 1.1423,
+      "step": 422500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.333071053796307e-05,
+      "loss": 1.1393,
+      "step": 423000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3299186555147427e-05,
+      "loss": 1.1271,
+      "step": 423500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.326766257233178e-05,
+      "loss": 1.1357,
+      "step": 424000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.323613858951613e-05,
+      "loss": 1.1344,
+      "step": 424500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3204614606700487e-05,
+      "loss": 1.1291,
+      "step": 425000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.317309062388484e-05,
+      "loss": 1.1355,
+      "step": 425500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3141566641069194e-05,
+      "loss": 1.1287,
+      "step": 426000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3110042658253548e-05,
+      "loss": 1.135,
+      "step": 426500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.30785186754379e-05,
+      "loss": 1.1446,
+      "step": 427000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3046994692622255e-05,
+      "loss": 1.1432,
+      "step": 427500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3015470709806608e-05,
+      "loss": 1.1303,
+      "step": 428000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.298394672699096e-05,
+      "loss": 1.1442,
+      "step": 428500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.2952422744175315e-05,
+      "loss": 1.1309,
+      "step": 429000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.2920898761359668e-05,
+      "loss": 1.1268,
+      "step": 429500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.288937477854402e-05,
+      "loss": 1.1306,
+      "step": 430000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2857850795728375e-05,
+      "loss": 1.1389,
+      "step": 430500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.282632681291273e-05,
+      "loss": 1.1343,
+      "step": 431000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2794802830097082e-05,
+      "loss": 1.1333,
+      "step": 431500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2763278847281435e-05,
+      "loss": 1.1383,
+      "step": 432000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.273175486446579e-05,
+      "loss": 1.1341,
+      "step": 432500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2700230881650142e-05,
+      "loss": 1.1343,
+      "step": 433000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2668706898834496e-05,
+      "loss": 1.1287,
+      "step": 433500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2637182916018853e-05,
+      "loss": 1.1397,
+      "step": 434000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2605658933203203e-05,
+      "loss": 1.136,
+      "step": 434500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2574134950387556e-05,
+      "loss": 1.1247,
+      "step": 435000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2542610967571913e-05,
+      "loss": 1.1283,
+      "step": 435500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2511086984756263e-05,
+      "loss": 1.1349,
+      "step": 436000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2479563001940616e-05,
+      "loss": 1.1261,
+      "step": 436500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2448039019124973e-05,
+      "loss": 1.1287,
+      "step": 437000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2416515036309323e-05,
+      "loss": 1.1357,
+      "step": 437500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.238499105349368e-05,
+      "loss": 1.1362,
+      "step": 438000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.235346707067803e-05,
+      "loss": 1.1276,
+      "step": 438500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2321943087862384e-05,
+      "loss": 1.1325,
+      "step": 439000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.229041910504674e-05,
+      "loss": 1.1226,
+      "step": 439500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.225889512223109e-05,
+      "loss": 1.1207,
+      "step": 440000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2227371139415444e-05,
+      "loss": 1.1301,
+      "step": 440500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.21958471565998e-05,
+      "loss": 1.1343,
+      "step": 441000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.216432317378415e-05,
+      "loss": 1.1279,
+      "step": 441500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2132799190968508e-05,
+      "loss": 1.124,
+      "step": 442000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.210127520815286e-05,
+      "loss": 1.1229,
+      "step": 442500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.206975122533721e-05,
+      "loss": 1.1377,
+      "step": 443000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2038227242521568e-05,
+      "loss": 1.1277,
+      "step": 443500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.200670325970592e-05,
+      "loss": 1.1284,
+      "step": 444000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.1975179276890275e-05,
+      "loss": 1.1266,
+      "step": 444500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.1943655294074628e-05,
+      "loss": 1.1242,
+      "step": 445000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1912131311258978e-05,
+      "loss": 1.119,
+      "step": 445500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1880607328443335e-05,
+      "loss": 1.1285,
+      "step": 446000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.184908334562769e-05,
+      "loss": 1.133,
+      "step": 446500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.181755936281204e-05,
+      "loss": 1.1283,
+      "step": 447000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1786035379996395e-05,
+      "loss": 1.1312,
+      "step": 447500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.175451139718075e-05,
+      "loss": 1.1295,
+      "step": 448000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1722987414365102e-05,
+      "loss": 1.1331,
+      "step": 448500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1691463431549456e-05,
+      "loss": 1.1275,
+      "step": 449000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.165993944873381e-05,
+      "loss": 1.1229,
+      "step": 449500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1628415465918163e-05,
+      "loss": 1.1197,
+      "step": 450000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1596891483102516e-05,
+      "loss": 1.1211,
+      "step": 450500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.156536750028687e-05,
+      "loss": 1.1299,
+      "step": 451000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1533843517471223e-05,
+      "loss": 1.1399,
+      "step": 451500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1502319534655576e-05,
+      "loss": 1.1205,
+      "step": 452000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.147079555183993e-05,
+      "loss": 1.1271,
+      "step": 452500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1439271569024283e-05,
+      "loss": 1.1253,
+      "step": 453000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1407747586208637e-05,
+      "loss": 1.1318,
+      "step": 453500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.137622360339299e-05,
+      "loss": 1.124,
+      "step": 454000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1344699620577343e-05,
+      "loss": 1.1271,
+      "step": 454500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1313175637761697e-05,
+      "loss": 1.1284,
+      "step": 455000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.128165165494605e-05,
+      "loss": 1.1278,
+      "step": 455500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1250127672130404e-05,
+      "loss": 1.1375,
+      "step": 456000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1218603689314757e-05,
+      "loss": 1.1224,
+      "step": 456500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.118707970649911e-05,
+      "loss": 1.1182,
+      "step": 457000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1155555723683464e-05,
+      "loss": 1.1236,
+      "step": 457500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.112403174086782e-05,
+      "loss": 1.1216,
+      "step": 458000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.109250775805217e-05,
+      "loss": 1.1148,
+      "step": 458500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1060983775236524e-05,
+      "loss": 1.1259,
+      "step": 459000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.102945979242088e-05,
+      "loss": 1.1209,
+      "step": 459500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.099793580960523e-05,
+      "loss": 1.1278,
+      "step": 460000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.0966411826789588e-05,
+      "loss": 1.1265,
+      "step": 460500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.0934887843973938e-05,
+      "loss": 1.1267,
+      "step": 461000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.090336386115829e-05,
+      "loss": 1.1236,
+      "step": 461500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.087183987834265e-05,
+      "loss": 1.1223,
+      "step": 462000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0840315895527e-05,
+      "loss": 1.1282,
+      "step": 462500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0808791912711352e-05,
+      "loss": 1.1196,
+      "step": 463000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.077726792989571e-05,
+      "loss": 1.1282,
+      "step": 463500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.074574394708006e-05,
+      "loss": 1.1278,
+      "step": 464000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0714219964264416e-05,
+      "loss": 1.1204,
+      "step": 464500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.068269598144877e-05,
+      "loss": 1.1181,
+      "step": 465000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.065117199863312e-05,
+      "loss": 1.1217,
+      "step": 465500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0619648015817476e-05,
+      "loss": 1.1149,
+      "step": 466000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.058812403300183e-05,
+      "loss": 1.1294,
+      "step": 466500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0556600050186183e-05,
+      "loss": 1.1255,
+      "step": 467000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0525076067370536e-05,
+      "loss": 1.115,
+      "step": 467500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0493552084554886e-05,
+      "loss": 1.1232,
+      "step": 468000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0462028101739243e-05,
+      "loss": 1.1268,
+      "step": 468500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0430504118923596e-05,
+      "loss": 1.1208,
