diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,53938 @@
+{
+  "best_metric": 0.269069105386734,
+  "best_model_checkpoint": "ai-light-dance_singing2_ft_wav2vec2-large-xlsr-53-5gram-v4-2/checkpoint-62783",
+  "epoch": 10.0,
+  "global_step": 89690,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7e-06,
+      "loss": 0.1333,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.4e-06,
+      "loss": 0.2225,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 0.1014,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1400000000000001e-05,
+      "loss": 0.0747,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.44e-05,
+      "loss": 0.1475,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.74e-05,
+      "loss": 0.1396,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.04e-05,
+      "loss": 0.0822,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3400000000000003e-05,
+      "loss": 0.1195,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.64e-05,
+      "loss": 0.1329,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.94e-05,
+      "loss": 0.1644,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9997321129590358e-05,
+      "loss": 0.3298,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9993972541578302e-05,
+      "loss": 0.2061,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9990623953566246e-05,
+      "loss": 0.2549,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998727536555419e-05,
+      "loss": 0.27,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9983926777542134e-05,
+      "loss": 0.2936,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9980578189530082e-05,
+      "loss": 0.3404,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997722960151803e-05,
+      "loss": 0.2546,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9974215872307177e-05,
+      "loss": 0.21,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997086728429512e-05,
+      "loss": 0.2806,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9967518696283068e-05,
+      "loss": 0.2948,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9964170108271016e-05,
+      "loss": 0.1859,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.996082152025896e-05,
+      "loss": 0.2168,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9957472932246904e-05,
+      "loss": 0.1699,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9954124344234848e-05,
+      "loss": 0.2094,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9950775756222792e-05,
+      "loss": 0.2713,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9947427168210736e-05,
+      "loss": 0.3365,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9944078580198683e-05,
+      "loss": 0.2232,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.994072999218663e-05,
+      "loss": 0.3366,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9937381404174575e-05,
+      "loss": 0.179,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.993403281616252e-05,
+      "loss": 0.4673,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9930684228150463e-05,
+      "loss": 0.2592,
+      "step": 310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9927335640138407e-05,
+      "loss": 0.3298,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.992398705212635e-05,
+      "loss": 0.2496,
+      "step": 330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9920638464114302e-05,
+      "loss": 0.2553,
+      "step": 340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9917289876102246e-05,
+      "loss": 0.4419,
+      "step": 350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.991394128809019e-05,
+      "loss": 0.2428,
+      "step": 360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9910592700078134e-05,
+      "loss": 0.2564,
+      "step": 370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.990724411206608e-05,
+      "loss": 0.3407,
+      "step": 380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9903895524054022e-05,
+      "loss": 0.1846,
+      "step": 390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9900546936041967e-05,
+      "loss": 0.3247,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9897198348029917e-05,
+      "loss": 0.3679,
+      "step": 410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.989384976001786e-05,
+      "loss": 0.4175,
+      "step": 420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9890501172005806e-05,
+      "loss": 0.3292,
+      "step": 430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.988715258399375e-05,
+      "loss": 0.2351,
+      "step": 440
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9883803995981694e-05,
+      "loss": 0.4042,
+      "step": 450
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9880455407969638e-05,
+      "loss": 0.4474,
+      "step": 460
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.987710681995759e-05,
+      "loss": 0.2459,
+      "step": 470
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9873758231945533e-05,
+      "loss": 0.2105,
+      "step": 480
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9870409643933477e-05,
+      "loss": 0.4974,
+      "step": 490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9867395914722624e-05,
+      "loss": 0.4537,
+      "step": 500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9864047326710568e-05,
+      "loss": 0.3313,
+      "step": 510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.986069873869852e-05,
+      "loss": 0.252,
+      "step": 520
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9857350150686463e-05,
+      "loss": 0.3456,
+      "step": 530
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9854001562674407e-05,
+      "loss": 0.2086,
+      "step": 540
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.985065297466235e-05,
+      "loss": 0.3816,
+      "step": 550
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9847304386650295e-05,
+      "loss": 0.3016,
+      "step": 560
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.984395579863824e-05,
+      "loss": 0.343,
+      "step": 570
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9840607210626187e-05,
+      "loss": 0.3666,
+      "step": 580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9837258622614134e-05,
+      "loss": 0.2591,
+      "step": 590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.983391003460208e-05,
+      "loss": 0.3751,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9830561446590022e-05,
+      "loss": 0.4188,
+      "step": 610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9827212858577966e-05,
+      "loss": 0.2572,
+      "step": 620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.982386427056591e-05,
+      "loss": 0.2217,
+      "step": 630
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9820515682553855e-05,
+      "loss": 0.3142,
+      "step": 640
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9817167094541802e-05,
+      "loss": 0.2939,
+      "step": 650
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.981381850652975e-05,
+      "loss": 0.2342,
+      "step": 660
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9810469918517694e-05,
+      "loss": 0.4239,
+      "step": 670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9807121330505638e-05,
+      "loss": 0.3248,
+      "step": 680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9803772742493582e-05,
+      "loss": 0.2975,
+      "step": 690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9800424154481526e-05,
+      "loss": 0.3044,
+      "step": 700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.979707556646947e-05,
+      "loss": 0.2869,
+      "step": 710
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9793726978457417e-05,
+      "loss": 0.263,
+      "step": 720
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9790378390445365e-05,
+      "loss": 0.3302,
+      "step": 730
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.978702980243331e-05,
+      "loss": 0.3397,
+      "step": 740
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9783681214421253e-05,
+      "loss": 0.3027,
+      "step": 750
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9780332626409197e-05,
+      "loss": 0.1899,
+      "step": 760
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.977698403839714e-05,
+      "loss": 0.2576,
+      "step": 770
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.977363545038509e-05,
+      "loss": 0.4207,
+      "step": 780
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9770286862373036e-05,
+      "loss": 0.2347,
+      "step": 790
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.976693827436098e-05,
+      "loss": 0.2507,
+      "step": 800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9763589686348924e-05,
+      "loss": 0.49,
+      "step": 810
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9760241098336868e-05,
+      "loss": 0.3139,
+      "step": 820
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9756892510324812e-05,
+      "loss": 0.3383,
+      "step": 830
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9753543922312756e-05,
+      "loss": 0.2786,
+      "step": 840
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9750195334300704e-05,
+      "loss": 0.3292,
+      "step": 850
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.974684674628865e-05,
+      "loss": 0.2632,
+      "step": 860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9743498158276595e-05,
+      "loss": 0.3325,
+      "step": 870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.974014957026454e-05,
+      "loss": 0.2559,
+      "step": 880
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9736800982252484e-05,
+      "loss": 0.2241,
+      "step": 890
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9733452394240428e-05,
+      "loss": 0.2325,
+      "step": 900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9730103806228375e-05,
+      "loss": 0.3107,
+      "step": 910
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.972675521821632e-05,
+      "loss": 0.1823,
+      "step": 920
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9723406630204267e-05,
+      "loss": 0.2831,
+      "step": 930
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.972005804219221e-05,
+      "loss": 0.2745,
+      "step": 940
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9716709454180155e-05,
+      "loss": 0.2689,
+      "step": 950
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.97133608661681e-05,
+      "loss": 0.2481,
+      "step": 960
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9710012278156043e-05,
+      "loss": 0.2514,
+      "step": 970
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.970666369014399e-05,
+      "loss": 0.3045,
+      "step": 980
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9703315102131934e-05,
+      "loss": 0.3063,
+      "step": 990
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9699966514119882e-05,
+      "loss": 0.4621,
+      "step": 1000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9696617926107826e-05,
+      "loss": 0.1746,
+      "step": 1010
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.969326933809577e-05,
+      "loss": 0.1882,
+      "step": 1020
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9689920750083714e-05,
+      "loss": 0.2933,
+      "step": 1030
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.968657216207166e-05,
+      "loss": 0.2065,
+      "step": 1040
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9683223574059606e-05,
+      "loss": 0.2423,
+      "step": 1050
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.967987498604755e-05,
+      "loss": 0.2428,
+      "step": 1060
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9676526398035497e-05,
+      "loss": 0.3423,
+      "step": 1070
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.967317781002344e-05,
+      "loss": 0.2873,
+      "step": 1080
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9669829222011385e-05,
+      "loss": 0.2201,
+      "step": 1090
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.966648063399933e-05,
+      "loss": 0.34,
+      "step": 1100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9663132045987277e-05,
+      "loss": 0.2715,
+      "step": 1110
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.965978345797522e-05,
+      "loss": 0.2599,
+      "step": 1120
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9656434869963165e-05,
+      "loss": 0.3173,
+      "step": 1130
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9653086281951113e-05,
+      "loss": 0.2941,
+      "step": 1140
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9649737693939057e-05,
+      "loss": 0.4499,
+      "step": 1150
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9646389105927e-05,
+      "loss": 0.1807,
+      "step": 1160
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9643040517914948e-05,
+      "loss": 0.3558,
+      "step": 1170
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9639691929902892e-05,
+      "loss": 0.2932,
+      "step": 1180
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9636343341890836e-05,
+      "loss": 0.2372,
+      "step": 1190
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.963299475387878e-05,
+      "loss": 0.3785,
+      "step": 1200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9629646165866728e-05,
+      "loss": 0.3696,
+      "step": 1210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9626297577854672e-05,
+      "loss": 0.3875,
+      "step": 1220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9622948989842616e-05,
+      "loss": 0.2338,
+      "step": 1230
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9619600401830563e-05,
+      "loss": 0.2742,
+      "step": 1240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9616251813818508e-05,
+      "loss": 0.3079,
+      "step": 1250
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.961290322580645e-05,
+      "loss": 0.2413,
+      "step": 1260
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9609554637794396e-05,
+      "loss": 0.2657,
+      "step": 1270
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9606206049782343e-05,
+      "loss": 0.4674,
+      "step": 1280
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9602857461770287e-05,
+      "loss": 0.3178,
+      "step": 1290
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9599508873758235e-05,
+      "loss": 0.2895,
+      "step": 1300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.959616028574618e-05,
+      "loss": 0.3187,
+      "step": 1310
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9592811697734123e-05,
+      "loss": 0.2595,
+      "step": 1320
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9589463109722067e-05,
+      "loss": 0.2471,
+      "step": 1330
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.958611452171001e-05,
+      "loss": 0.354,
+      "step": 1340
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.958276593369796e-05,
+      "loss": 0.4089,
+      "step": 1350
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9579417345685903e-05,
+      "loss": 0.2774,
+      "step": 1360
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.957606875767385e-05,
+      "loss": 0.3519,
+      "step": 1370
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9572720169661794e-05,
+      "loss": 0.3929,
+      "step": 1380
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9569371581649738e-05,
+      "loss": 0.3293,
+      "step": 1390
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9566022993637682e-05,
+      "loss": 0.4493,
+      "step": 1400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9562674405625626e-05,
+      "loss": 0.1704,
+      "step": 1410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9559325817613574e-05,
+      "loss": 0.2484,
+      "step": 1420
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.955597722960152e-05,
+      "loss": 0.2995,
+      "step": 1430
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9552628641589465e-05,
+      "loss": 0.3295,
+      "step": 1440
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.954928005357741e-05,
+      "loss": 0.5011,
+      "step": 1450
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9545931465565353e-05,
+      "loss": 0.218,
+      "step": 1460
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9542582877553297e-05,
+      "loss": 0.308,
+      "step": 1470
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.953923428954124e-05,
+      "loss": 0.2326,
+      "step": 1480
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.953588570152919e-05,
+      "loss": 0.3292,
+      "step": 1490
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9532537113517136e-05,
+      "loss": 0.1913,
+      "step": 1500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.952918852550508e-05,
+      "loss": 0.2599,
+      "step": 1510
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9525839937493025e-05,
+      "loss": 0.3461,
+      "step": 1520
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.952249134948097e-05,
+      "loss": 0.2685,
+      "step": 1530
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9519142761468913e-05,
+      "loss": 0.3459,
+      "step": 1540
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9515794173456857e-05,
+      "loss": 0.2864,
+      "step": 1550
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9512445585444808e-05,
+      "loss": 0.3786,
+      "step": 1560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9509096997432752e-05,
+      "loss": 0.3324,
+      "step": 1570
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9505748409420696e-05,
+      "loss": 0.2144,
+      "step": 1580
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.950239982140864e-05,
+      "loss": 0.3589,
+      "step": 1590
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9499051233396584e-05,
+      "loss": 0.3423,
+      "step": 1600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9495702645384528e-05,
+      "loss": 0.3592,
+      "step": 1610
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9492354057372472e-05,
+      "loss": 0.3225,
+      "step": 1620
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9489005469360423e-05,
+      "loss": 0.271,
+      "step": 1630
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.948599174014957e-05,
+      "loss": 0.6352,
+      "step": 1640
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9482643152137514e-05,
+      "loss": 0.369,
+      "step": 1650
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.947929456412546e-05,
+      "loss": 0.3149,
+      "step": 1660
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9475945976113406e-05,
+      "loss": 0.2427,
+      "step": 1670
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9472597388101353e-05,
+      "loss": 0.2249,
+      "step": 1680
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9469248800089297e-05,
+      "loss": 0.367,
+      "step": 1690
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.946590021207724e-05,
+      "loss": 0.3768,
+      "step": 1700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9462551624065186e-05,
+      "loss": 0.2897,
+      "step": 1710
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.945920303605313e-05,
+      "loss": 0.2873,
+      "step": 1720
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9455854448041074e-05,
+      "loss": 0.241,
+      "step": 1730
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9452505860029025e-05,
+      "loss": 0.2226,
+      "step": 1740
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.944915727201697e-05,
+      "loss": 0.4106,
+      "step": 1750
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9445808684004913e-05,
+      "loss": 0.2407,
+      "step": 1760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9442460095992857e-05,
+      "loss": 0.2977,
+      "step": 1770
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.94391115079808e-05,
+      "loss": 0.4242,
+      "step": 1780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9435762919968745e-05,
+      "loss": 0.3121,
+      "step": 1790
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9432414331956692e-05,
+      "loss": 0.3337,
+      "step": 1800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.942906574394464e-05,
+      "loss": 0.2749,
+      "step": 1810
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9425717155932584e-05,
+      "loss": 0.4066,
+      "step": 1820
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9422368567920528e-05,
+      "loss": 0.3787,
+      "step": 1830
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9419019979908472e-05,
+      "loss": 0.2425,
+      "step": 1840
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9416006250697626e-05,
+      "loss": 0.5961,
+      "step": 1850
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.941265766268557e-05,
+      "loss": 0.226,
+      "step": 1860
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9409309074673514e-05,
+      "loss": 0.2778,
+      "step": 1870
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.940596048666146e-05,
+      "loss": 0.2189,
+      "step": 1880
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9402611898649402e-05,
+      "loss": 0.4383,
+      "step": 1890
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9399263310637347e-05,
+      "loss": 0.3946,
+      "step": 1900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.939591472262529e-05,
+      "loss": 0.3077,
+      "step": 1910
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.939256613461324e-05,
+      "loss": 0.2786,
+      "step": 1920
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9389217546601186e-05,
+      "loss": 0.2595,
+      "step": 1930
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.938586895858913e-05,
+      "loss": 0.1824,
+      "step": 1940
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9382520370577074e-05,
+      "loss": 0.3413,
+      "step": 1950
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9379171782565018e-05,
+      "loss": 0.3299,
+      "step": 1960
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9375823194552962e-05,
+      "loss": 0.2578,
+      "step": 1970
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.937247460654091e-05,
+      "loss": 0.3664,
+      "step": 1980
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9369126018528857e-05,
+      "loss": 0.3568,
+      "step": 1990
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.93657774305168e-05,
+      "loss": 0.3332,
+      "step": 2000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9362428842504745e-05,
+      "loss": 0.3657,
+      "step": 2010
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.935908025449269e-05,
+      "loss": 0.3801,
+      "step": 2020
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9355731666480633e-05,
+      "loss": 0.247,
+      "step": 2030
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9352383078468577e-05,
+      "loss": 0.2187,
+      "step": 2040
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9349034490456525e-05,
+      "loss": 0.3166,
+      "step": 2050
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9345685902444472e-05,
+      "loss": 0.2663,
+      "step": 2060
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9342337314432416e-05,
+      "loss": 0.2978,
+      "step": 2070
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.933898872642036e-05,
+      "loss": 0.3683,
+      "step": 2080
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9335640138408304e-05,
+      "loss": 0.4044,
+      "step": 2090
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.933229155039625e-05,
+      "loss": 0.3323,
+      "step": 2100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9328942962384196e-05,
+      "loss": 0.1943,
+      "step": 2110
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.932559437437214e-05,
+      "loss": 0.1952,
+      "step": 2120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9322245786360087e-05,
+      "loss": 0.3089,
+      "step": 2130
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.931889719834803e-05,
+      "loss": 0.3246,
+      "step": 2140
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9315548610335976e-05,
+      "loss": 0.3987,
+      "step": 2150
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.931220002232392e-05,
+      "loss": 0.2745,
+      "step": 2160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9308851434311864e-05,
+      "loss": 0.3004,
+      "step": 2170
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.930550284629981e-05,
+      "loss": 0.2344,
+      "step": 2180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.930215425828776e-05,
+      "loss": 0.2571,
+      "step": 2190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9298805670275703e-05,
+      "loss": 0.3107,
+      "step": 2200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9295457082263647e-05,
+      "loss": 0.2853,
+      "step": 2210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.929210849425159e-05,
+      "loss": 0.3413,
+      "step": 2220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9288759906239535e-05,
+      "loss": 0.331,
+      "step": 2230
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9285411318227482e-05,
+      "loss": 0.3168,
+      "step": 2240
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9282062730215426e-05,
+      "loss": 0.2942,
+      "step": 2250
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9278714142203374e-05,
+      "loss": 0.2923,
+      "step": 2260
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9275365554191318e-05,
+      "loss": 0.193,
+      "step": 2270
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9272016966179262e-05,
+      "loss": 0.2523,
+      "step": 2280
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9268668378167206e-05,
+      "loss": 0.2988,
+      "step": 2290
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.926531979015515e-05,
+      "loss": 0.6501,
+      "step": 2300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9261971202143098e-05,
+      "loss": 0.2306,
+      "step": 2310
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9258622614131042e-05,
+      "loss": 0.2573,
+      "step": 2320
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.925527402611899e-05,
+      "loss": 0.2034,
+      "step": 2330
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9251925438106933e-05,
+      "loss": 0.3989,
+      "step": 2340
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9248576850094877e-05,
+      "loss": 0.2532,
+      "step": 2350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.924522826208282e-05,
+      "loss": 0.3798,
+      "step": 2360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.924187967407077e-05,
+      "loss": 0.3105,
+      "step": 2370
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9238531086058713e-05,
+      "loss": 0.382,
+      "step": 2380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9235182498046657e-05,
+      "loss": 0.2086,
+      "step": 2390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9231833910034604e-05,
+      "loss": 0.3751,
+      "step": 2400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.922848532202255e-05,
+      "loss": 0.2725,
+      "step": 2410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9225136734010493e-05,
+      "loss": 0.274,
+      "step": 2420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9221788145998437e-05,
+      "loss": 0.3629,
+      "step": 2430
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9218439557986384e-05,
+      "loss": 0.2539,
+      "step": 2440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9215090969974328e-05,
+      "loss": 0.3229,
+      "step": 2450
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9211742381962272e-05,
+      "loss": 0.2418,
+      "step": 2460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.920839379395022e-05,
+      "loss": 0.2635,
+      "step": 2470
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9205045205938164e-05,
+      "loss": 0.2654,
+      "step": 2480
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9201696617926108e-05,
+      "loss": 0.2423,
+      "step": 2490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9198348029914055e-05,
+      "loss": 0.3214,
+      "step": 2500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9194999441902e-05,
+      "loss": 0.3244,
+      "step": 2510
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9191650853889944e-05,
+      "loss": 0.3542,
+      "step": 2520
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9188302265877888e-05,
+      "loss": 0.3033,
+      "step": 2530
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9184953677865835e-05,
+      "loss": 0.2417,
+      "step": 2540
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.918160508985378e-05,
+      "loss": 0.322,
+      "step": 2550
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9178256501841723e-05,
+      "loss": 0.1738,
+      "step": 2560
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.917490791382967e-05,
+      "loss": 0.3273,
+      "step": 2570
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9171559325817615e-05,
+      "loss": 0.2372,
+      "step": 2580
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.916821073780556e-05,
+      "loss": 0.3597,
+      "step": 2590
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9164862149793503e-05,
+      "loss": 0.2917,
+      "step": 2600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.916151356178145e-05,
+      "loss": 0.2962,
+      "step": 2610
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9158164973769394e-05,
+      "loss": 0.2528,
+      "step": 2620
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9154816385757342e-05,
+      "loss": 0.4564,
+      "step": 2630
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9151467797745286e-05,
+      "loss": 0.4208,
+      "step": 2640
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.914811920973323e-05,
+      "loss": 0.3509,
+      "step": 2650
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9144770621721174e-05,
+      "loss": 0.3555,
+      "step": 2660
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9141422033709118e-05,
+      "loss": 0.1908,
+      "step": 2670
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9138073445697066e-05,
+      "loss": 0.3751,
+      "step": 2680
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.913472485768501e-05,
+      "loss": 0.2079,
+      "step": 2690
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9131376269672957e-05,
+      "loss": 0.4672,
+      "step": 2700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.91280276816609e-05,
+      "loss": 0.368,
+      "step": 2710
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9124679093648845e-05,
+      "loss": 0.3923,
+      "step": 2720
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.912133050563679e-05,
+      "loss": 0.2806,
+      "step": 2730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9117981917624733e-05,
+      "loss": 0.3179,
+      "step": 2740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.911463332961268e-05,
+      "loss": 0.3784,
+      "step": 2750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9111284741600625e-05,
+      "loss": 0.2582,
+      "step": 2760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9107936153588572e-05,
+      "loss": 0.2694,
+      "step": 2770
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9104587565576517e-05,
+      "loss": 0.226,
+      "step": 2780
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.910123897756446e-05,
+      "loss": 0.3128,
+      "step": 2790
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9097890389552405e-05,
+      "loss": 0.3383,
+      "step": 2800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.909454180154035e-05,
+      "loss": 0.2754,
+      "step": 2810
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9091193213528296e-05,
+      "loss": 0.2628,
+      "step": 2820
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9087844625516244e-05,
+      "loss": 0.2788,
+      "step": 2830
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9084496037504188e-05,
+      "loss": 0.238,
+      "step": 2840
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9081147449492132e-05,
+      "loss": 0.4032,
+      "step": 2850
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9077798861480076e-05,
+      "loss": 0.2436,
+      "step": 2860
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.907445027346802e-05,
+      "loss": 0.1706,
+      "step": 2870
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9071101685455964e-05,
+      "loss": 0.3794,
+      "step": 2880
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.906775309744391e-05,
+      "loss": 0.4788,
+      "step": 2890
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.906440450943186e-05,
+      "loss": 0.3913,
+      "step": 2900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9061055921419803e-05,
+      "loss": 0.3332,
+      "step": 2910
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9057707333407747e-05,
+      "loss": 0.2701,
+      "step": 2920
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.905435874539569e-05,
+      "loss": 0.2956,
+      "step": 2930
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9051010157383635e-05,
+      "loss": 0.3244,
+      "step": 2940
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.904766156937158e-05,
+      "loss": 0.2967,
+      "step": 2950
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.904431298135953e-05,
+      "loss": 0.3729,
+      "step": 2960
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9040964393347474e-05,
+      "loss": 0.2844,
+      "step": 2970
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.903761580533542e-05,
+      "loss": 0.2278,
+      "step": 2980
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9034267217323362e-05,
+      "loss": 0.2514,
+      "step": 2990
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9030918629311307e-05,
+      "loss": 0.3162,
+      "step": 3000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.902757004129925e-05,
+      "loss": 0.3212,
+      "step": 3010
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9024221453287195e-05,
+      "loss": 0.3072,
+      "step": 3020
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9020872865275145e-05,
+      "loss": 0.3281,
+      "step": 3030
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.901752427726309e-05,
+      "loss": 0.309,
+      "step": 3040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9014175689251034e-05,
+      "loss": 0.309,
+      "step": 3050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9010827101238978e-05,
+      "loss": 0.287,
+      "step": 3060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9007478513226922e-05,
+      "loss": 0.1877,
+      "step": 3070
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9004129925214866e-05,
+      "loss": 0.4466,
+      "step": 3080
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9000781337202813e-05,
+      "loss": 0.4413,
+      "step": 3090
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.899743274919076e-05,
+      "loss": 0.4515,
+      "step": 3100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8994084161178705e-05,
+      "loss": 0.2391,
+      "step": 3110
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.899073557316665e-05,
+      "loss": 0.2804,
+      "step": 3120
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8987386985154593e-05,
+      "loss": 0.3225,
+      "step": 3130
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8984038397142537e-05,
+      "loss": 0.2896,
+      "step": 3140
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.898068980913048e-05,
+      "loss": 0.3319,
+      "step": 3150
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8977341221118432e-05,
+      "loss": 0.2722,
+      "step": 3160
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8973992633106376e-05,
+      "loss": 0.2571,
+      "step": 3170
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.897064404509432e-05,
+      "loss": 0.4143,
+      "step": 3180
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8967295457082264e-05,
+      "loss": 0.3223,
+      "step": 3190
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.896394686907021e-05,
+      "loss": 0.5021,
+      "step": 3200
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8960598281058152e-05,
+      "loss": 0.6432,
+      "step": 3210
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.89572496930461e-05,
+      "loss": 0.3037,
+      "step": 3220
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8953901105034047e-05,
+      "loss": 0.2806,
+      "step": 3230
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.895055251702199e-05,
+      "loss": 0.2894,
+      "step": 3240
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8947203929009935e-05,
+      "loss": 0.2184,
+      "step": 3250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.894385534099788e-05,
+      "loss": 0.2069,
+      "step": 3260
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8940506752985824e-05,
+      "loss": 0.307,
+      "step": 3270
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8937158164973768e-05,
+      "loss": 0.2571,
+      "step": 3280
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8933809576961715e-05,
+      "loss": 0.1343,
+      "step": 3290
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8930460988949663e-05,
+      "loss": 0.3783,
+      "step": 3300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8927112400937607e-05,
+      "loss": 0.4052,
+      "step": 3310
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.892376381292555e-05,
+      "loss": 0.2773,
+      "step": 3320
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8920415224913495e-05,
+      "loss": 0.2517,
+      "step": 3330
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.891706663690144e-05,
+      "loss": 0.278,
+      "step": 3340
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8913718048889386e-05,
+      "loss": 0.1903,
+      "step": 3350
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.891036946087733e-05,
+      "loss": 0.1556,
+      "step": 3360
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8907020872865278e-05,
+      "loss": 0.3829,
+      "step": 3370
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8903672284853222e-05,
+      "loss": 0.2567,
+      "step": 3380
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8900323696841166e-05,
+      "loss": 0.4417,
+      "step": 3390
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.889697510882911e-05,
+      "loss": 0.3469,
+      "step": 3400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8893626520817054e-05,
+      "loss": 0.1888,
+      "step": 3410
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8890277932805e-05,
+      "loss": 0.3623,
+      "step": 3420
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8886929344792946e-05,
+      "loss": 0.3189,
+      "step": 3430
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8883580756780893e-05,
