diff --git "a/checkpoint-14000/trainer_state.json" "b/checkpoint-14000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-14000/trainer_state.json"
@@ -0,0 +1,8416 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.518273506127318,
+  "global_step": 14000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999996372615291e-05,
+      "loss": 1.4261,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999985490471688e-05,
+      "loss": 1.3752,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999673536007715e-05,
+      "loss": 1.3193,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999941962055173e-05,
+      "loss": 1.2661,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9999093159085765e-05,
+      "loss": 1.2613,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999869415255717e-05,
+      "loss": 1.2233,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9998222602123825e-05,
+      "loss": 1.2184,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9997678509154145e-05,
+      "loss": 1.2031,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999706187522701e-05,
+      "loss": 1.1744,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996372702131866e-05,
+      "loss": 1.1756,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995610991868615e-05,
+      "loss": 1.1611,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994776746647656e-05,
+      "loss": 1.1384,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999386996888991e-05,
+      "loss": 1.1275,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999289066122676e-05,
+      "loss": 1.1227,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999183882650006e-05,
+      "loss": 1.1135,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999071446776215e-05,
+      "loss": 1.1253,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998951758827579e-05,
+      "loss": 1.11,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998824819151424e-05,
+      "loss": 1.0942,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998690628116115e-05,
+      "loss": 1.0963,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998549186111065e-05,
+      "loss": 1.0676,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998400493546723e-05,
+      "loss": 1.0891,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998244550854582e-05,
+      "loss": 1.0612,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9980813584871734e-05,
+      "loss": 1.0726,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9979109169180664e-05,
+      "loss": 1.0721,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997733226641867e-05,
+      "loss": 1.0757,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9975482881742155e-05,
+      "loss": 1.0606,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9973561020517854e-05,
+      "loss": 1.0624,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9971566688322854e-05,
+      "loss": 1.0503,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9969499890944505e-05,
+      "loss": 1.0376,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996736063438047e-05,
+      "loss": 1.0402,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996514892483866e-05,
+      "loss": 1.0362,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996286476873727e-05,
+      "loss": 1.0363,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.996050817270471e-05,
+      "loss": 1.0354,
+      "step": 330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9958079143579594e-05,
+      "loss": 1.0437,
+      "step": 340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9955577688410744e-05,
+      "loss": 1.0223,
+      "step": 350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9953003814457154e-05,
+      "loss": 1.017,
+      "step": 360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.995035752918797e-05,
+      "loss": 1.013,
+      "step": 370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994763884028246e-05,
+      "loss": 1.0177,
+      "step": 380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994484775563002e-05,
+      "loss": 1.0445,
+      "step": 390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9941984283330115e-05,
+      "loss": 1.0078,
+      "step": 400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993904843169227e-05,
+      "loss": 1.0033,
+      "step": 410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.993604020923607e-05,
+      "loss": 1.0108,
+      "step": 420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.993295962469109e-05,
+      "loss": 1.0085,
+      "step": 430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.99298066869969e-05,
+      "loss": 1.0115,
+      "step": 440
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.992658140530304e-05,
+      "loss": 0.9952,
+      "step": 450
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.992328378896898e-05,
+      "loss": 1.0234,
+      "step": 460
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.991991384756409e-05,
+      "loss": 0.9995,
+      "step": 470
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.991647159086764e-05,
+      "loss": 0.9844,
+      "step": 480
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.991295702886874e-05,
+      "loss": 0.9989,
+      "step": 490
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.990937017176632e-05,
+      "loss": 1.0023,
+      "step": 500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.990571102996911e-05,
+      "loss": 0.9729,
+      "step": 510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.990197961409561e-05,
+      "loss": 0.9793,
+      "step": 520
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.989817593497404e-05,
+      "loss": 0.9876,
+      "step": 530
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9894300003642304e-05,
+      "loss": 0.9883,
+      "step": 540
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.989035183134802e-05,
+      "loss": 0.9997,
+      "step": 550
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9886331429548434e-05,
+      "loss": 0.9998,
+      "step": 560
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.988223880991035e-05,
+      "loss": 0.988,
+      "step": 570
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9878073984310194e-05,
+      "loss": 1.0108,
+      "step": 580
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.987383696483389e-05,
+      "loss": 0.9929,
+      "step": 590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.987126010519375e-05,
+      "loss": 1.0691,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9866907601177285e-05,
+      "loss": 0.9875,
+      "step": 610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.986248293568848e-05,
+      "loss": 0.9738,
+      "step": 620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9857986121567325e-05,
+      "loss": 0.9682,
+      "step": 630
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.985341717186316e-05,
+      "loss": 0.9955,
+      "step": 640
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.984877609983465e-05,
+      "loss": 0.9865,
+      "step": 650
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.984406291894975e-05,
+      "loss": 0.981,
+      "step": 660
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9839277642885694e-05,
+      "loss": 0.9725,
+      "step": 670
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9834420285528904e-05,
+      "loss": 0.9681,
+      "step": 680
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9829490860974975e-05,
+      "loss": 0.9632,
+      "step": 690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.982448938352865e-05,
+      "loss": 0.9877,
+      "step": 700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.981941586770376e-05,
+      "loss": 0.9808,
+      "step": 710
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9814270328223166e-05,
+      "loss": 0.9677,
+      "step": 720
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.980905278001876e-05,
+      "loss": 0.9699,
+      "step": 730
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.980376323823139e-05,
+      "loss": 0.9496,
+      "step": 740
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.97984017182108e-05,
+      "loss": 0.9669,
+      "step": 750
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.979296823551565e-05,
+      "loss": 0.9784,
+      "step": 760
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.978746280591338e-05,
+      "loss": 0.9626,
+      "step": 770
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.978188544538026e-05,
+      "loss": 0.9695,
+      "step": 780
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.977623617010127e-05,
+      "loss": 0.9575,
+      "step": 790
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.977051499647008e-05,
+      "loss": 0.9506,
+      "step": 800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9764721941089016e-05,
+      "loss": 0.9489,
+      "step": 810
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.975885702076898e-05,
+      "loss": 0.9454,
+      "step": 820
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.975292025252945e-05,
+      "loss": 0.9427,
+      "step": 830
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.974691165359836e-05,
+      "loss": 0.944,
+      "step": 840
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.974083124141211e-05,
+      "loss": 0.9147,
+      "step": 850
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9734679033615505e-05,
+      "loss": 0.9487,
+      "step": 860
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9728455048061686e-05,
+      "loss": 0.9595,
+      "step": 870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9722159302812074e-05,
+      "loss": 0.9522,
+      "step": 880
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.971579181613635e-05,
+      "loss": 0.9575,
+      "step": 890
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.970935260651237e-05,
+      "loss": 0.9474,
+      "step": 900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9702841692626125e-05,
+      "loss": 0.9533,
+      "step": 910
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.969625909337169e-05,
+      "loss": 0.9674,
+      "step": 920
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.968960482785116e-05,
+      "loss": 0.9196,
+      "step": 930
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.96828789153746e-05,
+      "loss": 0.9326,
+      "step": 940
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.967608137545998e-05,
+      "loss": 0.9546,
+      "step": 950
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.966921222783315e-05,
+      "loss": 0.937,
+      "step": 960
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.966227149242773e-05,
+      "loss": 0.9298,
+      "step": 970
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.96552591893851e-05,
+      "loss": 0.944,
+      "step": 980
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9648175339054314e-05,
+      "loss": 0.9715,
+      "step": 990
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.964101996199205e-05,
+      "loss": 0.9465,
+      "step": 1000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.963379307896255e-05,
+      "loss": 0.9526,
+      "step": 1010
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9626494710937575e-05,
+      "loss": 0.9149,
+      "step": 1020
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.96191248790963e-05,
+      "loss": 0.9608,
+      "step": 1030
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.961168360482531e-05,
+      "loss": 0.9236,
+      "step": 1040
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9604170909718494e-05,
+      "loss": 0.9455,
+      "step": 1050
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9596586815577e-05,
+      "loss": 0.9637,
+      "step": 1060
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.958893134440915e-05,
+      "loss": 0.9477,
+      "step": 1070
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.958120451843045e-05,
+      "loss": 0.9275,
+      "step": 1080
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.957340636006341e-05,
+      "loss": 0.946,
+      "step": 1090
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.956553689193758e-05,
+      "loss": 0.9226,
+      "step": 1100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9557596136889425e-05,
+      "loss": 0.9463,
+      "step": 1110
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.954958411796229e-05,
+      "loss": 0.9514,
+      "step": 1120
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.95415008584063e-05,
+      "loss": 0.9169,
+      "step": 1130
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9533346381678346e-05,
+      "loss": 0.9345,
+      "step": 1140
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.952512071144195e-05,
+      "loss": 0.9313,
+      "step": 1150
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.951682387156727e-05,
+      "loss": 0.931,
+      "step": 1160
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.950845588613096e-05,
+      "loss": 0.9292,
+      "step": 1170
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9500016779416124e-05,
+      "loss": 0.9442,
+      "step": 1180
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9491506575912306e-05,
+      "loss": 0.9307,
+      "step": 1190
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9482925300315296e-05,
+      "loss": 0.9473,
+      "step": 1200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9474272977527194e-05,
+      "loss": 0.9279,
+      "step": 1210
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.946554963265623e-05,
+      "loss": 0.9433,
+      "step": 1220
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9456755291016744e-05,
+      "loss": 0.9286,
+      "step": 1230
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9447889978129105e-05,
+      "loss": 0.9307,
+      "step": 1240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.943895371971963e-05,
+      "loss": 0.9207,
+      "step": 1250
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.942994654172053e-05,
+      "loss": 0.9067,
+      "step": 1260
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.942086847026979e-05,
+      "loss": 0.9253,
+      "step": 1270
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.941171953171115e-05,
+      "loss": 0.9072,
+      "step": 1280
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9402499752593975e-05,
+      "loss": 0.9051,
+      "step": 1290
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9393209159673217e-05,
+      "loss": 0.9076,
+      "step": 1300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.938384777990932e-05,
+      "loss": 0.9214,
+      "step": 1310
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.937441564046814e-05,
+      "loss": 0.9249,
+      "step": 1320
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.936491276872089e-05,
+      "loss": 0.9379,
+      "step": 1330
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.935533919224401e-05,
+      "loss": 0.933,
+      "step": 1340
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9345694938819146e-05,
+      "loss": 0.9228,
+      "step": 1350
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.933598003643303e-05,
+      "loss": 0.9327,
+      "step": 1360
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9326194513277415e-05,
+      "loss": 0.9198,
+      "step": 1370
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.931633839774898e-05,
+      "loss": 0.9349,
+      "step": 1380
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.930641171844926e-05,
+      "loss": 0.9296,
+      "step": 1390
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.929641450418459e-05,
+      "loss": 0.9283,
+      "step": 1400
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9286346783965925e-05,
+      "loss": 0.9015,
+      "step": 1410
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9276208587008885e-05,
+      "loss": 0.9227,
+      "step": 1420
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.926599994273358e-05,
+      "loss": 0.9059,
+      "step": 1430
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.925572088076454e-05,
+      "loss": 0.9038,
+      "step": 1440
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.924537143093068e-05,
+      "loss": 0.9109,
+      "step": 1450
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.923495162326512e-05,
+      "loss": 0.9094,
+      "step": 1460
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.92244614880052e-05,
+      "loss": 0.9292,
+      "step": 1470
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.921390105559233e-05,
+      "loss": 0.9266,
+      "step": 1480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9203270356671894e-05,
+      "loss": 0.8977,
+      "step": 1490
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.91925694220932e-05,
+      "loss": 0.9151,
+      "step": 1500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.918179828290939e-05,
+      "loss": 0.9118,
+      "step": 1510
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.91709569703773e-05,
+      "loss": 0.9259,
+      "step": 1520
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9160045515957435e-05,
+      "loss": 0.9295,
+      "step": 1530
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.914906395131381e-05,
+      "loss": 0.9152,
+      "step": 1540
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.913801230831392e-05,
+      "loss": 0.9187,
+      "step": 1550
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9126890619028623e-05,
+      "loss": 0.9371,
+      "step": 1560
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9115698915732025e-05,
+      "loss": 0.909,
+      "step": 1570
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9104437230901424e-05,
+      "loss": 0.9044,
+      "step": 1580
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.909310559721718e-05,
+      "loss": 0.9031,
+      "step": 1590
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.908170404756267e-05,
+      "loss": 0.9221,
+      "step": 1600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9070232615024115e-05,
+      "loss": 0.9227,
+      "step": 1610
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.905869133289056e-05,
+      "loss": 0.928,
+      "step": 1620
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.904708023465375e-05,
+      "loss": 0.8813,
+      "step": 1630
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.903539935400803e-05,
+      "loss": 0.8944,
+      "step": 1640
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9023648724850214e-05,
+      "loss": 0.8901,
+      "step": 1650
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.901182838127955e-05,
+      "loss": 0.923,
+      "step": 1660
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8999938357597594e-05,
+      "loss": 0.9103,
+      "step": 1670
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.89879786883081e-05,
+      "loss": 0.9013,
+      "step": 1680
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8975949408116914e-05,
+      "loss": 0.8921,
+      "step": 1690
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8963850551931904e-05,
+      "loss": 0.9103,
+      "step": 1700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8951682154862836e-05,
+      "loss": 0.9057,
+      "step": 1710
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.893944425222128e-05,
+      "loss": 0.9005,
+      "step": 1720
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.892713687952048e-05,
+      "loss": 0.8894,
+      "step": 1730
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8914760072475316e-05,
+      "loss": 0.9067,
+      "step": 1740
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.890231386700214e-05,
+      "loss": 0.9206,
+      "step": 1750
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.888979829921869e-05,
+      "loss": 0.9143,
+      "step": 1760
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.887721340544398e-05,
+      "loss": 0.8992,
+      "step": 1770
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.886455922219822e-05,
+      "loss": 0.9068,
+      "step": 1780
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.885183578620268e-05,
+      "loss": 0.8978,
+      "step": 1790
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8839043134379604e-05,
+      "loss": 0.9198,