+      "step": 469000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0398980136107947e-05,
+      "loss": 1.1205,
+      "step": 469500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0367456153292303e-05,
+      "loss": 1.127,
+      "step": 470000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0335932170476657e-05,
+      "loss": 1.1257,
+      "step": 470500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.030440818766101e-05,
+      "loss": 1.1201,
+      "step": 471000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0272884204845364e-05,
+      "loss": 1.128,
+      "step": 471500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0241360222029717e-05,
+      "loss": 1.1132,
+      "step": 472000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.020983623921407e-05,
+      "loss": 1.1115,
+      "step": 472500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0178312256398424e-05,
+      "loss": 1.1164,
+      "step": 473000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0146788273582777e-05,
+      "loss": 1.1181,
+      "step": 473500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.011526429076713e-05,
+      "loss": 1.1184,
+      "step": 474000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0083740307951484e-05,
+      "loss": 1.1179,
+      "step": 474500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0052216325135838e-05,
+      "loss": 1.1156,
+      "step": 475000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.002069234232019e-05,
+      "loss": 1.1261,
+      "step": 475500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9989168359504545e-05,
+      "loss": 1.1247,
+      "step": 476000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9957644376688898e-05,
+      "loss": 1.1248,
+      "step": 476500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.992612039387325e-05,
+      "loss": 1.1202,
+      "step": 477000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9894596411057605e-05,
+      "loss": 1.1165,
+      "step": 477500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.986307242824196e-05,
+      "loss": 1.1134,
+      "step": 478000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9831548445426312e-05,
+      "loss": 1.1234,
+      "step": 478500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9800024462610665e-05,
+      "loss": 1.1199,
+      "step": 479000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.976850047979502e-05,
+      "loss": 1.1192,
+      "step": 479500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9736976496979372e-05,
+      "loss": 1.1215,
+      "step": 480000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.970545251416373e-05,
+      "loss": 1.1169,
+      "step": 480500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.967392853134808e-05,
+      "loss": 1.1111,
+      "step": 481000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9642404548532432e-05,
+      "loss": 1.1167,
+      "step": 481500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.961088056571679e-05,
+      "loss": 1.1197,
+      "step": 482000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.957935658290114e-05,
+      "loss": 1.1203,
+      "step": 482500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9547832600085496e-05,
+      "loss": 1.1139,
+      "step": 483000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9516308617269846e-05,
+      "loss": 1.1198,
+      "step": 483500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.94847846344542e-05,
+      "loss": 1.117,
+      "step": 484000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9453260651638556e-05,
+      "loss": 1.1217,
+      "step": 484500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9421736668822906e-05,
+      "loss": 1.1194,
+      "step": 485000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.939021268600726e-05,
+      "loss": 1.1257,
+      "step": 485500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9358688703191617e-05,
+      "loss": 1.109,
+      "step": 486000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9327164720375967e-05,
+      "loss": 1.1206,
+      "step": 486500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9295640737560324e-05,
+      "loss": 1.1119,
+      "step": 487000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9264116754744677e-05,
+      "loss": 1.1128,
+      "step": 487500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9232592771929027e-05,
+      "loss": 1.1165,
+      "step": 488000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9201068789113384e-05,
+      "loss": 1.1098,
+      "step": 488500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9169544806297737e-05,
+      "loss": 1.1197,
+      "step": 489000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9138020823482087e-05,
+      "loss": 1.1133,
+      "step": 489500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9106496840666444e-05,
+      "loss": 1.1117,
+      "step": 490000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9074972857850798e-05,
+      "loss": 1.1185,
+      "step": 490500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.904344887503515e-05,
+      "loss": 1.1189,
+      "step": 491000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9011924892219504e-05,
+      "loss": 1.1308,
+      "step": 491500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8980400909403855e-05,
+      "loss": 1.112,
+      "step": 492000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.894887692658821e-05,
+      "loss": 1.1164,
+      "step": 492500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8917352943772565e-05,
+      "loss": 1.1147,
+      "step": 493000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8885828960956918e-05,
+      "loss": 1.1175,
+      "step": 493500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.885430497814127e-05,
+      "loss": 1.1223,
+      "step": 494000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8822780995325625e-05,
+      "loss": 1.1118,
+      "step": 494500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.879125701250998e-05,
+      "loss": 1.1173,
+      "step": 495000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8759733029694332e-05,
+      "loss": 1.1157,
+      "step": 495500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8728209046878685e-05,
+      "loss": 1.1181,
+      "step": 496000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.869668506406304e-05,
+      "loss": 1.109,
+      "step": 496500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8665161081247392e-05,
+      "loss": 1.1123,
+      "step": 497000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8633637098431746e-05,
+      "loss": 1.1169,
+      "step": 497500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.86021131156161e-05,
+      "loss": 1.1142,
+      "step": 498000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8570589132800453e-05,
+      "loss": 1.1084,
+      "step": 498500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8539065149984806e-05,
+      "loss": 1.1145,
+      "step": 499000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.850754116716916e-05,
+      "loss": 1.1159,
+      "step": 499500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8476017184353513e-05,
+      "loss": 1.1052,
+      "step": 500000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8444493201537866e-05,
+      "loss": 1.1137,
+      "step": 500500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.841296921872222e-05,
+      "loss": 1.1116,
+      "step": 501000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8381445235906573e-05,
+      "loss": 1.1164,
+      "step": 501500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8349921253090927e-05,
+      "loss": 1.1186,
+      "step": 502000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.831839727027528e-05,
+      "loss": 1.1086,
+      "step": 502500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8286873287459637e-05,
+      "loss": 1.1108,
+      "step": 503000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8255349304643987e-05,
+      "loss": 1.1076,
+      "step": 503500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.822382532182834e-05,
+      "loss": 1.1165,
+      "step": 504000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8192301339012697e-05,
+      "loss": 1.1067,
+      "step": 504500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8160777356197047e-05,
+      "loss": 1.1057,
+      "step": 505000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8129253373381404e-05,
+      "loss": 1.1001,
+      "step": 505500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8097729390565754e-05,
+      "loss": 1.1077,
+      "step": 506000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8066205407750108e-05,
+      "loss": 1.1201,
+      "step": 506500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8034681424934464e-05,
+      "loss": 1.1105,
+      "step": 507000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8003157442118814e-05,
+      "loss": 1.1053,
+      "step": 507500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.7971633459303168e-05,
+      "loss": 1.1007,
+      "step": 508000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.7940109476487525e-05,
+      "loss": 1.1128,
+      "step": 508500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7908585493671875e-05,
+      "loss": 1.1074,
+      "step": 509000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.787706151085623e-05,
+      "loss": 1.0999,
+      "step": 509500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7845537528040585e-05,
+      "loss": 1.1167,
+      "step": 510000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7814013545224935e-05,
+      "loss": 1.1109,
+      "step": 510500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7782489562409292e-05,
+      "loss": 1.115,
+      "step": 511000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7750965579593645e-05,
+      "loss": 1.1081,
+      "step": 511500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7719441596777995e-05,
+      "loss": 1.1029,
+      "step": 512000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7687917613962352e-05,
+      "loss": 1.1092,
+      "step": 512500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7656393631146706e-05,
+      "loss": 1.1162,
+      "step": 513000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.762486964833106e-05,
+      "loss": 1.1019,
+      "step": 513500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7593345665515412e-05,
+      "loss": 1.1105,
+      "step": 514000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7561821682699763e-05,
+      "loss": 1.1034,
+      "step": 514500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.753029769988412e-05,
+      "loss": 1.1069,
+      "step": 515000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7498773717068473e-05,
+      "loss": 1.1044,