+      "loss": 0.2527,
+      "step": 3440
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8880232168768837e-05,
+      "loss": 0.3846,
+      "step": 3450
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.887688358075678e-05,
+      "loss": 0.2763,
+      "step": 3460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8873534992744725e-05,
+      "loss": 0.4867,
+      "step": 3470
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8870186404732673e-05,
+      "loss": 0.3004,
+      "step": 3480
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8866837816720617e-05,
+      "loss": 0.2836,
+      "step": 3490
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.886348922870856e-05,
+      "loss": 0.3223,
+      "step": 3500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.886014064069651e-05,
+      "loss": 0.2879,
+      "step": 3510
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8856792052684453e-05,
+      "loss": 0.4123,
+      "step": 3520
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8853443464672397e-05,
+      "loss": 0.3377,
+      "step": 3530
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.885009487666034e-05,
+      "loss": 0.2579,
+      "step": 3540
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.8846746288648288e-05,
+      "loss": 0.3297,
+      "step": 3550
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.8843397700636232e-05,
+      "loss": 0.6535,
+      "step": 3560
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.8840049112624176e-05,
+      "loss": 0.2304,
+      "step": 3570
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.8836700524612124e-05,
+      "loss": 0.2039,
+      "step": 3580
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.8833351936600068e-05,
+      "loss": 0.3441,
+      "step": 3590
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.8830003348588012e-05,
+      "loss": 0.2525,
+      "step": 3600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.882665476057596e-05,
+      "loss": 0.3579,
+      "step": 3610
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.8823306172563903e-05,
+      "loss": 0.372,
+      "step": 3620
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.8819957584551848e-05,
+      "loss": 0.3738,
+      "step": 3630
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.881660899653979e-05,
+      "loss": 0.332,
+      "step": 3640
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.881326040852774e-05,
+      "loss": 0.4288,
+      "step": 3650
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.8809911820515683e-05,
+      "loss": 0.307,
+      "step": 3660
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.8806563232503627e-05,
+      "loss": 0.29,
+      "step": 3670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.8803214644491575e-05,
+      "loss": 0.3302,
+      "step": 3680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.879986605647952e-05,
+      "loss": 0.2813,
+      "step": 3690
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.8796517468467463e-05,
+      "loss": 0.2062,
+      "step": 3700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.8793168880455407e-05,
+      "loss": 0.4751,
+      "step": 3710
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.8789820292443354e-05,
+      "loss": 0.2128,
+      "step": 3720
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.87864717044313e-05,
+      "loss": 0.2823,
+      "step": 3730
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8783123116419246e-05,
+      "loss": 0.2905,
+      "step": 3740
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.877977452840719e-05,
+      "loss": 0.3583,
+      "step": 3750
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8776425940395134e-05,
+      "loss": 0.3407,
+      "step": 3760
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8773077352383078e-05,
+      "loss": 0.2563,
+      "step": 3770
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8769728764371022e-05,
+      "loss": 0.2782,
+      "step": 3780
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.876638017635897e-05,
+      "loss": 0.2175,
+      "step": 3790
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8763031588346914e-05,
+      "loss": 0.3626,
+      "step": 3800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.875968300033486e-05,
+      "loss": 0.3096,
+      "step": 3810
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8756334412322805e-05,
+      "loss": 0.2807,
+      "step": 3820
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.875298582431075e-05,
+      "loss": 0.3498,
+      "step": 3830
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8749637236298693e-05,
+      "loss": 0.2311,
+      "step": 3840
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8746288648286638e-05,
+      "loss": 0.258,
+      "step": 3850
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8742940060274585e-05,
+      "loss": 0.4087,
+      "step": 3860
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8739591472262532e-05,
+      "loss": 0.2606,
+      "step": 3870
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8736242884250476e-05,
+      "loss": 0.2822,
+      "step": 3880
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.873289429623842e-05,
+      "loss": 0.2608,
+      "step": 3890
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8729545708226365e-05,
+      "loss": 0.4702,
+      "step": 3900
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.872619712021431e-05,
+      "loss": 0.2394,
+      "step": 3910
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8722848532202253e-05,
+      "loss": 0.3797,
+      "step": 3920
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.87194999441902e-05,
+      "loss": 0.305,
+      "step": 3930
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8716151356178148e-05,
+      "loss": 0.2368,
+      "step": 3940
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8712802768166092e-05,
+      "loss": 0.2145,
+      "step": 3950
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8709454180154036e-05,
+      "loss": 0.2994,
+      "step": 3960
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.870610559214198e-05,
+      "loss": 0.3664,
+      "step": 3970
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8702757004129924e-05,
+      "loss": 0.5603,
+      "step": 3980
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8699408416117868e-05,
+      "loss": 0.2815,
+      "step": 3990
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.869605982810582e-05,
+      "loss": 0.2993,
+      "step": 4000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8692711240093763e-05,
+      "loss": 0.2324,
+      "step": 4010
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8689362652081707e-05,
+      "loss": 0.2601,
+      "step": 4020
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.868601406406965e-05,
+      "loss": 0.2926,
+      "step": 4030
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8682665476057595e-05,
+      "loss": 0.355,
+      "step": 4040
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.867931688804554e-05,
+      "loss": 0.3526,
+      "step": 4050
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8675968300033487e-05,
+      "loss": 0.3313,
+      "step": 4060
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8672619712021434e-05,
+      "loss": 0.3634,
+      "step": 4070
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.866927112400938e-05,
+      "loss": 0.3835,
+      "step": 4080
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8665922535997322e-05,
+      "loss": 0.2421,
+      "step": 4090
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8662573947985266e-05,
+      "loss": 0.3454,
+      "step": 4100
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.865922535997321e-05,
+      "loss": 0.446,
+      "step": 4110
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8655876771961155e-05,
+      "loss": 0.2652,
+      "step": 4120
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8652528183949105e-05,
+      "loss": 0.4297,
+      "step": 4130
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.864917959593705e-05,
+      "loss": 0.2877,
+      "step": 4140
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8645831007924994e-05,
+      "loss": 0.4578,
+      "step": 4150
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8642482419912938e-05,
+      "loss": 0.3278,
+      "step": 4160
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8639133831900882e-05,
+      "loss": 0.3118,
+      "step": 4170
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8635785243888826e-05,
+      "loss": 0.5179,
+      "step": 4180
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.863243665587677e-05,
+      "loss": 0.2571,
+      "step": 4190
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.862908806786472e-05,
+      "loss": 0.3882,
+      "step": 4200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8625739479852665e-05,
+      "loss": 0.3314,
+      "step": 4210
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.862239089184061e-05,
+      "loss": 0.318,
+      "step": 4220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8619042303828553e-05,
+      "loss": 0.2294,
+      "step": 4230
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8615693715816497e-05,
+      "loss": 0.2306,
+      "step": 4240
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.861234512780444e-05,
+      "loss": 0.3997,
+      "step": 4250
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.860899653979239e-05,
+      "loss": 0.3924,
+      "step": 4260
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8605647951780336e-05,
+      "loss": 0.3444,
+      "step": 4270
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.860229936376828e-05,
+      "loss": 0.3532,
+      "step": 4280
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8598950775756224e-05,
+      "loss": 0.2894,
+      "step": 4290
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8595602187744168e-05,
+      "loss": 0.2314,
+      "step": 4300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8592253599732112e-05,
+      "loss": 0.2855,
+      "step": 4310
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8588905011720056e-05,
+      "loss": 0.3593,
+      "step": 4320
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8585556423708004e-05,
+      "loss": 0.3318,
+      "step": 4330
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.858220783569595e-05,
+      "loss": 0.3671,
+      "step": 4340
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.8578859247683895e-05,
+      "loss": 0.3766,
+      "step": 4350
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.857551065967184e-05,
+      "loss": 0.3343,
+      "step": 4360
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.8572162071659784e-05,
+      "loss": 0.3501,
+      "step": 4370
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.8568813483647728e-05,
+      "loss": 0.357,
+      "step": 4380
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.8565464895635675e-05,
+      "loss": 0.2835,
+      "step": 4390
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.856211630762362e-05,
+      "loss": 0.3577,
+      "step": 4400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.8558767719611567e-05,
+      "loss": 0.2895,
+      "step": 4410
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.855541913159951e-05,
+      "loss": 0.3987,
+      "step": 4420
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.8552070543587455e-05,
+      "loss": 0.1939,
+      "step": 4430
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.85487219555754e-05,
+      "loss": 0.2978,
+      "step": 4440
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.8545373367563343e-05,
+      "loss": 0.2691,
+      "step": 4450
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.854202477955129e-05,
+      "loss": 0.1818,
+      "step": 4460
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.8538676191539234e-05,
+      "loss": 0.3287,
+      "step": 4470
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.8535327603527182e-05,
+      "loss": 0.2582,
+      "step": 4480
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.8531979015515126e-05,
+      "loss": 0.2541,
+      "step": 4490
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.852863042750307e-05,
+      "loss": 0.3397,
+      "step": 4500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.8525281839491014e-05,
+      "loss": 0.5791,
+      "step": 4510
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.8521933251478958e-05,
+      "loss": 0.2502,
+      "step": 4520
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8518584663466906e-05,
+      "loss": 0.3114,
+      "step": 4530
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.851523607545485e-05,
+      "loss": 0.2173,
+      "step": 4540
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8511887487442797e-05,
+      "loss": 0.2588,
+      "step": 4550
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.850853889943074e-05,
+      "loss": 0.3148,
+      "step": 4560
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8505190311418685e-05,
+      "loss": 0.342,
+      "step": 4570
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.850184172340663e-05,
+      "loss": 0.273,
+      "step": 4580
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8498493135394577e-05,
+      "loss": 0.2818,
+      "step": 4590
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.849514454738252e-05,
+      "loss": 0.3846,
+      "step": 4600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.8491795959370465e-05,
+      "loss": 0.3392,
+      "step": 4610
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.8488447371358413e-05,
+      "loss": 0.3168,
+      "step": 4620
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.8485098783346357e-05,
+      "loss": 0.1967,
+      "step": 4630
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.84817501953343e-05,
+      "loss": 0.3814,
+      "step": 4640
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.8478401607322245e-05,
+      "loss": 0.4463,
+      "step": 4650
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.8475053019310192e-05,
+      "loss": 0.2366,
+      "step": 4660
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.8471704431298136e-05,
+      "loss": 0.3218,
+      "step": 4670
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.846835584328608e-05,
+      "loss": 0.3682,
+      "step": 4680
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.8465007255274028e-05,
+      "loss": 0.299,
+      "step": 4690
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.8461658667261972e-05,
+      "loss": 0.4885,
+      "step": 4700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.8458310079249916e-05,
+      "loss": 0.3029,
+      "step": 4710
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.8454961491237863e-05,
+      "loss": 0.2412,
+      "step": 4720
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.8451612903225808e-05,
+      "loss": 0.3781,
+      "step": 4730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.844826431521375e-05,
+      "loss": 0.2935,
+      "step": 4740
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.8444915727201696e-05,
+      "loss": 0.5355,
+      "step": 4750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.8441567139189643e-05,
+      "loss": 0.3359,
+      "step": 4760
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.8438218551177587e-05,
+      "loss": 0.1742,
+      "step": 4770
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.843486996316553e-05,
+      "loss": 0.2968,
+      "step": 4780
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.843152137515348e-05,
+      "loss": 0.4391,
+      "step": 4790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.8428172787141423e-05,
+      "loss": 0.353,
+      "step": 4800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.8424824199129367e-05,
+      "loss": 0.358,
+      "step": 4810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.842147561111731e-05,
+      "loss": 0.4561,
+      "step": 4820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.841812702310526e-05,
+      "loss": 0.328,
+      "step": 4830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.8414778435093202e-05,
+      "loss": 0.2599,
+      "step": 4840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.841142984708115e-05,
+      "loss": 0.2283,
+      "step": 4850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.8408081259069094e-05,
+      "loss": 0.4592,
+      "step": 4860
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.8404732671057038e-05,
+      "loss": 0.2888,
+      "step": 4870
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.8401384083044982e-05,
+      "loss": 0.293,
+      "step": 4880
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.8398035495032926e-05,
+      "loss": 0.2979,
+      "step": 4890
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.8394686907020874e-05,
+      "loss": 0.4657,
+      "step": 4900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.8391338319008818e-05,
+      "loss": 0.3231,
+      "step": 4910
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.8387989730996765e-05,
+      "loss": 0.3339,
+      "step": 4920
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.838464114298471e-05,
+      "loss": 0.421,
+      "step": 4930
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.8381292554972653e-05,
+      "loss": 0.3323,
+      "step": 4940
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.8377943966960597e-05,
+      "loss": 0.3351,
+      "step": 4950
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.8374595378948545e-05,
+      "loss": 0.2613,
+      "step": 4960
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.837124679093649e-05,
+      "loss": 0.2568,
+      "step": 4970
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.8367898202924436e-05,
+      "loss": 0.3649,
+      "step": 4980
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.836454961491238e-05,
+      "loss": 0.2827,
+      "step": 4990
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.8361201026900325e-05,
+      "loss": 0.6217,
+      "step": 5000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.835785243888827e-05,
+      "loss": 0.3068,
+      "step": 5010
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.8354503850876213e-05,
+      "loss": 0.2597,
+      "step": 5020
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.835115526286416e-05,
+      "loss": 0.239,
+      "step": 5030
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.8347806674852104e-05,
+      "loss": 0.3,
+      "step": 5040
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.8344458086840052e-05,
+      "loss": 0.273,
+      "step": 5050
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.8341109498827996e-05,
+      "loss": 0.3774,
+      "step": 5060
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.833776091081594e-05,
+      "loss": 0.2405,
+      "step": 5070
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.8334412322803884e-05,
+      "loss": 0.3284,
+      "step": 5080
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.8331063734791828e-05,
+      "loss": 0.2678,
+      "step": 5090
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.8327715146779776e-05,
+      "loss": 0.4018,
+      "step": 5100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.8324366558767723e-05,
+      "loss": 0.2179,
+      "step": 5110
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.8321017970755667e-05,
+      "loss": 0.2904,
+      "step": 5120
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.831766938274361e-05,
+      "loss": 0.2636,
+      "step": 5130
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.8314320794731555e-05,
+      "loss": 0.3581,
+      "step": 5140
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.83109722067195e-05,
+      "loss": 0.394,
+      "step": 5150
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.8307623618707443e-05,
+      "loss": 0.2686,
+      "step": 5160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.830427503069539e-05,
+      "loss": 0.3395,
+      "step": 5170
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.8300926442683338e-05,
+      "loss": 0.4244,
+      "step": 5180
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.8297577854671282e-05,
+      "loss": 0.2993,
+      "step": 5190
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.8294229266659226e-05,
+      "loss": 0.563,
+      "step": 5200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.829088067864717e-05,
+      "loss": 0.2876,
+      "step": 5210
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.8287532090635115e-05,
+      "loss": 0.2458,
+      "step": 5220
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.828418350262306e-05,
+      "loss": 0.3001,
+      "step": 5230
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.828083491461101e-05,
+      "loss": 0.6134,
+      "step": 5240
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.8277486326598954e-05,
+      "loss": 0.2871,
+      "step": 5250
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.8274137738586898e-05,
+      "loss": 0.6178,
+      "step": 5260
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.8270789150574842e-05,
+      "loss": 0.2693,
+      "step": 5270
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.8267440562562786e-05,
+      "loss": 0.4363,
+      "step": 5280
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.826409197455073e-05,
+      "loss": 0.2794,
+      "step": 5290
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.8260743386538674e-05,
+      "loss": 0.6376,
+      "step": 5300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.8257394798526625e-05,
+      "loss": 0.3625,
+      "step": 5310
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.825404621051457e-05,
+      "loss": 0.3627,
+      "step": 5320
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.8250697622502513e-05,
+      "loss": 0.4817,
+      "step": 5330
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.8247349034490457e-05,
+      "loss": 0.3692,
+      "step": 5340
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.82440004464784e-05,
+      "loss": 0.2642,
+      "step": 5350
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.8240651858466345e-05,
+      "loss": 0.2943,
+      "step": 5360
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.8237303270454293e-05,
+      "loss": 0.2088,
+      "step": 5370
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.823395468244224e-05,
+      "loss": 0.2777,
+      "step": 5380
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.8230606094430184e-05,
+      "loss": 0.3062,
+      "step": 5390
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.8227257506418128e-05,
+      "loss": 0.2769,
+      "step": 5400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.8223908918406072e-05,
+      "loss": 0.291,
+      "step": 5410
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.8220560330394016e-05,
+      "loss": 0.2674,
+      "step": 5420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.821721174238196e-05,
+      "loss": 0.3713,
+      "step": 5430
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.8213863154369908e-05,
+      "loss": 0.2747,
+      "step": 5440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.8210514566357855e-05,
+      "loss": 0.3134,
+      "step": 5450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.82071659783458e-05,
+      "loss": 0.2267,
+      "step": 5460
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.8203817390333744e-05,
+      "loss": 0.2755,
+      "step": 5470
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.8200468802321688e-05,
+      "loss": 0.3589,
+      "step": 5480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.819712021430963e-05,
+      "loss": 0.2279,
+      "step": 5490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.819377162629758e-05,
+      "loss": 0.3522,
+      "step": 5500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.8190423038285523e-05,
+      "loss": 0.298,
+      "step": 5510
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.818707445027347e-05,
+      "loss": 0.3389,
+      "step": 5520
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.8183725862261415e-05,
+      "loss": 0.1941,
+      "step": 5530
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.818037727424936e-05,
+      "loss": 0.2136,
+      "step": 5540
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.8177028686237303e-05,
+      "loss": 0.389,
+      "step": 5550
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.8173680098225247e-05,
+      "loss": 0.299,
+      "step": 5560
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.8170331510213194e-05,
+      "loss": 0.3413,
+      "step": 5570
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.816698292220114e-05,
+      "loss": 0.3061,
+      "step": 5580
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.8163634334189086e-05,
+      "loss": 0.2108,
+      "step": 5590
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.816028574617703e-05,
+      "loss": 0.6051,
+      "step": 5600
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.8156937158164974e-05,
+      "loss": 0.3605,
+      "step": 5610
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.8153588570152918e-05,
+      "loss": 0.3549,
+      "step": 5620
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.8150239982140866e-05,
+      "loss": 0.3417,
+      "step": 5630
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.814689139412881e-05,
+      "loss": 0.3333,
+      "step": 5640
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.8143542806116754e-05,
+      "loss": 0.4176,
+      "step": 5650
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.81401942181047e-05,
+      "loss": 0.3266,
+      "step": 5660
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.8136845630092645e-05,
+      "loss": 0.1713,
+      "step": 5670
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.813349704208059e-05,
+      "loss": 0.4737,
+      "step": 5680
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.8130148454068533e-05,
+      "loss": 0.3058,
+      "step": 5690
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.812679986605648e-05,
+      "loss": 0.6805,
+      "step": 5700
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8123451278044425e-05,
+      "loss": 0.2173,
+      "step": 5710
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.812010269003237e-05,
+      "loss": 0.3184,
+      "step": 5720
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8116754102020317e-05,
+      "loss": 0.3563,
+      "step": 5730
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.811340551400826e-05,
+      "loss": 0.3942,
+      "step": 5740
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8110056925996205e-05,
+      "loss": 0.3746,
+      "step": 5750
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8106708337984152e-05,
+      "loss": 0.214,
+      "step": 5760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8103359749972096e-05,
+      "loss": 0.2454,
+      "step": 5770
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.810001116196004e-05,
+      "loss": 0.2263,
+      "step": 5780
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8096662573947984e-05,
+      "loss": 0.3404,
+      "step": 5790
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8093313985935932e-05,
+      "loss": 0.3125,
+      "step": 5800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8089965397923876e-05,
+      "loss": 0.322,
+      "step": 5810
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.808661680991182e-05,
+      "loss": 0.3425,
+      "step": 5820
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8083268221899767e-05,
+      "loss": 0.2768,
+      "step": 5830
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.807991963388771e-05,
+      "loss": 0.2681,
+      "step": 5840
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8076571045875656e-05,
+      "loss": 0.3835,
+      "step": 5850
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8073222457863603e-05,
+      "loss": 0.2731,
+      "step": 5860
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8069873869851547e-05,
+      "loss": 0.215,
+      "step": 5870
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.806652528183949e-05,
+      "loss": 0.2777,
+      "step": 5880
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.806317669382744e-05,
+      "loss": 0.2691,
+      "step": 5890
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8059828105815383e-05,
+      "loss": 0.3341,
+      "step": 5900
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8056479517803327e-05,
+      "loss": 0.1912,
+      "step": 5910
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.805313092979127e-05,
+      "loss": 0.225,
+      "step": 5920
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.804978234177922e-05,
+      "loss": 0.2653,
+      "step": 5930
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8046433753767162e-05,
+      "loss": 0.2225,
+      "step": 5940
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8043085165755107e-05,
+      "loss": 0.419,
+      "step": 5950
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8039736577743054e-05,
+      "loss": 0.3128,
+      "step": 5960
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8036387989730998e-05,
+      "loss": 0.3426,
+      "step": 5970
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8033039401718942e-05,
+      "loss": 0.3459,
+      "step": 5980
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8029690813706886e-05,
+      "loss": 0.2858,
+      "step": 5990
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8026342225694834e-05,
+      "loss": 0.3456,
+      "step": 6000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8022993637682778e-05,
+      "loss": 0.3273,
+      "step": 6010
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8019645049670722e-05,
+      "loss": 0.4041,
+      "step": 6020
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.801629646165867e-05,
+      "loss": 0.2328,
+      "step": 6030
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8012947873646613e-05,
+      "loss": 0.266,
+      "step": 6040
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8009599285634557e-05,
+      "loss": 0.3212,
+      "step": 6050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.80062506976225e-05,
+      "loss": 0.3876,
+      "step": 6060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.800290210961045e-05,
+      "loss": 0.1836,
+      "step": 6070
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.7999553521598393e-05,
+      "loss": 0.2372,
+      "step": 6080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.799620493358634e-05,
+      "loss": 0.3036,
+      "step": 6090
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.7992856345574285e-05,
+      "loss": 0.3021,
+      "step": 6100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.798950775756223e-05,
+      "loss": 0.3691,
+      "step": 6110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.7986159169550173e-05,
+      "loss": 0.4164,
+      "step": 6120
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.7982810581538117e-05,
+      "loss": 0.2728,
+      "step": 6130
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.7979461993526064e-05,
+      "loss": 0.2764,
+      "step": 6140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.797611340551401e-05,
+      "loss": 0.3875,
+      "step": 6150
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.7972764817501956e-05,
+      "loss": 0.3168,
+      "step": 6160
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.79694162294899e-05,
+      "loss": 0.7763,
+      "step": 6170
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.7966067641477844e-05,
+      "loss": 0.2738,
+      "step": 6180
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.7962719053465788e-05,
+      "loss": 0.1838,
+      "step": 6190
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.7959370465453732e-05,
+      "loss": 0.3994,
+      "step": 6200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.795602187744168e-05,
+      "loss": 0.2111,
+      "step": 6210
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.7952673289429627e-05,
+      "loss": 0.412,
+      "step": 6220
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.794932470141757e-05,
+      "loss": 0.2002,
+      "step": 6230
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.7945976113405515e-05,
+      "loss": 0.307,
+      "step": 6240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.794262752539346e-05,
+      "loss": 0.2767,
+      "step": 6250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.7939278937381403e-05,
+      "loss": 0.2908,
+      "step": 6260
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.7935930349369347e-05,
+      "loss": 0.2108,
+      "step": 6270
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.7932581761357295e-05,
+      "loss": 0.2857,
+      "step": 6280
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.7929233173345242e-05,
+      "loss": 0.3357,
+      "step": 6290
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.7925884585333186e-05,
+      "loss": 0.3486,
+      "step": 6300
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.792253599732113e-05,
+      "loss": 0.2336,
+      "step": 6310
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.7919187409309075e-05,
+      "loss": 0.3682,
+      "step": 6320
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.791583882129702e-05,
+      "loss": 0.3267,
+      "step": 6330
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.7912490233284963e-05,
+      "loss": 0.2309,
+      "step": 6340
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.7909141645272914e-05,
+      "loss": 0.228,
+      "step": 6350
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.7905793057260858e-05,
+      "loss": 0.1805,
+      "step": 6360
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.79024444692488e-05,
+      "loss": 0.3414,
+      "step": 6370
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.7899095881236746e-05,
+      "loss": 0.4235,
+      "step": 6380
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.789574729322469e-05,
+      "loss": 0.2132,
+      "step": 6390
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.7892398705212634e-05,
+      "loss": 0.4204,
+      "step": 6400
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.7889050117200578e-05,
+      "loss": 0.2906,
+      "step": 6410
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.788570152918853e-05,
+      "loss": 0.2611,
+      "step": 6420
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7882352941176473e-05,
+      "loss": 0.3169,
+      "step": 6430
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7879004353164417e-05,
+      "loss": 0.2593,
+      "step": 6440
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.787565576515236e-05,
+      "loss": 0.2472,
+      "step": 6450
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7872307177140305e-05,
+      "loss": 0.3164,
+      "step": 6460
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.786895858912825e-05,
+      "loss": 0.3101,
+      "step": 6470
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7865610001116197e-05,
+      "loss": 0.2343,
+      "step": 6480
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7862261413104144e-05,
+      "loss": 0.2376,
+      "step": 6490
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.7858912825092088e-05,
+      "loss": 0.3243,
+      "step": 6500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7855564237080032e-05,
+      "loss": 0.4699,
+      "step": 6510
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7852215649067976e-05,
+      "loss": 0.3653,
+      "step": 6520
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.784886706105592e-05,
+      "loss": 0.3422,
+      "step": 6530
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7845518473043864e-05,
+      "loss": 0.2729,
+      "step": 6540
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7842169885031812e-05,
+      "loss": 0.2444,
+      "step": 6550
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.783882129701976e-05,
+      "loss": 0.2942,
+      "step": 6560
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7835472709007703e-05,
+      "loss": 0.5227,
+      "step": 6570
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7832124120995648e-05,
+      "loss": 0.2907,
+      "step": 6580
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.782877553298359e-05,
+      "loss": 0.2217,
+      "step": 6590
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.7825426944971536e-05,
+      "loss": 0.3995,
+      "step": 6600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.7822078356959483e-05,
+      "loss": 0.2314,
+      "step": 6610
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.7818729768947427e-05,
+      "loss": 0.2926,
+      "step": 6620
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.7815381180935375e-05,
+      "loss": 0.7338,
+      "step": 6630
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.781203259292332e-05,
+      "loss": 0.3562,
+      "step": 6640
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.7808684004911263e-05,
+      "loss": 0.2689,
+      "step": 6650
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.7805335416899207e-05,
+      "loss": 0.3426,
+      "step": 6660
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.780198682888715e-05,
+      "loss": 0.4198,
+      "step": 6670
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.77986382408751e-05,
+      "loss": 0.3625,
+      "step": 6680
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.7795289652863043e-05,
+      "loss": 0.3599,
+      "step": 6690
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.779194106485099e-05,
+      "loss": 0.2571,
+      "step": 6700
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.7788592476838934e-05,
+      "loss": 0.3255,
+      "step": 6710
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.7785243888826878e-05,
+      "loss": 0.3218,
+      "step": 6720
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.7781895300814822e-05,
+      "loss": 0.3153,
+      "step": 6730
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.777854671280277e-05,
+      "loss": 0.2871,
+      "step": 6740
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.7775198124790714e-05,
+      "loss": 0.4455,
+      "step": 6750
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.777184953677866e-05,
+      "loss": 0.3346,
+      "step": 6760
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.7768500948766605e-05,
+      "loss": 0.2649,
+      "step": 6770
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.776515236075455e-05,
+      "loss": 0.2111,
+      "step": 6780
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7761803772742493e-05,
+      "loss": 0.3545,
+      "step": 6790
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7758455184730438e-05,
+      "loss": 0.2561,
+      "step": 6800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7755106596718385e-05,