+      "step": 1800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8826181303852075e-05,
+      "loss": 0.9042,
+      "step": 1810
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.881325033194395e-05,
+      "loss": 0.8947,
+      "step": 1820
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.880025025617971e-05,
+      "loss": 0.9095,
+      "step": 1830
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.878718111428438e-05,
+      "loss": 0.9077,
+      "step": 1840
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8774042944183404e-05,
+      "loss": 0.9092,
+      "step": 1850
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.876083578400255e-05,
+      "loss": 0.894,
+      "step": 1860
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.874755967206776e-05,
+      "loss": 0.9111,
+      "step": 1870
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8734214646905086e-05,
+      "loss": 0.8819,
+      "step": 1880
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.872080074724058e-05,
+      "loss": 0.8981,
+      "step": 1890
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.870731801200012e-05,
+      "loss": 0.8966,
+      "step": 1900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.869376648030938e-05,
+      "loss": 0.8935,
+      "step": 1910
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.868014619149364e-05,
+      "loss": 0.9013,
+      "step": 1920
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.866645718507772e-05,
+      "loss": 0.9019,
+      "step": 1930
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.865269950078587e-05,
+      "loss": 0.913,
+      "step": 1940
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.863887317854161e-05,
+      "loss": 0.893,
+      "step": 1950
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8624978258467655e-05,
+      "loss": 0.8725,
+      "step": 1960
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.861101478088578e-05,
+      "loss": 0.9065,
+      "step": 1970
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.859698278631671e-05,
+      "loss": 0.8794,
+      "step": 1980
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.858288231547999e-05,
+      "loss": 0.8948,
+      "step": 1990
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8568713409293906e-05,
+      "loss": 0.9128,
+      "step": 2000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.85544761088753e-05,
+      "loss": 0.9076,
+      "step": 2010
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8540170455539514e-05,
+      "loss": 0.8837,
+      "step": 2020
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8525796490800235e-05,
+      "loss": 0.8786,
+      "step": 2030
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.851135425636937e-05,
+      "loss": 0.8859,
+      "step": 2040
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.849684379415696e-05,
+      "loss": 0.8995,
+      "step": 2050
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8482265146271036e-05,
+      "loss": 0.873,
+      "step": 2060
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.846761835501747e-05,
+      "loss": 0.9225,
+      "step": 2070
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8452903462899914e-05,
+      "loss": 0.9137,
+      "step": 2080
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8438120512619624e-05,
+      "loss": 0.9071,
+      "step": 2090
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.842326954707536e-05,
+      "loss": 0.8829,
+      "step": 2100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.840835060936325e-05,
+      "loss": 0.8959,
+      "step": 2110
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8393363742776685e-05,
+      "loss": 0.9036,
+      "step": 2120
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.837830899080616e-05,
+      "loss": 0.8982,
+      "step": 2130
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.836318639713918e-05,
+      "loss": 0.8851,
+      "step": 2140
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.834799600566011e-05,
+      "loss": 0.8718,
+      "step": 2150
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8332737860450085e-05,
+      "loss": 0.886,
+      "step": 2160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.831741200578681e-05,
+      "loss": 0.8869,
+      "step": 2170
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.830201848614452e-05,
+      "loss": 0.8877,
+      "step": 2180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.828655734619377e-05,
+      "loss": 0.8794,
+      "step": 2190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.827102863080138e-05,
+      "loss": 0.8879,
+      "step": 2200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.825543238503025e-05,
+      "loss": 0.8873,
+      "step": 2210
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8239768654139236e-05,
+      "loss": 0.8939,
+      "step": 2220
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.822403748358305e-05,
+      "loss": 0.8906,
+      "step": 2230
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8208238919012094e-05,
+      "loss": 0.8935,
+      "step": 2240
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.819237300627234e-05,
+      "loss": 0.8671,
+      "step": 2250
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.817643979140521e-05,
+      "loss": 0.8884,
+      "step": 2260
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8160439320647426e-05,
+      "loss": 0.8919,
+      "step": 2270
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.814437164043087e-05,
+      "loss": 0.8709,
+      "step": 2280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8128236797382474e-05,
+      "loss": 0.9036,
+      "step": 2290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.811203483832406e-05,
+      "loss": 0.884,
+      "step": 2300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.809576581027223e-05,
+      "loss": 0.8862,
+      "step": 2310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8079429760438186e-05,
+      "loss": 0.8852,
+      "step": 2320
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8063026736227656e-05,
+      "loss": 0.9042,
+      "step": 2330
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.804655678524068e-05,
+      "loss": 0.8793,
+      "step": 2340
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8030019955271565e-05,
+      "loss": 0.876,
+      "step": 2350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8013416294308645e-05,
+      "loss": 0.8822,
+      "step": 2360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.799674585053422e-05,
+      "loss": 0.8813,
+      "step": 2370
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.798000867232438e-05,
+      "loss": 0.8957,
+      "step": 2380
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.796320480824887e-05,
+      "loss": 0.8813,
+      "step": 2390
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.794633430707096e-05,
+      "loss": 0.8893,
+      "step": 2400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.792939721774728e-05,
+      "loss": 0.8887,
+      "step": 2410
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.791239358942771e-05,
+      "loss": 0.8673,
+      "step": 2420
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.78953234714552e-05,
+      "loss": 0.8823,
+      "step": 2430
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7878186913365676e-05,
+      "loss": 0.8682,
+      "step": 2440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7860983964887826e-05,
+      "loss": 0.8696,
+      "step": 2450
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.784371467594304e-05,
+      "loss": 0.892,
+      "step": 2460
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.782637909664519e-05,
+      "loss": 0.8895,
+      "step": 2470
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7808977277300535e-05,
+      "loss": 0.8871,
+      "step": 2480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.779150926840754e-05,
+      "loss": 0.8726,
+      "step": 2490
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7773975120656765e-05,
+      "loss": 0.8728,
+      "step": 2500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.775637488493069e-05,
+      "loss": 0.8856,
+      "step": 2510
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7738708612303576e-05,
+      "loss": 0.8781,
+      "step": 2520
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.77209763540413e-05,
+      "loss": 0.8933,
+      "step": 2530
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.7703178161601256e-05,
+      "loss": 0.8646,
+      "step": 2540
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.768531408663216e-05,
+      "loss": 0.8702,
+      "step": 2550
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.7667384180973895e-05,
+      "loss": 0.8642,
+      "step": 2560
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.764938849665741e-05,
+      "loss": 0.8651,
+      "step": 2570
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.763132708590451e-05,
+      "loss": 0.8674,
+      "step": 2580
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.761320000112775e-05,
+      "loss": 0.8818,
+      "step": 2590
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.759500729493025e-05,
+      "loss": 0.8849,
+      "step": 2600
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.757674902010557e-05,
+      "loss": 0.8696,
+      "step": 2610
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.755842522963755e-05,
+      "loss": 0.8895,
+      "step": 2620
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7540035976700117e-05,
+      "loss": 0.88,
+      "step": 2630
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7521581314657206e-05,
+      "loss": 0.8869,
+      "step": 2640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7503061297062535e-05,
+      "loss": 0.9043,
+      "step": 2650
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.748447597765949e-05,
+      "loss": 0.8806,
+      "step": 2660
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.746582541038095e-05,
+      "loss": 0.9031,
+      "step": 2670
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7447109649349145e-05,
+      "loss": 0.8778,
+      "step": 2680
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7428328748875494e-05,
+      "loss": 0.8805,
+      "step": 2690
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.740948276346042e-05,
+      "loss": 0.876,
+      "step": 2700
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.739057174779325e-05,
+      "loss": 0.8669,
+      "step": 2710
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.737159575675201e-05,
+      "loss": 0.873,
+      "step": 2720
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.7352554845403255e-05,
+      "loss": 0.8608,
+      "step": 2730
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.733344906900197e-05,
+      "loss": 0.8711,
+      "step": 2740
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.731427848299136e-05,
+      "loss": 0.8768,
+      "step": 2750
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.729504314300268e-05,
+      "loss": 0.8557,
+      "step": 2760
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.727574310485513e-05,
+      "loss": 0.8937,
+      "step": 2770
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.725637842455563e-05,
+      "loss": 0.8685,
+      "step": 2780
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.7236949158298703e-05,
+      "loss": 0.8662,
+      "step": 2790
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.721745536246628e-05,
+      "loss": 0.8777,
+      "step": 2800
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.719789709362756e-05,
+      "loss": 0.881,
+      "step": 2810
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.717827440853885e-05,
+      "loss": 0.8807,
+      "step": 2820
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.7158587364143344e-05,
+      "loss": 0.8744,
+      "step": 2830
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.713883601757105e-05,
+      "loss": 0.8703,
+      "step": 2840
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.711902042613856e-05,
+      "loss": 0.8672,
+      "step": 2850
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.7099140647348874e-05,
+      "loss": 0.8811,
+      "step": 2860
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.7079196738891286e-05,
+      "loss": 0.8776,
+      "step": 2870
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.705918875864117e-05,
+      "loss": 0.8625,
+      "step": 2880
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.703911676465986e-05,
+      "loss": 0.8663,
+      "step": 2890
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.7018980815194405e-05,
+      "loss": 0.8666,
+      "step": 2900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.699878096867749e-05,
+      "loss": 0.8653,
+      "step": 2910
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.697851728372721e-05,
+      "loss": 0.8744,
+      "step": 2920
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.695818981914689e-05,
+      "loss": 0.8739,
+      "step": 2930
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.693779863392498e-05,
+      "loss": 0.8724,
+      "step": 2940
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.69173437872348e-05,
+      "loss": 0.865,
+      "step": 2950
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.689682533843445e-05,
+      "loss": 0.8774,
+      "step": 2960
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6876243347066554e-05,
+      "loss": 0.8527,
+      "step": 2970
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.685559787285816e-05,
+      "loss": 0.8809,
+      "step": 2980
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.683488897572052e-05,
+      "loss": 0.8898,
+      "step": 2990
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.681411671574898e-05,
+      "loss": 0.8756,
+      "step": 3000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6793281153222676e-05,
+      "loss": 0.8809,
+      "step": 3010
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6772382348604524e-05,
+      "loss": 0.8908,
+      "step": 3020
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.675142036254091e-05,
+      "loss": 0.847,
+      "step": 3030
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.673039525586159e-05,
+      "loss": 0.8743,
+      "step": 3040
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6709307089579476e-05,
+      "loss": 0.8658,
+      "step": 3050
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.668815592489049e-05,
+      "loss": 0.8778,
+      "step": 3060
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.666694182317336e-05,
+      "loss": 0.8531,
+      "step": 3070
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.664566484598946e-05,
+      "loss": 0.8645,
+      "step": 3080
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6624325055082595e-05,
+      "loss": 0.8646,
+      "step": 3090
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.660292251237889e-05,
+      "loss": 0.8622,
+      "step": 3100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.658145727998653e-05,
+      "loss": 0.8676,
+      "step": 3110
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.655992942019566e-05,
+      "loss": 0.8538,
+      "step": 3120
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.653833899547814e-05,
+      "loss": 0.8445,
+      "step": 3130
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.651668606848737e-05,
+      "loss": 0.8674,
+      "step": 3140
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6494970702058175e-05,
+      "loss": 0.8492,
+      "step": 3150
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6473192959206536e-05,
+      "loss": 0.8583,
+      "step": 3160
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.645135290312944e-05,
+      "loss": 0.8727,
+      "step": 3170
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6429450597204736e-05,
+      "loss": 0.8369,
+      "step": 3180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6407486104990887e-05,
+      "loss": 0.8857,
+      "step": 3190
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.638545949022682e-05,
+      "loss": 0.8415,
+      "step": 3200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6363370816831746e-05,
+      "loss": 0.8503,
+      "step": 3210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6341220148904954e-05,
+      "loss": 0.8828,
+      "step": 3220
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.631900755072564e-05,
+      "loss": 0.8664,
+      "step": 3230
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.629673308675271e-05,
+      "loss": 0.8692,
+      "step": 3240
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.627439682162462e-05,
+      "loss": 0.8622,
+      "step": 3250
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.625199882015913e-05,
+      "loss": 0.8896,
+      "step": 3260
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6229539147353194e-05,
+      "loss": 0.8451,
+      "step": 3270
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.62070178683827e-05,
+      "loss": 0.8462,
+      "step": 3280
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.618443504860232e-05,
+      "loss": 0.8581,
+      "step": 3290
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.616179075354532e-05,
+      "loss": 0.8679,
+      "step": 3300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.613908504892336e-05,
+      "loss": 0.8572,
+      "step": 3310
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.611631800062629e-05,
+      "loss": 0.8519,
+      "step": 3320
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.609348967472198e-05,
+      "loss": 0.8737,
+      "step": 3330
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.6070600137456147e-05,
+      "loss": 0.868,
+      "step": 3340
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.6047649455252105e-05,
+      "loss": 0.8675,
+      "step": 3350
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.6024637694710614e-05,
+      "loss": 0.8649,
+      "step": 3360
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6001564922609684e-05,
+      "loss": 0.8622,
+      "step": 3370
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.597843120590438e-05,
+      "loss": 0.8473,
+      "step": 3380
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.59552366117266e-05,
+      "loss": 0.8624,
+      "step": 3390
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.593198120738492e-05,
+      "loss": 0.8668,
+      "step": 3400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.5908665060364386e-05,
+      "loss": 0.8687,
+      "step": 3410
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.58852882383263e-05,
+      "loss": 0.8686,
+      "step": 3420
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.586185080910804e-05,
+      "loss": 0.8451,
+      "step": 3430
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.583835284072288e-05,
+      "loss": 0.8617,
+      "step": 3440
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.581479440135973e-05,
+      "loss": 0.8603,
+      "step": 3450
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.579117555938304e-05,
+      "loss": 0.8622,
+      "step": 3460
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.576749638333248e-05,
+      "loss": 0.8588,
+      "step": 3470
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.574375694192286e-05,
+      "loss": 0.8318,
+      "step": 3480
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.571995730404384e-05,
+      "loss": 0.8457,
+      "step": 3490
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.569609753875976e-05,
+      "loss": 0.8571,
+      "step": 3500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.567217771530949e-05,
+      "loss": 0.8579,
+      "step": 3510
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.564819790310613e-05,
+      "loss": 0.8474,
+      "step": 3520
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.562415817173687e-05,
+      "loss": 0.8635,
+      "step": 3530
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.560005859096282e-05,
+      "loss": 0.8867,
+      "step": 3540
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.557589923071874e-05,
+      "loss": 0.8527,
+      "step": 3550