+      "step": 515500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7467249734252826e-05,
+      "loss": 1.1081,
+      "step": 516000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.743572575143718e-05,
+      "loss": 1.109,
+      "step": 516500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7404201768621533e-05,
+      "loss": 1.1102,
+      "step": 517000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7372677785805887e-05,
+      "loss": 1.0989,
+      "step": 517500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.734115380299024e-05,
+      "loss": 1.0978,
+      "step": 518000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7309629820174593e-05,
+      "loss": 1.1085,
+      "step": 518500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7278105837358947e-05,
+      "loss": 1.1097,
+      "step": 519000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.72465818545433e-05,
+      "loss": 1.1037,
+      "step": 519500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7215057871727654e-05,
+      "loss": 1.1099,
+      "step": 520000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7183533888912007e-05,
+      "loss": 1.1201,
+      "step": 520500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.715200990609636e-05,
+      "loss": 1.1012,
+      "step": 521000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7120485923280714e-05,
+      "loss": 1.1104,
+      "step": 521500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7088961940465067e-05,
+      "loss": 1.11,
+      "step": 522000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.705743795764942e-05,
+      "loss": 1.1128,
+      "step": 522500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7025913974833774e-05,
+      "loss": 1.1035,
+      "step": 523000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6994389992018128e-05,
+      "loss": 1.0976,
+      "step": 523500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.696286600920248e-05,
+      "loss": 1.1013,
+      "step": 524000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6931342026386835e-05,
+      "loss": 1.1072,
+      "step": 524500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6899818043571188e-05,
+      "loss": 1.1065,
+      "step": 525000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6868294060755545e-05,
+      "loss": 1.0969,
+      "step": 525500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6836770077939895e-05,
+      "loss": 1.1157,
+      "step": 526000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.680524609512425e-05,
+      "loss": 1.1016,
+      "step": 526500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6773722112308605e-05,
+      "loss": 1.0994,
+      "step": 527000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6742198129492955e-05,
+      "loss": 1.1045,
+      "step": 527500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.671067414667731e-05,
+      "loss": 1.1107,
+      "step": 528000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6679150163861666e-05,
+      "loss": 1.0976,
+      "step": 528500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6647626181046016e-05,
+      "loss": 1.0981,
+      "step": 529000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6616102198230372e-05,
+      "loss": 1.0969,
+      "step": 529500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6584578215414722e-05,
+      "loss": 1.1059,
+      "step": 530000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6553054232599076e-05,
+      "loss": 1.1098,
+      "step": 530500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6521530249783433e-05,
+      "loss": 1.0973,
+      "step": 531000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6490006266967783e-05,
+      "loss": 1.0941,
+      "step": 531500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.645848228415214e-05,
+      "loss": 1.1041,
+      "step": 532000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6426958301336493e-05,
+      "loss": 1.0995,
+      "step": 532500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6395434318520843e-05,
+      "loss": 1.1055,
+      "step": 533000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.63639103357052e-05,
+      "loss": 1.0983,
+      "step": 533500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6332386352889553e-05,
+      "loss": 1.0968,
+      "step": 534000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6300862370073903e-05,
+      "loss": 1.1039,
+      "step": 534500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.626933838725826e-05,
+      "loss": 1.0923,
+      "step": 535000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6237814404442614e-05,
+      "loss": 1.0991,
+      "step": 535500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6206290421626967e-05,
+      "loss": 1.1015,
+      "step": 536000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.617476643881132e-05,
+      "loss": 1.1005,
+      "step": 536500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.614324245599567e-05,
+      "loss": 1.1046,
+      "step": 537000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6111718473180027e-05,
+      "loss": 1.0857,
+      "step": 537500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.608019449036438e-05,
+      "loss": 1.0974,
+      "step": 538000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6048670507548734e-05,
+      "loss": 1.0986,
+      "step": 538500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6017146524733088e-05,
+      "loss": 1.0967,
+      "step": 539000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.598562254191744e-05,
+      "loss": 1.0952,
+      "step": 539500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5954098559101795e-05,
+      "loss": 1.0955,
+      "step": 540000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5922574576286148e-05,
+      "loss": 1.0955,
+      "step": 540500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.58910505934705e-05,
+      "loss": 1.098,
+      "step": 541000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5859526610654855e-05,
+      "loss": 1.0937,
+      "step": 541500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5828002627839208e-05,
+      "loss": 1.09,
+      "step": 542000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5796478645023562e-05,
+      "loss": 1.0914,
+      "step": 542500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5764954662207915e-05,
+      "loss": 1.0964,
+      "step": 543000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.573343067939227e-05,
+      "loss": 1.092,
+      "step": 543500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5701906696576622e-05,
+      "loss": 1.0949,
+      "step": 544000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5670382713760975e-05,
+      "loss": 1.0893,
+      "step": 544500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.563885873094533e-05,
+      "loss": 1.1014,
+      "step": 545000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5607334748129682e-05,
+      "loss": 1.089,
+      "step": 545500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5575810765314036e-05,
+      "loss": 1.088,
+      "step": 546000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.554428678249839e-05,
+      "loss": 1.1041,
+      "step": 546500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5512762799682743e-05,
+      "loss": 1.0843,
+      "step": 547000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5481238816867096e-05,
+      "loss": 1.1017,
+      "step": 547500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5449714834051453e-05,
+      "loss": 1.0963,
+      "step": 548000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5418190851235803e-05,
+      "loss": 1.0915,
+      "step": 548500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5386666868420156e-05,
+      "loss": 1.0932,
+      "step": 549000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5355142885604513e-05,
+      "loss": 1.0918,
+      "step": 549500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5323618902788863e-05,
+      "loss": 1.0964,
+      "step": 550000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5292094919973217e-05,
+      "loss": 1.0858,
+      "step": 550500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5260570937157574e-05,
+      "loss": 1.0971,
+      "step": 551000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5229046954341925e-05,
+      "loss": 1.0964,
+      "step": 551500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5197522971526279e-05,
+      "loss": 1.0904,
+      "step": 552000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.516599898871063e-05,
+      "loss": 1.099,
+      "step": 552500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5134475005894986e-05,
+      "loss": 1.0942,
+      "step": 553000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5102951023079339e-05,
+      "loss": 1.0921,
+      "step": 553500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.507142704026369e-05,
+      "loss": 1.0842,
+      "step": 554000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5039903057448046e-05,
+      "loss": 1.0905,
+      "step": 554500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5008379074632401e-05,
+      "loss": 1.0972,
+      "step": 555000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.4976855091816753e-05,
+      "loss": 1.0933,
+      "step": 555500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.4945331109001106e-05,
+      "loss": 1.0964,
+      "step": 556000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4913807126185461e-05,
+      "loss": 1.0975,
+      "step": 556500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4882283143369813e-05,
+      "loss": 1.1022,
+      "step": 557000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4850759160554168e-05,
+      "loss": 1.0903,
+      "step": 557500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4819235177738522e-05,
+      "loss": 1.0947,
+      "step": 558000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4787711194922873e-05,
+      "loss": 1.0968,
+      "step": 558500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4756187212107228e-05,
+      "loss": 1.0953,
+      "step": 559000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.472466322929158e-05,
+      "loss": 1.0967,
+      "step": 559500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4693139246475934e-05,
+      "loss": 1.0867,
+      "step": 560000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4661615263660289e-05,
+      "loss": 1.0931,
+      "step": 560500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.463009128084464e-05,
+      "loss": 1.0917,
+      "step": 561000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4598567298028996e-05,
+      "loss": 1.0869,
+      "step": 561500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4567043315213349e-05,
+      "loss": 1.089,