+      "loss": 0.2615,
+      "step": 6810
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.775175800870633e-05,
+      "loss": 0.2699,
+      "step": 6820
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7748409420694277e-05,
+      "loss": 0.2553,
+      "step": 6830
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.774506083268222e-05,
+      "loss": 0.2651,
+      "step": 6840
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7741712244670165e-05,
+      "loss": 0.478,
+      "step": 6850
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.773836365665811e-05,
+      "loss": 0.6018,
+      "step": 6860
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7735015068646056e-05,
+      "loss": 0.404,
+      "step": 6870
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7731666480634e-05,
+      "loss": 0.2023,
+      "step": 6880
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7728317892621944e-05,
+      "loss": 0.3029,
+      "step": 6890
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7724969304609892e-05,
+      "loss": 0.3345,
+      "step": 6900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7721620716597836e-05,
+      "loss": 0.469,
+      "step": 6910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.771827212858578e-05,
+      "loss": 0.2554,
+      "step": 6920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7714923540573724e-05,
+      "loss": 0.2283,
+      "step": 6930
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.771157495256167e-05,
+      "loss": 0.2976,
+      "step": 6940
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7708226364549616e-05,
+      "loss": 0.3634,
+      "step": 6950
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.770487777653756e-05,
+      "loss": 0.3067,
+      "step": 6960
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7701529188525507e-05,
+      "loss": 0.2931,
+      "step": 6970
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.769818060051345e-05,
+      "loss": 0.3567,
+      "step": 6980
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7694832012501395e-05,
+      "loss": 0.3428,
+      "step": 6990
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7691483424489343e-05,
+      "loss": 0.398,
+      "step": 7000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7688134836477287e-05,
+      "loss": 0.2065,
+      "step": 7010
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.768478624846523e-05,
+      "loss": 0.3929,
+      "step": 7020
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7681437660453175e-05,
+      "loss": 0.2601,
+      "step": 7030
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7678089072441122e-05,
+      "loss": 0.3549,
+      "step": 7040
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7674740484429066e-05,
+      "loss": 0.3289,
+      "step": 7050
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.767139189641701e-05,
+      "loss": 0.3119,
+      "step": 7060
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7668043308404958e-05,
+      "loss": 0.2307,
+      "step": 7070
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7664694720392902e-05,
+      "loss": 0.3178,
+      "step": 7080
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7661346132380846e-05,
+      "loss": 0.4569,
+      "step": 7090
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.765799754436879e-05,
+      "loss": 0.3105,
+      "step": 7100
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7654648956356738e-05,
+      "loss": 0.2964,
+      "step": 7110
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7651300368344682e-05,
+      "loss": 0.3224,
+      "step": 7120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.764795178033263e-05,
+      "loss": 0.4501,
+      "step": 7130
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7644603192320573e-05,
+      "loss": 0.3529,
+      "step": 7140
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7641254604308517e-05,
+      "loss": 0.315,
+      "step": 7150
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.763790601629646e-05,
+      "loss": 0.2057,
+      "step": 7160
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7634892287085612e-05,
+      "loss": 0.2861,
+      "step": 7170
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.763154369907356e-05,
+      "loss": 0.5034,
+      "step": 7180
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7628195111061504e-05,
+      "loss": 0.2359,
+      "step": 7190
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7624846523049448e-05,
+      "loss": 0.7508,
+      "step": 7200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7621497935037392e-05,
+      "loss": 0.2851,
+      "step": 7210
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.761814934702534e-05,
+      "loss": 0.2968,
+      "step": 7220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.7614800759013283e-05,
+      "loss": 0.2937,
+      "step": 7230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.7611452171001227e-05,
+      "loss": 0.333,
+      "step": 7240
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.7608103582989175e-05,
+      "loss": 0.392,
+      "step": 7250
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.760475499497712e-05,
+      "loss": 0.3851,
+      "step": 7260
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.7601406406965063e-05,
+      "loss": 0.35,
+      "step": 7270
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.7598057818953007e-05,
+      "loss": 0.3345,
+      "step": 7280
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.7594709230940955e-05,
+      "loss": 0.5123,
+      "step": 7290
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.75913606429289e-05,
+      "loss": 0.3653,
+      "step": 7300
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.7588012054916846e-05,
+      "loss": 0.2177,
+      "step": 7310
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.758466346690479e-05,
+      "loss": 0.3021,
+      "step": 7320
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.7581314878892734e-05,
+      "loss": 0.2501,
+      "step": 7330
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.757796629088068e-05,
+      "loss": 0.2917,
+      "step": 7340
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.7574617702868622e-05,
+      "loss": 0.3049,
+      "step": 7350
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.757126911485657e-05,
+      "loss": 0.1929,
+      "step": 7360
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.7567920526844514e-05,
+      "loss": 0.3441,
+      "step": 7370
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.756457193883246e-05,
+      "loss": 0.2256,
+      "step": 7380
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.7561223350820405e-05,
+      "loss": 0.3619,
+      "step": 7390
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.755787476280835e-05,
+      "loss": 0.4454,
+      "step": 7400
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.7554526174796294e-05,
+      "loss": 0.2736,
+      "step": 7410
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.7551177586784238e-05,
+      "loss": 0.3821,
+      "step": 7420
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.7547828998772185e-05,
+      "loss": 0.2481,
+      "step": 7430
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.7544480410760133e-05,
+      "loss": 0.3622,
+      "step": 7440
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.7541131822748077e-05,
+      "loss": 0.3723,
+      "step": 7450
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.753778323473602e-05,
+      "loss": 0.2469,
+      "step": 7460
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.7534434646723965e-05,
+      "loss": 0.3241,
+      "step": 7470
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.753108605871191e-05,
+      "loss": 0.5232,
+      "step": 7480
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7527737470699853e-05,
+      "loss": 0.2332,
+      "step": 7490
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.75243888826878e-05,
+      "loss": 0.2393,
+      "step": 7500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7521040294675748e-05,
+      "loss": 0.3177,
+      "step": 7510
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7517691706663692e-05,
+      "loss": 0.313,
+      "step": 7520
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7514343118651636e-05,
+      "loss": 0.2987,
+      "step": 7530
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.751099453063958e-05,
+      "loss": 0.3476,
+      "step": 7540
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7507645942627524e-05,
+      "loss": 0.3342,
+      "step": 7550
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7504297354615468e-05,
+      "loss": 0.292,
+      "step": 7560
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.750094876660342e-05,
+      "loss": 0.2421,
+      "step": 7570
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7497600178591363e-05,
+      "loss": 0.2247,
+      "step": 7580
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7494251590579307e-05,
+      "loss": 0.3347,
+      "step": 7590
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.749090300256725e-05,
+      "loss": 0.3745,
+      "step": 7600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7487554414555195e-05,
+      "loss": 0.2326,
+      "step": 7610
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.748420582654314e-05,
+      "loss": 0.2776,
+      "step": 7620
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7480857238531084e-05,
+      "loss": 0.3382,
+      "step": 7630
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7477508650519034e-05,
+      "loss": 0.2694,
+      "step": 7640
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.747416006250698e-05,
+      "loss": 0.3277,
+      "step": 7650
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7470811474494923e-05,
+      "loss": 0.2224,
+      "step": 7660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.7467462886482867e-05,
+      "loss": 0.407,
+      "step": 7670
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.746411429847081e-05,
+      "loss": 0.2355,
+      "step": 7680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.7460765710458755e-05,
+      "loss": 0.3267,
+      "step": 7690
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.7457417122446702e-05,
+      "loss": 0.2821,
+      "step": 7700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.745406853443465e-05,
+      "loss": 0.3184,
+      "step": 7710
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.7450719946422594e-05,
+      "loss": 0.2352,
+      "step": 7720
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.7447371358410538e-05,
+      "loss": 0.2283,
+      "step": 7730
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.7444022770398482e-05,
+      "loss": 0.7634,
+      "step": 7740
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.7440674182386426e-05,
+      "loss": 0.2692,
+      "step": 7750
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.743732559437437e-05,
+      "loss": 0.3391,
+      "step": 7760
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7433977006362318e-05,
+      "loss": 0.2188,
+      "step": 7770
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7430628418350265e-05,
+      "loss": 0.2638,
+      "step": 7780
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.742727983033821e-05,
+      "loss": 0.2916,
+      "step": 7790
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7423931242326153e-05,
+      "loss": 0.6189,
+      "step": 7800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7420582654314097e-05,
+      "loss": 0.2545,
+      "step": 7810
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.741723406630204e-05,
+      "loss": 0.1335,
+      "step": 7820
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.741388547828999e-05,
+      "loss": 0.2259,
+      "step": 7830
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7410536890277933e-05,
+      "loss": 0.2655,
+      "step": 7840
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.740718830226588e-05,
+      "loss": 0.6887,
+      "step": 7850
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7403839714253824e-05,
+      "loss": 0.165,
+      "step": 7860
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.740049112624177e-05,
+      "loss": 0.3059,
+      "step": 7870
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7397142538229713e-05,
+      "loss": 0.5218,
+      "step": 7880
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7393793950217657e-05,
+      "loss": 0.2873,
+      "step": 7890
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7390445362205604e-05,
+      "loss": 0.3783,
+      "step": 7900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.738709677419355e-05,
+      "loss": 0.3495,
+      "step": 7910
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7383748186181496e-05,
+      "loss": 0.2612,
+      "step": 7920
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.738039959816944e-05,
+      "loss": 0.2643,
+      "step": 7930
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.7377051010157384e-05,
+      "loss": 0.3061,
+      "step": 7940
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.7373702422145328e-05,
+      "loss": 0.3945,
+      "step": 7950
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.7370353834133275e-05,
+      "loss": 0.3046,
+      "step": 7960
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.736700524612122e-05,
+      "loss": 0.2879,
+      "step": 7970
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.7363656658109167e-05,
+      "loss": 0.3325,
+      "step": 7980
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.736030807009711e-05,
+      "loss": 0.2016,
+      "step": 7990
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.7356959482085055e-05,
+      "loss": 0.3561,
+      "step": 8000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.7353610894073e-05,
+      "loss": 0.2973,
+      "step": 8010
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.7350262306060943e-05,
+      "loss": 0.2704,
+      "step": 8020
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.734691371804889e-05,
+      "loss": 0.4374,
+      "step": 8030
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7343565130036835e-05,
+      "loss": 0.3368,
+      "step": 8040
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7340216542024782e-05,
+      "loss": 0.408,
+      "step": 8050
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7336867954012726e-05,
+      "loss": 0.5914,
+      "step": 8060
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.733351936600067e-05,
+      "loss": 0.3722,
+      "step": 8070
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7330170777988614e-05,
+      "loss": 0.2576,
+      "step": 8080
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7326822189976562e-05,
+      "loss": 0.2548,
+      "step": 8090
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7323473601964506e-05,
+      "loss": 0.2349,
+      "step": 8100
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.732012501395245e-05,
+      "loss": 0.2487,
+      "step": 8110
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7316776425940397e-05,
+      "loss": 0.2734,
+      "step": 8120
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.731342783792834e-05,
+      "loss": 0.2177,
+      "step": 8130
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7310079249916286e-05,
+      "loss": 0.3506,
+      "step": 8140
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.730673066190423e-05,
+      "loss": 0.2663,
+      "step": 8150
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7303382073892177e-05,
+      "loss": 0.3138,
+      "step": 8160
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.730003348588012e-05,
+      "loss": 0.3519,
+      "step": 8170
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7296684897868065e-05,
+      "loss": 0.385,
+      "step": 8180
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7293336309856013e-05,
+      "loss": 0.2028,
+      "step": 8190
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7289987721843957e-05,
+      "loss": 0.3046,
+      "step": 8200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.72866391338319e-05,
+      "loss": 0.3495,
+      "step": 8210
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.728329054581985e-05,
+      "loss": 0.2663,
+      "step": 8220
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7279941957807792e-05,
+      "loss": 0.3339,
+      "step": 8230
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7276593369795736e-05,
+      "loss": 0.2028,
+      "step": 8240
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.727324478178368e-05,
+      "loss": 0.3159,
+      "step": 8250
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7269896193771628e-05,
+      "loss": 0.2089,
+      "step": 8260
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7266547605759572e-05,
+      "loss": 0.2452,
+      "step": 8270
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7263199017747516e-05,
+      "loss": 0.318,
+      "step": 8280
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7259850429735464e-05,
+      "loss": 0.3348,
+      "step": 8290
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7256501841723408e-05,
+      "loss": 0.2867,
+      "step": 8300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7253153253711352e-05,
+      "loss": 0.4072,
+      "step": 8310
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7249804665699296e-05,
+      "loss": 0.2086,
+      "step": 8320
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7246456077687243e-05,
+      "loss": 0.2383,
+      "step": 8330
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7243107489675187e-05,
+      "loss": 0.3159,
+      "step": 8340
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7239758901663135e-05,
+      "loss": 0.5242,
+      "step": 8350
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.723641031365108e-05,
+      "loss": 0.3471,
+      "step": 8360
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7233061725639023e-05,
+      "loss": 0.3031,
+      "step": 8370
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7229713137626967e-05,
+      "loss": 0.3838,
+      "step": 8380
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.722636454961491e-05,
+      "loss": 0.2704,
+      "step": 8390
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.722301596160286e-05,
+      "loss": 0.2246,
+      "step": 8400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.7219667373590803e-05,
+      "loss": 0.3057,
+      "step": 8410
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.721631878557875e-05,
+      "loss": 0.268,
+      "step": 8420
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.7212970197566694e-05,
+      "loss": 0.2617,
+      "step": 8430
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.7209621609554638e-05,
+      "loss": 0.3125,
+      "step": 8440
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.7206273021542582e-05,
+      "loss": 0.3471,
+      "step": 8450
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.7202924433530526e-05,
+      "loss": 0.2196,
+      "step": 8460
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.7199575845518474e-05,
+      "loss": 0.3354,
+      "step": 8470
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.719622725750642e-05,
+      "loss": 0.3732,
+      "step": 8480
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7192878669494365e-05,
+      "loss": 0.276,
+      "step": 8490
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.718953008148231e-05,
+      "loss": 0.7655,
+      "step": 8500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7186181493470254e-05,
+      "loss": 0.3478,
+      "step": 8510
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7182832905458198e-05,
+      "loss": 0.3035,
+      "step": 8520
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7179484317446142e-05,
+      "loss": 0.2893,
+      "step": 8530
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.717613572943409e-05,
+      "loss": 0.2715,
+      "step": 8540
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7172787141422037e-05,
+      "loss": 0.4414,
+      "step": 8550
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.716943855340998e-05,
+      "loss": 0.3527,
+      "step": 8560
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.7166089965397925e-05,
+      "loss": 0.3223,
+      "step": 8570
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.716274137738587e-05,
+      "loss": 0.3145,
+      "step": 8580
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.7159392789373813e-05,
+      "loss": 0.3341,
+      "step": 8590
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.7156044201361757e-05,
+      "loss": 0.3911,
+      "step": 8600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.7152695613349704e-05,
+      "loss": 0.3075,
+      "step": 8610
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.7149347025337652e-05,
+      "loss": 0.2984,
+      "step": 8620
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.7145998437325596e-05,
+      "loss": 0.2122,
+      "step": 8630
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.714264984931354e-05,
+      "loss": 0.3097,
+      "step": 8640
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.7139301261301484e-05,
+      "loss": 0.4159,
+      "step": 8650
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7135952673289428e-05,
+      "loss": 0.3769,
+      "step": 8660
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7132604085277372e-05,
+      "loss": 0.3375,
+      "step": 8670
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7129255497265323e-05,
+      "loss": 0.2706,
+      "step": 8680
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7125906909253267e-05,
+      "loss": 0.2798,
+      "step": 8690
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.712255832124121e-05,
+      "loss": 0.4084,
+      "step": 8700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7119209733229155e-05,
+      "loss": 0.4589,
+      "step": 8710
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.71158611452171e-05,
+      "loss": 0.287,
+      "step": 8720
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7112512557205044e-05,
+      "loss": 0.3491,
+      "step": 8730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.710916396919299e-05,
+      "loss": 0.1877,
+      "step": 8740
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.710581538118094e-05,
+      "loss": 0.3611,
+      "step": 8750
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7102466793168883e-05,
+      "loss": 0.2345,
+      "step": 8760
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7099118205156827e-05,
+      "loss": 0.2327,
+      "step": 8770
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.709576961714477e-05,
+      "loss": 0.2995,
+      "step": 8780
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7092421029132715e-05,
+      "loss": 0.3004,
+      "step": 8790
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.708907244112066e-05,
+      "loss": 0.2853,
+      "step": 8800
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.708572385310861e-05,
+      "loss": 0.2752,
+      "step": 8810
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7082375265096554e-05,
+      "loss": 0.2455,
+      "step": 8820
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7079026677084498e-05,
+      "loss": 0.3564,
+      "step": 8830
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7075678089072442e-05,
+      "loss": 0.3099,
+      "step": 8840
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7072329501060386e-05,
+      "loss": 0.4908,
+      "step": 8850
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.706898091304833e-05,
+      "loss": 0.2684,
+      "step": 8860
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7065632325036278e-05,
+      "loss": 0.2894,
+      "step": 8870
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7062283737024225e-05,
+      "loss": 0.2853,
+      "step": 8880
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.705893514901217e-05,
+      "loss": 0.3977,
+      "step": 8890
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7055586561000113e-05,
+      "loss": 0.4053,
+      "step": 8900
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7052237972988057e-05,
+      "loss": 0.2505,
+      "step": 8910
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7048889384976e-05,
+      "loss": 0.576,
+      "step": 8920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7045540796963945e-05,
+      "loss": 0.2776,
+      "step": 8930
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7042192208951893e-05,
+      "loss": 0.3915,
+      "step": 8940
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.703884362093984e-05,
+      "loss": 0.3445,
+      "step": 8950
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7035495032927784e-05,
+      "loss": 0.2664,
+      "step": 8960
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 0.3346670866012573,
+      "eval_runtime": 249.3241,
+      "eval_samples_per_second": 7.998,
+      "eval_steps_per_second": 3.999,
+      "eval_wer": 0.16450466324391652,
+      "step": 8969
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.703214644491573e-05,
+      "loss": 0.2525,
+      "step": 8970
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7028797856903672e-05,
+      "loss": 0.2641,
+      "step": 8980
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7025449268891617e-05,
+      "loss": 0.2675,
+      "step": 8990
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.702210068087956e-05,
+      "loss": 0.2473,
+      "step": 9000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7018752092867508e-05,
+      "loss": 0.2298,
+      "step": 9010
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7015403504855456e-05,
+      "loss": 0.3689,
+      "step": 9020
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.70120549168434e-05,
+      "loss": 0.2625,
+      "step": 9030
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7008706328831344e-05,
+      "loss": 0.1601,
+      "step": 9040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7005357740819288e-05,
+      "loss": 0.1862,
+      "step": 9050
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7002009152807232e-05,
+      "loss": 0.2219,
+      "step": 9060
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.699866056479518e-05,
+      "loss": 0.178,
+      "step": 9070
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6995311976783123e-05,
+      "loss": 0.3174,
+      "step": 9080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.699196338877107e-05,
+      "loss": 0.3293,
+      "step": 9090
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6988614800759015e-05,
+      "loss": 0.2257,
+      "step": 9100
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.698526621274696e-05,
+      "loss": 0.2686,
+      "step": 9110
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6981917624734903e-05,
+      "loss": 0.3347,
+      "step": 9120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6978569036722847e-05,
+      "loss": 0.2804,
+      "step": 9130
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6975220448710795e-05,
+      "loss": 0.2923,
+      "step": 9140
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.697187186069874e-05,
+      "loss": 0.4099,
+      "step": 9150
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6968523272686686e-05,
+      "loss": 0.2137,
+      "step": 9160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.696517468467463e-05,
+      "loss": 0.2378,
+      "step": 9170
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.696216095546378e-05,
+      "loss": 0.1275,
+      "step": 9180
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6958812367451725e-05,
+      "loss": 0.3359,
+      "step": 9190
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.6955463779439672e-05,
+      "loss": 0.2222,
+      "step": 9200
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.6952115191427617e-05,
+      "loss": 0.1577,
+      "step": 9210
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.694876660341556e-05,
+      "loss": 0.294,
+      "step": 9220
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.6945418015403505e-05,
+      "loss": 0.3196,
+      "step": 9230
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.694206942739145e-05,
+      "loss": 0.2558,
+      "step": 9240
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.6938720839379396e-05,
+      "loss": 0.2528,
+      "step": 9250
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.693537225136734e-05,
+      "loss": 0.2033,
+      "step": 9260
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.6932023663355288e-05,
+      "loss": 0.31,
+      "step": 9270
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.6928675075343232e-05,
+      "loss": 0.2429,
+      "step": 9280
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.6925326487331176e-05,
+      "loss": 0.1972,
+      "step": 9290
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.692197789931912e-05,
+      "loss": 0.2162,
+      "step": 9300
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.6918629311307067e-05,
+      "loss": 0.2379,
+      "step": 9310
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.691528072329501e-05,
+      "loss": 0.2992,
+      "step": 9320
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.6911932135282956e-05,
+      "loss": 0.239,
+      "step": 9330
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.6908583547270903e-05,
+      "loss": 0.293,
+      "step": 9340
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.6905234959258847e-05,
+      "loss": 0.2968,
+      "step": 9350
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.690188637124679e-05,
+      "loss": 0.3057,
+      "step": 9360
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.6898537783234735e-05,
+      "loss": 0.1848,
+      "step": 9370
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.6895189195222683e-05,
+      "loss": 0.149,
+      "step": 9380
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.6891840607210627e-05,
+      "loss": 0.2216,
+      "step": 9390
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.688849201919857e-05,
+      "loss": 0.1308,
+      "step": 9400
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.688514343118652e-05,
+      "loss": 0.2552,
+      "step": 9410
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.6881794843174462e-05,
+      "loss": 0.507,
+      "step": 9420
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.6878446255162406e-05,
+      "loss": 0.3108,
+      "step": 9430
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.6875097667150354e-05,
+      "loss": 0.2446,
+      "step": 9440
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.6871749079138298e-05,
+      "loss": 0.2607,
+      "step": 9450
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.6868400491126242e-05,
+      "loss": 0.2737,
+      "step": 9460
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.6865051903114186e-05,
+      "loss": 0.2506,
+      "step": 9470
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.6861703315102134e-05,
+      "loss": 0.1653,
+      "step": 9480
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.6858354727090078e-05,
+      "loss": 0.205,
+      "step": 9490
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.6855006139078022e-05,
+      "loss": 0.2554,
+      "step": 9500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.685165755106597e-05,
+      "loss": 0.142,
+      "step": 9510
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.6848308963053913e-05,
+      "loss": 0.2084,
+      "step": 9520
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.6844960375041857e-05,
+      "loss": 0.1318,
+      "step": 9530
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.68416117870298e-05,
+      "loss": 0.1304,
+      "step": 9540
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.683826319901775e-05,
+      "loss": 0.2451,
+      "step": 9550
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.6834914611005693e-05,
+      "loss": 0.2463,
+      "step": 9560
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.683156602299364e-05,
+      "loss": 0.2715,
+      "step": 9570
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.6828217434981585e-05,
+      "loss": 0.2025,
+      "step": 9580
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.682486884696953e-05,
+      "loss": 0.2108,
+      "step": 9590
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.6821520258957473e-05,
+      "loss": 0.2515,
+      "step": 9600
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.6818171670945417e-05,
+      "loss": 0.256,
+      "step": 9610
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.6814823082933364e-05,
+      "loss": 0.2701,
+      "step": 9620
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.6811474494921308e-05,
+      "loss": 0.2235,
+      "step": 9630
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.6808125906909256e-05,
+      "loss": 0.2637,
+      "step": 9640
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.68047773188972e-05,
+      "loss": 0.2216,
+      "step": 9650
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.6801428730885144e-05,
+      "loss": 0.1757,
+      "step": 9660
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.6798080142873088e-05,
+      "loss": 0.3206,
+      "step": 9670
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.6794731554861032e-05,
+      "loss": 0.1884,
+      "step": 9680
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.679138296684898e-05,
+      "loss": 0.1755,
+      "step": 9690
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.6788034378836927e-05,
+      "loss": 0.2358,
+      "step": 9700
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.678468579082487e-05,
+      "loss": 0.2804,
+      "step": 9710
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.6781337202812815e-05,
+      "loss": 0.3231,
+      "step": 9720
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.677798861480076e-05,
+      "loss": 0.2914,
+      "step": 9730
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.6774640026788703e-05,
+      "loss": 0.2383,
+      "step": 9740
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.6771291438776647e-05,
+      "loss": 0.2889,
+      "step": 9750
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.6767942850764595e-05,
+      "loss": 0.3235,
+      "step": 9760
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.6764594262752542e-05,
+      "loss": 0.3888,
+      "step": 9770
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.6761245674740486e-05,
+      "loss": 0.2638,
+      "step": 9780
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.675789708672843e-05,
+      "loss": 0.3879,
+      "step": 9790
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.6754548498716374e-05,
+      "loss": 0.3365,
+      "step": 9800
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.675119991070432e-05,
+      "loss": 0.1908,
+      "step": 9810
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.6747851322692263e-05,
+      "loss": 0.3094,
+      "step": 9820
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.674450273468021e-05,
+      "loss": 0.3355,
+      "step": 9830
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.6741154146668158e-05,
+      "loss": 0.2109,
+      "step": 9840
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.6738140417457305e-05,
+      "loss": 0.3598,
+      "step": 9850
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.673479182944525e-05,
+      "loss": 0.218,
+      "step": 9860
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.6731443241433196e-05,
+      "loss": 0.4365,
+      "step": 9870
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.6728094653421144e-05,
+      "loss": 0.2864,
+      "step": 9880
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.6724746065409088e-05,
+      "loss": 0.2394,
+      "step": 9890
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.6721397477397032e-05,
+      "loss": 0.205,
+      "step": 9900
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.6718048889384976e-05,
+      "loss": 0.241,
+      "step": 9910
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.671470030137292e-05,
+      "loss": 0.2219,
+      "step": 9920
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.6711351713360864e-05,
+      "loss": 0.1895,
+      "step": 9930
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.6708003125348812e-05,
+      "loss": 0.3271,
+      "step": 9940
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.670465453733676e-05,
+      "loss": 0.1669,
+      "step": 9950
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.6701305949324703e-05,
+      "loss": 0.4085,
+      "step": 9960
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.6697957361312647e-05,
+      "loss": 0.3369,
+      "step": 9970
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.669460877330059e-05,
+      "loss": 0.3931,
+      "step": 9980
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.6691260185288535e-05,
+      "loss": 0.224,
+      "step": 9990
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.668791159727648e-05,
+      "loss": 0.3224,
+      "step": 10000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.668456300926443e-05,
+      "loss": 0.2636,
+      "step": 10010
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.6681214421252374e-05,
+      "loss": 0.3033,
+      "step": 10020
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.667786583324032e-05,
+      "loss": 0.2237,
+      "step": 10030
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.6674517245228263e-05,
+      "loss": 0.3265,
+      "step": 10040
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.6671168657216207e-05,
+      "loss": 0.3295,
+      "step": 10050
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.666782006920415e-05,
+      "loss": 0.2043,
+      "step": 10060
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.6664471481192095e-05,
+      "loss": 0.2789,
+      "step": 10070
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.6661122893180046e-05,
+      "loss": 0.3074,
+      "step": 10080
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.665777430516799e-05,
+      "loss": 0.2384,
+      "step": 10090
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.6654425717155934e-05,
+      "loss": 0.2177,
+      "step": 10100
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.6651077129143878e-05,
+      "loss": 0.2484,
+      "step": 10110
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.6647728541131822e-05,
+      "loss": 0.2978,
+      "step": 10120
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.6644379953119766e-05,
+      "loss": 0.343,
+      "step": 10130
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.6641031365107714e-05,
+      "loss": 0.2244,
+      "step": 10140
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.663768277709566e-05,
+      "loss": 0.2795,
+      "step": 10150
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.6634334189083605e-05,