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.555168016111284e-05,
+      "loss": 0.8579,
+      "step": 3560
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.552740145242665e-05,
+      "loss": 0.8627,
+      "step": 3570
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.550306317511474e-05,
+      "loss": 0.8579,
+      "step": 3580
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.547866539980453e-05,
+      "loss": 0.8648,
+      "step": 3590
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.5454208197296136e-05,
+      "loss": 0.8592,
+      "step": 3600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.542969163856208e-05,
+      "loss": 0.8648,
+      "step": 3610
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.540511579474717e-05,
+      "loss": 0.8578,
+      "step": 3620
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.538048073716824e-05,
+      "loss": 0.8463,
+      "step": 3630
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.5355786537313946e-05,
+      "loss": 0.8689,
+      "step": 3640
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.533103326684458e-05,
+      "loss": 0.8562,
+      "step": 3650
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.530622099759186e-05,
+      "loss": 0.8449,
+      "step": 3660
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.528134980155869e-05,
+      "loss": 0.8482,
+      "step": 3670
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.525641975091899e-05,
+      "loss": 0.8606,
+      "step": 3680
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.523143091801747e-05,
+      "loss": 0.8334,
+      "step": 3690
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.520638337536942e-05,
+      "loss": 0.8408,
+      "step": 3700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.5181277195660495e-05,
+      "loss": 0.8431,
+      "step": 3710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.5156112451746515e-05,
+      "loss": 0.8542,
+      "step": 3720
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.513088921665324e-05,
+      "loss": 0.8608,
+      "step": 3730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.510560756357619e-05,
+      "loss": 0.8454,
+      "step": 3740
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.508026756588036e-05,
+      "loss": 0.8771,
+      "step": 3750
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5054869297100116e-05,
+      "loss": 0.8584,
+      "step": 3760
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.502941283093888e-05,
+      "loss": 0.852,
+      "step": 3770
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.500389824126896e-05,
+      "loss": 0.8551,
+      "step": 3780
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.497832560213135e-05,
+      "loss": 0.8554,
+      "step": 3790
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.495269498773549e-05,
+      "loss": 0.8437,
+      "step": 3800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.492700647245906e-05,
+      "loss": 0.8695,
+      "step": 3810
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.4901260130847766e-05,
+      "loss": 0.8498,
+      "step": 3820
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.487545603761511e-05,
+      "loss": 0.84,
+      "step": 3830
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.484959426764219e-05,
+      "loss": 0.8524,
+      "step": 3840
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.482367489597748e-05,
+      "loss": 0.868,
+      "step": 3850
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.4797697997836604e-05,
+      "loss": 0.8335,
+      "step": 3860
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.477166364860213e-05,
+      "loss": 0.8438,
+      "step": 3870
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.474557192382333e-05,
+      "loss": 0.8461,
+      "step": 3880
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.471942289921599e-05,
+      "loss": 0.8529,
+      "step": 3890
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.469321665066216e-05,
+      "loss": 0.8669,
+      "step": 3900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.4666953254209965e-05,
+      "loss": 0.8252,
+      "step": 3910
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.4640632786073357e-05,
+      "loss": 0.8377,
+      "step": 3920
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.46142553226319e-05,
+      "loss": 0.8413,
+      "step": 3930
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.458782094043057e-05,
+      "loss": 0.8481,
+      "step": 3940
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.45613297161795e-05,
+      "loss": 0.8455,
+      "step": 3950
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.4534781726753775e-05,
+      "loss": 0.8496,
+      "step": 3960
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.450817704919322e-05,
+      "loss": 0.8552,
+      "step": 3970
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.4481515760702155e-05,
+      "loss": 0.8625,
+      "step": 3980
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.4454797938649174e-05,
+      "loss": 0.8525,
+      "step": 3990
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.442802366056694e-05,
+      "loss": 0.8349,
+      "step": 4000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.440119300415193e-05,
+      "loss": 0.8429,
+      "step": 4010
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.4374306047264245e-05,
+      "loss": 0.8593,
+      "step": 4020
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.4347362867927344e-05,
+      "loss": 0.845,
+      "step": 4030
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.432036354432786e-05,
+      "loss": 0.8258,
+      "step": 4040
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.4293308154815315e-05,
+      "loss": 0.8676,
+      "step": 4050
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.426619677790199e-05,
+      "loss": 0.8197,
+      "step": 4060
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.4239029492262576e-05,
+      "loss": 0.8299,
+      "step": 4070
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.421180637673403e-05,
+      "loss": 0.855,
+      "step": 4080
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.418452751031533e-05,
+      "loss": 0.865,
+      "step": 4090
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.415719297216723e-05,
+      "loss": 0.8631,
+      "step": 4100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.4129802841612036e-05,
+      "loss": 0.8532,
+      "step": 4110
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.410235719813338e-05,
+      "loss": 0.827,
+      "step": 4120
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.407485612137599e-05,
+      "loss": 0.8391,
+      "step": 4130
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.404729969114547e-05,
+      "loss": 0.8395,
+      "step": 4140
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.401968798740801e-05,
+      "loss": 0.8262,
+      "step": 4150
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.3992021090290235e-05,
+      "loss": 0.8384,
+      "step": 4160
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.396429908007894e-05,
+      "loss": 0.8218,
+      "step": 4170
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.393652203722084e-05,
+      "loss": 0.8292,
+      "step": 4180
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.390869004232234e-05,
+      "loss": 0.8496,
+      "step": 4190
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.3880803176149334e-05,
+      "loss": 0.8454,
+      "step": 4200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.385286151962693e-05,
+      "loss": 0.8475,
+      "step": 4210
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.3824865153839234e-05,
+      "loss": 0.838,
+      "step": 4220
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.3796814160029134e-05,
+      "loss": 0.8539,
+      "step": 4230
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.3768708619598e-05,
+      "loss": 0.8607,
+      "step": 4240
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.3740548614105545e-05,
+      "loss": 0.8607,
+      "step": 4250
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.3712334225269494e-05,
+      "loss": 0.8502,
+      "step": 4260
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.3684065534965405e-05,
+      "loss": 0.8495,
+      "step": 4270
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.3655742625226404e-05,
+      "loss": 0.8401,
+      "step": 4280
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.362736557824297e-05,
+      "loss": 0.8396,
+      "step": 4290
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.3598934476362676e-05,
+      "loss": 0.8483,
+      "step": 4300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.357044940208995e-05,
+      "loss": 0.8655,
+      "step": 4310
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.354191043808585e-05,
+      "loss": 0.8543,
+      "step": 4320
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.351331766716783e-05,
+      "loss": 0.8427,
+      "step": 4330
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.348467117230947e-05,
+      "loss": 0.8545,
+      "step": 4340
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.345597103664024e-05,
+      "loss": 0.8718,
+      "step": 4350
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.34272173434453e-05,
+      "loss": 0.8321,
+      "step": 4360
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.339841017616522e-05,
+      "loss": 0.8458,
+      "step": 4370
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.3369549618395735e-05,
+      "loss": 0.8447,
+      "step": 4380
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.3340635753887525e-05,
+      "loss": 0.8323,
+      "step": 4390
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.3311668666545954e-05,
+      "loss": 0.8239,
+      "step": 4400
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.328264844043084e-05,
+      "loss": 0.8496,
+      "step": 4410
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.325357515975621e-05,
+      "loss": 0.82,
+      "step": 4420
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.322444890889004e-05,
+      "loss": 0.8389,
+      "step": 4430
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.3195269772354e-05,
+      "loss": 0.8532,
+      "step": 4440
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.316603783482329e-05,
+      "loss": 0.8407,
+      "step": 4450
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.313675318112627e-05,
+      "loss": 0.8399,
+      "step": 4460
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.3107415896244316e-05,
+      "loss": 0.8303,
+      "step": 4470
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.307802606531151e-05,
+      "loss": 0.846,
+      "step": 4480
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.304858377361446e-05,
+      "loss": 0.8168,
+      "step": 4490
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.301908910659195e-05,
+      "loss": 0.8436,
+      "step": 4500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.2989542149834796e-05,
+      "loss": 0.858,
+      "step": 4510
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.2959942989085546e-05,
+      "loss": 0.8328,
+      "step": 4520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.2930291710238224e-05,
+      "loss": 0.8159,
+      "step": 4530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.290058839933811e-05,
+      "loss": 0.8253,
+      "step": 4540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.2870833142581476e-05,
+      "loss": 0.8485,
+      "step": 4550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.284102602631533e-05,
+      "loss": 0.8252,
+      "step": 4560
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.2811167137037175e-05,
+      "loss": 0.8548,
+      "step": 4570
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.278125656139476e-05,
+      "loss": 0.8376,
+      "step": 4580
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.2751294386185796e-05,
+      "loss": 0.8225,
+      "step": 4590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.272128069835778e-05,
+      "loss": 0.8409,
+      "step": 4600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.2691215585007645e-05,
+      "loss": 0.8179,
+      "step": 4610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.266109913338159e-05,
+      "loss": 0.842,
+      "step": 4620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.263093143087477e-05,
+      "loss": 0.8751,
+      "step": 4630
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.2600712565031074e-05,
+      "loss": 0.818,
+      "step": 4640
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.257044262354288e-05,
+      "loss": 0.8358,
+      "step": 4650
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.254012169425075e-05,
+      "loss": 0.8478,
+      "step": 4660
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.250974986514322e-05,
+      "loss": 0.8399,
+      "step": 4670
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.2479327224356556e-05,
+      "loss": 0.8575,
+      "step": 4680
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.244885386017445e-05,
+      "loss": 0.8392,
+      "step": 4690
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.241832986102778e-05,
+      "loss": 0.816,
+      "step": 4700
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.2387755315494394e-05,
+      "loss": 0.8579,
+      "step": 4710
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.2357130312298795e-05,
+      "loss": 0.83,
+      "step": 4720
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.232645494031191e-05,
+      "loss": 0.8281,
+      "step": 4730
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.2295729288550855e-05,
+      "loss": 0.8213,
+      "step": 4740
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.226495344617863e-05,
+      "loss": 0.8265,
+      "step": 4750
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.223412750250389e-05,
+      "loss": 0.8283,
+      "step": 4760
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.220325154698068e-05,
+      "loss": 0.8392,
+      "step": 4770
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.2172325669208176e-05,
+      "loss": 0.8189,
+      "step": 4780
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.214134995893042e-05,
+      "loss": 0.8317,
+      "step": 4790
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.211032450603608e-05,
+      "loss": 0.8369,
+      "step": 4800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.207924940055814e-05,
+      "loss": 0.8331,
+      "step": 4810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.204812473267371e-05,
+      "loss": 0.843,
+      "step": 4820
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.201695059270367e-05,
+      "loss": 0.8413,
+      "step": 4830
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1985727071112543e-05,
+      "loss": 0.8208,
+      "step": 4840
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1954454258508086e-05,
+      "loss": 0.8346,
+      "step": 4850
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.192313224564111e-05,
+      "loss": 0.8476,
+      "step": 4860
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1891761123405224e-05,
+      "loss": 0.8354,
+      "step": 4870
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1860340982836526e-05,
+      "loss": 0.8132,
+      "step": 4880
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1828871915113365e-05,
+      "loss": 0.8448,
+      "step": 4890
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1797354011556065e-05,
+      "loss": 0.8401,
+      "step": 4900
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.176578736362669e-05,
+      "loss": 0.8241,
+      "step": 4910
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.173417206292873e-05,
+      "loss": 0.8337,
+      "step": 4920
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.170250820120687e-05,
+      "loss": 0.8248,
+      "step": 4930
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.167079587034673e-05,
+      "loss": 0.8297,
+      "step": 4940
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.163903516237455e-05,
+      "loss": 0.8414,
+      "step": 4950
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1607226169457e-05,
+      "loss": 0.8399,
+      "step": 4960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.157536898390082e-05,
+      "loss": 0.828,
+      "step": 4970
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.154346369815263e-05,
+      "loss": 0.815,
+      "step": 4980
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1511510404798636e-05,
+      "loss": 0.8268,
+      "step": 4990
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.147950919656434e-05,
+      "loss": 0.8168,
+      "step": 5000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.14474601663143e-05,
+      "loss": 0.8436,
+      "step": 5010
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.141536340705184e-05,
+      "loss": 0.8193,
+      "step": 5020
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.138321901191881e-05,
+      "loss": 0.8148,
+      "step": 5030
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1351027074195266e-05,
+      "loss": 0.8288,
+      "step": 5040
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.131878768729924e-05,
+      "loss": 0.8342,
+      "step": 5050
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.128650094478647e-05,
+      "loss": 0.8439,
+      "step": 5060
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.12541669403501e-05,
+      "loss": 0.8242,
+      "step": 5070
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.122178576782042e-05,
+      "loss": 0.8328,
+      "step": 5080
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.118935752116463e-05,
+      "loss": 0.8305,
+      "step": 5090
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1156882294486484e-05,
+      "loss": 0.8179,
+      "step": 5100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.112436018202612e-05,
+      "loss": 0.8465,
+      "step": 5110
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.109179127815968e-05,
+      "loss": 0.839,
+      "step": 5120
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1059175677399136e-05,
+      "loss": 0.8353,
+      "step": 5130
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1026513474391956e-05,
+      "loss": 0.8454,
+      "step": 5140
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.099380476392083e-05,
+      "loss": 0.8232,
+      "step": 5150
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.096104964090344e-05,
+      "loss": 0.8273,
+      "step": 5160
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.09282482003921e-05,
+      "loss": 0.8308,
+      "step": 5170
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.089540053757359e-05,
+      "loss": 0.8235,
+      "step": 5180
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0862506747768784e-05,
+      "loss": 0.8293,
+      "step": 5190
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.082956692643244e-05,
+      "loss": 0.8282,
+      "step": 5200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.079658116915286e-05,
+      "loss": 0.8397,
+      "step": 5210
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.076354957165169e-05,
+      "loss": 0.8473,
+      "step": 5220
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.073047222978358e-05,
+      "loss": 0.814,
+      "step": 5230
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0697349239535905e-05,
+      "loss": 0.8313,
+      "step": 5240
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.066418069702854e-05,
+      "loss": 0.8264,
+      "step": 5250
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0630966698513545e-05,
+      "loss": 0.8327,
+      "step": 5260
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.059770734037487e-05,
+      "loss": 0.8266,
+      "step": 5270
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.05644027191281e-05,
+      "loss": 0.8322,
+      "step": 5280
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.053105293142019e-05,
+      "loss": 0.8118,
+      "step": 5290
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.049765807402912e-05,
+      "loss": 0.8257,