+      "step": 562000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4535519332397701e-05,
+      "loss": 1.0868,
+      "step": 562500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4503995349582056e-05,
+      "loss": 1.0973,
+      "step": 563000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.447247136676641e-05,
+      "loss": 1.0955,
+      "step": 563500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4440947383950761e-05,
+      "loss": 1.0871,
+      "step": 564000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4409423401135116e-05,
+      "loss": 1.0893,
+      "step": 564500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4377899418319471e-05,
+      "loss": 1.0931,
+      "step": 565000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4346375435503823e-05,
+      "loss": 1.0915,
+      "step": 565500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4314851452688177e-05,
+      "loss": 1.0816,
+      "step": 566000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4283327469872532e-05,
+      "loss": 1.0903,
+      "step": 566500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4251803487056883e-05,
+      "loss": 1.0905,
+      "step": 567000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4220279504241239e-05,
+      "loss": 1.0958,
+      "step": 567500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.418875552142559e-05,
+      "loss": 1.0923,
+      "step": 568000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4157231538609944e-05,
+      "loss": 1.0845,
+      "step": 568500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4125707555794299e-05,
+      "loss": 1.0887,
+      "step": 569000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.409418357297865e-05,
+      "loss": 1.0908,
+      "step": 569500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4062659590163004e-05,
+      "loss": 1.0826,
+      "step": 570000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.403113560734736e-05,
+      "loss": 1.0899,
+      "step": 570500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.3999611624531711e-05,
+      "loss": 1.0909,
+      "step": 571000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.3968087641716066e-05,
+      "loss": 1.0922,
+      "step": 571500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.393656365890042e-05,
+      "loss": 1.0922,
+      "step": 572000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3905039676084771e-05,
+      "loss": 1.0844,
+      "step": 572500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3873515693269126e-05,
+      "loss": 1.0919,
+      "step": 573000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3841991710453482e-05,
+      "loss": 1.0872,
+      "step": 573500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3810467727637833e-05,
+      "loss": 1.0807,
+      "step": 574000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3778943744822187e-05,
+      "loss": 1.0824,
+      "step": 574500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3747419762006538e-05,
+      "loss": 1.0874,
+      "step": 575000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3715895779190894e-05,
+      "loss": 1.0842,
+      "step": 575500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3684371796375247e-05,
+      "loss": 1.0887,
+      "step": 576000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3652847813559599e-05,
+      "loss": 1.0903,
+      "step": 576500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3621323830743954e-05,
+      "loss": 1.0836,
+      "step": 577000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3589799847928309e-05,
+      "loss": 1.0807,
+      "step": 577500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.355827586511266e-05,
+      "loss": 1.0868,
+      "step": 578000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3526751882297014e-05,
+      "loss": 1.0871,
+      "step": 578500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.349522789948137e-05,
+      "loss": 1.0815,
+      "step": 579000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3463703916665721e-05,
+      "loss": 1.0849,
+      "step": 579500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3432179933850074e-05,
+      "loss": 1.0831,
+      "step": 580000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.340065595103443e-05,
+      "loss": 1.0814,
+      "step": 580500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3369131968218781e-05,
+      "loss": 1.0887,
+      "step": 581000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3337607985403136e-05,
+      "loss": 1.0825,
+      "step": 581500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3306084002587488e-05,
+      "loss": 1.0776,
+      "step": 582000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3274560019771842e-05,
+      "loss": 1.0856,
+      "step": 582500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3243036036956197e-05,
+      "loss": 1.0776,
+      "step": 583000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3211512054140549e-05,
+      "loss": 1.0791,
+      "step": 583500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3179988071324904e-05,
+      "loss": 1.0893,
+      "step": 584000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3148464088509257e-05,
+      "loss": 1.0837,
+      "step": 584500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3116940105693609e-05,
+      "loss": 1.0708,
+      "step": 585000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3085416122877964e-05,
+      "loss": 1.0831,
+      "step": 585500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3053892140062317e-05,
+      "loss": 1.0876,
+      "step": 586000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.302236815724667e-05,
+      "loss": 1.0829,
+      "step": 586500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2990844174431024e-05,
+      "loss": 1.0824,
+      "step": 587000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.295932019161538e-05,
+      "loss": 1.0865,
+      "step": 587500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2927796208799731e-05,
+      "loss": 1.0822,
+      "step": 588000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2896272225984085e-05,
+      "loss": 1.0788,
+      "step": 588500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.286474824316844e-05,
+      "loss": 1.0901,
+      "step": 589000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2833224260352791e-05,
+      "loss": 1.0816,
+      "step": 589500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2801700277537147e-05,
+      "loss": 1.0801,
+      "step": 590000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2770176294721498e-05,
+      "loss": 1.0791,
+      "step": 590500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2738652311905852e-05,
+      "loss": 1.0807,
+      "step": 591000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2707128329090207e-05,
+      "loss": 1.0823,
+      "step": 591500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2675604346274559e-05,
+      "loss": 1.0822,
+      "step": 592000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2644080363458912e-05,
+      "loss": 1.0845,
+      "step": 592500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2612556380643267e-05,
+      "loss": 1.0854,
+      "step": 593000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2581032397827619e-05,
+      "loss": 1.0809,
+      "step": 593500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2549508415011974e-05,
+      "loss": 1.0927,
+      "step": 594000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2517984432196328e-05,
+      "loss": 1.0874,
+      "step": 594500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2486460449380681e-05,
+      "loss": 1.0818,
+      "step": 595000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2454936466565034e-05,
+      "loss": 1.0845,
+      "step": 595500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2423412483749388e-05,
+      "loss": 1.0699,
+      "step": 596000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2391888500933741e-05,
+      "loss": 1.0745,
+      "step": 596500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2360364518118095e-05,
+      "loss": 1.0893,
+      "step": 597000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2328840535302448e-05,
+      "loss": 1.0815,
+      "step": 597500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2297316552486802e-05,
+      "loss": 1.0809,
+      "step": 598000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2265792569671155e-05,
+      "loss": 1.0717,
+      "step": 598500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2234268586855508e-05,
+      "loss": 1.0795,
+      "step": 599000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2202744604039862e-05,
+      "loss": 1.0765,
+      "step": 599500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2171220621224215e-05,
+      "loss": 1.0786,
+      "step": 600000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2139696638408569e-05,
+      "loss": 1.0821,
+      "step": 600500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2108172655592922e-05,
+      "loss": 1.0718,
+      "step": 601000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2076648672777276e-05,
+      "loss": 1.0853,
+      "step": 601500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.204512468996163e-05,
+      "loss": 1.0858,
+      "step": 602000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2013600707145983e-05,
+      "loss": 1.0842,
+      "step": 602500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1982076724330336e-05,
+      "loss": 1.0794,
+      "step": 603000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.195055274151469e-05,
+      "loss": 1.0829,
+      "step": 603500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1919028758699045e-05,
+      "loss": 1.0785,
+      "step": 604000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1887504775883398e-05,
+      "loss": 1.0776,
+      "step": 604500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.185598079306775e-05,
+      "loss": 1.0774,
+      "step": 605000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1824456810252105e-05,
+      "loss": 1.077,
+      "step": 605500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1792932827436458e-05,
+      "loss": 1.0753,
+      "step": 606000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1761408844620812e-05,
+      "loss": 1.0839,
+      "step": 606500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1729884861805163e-05,
+      "loss": 1.075,
+      "step": 607000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1698360878989519e-05,
+      "loss": 1.0784,
+      "step": 607500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1666836896173872e-05,
+      "loss": 1.0788,
+      "step": 608000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1635312913358225e-05,
+      "loss": 1.0812,