+      "loss": 0.339,
+      "step": 10160
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.663098560107155e-05,
+      "loss": 0.2588,
+      "step": 10170
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.6627637013059493e-05,
+      "loss": 0.2395,
+      "step": 10180
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.6624288425047437e-05,
+      "loss": 0.2507,
+      "step": 10190
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.662093983703538e-05,
+      "loss": 0.1724,
+      "step": 10200
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.661759124902333e-05,
+      "loss": 0.2143,
+      "step": 10210
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.6614242661011276e-05,
+      "loss": 0.429,
+      "step": 10220
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.661089407299922e-05,
+      "loss": 0.3569,
+      "step": 10230
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.6607545484987164e-05,
+      "loss": 0.2578,
+      "step": 10240
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.660419689697511e-05,
+      "loss": 0.2231,
+      "step": 10250
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.6600848308963053e-05,
+      "loss": 0.2667,
+      "step": 10260
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.6597499720951e-05,
+      "loss": 0.3537,
+      "step": 10270
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.6594151132938947e-05,
+      "loss": 0.3675,
+      "step": 10280
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.659080254492689e-05,
+      "loss": 0.1742,
+      "step": 10290
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.6587453956914836e-05,
+      "loss": 0.2174,
+      "step": 10300
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.658410536890278e-05,
+      "loss": 0.2295,
+      "step": 10310
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.6580756780890724e-05,
+      "loss": 0.2379,
+      "step": 10320
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.6577408192878668e-05,
+      "loss": 0.2454,
+      "step": 10330
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.6574059604866615e-05,
+      "loss": 0.2329,
+      "step": 10340
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.6570711016854563e-05,
+      "loss": 0.213,
+      "step": 10350
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.6567362428842507e-05,
+      "loss": 0.1049,
+      "step": 10360
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.656401384083045e-05,
+      "loss": 0.2576,
+      "step": 10370
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.6560665252818395e-05,
+      "loss": 0.2028,
+      "step": 10380
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.655731666480634e-05,
+      "loss": 0.3962,
+      "step": 10390
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.6553968076794287e-05,
+      "loss": 0.1467,
+      "step": 10400
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.655061948878223e-05,
+      "loss": 0.2353,
+      "step": 10410
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.6547270900770178e-05,
+      "loss": 0.1772,
+      "step": 10420
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.6543922312758122e-05,
+      "loss": 0.1847,
+      "step": 10430
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.6540573724746066e-05,
+      "loss": 0.4424,
+      "step": 10440
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.653722513673401e-05,
+      "loss": 0.2931,
+      "step": 10450
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.6533876548721954e-05,
+      "loss": 0.23,
+      "step": 10460
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.6530527960709902e-05,
+      "loss": 0.3164,
+      "step": 10470
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.6527514231499053e-05,
+      "loss": 0.4672,
+      "step": 10480
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.6524165643486997e-05,
+      "loss": 0.2805,
+      "step": 10490
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.652081705547494e-05,
+      "loss": 0.2335,
+      "step": 10500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.6517468467462888e-05,
+      "loss": 0.1888,
+      "step": 10510
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.6514119879450832e-05,
+      "loss": 0.184,
+      "step": 10520
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.651077129143878e-05,
+      "loss": 0.2704,
+      "step": 10530
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.6507422703426724e-05,
+      "loss": 0.1775,
+      "step": 10540
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.6504074115414668e-05,
+      "loss": 0.3833,
+      "step": 10550
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.6500725527402612e-05,
+      "loss": 0.2507,
+      "step": 10560
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.6497376939390556e-05,
+      "loss": 0.3567,
+      "step": 10570
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.6494028351378503e-05,
+      "loss": 0.1899,
+      "step": 10580
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.6490679763366447e-05,
+      "loss": 0.293,
+      "step": 10590
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.6487331175354395e-05,
+      "loss": 0.188,
+      "step": 10600
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.648398258734234e-05,
+      "loss": 0.2296,
+      "step": 10610
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.6480633999330283e-05,
+      "loss": 0.2484,
+      "step": 10620
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.6477285411318227e-05,
+      "loss": 0.2253,
+      "step": 10630
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.647393682330617e-05,
+      "loss": 0.3108,
+      "step": 10640
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.647058823529412e-05,
+      "loss": 0.3307,
+      "step": 10650
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.6467239647282063e-05,
+      "loss": 0.3089,
+      "step": 10660
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.646389105927001e-05,
+      "loss": 0.3091,
+      "step": 10670
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.6460542471257954e-05,
+      "loss": 0.2912,
+      "step": 10680
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.64571938832459e-05,
+      "loss": 0.1722,
+      "step": 10690
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.6453845295233842e-05,
+      "loss": 0.1829,
+      "step": 10700
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.645049670722179e-05,
+      "loss": 0.2818,
+      "step": 10710
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.6447148119209734e-05,
+      "loss": 0.2833,
+      "step": 10720
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.6443799531197678e-05,
+      "loss": 0.2131,
+      "step": 10730
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.6440450943185626e-05,
+      "loss": 0.2096,
+      "step": 10740
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.643710235517357e-05,
+      "loss": 0.2213,
+      "step": 10750
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.6433753767161514e-05,
+      "loss": 0.2232,
+      "step": 10760
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.6430405179149458e-05,
+      "loss": 0.3062,
+      "step": 10770
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.6427056591137405e-05,
+      "loss": 0.2037,
+      "step": 10780
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.642370800312535e-05,
+      "loss": 0.191,
+      "step": 10790
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.6420359415113293e-05,
+      "loss": 0.1917,
+      "step": 10800
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.641701082710124e-05,
+      "loss": 0.2108,
+      "step": 10810
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.6413662239089185e-05,
+      "loss": 0.4272,
+      "step": 10820
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.641031365107713e-05,
+      "loss": 0.2584,
+      "step": 10830
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.6406965063065076e-05,
+      "loss": 0.2211,
+      "step": 10840
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.640361647505302e-05,
+      "loss": 0.2963,
+      "step": 10850
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.6400267887040965e-05,
+      "loss": 0.2631,
+      "step": 10860
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.639691929902891e-05,
+      "loss": 0.3302,
+      "step": 10870
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.6393570711016856e-05,
+      "loss": 0.1095,
+      "step": 10880
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.63902221230048e-05,
+      "loss": 0.1613,
+      "step": 10890
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.6386873534992744e-05,
+      "loss": 0.2672,
+      "step": 10900
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.6383524946980692e-05,
+      "loss": 0.3972,
+      "step": 10910
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.6380176358968636e-05,
+      "loss": 0.4338,
+      "step": 10920
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.637682777095658e-05,
+      "loss": 0.1499,
+      "step": 10930
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.6373479182944524e-05,
+      "loss": 0.1939,
+      "step": 10940
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.637013059493247e-05,
+      "loss": 0.3769,
+      "step": 10950
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.6366782006920416e-05,
+      "loss": 0.1755,
+      "step": 10960
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.6363433418908363e-05,
+      "loss": 0.3183,
+      "step": 10970
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.6360084830896307e-05,
+      "loss": 0.3501,
+      "step": 10980
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.635673624288425e-05,
+      "loss": 0.3053,
+      "step": 10990
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.6353387654872195e-05,
+      "loss": 0.2702,
+      "step": 11000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.635003906686014e-05,
+      "loss": 0.2791,
+      "step": 11010
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.6346690478848087e-05,
+      "loss": 0.244,
+      "step": 11020
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.634334189083603e-05,
+      "loss": 0.2047,
+      "step": 11030
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.6339993302823978e-05,
+      "loss": 0.6025,
+      "step": 11040
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.6336644714811922e-05,
+      "loss": 0.2733,
+      "step": 11050
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.6333296126799866e-05,
+      "loss": 0.3544,
+      "step": 11060
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.632994753878781e-05,
+      "loss": 0.1856,
+      "step": 11070
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.6326598950775755e-05,
+      "loss": 0.2224,
+      "step": 11080
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.6323250362763702e-05,
+      "loss": 0.3794,
+      "step": 11090
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.631990177475165e-05,
+      "loss": 0.2476,
+      "step": 11100
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.6316553186739594e-05,
+      "loss": 0.294,
+      "step": 11110
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.6313204598727538e-05,
+      "loss": 0.3297,
+      "step": 11120
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.6309856010715482e-05,
+      "loss": 0.1871,
+      "step": 11130
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.6306507422703426e-05,
+      "loss": 0.2212,
+      "step": 11140
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.630315883469137e-05,
+      "loss": 0.3105,
+      "step": 11150
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.6299810246679317e-05,
+      "loss": 0.2786,
+      "step": 11160
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.6296461658667265e-05,
+      "loss": 0.2739,
+      "step": 11170
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.629311307065521e-05,
+      "loss": 0.3255,
+      "step": 11180
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.6289764482643153e-05,
+      "loss": 0.2514,
+      "step": 11190
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.6286415894631097e-05,
+      "loss": 0.1755,
+      "step": 11200
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.628306730661904e-05,
+      "loss": 0.2873,
+      "step": 11210
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.6279718718606985e-05,
+      "loss": 0.8128,
+      "step": 11220
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.6276370130594936e-05,
+      "loss": 0.2635,
+      "step": 11230
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.627302154258288e-05,
+      "loss": 0.2001,
+      "step": 11240
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.6269672954570824e-05,
+      "loss": 0.1692,
+      "step": 11250
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.6266324366558768e-05,
+      "loss": 0.1837,
+      "step": 11260
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.6262975778546712e-05,
+      "loss": 0.3403,
+      "step": 11270
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.6259627190534656e-05,
+      "loss": 0.2893,
+      "step": 11280
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.62562786025226e-05,
+      "loss": 0.1966,
+      "step": 11290
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.625293001451055e-05,
+      "loss": 0.3653,
+      "step": 11300
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.6249581426498495e-05,
+      "loss": 0.2678,
+      "step": 11310
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.624623283848644e-05,
+      "loss": 0.2362,
+      "step": 11320
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.6242884250474384e-05,
+      "loss": 0.3632,
+      "step": 11330
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.6239535662462328e-05,
+      "loss": 0.2101,
+      "step": 11340
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.623618707445027e-05,
+      "loss": 0.1787,
+      "step": 11350
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.623283848643822e-05,
+      "loss": 0.2556,
+      "step": 11360
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.6229489898426167e-05,
+      "loss": 0.3298,
+      "step": 11370
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.622614131041411e-05,
+      "loss": 0.2385,
+      "step": 11380
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.6222792722402055e-05,
+      "loss": 0.3332,
+      "step": 11390
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.621944413439e-05,
+      "loss": 0.2642,
+      "step": 11400
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.6216095546377943e-05,
+      "loss": 0.2497,
+      "step": 11410
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.6212746958365887e-05,
+      "loss": 0.2383,
+      "step": 11420
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.6209398370353838e-05,
+      "loss": 0.1677,
+      "step": 11430
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.6206049782341782e-05,
+      "loss": 0.2035,
+      "step": 11440
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.6202701194329726e-05,
+      "loss": 0.1926,
+      "step": 11450
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.619935260631767e-05,
+      "loss": 0.241,
+      "step": 11460
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.6196004018305614e-05,
+      "loss": 0.2649,
+      "step": 11470
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.6192655430293558e-05,
+      "loss": 0.3361,
+      "step": 11480
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.6189306842281506e-05,
+      "loss": 0.1692,
+      "step": 11490
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.6185958254269453e-05,
+      "loss": 0.2839,
+      "step": 11500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.6182609666257397e-05,
+      "loss": 0.1959,
+      "step": 11510
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.617926107824534e-05,
+      "loss": 0.3348,
+      "step": 11520
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6175912490233285e-05,
+      "loss": 0.215,
+      "step": 11530
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.617256390222123e-05,
+      "loss": 0.2377,
+      "step": 11540
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6169215314209173e-05,
+      "loss": 0.2106,
+      "step": 11550
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.616586672619712e-05,
+      "loss": 0.1631,
+      "step": 11560
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.616251813818507e-05,
+      "loss": 0.367,
+      "step": 11570
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6159169550173012e-05,
+      "loss": 0.2345,
+      "step": 11580
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6155820962160957e-05,
+      "loss": 0.252,
+      "step": 11590
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.61524723741489e-05,
+      "loss": 0.1645,
+      "step": 11600
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6149123786136845e-05,
+      "loss": 0.1478,
+      "step": 11610
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6145775198124792e-05,
+      "loss": 0.3025,
+      "step": 11620
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6142426610112736e-05,
+      "loss": 0.2617,
+      "step": 11630
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6139078022100684e-05,
+      "loss": 0.1826,
+      "step": 11640
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6135729434088628e-05,
+      "loss": 0.1867,
+      "step": 11650
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6132380846076572e-05,
+      "loss": 0.1995,
+      "step": 11660
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6129032258064516e-05,
+      "loss": 0.2595,
+      "step": 11670
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.612568367005246e-05,
+      "loss": 0.2458,
+      "step": 11680
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6122335082040407e-05,
+      "loss": 0.2704,
+      "step": 11690
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.611898649402835e-05,
+      "loss": 0.1545,
+      "step": 11700
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.61156379060163e-05,
+      "loss": 0.2383,
+      "step": 11710
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6112289318004243e-05,
+      "loss": 0.3162,
+      "step": 11720
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6108940729992187e-05,
+      "loss": 0.2845,
+      "step": 11730
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.610559214198013e-05,
+      "loss": 0.2955,
+      "step": 11740
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.610224355396808e-05,
+      "loss": 0.3526,
+      "step": 11750
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6098894965956023e-05,
+      "loss": 0.1835,
+      "step": 11760
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6095546377943967e-05,
+      "loss": 0.2168,
+      "step": 11770
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6092197789931914e-05,
+      "loss": 0.1172,
+      "step": 11780
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.608884920191986e-05,
+      "loss": 0.255,
+      "step": 11790
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6085500613907802e-05,
+      "loss": 0.3347,
+      "step": 11800
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6082152025895747e-05,
+      "loss": 0.2098,
+      "step": 11810
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6078803437883694e-05,
+      "loss": 0.2661,
+      "step": 11820
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6075454849871638e-05,
+      "loss": 0.1786,
+      "step": 11830
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6072106261859582e-05,
+      "loss": 0.263,
+      "step": 11840
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.606875767384753e-05,
+      "loss": 0.3034,
+      "step": 11850
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6065409085835474e-05,
+      "loss": 0.2173,
+      "step": 11860
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6062060497823418e-05,
+      "loss": 0.1988,
+      "step": 11870
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6058711909811365e-05,
+      "loss": 0.3455,
+      "step": 11880
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.605536332179931e-05,
+      "loss": 0.2738,
+      "step": 11890
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6052014733787253e-05,
+      "loss": 0.2524,
+      "step": 11900
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6048666145775197e-05,
+      "loss": 0.1966,
+      "step": 11910
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6045317557763145e-05,
+      "loss": 0.2896,
+      "step": 11920
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.604196896975109e-05,
+      "loss": 0.2451,
+      "step": 11930
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6038620381739033e-05,
+      "loss": 0.2504,
+      "step": 11940
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.603527179372698e-05,
+      "loss": 0.314,
+      "step": 11950
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6031923205714925e-05,
+      "loss": 0.3184,
+      "step": 11960
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.602857461770287e-05,
+      "loss": 0.3798,
+      "step": 11970
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6025226029690813e-05,
+      "loss": 0.3485,
+      "step": 11980
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.602187744167876e-05,
+      "loss": 0.1886,
+      "step": 11990
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6018528853666704e-05,
+      "loss": 0.2415,
+      "step": 12000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6015180265654652e-05,
+      "loss": 0.1818,
+      "step": 12010
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6011831677642596e-05,
+      "loss": 0.3149,
+      "step": 12020
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.600848308963054e-05,
+      "loss": 0.185,
+      "step": 12030
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6005134501618484e-05,
+      "loss": 0.1747,
+      "step": 12040
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6001785913606428e-05,
+      "loss": 0.2026,
+      "step": 12050
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.5998437325594375e-05,
+      "loss": 0.2257,
+      "step": 12060
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.599508873758232e-05,
+      "loss": 0.4751,
+      "step": 12070
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.5991740149570267e-05,
+      "loss": 0.2242,
+      "step": 12080
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.598839156155821e-05,
+      "loss": 0.1573,
+      "step": 12090
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.5985042973546155e-05,
+      "loss": 0.4127,
+      "step": 12100
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.59816943855341e-05,
+      "loss": 0.2845,
+      "step": 12110
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.5978345797522043e-05,
+      "loss": 0.219,
+      "step": 12120
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.597499720950999e-05,
+      "loss": 0.3035,
+      "step": 12130
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.5971648621497935e-05,
+      "loss": 0.1814,
+      "step": 12140
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.5968300033485882e-05,
+      "loss": 0.276,
+      "step": 12150
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.5964951445473826e-05,
+      "loss": 0.221,
+      "step": 12160
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.596160285746177e-05,
+      "loss": 0.2125,
+      "step": 12170
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.5958254269449715e-05,
+      "loss": 0.1502,
+      "step": 12180
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.595490568143766e-05,
+      "loss": 0.2054,
+      "step": 12190
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.5951557093425606e-05,
+      "loss": 0.3908,
+      "step": 12200
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.5948208505413554e-05,
+      "loss": 0.2458,
+      "step": 12210
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.5944859917401498e-05,
+      "loss": 0.251,
+      "step": 12220
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.594151132938944e-05,
+      "loss": 0.4146,
+      "step": 12230
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.5938162741377386e-05,
+      "loss": 0.1472,
+      "step": 12240
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.593481415336533e-05,
+      "loss": 0.1655,
+      "step": 12250
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.5931465565353274e-05,
+      "loss": 0.1888,
+      "step": 12260
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.592811697734122e-05,
+      "loss": 0.2946,
+      "step": 12270
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.592476838932917e-05,
+      "loss": 0.2573,
+      "step": 12280
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.5921419801317113e-05,
+      "loss": 0.3862,
+      "step": 12290
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.5918071213305057e-05,
+      "loss": 0.1885,
+      "step": 12300
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.5914722625293e-05,
+      "loss": 0.3278,
+      "step": 12310
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.5911374037280945e-05,
+      "loss": 0.3041,
+      "step": 12320
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.5908025449268893e-05,
+      "loss": 0.2991,
+      "step": 12330
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.590467686125684e-05,
+      "loss": 0.2577,
+      "step": 12340
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.5901328273244784e-05,
+      "loss": 0.279,
+      "step": 12350
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.5897979685232728e-05,
+      "loss": 0.254,
+      "step": 12360
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.5894631097220672e-05,
+      "loss": 0.4532,
+      "step": 12370
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.5891282509208616e-05,
+      "loss": 0.1738,
+      "step": 12380
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.588793392119656e-05,
+      "loss": 0.2788,
+      "step": 12390
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.5884585333184508e-05,
+      "loss": 0.1727,
+      "step": 12400
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.5881236745172455e-05,
+      "loss": 0.2229,
+      "step": 12410
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.58778881571604e-05,
+      "loss": 0.2345,
+      "step": 12420
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.5874539569148343e-05,
+      "loss": 0.32,
+      "step": 12430
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.5871190981136288e-05,
+      "loss": 0.1828,
+      "step": 12440
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.586784239312423e-05,
+      "loss": 0.1644,
+      "step": 12450
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.5864493805112176e-05,
+      "loss": 0.1752,
+      "step": 12460
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.5861145217100127e-05,
+      "loss": 0.2824,
+      "step": 12470
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.585779662908807e-05,
+      "loss": 0.466,
+      "step": 12480
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.5854448041076015e-05,
+      "loss": 0.2194,
+      "step": 12490
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.585109945306396e-05,
+      "loss": 0.3545,
+      "step": 12500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.5847750865051903e-05,
+      "loss": 0.2453,
+      "step": 12510
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.5844402277039847e-05,
+      "loss": 0.2388,
+      "step": 12520
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.584105368902779e-05,
+      "loss": 0.2563,
+      "step": 12530
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.5837705101015742e-05,
+      "loss": 0.2445,
+      "step": 12540
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.5834356513003686e-05,
+      "loss": 0.3609,
+      "step": 12550
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.583100792499163e-05,
+      "loss": 0.2291,
+      "step": 12560
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.5827659336979574e-05,
+      "loss": 0.2612,
+      "step": 12570
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.5824310748967518e-05,
+      "loss": 0.3343,
+      "step": 12580
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.5820962160955462e-05,
+      "loss": 0.3587,
+      "step": 12590
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.581761357294341e-05,
+      "loss": 0.4618,
+      "step": 12600
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.5814264984931357e-05,
+      "loss": 0.4394,
+      "step": 12610
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.58109163969193e-05,
+      "loss": 0.3174,
+      "step": 12620
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.5807567808907245e-05,
+      "loss": 0.336,
+      "step": 12630
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.580421922089519e-05,
+      "loss": 0.1491,
+      "step": 12640
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.5800870632883133e-05,
+      "loss": 0.2612,
+      "step": 12650
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.5797522044871078e-05,
+      "loss": 0.2121,
+      "step": 12660
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.5794173456859025e-05,
+      "loss": 0.28,
+      "step": 12670
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.5790824868846972e-05,
+      "loss": 0.2003,
+      "step": 12680
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.5787476280834917e-05,
+      "loss": 0.2147,
+      "step": 12690
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.578412769282286e-05,
+      "loss": 0.2495,
+      "step": 12700
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.5780779104810805e-05,
+      "loss": 0.2867,
+      "step": 12710
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.577743051679875e-05,
+      "loss": 0.3134,
+      "step": 12720
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.5774081928786696e-05,
+      "loss": 0.1671,
+      "step": 12730
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.577073334077464e-05,
+      "loss": 0.2246,
+      "step": 12740
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.5767384752762588e-05,
+      "loss": 0.2452,
+      "step": 12750
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.5764036164750532e-05,
+      "loss": 0.2273,
+      "step": 12760
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.5760687576738476e-05,
+      "loss": 0.3706,
+      "step": 12770
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.575733898872642e-05,
+      "loss": 0.3918,
+      "step": 12780
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.5753990400714364e-05,
+      "loss": 0.293,
+      "step": 12790
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.575064181270231e-05,
+      "loss": 0.2164,
+      "step": 12800
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.5747293224690256e-05,
+      "loss": 0.2064,
+      "step": 12810
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.5743944636678203e-05,
+      "loss": 0.1592,
+      "step": 12820
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.5740596048666147e-05,
+      "loss": 0.2461,
+      "step": 12830
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.573724746065409e-05,
+      "loss": 0.1319,
+      "step": 12840
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.5733898872642035e-05,
+      "loss": 0.1968,
+      "step": 12850
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.5730550284629983e-05,
+      "loss": 0.413,
+      "step": 12860
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.5727201696617927e-05,
+      "loss": 0.2698,
+      "step": 12870
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.572385310860587e-05,
+      "loss": 0.2074,
+      "step": 12880
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.572050452059382e-05,
+      "loss": 0.2177,
+      "step": 12890
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.5717155932581762e-05,
+      "loss": 0.2366,
+      "step": 12900
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.5713807344569706e-05,
+      "loss": 0.165,
+      "step": 12910
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.571045875655765e-05,
+      "loss": 0.3743,
+      "step": 12920
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.5707110168545598e-05,
+      "loss": 0.234,
+      "step": 12930
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.5703761580533542e-05,
+      "loss": 0.2229,
+      "step": 12940
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.5700412992521486e-05,
+      "loss": 0.2773,
+      "step": 12950
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.5697064404509434e-05,
+      "loss": 0.2247,
+      "step": 12960
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5693715816497378e-05,
+      "loss": 0.3557,
+      "step": 12970
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5690367228485322e-05,
+      "loss": 0.3974,
+      "step": 12980
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.568701864047327e-05,
+      "loss": 0.3407,
+      "step": 12990
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5683670052461213e-05,
+      "loss": 0.1948,
+      "step": 13000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5680321464449157e-05,
+      "loss": 0.2094,
+      "step": 13010
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.56769728764371e-05,
+      "loss": 0.3835,
+      "step": 13020
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.567362428842505e-05,
+      "loss": 0.2477,
+      "step": 13030
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5670275700412993e-05,
+      "loss": 0.28,
+      "step": 13040
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5666927112400937e-05,
+      "loss": 0.2252,
+      "step": 13050
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5663578524388885e-05,
+      "loss": 0.328,
+      "step": 13060
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.566022993637683e-05,
+      "loss": 0.3765,
+      "step": 13070
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5656881348364773e-05,
+      "loss": 0.2278,
+      "step": 13080
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5653532760352717e-05,
+      "loss": 0.2739,
+      "step": 13090
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5650184172340664e-05,
+      "loss": 0.1679,
+      "step": 13100
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5646835584328608e-05,
+      "loss": 0.1913,
+      "step": 13110
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5643486996316556e-05,
+      "loss": 0.3977,
+      "step": 13120
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.56401384083045e-05,
+      "loss": 0.3066,
+      "step": 13130
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5636789820292444e-05,
+      "loss": 0.2201,
+      "step": 13140
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5633441232280388e-05,
+      "loss": 0.2209,
+      "step": 13150
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5630092644268332e-05,
+      "loss": 0.3547,
+      "step": 13160
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.562674405625628e-05,
+      "loss": 0.3034,
+      "step": 13170
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5623395468244224e-05,
+      "loss": 0.1941,
+      "step": 13180
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.562004688023217e-05,
+      "loss": 0.1202,
+      "step": 13190
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5616698292220115e-05,
+      "loss": 0.1786,
+      "step": 13200
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.561334970420806e-05,
+      "loss": 0.2904,
+      "step": 13210
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5610001116196003e-05,
+      "loss": 0.2422,
+      "step": 13220
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5606652528183947e-05,
+      "loss": 0.2462,
+      "step": 13230
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5603303940171895e-05,
+      "loss": 0.1978,
+      "step": 13240
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5599955352159842e-05,
+      "loss": 0.1832,
+      "step": 13250
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5596606764147786e-05,
+      "loss": 0.1735,
+      "step": 13260
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.559325817613573e-05,
+      "loss": 0.2967,
+      "step": 13270
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5589909588123674e-05,
+      "loss": 0.1522,
+      "step": 13280
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.558656100011162e-05,
+      "loss": 0.2278,
+      "step": 13290
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5583212412099566e-05,
+      "loss": 0.2027,
+      "step": 13300
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.557986382408751e-05,
+      "loss": 0.3648,
+      "step": 13310
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5576515236075458e-05,
+      "loss": 0.3553,
+      "step": 13320
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.55731666480634e-05,
+      "loss": 0.1655,
+      "step": 13330
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5569818060051346e-05,
+      "loss": 0.2337,
+      "step": 13340
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.556646947203929e-05,
+      "loss": 0.2279,
+      "step": 13350
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5563120884027234e-05,
+      "loss": 0.1711,
+      "step": 13360
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.555977229601518e-05,
+      "loss": 0.3461,
+      "step": 13370
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.555642370800313e-05,
+      "loss": 0.2103,
+      "step": 13380
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5553075119991073e-05,
+      "loss": 0.2172,
+      "step": 13390
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5549726531979017e-05,
+      "loss": 0.3447,
+      "step": 13400
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.554637794396696e-05,
+      "loss": 0.2426,
+      "step": 13410
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5543029355954905e-05,
+      "loss": 0.3156,
+      "step": 13420
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.553968076794285e-05,
+      "loss": 0.2658,
+      "step": 13430
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5536332179930797e-05,
+      "loss": 0.3205,
+      "step": 13440
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5532983591918744e-05,
+      "loss": 0.2989,
+      "step": 13450
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5529635003906688e-05,
+      "loss": 0.2103,
+      "step": 13460
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5526286415894632e-05,
+      "loss": 0.2755,
+      "step": 13470
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5522937827882576e-05,
+      "loss": 0.3563,
+      "step": 13480
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.551958923987052e-05,
+      "loss": 0.314,
+      "step": 13490