+      "step": 5300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.046421824386371e-05,
+      "loss": 0.8144,
+      "step": 5310
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.043073353796327e-05,
+      "loss": 0.8432,
+      "step": 5320
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0397204053497304e-05,
+      "loss": 0.8378,
+      "step": 5330
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.03636298877653e-05,
+      "loss": 0.8297,
+      "step": 5340
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.033001113819638e-05,
+      "loss": 0.8186,
+      "step": 5350
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.029634790234906e-05,
+      "loss": 0.8163,
+      "step": 5360
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0262640277910953e-05,
+      "loss": 0.824,
+      "step": 5370
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0228888362698455e-05,
+      "loss": 0.8153,
+      "step": 5380
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.019509225465654e-05,
+      "loss": 0.8221,
+      "step": 5390
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0161252051858366e-05,
+      "loss": 0.811,
+      "step": 5400
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0127367852505106e-05,
+      "loss": 0.8121,
+      "step": 5410
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0093439754925566e-05,
+      "loss": 0.8324,
+      "step": 5420
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.005946785757596e-05,
+      "loss": 0.8439,
+      "step": 5430
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0025452259039596e-05,
+      "loss": 0.8072,
+      "step": 5440
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9991393058026614e-05,
+      "loss": 0.8491,
+      "step": 5450
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9957290353373666e-05,
+      "loss": 0.8209,
+      "step": 5460
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.992314424404366e-05,
+      "loss": 0.7993,
+      "step": 5470
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9888954829125456e-05,
+      "loss": 0.8192,
+      "step": 5480
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.985472220783358e-05,
+      "loss": 0.8375,
+      "step": 5490
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.982044647950794e-05,
+      "loss": 0.8465,
+      "step": 5500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.978612774361354e-05,
+      "loss": 0.8105,
+      "step": 5510
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.975176609974019e-05,
+      "loss": 0.8238,
+      "step": 5520
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9717361647602205e-05,
+      "loss": 0.8311,
+      "step": 5530
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9682914487038135e-05,
+      "loss": 0.8303,
+      "step": 5540
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9648424718010467e-05,
+      "loss": 0.8457,
+      "step": 5550
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9613892440605324e-05,
+      "loss": 0.8166,
+      "step": 5560
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.957931775503219e-05,
+      "loss": 0.8347,
+      "step": 5570
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.954470076162362e-05,
+      "loss": 0.8253,
+      "step": 5580
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.951004156083492e-05,
+      "loss": 0.8439,
+      "step": 5590
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9475340253243906e-05,
+      "loss": 0.8354,
+      "step": 5600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.944059693955059e-05,
+      "loss": 0.8237,
+      "step": 5610
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.940581172057683e-05,
+      "loss": 0.8298,
+      "step": 5620
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.937098469726615e-05,
+      "loss": 0.8359,
+      "step": 5630
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.933611597068333e-05,
+      "loss": 0.8229,
+      "step": 5640
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.930120564201423e-05,
+      "loss": 0.8199,
+      "step": 5650
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.926625381256538e-05,
+      "loss": 0.8275,
+      "step": 5660
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9231260583763786e-05,
+      "loss": 0.8351,
+      "step": 5670
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.919622605715655e-05,
+      "loss": 0.8237,
+      "step": 5680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.916115033441066e-05,
+      "loss": 0.8156,
+      "step": 5690
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.91260335173126e-05,
+      "loss": 0.8413,
+      "step": 5700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.909087570776816e-05,
+      "loss": 0.8343,
+      "step": 5710
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.905567700780205e-05,
+      "loss": 0.8333,
+      "step": 5720
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.902043751955764e-05,
+      "loss": 0.8067,
+      "step": 5730
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.89851573452967e-05,
+      "loss": 0.8163,
+      "step": 5740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.894983658739902e-05,
+      "loss": 0.8073,
+      "step": 5750
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8914475348362194e-05,
+      "loss": 0.8144,
+      "step": 5760
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.887907373080126e-05,
+      "loss": 0.8259,
+      "step": 5770
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.884363183744847e-05,
+      "loss": 0.8102,
+      "step": 5780
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.8808149771152916e-05,
+      "loss": 0.8212,
+      "step": 5790
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.877262763488028e-05,
+      "loss": 0.8357,
+      "step": 5800
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.8737065531712535e-05,
+      "loss": 0.8211,
+      "step": 5810
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.8701463564847615e-05,
+      "loss": 0.8132,
+      "step": 5820
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.866582183759915e-05,
+      "loss": 0.796,
+      "step": 5830
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.8630140453396134e-05,
+      "loss": 0.8204,
+      "step": 5840
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.859441951578268e-05,
+      "loss": 0.8276,
+      "step": 5850
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.855865912841762e-05,
+      "loss": 0.8148,
+      "step": 5860
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.852285939507432e-05,
+      "loss": 0.7977,
+      "step": 5870
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.8487020419640296e-05,
+      "loss": 0.8046,
+      "step": 5880
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.845114230611696e-05,
+      "loss": 0.8269,
+      "step": 5890
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.841522515861927e-05,
+      "loss": 0.8089,
+      "step": 5900
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.83792690813755e-05,
+      "loss": 0.8217,
+      "step": 5910
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.834327417872686e-05,
+      "loss": 0.8226,
+      "step": 5920
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.830724055512722e-05,
+      "loss": 0.7927,
+      "step": 5930
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.827116831514286e-05,
+      "loss": 0.8212,
+      "step": 5940
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.8235057563452075e-05,
+      "loss": 0.8079,
+      "step": 5950
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.819890840484494e-05,
+      "loss": 0.8186,
+      "step": 5960
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.816272094422298e-05,
+      "loss": 0.8124,
+      "step": 5970
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.812649528659887e-05,
+      "loss": 0.8115,
+      "step": 5980
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.809023153709612e-05,
+      "loss": 0.8117,
+      "step": 5990
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.80539298009488e-05,
+      "loss": 0.7797,
+      "step": 6000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.801759018350118e-05,
+      "loss": 0.8102,
+      "step": 6010
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.798121279020751e-05,
+      "loss": 0.8157,
+      "step": 6020
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.79447977266316e-05,
+      "loss": 0.8102,
+      "step": 6030
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.790834509844661e-05,
+      "loss": 0.7939,
+      "step": 6040
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.787185501143471e-05,
+      "loss": 0.8153,
+      "step": 6050
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.7835327571486774e-05,
+      "loss": 0.8162,
+      "step": 6060
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.7798762884602054e-05,
+      "loss": 0.8358,
+      "step": 6070
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.7762161056887914e-05,
+      "loss": 0.8224,
+      "step": 6080
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.772552219455946e-05,
+      "loss": 0.8151,
+      "step": 6090
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.7688846403939296e-05,
+      "loss": 0.8112,
+      "step": 6100
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.76521337914572e-05,
+      "loss": 0.8064,
+      "step": 6110
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.761538446364977e-05,
+      "loss": 0.8182,
+      "step": 6120
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.757859852716018e-05,
+      "loss": 0.8244,
+      "step": 6130
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.7541776088737816e-05,
+      "loss": 0.8231,
+      "step": 6140
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.750491725523799e-05,
+      "loss": 0.8087,
+      "step": 6150
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.746802213362166e-05,
+      "loss": 0.8106,
+      "step": 6160
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.743109083095504e-05,
+      "loss": 0.8342,
+      "step": 6170
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.739412345440939e-05,
+      "loss": 0.8299,
+      "step": 6180
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.735712011126061e-05,
+      "loss": 0.8259,
+      "step": 6190
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.7320080908888986e-05,
+      "loss": 0.8049,
+      "step": 6200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.728300595477888e-05,
+      "loss": 0.8252,
+      "step": 6210
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.7245895356518385e-05,
+      "loss": 0.818,
+      "step": 6220
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.7208749221799036e-05,
+      "loss": 0.8304,
+      "step": 6230
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.717156765841548e-05,
+      "loss": 0.8352,
+      "step": 6240
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.713435077426519e-05,
+      "loss": 0.8114,
+      "step": 6250
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.709709867734813e-05,
+      "loss": 0.8306,
+      "step": 6260
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.7059811475766464e-05,
+      "loss": 0.8285,
+      "step": 6270
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.702248927772419e-05,
+      "loss": 0.8033,
+      "step": 6280
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.6985132191526896e-05,
+      "loss": 0.8164,
+      "step": 6290
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.69477403255814e-05,
+      "loss": 0.8201,
+      "step": 6300
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.6910313788395445e-05,
+      "loss": 0.8349,
+      "step": 6310
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.6872852688577395e-05,
+      "loss": 0.8241,
+      "step": 6320
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.68353571348359e-05,
+      "loss": 0.8308,
+      "step": 6330
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.679782723597961e-05,
+      "loss": 0.8349,
+      "step": 6340
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.6760263100916806e-05,
+      "loss": 0.8195,
+      "step": 6350
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.6722664838655176e-05,
+      "loss": 0.8272,
+      "step": 6360
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.668503255830138e-05,
+      "loss": 0.8048,
+      "step": 6370
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.6647366369060845e-05,
+      "loss": 0.808,
+      "step": 6380
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.6609666380237376e-05,
+      "loss": 0.8172,
+      "step": 6390
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.657193270123284e-05,
+      "loss": 0.8215,
+      "step": 6400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.653416544154693e-05,
+      "loss": 0.8018,
+      "step": 6410
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.649636471077672e-05,
+      "loss": 0.8089,
+      "step": 6420
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.6458530618616465e-05,
+      "loss": 0.8219,
+      "step": 6430
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.642066327485719e-05,
+      "loss": 0.8251,
+      "step": 6440
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.6382762789386456e-05,
+      "loss": 0.8173,
+      "step": 6450
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.634482927218796e-05,
+      "loss": 0.8022,
+      "step": 6460
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.6306862833341275e-05,
+      "loss": 0.8208,
+      "step": 6470
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.626886358302152e-05,
+      "loss": 0.8282,
+      "step": 6480
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.623083163149899e-05,
+      "loss": 0.8009,
+      "step": 6490
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.61927670891389e-05,
+      "loss": 0.8161,
+      "step": 6500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.615467006640107e-05,
+      "loss": 0.8307,
+      "step": 6510
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.611654067383952e-05,
+      "loss": 0.8315,
+      "step": 6520
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.607837902210224e-05,
+      "loss": 0.7931,
+      "step": 6530
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.604018522193082e-05,
+      "loss": 0.8274,
+      "step": 6540
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.6001959384160145e-05,
+      "loss": 0.8165,
+      "step": 6550
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.596370161971808e-05,
+      "loss": 0.8182,
+      "step": 6560
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.592541203962512e-05,
+      "loss": 0.8211,
+      "step": 6570
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.58870907549941e-05,
+      "loss": 0.8202,
+      "step": 6580
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.584873787702985e-05,
+      "loss": 0.8003,
+      "step": 6590
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.581035351702888e-05,
+      "loss": 0.792,
+      "step": 6600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.577193778637907e-05,
+      "loss": 0.8028,
+      "step": 6610
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.573349079655933e-05,
+      "loss": 0.809,
+      "step": 6620
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.569501265913927e-05,
+      "loss": 0.8005,
+      "step": 6630
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.56565034857789e-05,
+      "loss": 0.8029,
+      "step": 6640
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5617963388228284e-05,
+      "loss": 0.8072,
+      "step": 6650
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5579392478327235e-05,
+      "loss": 0.828,
+      "step": 6660
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.554079086800498e-05,
+      "loss": 0.805,
+      "step": 6670
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.550215866927983e-05,
+      "loss": 0.8007,
+      "step": 6680
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.546349599425885e-05,
+      "loss": 0.8285,
+      "step": 6690
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.542480295513757e-05,
+      "loss": 0.7823,
+      "step": 6700
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5386079664199624e-05,
+      "loss": 0.8109,
+      "step": 6710
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.534732623381642e-05,
+      "loss": 0.8086,
+      "step": 6720
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.530854277644684e-05,
+      "loss": 0.7941,
+      "step": 6730
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5269729404636914e-05,
+      "loss": 0.8192,
+      "step": 6740
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.523088623101945e-05,
+      "loss": 0.8069,
+      "step": 6750
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5192013368313764e-05,
+      "loss": 0.8119,
+      "step": 6760
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5153110929325316e-05,
+      "loss": 0.7996,
+      "step": 6770
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.511417902694539e-05,
+      "loss": 0.81,
+      "step": 6780
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.5075217774150796e-05,
+      "loss": 0.8106,
+      "step": 6790
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.503622728400347e-05,
+      "loss": 0.8099,
+      "step": 6800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.499720766965023e-05,
+      "loss": 0.8004,
+      "step": 6810
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.495815904432241e-05,
+      "loss": 0.8191,
+      "step": 6820
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.49190815213355e-05,
+      "loss": 0.8113,
+      "step": 6830
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.487997521408887e-05,
+      "loss": 0.8115,
+      "step": 6840
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.484084023606543e-05,
+      "loss": 0.8023,
+      "step": 6850
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.480167670083126e-05,
+      "loss": 0.8183,
+      "step": 6860
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.476248472203534e-05,
+      "loss": 0.7992,
+      "step": 6870
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.4723264413409183e-05,
+      "loss": 0.8117,
+      "step": 6880
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.468401588876648e-05,
+      "loss": 0.8263,
+      "step": 6890
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.464473926200286e-05,
+      "loss": 0.8038,
+      "step": 6900
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.4605434647095454e-05,
+      "loss": 0.8281,
+      "step": 6910
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.4566102158102636e-05,
+      "loss": 0.8256,
+      "step": 6920
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.4526741909163656e-05,
+      "loss": 0.8149,
+      "step": 6930
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.448735401449833e-05,
+      "loss": 0.8074,
+      "step": 6940
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.4447938588406693e-05,
+      "loss": 0.7954,
+      "step": 6950
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.4408495745268675e-05,
+      "loss": 0.8303,
+      "step": 6960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.436902559954378e-05,
+      "loss": 0.7915,
+      "step": 6970
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.4329528265770716e-05,
+      "loss": 0.8431,
+      "step": 6980
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.429000385856711e-05,
+      "loss": 0.8254,
+      "step": 6990
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.4250452492629147e-05,
+      "loss": 0.8296,
+      "step": 7000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.421087428273124e-05,
+      "loss": 0.8119,
+      "step": 7010
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.4171269343725703e-05,
+      "loss": 0.8029,
+      "step": 7020
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.413163779054242e-05,
+      "loss": 0.8112,
+      "step": 7030
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.40919797381885e-05,
+      "loss": 0.7941,
+      "step": 7040
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.405229530174795e-05,
+      "loss": 0.8016,
+      "step": 7050