+      "step": 608500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1603788930542579e-05,
+      "loss": 1.0786,
+      "step": 609000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1572264947726932e-05,
+      "loss": 1.0876,
+      "step": 609500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1540740964911286e-05,
+      "loss": 1.0838,
+      "step": 610000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.150921698209564e-05,
+      "loss": 1.0773,
+      "step": 610500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1477692999279993e-05,
+      "loss": 1.072,
+      "step": 611000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1446169016464346e-05,
+      "loss": 1.0727,
+      "step": 611500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.14146450336487e-05,
+      "loss": 1.0839,
+      "step": 612000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1383121050833055e-05,
+      "loss": 1.0851,
+      "step": 612500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1351597068017406e-05,
+      "loss": 1.078,
+      "step": 613000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.132007308520176e-05,
+      "loss": 1.0738,
+      "step": 613500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1288549102386115e-05,
+      "loss": 1.077,
+      "step": 614000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1257025119570468e-05,
+      "loss": 1.0792,
+      "step": 614500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.122550113675482e-05,
+      "loss": 1.0807,
+      "step": 615000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1193977153939174e-05,
+      "loss": 1.078,
+      "step": 615500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1162453171123529e-05,
+      "loss": 1.0773,
+      "step": 616000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1130929188307882e-05,
+      "loss": 1.0827,
+      "step": 616500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1099405205492234e-05,
+      "loss": 1.0712,
+      "step": 617000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1067881222676589e-05,
+      "loss": 1.0739,
+      "step": 617500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1036357239860942e-05,
+      "loss": 1.0767,
+      "step": 618000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1004833257045296e-05,
+      "loss": 1.0806,
+      "step": 618500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0973309274229648e-05,
+      "loss": 1.078,
+      "step": 619000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0941785291414003e-05,
+      "loss": 1.0696,
+      "step": 619500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0910261308598356e-05,
+      "loss": 1.0727,
+      "step": 620000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.087873732578271e-05,
+      "loss": 1.0634,
+      "step": 620500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0847213342967063e-05,
+      "loss": 1.0717,
+      "step": 621000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0815689360151416e-05,
+      "loss": 1.0734,
+      "step": 621500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.078416537733577e-05,
+      "loss": 1.0742,
+      "step": 622000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0752641394520123e-05,
+      "loss": 1.0776,
+      "step": 622500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0721117411704477e-05,
+      "loss": 1.0619,
+      "step": 623000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.068959342888883e-05,
+      "loss": 1.072,
+      "step": 623500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0658069446073184e-05,
+      "loss": 1.0676,
+      "step": 624000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0626545463257539e-05,
+      "loss": 1.0761,
+      "step": 624500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.059502148044189e-05,
+      "loss": 1.0741,
+      "step": 625000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0563497497626244e-05,
+      "loss": 1.0759,
+      "step": 625500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0531973514810597e-05,
+      "loss": 1.0691,
+      "step": 626000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0500449531994953e-05,
+      "loss": 1.0722,
+      "step": 626500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0468925549179304e-05,
+      "loss": 1.0736,
+      "step": 627000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0437401566363658e-05,
+      "loss": 1.0748,
+      "step": 627500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0405877583548013e-05,
+      "loss": 1.0707,
+      "step": 628000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0374353600732366e-05,
+      "loss": 1.0712,
+      "step": 628500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.034282961791672e-05,
+      "loss": 1.0782,
+      "step": 629000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0311305635101071e-05,
+      "loss": 1.0667,
+      "step": 629500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0279781652285427e-05,
+      "loss": 1.0692,
+      "step": 630000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.024825766946978e-05,
+      "loss": 1.0671,
+      "step": 630500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0216733686654133e-05,
+      "loss": 1.0794,
+      "step": 631000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0185209703838487e-05,
+      "loss": 1.069,
+      "step": 631500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.015368572102284e-05,
+      "loss": 1.0762,
+      "step": 632000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0122161738207194e-05,
+      "loss": 1.0761,
+      "step": 632500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0090637755391547e-05,
+      "loss": 1.0722,
+      "step": 633000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.00591137725759e-05,
+      "loss": 1.0746,
+      "step": 633500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0027589789760254e-05,
+      "loss": 1.0727,
+      "step": 634000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.996065806944607e-06,
+      "loss": 1.0706,
+      "step": 634500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.964541824128961e-06,
+      "loss": 1.0719,
+      "step": 635000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.933017841313314e-06,
+      "loss": 1.0761,
+      "step": 635500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.901493858497668e-06,
+      "loss": 1.0737,
+      "step": 636000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.869969875682023e-06,
+      "loss": 1.0711,
+      "step": 636500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.838445892866376e-06,
+      "loss": 1.075,
+      "step": 637000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.806921910050728e-06,
+      "loss": 1.072,
+      "step": 637500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.775397927235082e-06,
+      "loss": 1.0734,
+      "step": 638000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.743873944419437e-06,
+      "loss": 1.0737,
+      "step": 638500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.71234996160379e-06,
+      "loss": 1.0697,
+      "step": 639000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.680825978788142e-06,
+      "loss": 1.0767,
+      "step": 639500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.649301995972497e-06,
+      "loss": 1.0688,
+      "step": 640000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.61777801315685e-06,
+      "loss": 1.0666,
+      "step": 640500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.586254030341204e-06,
+      "loss": 1.0604,
+      "step": 641000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.554730047525556e-06,
+      "loss": 1.0782,
+      "step": 641500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.52320606470991e-06,
+      "loss": 1.0756,
+      "step": 642000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.491682081894264e-06,
+      "loss": 1.0665,
+      "step": 642500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.460158099078618e-06,
+      "loss": 1.0649,
+      "step": 643000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.428634116262971e-06,
+      "loss": 1.068,
+      "step": 643500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.397110133447324e-06,
+      "loss": 1.0761,
+      "step": 644000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.365586150631678e-06,
+      "loss": 1.0633,
+      "step": 644500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.334062167816031e-06,
+      "loss": 1.0723,
+      "step": 645000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.302538185000385e-06,
+      "loss": 1.064,
+      "step": 645500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.271014202184738e-06,
+      "loss": 1.0751,
+      "step": 646000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.239490219369092e-06,
+      "loss": 1.0682,
+      "step": 646500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.207966236553447e-06,
+      "loss": 1.0813,
+      "step": 647000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.176442253737799e-06,
+      "loss": 1.073,
+      "step": 647500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.144918270922152e-06,
+      "loss": 1.0689,
+      "step": 648000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.113394288106505e-06,
+      "loss": 1.0619,
+      "step": 648500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.08187030529086e-06,
+      "loss": 1.0722,
+      "step": 649000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.050346322475212e-06,
+      "loss": 1.065,
+      "step": 649500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.018822339659566e-06,
+      "loss": 1.0622,
+      "step": 650000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.98729835684392e-06,
+      "loss": 1.0621,
+      "step": 650500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.955774374028274e-06,
+      "loss": 1.0701,
+      "step": 651000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.924250391212628e-06,
+      "loss": 1.0673,
+      "step": 651500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.892726408396981e-06,
+      "loss": 1.063,
+      "step": 652000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.861202425581335e-06,
+      "loss": 1.0746,
+      "step": 652500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.829678442765688e-06,
+      "loss": 1.0715,
+      "step": 653000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.798154459950041e-06,
+      "loss": 1.064,
+      "step": 653500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.766630477134395e-06,
+      "loss": 1.0637,
+      "step": 654000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.735106494318748e-06,
+      "loss": 1.0663,
+      "step": 654500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.703582511503102e-06,
+      "loss": 1.0693,
+      "step": 655000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.672058528687455e-06,