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5516240651858464e-05,
+      "loss": 0.1967,
+      "step": 13500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5512892063846415e-05,
+      "loss": 0.3949,
+      "step": 13510
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.550954347583436e-05,
+      "loss": 0.3139,
+      "step": 13520
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5506194887822303e-05,
+      "loss": 0.2026,
+      "step": 13530
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5502846299810248e-05,
+      "loss": 0.2773,
+      "step": 13540
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.549949771179819e-05,
+      "loss": 0.3019,
+      "step": 13550
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5496149123786136e-05,
+      "loss": 0.1785,
+      "step": 13560
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.549280053577408e-05,
+      "loss": 0.4136,
+      "step": 13570
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.548945194776203e-05,
+      "loss": 0.2068,
+      "step": 13580
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5486103359749975e-05,
+      "loss": 0.2242,
+      "step": 13590
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.548275477173792e-05,
+      "loss": 0.2436,
+      "step": 13600
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5479406183725863e-05,
+      "loss": 0.2374,
+      "step": 13610
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5476057595713807e-05,
+      "loss": 0.3677,
+      "step": 13620
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.547270900770175e-05,
+      "loss": 0.2935,
+      "step": 13630
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.54693604196897e-05,
+      "loss": 0.4163,
+      "step": 13640
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5466011831677646e-05,
+      "loss": 0.3176,
+      "step": 13650
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.546266324366559e-05,
+      "loss": 0.5722,
+      "step": 13660
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5459314655653534e-05,
+      "loss": 0.1939,
+      "step": 13670
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5455966067641478e-05,
+      "loss": 0.2586,
+      "step": 13680
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5452617479629422e-05,
+      "loss": 0.2413,
+      "step": 13690
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5449268891617366e-05,
+      "loss": 0.184,
+      "step": 13700
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5445920303605314e-05,
+      "loss": 0.2603,
+      "step": 13710
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.544257171559326e-05,
+      "loss": 0.2747,
+      "step": 13720
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5439223127581205e-05,
+      "loss": 0.4919,
+      "step": 13730
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.543587453956915e-05,
+      "loss": 0.2131,
+      "step": 13740
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5432525951557093e-05,
+      "loss": 0.2699,
+      "step": 13750
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5429177363545037e-05,
+      "loss": 0.2878,
+      "step": 13760
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5425828775532985e-05,
+      "loss": 0.2312,
+      "step": 13770
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.542248018752093e-05,
+      "loss": 0.1322,
+      "step": 13780
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5419131599508876e-05,
+      "loss": 0.1989,
+      "step": 13790
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.541578301149682e-05,
+      "loss": 0.1373,
+      "step": 13800
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5412434423484765e-05,
+      "loss": 0.2084,
+      "step": 13810
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.540908583547271e-05,
+      "loss": 0.355,
+      "step": 13820
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5405737247460653e-05,
+      "loss": 0.4697,
+      "step": 13830
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.54023886594486e-05,
+      "loss": 0.1735,
+      "step": 13840
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5399040071436544e-05,
+      "loss": 0.2753,
+      "step": 13850
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5395691483424492e-05,
+      "loss": 0.2408,
+      "step": 13860
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5392342895412436e-05,
+      "loss": 0.2325,
+      "step": 13870
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.538899430740038e-05,
+      "loss": 0.2957,
+      "step": 13880
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5385645719388324e-05,
+      "loss": 0.2527,
+      "step": 13890
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5382297131376268e-05,
+      "loss": 0.3843,
+      "step": 13900
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5378948543364216e-05,
+      "loss": 0.3347,
+      "step": 13910
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.537559995535216e-05,
+      "loss": 0.2413,
+      "step": 13920
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5372251367340107e-05,
+      "loss": 0.2775,
+      "step": 13930
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.536890277932805e-05,
+      "loss": 0.1577,
+      "step": 13940
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5365554191315995e-05,
+      "loss": 0.202,
+      "step": 13950
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.536220560330394e-05,
+      "loss": 0.2348,
+      "step": 13960
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5358857015291887e-05,
+      "loss": 0.2854,
+      "step": 13970
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.535550842727983e-05,
+      "loss": 0.2324,
+      "step": 13980
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5352159839267775e-05,
+      "loss": 0.2897,
+      "step": 13990
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5348811251255722e-05,
+      "loss": 0.2232,
+      "step": 14000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5345462663243666e-05,
+      "loss": 0.1877,
+      "step": 14010
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.534211407523161e-05,
+      "loss": 0.3429,
+      "step": 14020
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5338765487219555e-05,
+      "loss": 0.2416,
+      "step": 14030
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5335416899207502e-05,
+      "loss": 0.3435,
+      "step": 14040
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5332068311195446e-05,
+      "loss": 0.2503,
+      "step": 14050
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.532871972318339e-05,
+      "loss": 0.2438,
+      "step": 14060
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5325371135171338e-05,
+      "loss": 0.4095,
+      "step": 14070
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5322022547159282e-05,
+      "loss": 0.2068,
+      "step": 14080
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5318673959147226e-05,
+      "loss": 0.1927,
+      "step": 14090
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5315325371135173e-05,
+      "loss": 0.1533,
+      "step": 14100
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5311976783123117e-05,
+      "loss": 0.3457,
+      "step": 14110
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.530862819511106e-05,
+      "loss": 0.374,
+      "step": 14120
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5305279607099005e-05,
+      "loss": 0.2077,
+      "step": 14130
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5301931019086953e-05,
+      "loss": 0.3327,
+      "step": 14140
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5298582431074897e-05,
+      "loss": 0.1877,
+      "step": 14150
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.529523384306284e-05,
+      "loss": 0.2488,
+      "step": 14160
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.529188525505079e-05,
+      "loss": 0.4979,
+      "step": 14170
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5288536667038733e-05,
+      "loss": 0.2071,
+      "step": 14180
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5285188079026677e-05,
+      "loss": 0.204,
+      "step": 14190
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5281839491014624e-05,
+      "loss": 0.1821,
+      "step": 14200
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5278490903002568e-05,
+      "loss": 0.1792,
+      "step": 14210
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5275142314990512e-05,
+      "loss": 0.442,
+      "step": 14220
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.527179372697846e-05,
+      "loss": 0.2312,
+      "step": 14230
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5268445138966404e-05,
+      "loss": 0.3251,
+      "step": 14240
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5265096550954348e-05,
+      "loss": 0.3104,
+      "step": 14250
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5261747962942292e-05,
+      "loss": 0.1941,
+      "step": 14260
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.525839937493024e-05,
+      "loss": 0.4336,
+      "step": 14270
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5255050786918184e-05,
+      "loss": 0.3383,
+      "step": 14280
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5251702198906128e-05,
+      "loss": 0.2085,
+      "step": 14290
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5248353610894075e-05,
+      "loss": 0.1892,
+      "step": 14300
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.524500502288202e-05,
+      "loss": 0.1607,
+      "step": 14310
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5241656434869963e-05,
+      "loss": 0.3224,
+      "step": 14320
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5238307846857907e-05,
+      "loss": 0.2352,
+      "step": 14330
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5234959258845855e-05,
+      "loss": 0.2235,
+      "step": 14340
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.52316106708338e-05,
+      "loss": 0.2282,
+      "step": 14350
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5228262082821746e-05,
+      "loss": 0.2575,
+      "step": 14360
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.522491349480969e-05,
+      "loss": 0.2346,
+      "step": 14370
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5221564906797634e-05,
+      "loss": 0.3224,
+      "step": 14380
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.521821631878558e-05,
+      "loss": 0.2141,
+      "step": 14390
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5214867730773523e-05,
+      "loss": 0.1691,
+      "step": 14400
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.521151914276147e-05,
+      "loss": 0.1564,
+      "step": 14410
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5208170554749414e-05,
+      "loss": 0.3922,
+      "step": 14420
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.520482196673736e-05,
+      "loss": 0.2826,
+      "step": 14430
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5201473378725306e-05,
+      "loss": 0.141,
+      "step": 14440
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.519812479071325e-05,
+      "loss": 0.1718,
+      "step": 14450
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5194776202701194e-05,
+      "loss": 0.2152,
+      "step": 14460
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5191427614689138e-05,
+      "loss": 0.3448,
+      "step": 14470
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5188079026677085e-05,
+      "loss": 0.3827,
+      "step": 14480
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5184730438665033e-05,
+      "loss": 0.2378,
+      "step": 14490
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5181381850652977e-05,
+      "loss": 0.1593,
+      "step": 14500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.517803326264092e-05,
+      "loss": 0.2891,
+      "step": 14510
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5174684674628865e-05,
+      "loss": 0.3391,
+      "step": 14520
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.517133608661681e-05,
+      "loss": 0.1603,
+      "step": 14530
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5167987498604753e-05,
+      "loss": 0.159,
+      "step": 14540
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.51646389105927e-05,
+      "loss": 0.1859,
+      "step": 14550
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5161290322580648e-05,
+      "loss": 0.3369,
+      "step": 14560
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5157941734568592e-05,
+      "loss": 0.3322,
+      "step": 14570
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5154593146556536e-05,
+      "loss": 0.2058,
+      "step": 14580
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.515124455854448e-05,
+      "loss": 0.3472,
+      "step": 14590
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5147895970532424e-05,
+      "loss": 0.2848,
+      "step": 14600
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.514454738252037e-05,
+      "loss": 0.2083,
+      "step": 14610
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.514119879450832e-05,
+      "loss": 0.3546,
+      "step": 14620
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5137850206496263e-05,
+      "loss": 0.2477,
+      "step": 14630
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5134501618484207e-05,
+      "loss": 0.2673,
+      "step": 14640
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.513115303047215e-05,
+      "loss": 0.304,
+      "step": 14650
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5127804442460096e-05,
+      "loss": 0.2229,
+      "step": 14660
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.512445585444804e-05,
+      "loss": 0.5046,
+      "step": 14670
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5121107266435984e-05,
+      "loss": 0.2466,
+      "step": 14680
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5117758678423935e-05,
+      "loss": 0.1877,
+      "step": 14690
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.511441009041188e-05,
+      "loss": 0.2844,
+      "step": 14700
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5111061502399823e-05,
+      "loss": 0.3642,
+      "step": 14710
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5107712914387767e-05,
+      "loss": 0.2625,
+      "step": 14720
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.510436432637571e-05,
+      "loss": 0.2002,
+      "step": 14730
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5101015738363655e-05,
+      "loss": 0.3685,
+      "step": 14740
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5097667150351602e-05,
+      "loss": 0.1458,
+      "step": 14750
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.509431856233955e-05,
+      "loss": 0.1837,
+      "step": 14760
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5090969974327494e-05,
+      "loss": 0.2816,
+      "step": 14770
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5087621386315438e-05,
+      "loss": 0.3334,
+      "step": 14780
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5084272798303382e-05,
+      "loss": 0.1856,
+      "step": 14790
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5080924210291326e-05,
+      "loss": 0.3269,
+      "step": 14800
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.507757562227927e-05,
+      "loss": 0.1666,
+      "step": 14810
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5074227034267218e-05,
+      "loss": 0.1729,
+      "step": 14820
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5070878446255165e-05,
+      "loss": 0.1845,
+      "step": 14830
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.506752985824311e-05,
+      "loss": 0.2421,
+      "step": 14840
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5064181270231053e-05,
+      "loss": 0.1634,
+      "step": 14850
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5060832682218997e-05,
+      "loss": 0.2236,
+      "step": 14860
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.505748409420694e-05,
+      "loss": 0.2967,
+      "step": 14870
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.505413550619489e-05,
+      "loss": 0.2741,
+      "step": 14880
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5050786918182833e-05,
+      "loss": 0.2085,
+      "step": 14890
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.504743833017078e-05,
+      "loss": 0.2012,
+      "step": 14900
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5044089742158725e-05,
+      "loss": 0.2636,
+      "step": 14910
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.504074115414667e-05,
+      "loss": 0.2821,
+      "step": 14920
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5037392566134613e-05,
+      "loss": 0.2653,
+      "step": 14930
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5034043978122557e-05,
+      "loss": 0.1688,
+      "step": 14940
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5030695390110504e-05,
+      "loss": 0.4483,
+      "step": 14950
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.502734680209845e-05,
+      "loss": 0.2806,
+      "step": 14960
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5023998214086396e-05,
+      "loss": 0.3075,
+      "step": 14970
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.502064962607434e-05,
+      "loss": 0.2917,
+      "step": 14980
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5017301038062284e-05,
+      "loss": 0.4458,
+      "step": 14990
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5013952450050228e-05,
+      "loss": 0.2021,
+      "step": 15000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5010603862038175e-05,
+      "loss": 0.3231,
+      "step": 15010
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.500725527402612e-05,
+      "loss": 0.2956,
+      "step": 15020
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5003906686014064e-05,
+      "loss": 0.2345,
+      "step": 15030
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.500055809800201e-05,
+      "loss": 0.1755,
+      "step": 15040
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4997209509989955e-05,
+      "loss": 0.1731,
+      "step": 15050
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.49938609219779e-05,
+      "loss": 0.216,
+      "step": 15060
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4990512333965843e-05,
+      "loss": 0.1838,
+      "step": 15070
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.498716374595379e-05,
+      "loss": 0.2436,
+      "step": 15080
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4983815157941735e-05,
+      "loss": 0.2266,
+      "step": 15090
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4980466569929682e-05,
+      "loss": 0.168,
+      "step": 15100
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4977117981917626e-05,
+      "loss": 0.2028,
+      "step": 15110
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.497376939390557e-05,
+      "loss": 0.1997,
+      "step": 15120
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4970420805893515e-05,
+      "loss": 0.2185,
+      "step": 15130
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4967072217881462e-05,
+      "loss": 0.2154,
+      "step": 15140
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4963723629869406e-05,
+      "loss": 0.3875,
+      "step": 15150
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.496037504185735e-05,
+      "loss": 0.1937,
+      "step": 15160
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4957026453845298e-05,
+      "loss": 0.3061,
+      "step": 15170
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.495367786583324e-05,
+      "loss": 0.2788,
+      "step": 15180
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4950329277821186e-05,
+      "loss": 0.1764,
+      "step": 15190
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.494698068980913e-05,
+      "loss": 0.149,
+      "step": 15200
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4943632101797077e-05,
+      "loss": 0.2655,
+      "step": 15210
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.494028351378502e-05,
+      "loss": 0.4143,
+      "step": 15220
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4936934925772965e-05,
+      "loss": 0.2132,
+      "step": 15230
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4933586337760913e-05,
+      "loss": 0.7279,
+      "step": 15240
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4930237749748857e-05,
+      "loss": 0.2577,
+      "step": 15250
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.49268891617368e-05,
+      "loss": 0.2089,
+      "step": 15260
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.492354057372475e-05,
+      "loss": 0.2176,
+      "step": 15270
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4920191985712693e-05,
+      "loss": 0.2521,
+      "step": 15280
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4916843397700637e-05,
+      "loss": 0.2438,
+      "step": 15290
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4913829668489787e-05,
+      "loss": 0.4015,
+      "step": 15300
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.491048108047773e-05,
+      "loss": 0.2749,
+      "step": 15310
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.490713249246568e-05,
+      "loss": 0.7336,
+      "step": 15320
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4903783904453623e-05,
+      "loss": 0.2584,
+      "step": 15330
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4900435316441567e-05,
+      "loss": 0.1873,
+      "step": 15340
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4897086728429514e-05,
+      "loss": 0.179,
+      "step": 15350
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.489373814041746e-05,
+      "loss": 0.157,
+      "step": 15360
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4890389552405403e-05,
+      "loss": 0.255,
+      "step": 15370
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4887040964393347e-05,
+      "loss": 0.1729,
+      "step": 15380
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4883692376381294e-05,
+      "loss": 0.1955,
+      "step": 15390
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4880343788369238e-05,
+      "loss": 0.2497,
+      "step": 15400
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4876995200357182e-05,
+      "loss": 0.2014,
+      "step": 15410
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.487364661234513e-05,
+      "loss": 0.2888,
+      "step": 15420
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4870298024333074e-05,
+      "loss": 0.2856,
+      "step": 15430
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4866949436321018e-05,
+      "loss": 0.2674,
+      "step": 15440
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4863600848308965e-05,
+      "loss": 0.4767,
+      "step": 15450
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.486025226029691e-05,
+      "loss": 0.1534,
+      "step": 15460
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4856903672284854e-05,
+      "loss": 0.5782,
+      "step": 15470
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4853555084272798e-05,
+      "loss": 0.2876,
+      "step": 15480
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4850206496260745e-05,
+      "loss": 0.2744,
+      "step": 15490
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.484685790824869e-05,
+      "loss": 0.1906,
+      "step": 15500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4843509320236633e-05,
+      "loss": 0.229,
+      "step": 15510
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.484016073222458e-05,
+      "loss": 0.2859,
+      "step": 15520
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4836812144212525e-05,
+      "loss": 0.181,
+      "step": 15530
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.483346355620047e-05,
+      "loss": 0.2428,
+      "step": 15540
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4830114968188413e-05,
+      "loss": 0.2265,
+      "step": 15550
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.482676638017636e-05,
+      "loss": 0.2244,
+      "step": 15560
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4823417792164304e-05,
+      "loss": 0.242,
+      "step": 15570
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4820069204152252e-05,
+      "loss": 0.3,
+      "step": 15580
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4816720616140196e-05,
+      "loss": 0.1326,
+      "step": 15590
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.481337202812814e-05,
+      "loss": 0.4208,
+      "step": 15600
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4810023440116084e-05,
+      "loss": 0.2496,
+      "step": 15610
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4806674852104028e-05,
+      "loss": 0.2005,
+      "step": 15620
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4803326264091976e-05,
+      "loss": 0.3638,
+      "step": 15630
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.479997767607992e-05,
+      "loss": 0.1845,
+      "step": 15640
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4796629088067867e-05,
+      "loss": 0.2173,
+      "step": 15650
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.479328050005581e-05,
+      "loss": 0.2781,
+      "step": 15660
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4789931912043755e-05,
+      "loss": 0.3278,
+      "step": 15670
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.47865833240317e-05,
+      "loss": 0.2758,
+      "step": 15680
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4783234736019643e-05,
+      "loss": 0.3226,
+      "step": 15690
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.477988614800759e-05,
+      "loss": 0.1695,
+      "step": 15700
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.477653755999554e-05,
+      "loss": 0.2082,
+      "step": 15710
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4773188971983482e-05,
+      "loss": 0.2451,
+      "step": 15720
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4769840383971427e-05,
+      "loss": 0.2303,
+      "step": 15730
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.476649179595937e-05,
+      "loss": 0.1262,
+      "step": 15740
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4763143207947315e-05,
+      "loss": 0.1567,
+      "step": 15750
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.475979461993526e-05,
+      "loss": 0.3801,
+      "step": 15760
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4756446031923206e-05,
+      "loss": 0.2759,
+      "step": 15770
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4753097443911154e-05,
+      "loss": 0.1889,
+      "step": 15780
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4749748855899098e-05,
+      "loss": 0.2737,
+      "step": 15790
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4746400267887042e-05,
+      "loss": 0.3452,
+      "step": 15800
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4743051679874986e-05,
+      "loss": 0.2146,
+      "step": 15810
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.473970309186293e-05,
+      "loss": 0.2117,
+      "step": 15820
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4736354503850874e-05,
+      "loss": 0.2001,
+      "step": 15830
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4733005915838825e-05,
+      "loss": 0.3207,
+      "step": 15840
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.472965732782677e-05,
+      "loss": 0.2375,
+      "step": 15850
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4726308739814713e-05,
+      "loss": 0.2296,
+      "step": 15860
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4722960151802657e-05,
+      "loss": 0.3185,
+      "step": 15870
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.47196115637906e-05,
+      "loss": 0.1913,
+      "step": 15880
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4716262975778545e-05,
+      "loss": 0.447,
+      "step": 15890
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.471291438776649e-05,
+      "loss": 0.2822,
+      "step": 15900
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.470956579975444e-05,
+      "loss": 0.175,
+      "step": 15910
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4706217211742384e-05,
+      "loss": 0.3266,
+      "step": 15920
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.470286862373033e-05,
+      "loss": 0.2662,
+      "step": 15930
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4699520035718272e-05,
+      "loss": 0.1479,
+      "step": 15940
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4696171447706217e-05,
+      "loss": 0.2155,
+      "step": 15950
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.469282285969416e-05,
+      "loss": 0.2171,
+      "step": 15960
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4689474271682108e-05,
+      "loss": 0.2291,
+      "step": 15970
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4686125683670056e-05,
+      "loss": 0.1864,
+      "step": 15980
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4682777095658e-05,
+      "loss": 0.4868,
+      "step": 15990
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4679428507645944e-05,
+      "loss": 0.1781,
+      "step": 16000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4676079919633888e-05,
+      "loss": 0.3644,
+      "step": 16010
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4672731331621832e-05,
+      "loss": 0.2667,
+      "step": 16020
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4669382743609776e-05,
+      "loss": 0.243,
+      "step": 16030
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4666034155597723e-05,
+      "loss": 0.2361,
+      "step": 16040
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.466268556758567e-05,
+      "loss": 0.27,
+      "step": 16050
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4659336979573615e-05,
+      "loss": 0.3025,
+      "step": 16060
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.465598839156156e-05,
+      "loss": 0.2635,
+      "step": 16070
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4652639803549503e-05,
+      "loss": 0.2561,
+      "step": 16080
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4649291215537447e-05,
+      "loss": 0.2705,
+      "step": 16090
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4645942627525395e-05,
+      "loss": 0.205,
+      "step": 16100
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.464259403951334e-05,
+      "loss": 0.1893,
+      "step": 16110
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4639245451501286e-05,
+      "loss": 0.237,
+      "step": 16120
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.463589686348923e-05,
+      "loss": 0.2394,
+      "step": 16130
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4632548275477174e-05,
+      "loss": 0.1805,
+      "step": 16140
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.462919968746512e-05,
+      "loss": 0.3416,
+      "step": 16150
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4625851099453062e-05,
+      "loss": 0.2727,
+      "step": 16160
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.462250251144101e-05,
+      "loss": 0.1805,
+      "step": 16170
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4619153923428954e-05,
+      "loss": 0.223,
+      "step": 16180
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.46158053354169e-05,
+      "loss": 0.2032,
+      "step": 16190
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4612456747404845e-05,
+      "loss": 0.2161,
+      "step": 16200
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.460910815939279e-05,
+      "loss": 0.1455,
+      "step": 16210
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4605759571380734e-05,
+      "loss": 0.2787,
+      "step": 16220
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.460241098336868e-05,
+      "loss": 0.1933,
+      "step": 16230
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4599062395356625e-05,
+      "loss": 0.2542,
+      "step": 16240
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4595713807344573e-05,
+      "loss": 0.3503,
+      "step": 16250
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4592365219332517e-05,
+      "loss": 0.2565,
+      "step": 16260
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.458901663132046e-05,
+      "loss": 0.2761,
+      "step": 16270
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4585668043308405e-05,
+      "loss": 0.26,
+      "step": 16280
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.458231945529635e-05,
+      "loss": 0.334,
+      "step": 16290
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4578970867284296e-05,
+      "loss": 0.143,
+      "step": 16300
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.457562227927224e-05,
+      "loss": 0.1918,
+      "step": 16310
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4572273691260188e-05,
+      "loss": 0.3194,
+      "step": 16320
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4568925103248132e-05,
+      "loss": 0.2377,
+      "step": 16330
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4565576515236076e-05,
+      "loss": 0.1804,
+      "step": 16340
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.456222792722402e-05,
+      "loss": 0.287,
+      "step": 16350
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4558879339211968e-05,
+      "loss": 0.1799,
+      "step": 16360
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.455553075119991e-05,
+      "loss": 0.2733,
+      "step": 16370
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4552182163187856e-05,
+      "loss": 0.1816,
+      "step": 16380
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4548833575175803e-05,
+      "loss": 0.2944,
+      "step": 16390
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4545484987163747e-05,
+      "loss": 0.2332,
+      "step": 16400
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.454213639915169e-05,
+      "loss": 0.2766,
+      "step": 16410
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4538787811139635e-05,
+      "loss": 0.3646,
+      "step": 16420
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4535439223127583e-05,
+      "loss": 0.1868,
+      "step": 16430
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4532090635115527e-05,
+      "loss": 0.2246,
+      "step": 16440
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.452874204710347e-05,
+      "loss": 0.1876,
+      "step": 16450
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.452539345909142e-05,
+      "loss": 0.1625,
+      "step": 16460
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4522044871079363e-05,
+      "loss": 0.2588,
+      "step": 16470
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4518696283067307e-05,
+      "loss": 0.2766,
+      "step": 16480
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.451534769505525e-05,
+      "loss": 0.2731,
+      "step": 16490
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4511999107043198e-05,
+      "loss": 0.2085,
+      "step": 16500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4508650519031142e-05,
+      "loss": 0.137,
+      "step": 16510
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4505301931019086e-05,
+      "loss": 0.4952,
+      "step": 16520
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4501953343007034e-05,
+      "loss": 0.1352,
+      "step": 16530
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4498604754994978e-05,
+      "loss": 0.2092,
+      "step": 16540
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.4495256166982922e-05,
+      "loss": 0.2666,
+      "step": 16550
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.449190757897087e-05,
+      "loss": 0.2365,
+      "step": 16560
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.4488558990958813e-05,
+      "loss": 0.2749,
+      "step": 16570
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.4485210402946758e-05,
+      "loss": 0.1916,
+      "step": 16580
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.44818618149347e-05,
+      "loss": 0.2438,
+      "step": 16590
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.447851322692265e-05,
+      "loss": 0.171,
+      "step": 16600
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.4475164638910593e-05,
+      "loss": 0.1846,
+      "step": 16610
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.4471816050898537e-05,
+      "loss": 0.2638,
+      "step": 16620
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.4468467462886485e-05,
+      "loss": 0.2249,
+      "step": 16630
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.446511887487443e-05,
+      "loss": 0.1287,
+      "step": 16640
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4461770286862373e-05,
+      "loss": 0.1301,
+      "step": 16650
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4458421698850317e-05,
+      "loss": 0.2109,
+      "step": 16660
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4455073110838264e-05,
+      "loss": 0.3366,
+      "step": 16670
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.445172452282621e-05,
+      "loss": 0.2392,
+      "step": 16680
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4448375934814156e-05,
+      "loss": 0.2276,
+      "step": 16690
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.44450273468021e-05,
+      "loss": 0.2051,
+      "step": 16700
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4441678758790044e-05,
+      "loss": 0.2742,
+      "step": 16710
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4438330170777988e-05,
+      "loss": 0.5131,
+      "step": 16720
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.4434981582765932e-05,
+      "loss": 0.4483,
+      "step": 16730
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.443163299475388e-05,
+      "loss": 0.2586,
+      "step": 16740
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.4428284406741824e-05,
+      "loss": 0.2529,
+      "step": 16750
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.442493581872977e-05,
+      "loss": 0.2896,
+      "step": 16760
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.4421587230717715e-05,
+      "loss": 0.2247,
+      "step": 16770
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.441823864270566e-05,
+      "loss": 0.2416,
+      "step": 16780
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.4414890054693603e-05,
+      "loss": 0.287,
+      "step": 16790
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.4411541466681548e-05,
+      "loss": 0.2538,
+      "step": 16800
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.4408192878669495e-05,
+      "loss": 0.1377,
+      "step": 16810
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.4404844290657442e-05,
+      "loss": 0.3045,
+      "step": 16820
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.4401495702645387e-05,
+      "loss": 0.2458,
+      "step": 16830
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.439814711463333e-05,
+      "loss": 0.1462,
+      "step": 16840