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.401258459638136e-05,
+      "loss": 0.8219,
+      "step": 7060
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.3972847737325514e-05,
+      "loss": 0.7907,
+      "step": 7070
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.393308483989313e-05,
+      "loss": 0.8045,
+      "step": 7080
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.3893296019472446e-05,
+      "loss": 0.8052,
+      "step": 7090
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.385348139152697e-05,
+      "loss": 0.8373,
+      "step": 7100
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.381364107159507e-05,
+      "loss": 0.8262,
+      "step": 7110
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.377377517528969e-05,
+      "loss": 0.8009,
+      "step": 7120
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.373388381829797e-05,
+      "loss": 0.8099,
+      "step": 7130
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.369396711638096e-05,
+      "loss": 0.8006,
+      "step": 7140
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.365402518537324e-05,
+      "loss": 0.8048,
+      "step": 7150
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.361405814118262e-05,
+      "loss": 0.8147,
+      "step": 7160
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.357406609978976e-05,
+      "loss": 0.8474,
+      "step": 7170
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.353404917724789e-05,
+      "loss": 0.8024,
+      "step": 7180
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.349400748968241e-05,
+      "loss": 0.8363,
+      "step": 7190
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.345394115329062e-05,
+      "loss": 0.8022,
+      "step": 7200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.341385028434135e-05,
+      "loss": 0.7881,
+      "step": 7210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.337373499917457e-05,
+      "loss": 0.8024,
+      "step": 7220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.3333595414201146e-05,
+      "loss": 0.801,
+      "step": 7230
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.329343164590246e-05,
+      "loss": 0.7999,
+      "step": 7240
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.325324381083007e-05,
+      "loss": 0.8078,
+      "step": 7250
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.321303202560535e-05,
+      "loss": 0.7796,
+      "step": 7260
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.317279640691919e-05,
+      "loss": 0.8299,
+      "step": 7270
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.3132537071531664e-05,
+      "loss": 0.8144,
+      "step": 7280
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.309225413627163e-05,
+      "loss": 0.8113,
+      "step": 7290
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.305194771803647e-05,
+      "loss": 0.8137,
+      "step": 7300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.301161793379168e-05,
+      "loss": 0.7906,
+      "step": 7310
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.297126490057058e-05,
+      "loss": 0.841,
+      "step": 7320
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.293088873547394e-05,
+      "loss": 0.8137,
+      "step": 7330
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.289048955566967e-05,
+      "loss": 0.7998,
+      "step": 7340
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.2850067478392474e-05,
+      "loss": 0.7984,
+      "step": 7350
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.280962262094349e-05,
+      "loss": 0.8088,
+      "step": 7360
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.276915510068995e-05,
+      "loss": 0.8102,
+      "step": 7370
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.272866503506488e-05,
+      "loss": 0.7932,
+      "step": 7380
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.268815254156671e-05,
+      "loss": 0.8099,
+      "step": 7390
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.2647617737758964e-05,
+      "loss": 0.7982,
+      "step": 7400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.26070607412699e-05,
+      "loss": 0.8192,
+      "step": 7410
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.256648166979218e-05,
+      "loss": 0.8249,
+      "step": 7420
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.2525880641082516e-05,
+      "loss": 0.8052,
+      "step": 7430
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.2485257772961374e-05,
+      "loss": 0.8172,
+      "step": 7440
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.244461318331255e-05,
+      "loss": 0.8167,
+      "step": 7450
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.2403946990082916e-05,
+      "loss": 0.7899,
+      "step": 7460
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.236325931128199e-05,
+      "loss": 0.7975,
+      "step": 7470
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.232255026498168e-05,
+      "loss": 0.7727,
+      "step": 7480
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.228181996931587e-05,
+      "loss": 0.8108,
+      "step": 7490
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.2241068542480135e-05,
+      "loss": 0.8195,
+      "step": 7500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.220029610273134e-05,
+      "loss": 0.8102,
+      "step": 7510
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.215950276838737e-05,
+      "loss": 0.8079,
+      "step": 7520
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.211868865782669e-05,
+      "loss": 0.8081,
+      "step": 7530
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.207785388948809e-05,
+      "loss": 0.8082,
+      "step": 7540
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.203699858187033e-05,
+      "loss": 0.8188,
+      "step": 7550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.199612285353171e-05,
+      "loss": 0.8401,
+      "step": 7560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.195522682308985e-05,
+      "loss": 0.8001,
+      "step": 7570
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.191431060922123e-05,
+      "loss": 0.8197,
+      "step": 7580
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.187337433066097e-05,
+      "loss": 0.8093,
+      "step": 7590
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1832418106202325e-05,
+      "loss": 0.8188,
+      "step": 7600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.179144205469652e-05,
+      "loss": 0.8137,
+      "step": 7610
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.175044629505226e-05,
+      "loss": 0.8098,
+      "step": 7620
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.1709430946235454e-05,
+      "loss": 0.8035,
+      "step": 7630
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.1668396127268874e-05,
+      "loss": 0.7968,
+      "step": 7640
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.162734195723177e-05,
+      "loss": 0.7908,
+      "step": 7650
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.1586268555259565e-05,
+      "loss": 0.7893,
+      "step": 7660
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.154517604054348e-05,
+      "loss": 0.7937,
+      "step": 7670
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.15040645323302e-05,
+      "loss": 0.802,
+      "step": 7680
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.146293414992154e-05,
+      "loss": 0.8263,
+      "step": 7690
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.142178501267406e-05,
+      "loss": 0.8153,
+      "step": 7700
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1380617239998765e-05,
+      "loss": 0.7887,
+      "step": 7710
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.133943095136075e-05,
+      "loss": 0.8151,
+      "step": 7720
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.129822626627881e-05,
+      "loss": 0.797,
+      "step": 7730
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.125700330432514e-05,
+      "loss": 0.7991,
+      "step": 7740
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.121576218512498e-05,
+      "loss": 0.8083,
+      "step": 7750
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1174503028356264e-05,
+      "loss": 0.8359,
+      "step": 7760
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1133225953749236e-05,
+      "loss": 0.7904,
+      "step": 7770
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.109193108108618e-05,
+      "loss": 0.7883,
+      "step": 7780
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1050618530201e-05,
+      "loss": 0.7799,
+      "step": 7790
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.10092884209789e-05,
+      "loss": 0.8149,
+      "step": 7800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.096794087335606e-05,
+      "loss": 0.7949,
+      "step": 7810
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.092657600731924e-05,
+      "loss": 0.8169,
+      "step": 7820
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0885193942905464e-05,
+      "loss": 0.8379,
+      "step": 7830
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0843794800201676e-05,
+      "loss": 0.8128,
+      "step": 7840
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0802378699344365e-05,
+      "loss": 0.8,
+      "step": 7850
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0760945760519233e-05,
+      "loss": 0.8107,
+      "step": 7860
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.071949610396085e-05,
+      "loss": 0.8005,
+      "step": 7870
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0678029849952296e-05,
+      "loss": 0.798,
+      "step": 7880
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.063654711882482e-05,
+      "loss": 0.7975,
+      "step": 7890
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0595048030957474e-05,
+      "loss": 0.8053,
+      "step": 7900
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0553532706776786e-05,
+      "loss": 0.7641,
+      "step": 7910
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.051200126675639e-05,
+      "loss": 0.7971,
+      "step": 7920
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0470453831416713e-05,
+      "loss": 0.8071,
+      "step": 7930
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.042889052132456e-05,
+      "loss": 0.8083,
+      "step": 7940
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0387311457092842e-05,
+      "loss": 0.8154,
+      "step": 7950
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0345716759380162e-05,
+      "loss": 0.8071,
+      "step": 7960
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.030410654889049e-05,
+      "loss": 0.8047,
+      "step": 7970
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0262480946372818e-05,
+      "loss": 0.8147,
+      "step": 7980
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.022084007262082e-05,
+      "loss": 0.8144,
+      "step": 7990
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0179184048472453e-05,
+      "loss": 0.816,
+      "step": 8000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0137512994809673e-05,
+      "loss": 0.7913,
+      "step": 8010
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0095827032558017e-05,
+      "loss": 0.8145,
+      "step": 8020
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0054126282686318e-05,
+      "loss": 0.8089,
+      "step": 8030
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.0012410866206302e-05,
+      "loss": 0.7993,
+      "step": 8040
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.997068090417226e-05,
+      "loss": 0.7967,
+      "step": 8050
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9928936517680695e-05,
+      "loss": 0.8091,
+      "step": 8060
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.988717782786995e-05,
+      "loss": 0.8014,
+      "step": 8070
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9845404955919915e-05,
+      "loss": 0.7997,
+      "step": 8080
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9803618023051595e-05,
+      "loss": 0.7891,
+      "step": 8090
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9761817150526828e-05,
+      "loss": 0.802,
+      "step": 8100
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.972000245964788e-05,
+      "loss": 0.8076,
+      "step": 8110
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9678174071757127e-05,
+      "loss": 0.7809,
+      "step": 8120
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9636332108236693e-05,
+      "loss": 0.8037,
+      "step": 8130
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9594476690508106e-05,
+      "loss": 0.7991,
+      "step": 8140
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9552607940031928e-05,
+      "loss": 0.788,
+      "step": 8150
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9510725978307398e-05,
+      "loss": 0.8069,
+      "step": 8160
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9468830926872115e-05,
+      "loss": 0.782,
+      "step": 8170
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.942692290730165e-05,
+      "loss": 0.8017,
+      "step": 8180
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9385002041209226e-05,
+      "loss": 0.7963,
+      "step": 8190
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.934306845024531e-05,
+      "loss": 0.7963,
+      "step": 8200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9301122256097318e-05,
+      "loss": 0.7975,
+      "step": 8210
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.925916358048924e-05,
+      "loss": 0.7849,
+      "step": 8220
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9217192545181295e-05,
+      "loss": 0.7881,
+      "step": 8230
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.917520927196954e-05,
+      "loss": 0.7844,
+      "step": 8240
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9133213882685563e-05,
+      "loss": 0.7898,
+      "step": 8250
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9091206499196123e-05,
+      "loss": 0.814,
+      "step": 8260
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9049187243402755e-05,
+      "loss": 0.8039,
+      "step": 8270
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9007156237241472e-05,
+      "loss": 0.8094,
+      "step": 8280
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8965113602682387e-05,
+      "loss": 0.8005,
+      "step": 8290
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8923059461729334e-05,
+      "loss": 0.7751,
+      "step": 8300
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8880993936419553e-05,
+      "loss": 0.8017,
+      "step": 8310
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.883891714882332e-05,
+      "loss": 0.8107,
+      "step": 8320
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8796829221043603e-05,
+      "loss": 0.7895,
+      "step": 8330
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8754730275215668e-05,
+      "loss": 0.8041,
+      "step": 8340
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.871262043350679e-05,
+      "loss": 0.8119,
+      "step": 8350
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8670499818115832e-05,
+      "loss": 0.7923,
+      "step": 8360
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.862836855127294e-05,
+      "loss": 0.8087,
+      "step": 8370
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8586226755239165e-05,
+      "loss": 0.8013,
+      "step": 8380
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8544074552306114e-05,
+      "loss": 0.7797,
+      "step": 8390
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.850191206479559e-05,
+      "loss": 0.8043,
+      "step": 8400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.845973941505924e-05,
+      "loss": 0.7955,
+      "step": 8410
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.841755672547821e-05,
+      "loss": 0.8173,
+      "step": 8420
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8375364118462773e-05,
+      "loss": 0.7999,
+      "step": 8430
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8333161716451985e-05,
+      "loss": 0.7858,
+      "step": 8440
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8290949641913315e-05,
+      "loss": 0.7915,
+      "step": 8450
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8248728017342318e-05,
+      "loss": 0.8127,
+      "step": 8460
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8206496965262254e-05,
+      "loss": 0.8128,
+      "step": 8470
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8164256608223738e-05,
+      "loss": 0.8045,
+      "step": 8480
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8122007068804397e-05,
+      "loss": 0.7982,
+      "step": 8490
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.807974846960848e-05,
+      "loss": 0.8137,
+      "step": 8500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.803748093326657e-05,
+      "loss": 0.8116,
+      "step": 8510
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7995204582435132e-05,
+      "loss": 0.7922,
+      "step": 8520
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7952919539796262e-05,
+      "loss": 0.8051,
+      "step": 8530
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7910625928057238e-05,
+      "loss": 0.8025,
+      "step": 8540
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7868323869950232e-05,
+      "loss": 0.7915,
+      "step": 8550
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7826013488231895e-05,
+      "loss": 0.7887,
+      "step": 8560
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.778369490568307e-05,
+      "loss": 0.8079,
+      "step": 8570
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7741368245108374e-05,
+      "loss": 0.8019,
+      "step": 8580
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.769903362933588e-05,
+      "loss": 0.8015,
+      "step": 8590
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7656691181216727e-05,
+      "loss": 0.7921,
+      "step": 8600
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7614341023624796e-05,
+      "loss": 0.8096,
+      "step": 8610
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7571983279456342e-05,
+      "loss": 0.7895,
+      "step": 8620
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.752961807162963e-05,
+      "loss": 0.8286,
+      "step": 8630
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.7487245523084593e-05,
+      "loss": 0.794,
+      "step": 8640
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.744486575678245e-05,
+      "loss": 0.8082,
+      "step": 8650
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.7402478895705375e-05,
+      "loss": 0.7944,
+      "step": 8660
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.7360085062856128e-05,
+      "loss": 0.7839,
+      "step": 8670
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.7317684381257713e-05,
+      "loss": 0.7963,
+      "step": 8680
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.7275276973952985e-05,
+      "loss": 0.8032,
+      "step": 8690
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.723286296400433e-05,
+      "loss": 0.798,
+      "step": 8700
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.7190442474493293e-05,
+      "loss": 0.8079,
+      "step": 8710
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.714801562852023e-05,
+      "loss": 0.8168,
+      "step": 8720
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.710558254920393e-05,
+      "loss": 0.8057,
+      "step": 8730
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.706314335968127e-05,
+      "loss": 0.7757,
+      "step": 8740
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7020698183106878e-05,
+      "loss": 0.8014,
+      "step": 8750
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.697824714265272e-05,
+      "loss": 0.7877,
+      "step": 8760
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6935790361507824e-05,
+      "loss": 0.7899,
+      "step": 8770
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6893327962877835e-05,
+      "loss": 0.8005,
+      "step": 8780
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6850860069984723e-05,
+      "loss": 0.8048,
+      "step": 8790
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6808386806066406e-05,
+      "loss": 0.7824,
+      "step": 8800