+      "loss": 1.0673,
+      "step": 655500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.640534545871809e-06,
+      "loss": 1.0606,
+      "step": 656000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.609010563056162e-06,
+      "loss": 1.0609,
+      "step": 656500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.577486580240515e-06,
+      "loss": 1.0726,
+      "step": 657000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.545962597424869e-06,
+      "loss": 1.0634,
+      "step": 657500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.514438614609222e-06,
+      "loss": 1.0589,
+      "step": 658000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.482914631793576e-06,
+      "loss": 1.0693,
+      "step": 658500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.451390648977931e-06,
+      "loss": 1.0608,
+      "step": 659000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.419866666162284e-06,
+      "loss": 1.07,
+      "step": 659500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.388342683346636e-06,
+      "loss": 1.0728,
+      "step": 660000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.35681870053099e-06,
+      "loss": 1.0619,
+      "step": 660500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.325294717715345e-06,
+      "loss": 1.0645,
+      "step": 661000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.293770734899698e-06,
+      "loss": 1.0606,
+      "step": 661500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.26224675208405e-06,
+      "loss": 1.0631,
+      "step": 662000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.230722769268405e-06,
+      "loss": 1.0733,
+      "step": 662500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.199198786452758e-06,
+      "loss": 1.064,
+      "step": 663000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.167674803637112e-06,
+      "loss": 1.0628,
+      "step": 663500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.136150820821464e-06,
+      "loss": 1.0653,
+      "step": 664000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.104626838005819e-06,
+      "loss": 1.0561,
+      "step": 664500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.073102855190172e-06,
+      "loss": 1.0631,
+      "step": 665000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.041578872374526e-06,
+      "loss": 1.0613,
+      "step": 665500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.010054889558879e-06,
+      "loss": 1.0615,
+      "step": 666000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.978530906743232e-06,
+      "loss": 1.0692,
+      "step": 666500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.947006923927586e-06,
+      "loss": 1.0685,
+      "step": 667000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.91548294111194e-06,
+      "loss": 1.0625,
+      "step": 667500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.883958958296293e-06,
+      "loss": 1.0702,
+      "step": 668000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.852434975480646e-06,
+      "loss": 1.0648,
+      "step": 668500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.820910992665e-06,
+      "loss": 1.0631,
+      "step": 669000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.789387009849355e-06,
+      "loss": 1.0653,
+      "step": 669500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.757863027033707e-06,
+      "loss": 1.0647,
+      "step": 670000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.72633904421806e-06,
+      "loss": 1.0595,
+      "step": 670500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.694815061402415e-06,
+      "loss": 1.0665,
+      "step": 671000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.663291078586769e-06,
+      "loss": 1.0611,
+      "step": 671500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.63176709577112e-06,
+      "loss": 1.0597,
+      "step": 672000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.600243112955474e-06,
+      "loss": 1.0616,
+      "step": 672500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.568719130139829e-06,
+      "loss": 1.0685,
+      "step": 673000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.537195147324181e-06,
+      "loss": 1.0672,
+      "step": 673500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.505671164508535e-06,
+      "loss": 1.0586,
+      "step": 674000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.474147181692889e-06,
+      "loss": 1.063,
+      "step": 674500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.4426231988772426e-06,
+      "loss": 1.0588,
+      "step": 675000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.411099216061595e-06,
+      "loss": 1.0593,
+      "step": 675500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.379575233245949e-06,
+      "loss": 1.0564,
+      "step": 676000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.348051250430303e-06,
+      "loss": 1.065,
+      "step": 676500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.316527267614656e-06,
+      "loss": 1.0657,
+      "step": 677000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.28500328479901e-06,
+      "loss": 1.0555,
+      "step": 677500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.253479301983364e-06,
+      "loss": 1.062,
+      "step": 678000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.221955319167717e-06,
+      "loss": 1.0612,
+      "step": 678500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.19043133635207e-06,
+      "loss": 1.0648,
+      "step": 679000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.1589073535364235e-06,
+      "loss": 1.0595,
+      "step": 679500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.127383370720778e-06,
+      "loss": 1.0601,
+      "step": 680000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.09585938790513e-06,
+      "loss": 1.0595,
+      "step": 680500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.064335405089484e-06,
+      "loss": 1.0532,
+      "step": 681000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.032811422273838e-06,
+      "loss": 1.0577,
+      "step": 681500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.0012874394581915e-06,
+      "loss": 1.0652,
+      "step": 682000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.969763456642545e-06,
+      "loss": 1.0537,
+      "step": 682500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.9382394738268975e-06,
+      "loss": 1.0668,
+      "step": 683000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.906715491011252e-06,
+      "loss": 1.07,
+      "step": 683500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.875191508195605e-06,
+      "loss": 1.0588,
+      "step": 684000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.843667525379959e-06,
+      "loss": 1.0543,
+      "step": 684500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.812143542564313e-06,
+      "loss": 1.0557,
+      "step": 685000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.780619559748666e-06,
+      "loss": 1.0475,
+      "step": 685500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.749095576933019e-06,
+      "loss": 1.0608,
+      "step": 686000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.7175715941173724e-06,
+      "loss": 1.0565,
+      "step": 686500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.686047611301727e-06,
+      "loss": 1.0513,
+      "step": 687000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.65452362848608e-06,
+      "loss": 1.0601,
+      "step": 687500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.622999645670433e-06,
+      "loss": 1.0524,
+      "step": 688000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.591475662854787e-06,
+      "loss": 1.0611,
+      "step": 688500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.5599516800391405e-06,
+      "loss": 1.0514,
+      "step": 689000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.528427697223494e-06,
+      "loss": 1.0609,
+      "step": 689500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.496903714407848e-06,
+      "loss": 1.0616,
+      "step": 690000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.465379731592202e-06,
+      "loss": 1.0495,
+      "step": 690500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.433855748776554e-06,
+      "loss": 1.0681,
+      "step": 691000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.402331765960908e-06,
+      "loss": 1.0629,
+      "step": 691500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.370807783145262e-06,
+      "loss": 1.0577,
+      "step": 692000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.339283800329615e-06,
+      "loss": 1.0577,
+      "step": 692500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.307759817513968e-06,
+      "loss": 1.062,
+      "step": 693000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.276235834698323e-06,
+      "loss": 1.0612,
+      "step": 693500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.244711851882676e-06,
+      "loss": 1.0552,
+      "step": 694000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.213187869067029e-06,
+      "loss": 1.0621,
+      "step": 694500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.1816638862513826e-06,
+      "loss": 1.0647,
+      "step": 695000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.150139903435736e-06,
+      "loss": 1.0584,
+      "step": 695500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.1186159206200894e-06,
+      "loss": 1.0503,
+      "step": 696000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.087091937804444e-06,
+      "loss": 1.057,
+      "step": 696500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.055567954988796e-06,
+      "loss": 1.0624,
+      "step": 697000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.024043972173151e-06,
+      "loss": 1.0536,
+      "step": 697500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.992519989357503e-06,
+      "loss": 1.0544,
+      "step": 698000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.9609960065418575e-06,
+      "loss": 1.046,
+      "step": 698500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.92947202372621e-06,
+      "loss": 1.06,
+      "step": 699000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.897948040910564e-06,
+      "loss": 1.0442,
+      "step": 699500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.866424058094918e-06,
+      "loss": 1.0605,
+      "step": 700000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.834900075279271e-06,
+      "loss": 1.0518,
+      "step": 700500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.803376092463625e-06,
+      "loss": 1.0561,
+      "step": 701000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.771852109647978e-06,
+      "loss": 1.0522,
+      "step": 701500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.7403281268323315e-06,