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.4394798526621275e-05,
+      "loss": 0.2269,
+      "step": 16850
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.439144993860922e-05,
+      "loss": 0.2,
+      "step": 16860
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.4388101350597163e-05,
+      "loss": 0.3443,
+      "step": 16870
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.438475276258511e-05,
+      "loss": 0.2049,
+      "step": 16880
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.4381404174573058e-05,
+      "loss": 0.1843,
+      "step": 16890
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.4378055586561002e-05,
+      "loss": 0.1986,
+      "step": 16900
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.4374706998548946e-05,
+      "loss": 0.301,
+      "step": 16910
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.437135841053689e-05,
+      "loss": 0.2527,
+      "step": 16920
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.4368009822524834e-05,
+      "loss": 0.2295,
+      "step": 16930
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.4364661234512778e-05,
+      "loss": 0.2075,
+      "step": 16940
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.436131264650073e-05,
+      "loss": 0.1578,
+      "step": 16950
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.4357964058488673e-05,
+      "loss": 0.4099,
+      "step": 16960
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.4354615470476617e-05,
+      "loss": 0.3893,
+      "step": 16970
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.435126688246456e-05,
+      "loss": 0.1684,
+      "step": 16980
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.4347918294452505e-05,
+      "loss": 0.1464,
+      "step": 16990
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.434456970644045e-05,
+      "loss": 0.1472,
+      "step": 17000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.4341221118428393e-05,
+      "loss": 0.2242,
+      "step": 17010
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.4337872530416344e-05,
+      "loss": 0.3459,
+      "step": 17020
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.433452394240429e-05,
+      "loss": 0.2472,
+      "step": 17030
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.4331175354392232e-05,
+      "loss": 0.1551,
+      "step": 17040
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.4327826766380176e-05,
+      "loss": 0.2506,
+      "step": 17050
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.432447817836812e-05,
+      "loss": 0.2476,
+      "step": 17060
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.4321129590356065e-05,
+      "loss": 0.3336,
+      "step": 17070
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.4317781002344012e-05,
+      "loss": 0.3019,
+      "step": 17080
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.431443241433196e-05,
+      "loss": 0.1988,
+      "step": 17090
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.4311083826319904e-05,
+      "loss": 0.2793,
+      "step": 17100
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.4307735238307848e-05,
+      "loss": 0.2387,
+      "step": 17110
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.4304386650295792e-05,
+      "loss": 0.4961,
+      "step": 17120
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.4301038062283736e-05,
+      "loss": 0.2364,
+      "step": 17130
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.429768947427168e-05,
+      "loss": 0.2811,
+      "step": 17140
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.429434088625963e-05,
+      "loss": 0.1455,
+      "step": 17150
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.4290992298247575e-05,
+      "loss": 0.2782,
+      "step": 17160
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.428764371023552e-05,
+      "loss": 0.2569,
+      "step": 17170
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.4284295122223463e-05,
+      "loss": 0.2429,
+      "step": 17180
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.4280946534211407e-05,
+      "loss": 0.3011,
+      "step": 17190
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.427759794619935e-05,
+      "loss": 0.1915,
+      "step": 17200
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.42742493581873e-05,
+      "loss": 0.1497,
+      "step": 17210
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.4270900770175246e-05,
+      "loss": 0.2915,
+      "step": 17220
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.426755218216319e-05,
+      "loss": 0.1494,
+      "step": 17230
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.4264203594151134e-05,
+      "loss": 0.3504,
+      "step": 17240
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.4260855006139078e-05,
+      "loss": 0.2922,
+      "step": 17250
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.4257506418127022e-05,
+      "loss": 0.2152,
+      "step": 17260
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.4254157830114966e-05,
+      "loss": 0.2229,
+      "step": 17270
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.4250809242102914e-05,
+      "loss": 0.16,
+      "step": 17280
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.424746065409086e-05,
+      "loss": 0.1766,
+      "step": 17290
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.4244112066078805e-05,
+      "loss": 0.3227,
+      "step": 17300
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.424076347806675e-05,
+      "loss": 0.2095,
+      "step": 17310
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.4237414890054694e-05,
+      "loss": 0.2119,
+      "step": 17320
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.4234066302042638e-05,
+      "loss": 0.2383,
+      "step": 17330
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.4230717714030585e-05,
+      "loss": 0.2588,
+      "step": 17340
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.422736912601853e-05,
+      "loss": 0.2945,
+      "step": 17350
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.4224020538006477e-05,
+      "loss": 0.2466,
+      "step": 17360
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.422067194999442e-05,
+      "loss": 0.1611,
+      "step": 17370
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.4217323361982365e-05,
+      "loss": 0.247,
+      "step": 17380
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.421397477397031e-05,
+      "loss": 0.168,
+      "step": 17390
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.4210626185958253e-05,
+      "loss": 0.2337,
+      "step": 17400
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.42072775979462e-05,
+      "loss": 0.1627,
+      "step": 17410
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.4203929009934144e-05,
+      "loss": 0.2245,
+      "step": 17420
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.4200580421922092e-05,
+      "loss": 0.152,
+      "step": 17430
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.4197231833910036e-05,
+      "loss": 0.2017,
+      "step": 17440
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.419388324589798e-05,
+      "loss": 0.2105,
+      "step": 17450
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.4190534657885924e-05,
+      "loss": 0.2239,
+      "step": 17460
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.418718606987387e-05,
+      "loss": 0.2815,
+      "step": 17470
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.4183837481861816e-05,
+      "loss": 0.1997,
+      "step": 17480
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.418048889384976e-05,
+      "loss": 0.2331,
+      "step": 17490
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.4177140305837707e-05,
+      "loss": 0.1913,
+      "step": 17500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.417379171782565e-05,
+      "loss": 0.1674,
+      "step": 17510
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.4170443129813595e-05,
+      "loss": 0.2552,
+      "step": 17520
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.416709454180154e-05,
+      "loss": 0.2199,
+      "step": 17530
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4163745953789487e-05,
+      "loss": 0.2005,
+      "step": 17540
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.416039736577743e-05,
+      "loss": 0.1288,
+      "step": 17550
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4157048777765375e-05,
+      "loss": 0.2139,
+      "step": 17560
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4153700189753323e-05,
+      "loss": 0.3368,
+      "step": 17570
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4150351601741267e-05,
+      "loss": 0.2506,
+      "step": 17580
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.414700301372921e-05,
+      "loss": 0.364,
+      "step": 17590
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4143654425717158e-05,
+      "loss": 0.2019,
+      "step": 17600
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4140305837705102e-05,
+      "loss": 0.2541,
+      "step": 17610
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4136957249693046e-05,
+      "loss": 0.2068,
+      "step": 17620
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.413360866168099e-05,
+      "loss": 0.2912,
+      "step": 17630
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.4130260073668938e-05,
+      "loss": 0.2478,
+      "step": 17640
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.4126911485656882e-05,
+      "loss": 0.3966,
+      "step": 17650
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.4123562897644826e-05,
+      "loss": 0.1966,
+      "step": 17660
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.4120214309632773e-05,
+      "loss": 0.2986,
+      "step": 17670
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.4116865721620718e-05,
+      "loss": 0.2103,
+      "step": 17680
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.411351713360866e-05,
+      "loss": 0.2506,
+      "step": 17690
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.4110168545596606e-05,
+      "loss": 0.2546,
+      "step": 17700
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.4106819957584553e-05,
+      "loss": 0.1667,
+      "step": 17710
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.4103471369572497e-05,
+      "loss": 0.3752,
+      "step": 17720
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.4100122781560445e-05,
+      "loss": 0.2648,
+      "step": 17730
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.409677419354839e-05,
+      "loss": 0.1921,
+      "step": 17740
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.4093425605536333e-05,
+      "loss": 0.442,
+      "step": 17750
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.4090077017524277e-05,
+      "loss": 0.2634,
+      "step": 17760
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.408672842951222e-05,
+      "loss": 0.3248,
+      "step": 17770
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.408337984150017e-05,
+      "loss": 0.2474,
+      "step": 17780
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.4080031253488113e-05,
+      "loss": 0.142,
+      "step": 17790
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.407668266547606e-05,
+      "loss": 0.219,
+      "step": 17800
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.4073334077464004e-05,
+      "loss": 0.3037,
+      "step": 17810
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.4069985489451948e-05,
+      "loss": 0.185,
+      "step": 17820
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.4066636901439892e-05,
+      "loss": 0.186,
+      "step": 17830
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.4063288313427836e-05,
+      "loss": 0.2247,
+      "step": 17840
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.4059939725415784e-05,
+      "loss": 0.2999,
+      "step": 17850
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.405659113740373e-05,
+      "loss": 0.1851,
+      "step": 17860
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.4053242549391675e-05,
+      "loss": 0.283,
+      "step": 17870
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.404989396137962e-05,
+      "loss": 0.2374,
+      "step": 17880
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.4046545373367563e-05,
+      "loss": 0.228,
+      "step": 17890
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.4043196785355507e-05,
+      "loss": 0.1669,
+      "step": 17900
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.403984819734345e-05,
+      "loss": 0.2723,
+      "step": 17910
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.40364996093314e-05,
+      "loss": 0.2295,
+      "step": 17920
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.4033151021319346e-05,
+      "loss": 0.2032,
+      "step": 17930
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 0.3169889748096466,
+      "eval_runtime": 252.8555,
+      "eval_samples_per_second": 7.886,
+      "eval_steps_per_second": 3.943,
+      "eval_wer": 0.1661755323996719,
+      "step": 17938
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.402980243330729e-05,
+      "loss": 0.3903,
+      "step": 17940
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.4026453845295235e-05,
+      "loss": 0.1388,
+      "step": 17950
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.402310525728318e-05,
+      "loss": 0.1692,
+      "step": 17960
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.4019756669271123e-05,
+      "loss": 0.2189,
+      "step": 17970
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.4016408081259067e-05,
+      "loss": 0.4362,
+      "step": 17980
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.401339435204822e-05,
+      "loss": 0.2512,
+      "step": 17990
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.4010045764036165e-05,
+      "loss": 0.1432,
+      "step": 18000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.400669717602411e-05,
+      "loss": 0.2571,
+      "step": 18010
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.4003348588012053e-05,
+      "loss": 0.1507,
+      "step": 18020
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.4e-05,
+      "loss": 0.2606,
+      "step": 18030
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.3996651411987948e-05,
+      "loss": 0.1356,
+      "step": 18040
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.3993302823975892e-05,
+      "loss": 0.1232,
+      "step": 18050
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.3989954235963836e-05,
+      "loss": 0.2641,
+      "step": 18060
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.398660564795178e-05,
+      "loss": 0.231,
+      "step": 18070
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.3983257059939724e-05,
+      "loss": 0.1476,
+      "step": 18080
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.397990847192767e-05,
+      "loss": 0.2333,
+      "step": 18090
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.3976559883915616e-05,
+      "loss": 0.216,
+      "step": 18100
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.3973211295903563e-05,
+      "loss": 0.1557,
+      "step": 18110
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.3969862707891507e-05,
+      "loss": 0.2321,
+      "step": 18120
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.396651411987945e-05,
+      "loss": 0.1303,
+      "step": 18130
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.3963165531867396e-05,
+      "loss": 0.2607,
+      "step": 18140
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.395981694385534e-05,
+      "loss": 0.1706,
+      "step": 18150
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.3956468355843284e-05,
+      "loss": 0.193,
+      "step": 18160
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.3953119767831235e-05,
+      "loss": 0.1184,
+      "step": 18170
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.394977117981918e-05,
+      "loss": 0.1214,
+      "step": 18180
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.3946422591807123e-05,
+      "loss": 0.2365,
+      "step": 18190
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.3943074003795067e-05,
+      "loss": 0.2137,
+      "step": 18200
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.393972541578301e-05,
+      "loss": 0.1962,
+      "step": 18210
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.3936376827770955e-05,
+      "loss": 0.2192,
+      "step": 18220
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.39330282397589e-05,
+      "loss": 0.1621,
+      "step": 18230
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.392967965174685e-05,
+      "loss": 0.2191,
+      "step": 18240
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.3926331063734794e-05,
+      "loss": 0.1598,
+      "step": 18250
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.3922982475722738e-05,
+      "loss": 0.131,
+      "step": 18260
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.3919633887710682e-05,
+      "loss": 0.182,
+      "step": 18270
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.3916285299698626e-05,
+      "loss": 0.1546,
+      "step": 18280
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.391293671168657e-05,
+      "loss": 0.48,
+      "step": 18290
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.3909588123674518e-05,
+      "loss": 0.1908,
+      "step": 18300
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.3906239535662465e-05,
+      "loss": 0.196,
+      "step": 18310
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.390289094765041e-05,
+      "loss": 0.1403,
+      "step": 18320
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.3899542359638353e-05,
+      "loss": 0.1773,
+      "step": 18330
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.3896193771626297e-05,
+      "loss": 0.3454,
+      "step": 18340
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.389284518361424e-05,
+      "loss": 0.3671,
+      "step": 18350
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.3889496595602186e-05,
+      "loss": 0.2724,
+      "step": 18360
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.3886148007590136e-05,
+      "loss": 0.2325,
+      "step": 18370
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.388279941957808e-05,
+      "loss": 0.1505,
+      "step": 18380
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.3879450831566025e-05,
+      "loss": 0.2328,
+      "step": 18390
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.387610224355397e-05,
+      "loss": 0.2469,
+      "step": 18400
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.3872753655541913e-05,
+      "loss": 0.241,
+      "step": 18410
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.3869405067529857e-05,
+      "loss": 0.4174,
+      "step": 18420
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.3866056479517804e-05,
+      "loss": 0.2046,
+      "step": 18430
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.386270789150575e-05,
+      "loss": 0.2187,
+      "step": 18440
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.3859359303493696e-05,
+      "loss": 0.1694,
+      "step": 18450
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.385601071548164e-05,
+      "loss": 0.2542,
+      "step": 18460
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.3852662127469584e-05,
+      "loss": 0.2458,
+      "step": 18470
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.3849313539457528e-05,
+      "loss": 0.1263,
+      "step": 18480
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.3845964951445472e-05,
+      "loss": 0.1817,
+      "step": 18490
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.384261636343342e-05,
+      "loss": 0.0976,
+      "step": 18500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.3839267775421367e-05,
+      "loss": 0.2477,
+      "step": 18510
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.383591918740931e-05,
+      "loss": 0.1134,
+      "step": 18520
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.3832570599397255e-05,
+      "loss": 0.1365,
+      "step": 18530
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.38292220113852e-05,
+      "loss": 0.3007,
+      "step": 18540
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.3825873423373143e-05,
+      "loss": 0.1396,
+      "step": 18550
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.382252483536109e-05,
+      "loss": 0.2177,
+      "step": 18560
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.3819176247349035e-05,
+      "loss": 0.1354,
+      "step": 18570
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.3815827659336982e-05,
+      "loss": 0.2465,
+      "step": 18580
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.3812479071324926e-05,
+      "loss": 0.2535,
+      "step": 18590
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.380913048331287e-05,
+      "loss": 0.2804,
+      "step": 18600
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.3805781895300815e-05,
+      "loss": 0.2155,
+      "step": 18610
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.380243330728876e-05,
+      "loss": 0.1384,
+      "step": 18620
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.3799084719276706e-05,
+      "loss": 0.1374,
+      "step": 18630
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.379573613126465e-05,
+      "loss": 0.3652,
+      "step": 18640
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.3792387543252598e-05,
+      "loss": 0.222,
+      "step": 18650
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.378903895524054e-05,
+      "loss": 0.1886,
+      "step": 18660
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.3785690367228486e-05,
+      "loss": 0.2145,
+      "step": 18670
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.378234177921643e-05,
+      "loss": 0.2054,
+      "step": 18680
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.3778993191204377e-05,
+      "loss": 0.1538,
+      "step": 18690
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.377564460319232e-05,
+      "loss": 0.103,
+      "step": 18700
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.3772296015180265e-05,
+      "loss": 0.3087,
+      "step": 18710
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.3768947427168213e-05,
+      "loss": 0.389,
+      "step": 18720
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.3765598839156157e-05,
+      "loss": 0.1537,
+      "step": 18730
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.37622502511441e-05,
+      "loss": 0.1066,
+      "step": 18740
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.3758901663132045e-05,
+      "loss": 0.2147,
+      "step": 18750
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.3755553075119993e-05,
+      "loss": 0.2294,
+      "step": 18760
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.3752204487107937e-05,
+      "loss": 0.2606,
+      "step": 18770
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.374885589909588e-05,
+      "loss": 0.3001,
+      "step": 18780
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.3745507311083828e-05,
+      "loss": 0.231,
+      "step": 18790
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.3742158723071772e-05,
+      "loss": 0.2298,
+      "step": 18800
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.3738810135059716e-05,
+      "loss": 0.1683,
+      "step": 18810
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.3735461547047664e-05,
+      "loss": 0.1988,
+      "step": 18820
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.3732112959035608e-05,
+      "loss": 0.1596,
+      "step": 18830
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.3728764371023552e-05,
+      "loss": 0.3044,
+      "step": 18840
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.3725415783011496e-05,
+      "loss": 0.2523,
+      "step": 18850
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.3722067194999443e-05,
+      "loss": 0.1904,
+      "step": 18860
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.3718718606987388e-05,
+      "loss": 0.305,
+      "step": 18870
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.371537001897533e-05,
+      "loss": 0.1059,
+      "step": 18880
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.371202143096328e-05,
+      "loss": 0.196,
+      "step": 18890
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.3708672842951223e-05,
+      "loss": 0.1633,
+      "step": 18900
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.3705324254939167e-05,
+      "loss": 0.136,
+      "step": 18910
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.370197566692711e-05,
+      "loss": 0.2179,
+      "step": 18920
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.369862707891506e-05,
+      "loss": 0.2588,
+      "step": 18930
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.3695278490903003e-05,
+      "loss": 0.4989,
+      "step": 18940
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.369192990289095e-05,
+      "loss": 0.1846,
+      "step": 18950
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.3688581314878894e-05,
+      "loss": 0.1443,
+      "step": 18960
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.368523272686684e-05,
+      "loss": 0.1263,
+      "step": 18970
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.3681884138854783e-05,
+      "loss": 0.1341,
+      "step": 18980
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.3678535550842727e-05,
+      "loss": 0.2679,
+      "step": 18990
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.3675186962830674e-05,
+      "loss": 0.2571,
+      "step": 19000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.3671838374818618e-05,
+      "loss": 0.1547,
+      "step": 19010
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.3668489786806566e-05,
+      "loss": 0.1412,
+      "step": 19020
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.366514119879451e-05,
+      "loss": 0.2035,
+      "step": 19030
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.3661792610782454e-05,
+      "loss": 0.1474,
+      "step": 19040
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.3658444022770398e-05,
+      "loss": 0.1372,
+      "step": 19050
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.3655095434758342e-05,
+      "loss": 0.1983,
+      "step": 19060
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.365174684674629e-05,
+      "loss": 0.1693,
+      "step": 19070
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.3648398258734233e-05,
+      "loss": 0.2607,
+      "step": 19080
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.364504967072218e-05,
+      "loss": 0.1976,
+      "step": 19090
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.3641701082710125e-05,
+      "loss": 0.1617,
+      "step": 19100
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.363835249469807e-05,
+      "loss": 0.2079,
+      "step": 19110
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.3635003906686013e-05,
+      "loss": 0.2167,
+      "step": 19120
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.3631655318673957e-05,
+      "loss": 0.1059,
+      "step": 19130
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.3628306730661905e-05,
+      "loss": 0.3185,
+      "step": 19140
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.3624958142649852e-05,
+      "loss": 0.1623,
+      "step": 19150
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.3621609554637796e-05,
+      "loss": 0.1744,
+      "step": 19160
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.361826096662574e-05,
+      "loss": 0.0923,
+      "step": 19170
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.3614912378613684e-05,
+      "loss": 0.172,
+      "step": 19180
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.361156379060163e-05,
+      "loss": 0.2785,
+      "step": 19190
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.3608215202589572e-05,
+      "loss": 0.1711,
+      "step": 19200
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.360486661457752e-05,
+      "loss": 0.1397,
+      "step": 19210
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.3601518026565467e-05,
+      "loss": 0.2273,
+      "step": 19220
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.359816943855341e-05,
+      "loss": 0.1407,
+      "step": 19230
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.3594820850541356e-05,
+      "loss": 0.21,
+      "step": 19240
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.35914722625293e-05,
+      "loss": 0.103,
+      "step": 19250
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.3588123674517244e-05,
+      "loss": 0.197,
+      "step": 19260
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.358477508650519e-05,
+      "loss": 0.1228,
+      "step": 19270
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.358142649849314e-05,
+      "loss": 0.1731,
+      "step": 19280
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.3578077910481083e-05,
+      "loss": 0.3058,
+      "step": 19290
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.3574729322469027e-05,
+      "loss": 0.149,
+      "step": 19300
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.357138073445697e-05,
+      "loss": 0.2642,
+      "step": 19310
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.3568032146444915e-05,
+      "loss": 0.2311,
+      "step": 19320
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.356468355843286e-05,
+      "loss": 0.2487,
+      "step": 19330
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.3561334970420806e-05,
+      "loss": 0.1473,
+      "step": 19340
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.3557986382408754e-05,
+      "loss": 0.2655,
+      "step": 19350
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.3554637794396698e-05,
+      "loss": 0.3396,
+      "step": 19360
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.3551289206384642e-05,
+      "loss": 0.2142,
+      "step": 19370
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.3547940618372586e-05,
+      "loss": 0.2058,
+      "step": 19380
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.354459203036053e-05,
+      "loss": 0.2343,
+      "step": 19390
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.3541243442348474e-05,
+      "loss": 0.1671,
+      "step": 19400
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.3537894854336425e-05,
+      "loss": 0.1995,
+      "step": 19410
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.353454626632437e-05,
+      "loss": 0.1659,
+      "step": 19420
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.3531197678312313e-05,
+      "loss": 0.2718,
+      "step": 19430
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.3527849090300257e-05,
+      "loss": 0.1871,
+      "step": 19440
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.35245005022882e-05,
+      "loss": 0.2816,
+      "step": 19450
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.3521151914276146e-05,
+      "loss": 0.3087,
+      "step": 19460
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.351780332626409e-05,
+      "loss": 0.291,
+      "step": 19470
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.351445473825204e-05,
+      "loss": 0.1995,
+      "step": 19480
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.3511106150239985e-05,
+      "loss": 0.2475,
+      "step": 19490
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.350775756222793e-05,
+      "loss": 0.1935,
+      "step": 19500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.3504408974215873e-05,
+      "loss": 0.1282,
+      "step": 19510
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.3501060386203817e-05,
+      "loss": 0.2189,
+      "step": 19520
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.349771179819176e-05,
+      "loss": 0.2106,
+      "step": 19530
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.3494363210179708e-05,
+      "loss": 0.2331,
+      "step": 19540
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.3491014622167656e-05,
+      "loss": 0.1196,
+      "step": 19550
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.34876660341556e-05,
+      "loss": 0.2308,
+      "step": 19560
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.3484317446143544e-05,
+      "loss": 0.16,
+      "step": 19570
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.3480968858131488e-05,
+      "loss": 0.2237,
+      "step": 19580
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.3477620270119432e-05,
+      "loss": 0.2803,
+      "step": 19590
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.3474271682107376e-05,
+      "loss": 0.2252,
+      "step": 19600
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.3470923094095324e-05,
+      "loss": 0.1919,
+      "step": 19610
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.346757450608327e-05,
+      "loss": 0.2077,
+      "step": 19620
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.3464225918071215e-05,
+      "loss": 0.1421,
+      "step": 19630
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.346087733005916e-05,
+      "loss": 0.3385,
+      "step": 19640
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.3457528742047103e-05,
+      "loss": 0.1911,
+      "step": 19650
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.3454180154035047e-05,
+      "loss": 0.1132,
+      "step": 19660
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.3450831566022995e-05,
+      "loss": 0.2937,
+      "step": 19670
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.344748297801094e-05,
+      "loss": 0.2677,
+      "step": 19680
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.3444134389998886e-05,
+      "loss": 0.1919,
+      "step": 19690
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.344078580198683e-05,
+      "loss": 0.155,
+      "step": 19700
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.3437437213974774e-05,
+      "loss": 0.1339,
+      "step": 19710
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.343408862596272e-05,
+      "loss": 0.1391,
+      "step": 19720
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.3430740037950663e-05,
+      "loss": 0.355,
+      "step": 19730
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.342739144993861e-05,
+      "loss": 0.2515,
+      "step": 19740
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.3424042861926554e-05,
+      "loss": 0.163,
+      "step": 19750
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.34206942739145e-05,
+      "loss": 0.1436,
+      "step": 19760
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.3417345685902446e-05,
+      "loss": 0.196,
+      "step": 19770
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.341399709789039e-05,
+      "loss": 0.1801,
+      "step": 19780
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.3410648509878334e-05,
+      "loss": 0.2532,
+      "step": 19790
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.340729992186628e-05,
+      "loss": 0.2135,
+      "step": 19800
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.3403951333854225e-05,
+      "loss": 0.1654,
+      "step": 19810
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.340060274584217e-05,
+      "loss": 0.1636,
+      "step": 19820
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.3397254157830117e-05,
+      "loss": 0.2862,
+      "step": 19830
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.339390556981806e-05,
+      "loss": 0.161,
+      "step": 19840
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.3390556981806005e-05,
+      "loss": 0.2013,
+      "step": 19850
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.338720839379395e-05,
+      "loss": 0.1413,
+      "step": 19860
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.3383859805781897e-05,
+      "loss": 0.2148,
+      "step": 19870
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.338051121776984e-05,
+      "loss": 0.1584,
+      "step": 19880
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.3377162629757785e-05,
+      "loss": 0.2945,
+      "step": 19890
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.3373814041745732e-05,
+      "loss": 0.2061,
+      "step": 19900
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.3370465453733676e-05,
+      "loss": 0.1533,
+      "step": 19910
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.336711686572162e-05,
+      "loss": 0.1532,
+      "step": 19920
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.3363768277709568e-05,
+      "loss": 0.1851,
+      "step": 19930
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.3360419689697512e-05,
+      "loss": 0.247,
+      "step": 19940
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.3357071101685456e-05,
+      "loss": 0.1713,
+      "step": 19950
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.33537225136734e-05,
+      "loss": 0.2539,
+      "step": 19960
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.3350373925661347e-05,
+      "loss": 0.1647,
+      "step": 19970
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.334702533764929e-05,
+      "loss": 0.2518,
+      "step": 19980
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.3343676749637236e-05,
+      "loss": 0.3284,
+      "step": 19990
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.3340328161625183e-05,
+      "loss": 0.158,
+      "step": 20000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.3336979573613127e-05,
+      "loss": 0.1258,
+      "step": 20010
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.333363098560107e-05,
+      "loss": 0.1256,
+      "step": 20020
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.3330282397589015e-05,
+      "loss": 0.2773,
+      "step": 20030
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.3326933809576963e-05,
+      "loss": 0.2638,
+      "step": 20040
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.3323585221564907e-05,
+      "loss": 0.218,
+      "step": 20050
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.3320236633552854e-05,
+      "loss": 0.2235,
+      "step": 20060
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.33168880455408e-05,
+      "loss": 0.1189,
+      "step": 20070
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.3313539457528742e-05,
+      "loss": 0.1263,
+      "step": 20080
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.3310190869516687e-05,
+      "loss": 0.105,
+      "step": 20090
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.330684228150463e-05,
+      "loss": 0.2745,
+      "step": 20100
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.3303493693492578e-05,
+      "loss": 0.1581,
+      "step": 20110
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.3300145105480522e-05,
+      "loss": 0.1344,
+      "step": 20120
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.329679651746847e-05,
+      "loss": 0.1733,
+      "step": 20130
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.3293447929456414e-05,
+      "loss": 0.2026,
+      "step": 20140
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.3290099341444358e-05,
+      "loss": 0.17,
+      "step": 20150