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6765908294376356e-05,
+      "loss": 0.7823,
+      "step": 8810
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6723424658183318e-05,
+      "loss": 0.8098,
+      "step": 8820
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.668093602077087e-05,
+      "loss": 0.8036,
+      "step": 8830
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.663844250543714e-05,
+      "loss": 0.788,
+      "step": 8840
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6595944235494364e-05,
+      "loss": 0.8005,
+      "step": 8850
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6553441334268618e-05,
+      "loss": 0.8036,
+      "step": 8860
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6510933925099397e-05,
+      "loss": 0.7907,
+      "step": 8870
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6468422131339288e-05,
+      "loss": 0.7896,
+      "step": 8880
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.642590607635358e-05,
+      "loss": 0.7986,
+      "step": 8890
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6383385883519956e-05,
+      "loss": 0.8074,
+      "step": 8900
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6340861676228097e-05,
+      "loss": 0.8028,
+      "step": 8910
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6298333577879313e-05,
+      "loss": 0.7937,
+      "step": 8920
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.625580171188625e-05,
+      "loss": 0.8017,
+      "step": 8930
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.621326620167243e-05,
+      "loss": 0.786,
+      "step": 8940
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.617072717067201e-05,
+      "loss": 0.7993,
+      "step": 8950
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6128184742329305e-05,
+      "loss": 0.7812,
+      "step": 8960
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6085639040098543e-05,
+      "loss": 0.7905,
+      "step": 8970
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6043090187443408e-05,
+      "loss": 0.8141,
+      "step": 8980
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6000538307836764e-05,
+      "loss": 0.7958,
+      "step": 8990
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5957983524760233e-05,
+      "loss": 0.79,
+      "step": 9000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.591542596170386e-05,
+      "loss": 0.7787,
+      "step": 9010
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.587286574216579e-05,
+      "loss": 0.8162,
+      "step": 9020
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5830302989651833e-05,
+      "loss": 0.8013,
+      "step": 9030
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.578773782767519e-05,
+      "loss": 0.7936,
+      "step": 9040
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5745170379756024e-05,
+      "loss": 0.8022,
+      "step": 9050
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5702600769421138e-05,
+      "loss": 0.7968,
+      "step": 9060
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5660029120203626e-05,
+      "loss": 0.8022,
+      "step": 9070
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5617455555642484e-05,
+      "loss": 0.7955,
+      "step": 9080
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5574880199282268e-05,
+      "loss": 0.7936,
+      "step": 9090
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5532303174672733e-05,
+      "loss": 0.8086,
+      "step": 9100
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5489724605368488e-05,
+      "loss": 0.8017,
+      "step": 9110
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.54471446149286e-05,
+      "loss": 0.8144,
+      "step": 9120
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5404563326916286e-05,
+      "loss": 0.8061,
+      "step": 9130
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5361980864898506e-05,
+      "loss": 0.7975,
+      "step": 9140
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5319397352445644e-05,
+      "loss": 0.7965,
+      "step": 9150
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.527681291313112e-05,
+      "loss": 0.7983,
+      "step": 9160
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5234227670531065e-05,
+      "loss": 0.7867,
+      "step": 9170
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5191641748223904e-05,
+      "loss": 0.8032,
+      "step": 9180
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5149055269790066e-05,
+      "loss": 0.7993,
+      "step": 9190
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5106468358811586e-05,
+      "loss": 0.7942,
+      "step": 9200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5063881138871743e-05,
+      "loss": 0.8025,
+      "step": 9210
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5021293733554735e-05,
+      "loss": 0.7834,
+      "step": 9220
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4978706266445277e-05,
+      "loss": 0.7778,
+      "step": 9230
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4936118861128263e-05,
+      "loss": 0.7983,
+      "step": 9240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4893531641188423e-05,
+      "loss": 0.7982,
+      "step": 9250
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4850944730209933e-05,
+      "loss": 0.7904,
+      "step": 9260
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4808358251776108e-05,
+      "loss": 0.811,
+      "step": 9270
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4765772329468945e-05,
+      "loss": 0.7693,
+      "step": 9280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4723187086868882e-05,
+      "loss": 0.7996,
+      "step": 9290
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.468060264755436e-05,
+      "loss": 0.7812,
+      "step": 9300
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4638019135101493e-05,
+      "loss": 0.7884,
+      "step": 9310
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4595436673083724e-05,
+      "loss": 0.7707,
+      "step": 9320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4552855385071407e-05,
+      "loss": 0.7846,
+      "step": 9330
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.451027539463152e-05,
+      "loss": 0.7941,
+      "step": 9340
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.446769682532727e-05,
+      "loss": 0.8003,
+      "step": 9350
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.442511980071774e-05,
+      "loss": 0.79,
+      "step": 9360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4382544444357526e-05,
+      "loss": 0.8067,
+      "step": 9370
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.433997087979638e-05,
+      "loss": 0.7922,
+      "step": 9380
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4297399230578864e-05,
+      "loss": 0.7938,
+      "step": 9390
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.425482962024398e-05,
+      "loss": 0.7946,
+      "step": 9400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.421226217232482e-05,
+      "loss": 0.7872,
+      "step": 9410
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.416969701034817e-05,
+      "loss": 0.7982,
+      "step": 9420
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4127134257834215e-05,
+      "loss": 0.7831,
+      "step": 9430
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.408457403829614e-05,
+      "loss": 0.808,
+      "step": 9440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.404201647523978e-05,
+      "loss": 0.8034,
+      "step": 9450
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.3999461692163242e-05,
+      "loss": 0.794,
+      "step": 9460
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.3956909812556594e-05,
+      "loss": 0.7959,
+      "step": 9470
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.3914360959901463e-05,
+      "loss": 0.7757,
+      "step": 9480
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.3871815257670697e-05,
+      "loss": 0.8065,
+      "step": 9490
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.3829272829328e-05,
+      "loss": 0.7843,
+      "step": 9500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.378673379832757e-05,
+      "loss": 0.8021,
+      "step": 9510
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.3744198288113758e-05,
+      "loss": 0.7796,
+      "step": 9520
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.3701666422120683e-05,
+      "loss": 0.7868,
+      "step": 9530
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.3659138323771913e-05,
+      "loss": 0.7755,
+      "step": 9540
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.3616614116480046e-05,
+      "loss": 0.8032,
+      "step": 9550
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.3574093923646423e-05,
+      "loss": 0.7865,
+      "step": 9560
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.353157786866072e-05,
+      "loss": 0.778,
+      "step": 9570
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.3489066074900605e-05,
+      "loss": 0.7935,
+      "step": 9580
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.3446558665731388e-05,
+      "loss": 0.7779,
+      "step": 9590
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.340405576450564e-05,
+      "loss": 0.7938,
+      "step": 9600
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.336155749456287e-05,
+      "loss": 0.7763,
+      "step": 9610
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.331906397922913e-05,
+      "loss": 0.7762,
+      "step": 9620
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.3276575341816695e-05,
+      "loss": 0.8031,
+      "step": 9630
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.3234091705623647e-05,
+      "loss": 0.7876,
+      "step": 9640
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.3191613193933603e-05,
+      "loss": 0.7888,
+      "step": 9650
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.314913993001528e-05,
+      "loss": 0.7854,
+      "step": 9660
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.3106672037122164e-05,
+      "loss": 0.7673,
+      "step": 9670
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.306420963849219e-05,
+      "loss": 0.8118,
+      "step": 9680
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.3021752857347283e-05,
+      "loss": 0.7974,
+      "step": 9690
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.2979301816893124e-05,
+      "loss": 0.8034,
+      "step": 9700
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.2936856640318727e-05,
+      "loss": 0.7781,
+      "step": 9710
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.289441745079608e-05,
+      "loss": 0.7892,
+      "step": 9720
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.2851984371479777e-05,
+      "loss": 0.793,
+      "step": 9730
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.280955752550671e-05,
+      "loss": 0.7948,
+      "step": 9740
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.2767137035995674e-05,
+      "loss": 0.7838,
+      "step": 9750
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.2724723026047018e-05,
+      "loss": 0.7744,
+      "step": 9760
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.2682315618742297e-05,
+      "loss": 0.7869,
+      "step": 9770
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.2639914937143878e-05,
+      "loss": 0.7912,
+      "step": 9780
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.2597521104294634e-05,
+      "loss": 0.7919,
+      "step": 9790
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.2555134243217552e-05,
+      "loss": 0.7968,
+      "step": 9800
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.2512754476915416e-05,
+      "loss": 0.8021,
+      "step": 9810
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.2470381928370372e-05,
+      "loss": 0.7978,
+      "step": 9820
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.2428016720543664e-05,
+      "loss": 0.7961,
+      "step": 9830
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.238565897637521e-05,
+      "loss": 0.7685,
+      "step": 9840
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.2343308818783286e-05,
+      "loss": 0.7759,
+      "step": 9850
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.230096637066413e-05,
+      "loss": 0.7868,
+      "step": 9860
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.225863175489163e-05,
+      "loss": 0.7858,
+      "step": 9870
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.2216305094316935e-05,
+      "loss": 0.7827,
+      "step": 9880
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.217398651176811e-05,
+      "loss": 0.8009,
+      "step": 9890
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.2131676130049784e-05,
+      "loss": 0.8079,
+      "step": 9900
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.2089374071942768e-05,
+      "loss": 0.7905,
+      "step": 9910
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.2047080460203744e-05,
+      "loss": 0.7882,
+      "step": 9920
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.2004795417564867e-05,
+      "loss": 0.7709,
+      "step": 9930
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.196251906673344e-05,
+      "loss": 0.7924,
+      "step": 9940
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.1920251530391523e-05,
+      "loss": 0.781,
+      "step": 9950
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.1877992931195612e-05,
+      "loss": 0.782,
+      "step": 9960
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.1835743391776265e-05,
+      "loss": 0.7779,
+      "step": 9970
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.179350303473775e-05,
+      "loss": 0.7873,
+      "step": 9980
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.1751271982657688e-05,
+      "loss": 0.7751,
+      "step": 9990
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.1709050358086687e-05,
+      "loss": 0.8038,
+      "step": 10000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.1666838283548017e-05,
+      "loss": 0.7752,
+      "step": 10010
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.162463588153723e-05,
+      "loss": 0.8003,
+      "step": 10020
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.1582443274521792e-05,
+      "loss": 0.7957,
+      "step": 10030
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.1540260584940762e-05,
+      "loss": 0.7894,
+      "step": 10040
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.1498087935204413e-05,
+      "loss": 0.7883,
+      "step": 10050
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.1455925447693888e-05,
+      "loss": 0.7772,
+      "step": 10060
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.1413773244760837e-05,
+      "loss": 0.7714,
+      "step": 10070
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.1371631448727066e-05,
+      "loss": 0.7829,
+      "step": 10080
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.1329500181884177e-05,
+      "loss": 0.7617,
+      "step": 10090
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.1287379566493214e-05,
+      "loss": 0.7938,
+      "step": 10100
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.1245269724784334e-05,
+      "loss": 0.7639,
+      "step": 10110
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.120317077895641e-05,
+      "loss": 0.79,
+      "step": 10120
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.1161082851176687e-05,
+      "loss": 0.7532,
+      "step": 10130
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.111900606358045e-05,
+      "loss": 0.7974,
+      "step": 10140
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.1076940538270672e-05,
+      "loss": 0.7839,
+      "step": 10150
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.1034886397317615e-05,
+      "loss": 0.7922,
+      "step": 10160
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.0992843762758533e-05,
+      "loss": 0.7867,
+      "step": 10170
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.0950812756597254e-05,
+      "loss": 0.7931,
+      "step": 10180
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.0908793500803883e-05,
+      "loss": 0.7888,
+      "step": 10190
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.0866786117314436e-05,
+      "loss": 0.7798,
+      "step": 10200
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.0824790728030473e-05,
+      "loss": 0.7994,
+      "step": 10210
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.0782807454818715e-05,
+      "loss": 0.7991,
+      "step": 10220
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.074083641951076e-05,
+      "loss": 0.7778,
+      "step": 10230
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.0698877743902688e-05,
+      "loss": 0.7943,
+      "step": 10240
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.0656931549754693e-05,
+      "loss": 0.7848,
+      "step": 10250
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.0614997958790786e-05,
+      "loss": 0.7698,
+      "step": 10260
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.0573077092698352e-05,
+      "loss": 0.7853,
+      "step": 10270
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.053116907312789e-05,
+      "loss": 0.7852,
+      "step": 10280
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.0489274021692605e-05,
+      "loss": 0.7626,
+      "step": 10290
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.0447392059968084e-05,
+      "loss": 0.781,
+      "step": 10300
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.04055233094919e-05,
+      "loss": 0.7785,
+      "step": 10310
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.036366789176331e-05,
+      "loss": 0.7971,
+      "step": 10320
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.0321825928242882e-05,
+      "loss": 0.783,
+      "step": 10330
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.027999754035213e-05,
+      "loss": 0.7816,
+      "step": 10340
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.0238182849473188e-05,
+      "loss": 0.7995,
+      "step": 10350
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.019638197694841e-05,
+      "loss": 0.7744,
+      "step": 10360
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.0154595044080094e-05,
+      "loss": 0.7955,
+      "step": 10370
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.0112822172130053e-05,
+      "loss": 0.7629,
+      "step": 10380
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.0071063482319317e-05,
+      "loss": 0.7978,
+      "step": 10390
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.0029319095827748e-05,
+      "loss": 0.782,
+      "step": 10400
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.99875891337937e-05,
+      "loss": 0.7997,
+      "step": 10410
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.994587371731368e-05,
+      "loss": 0.7931,
+      "step": 10420
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9904172967441986e-05,
+      "loss": 0.7998,
+      "step": 10430
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9862487005190336e-05,
+      "loss": 0.778,
+      "step": 10440
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9820815951527552e-05,
+      "loss": 0.791,
+      "step": 10450
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9779159927379186e-05,
+      "loss": 0.7715,
+      "step": 10460
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.973751905362718e-05,
+      "loss": 0.7834,
+      "step": 10470
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.969589345110952e-05,
+      "loss": 0.7691,
+      "step": 10480
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9654283240619847e-05,
+      "loss": 0.7903,
+      "step": 10490
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.961268854290716e-05,
+      "loss": 0.7793,
+      "step": 10500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.957110947867544e-05,
+      "loss": 0.7856,
+      "step": 10510
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9529546168583303e-05,
+      "loss": 0.7966,
+      "step": 10520
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9487998733243613e-05,
+      "loss": 0.7841,
+      "step": 10530
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9446467293223224e-05,
+      "loss": 0.7884,
+      "step": 10540