+      "loss": 1.0593,
+      "step": 702000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.708804144016686e-06,
+      "loss": 1.055,
+      "step": 702500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.677280161201038e-06,
+      "loss": 1.0623,
+      "step": 703000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.645756178385393e-06,
+      "loss": 1.0621,
+      "step": 703500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.614232195569745e-06,
+      "loss": 1.0551,
+      "step": 704000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.5827082127540995e-06,
+      "loss": 1.0457,
+      "step": 704500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.551184229938453e-06,
+      "loss": 1.0516,
+      "step": 705000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.519660247122806e-06,
+      "loss": 1.0605,
+      "step": 705500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.48813626430716e-06,
+      "loss": 1.0568,
+      "step": 706000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.456612281491513e-06,
+      "loss": 1.0498,
+      "step": 706500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.425088298675867e-06,
+      "loss": 1.0547,
+      "step": 707000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.39356431586022e-06,
+      "loss": 1.0549,
+      "step": 707500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.362040333044574e-06,
+      "loss": 1.0595,
+      "step": 708000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.330516350228927e-06,
+      "loss": 1.0626,
+      "step": 708500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.298992367413281e-06,
+      "loss": 1.0501,
+      "step": 709000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.267468384597635e-06,
+      "loss": 1.0593,
+      "step": 709500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.235944401781988e-06,
+      "loss": 1.0556,
+      "step": 710000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.204420418966342e-06,
+      "loss": 1.0623,
+      "step": 710500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.172896436150695e-06,
+      "loss": 1.0494,
+      "step": 711000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.1413724533350485e-06,
+      "loss": 1.0515,
+      "step": 711500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.109848470519402e-06,
+      "loss": 1.0591,
+      "step": 712000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.078324487703755e-06,
+      "loss": 1.0514,
+      "step": 712500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.04680050488811e-06,
+      "loss": 1.0553,
+      "step": 713000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.015276522072462e-06,
+      "loss": 1.0578,
+      "step": 713500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.9837525392568165e-06,
+      "loss": 1.0506,
+      "step": 714000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.952228556441169e-06,
+      "loss": 1.0539,
+      "step": 714500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.920704573625523e-06,
+      "loss": 1.0544,
+      "step": 715000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.889180590809877e-06,
+      "loss": 1.0539,
+      "step": 715500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.85765660799423e-06,
+      "loss": 1.0479,
+      "step": 716000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.826132625178584e-06,
+      "loss": 1.0552,
+      "step": 716500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.794608642362937e-06,
+      "loss": 1.0532,
+      "step": 717000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7630846595472906e-06,
+      "loss": 1.0511,
+      "step": 717500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.731560676731644e-06,
+      "loss": 1.0522,
+      "step": 718000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.7000366939159974e-06,
+      "loss": 1.0478,
+      "step": 718500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.668512711100352e-06,
+      "loss": 1.0505,
+      "step": 719000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.636988728284704e-06,
+      "loss": 1.0561,
+      "step": 719500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.605464745469059e-06,
+      "loss": 1.0526,
+      "step": 720000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.573940762653411e-06,
+      "loss": 1.0546,
+      "step": 720500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.5424167798377655e-06,
+      "loss": 1.057,
+      "step": 721000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.510892797022119e-06,
+      "loss": 1.0569,
+      "step": 721500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.479368814206472e-06,
+      "loss": 1.0517,
+      "step": 722000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.447844831390826e-06,
+      "loss": 1.0565,
+      "step": 722500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.416320848575179e-06,
+      "loss": 1.0603,
+      "step": 723000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.384796865759533e-06,
+      "loss": 1.0549,
+      "step": 723500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.353272882943886e-06,
+      "loss": 1.0549,
+      "step": 724000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3217489001282395e-06,
+      "loss": 1.0482,
+      "step": 724500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.290224917312594e-06,
+      "loss": 1.0547,
+      "step": 725000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.258700934496946e-06,
+      "loss": 1.0559,
+      "step": 725500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.227176951681301e-06,
+      "loss": 1.0486,
+      "step": 726000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.195652968865653e-06,
+      "loss": 1.0547,
+      "step": 726500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.1641289860500075e-06,
+      "loss": 1.0576,
+      "step": 727000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.13260500323436e-06,
+      "loss": 1.0508,
+      "step": 727500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.101081020418714e-06,
+      "loss": 1.055,
+      "step": 728000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.069557037603068e-06,
+      "loss": 1.053,
+      "step": 728500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.038033054787421e-06,
+      "loss": 1.0527,
+      "step": 729000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.006509071971775e-06,
+      "loss": 1.0536,
+      "step": 729500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.974985089156128e-06,
+      "loss": 1.0515,
+      "step": 730000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.943461106340482e-06,
+      "loss": 1.0488,
+      "step": 730500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.911937123524836e-06,
+      "loss": 1.0453,
+      "step": 731000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.8804131407091885e-06,
+      "loss": 1.0514,
+      "step": 731500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.848889157893543e-06,
+      "loss": 1.0507,
+      "step": 732000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.817365175077896e-06,
+      "loss": 1.0403,
+      "step": 732500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.7858411922622496e-06,
+      "loss": 1.0421,
+      "step": 733000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.7543172094466026e-06,
+      "loss": 1.0459,
+      "step": 733500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.7227932266309565e-06,
+      "loss": 1.0415,
+      "step": 734000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6912692438153103e-06,
+      "loss": 1.057,
+      "step": 734500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6597452609996634e-06,
+      "loss": 1.0531,
+      "step": 735000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6282212781840172e-06,
+      "loss": 1.0478,
+      "step": 735500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.5966972953683702e-06,
+      "loss": 1.0532,
+      "step": 736000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.565173312552724e-06,
+      "loss": 1.0481,
+      "step": 736500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.533649329737077e-06,
+      "loss": 1.043,
+      "step": 737000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.502125346921431e-06,
+      "loss": 1.0521,
+      "step": 737500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.470601364105785e-06,
+      "loss": 1.0489,
+      "step": 738000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.439077381290138e-06,
+      "loss": 1.0536,
+      "step": 738500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.4075533984744917e-06,
+      "loss": 1.0472,
+      "step": 739000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3760294156588447e-06,
+      "loss": 1.048,
+      "step": 739500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3445054328431986e-06,
+      "loss": 1.0474,
+      "step": 740000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3129814500275524e-06,
+      "loss": 1.042,
+      "step": 740500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.2814574672119054e-06,
+      "loss": 1.055,
+      "step": 741000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.2499334843962593e-06,
+      "loss": 1.048,
+      "step": 741500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.2184095015806123e-06,
+      "loss": 1.0434,
+      "step": 742000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.186885518764966e-06,
+      "loss": 1.0595,
+      "step": 742500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.155361535949319e-06,
+      "loss": 1.0464,
+      "step": 743000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.123837553133673e-06,
+      "loss": 1.0456,
+      "step": 743500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.0923135703180265e-06,
+      "loss": 1.0464,
+      "step": 744000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.06078958750238e-06,
+      "loss": 1.0556,
+      "step": 744500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0292656046867333e-06,
+      "loss": 1.0398,
+      "step": 745000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.997741621871087e-06,
+      "loss": 1.0468,
+      "step": 745500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9662176390554406e-06,
+      "loss": 1.0475,
+      "step": 746000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.934693656239794e-06,
+      "loss": 1.0468,
+      "step": 746500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.9031696734241475e-06,
+      "loss": 1.0412,
+      "step": 747000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8716456906085014e-06,
+      "loss": 1.041,
+      "step": 747500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.840121707792855e-06,
+      "loss": 1.0431,
+      "step": 748000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8085977249772082e-06,