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.3286750753432302e-05,
+      "loss": 0.2552,
+      "step": 20160
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.328340216542025e-05,
+      "loss": 0.2339,
+      "step": 20170
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.3280053577408193e-05,
+      "loss": 0.1072,
+      "step": 20180
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.327670498939614e-05,
+      "loss": 0.2673,
+      "step": 20190
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.3273356401384085e-05,
+      "loss": 0.152,
+      "step": 20200
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.327000781337203e-05,
+      "loss": 0.1985,
+      "step": 20210
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.3266659225359973e-05,
+      "loss": 0.0871,
+      "step": 20220
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.3263310637347917e-05,
+      "loss": 0.2551,
+      "step": 20230
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.3259962049335865e-05,
+      "loss": 0.1015,
+      "step": 20240
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.325661346132381e-05,
+      "loss": 0.1986,
+      "step": 20250
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.3253264873311756e-05,
+      "loss": 0.1507,
+      "step": 20260
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.32499162852997e-05,
+      "loss": 0.2271,
+      "step": 20270
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.3246567697287644e-05,
+      "loss": 0.2806,
+      "step": 20280
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.324321910927559e-05,
+      "loss": 0.3545,
+      "step": 20290
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.3239870521263532e-05,
+      "loss": 0.158,
+      "step": 20300
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.323652193325148e-05,
+      "loss": 0.1064,
+      "step": 20310
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.3233173345239427e-05,
+      "loss": 0.1311,
+      "step": 20320
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.322982475722737e-05,
+      "loss": 0.2652,
+      "step": 20330
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.3226476169215316e-05,
+      "loss": 0.2521,
+      "step": 20340
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.322312758120326e-05,
+      "loss": 0.1544,
+      "step": 20350
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.3219778993191204e-05,
+      "loss": 0.2242,
+      "step": 20360
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.3216430405179148e-05,
+      "loss": 0.1413,
+      "step": 20370
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.3213081817167095e-05,
+      "loss": 0.2181,
+      "step": 20380
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.3209733229155043e-05,
+      "loss": 0.3582,
+      "step": 20390
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.3206384641142987e-05,
+      "loss": 0.178,
+      "step": 20400
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.320303605313093e-05,
+      "loss": 0.2052,
+      "step": 20410
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3199687465118875e-05,
+      "loss": 0.146,
+      "step": 20420
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.319633887710682e-05,
+      "loss": 0.2236,
+      "step": 20430
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3192990289094763e-05,
+      "loss": 0.2728,
+      "step": 20440
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3189641701082714e-05,
+      "loss": 0.237,
+      "step": 20450
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3186293113070658e-05,
+      "loss": 0.1828,
+      "step": 20460
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3182944525058602e-05,
+      "loss": 0.2292,
+      "step": 20470
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3179595937046546e-05,
+      "loss": 0.1412,
+      "step": 20480
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.317624734903449e-05,
+      "loss": 0.1236,
+      "step": 20490
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3172898761022434e-05,
+      "loss": 0.1095,
+      "step": 20500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.316955017301038e-05,
+      "loss": 0.1962,
+      "step": 20510
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.316620158499833e-05,
+      "loss": 0.2042,
+      "step": 20520
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3162852996986273e-05,
+      "loss": 0.1424,
+      "step": 20530
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3159504408974217e-05,
+      "loss": 0.1936,
+      "step": 20540
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.315615582096216e-05,
+      "loss": 0.1541,
+      "step": 20550
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3152807232950105e-05,
+      "loss": 0.1711,
+      "step": 20560
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.314945864493805e-05,
+      "loss": 0.216,
+      "step": 20570
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3146110056925997e-05,
+      "loss": 0.2056,
+      "step": 20580
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3142761468913944e-05,
+      "loss": 0.1952,
+      "step": 20590
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.313941288090189e-05,
+      "loss": 0.2907,
+      "step": 20600
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3136064292889833e-05,
+      "loss": 0.1384,
+      "step": 20610
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3132715704877777e-05,
+      "loss": 0.2163,
+      "step": 20620
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.312936711686572e-05,
+      "loss": 0.1224,
+      "step": 20630
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3126018528853665e-05,
+      "loss": 0.1796,
+      "step": 20640
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3122669940841612e-05,
+      "loss": 0.1568,
+      "step": 20650
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.311932135282956e-05,
+      "loss": 0.1518,
+      "step": 20660
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3115972764817504e-05,
+      "loss": 0.2339,
+      "step": 20670
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3112624176805448e-05,
+      "loss": 0.1787,
+      "step": 20680
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3109275588793392e-05,
+      "loss": 0.2455,
+      "step": 20690
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3105927000781336e-05,
+      "loss": 0.1642,
+      "step": 20700
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3102578412769284e-05,
+      "loss": 0.1191,
+      "step": 20710
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3099229824757228e-05,
+      "loss": 0.171,
+      "step": 20720
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3095881236745175e-05,
+      "loss": 0.2184,
+      "step": 20730
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.309253264873312e-05,
+      "loss": 0.1879,
+      "step": 20740
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3089184060721063e-05,
+      "loss": 0.1862,
+      "step": 20750
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3085835472709007e-05,
+      "loss": 0.2457,
+      "step": 20760
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.308248688469695e-05,
+      "loss": 0.1564,
+      "step": 20770
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.30791382966849e-05,
+      "loss": 0.3087,
+      "step": 20780
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3075789708672843e-05,
+      "loss": 0.2484,
+      "step": 20790
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.307244112066079e-05,
+      "loss": 0.1795,
+      "step": 20800
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3069092532648734e-05,
+      "loss": 0.1614,
+      "step": 20810
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.306574394463668e-05,
+      "loss": 0.2393,
+      "step": 20820
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3062395356624623e-05,
+      "loss": 0.1118,
+      "step": 20830
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3059046768612567e-05,
+      "loss": 0.1555,
+      "step": 20840
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3055698180600514e-05,
+      "loss": 0.1645,
+      "step": 20850
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3052349592588458e-05,
+      "loss": 0.3751,
+      "step": 20860
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3049001004576406e-05,
+      "loss": 0.1434,
+      "step": 20870
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.304565241656435e-05,
+      "loss": 0.2973,
+      "step": 20880
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3042303828552294e-05,
+      "loss": 0.224,
+      "step": 20890
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3038955240540238e-05,
+      "loss": 0.2241,
+      "step": 20900
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3035606652528185e-05,
+      "loss": 0.133,
+      "step": 20910
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.303225806451613e-05,
+      "loss": 0.4102,
+      "step": 20920
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3028909476504073e-05,
+      "loss": 0.1994,
+      "step": 20930
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.302556088849202e-05,
+      "loss": 0.1903,
+      "step": 20940
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3022212300479965e-05,
+      "loss": 0.2058,
+      "step": 20950
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.301886371246791e-05,
+      "loss": 0.1739,
+      "step": 20960
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3015515124455853e-05,
+      "loss": 0.1892,
+      "step": 20970
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.30121665364438e-05,
+      "loss": 0.2009,
+      "step": 20980
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3008817948431745e-05,
+      "loss": 0.2207,
+      "step": 20990
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.300546936041969e-05,
+      "loss": 0.1393,
+      "step": 21000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3002120772407636e-05,
+      "loss": 0.0721,
+      "step": 21010
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.299877218439558e-05,
+      "loss": 0.4065,
+      "step": 21020
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.2995423596383524e-05,
+      "loss": 0.1206,
+      "step": 21030
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2992075008371472e-05,
+      "loss": 0.3194,
+      "step": 21040
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2988726420359416e-05,
+      "loss": 0.3156,
+      "step": 21050
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.298537783234736e-05,
+      "loss": 0.2029,
+      "step": 21060
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2982029244335307e-05,
+      "loss": 0.1797,
+      "step": 21070
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.297868065632325e-05,
+      "loss": 0.1351,
+      "step": 21080
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2975332068311196e-05,
+      "loss": 0.1851,
+      "step": 21090
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.297198348029914e-05,
+      "loss": 0.2279,
+      "step": 21100
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2968634892287087e-05,
+      "loss": 0.2027,
+      "step": 21110
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.296528630427503e-05,
+      "loss": 0.2448,
+      "step": 21120
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2961937716262975e-05,
+      "loss": 0.1003,
+      "step": 21130
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2958589128250923e-05,
+      "loss": 0.1324,
+      "step": 21140
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2955240540238867e-05,
+      "loss": 0.1142,
+      "step": 21150
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.295189195222681e-05,
+      "loss": 0.2246,
+      "step": 21160
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.294854336421476e-05,
+      "loss": 0.1372,
+      "step": 21170
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2945194776202702e-05,
+      "loss": 0.2195,
+      "step": 21180
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2941846188190647e-05,
+      "loss": 0.2678,
+      "step": 21190
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.293849760017859e-05,
+      "loss": 0.1729,
+      "step": 21200
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2935149012166538e-05,
+      "loss": 0.1255,
+      "step": 21210
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2931800424154482e-05,
+      "loss": 0.1624,
+      "step": 21220
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2928451836142426e-05,
+      "loss": 0.2803,
+      "step": 21230
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2925103248130374e-05,
+      "loss": 0.2737,
+      "step": 21240
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2921754660118318e-05,
+      "loss": 0.1942,
+      "step": 21250
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2918406072106262e-05,
+      "loss": 0.1456,
+      "step": 21260
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2915057484094206e-05,
+      "loss": 0.1207,
+      "step": 21270
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2911708896082153e-05,
+      "loss": 0.1222,
+      "step": 21280
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2908360308070097e-05,
+      "loss": 0.1984,
+      "step": 21290
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2905011720058045e-05,
+      "loss": 0.1215,
+      "step": 21300
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.290166313204599e-05,
+      "loss": 0.2263,
+      "step": 21310
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2898314544033933e-05,
+      "loss": 0.2501,
+      "step": 21320
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2894965956021877e-05,
+      "loss": 0.2688,
+      "step": 21330
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.289161736800982e-05,
+      "loss": 0.3059,
+      "step": 21340
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.288826877999777e-05,
+      "loss": 0.198,
+      "step": 21350
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2884920191985713e-05,
+      "loss": 0.1607,
+      "step": 21360
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.288157160397366e-05,
+      "loss": 0.1081,
+      "step": 21370
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2878223015961604e-05,
+      "loss": 0.1475,
+      "step": 21380
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.287487442794955e-05,
+      "loss": 0.2798,
+      "step": 21390
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2871525839937492e-05,
+      "loss": 0.2308,
+      "step": 21400
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2868177251925436e-05,
+      "loss": 0.3438,
+      "step": 21410
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2864828663913384e-05,
+      "loss": 0.1813,
+      "step": 21420
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.286148007590133e-05,
+      "loss": 0.0871,
+      "step": 21430
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2858131487889275e-05,
+      "loss": 0.296,
+      "step": 21440
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.285478289987722e-05,
+      "loss": 0.1878,
+      "step": 21450
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2851434311865164e-05,
+      "loss": 0.198,
+      "step": 21460
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2848085723853108e-05,
+      "loss": 0.1265,
+      "step": 21470
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2844737135841052e-05,
+      "loss": 0.1741,
+      "step": 21480
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2841388547829e-05,
+      "loss": 0.2039,
+      "step": 21490
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2838039959816947e-05,
+      "loss": 0.2616,
+      "step": 21500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.283469137180489e-05,
+      "loss": 0.2405,
+      "step": 21510
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2831342783792835e-05,
+      "loss": 0.0992,
+      "step": 21520
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.282799419578078e-05,
+      "loss": 0.1739,
+      "step": 21530
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2824645607768723e-05,
+      "loss": 0.2459,
+      "step": 21540
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2821297019756667e-05,
+      "loss": 0.2282,
+      "step": 21550
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2817948431744618e-05,
+      "loss": 0.1325,
+      "step": 21560
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2814599843732562e-05,
+      "loss": 0.1382,
+      "step": 21570
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2811251255720506e-05,
+      "loss": 0.1922,
+      "step": 21580
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.280790266770845e-05,
+      "loss": 0.2845,
+      "step": 21590
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2804554079696394e-05,
+      "loss": 0.2823,
+      "step": 21600
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2801205491684338e-05,
+      "loss": 0.1916,
+      "step": 21610
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2797856903672282e-05,
+      "loss": 0.1872,
+      "step": 21620
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2794508315660233e-05,
+      "loss": 0.3424,
+      "step": 21630
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2791159727648177e-05,
+      "loss": 0.1159,
+      "step": 21640
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.278781113963612e-05,
+      "loss": 0.1958,
+      "step": 21650
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2784462551624065e-05,
+      "loss": 0.1768,
+      "step": 21660
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.278111396361201e-05,
+      "loss": 0.2246,
+      "step": 21670
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.2777765375599954e-05,
+      "loss": 0.1654,
+      "step": 21680
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.27744167875879e-05,
+      "loss": 0.1828,
+      "step": 21690
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.277106819957585e-05,
+      "loss": 0.1828,
+      "step": 21700
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.2767719611563793e-05,
+      "loss": 0.1392,
+      "step": 21710
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.2764371023551737e-05,
+      "loss": 0.1731,
+      "step": 21720
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.276102243553968e-05,
+      "loss": 0.0982,
+      "step": 21730
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.2757673847527625e-05,
+      "loss": 0.2941,
+      "step": 21740
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.275432525951557e-05,
+      "loss": 0.2129,
+      "step": 21750
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.2750976671503516e-05,
+      "loss": 0.1223,
+      "step": 21760
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.2747628083491464e-05,
+      "loss": 0.2097,
+      "step": 21770
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.2744279495479408e-05,
+      "loss": 0.2156,
+      "step": 21780
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.2740930907467352e-05,
+      "loss": 0.2485,
+      "step": 21790
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.2737582319455296e-05,
+      "loss": 0.1131,
+      "step": 21800
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.273423373144324e-05,
+      "loss": 0.2515,
+      "step": 21810
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.2730885143431188e-05,
+      "loss": 0.275,
+      "step": 21820
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.272753655541913e-05,
+      "loss": 0.1405,
+      "step": 21830
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.272418796740708e-05,
+      "loss": 0.1777,
+      "step": 21840
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.2720839379395023e-05,
+      "loss": 0.1472,
+      "step": 21850
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.2717490791382967e-05,
+      "loss": 0.118,
+      "step": 21860
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.271414220337091e-05,
+      "loss": 0.161,
+      "step": 21870
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.2710793615358855e-05,
+      "loss": 0.2752,
+      "step": 21880
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.2707445027346803e-05,
+      "loss": 0.1372,
+      "step": 21890
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.2704096439334747e-05,
+      "loss": 0.2341,
+      "step": 21900
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.2700747851322694e-05,
+      "loss": 0.241,
+      "step": 21910
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.269739926331064e-05,
+      "loss": 0.1465,
+      "step": 21920
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.2694050675298583e-05,
+      "loss": 0.1579,
+      "step": 21930
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.2690702087286527e-05,
+      "loss": 0.2959,
+      "step": 21940
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.2687353499274474e-05,
+      "loss": 0.2967,
+      "step": 21950
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.2684004911262418e-05,
+      "loss": 0.1545,
+      "step": 21960
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.2680656323250366e-05,
+      "loss": 0.1451,
+      "step": 21970
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.267730773523831e-05,
+      "loss": 0.1786,
+      "step": 21980
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.2673959147226254e-05,
+      "loss": 0.2467,
+      "step": 21990
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.2670610559214198e-05,
+      "loss": 0.1733,
+      "step": 22000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.2667261971202142e-05,
+      "loss": 0.1854,
+      "step": 22010
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.266391338319009e-05,
+      "loss": 0.2367,
+      "step": 22020
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.2660564795178033e-05,
+      "loss": 0.1866,
+      "step": 22030
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.265721620716598e-05,
+      "loss": 0.3088,
+      "step": 22040
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.2653867619153925e-05,
+      "loss": 0.2012,
+      "step": 22050
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.265051903114187e-05,
+      "loss": 0.1497,
+      "step": 22060
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.2647170443129813e-05,
+      "loss": 0.286,
+      "step": 22070
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.264382185511776e-05,
+      "loss": 0.1597,
+      "step": 22080
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.2640473267105705e-05,
+      "loss": 0.1902,
+      "step": 22090
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.263712467909365e-05,
+      "loss": 0.2244,
+      "step": 22100
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.2633776091081596e-05,
+      "loss": 0.2782,
+      "step": 22110
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.263042750306954e-05,
+      "loss": 0.2512,
+      "step": 22120
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.2627078915057484e-05,
+      "loss": 0.2092,
+      "step": 22130
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.262373032704543e-05,
+      "loss": 0.2197,
+      "step": 22140
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.2620381739033376e-05,
+      "loss": 0.1822,
+      "step": 22150
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.261703315102132e-05,
+      "loss": 0.1166,
+      "step": 22160
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.2613684563009264e-05,
+      "loss": 0.1919,
+      "step": 22170
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.261033597499721e-05,
+      "loss": 0.1125,
+      "step": 22180
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.2606987386985156e-05,
+      "loss": 0.2409,
+      "step": 22190
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.26036387989731e-05,
+      "loss": 0.1686,
+      "step": 22200
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.2600290210961047e-05,
+      "loss": 0.2285,
+      "step": 22210
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.259694162294899e-05,
+      "loss": 0.0979,
+      "step": 22220
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.2593593034936935e-05,
+      "loss": 0.151,
+      "step": 22230
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.259024444692488e-05,
+      "loss": 0.273,
+      "step": 22240
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.2586895858912827e-05,
+      "loss": 0.1408,
+      "step": 22250
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.258354727090077e-05,
+      "loss": 0.121,
+      "step": 22260
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.2580198682888715e-05,
+      "loss": 0.1693,
+      "step": 22270
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.2576850094876662e-05,
+      "loss": 0.165,
+      "step": 22280
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.2573501506864606e-05,
+      "loss": 0.3403,
+      "step": 22290
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.257015291885255e-05,
+      "loss": 0.1753,
+      "step": 22300
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.2566804330840495e-05,
+      "loss": 0.1785,
+      "step": 22310
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.2563455742828442e-05,
+      "loss": 0.1808,
+      "step": 22320
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.2560107154816386e-05,
+      "loss": 0.1199,
+      "step": 22330
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.255675856680433e-05,
+      "loss": 0.428,
+      "step": 22340
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.2553409978792278e-05,
+      "loss": 0.2806,
+      "step": 22350
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.2550061390780222e-05,
+      "loss": 0.1536,
+      "step": 22360
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.2546712802768166e-05,
+      "loss": 0.1997,
+      "step": 22370
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.254336421475611e-05,
+      "loss": 0.2746,
+      "step": 22380
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.2540015626744057e-05,
+      "loss": 0.1805,
+      "step": 22390
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.2536667038732e-05,
+      "loss": 0.2081,
+      "step": 22400
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.253331845071995e-05,
+      "loss": 0.2316,
+      "step": 22410
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.2529969862707893e-05,
+      "loss": 0.2194,
+      "step": 22420
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.2526621274695837e-05,
+      "loss": 0.2352,
+      "step": 22430
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.252327268668378e-05,
+      "loss": 0.3082,
+      "step": 22440
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.2519924098671725e-05,
+      "loss": 0.2052,
+      "step": 22450
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.2516575510659673e-05,
+      "loss": 0.1743,
+      "step": 22460
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.2513226922647617e-05,
+      "loss": 0.2059,
+      "step": 22470
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.2509878334635564e-05,
+      "loss": 0.1408,
+      "step": 22480
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.2506529746623508e-05,
+      "loss": 0.1827,
+      "step": 22490
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.2503181158611452e-05,
+      "loss": 0.2184,
+      "step": 22500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.2499832570599396e-05,
+      "loss": 0.2882,
+      "step": 22510
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.249648398258734e-05,
+      "loss": 0.2109,
+      "step": 22520
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.2493135394575288e-05,
+      "loss": 0.2145,
+      "step": 22530
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.2489786806563235e-05,
+      "loss": 0.176,
+      "step": 22540
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.248643821855118e-05,
+      "loss": 0.1853,
+      "step": 22550
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.2483089630539124e-05,
+      "loss": 0.2839,
+      "step": 22560
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.2479741042527068e-05,
+      "loss": 0.184,
+      "step": 22570
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.2476392454515012e-05,
+      "loss": 0.1407,
+      "step": 22580
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.2473043866502956e-05,
+      "loss": 0.2485,
+      "step": 22590
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.2469695278490903e-05,
+      "loss": 0.1848,
+      "step": 22600
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.246634669047885e-05,
+      "loss": 0.2464,
+      "step": 22610
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.2462998102466795e-05,
+      "loss": 0.2128,
+      "step": 22620
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.245964951445474e-05,
+      "loss": 0.1915,
+      "step": 22630
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.2456300926442683e-05,
+      "loss": 0.1544,
+      "step": 22640
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.2452952338430627e-05,
+      "loss": 0.1619,
+      "step": 22650
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.244960375041857e-05,
+      "loss": 0.2076,
+      "step": 22660
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.2446255162406522e-05,
+      "loss": 0.1731,
+      "step": 22670
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.2442906574394466e-05,
+      "loss": 0.1632,
+      "step": 22680
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.243955798638241e-05,
+      "loss": 0.2281,
+      "step": 22690
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.2436209398370354e-05,
+      "loss": 0.1113,
+      "step": 22700
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.2432860810358298e-05,
+      "loss": 0.2372,
+      "step": 22710
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.2429512222346242e-05,
+      "loss": 0.1814,
+      "step": 22720
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.2426163634334186e-05,
+      "loss": 0.1624,
+      "step": 22730
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.2422815046322137e-05,
+      "loss": 0.1787,
+      "step": 22740
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.241946645831008e-05,
+      "loss": 0.1024,
+      "step": 22750
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.2416117870298025e-05,
+      "loss": 0.1062,
+      "step": 22760
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.241276928228597e-05,
+      "loss": 0.2705,
+      "step": 22770
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.2409420694273914e-05,
+      "loss": 0.1866,
+      "step": 22780
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.2406072106261858e-05,
+      "loss": 0.26,
+      "step": 22790
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.2402723518249805e-05,
+      "loss": 0.229,
+      "step": 22800
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.2399374930237753e-05,
+      "loss": 0.1609,
+      "step": 22810
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.2396026342225697e-05,
+      "loss": 0.1315,
+      "step": 22820
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.239267775421364e-05,
+      "loss": 0.1735,
+      "step": 22830
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.2389329166201585e-05,
+      "loss": 0.2248,
+      "step": 22840
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.238598057818953e-05,
+      "loss": 0.0814,
+      "step": 22850
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.2382631990177473e-05,
+      "loss": 0.1728,
+      "step": 22860
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.2379283402165424e-05,
+      "loss": 0.2598,
+      "step": 22870
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.2375934814153368e-05,
+      "loss": 0.1915,
+      "step": 22880
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.2372586226141312e-05,
+      "loss": 0.1834,
+      "step": 22890
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.2369237638129256e-05,
+      "loss": 0.119,
+      "step": 22900
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.23658890501172e-05,
+      "loss": 0.0812,
+      "step": 22910
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.2362540462105144e-05,
+      "loss": 0.2315,
+      "step": 22920
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.235919187409309e-05,
+      "loss": 0.1451,
+      "step": 22930
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.235584328608104e-05,
+      "loss": 0.1918,
+      "step": 22940
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.2352494698068983e-05,
+      "loss": 0.2751,
+      "step": 22950
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.2349146110056927e-05,
+      "loss": 0.2158,
+      "step": 22960
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.234579752204487e-05,
+      "loss": 0.1563,
+      "step": 22970
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.2342448934032815e-05,
+      "loss": 0.2083,
+      "step": 22980
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.233910034602076e-05,
+      "loss": 0.164,
+      "step": 22990
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.2335751758008707e-05,
+      "loss": 0.2233,
+      "step": 23000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.2332403169996654e-05,
+      "loss": 0.2018,
+      "step": 23010
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.23290545819846e-05,
+      "loss": 0.1008,
+      "step": 23020
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.2325705993972542e-05,
+      "loss": 0.3177,
+      "step": 23030
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.2322357405960487e-05,
+      "loss": 0.2268,
+      "step": 23040
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.231900881794843e-05,
+      "loss": 0.1742,
+      "step": 23050
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.2315660229936378e-05,
+      "loss": 0.2632,
+      "step": 23060
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.2312311641924322e-05,
+      "loss": 0.1867,
+      "step": 23070
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.230896305391227e-05,
+      "loss": 0.1938,
+      "step": 23080
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.2305614465900214e-05,
+      "loss": 0.2058,
+      "step": 23090
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.2302265877888158e-05,
+      "loss": 0.2115,
+      "step": 23100
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.2298917289876102e-05,
+      "loss": 0.1592,
+      "step": 23110
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.2295568701864046e-05,
+      "loss": 0.1498,
+      "step": 23120
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.2292220113851993e-05,
+      "loss": 0.1602,
+      "step": 23130
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.2288871525839937e-05,
+      "loss": 0.3014,
+      "step": 23140
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.2285522937827885e-05,
+      "loss": 0.2133,
+      "step": 23150
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.228217434981583e-05,
+      "loss": 0.1667,
+      "step": 23160
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.2278825761803773e-05,
+      "loss": 0.1634,
+      "step": 23170
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.2275477173791717e-05,
+      "loss": 0.1331,
+      "step": 23180
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.2272128585779665e-05,
+      "loss": 0.1909,
+      "step": 23190
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.226877999776761e-05,
+      "loss": 0.0876,
+      "step": 23200
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.2265431409755553e-05,
+      "loss": 0.2056,
+      "step": 23210
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.22620828217435e-05,
+      "loss": 0.407,
+      "step": 23220
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.2258734233731444e-05,
+      "loss": 0.1599,
+      "step": 23230
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.225538564571939e-05,
+      "loss": 0.1803,
+      "step": 23240
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.2252037057707332e-05,
+      "loss": 0.1254,
+      "step": 23250
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.224868846969528e-05,
+      "loss": 0.3661,
+      "step": 23260
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.2245339881683224e-05,
+      "loss": 0.1758,
+      "step": 23270
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.2241991293671168e-05,
+      "loss": 0.1889,
+      "step": 23280
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.2238642705659116e-05,
+      "loss": 0.1764,
+      "step": 23290
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.223529411764706e-05,
+      "loss": 0.1367,
+      "step": 23300
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.2231945529635004e-05,
+      "loss": 0.1322,
+      "step": 23310
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.222859694162295e-05,
+      "loss": 0.1583,
+      "step": 23320
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.2225248353610895e-05,
+      "loss": 0.1858,
+      "step": 23330
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.222189976559884e-05,
+      "loss": 0.178,
+      "step": 23340
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.2218551177586783e-05,
+      "loss": 0.0966,
+      "step": 23350
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.221520258957473e-05,
+      "loss": 0.1293,
+      "step": 23360
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.2211854001562675e-05,
+      "loss": 0.268,
+      "step": 23370
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.220850541355062e-05,
+      "loss": 0.1983,
+      "step": 23380
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.2205156825538566e-05,
+      "loss": 0.4059,
+      "step": 23390
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.220180823752651e-05,
+      "loss": 0.1268,
+      "step": 23400
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.2198459649514455e-05,
+      "loss": 0.0882,
+      "step": 23410
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.21951110615024e-05,
+      "loss": 0.1308,
+      "step": 23420
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.2191762473490346e-05,
+      "loss": 0.3227,
+      "step": 23430
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.218841388547829e-05,
+      "loss": 0.2607,
+      "step": 23440
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.2185065297466238e-05,
+      "loss": 0.1825,
+      "step": 23450
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.2181716709454182e-05,
+      "loss": 0.1887,
+      "step": 23460
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.2178368121442126e-05,
+      "loss": 0.1522,
+      "step": 23470
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.217501953343007e-05,
+      "loss": 0.2297,
+      "step": 23480
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.2171670945418014e-05,
+      "loss": 0.132,
+      "step": 23490