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9404951969042535e-05,
+      "loss": 0.7869,
+      "step": 10550
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9363452881175186e-05,
+      "loss": 0.7733,
+      "step": 10560
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9321970150047713e-05,
+      "loss": 0.7872,
+      "step": 10570
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9280503896039154e-05,
+      "loss": 0.7844,
+      "step": 10580
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.923905423948077e-05,
+      "loss": 0.7834,
+      "step": 10590
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9197621300655637e-05,
+      "loss": 0.7861,
+      "step": 10600
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9156205199798333e-05,
+      "loss": 0.7867,
+      "step": 10610
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9114806057094546e-05,
+      "loss": 0.774,
+      "step": 10620
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9073423992680766e-05,
+      "loss": 0.7713,
+      "step": 10630
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9032059126643946e-05,
+      "loss": 0.7845,
+      "step": 10640
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8990711579021102e-05,
+      "loss": 0.7712,
+      "step": 10650
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8949381469799014e-05,
+      "loss": 0.7582,
+      "step": 10660
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.890806891891383e-05,
+      "loss": 0.788,
+      "step": 10670
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.886677404625077e-05,
+      "loss": 0.7715,
+      "step": 10680
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.882549697164374e-05,
+      "loss": 0.7896,
+      "step": 10690
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8784237814875026e-05,
+      "loss": 0.7864,
+      "step": 10700
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8742996695674867e-05,
+      "loss": 0.7759,
+      "step": 10710
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.87017737337212e-05,
+      "loss": 0.7659,
+      "step": 10720
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.866056904863926e-05,
+      "loss": 0.7772,
+      "step": 10730
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.861938276000123e-05,
+      "loss": 0.7864,
+      "step": 10740
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.857821498732595e-05,
+      "loss": 0.7808,
+      "step": 10750
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.853706585007847e-05,
+      "loss": 0.7719,
+      "step": 10760
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8495935467669803e-05,
+      "loss": 0.7581,
+      "step": 10770
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8454823959456525e-05,
+      "loss": 0.7798,
+      "step": 10780
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8413731444740444e-05,
+      "loss": 0.7903,
+      "step": 10790
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8372658042768238e-05,
+      "loss": 0.757,
+      "step": 10800
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8331603872731135e-05,
+      "loss": 0.7701,
+      "step": 10810
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.829056905376455e-05,
+      "loss": 0.7824,
+      "step": 10820
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8249553704947748e-05,
+      "loss": 0.7791,
+      "step": 10830
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.820855794530349e-05,
+      "loss": 0.7834,
+      "step": 10840
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.816758189379768e-05,
+      "loss": 0.7669,
+      "step": 10850
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8126625669339044e-05,
+      "loss": 0.7746,
+      "step": 10860
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8085689390778764e-05,
+      "loss": 0.7854,
+      "step": 10870
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8044773176910156e-05,
+      "loss": 0.776,
+      "step": 10880
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.800387714646829e-05,
+      "loss": 0.755,
+      "step": 10890
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7963001418129676e-05,
+      "loss": 0.7843,
+      "step": 10900
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7922146110511906e-05,
+      "loss": 0.7624,
+      "step": 10910
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.788131134217332e-05,
+      "loss": 0.7895,
+      "step": 10920
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7840497231612647e-05,
+      "loss": 0.7999,
+      "step": 10930
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7799703897268665e-05,
+      "loss": 0.7859,
+      "step": 10940
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7758931457519875e-05,
+      "loss": 0.8076,
+      "step": 10950
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7718180030684132e-05,
+      "loss": 0.787,
+      "step": 10960
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7677449735018327e-05,
+      "loss": 0.7622,
+      "step": 10970
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7636740688718013e-05,
+      "loss": 0.7788,
+      "step": 10980
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7596053009917086e-05,
+      "loss": 0.7642,
+      "step": 10990
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7555386816687443e-05,
+      "loss": 0.7939,
+      "step": 11000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.751474222703863e-05,
+      "loss": 0.7772,
+      "step": 11010
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7474119358917483e-05,
+      "loss": 0.7853,
+      "step": 11020
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7433518330207833e-05,
+      "loss": 0.7829,
+      "step": 11030
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7392939258730108e-05,
+      "loss": 0.7865,
+      "step": 11040
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.735238226224104e-05,
+      "loss": 0.8001,
+      "step": 11050
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.73118474584333e-05,
+      "loss": 0.7947,
+      "step": 11060
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7271334964935124e-05,
+      "loss": 0.7801,
+      "step": 11070
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.723084489931005e-05,
+      "loss": 0.7975,
+      "step": 11080
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7190377379056515e-05,
+      "loss": 0.8021,
+      "step": 11090
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7149932521607535e-05,
+      "loss": 0.7892,
+      "step": 11100
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7109510444330333e-05,
+      "loss": 0.8213,
+      "step": 11110
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.706911126452607e-05,
+      "loss": 0.7747,
+      "step": 11120
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7028735099429424e-05,
+      "loss": 0.7826,
+      "step": 11130
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.698838206620832e-05,
+      "loss": 0.7748,
+      "step": 11140
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.6948052281963536e-05,
+      "loss": 0.7814,
+      "step": 11150
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.690774586372837e-05,
+      "loss": 0.7813,
+      "step": 11160
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.686746292846834e-05,
+      "loss": 0.7845,
+      "step": 11170
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.6827203593080806e-05,
+      "loss": 0.8043,
+      "step": 11180
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.6786967974394664e-05,
+      "loss": 0.7844,
+      "step": 11190
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.674675618916994e-05,
+      "loss": 0.7825,
+      "step": 11200
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.6706568354097542e-05,
+      "loss": 0.7776,
+      "step": 11210
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.666640458579886e-05,
+      "loss": 0.7994,
+      "step": 11220
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.6626265000825435e-05,
+      "loss": 0.8007,
+      "step": 11230
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.6586149715658663e-05,
+      "loss": 0.7728,
+      "step": 11240
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.6546058846709376e-05,
+      "loss": 0.779,
+      "step": 11250
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.6505992510317593e-05,
+      "loss": 0.7843,
+      "step": 11260
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.6465950822752126e-05,
+      "loss": 0.7929,
+      "step": 11270
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.6425933900210257e-05,
+      "loss": 0.7664,
+      "step": 11280
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.6385941858817395e-05,
+      "loss": 0.788,
+      "step": 11290
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.634597481462677e-05,
+      "loss": 0.7671,
+      "step": 11300
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.6306032883619045e-05,
+      "loss": 0.7763,
+      "step": 11310
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.6266116181702035e-05,
+      "loss": 0.7918,
+      "step": 11320
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.622622482471032e-05,
+      "loss": 0.7752,
+      "step": 11330
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.6186358928404932e-05,
+      "loss": 0.7744,
+      "step": 11340
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.6146518608473034e-05,
+      "loss": 0.7846,
+      "step": 11350
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.6106703980527557e-05,
+      "loss": 0.7599,
+      "step": 11360
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.6066915160106884e-05,
+      "loss": 0.7817,
+      "step": 11370
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.6027152262674492e-05,
+      "loss": 0.787,
+      "step": 11380
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5987415403618643e-05,
+      "loss": 0.7793,
+      "step": 11390
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5947704698252048e-05,
+      "loss": 0.7803,
+      "step": 11400
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5908020261811503e-05,
+      "loss": 0.7697,
+      "step": 11410
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5868362209457585e-05,
+      "loss": 0.7955,
+      "step": 11420
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.58287306562743e-05,
+      "loss": 0.7656,
+      "step": 11430
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5789125717268766e-05,
+      "loss": 0.7746,
+      "step": 11440
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5749547507370856e-05,
+      "loss": 0.7747,
+      "step": 11450
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5709996141432895e-05,
+      "loss": 0.7813,
+      "step": 11460
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5670471734229286e-05,
+      "loss": 0.7939,
+      "step": 11470
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5630974400456224e-05,
+      "loss": 0.7791,
+      "step": 11480
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.5591504254731324e-05,
+      "loss": 0.7929,
+      "step": 11490
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.555206141159331e-05,
+      "loss": 0.7824,
+      "step": 11500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.551264598550168e-05,
+      "loss": 0.7793,
+      "step": 11510
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.547325809083635e-05,
+      "loss": 0.7801,
+      "step": 11520
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.543389784189737e-05,
+      "loss": 0.78,
+      "step": 11530
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.5394565352904545e-05,
+      "loss": 0.7722,
+      "step": 11540
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.5355260737997153e-05,
+      "loss": 0.781,
+      "step": 11550
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.5315984111233522e-05,
+      "loss": 0.7827,
+      "step": 11560
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.5276735586590822e-05,
+      "loss": 0.7861,
+      "step": 11570
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.5237515277964657e-05,
+      "loss": 0.7809,
+      "step": 11580
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.5198323299168746e-05,
+      "loss": 0.7631,
+      "step": 11590
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.515915976393458e-05,
+      "loss": 0.7656,
+      "step": 11600
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.5120024785911136e-05,
+      "loss": 0.7779,
+      "step": 11610
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.5080918478664503e-05,
+      "loss": 0.7623,
+      "step": 11620
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.5041840955677594e-05,
+      "loss": 0.776,
+      "step": 11630
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.5002792330349774e-05,
+      "loss": 0.7844,
+      "step": 11640
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.4963772715996538e-05,
+      "loss": 0.7758,
+      "step": 11650
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.4924782225849213e-05,
+      "loss": 0.7858,
+      "step": 11660
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4885820973054604e-05,
+      "loss": 0.7877,
+      "step": 11670
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4846889070674691e-05,
+      "loss": 0.7772,
+      "step": 11680
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4807986631686243e-05,
+      "loss": 0.7936,
+      "step": 11690
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4769113768980555e-05,
+      "loss": 0.7631,
+      "step": 11700
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.473027059536309e-05,
+      "loss": 0.7768,
+      "step": 11710
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4691457223553156e-05,
+      "loss": 0.7641,
+      "step": 11720
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4652673766183589e-05,
+      "loss": 0.7969,
+      "step": 11730
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4613920335800385e-05,
+      "loss": 0.7787,
+      "step": 11740
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.457519704486243e-05,
+      "loss": 0.7618,
+      "step": 11750
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.453650400574115e-05,
+      "loss": 0.7659,
+      "step": 11760
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4497841330720185e-05,
+      "loss": 0.7876,
+      "step": 11770
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4459209131995028e-05,
+      "loss": 0.763,
+      "step": 11780
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4420607521672772e-05,
+      "loss": 0.7735,
+      "step": 11790
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.438203661177172e-05,
+      "loss": 0.7835,
+      "step": 11800
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4343496514221108e-05,
+      "loss": 0.7677,
+      "step": 11810
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4304987340860746e-05,
+      "loss": 0.7941,
+      "step": 11820
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4266509203440681e-05,
+      "loss": 0.7674,
+      "step": 11830
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4228062213620926e-05,
+      "loss": 0.7994,
+      "step": 11840
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4189646482971117e-05,
+      "loss": 0.7698,
+      "step": 11850
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.415126212297016e-05,
+      "loss": 0.8061,
+      "step": 11860
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4112909245005906e-05,
+      "loss": 0.7919,
+      "step": 11870
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4074587960374885e-05,
+      "loss": 0.7882,
+      "step": 11880
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4036298380281922e-05,
+      "loss": 0.7952,
+      "step": 11890
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3998040615839854e-05,
+      "loss": 0.7793,
+      "step": 11900
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3959814778069195e-05,
+      "loss": 0.7892,
+      "step": 11910
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3921620977897776e-05,
+      "loss": 0.7942,
+      "step": 11920
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3883459326160492e-05,
+      "loss": 0.7777,
+      "step": 11930
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3845329933598944e-05,
+      "loss": 0.7884,
+      "step": 11940
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3807232910861103e-05,
+      "loss": 0.7879,
+      "step": 11950
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3769168368501022e-05,
+      "loss": 0.7625,
+      "step": 11960
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.373113641697849e-05,
+      "loss": 0.7798,
+      "step": 11970
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3693137166658724e-05,
+      "loss": 0.7587,
+      "step": 11980
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3655170727812039e-05,
+      "loss": 0.786,
+      "step": 11990
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3617237210613553e-05,
+      "loss": 0.786,
+      "step": 12000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3579336725142816e-05,
+      "loss": 0.7721,
+      "step": 12010
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3541469381383545e-05,
+      "loss": 0.7472,
+      "step": 12020
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3503635289223284e-05,
+      "loss": 0.7956,
+      "step": 12030
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3465834558453078e-05,
+      "loss": 0.7697,
+      "step": 12040
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3428067298767158e-05,
+      "loss": 0.7826,
+      "step": 12050
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3390333619762632e-05,
+      "loss": 0.7679,
+      "step": 12060
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3352633630939154e-05,
+      "loss": 0.7625,
+      "step": 12070
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3314967441698617e-05,
+      "loss": 0.7685,
+      "step": 12080
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3277335161344836e-05,
+      "loss": 0.7763,
+      "step": 12090
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.32397368990832e-05,
+      "loss": 0.7882,
+      "step": 12100
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3202172764020404e-05,
+      "loss": 0.7932,
+      "step": 12110
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3164642865164107e-05,
+      "loss": 0.7763,
+      "step": 12120
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.312714731142261e-05,
+      "loss": 0.7835,
+      "step": 12130
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3089686211604557e-05,
+      "loss": 0.774,
+      "step": 12140
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3052259674418605e-05,
+      "loss": 0.784,
+      "step": 12150
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3014867808473105e-05,
+      "loss": 0.7771,
+      "step": 12160
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.297751072227582e-05,
+      "loss": 0.7917,
+      "step": 12170
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2940188524233545e-05,
+      "loss": 0.7893,
+      "step": 12180
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2902901322651872e-05,
+      "loss": 0.7794,
+      "step": 12190
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2865649225734816e-05,
+      "loss": 0.7713,
+      "step": 12200
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2828432341584527e-05,
+      "loss": 0.7808,
+      "step": 12210
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.279125077820098e-05,
+      "loss": 0.778,
+      "step": 12220
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2754104643481612e-05,
+      "loss": 0.7782,
+      "step": 12230
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2716994045221114e-05,
+      "loss": 0.7921,
+      "step": 12240
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2679919091111008e-05,
+      "loss": 0.7843,
+      "step": 12250
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2642879888739398e-05,
+      "loss": 0.7911,
+      "step": 12260
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2605876545590615e-05,
+      "loss": 0.7656,
+      "step": 12270
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2568909169044956e-05,