+      "loss": 1.047,
+      "step": 748500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.777073742161562e-06,
+      "loss": 1.055,
+      "step": 749000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.7455497593459155e-06,
+      "loss": 1.0456,
+      "step": 749500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.714025776530269e-06,
+      "loss": 1.0545,
+      "step": 750000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6825017937146224e-06,
+      "loss": 1.0522,
+      "step": 750500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.650977810898976e-06,
+      "loss": 1.0421,
+      "step": 751000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6194538280833297e-06,
+      "loss": 1.0401,
+      "step": 751500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.587929845267683e-06,
+      "loss": 1.0476,
+      "step": 752000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5564058624520366e-06,
+      "loss": 1.0394,
+      "step": 752500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.52488187963639e-06,
+      "loss": 1.0383,
+      "step": 753000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4933578968207435e-06,
+      "loss": 1.0448,
+      "step": 753500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.461833914005097e-06,
+      "loss": 1.056,
+      "step": 754000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4303099311894503e-06,
+      "loss": 1.0415,
+      "step": 754500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.398785948373804e-06,
+      "loss": 1.0383,
+      "step": 755000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3672619655581576e-06,
+      "loss": 1.0481,
+      "step": 755500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.335737982742511e-06,
+      "loss": 1.0483,
+      "step": 756000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3042139999268645e-06,
+      "loss": 1.0469,
+      "step": 756500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.272690017111218e-06,
+      "loss": 1.0493,
+      "step": 757000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.2411660342955714e-06,
+      "loss": 1.0497,
+      "step": 757500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.2096420514799252e-06,
+      "loss": 1.0426,
+      "step": 758000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1781180686642787e-06,
+      "loss": 1.0399,
+      "step": 758500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.146594085848632e-06,
+      "loss": 1.0505,
+      "step": 759000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1150701030329855e-06,
+      "loss": 1.043,
+      "step": 759500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.083546120217339e-06,
+      "loss": 1.0407,
+      "step": 760000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0520221374016924e-06,
+      "loss": 1.0512,
+      "step": 760500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0204981545860463e-06,
+      "loss": 1.0439,
+      "step": 761000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9889741717703997e-06,
+      "loss": 1.0465,
+      "step": 761500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.957450188954753e-06,
+      "loss": 1.0413,
+      "step": 762000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9259262061391066e-06,
+      "loss": 1.0398,
+      "step": 762500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.89440222332346e-06,
+      "loss": 1.0443,
+      "step": 763000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8628782405078134e-06,
+      "loss": 1.0505,
+      "step": 763500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8313542576921669e-06,
+      "loss": 1.0475,
+      "step": 764000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7998302748765207e-06,
+      "loss": 1.0464,
+      "step": 764500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7683062920608742e-06,
+      "loss": 1.0379,
+      "step": 765000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7367823092452276e-06,
+      "loss": 1.0433,
+      "step": 765500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.705258326429581e-06,
+      "loss": 1.0483,
+      "step": 766000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6737343436139347e-06,
+      "loss": 1.0492,
+      "step": 766500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6422103607982881e-06,
+      "loss": 1.0481,
+      "step": 767000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6106863779826418e-06,
+      "loss": 1.0435,
+      "step": 767500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5791623951669952e-06,
+      "loss": 1.0428,
+      "step": 768000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5476384123513489e-06,
+      "loss": 1.0431,
+      "step": 768500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5161144295357023e-06,
+      "loss": 1.0402,
+      "step": 769000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4845904467200557e-06,
+      "loss": 1.0375,
+      "step": 769500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4530664639044094e-06,
+      "loss": 1.039,
+      "step": 770000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4215424810887628e-06,
+      "loss": 1.0465,
+      "step": 770500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3900184982731163e-06,
+      "loss": 1.0445,
+      "step": 771000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.35849451545747e-06,
+      "loss": 1.0495,
+      "step": 771500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3269705326418233e-06,
+      "loss": 1.0417,
+      "step": 772000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2954465498261768e-06,
+      "loss": 1.0418,
+      "step": 772500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2639225670105304e-06,
+      "loss": 1.0442,
+      "step": 773000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2323985841948839e-06,
+      "loss": 1.0474,
+      "step": 773500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2008746013792373e-06,
+      "loss": 1.0429,
+      "step": 774000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.169350618563591e-06,
+      "loss": 1.0394,
+      "step": 774500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1378266357479444e-06,
+      "loss": 1.0453,
+      "step": 775000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1063026529322978e-06,
+      "loss": 1.0387,
+      "step": 775500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0747786701166515e-06,
+      "loss": 1.0438,
+      "step": 776000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0432546873010049e-06,
+      "loss": 1.0435,
+      "step": 776500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0117307044853583e-06,
+      "loss": 1.046,
+      "step": 777000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.80206721669712e-07,
+      "loss": 1.0365,
+      "step": 777500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.486827388540655e-07,
+      "loss": 1.0443,
+      "step": 778000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.17158756038419e-07,
+      "loss": 1.0454,
+      "step": 778500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.856347732227724e-07,
+      "loss": 1.0427,
+      "step": 779000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.54110790407126e-07,
+      "loss": 1.0434,
+      "step": 779500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.225868075914795e-07,
+      "loss": 1.0393,
+      "step": 780000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.910628247758329e-07,
+      "loss": 1.0326,
+      "step": 780500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.595388419601865e-07,
+      "loss": 1.0494,
+      "step": 781000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.2801485914454e-07,
+      "loss": 1.0307,
+      "step": 781500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.964908763288935e-07,
+      "loss": 1.041,
+      "step": 782000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.64966893513247e-07,
+      "loss": 1.0451,
+      "step": 782500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.334429106976005e-07,
+      "loss": 1.0507,
+      "step": 783000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.019189278819541e-07,
+      "loss": 1.0379,
+      "step": 783500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.703949450663076e-07,
+      "loss": 1.0445,
+      "step": 784000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.388709622506611e-07,
+      "loss": 1.0423,
+      "step": 784500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.073469794350146e-07,
+      "loss": 1.0455,
+      "step": 785000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.758229966193681e-07,
+      "loss": 1.0432,
+      "step": 785500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.4429901380372166e-07,
+      "loss": 1.0409,
+      "step": 786000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.127750309880751e-07,
+      "loss": 1.0406,
+      "step": 786500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.8125104817242864e-07,
+      "loss": 1.0428,
+      "step": 787000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.497270653567821e-07,
+      "loss": 1.0337,
+      "step": 787500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.1820308254113567e-07,
+      "loss": 1.0387,
+      "step": 788000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.8667909972548916e-07,
+      "loss": 1.0403,
+      "step": 788500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.551551169098427e-07,
+      "loss": 1.046,
+      "step": 789000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.236311340941962e-07,
+      "loss": 1.0402,
+      "step": 789500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.921071512785497e-07,
+      "loss": 1.0478,
+      "step": 790000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.6058316846290322e-07,
+      "loss": 1.0432,
+      "step": 790500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.290591856472567e-07,
+      "loss": 1.0443,
+      "step": 791000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.753520283161024e-08,
+      "loss": 1.0442,
+      "step": 791500
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.601122001596375e-08,
+      "loss": 1.0512,
+      "step": 792000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.448723720031726e-08,
+      "loss": 1.0413,
+      "step": 792500
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.96325438467077e-09,
+      "loss": 1.0338,
+      "step": 793000
+    },
+    {
+      "epoch": 3.0,
+      "step": 793047,
+      "total_flos": 3.339750619344292e+18,
+      "train_loss": 0.9585861873120558,
+      "train_runtime": 247104.2719,
+      "train_samples_per_second": 51.35,
+      "train_steps_per_second": 3.209
+    }
+  ],
+  "max_steps": 793047,
+  "num_train_epochs": 3,
+  "total_flos": 3.339750619344292e+18,
+  "trial_name": null,
+  "trial_params": null
+}