+      "loss": 0.7763,
+      "step": 12280
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2531977866378345e-05,
+      "loss": 0.7986,
+      "step": 12290
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2495082744762007e-05,
+      "loss": 0.794,
+      "step": 12300
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2458223911262195e-05,
+      "loss": 0.7749,
+      "step": 12310
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2421401472839831e-05,
+      "loss": 0.7846,
+      "step": 12320
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2384615536350237e-05,
+      "loss": 0.7821,
+      "step": 12330
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.23478662085428e-05,
+      "loss": 0.7993,
+      "step": 12340
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2311153596060707e-05,
+      "loss": 0.7731,
+      "step": 12350
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.227447780544055e-05,
+      "loss": 0.7781,
+      "step": 12360
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2237838943112095e-05,
+      "loss": 0.7822,
+      "step": 12370
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2201237115397945e-05,
+      "loss": 0.7645,
+      "step": 12380
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2164672428513227e-05,
+      "loss": 0.7702,
+      "step": 12390
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2128144988565299e-05,
+      "loss": 0.7858,
+      "step": 12400
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2091654901553404e-05,
+      "loss": 0.7564,
+      "step": 12410
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2055202273368415e-05,
+      "loss": 0.7651,
+      "step": 12420
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.20187872097925e-05,
+      "loss": 0.7683,
+      "step": 12430
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.198240981649882e-05,
+      "loss": 0.7954,
+      "step": 12440
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1946070199051205e-05,
+      "loss": 0.7723,
+      "step": 12450
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1909768462903882e-05,
+      "loss": 0.7725,
+      "step": 12460
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1873504713401131e-05,
+      "loss": 0.786,
+      "step": 12470
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.183727905577702e-05,
+      "loss": 0.7729,
+      "step": 12480
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.180109159515507e-05,
+      "loss": 0.7702,
+      "step": 12490
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1764942436547935e-05,
+      "loss": 0.7497,
+      "step": 12500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1728831684857148e-05,
+      "loss": 0.7798,
+      "step": 12510
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.169275944487278e-05,
+      "loss": 0.766,
+      "step": 12520
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1656725821273145e-05,
+      "loss": 0.7757,
+      "step": 12530
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1620730918624498e-05,
+      "loss": 0.7756,
+      "step": 12540
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1584774841380724e-05,
+      "loss": 0.76,
+      "step": 12550
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1548857693883042e-05,
+      "loss": 0.7598,
+      "step": 12560
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1512979580359703e-05,
+      "loss": 0.7958,
+      "step": 12570
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.147714060492569e-05,
+      "loss": 0.7905,
+      "step": 12580
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.144134087158239e-05,
+      "loss": 0.7701,
+      "step": 12590
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1405580484217331e-05,
+      "loss": 0.7804,
+      "step": 12600
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1369859546603866e-05,
+      "loss": 0.7852,
+      "step": 12610
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1334178162400855e-05,
+      "loss": 0.7696,
+      "step": 12620
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1298536435152388e-05,
+      "loss": 0.7777,
+      "step": 12630
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1262934468287464e-05,
+      "loss": 0.7771,
+      "step": 12640
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1227372365119715e-05,
+      "loss": 0.7862,
+      "step": 12650
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1191850228847081e-05,
+      "loss": 0.7819,
+      "step": 12660
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1156368162551534e-05,
+      "loss": 0.7818,
+      "step": 12670
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.112092626919874e-05,
+      "loss": 0.7967,
+      "step": 12680
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1085524651637813e-05,
+      "loss": 0.7981,
+      "step": 12690
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1050163412600984e-05,
+      "loss": 0.7725,
+      "step": 12700
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1014842654703312e-05,
+      "loss": 0.7607,
+      "step": 12710
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.0979562480442356e-05,
+      "loss": 0.7679,
+      "step": 12720
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.0944322992197953e-05,
+      "loss": 0.7761,
+      "step": 12730
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.090912429223184e-05,
+      "loss": 0.7896,
+      "step": 12740
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.0873966482687407e-05,
+      "loss": 0.7592,
+      "step": 12750
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.0838849665589351e-05,
+      "loss": 0.7785,
+      "step": 12760
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.0803773942843454e-05,
+      "loss": 0.7832,
+      "step": 12770
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.0768739416236221e-05,
+      "loss": 0.7611,
+      "step": 12780
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.0733746187434621e-05,
+      "loss": 0.7601,
+      "step": 12790
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.0698794357985787e-05,
+      "loss": 0.7715,
+      "step": 12800
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.0663884029316679e-05,
+      "loss": 0.7782,
+      "step": 12810
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.0629015302733867e-05,
+      "loss": 0.7912,
+      "step": 12820
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.059418827942317e-05,
+      "loss": 0.7578,
+      "step": 12830
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.0559403060449418e-05,
+      "loss": 0.7747,
+      "step": 12840
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.052465974675609e-05,
+      "loss": 0.7816,
+      "step": 12850
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.0489958439165082e-05,
+      "loss": 0.7825,
+      "step": 12860
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0455299238376386e-05,
+      "loss": 0.7565,
+      "step": 12870
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0420682244967811e-05,
+      "loss": 0.7977,
+      "step": 12880
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0386107559394686e-05,
+      "loss": 0.7614,
+      "step": 12890
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0351575281989543e-05,
+      "loss": 0.7827,
+      "step": 12900
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0317085512961871e-05,
+      "loss": 0.7687,
+      "step": 12910
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.02826383523978e-05,
+      "loss": 0.7612,
+      "step": 12920
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0248233900259815e-05,
+      "loss": 0.7652,
+      "step": 12930
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0213872256386464e-05,
+      "loss": 0.7676,
+      "step": 12940
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0179553520492064e-05,
+      "loss": 0.7608,
+      "step": 12950
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0145277792166425e-05,
+      "loss": 0.7744,
+      "step": 12960
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0111045170874545e-05,
+      "loss": 0.78,
+      "step": 12970
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0076855755956347e-05,
+      "loss": 0.7747,
+      "step": 12980
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0042709646626339e-05,
+      "loss": 0.7887,
+      "step": 12990
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0008606941973391e-05,
+      "loss": 0.7876,
+      "step": 13000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.974547740960408e-06,
+      "loss": 0.7678,
+      "step": 13010
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.940532142424047e-06,
+      "loss": 0.7753,
+      "step": 13020
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.90656024507444e-06,
+      "loss": 0.7854,
+      "step": 13030
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.872632147494898e-06,
+      "loss": 0.7628,
+      "step": 13040
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.838747948141633e-06,
+      "loss": 0.7672,
+      "step": 13050
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.804907745343465e-06,
+      "loss": 0.7921,
+      "step": 13060
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.771111637301547e-06,
+      "loss": 0.8012,
+      "step": 13070
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.737359722089057e-06,
+      "loss": 0.7698,
+      "step": 13080
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.703652097650942e-06,
+      "loss": 0.7875,
+      "step": 13090
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.669988861803622e-06,
+      "loss": 0.7727,
+      "step": 13100
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.636370112234706e-06,
+      "loss": 0.7847,
+      "step": 13110
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.602795946502696e-06,
+      "loss": 0.7647,
+      "step": 13120
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.569266462036732e-06,
+      "loss": 0.7662,
+      "step": 13130
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.535781756136284e-06,
+      "loss": 0.7657,
+      "step": 13140
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.502341925970876e-06,
+      "loss": 0.7721,
+      "step": 13150
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.468947068579825e-06,
+      "loss": 0.779,
+      "step": 13160
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.435597280871905e-06,
+      "loss": 0.7731,
+      "step": 13170
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.402292659625137e-06,
+      "loss": 0.7902,
+      "step": 13180
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.369033301486458e-06,
+      "loss": 0.781,
+      "step": 13190
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.335819302971466e-06,
+      "loss": 0.7657,
+      "step": 13200
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.302650760464105e-06,
+      "loss": 0.7565,
+      "step": 13210
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.269527770216424e-06,
+      "loss": 0.7736,
+      "step": 13220
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.236450428348306e-06,
+      "loss": 0.7811,
+      "step": 13230
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.203418830847144e-06,
+      "loss": 0.7695,
+      "step": 13240
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.170433073567573e-06,
+      "loss": 0.7805,
+      "step": 13250
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.137493252231222e-06,
+      "loss": 0.7911,
+      "step": 13260
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.104599462426416e-06,
+      "loss": 0.7849,
+      "step": 13270
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.071751799607902e-06,
+      "loss": 0.7533,
+      "step": 13280
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.038950359096576e-06,
+      "loss": 0.8123,
+      "step": 13290
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.006195236079173e-06,
+      "loss": 0.7676,
+      "step": 13300
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.973486525608052e-06,
+      "loss": 0.7827,
+      "step": 13310
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.94082432260086e-06,
+      "loss": 0.7808,
+      "step": 13320
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.908208721840328e-06,
+      "loss": 0.7665,
+      "step": 13330
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.875639817973891e-06,
+      "loss": 0.77,
+      "step": 13340
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.843117705513515e-06,
+      "loss": 0.7919,
+      "step": 13350
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.810642478835374e-06,
+      "loss": 0.769,
+      "step": 13360
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.778214232179574e-06,
+      "loss": 0.7709,
+      "step": 13370
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.745833059649907e-06,
+      "loss": 0.7589,
+      "step": 13380
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.713499055213537e-06,
+      "loss": 0.7861,
+      "step": 13390
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.681212312700765e-06,
+      "loss": 0.7641,
+      "step": 13400
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.648972925804741e-06,
+      "loss": 0.7449,
+      "step": 13410
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.616780988081194e-06,
+      "loss": 0.7788,
+      "step": 13420
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.584636592948158e-06,
+      "loss": 0.7765,
+      "step": 13430
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.552539833685702e-06,
+      "loss": 0.7767,
+      "step": 13440
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.52049080343566e-06,
+      "loss": 0.788,
+      "step": 13450
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.488489595201363e-06,
+      "loss": 0.7816,
+      "step": 13460
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.456536301847377e-06,
+      "loss": 0.8021,
+      "step": 13470
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.424631016099188e-06,
+      "loss": 0.7684,
+      "step": 13480
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.39277383054301e-06,
+      "loss": 0.7845,
+      "step": 13490
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.360964837625452e-06,
+      "loss": 0.784,
+      "step": 13500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.329204129653276e-06,
+      "loss": 0.764,
+      "step": 13510
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.29749179879313e-06,
+      "loss": 0.7749,
+      "step": 13520
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.265827937071274e-06,
+      "loss": 0.7721,
+      "step": 13530
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.234212636373312e-06,
+      "loss": 0.7658,
+      "step": 13540
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.202645988443932e-06,
+      "loss": 0.7676,
+      "step": 13550
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.171128084886646e-06,
+      "loss": 0.7594,
+      "step": 13560
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.139659017163478e-06,
+      "loss": 0.7805,
+      "step": 13570
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.108238876594779e-06,
+      "loss": 0.766,
+      "step": 13580
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.076867754358891e-06,
+      "loss": 0.793,
+      "step": 13590
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.04554574149192e-06,
+      "loss": 0.7673,
+      "step": 13600
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.014272928887459e-06,
+      "loss": 0.7818,
+      "step": 13610
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.983049407296325e-06,
+      "loss": 0.7978,
+      "step": 13620
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.951875267326298e-06,
+      "loss": 0.7798,
+      "step": 13630
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.92075059944186e-06,
+      "loss": 0.7749,
+      "step": 13640
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.889675493963928e-06,
+      "loss": 0.7777,
+      "step": 13650
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.858650041069581e-06,
+      "loss": 0.7586,
+      "step": 13660
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.82767433079183e-06,
+      "loss": 0.8011,
+      "step": 13670
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.796748453019323e-06,
+      "loss": 0.7738,
+      "step": 13680
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.765872497496121e-06,
+      "loss": 0.7484,
+      "step": 13690
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.735046553821378e-06,
+      "loss": 0.7702,
+      "step": 13700
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.704270711449144e-06,
+      "loss": 0.7618,
+      "step": 13710
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.673545059688086e-06,
+      "loss": 0.7646,
+      "step": 13720
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.642869687701204e-06,
+      "loss": 0.781,
+      "step": 13730
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.612244684505612e-06,
+      "loss": 0.7646,
+      "step": 13740
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.5816701389722215e-06,
+      "loss": 0.7692,
+      "step": 13750
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.551146139825555e-06,
+      "loss": 0.7838,
+      "step": 13760
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.520672775643445e-06,
+      "loss": 0.7692,
+      "step": 13770
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.4902501348567855e-06,
+      "loss": 0.7789,
+      "step": 13780
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.459878305749262e-06,
+      "loss": 0.7834,
+      "step": 13790
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.429557376457133e-06,
+      "loss": 0.7778,
+      "step": 13800
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.399287434968932e-06,
+      "loss": 0.7659,
+      "step": 13810
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.36906856912524e-06,
+      "loss": 0.7675,
+      "step": 13820
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.33890086661842e-06,
+      "loss": 0.775,
+      "step": 13830
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.308784414992359e-06,
+      "loss": 0.7843,
+      "step": 13840
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.278719301642225e-06,
+      "loss": 0.7875,
+      "step": 13850
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.2487056138142025e-06,
+      "loss": 0.7715,
+      "step": 13860
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.218743438605255e-06,
+      "loss": 0.7633,
+      "step": 13870
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.188832862962833e-06,
+      "loss": 0.7882,
+      "step": 13880
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.158973973684677e-06,
+      "loss": 0.7945,
+      "step": 13890
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.129166857418529e-06,
+      "loss": 0.7832,
+      "step": 13900
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.099411600661895e-06,
+      "loss": 0.7596,
+      "step": 13910
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.069708289761781e-06,
+      "loss": 0.7738,
+      "step": 13920
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.040057010914458e-06,
+      "loss": 0.7756,
+      "step": 13930
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.010457850165203e-06,
+      "loss": 0.7853,
+      "step": 13940
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.980910893408049e-06,
+      "loss": 0.77,
+      "step": 13950
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.9514162263855485e-06,
+      "loss": 0.775,
+      "step": 13960
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.921973934688488e-06,
+      "loss": 0.7726,
+      "step": 13970
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.892584103755692e-06,
+      "loss": 0.7682,
+      "step": 13980
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.8632468188737365e-06,
+      "loss": 0.7787,
+      "step": 13990
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.833962165176719e-06,
+      "loss": 0.7633,
+      "step": 14000
+    }
+  ],
+  "max_steps": 18442,
+  "num_train_epochs": 2,
+  "total_flos": 1.4481626466070036e+19,
+  "trial_name": null,
+  "trial_params": null
+}