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.216832235740596e-05,
+      "loss": 0.1771,
+      "step": 23500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.2164973769393905e-05,
+      "loss": 0.3435,
+      "step": 23510
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.2161625181381853e-05,
+      "loss": 0.1768,
+      "step": 23520
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.2158276593369797e-05,
+      "loss": 0.2004,
+      "step": 23530
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.215492800535774e-05,
+      "loss": 0.2428,
+      "step": 23540
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.2151579417345685e-05,
+      "loss": 0.6497,
+      "step": 23550
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.214823082933363e-05,
+      "loss": 0.1788,
+      "step": 23560
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.2144882241321577e-05,
+      "loss": 0.1384,
+      "step": 23570
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.2141533653309524e-05,
+      "loss": 0.1218,
+      "step": 23580
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.2138185065297468e-05,
+      "loss": 0.1963,
+      "step": 23590
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.2134836477285412e-05,
+      "loss": 0.2014,
+      "step": 23600
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.2131487889273356e-05,
+      "loss": 0.1769,
+      "step": 23610
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.21281393012613e-05,
+      "loss": 0.1696,
+      "step": 23620
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.2124790713249245e-05,
+      "loss": 0.2211,
+      "step": 23630
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.2121442125237192e-05,
+      "loss": 0.1864,
+      "step": 23640
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.211809353722514e-05,
+      "loss": 0.1862,
+      "step": 23650
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.2114744949213084e-05,
+      "loss": 0.2001,
+      "step": 23660
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.2111396361201028e-05,
+      "loss": 0.2334,
+      "step": 23670
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.210804777318897e-05,
+      "loss": 0.1745,
+      "step": 23680
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.2104699185176916e-05,
+      "loss": 0.2131,
+      "step": 23690
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.210135059716486e-05,
+      "loss": 0.1815,
+      "step": 23700
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.209800200915281e-05,
+      "loss": 0.2684,
+      "step": 23710
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.2094653421140755e-05,
+      "loss": 0.1823,
+      "step": 23720
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.20913048331287e-05,
+      "loss": 0.1768,
+      "step": 23730
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.2087956245116643e-05,
+      "loss": 0.5496,
+      "step": 23740
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.2084607657104587e-05,
+      "loss": 0.1706,
+      "step": 23750
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.208125906909253e-05,
+      "loss": 0.2107,
+      "step": 23760
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.207791048108048e-05,
+      "loss": 0.2794,
+      "step": 23770
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.2074561893068426e-05,
+      "loss": 0.1545,
+      "step": 23780
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.207121330505637e-05,
+      "loss": 0.2845,
+      "step": 23790
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.2067864717044314e-05,
+      "loss": 0.1899,
+      "step": 23800
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.2064516129032258e-05,
+      "loss": 0.2358,
+      "step": 23810
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2061167541020202e-05,
+      "loss": 0.2199,
+      "step": 23820
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2057818953008146e-05,
+      "loss": 0.2927,
+      "step": 23830
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2054470364996094e-05,
+      "loss": 0.1707,
+      "step": 23840
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.205112177698404e-05,
+      "loss": 0.2701,
+      "step": 23850
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2047773188971985e-05,
+      "loss": 0.253,
+      "step": 23860
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.204442460095993e-05,
+      "loss": 0.3525,
+      "step": 23870
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2041076012947873e-05,
+      "loss": 0.1636,
+      "step": 23880
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2037727424935818e-05,
+      "loss": 0.1602,
+      "step": 23890
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.203437883692376e-05,
+      "loss": 0.2493,
+      "step": 23900
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2031030248911712e-05,
+      "loss": 0.1921,
+      "step": 23910
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2027681660899657e-05,
+      "loss": 0.1318,
+      "step": 23920
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.20243330728876e-05,
+      "loss": 0.2552,
+      "step": 23930
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2020984484875545e-05,
+      "loss": 0.174,
+      "step": 23940
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.201763589686349e-05,
+      "loss": 0.1548,
+      "step": 23950
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2014287308851433e-05,
+      "loss": 0.1372,
+      "step": 23960
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.201093872083938e-05,
+      "loss": 0.1777,
+      "step": 23970
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2007590132827328e-05,
+      "loss": 0.1488,
+      "step": 23980
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2004241544815272e-05,
+      "loss": 0.1175,
+      "step": 23990
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.2000892956803216e-05,
+      "loss": 0.2155,
+      "step": 24000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.199754436879116e-05,
+      "loss": 0.1374,
+      "step": 24010
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1994195780779104e-05,
+      "loss": 0.1042,
+      "step": 24020
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1990847192767048e-05,
+      "loss": 0.1502,
+      "step": 24030
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1987498604754996e-05,
+      "loss": 0.2947,
+      "step": 24040
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1984150016742943e-05,
+      "loss": 0.1939,
+      "step": 24050
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1980801428730887e-05,
+      "loss": 0.1874,
+      "step": 24060
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.197745284071883e-05,
+      "loss": 0.1528,
+      "step": 24070
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1974104252706775e-05,
+      "loss": 0.2495,
+      "step": 24080
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.197075566469472e-05,
+      "loss": 0.2496,
+      "step": 24090
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1967741935483873e-05,
+      "loss": 0.138,
+      "step": 24100
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1964393347471818e-05,
+      "loss": 0.1892,
+      "step": 24110
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.196104475945976e-05,
+      "loss": 0.1216,
+      "step": 24120
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1957696171447706e-05,
+      "loss": 0.1817,
+      "step": 24130
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.195434758343565e-05,
+      "loss": 0.2741,
+      "step": 24140
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1950998995423597e-05,
+      "loss": 0.1085,
+      "step": 24150
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1947650407411545e-05,
+      "loss": 0.1873,
+      "step": 24160
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.194430181939949e-05,
+      "loss": 0.1391,
+      "step": 24170
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.1940953231387433e-05,
+      "loss": 0.1175,
+      "step": 24180
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.1937604643375377e-05,
+      "loss": 0.2757,
+      "step": 24190
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.193425605536332e-05,
+      "loss": 0.2555,
+      "step": 24200
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.1930907467351265e-05,
+      "loss": 0.2266,
+      "step": 24210
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.1927558879339212e-05,
+      "loss": 0.0877,
+      "step": 24220
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.192421029132716e-05,
+      "loss": 0.3501,
+      "step": 24230
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.1920861703315104e-05,
+      "loss": 0.3327,
+      "step": 24240
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.1917513115303048e-05,
+      "loss": 0.1326,
+      "step": 24250
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.1914164527290992e-05,
+      "loss": 0.1689,
+      "step": 24260
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.1910815939278936e-05,
+      "loss": 0.3338,
+      "step": 24270
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.1907467351266884e-05,
+      "loss": 0.2726,
+      "step": 24280
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.1904118763254828e-05,
+      "loss": 0.305,
+      "step": 24290
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.1900770175242775e-05,
+      "loss": 0.2151,
+      "step": 24300
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.189742158723072e-05,
+      "loss": 0.1243,
+      "step": 24310
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.1894072999218663e-05,
+      "loss": 0.239,
+      "step": 24320
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.1890724411206607e-05,
+      "loss": 0.3735,
+      "step": 24330
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.188737582319455e-05,
+      "loss": 0.1929,
+      "step": 24340
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.18840272351825e-05,
+      "loss": 0.142,
+      "step": 24350
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.1880678647170443e-05,
+      "loss": 0.1247,
+      "step": 24360
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.187733005915839e-05,
+      "loss": 0.2285,
+      "step": 24370
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.1873981471146335e-05,
+      "loss": 0.2086,
+      "step": 24380
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.187063288313428e-05,
+      "loss": 0.2511,
+      "step": 24390
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.1867284295122223e-05,
+      "loss": 0.2063,
+      "step": 24400
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.186393570711017e-05,
+      "loss": 0.2732,
+      "step": 24410
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.1860587119098114e-05,
+      "loss": 0.1598,
+      "step": 24420
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.185723853108606e-05,
+      "loss": 0.1615,
+      "step": 24430
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.1853889943074006e-05,
+      "loss": 0.2813,
+      "step": 24440
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.185054135506195e-05,
+      "loss": 0.1326,
+      "step": 24450
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.1847192767049894e-05,
+      "loss": 0.334,
+      "step": 24460
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.1843844179037838e-05,
+      "loss": 0.266,
+      "step": 24470
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.1840495591025786e-05,
+      "loss": 0.1525,
+      "step": 24480
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.183714700301373e-05,
+      "loss": 0.2563,
+      "step": 24490
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.1833798415001674e-05,
+      "loss": 0.1608,
+      "step": 24500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.183044982698962e-05,
+      "loss": 0.1969,
+      "step": 24510
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.1827101238977565e-05,
+      "loss": 0.2693,
+      "step": 24520
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.182375265096551e-05,
+      "loss": 0.249,
+      "step": 24530
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.1820404062953457e-05,
+      "loss": 0.1888,
+      "step": 24540
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.18170554749414e-05,
+      "loss": 0.2031,
+      "step": 24550
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.1813706886929345e-05,
+      "loss": 0.2009,
+      "step": 24560
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.181035829891729e-05,
+      "loss": 0.1507,
+      "step": 24570
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.1807009710905236e-05,
+      "loss": 0.1905,
+      "step": 24580
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.180366112289318e-05,
+      "loss": 0.2415,
+      "step": 24590
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.1800312534881125e-05,
+      "loss": 0.1732,
+      "step": 24600
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.1796963946869072e-05,
+      "loss": 0.1495,
+      "step": 24610
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.1793615358857016e-05,
+      "loss": 0.1491,
+      "step": 24620
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.179026677084496e-05,
+      "loss": 0.1485,
+      "step": 24630
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.1786918182832904e-05,
+      "loss": 0.4126,
+      "step": 24640
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.1783569594820852e-05,
+      "loss": 0.215,
+      "step": 24650
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.1780221006808796e-05,
+      "loss": 0.1795,
+      "step": 24660
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.1776872418796743e-05,
+      "loss": 0.2362,
+      "step": 24670
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.1773523830784687e-05,
+      "loss": 0.1639,
+      "step": 24680
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.177017524277263e-05,
+      "loss": 0.2054,
+      "step": 24690
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.1766826654760575e-05,
+      "loss": 0.2373,
+      "step": 24700
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.176347806674852e-05,
+      "loss": 0.1793,
+      "step": 24710
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1760129478736467e-05,
+      "loss": 0.3513,
+      "step": 24720
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.175678089072441e-05,
+      "loss": 0.5214,
+      "step": 24730
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.175343230271236e-05,
+      "loss": 0.2454,
+      "step": 24740
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1750083714700303e-05,
+      "loss": 0.2219,
+      "step": 24750
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1746735126688247e-05,
+      "loss": 0.105,
+      "step": 24760
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.174338653867619e-05,
+      "loss": 0.1511,
+      "step": 24770
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1740037950664135e-05,
+      "loss": 0.3439,
+      "step": 24780
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1736689362652082e-05,
+      "loss": 0.3402,
+      "step": 24790
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.173334077464003e-05,
+      "loss": 0.1523,
+      "step": 24800
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1729992186627974e-05,
+      "loss": 0.2296,
+      "step": 24810
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1726643598615918e-05,
+      "loss": 0.3003,
+      "step": 24820
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1723295010603862e-05,
+      "loss": 0.2025,
+      "step": 24830
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1719946422591806e-05,
+      "loss": 0.2653,
+      "step": 24840
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.171659783457975e-05,
+      "loss": 0.2304,
+      "step": 24850
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1713249246567698e-05,
+      "loss": 0.1346,
+      "step": 24860
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1709900658555645e-05,
+      "loss": 0.2144,
+      "step": 24870
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.170655207054359e-05,
+      "loss": 0.1431,
+      "step": 24880
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1703203482531533e-05,
+      "loss": 0.4656,
+      "step": 24890
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1699854894519477e-05,
+      "loss": 0.1993,
+      "step": 24900
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.169650630650742e-05,
+      "loss": 0.2788,
+      "step": 24910
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.169315771849537e-05,
+      "loss": 0.1651,
+      "step": 24920
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1689809130483316e-05,
+      "loss": 0.1733,
+      "step": 24930
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.168646054247126e-05,
+      "loss": 0.1465,
+      "step": 24940
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1683111954459204e-05,
+      "loss": 0.1043,
+      "step": 24950
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.167976336644715e-05,
+      "loss": 0.147,
+      "step": 24960
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1676414778435093e-05,
+      "loss": 0.1306,
+      "step": 24970
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1673066190423037e-05,
+      "loss": 0.1206,
+      "step": 24980
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1669717602410984e-05,
+      "loss": 0.116,
+      "step": 24990
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.166636901439893e-05,
+      "loss": 0.1258,
+      "step": 25000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1663020426386876e-05,
+      "loss": 0.1781,
+      "step": 25010
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.165967183837482e-05,
+      "loss": 0.1357,
+      "step": 25020
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1656323250362764e-05,
+      "loss": 0.1539,
+      "step": 25030
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1652974662350708e-05,
+      "loss": 0.3825,
+      "step": 25040
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1649626074338652e-05,
+      "loss": 0.1573,
+      "step": 25050
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.16462774863266e-05,
+      "loss": 0.2451,
+      "step": 25060
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1642928898314547e-05,
+      "loss": 0.3092,
+      "step": 25070
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.163958031030249e-05,
+      "loss": 0.1358,
+      "step": 25080
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1636231722290435e-05,
+      "loss": 0.2046,
+      "step": 25090
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.163288313427838e-05,
+      "loss": 0.1677,
+      "step": 25100
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1629534546266323e-05,
+      "loss": 0.2233,
+      "step": 25110
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1626185958254267e-05,
+      "loss": 0.1267,
+      "step": 25120
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1622837370242218e-05,
+      "loss": 0.1865,
+      "step": 25130
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1619488782230162e-05,
+      "loss": 0.1549,
+      "step": 25140
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1616140194218106e-05,
+      "loss": 0.1614,
+      "step": 25150
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.161279160620605e-05,
+      "loss": 0.227,
+      "step": 25160
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1609443018193994e-05,
+      "loss": 0.1964,
+      "step": 25170
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.160609443018194e-05,
+      "loss": 0.2478,
+      "step": 25180
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1602745842169886e-05,
+      "loss": 0.2715,
+      "step": 25190
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1599397254157833e-05,
+      "loss": 0.2117,
+      "step": 25200
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1596048666145777e-05,
+      "loss": 0.1848,
+      "step": 25210
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.159270007813372e-05,
+      "loss": 0.282,
+      "step": 25220
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1589351490121666e-05,
+      "loss": 0.2752,
+      "step": 25230
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.158600290210961e-05,
+      "loss": 0.3509,
+      "step": 25240
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1582654314097554e-05,
+      "loss": 0.1844,
+      "step": 25250
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.15793057260855e-05,
+      "loss": 0.1808,
+      "step": 25260
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.157595713807345e-05,
+      "loss": 0.1573,
+      "step": 25270
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1572608550061393e-05,
+      "loss": 0.194,
+      "step": 25280
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1569259962049337e-05,
+      "loss": 0.2111,
+      "step": 25290
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.156591137403728e-05,
+      "loss": 0.428,
+      "step": 25300
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1562562786025225e-05,
+      "loss": 0.1769,
+      "step": 25310
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.155921419801317e-05,
+      "loss": 0.1019,
+      "step": 25320
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1555865610001117e-05,
+      "loss": 0.1909,
+      "step": 25330
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.1552517021989064e-05,
+      "loss": 0.3787,
+      "step": 25340
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.1549168433977008e-05,
+      "loss": 0.1302,
+      "step": 25350
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.1545819845964952e-05,
+      "loss": 0.2521,
+      "step": 25360
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.1542471257952896e-05,
+      "loss": 0.1866,
+      "step": 25370
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.153912266994084e-05,
+      "loss": 0.1919,
+      "step": 25380
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.1535774081928788e-05,
+      "loss": 0.185,
+      "step": 25390
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.1532425493916732e-05,
+      "loss": 0.1272,
+      "step": 25400
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.152907690590468e-05,
+      "loss": 0.1803,
+      "step": 25410
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.1525728317892623e-05,
+      "loss": 0.2311,
+      "step": 25420
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.1522379729880567e-05,
+      "loss": 0.1638,
+      "step": 25430
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.151903114186851e-05,
+      "loss": 0.21,
+      "step": 25440
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.1515682553856456e-05,
+      "loss": 0.135,
+      "step": 25450
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.1512333965844403e-05,
+      "loss": 0.3393,
+      "step": 25460
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.1508985377832347e-05,
+      "loss": 0.2452,
+      "step": 25470
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.1505636789820295e-05,
+      "loss": 0.2429,
+      "step": 25480
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.150228820180824e-05,
+      "loss": 0.1944,
+      "step": 25490
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.1498939613796183e-05,
+      "loss": 0.2104,
+      "step": 25500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.1495591025784127e-05,
+      "loss": 0.139,
+      "step": 25510
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.1492242437772074e-05,
+      "loss": 0.2061,
+      "step": 25520
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.148889384976002e-05,
+      "loss": 0.1461,
+      "step": 25530
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.1485545261747962e-05,
+      "loss": 0.1508,
+      "step": 25540
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.148219667373591e-05,
+      "loss": 0.1823,
+      "step": 25550
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.1478848085723854e-05,
+      "loss": 0.1182,
+      "step": 25560
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.1475499497711798e-05,
+      "loss": 0.2063,
+      "step": 25570
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.1472150909699742e-05,
+      "loss": 0.1576,
+      "step": 25580
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.146880232168769e-05,
+      "loss": 0.2196,
+      "step": 25590
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.1465453733675634e-05,
+      "loss": 0.1601,
+      "step": 25600
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1462105145663578e-05,
+      "loss": 0.1576,
+      "step": 25610
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1458756557651525e-05,
+      "loss": 0.2034,
+      "step": 25620
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.145540796963947e-05,
+      "loss": 0.1377,
+      "step": 25630
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1452059381627413e-05,
+      "loss": 0.156,
+      "step": 25640
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.144871079361536e-05,
+      "loss": 0.1087,
+      "step": 25650
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1445362205603305e-05,
+      "loss": 0.1452,
+      "step": 25660
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.144201361759125e-05,
+      "loss": 0.194,
+      "step": 25670
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1438665029579193e-05,
+      "loss": 0.227,
+      "step": 25680
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.143531644156714e-05,
+      "loss": 0.2041,
+      "step": 25690
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1431967853555085e-05,
+      "loss": 0.2211,
+      "step": 25700
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.142861926554303e-05,
+      "loss": 0.0834,
+      "step": 25710
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1425270677530976e-05,
+      "loss": 0.219,
+      "step": 25720
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.142192208951892e-05,
+      "loss": 0.1657,
+      "step": 25730
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1418573501506864e-05,
+      "loss": 0.204,
+      "step": 25740
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1415224913494808e-05,
+      "loss": 0.1188,
+      "step": 25750
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1411876325482756e-05,
+      "loss": 0.1856,
+      "step": 25760
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.14085277374707e-05,
+      "loss": 0.1969,
+      "step": 25770
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.140551400825985e-05,
+      "loss": 0.2362,
+      "step": 25780
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1402165420247795e-05,
+      "loss": 0.2292,
+      "step": 25790
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1398816832235742e-05,
+      "loss": 0.1978,
+      "step": 25800
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1395468244223686e-05,
+      "loss": 0.2654,
+      "step": 25810
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.139211965621163e-05,
+      "loss": 0.3282,
+      "step": 25820
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1388771068199578e-05,
+      "loss": 0.1583,
+      "step": 25830
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1385422480187522e-05,
+      "loss": 0.1362,
+      "step": 25840
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1382073892175466e-05,
+      "loss": 0.1169,
+      "step": 25850
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.137872530416341e-05,
+      "loss": 0.1969,
+      "step": 25860
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1375376716151357e-05,
+      "loss": 0.1625,
+      "step": 25870
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.13720281281393e-05,
+      "loss": 0.198,
+      "step": 25880
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.136867954012725e-05,
+      "loss": 0.1867,
+      "step": 25890
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.1365330952115193e-05,
+      "loss": 0.607,
+      "step": 25900
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.1361982364103137e-05,
+      "loss": 0.1897,
+      "step": 25910
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.135863377609108e-05,
+      "loss": 0.1637,
+      "step": 25920
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.1355285188079025e-05,
+      "loss": 0.3298,
+      "step": 25930
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.1351936600066973e-05,
+      "loss": 0.3816,
+      "step": 25940
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.1348588012054917e-05,
+      "loss": 0.1431,
+      "step": 25950
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.1345239424042864e-05,
+      "loss": 0.2491,
+      "step": 25960
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.1341890836030808e-05,
+      "loss": 0.1317,
+      "step": 25970
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.1338542248018752e-05,
+      "loss": 0.1697,
+      "step": 25980
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.1335193660006696e-05,
+      "loss": 0.2147,
+      "step": 25990
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.133184507199464e-05,
+      "loss": 0.1385,
+      "step": 26000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.1328496483982588e-05,
+      "loss": 0.1776,
+      "step": 26010
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.1325147895970535e-05,
+      "loss": 0.1717,
+      "step": 26020
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.132179930795848e-05,
+      "loss": 0.1722,
+      "step": 26030
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.1318450719946424e-05,
+      "loss": 0.1728,
+      "step": 26040
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.1315102131934368e-05,
+      "loss": 0.1115,
+      "step": 26050
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.131175354392231e-05,
+      "loss": 0.1558,
+      "step": 26060
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.1308404955910256e-05,
+      "loss": 0.2075,
+      "step": 26070
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.1305056367898203e-05,
+      "loss": 0.1988,
+      "step": 26080
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.130170777988615e-05,
+      "loss": 0.1653,
+      "step": 26090
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.1298359191874095e-05,
+      "loss": 0.1746,
+      "step": 26100
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.129501060386204e-05,
+      "loss": 0.1651,
+      "step": 26110
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.129199687465119e-05,
+      "loss": 0.1627,
+      "step": 26120
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.1288648286639134e-05,
+      "loss": 0.1835,
+      "step": 26130
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.128529969862708e-05,
+      "loss": 0.1361,
+      "step": 26140
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.1281951110615025e-05,
+      "loss": 0.3414,
+      "step": 26150
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.127860252260297e-05,
+      "loss": 0.1636,
+      "step": 26160
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.1275253934590913e-05,
+      "loss": 0.2356,
+      "step": 26170
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.1271905346578857e-05,
+      "loss": 0.164,
+      "step": 26180
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.1268556758566805e-05,
+      "loss": 0.2482,
+      "step": 26190
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.1265208170554752e-05,
+      "loss": 0.1571,
+      "step": 26200
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.1261859582542696e-05,
+      "loss": 0.1834,
+      "step": 26210
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.125851099453064e-05,
+      "loss": 0.2263,
+      "step": 26220
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.1255162406518584e-05,
+      "loss": 0.1769,
+      "step": 26230
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.125181381850653e-05,
+      "loss": 0.2147,
+      "step": 26240
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.1248465230494473e-05,
+      "loss": 0.1037,
+      "step": 26250
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.124511664248242e-05,
+      "loss": 0.1638,
+      "step": 26260
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.1241768054470368e-05,
+      "loss": 0.221,
+      "step": 26270
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.123841946645831e-05,
+      "loss": 0.1333,
+      "step": 26280
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.1235070878446256e-05,
+      "loss": 0.1495,
+      "step": 26290
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.12317222904342e-05,
+      "loss": 0.1731,
+      "step": 26300
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.1228373702422144e-05,
+      "loss": 0.1774,
+      "step": 26310
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.1225025114410088e-05,
+      "loss": 0.2397,
+      "step": 26320
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.122167652639804e-05,
+      "loss": 0.6903,
+      "step": 26330
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1218327938385983e-05,
+      "loss": 0.3114,
+      "step": 26340
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1214979350373927e-05,
+      "loss": 0.1671,
+      "step": 26350
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.121163076236187e-05,
+      "loss": 0.1716,
+      "step": 26360
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1208282174349815e-05,
+      "loss": 0.0964,
+      "step": 26370
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.120493358633776e-05,
+      "loss": 0.1484,
+      "step": 26380
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1201584998325707e-05,
+      "loss": 0.2494,
+      "step": 26390
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1198236410313654e-05,
+      "loss": 0.108,
+      "step": 26400
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1194887822301598e-05,
+      "loss": 0.1878,
+      "step": 26410
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.1191539234289542e-05,
+      "loss": 0.1816,
+      "step": 26420
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.1188190646277486e-05,
+      "loss": 0.378,
+      "step": 26430
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.118484205826543e-05,
+      "loss": 0.4913,
+      "step": 26440
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.1181493470253374e-05,
+      "loss": 0.1823,
+      "step": 26450
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.1178144882241325e-05,
+      "loss": 0.1577,
+      "step": 26460
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.117479629422927e-05,
+      "loss": 0.2493,
+      "step": 26470
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.1171447706217213e-05,
+      "loss": 0.1331,
+      "step": 26480
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.1168099118205158e-05,
+      "loss": 0.1928,
+      "step": 26490
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.11647505301931e-05,
+      "loss": 0.1214,
+      "step": 26500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.1161401942181046e-05,
+      "loss": 0.1381,
+      "step": 26510
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.115805335416899e-05,
+      "loss": 0.2226,
+      "step": 26520
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.115470476615694e-05,
+      "loss": 0.1473,
+      "step": 26530
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.1151356178144885e-05,
+      "loss": 0.1829,
+      "step": 26540
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.114800759013283e-05,
+      "loss": 0.1775,
+      "step": 26550
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.1144659002120773e-05,
+      "loss": 0.1657,
+      "step": 26560
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.1141310414108717e-05,
+      "loss": 0.1333,
+      "step": 26570
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.113796182609666e-05,
+      "loss": 0.142,
+      "step": 26580
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.113461323808461e-05,
+      "loss": 0.1557,
+      "step": 26590
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.1131264650072556e-05,
+      "loss": 0.157,
+      "step": 26600
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.11279160620605e-05,
+      "loss": 0.1785,
+      "step": 26610
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.1124567474048444e-05,
+      "loss": 0.1258,
+      "step": 26620
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.1121218886036388e-05,
+      "loss": 0.1971,
+      "step": 26630
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.1117870298024332e-05,
+      "loss": 0.1092,
+      "step": 26640
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.1114521710012276e-05,
+      "loss": 0.1333,
+      "step": 26650
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.1111173122000224e-05,
+      "loss": 0.18,
+      "step": 26660
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.110782453398817e-05,
+      "loss": 0.1298,
+      "step": 26670
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.1104475945976115e-05,
+      "loss": 0.1575,
+      "step": 26680
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.110112735796406e-05,
+      "loss": 0.157,
+      "step": 26690
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.1097778769952003e-05,
+      "loss": 0.1595,
+      "step": 26700
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.1094430181939947e-05,
+      "loss": 0.1455,
+      "step": 26710
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.1091081593927895e-05,
+      "loss": 0.2401,
+      "step": 26720
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.108773300591584e-05,
+      "loss": 0.2469,
+      "step": 26730
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.1084384417903786e-05,
+      "loss": 0.2636,
+      "step": 26740
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.108103582989173e-05,
+      "loss": 0.2341,
+      "step": 26750
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.1077687241879675e-05,
+      "loss": 0.2409,
+      "step": 26760
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.107433865386762e-05,
+      "loss": 0.1973,
+      "step": 26770
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.1070990065855563e-05,
+      "loss": 0.2077,
+      "step": 26780
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.106764147784351e-05,
+      "loss": 0.1884,
+      "step": 26790
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.1064292889831454e-05,
+      "loss": 0.0945,
+      "step": 26800
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.1060944301819402e-05,
+      "loss": 0.4159,
+      "step": 26810
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.1057595713807346e-05,
+      "loss": 0.3181,
+      "step": 26820
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.105424712579529e-05,
+      "loss": 0.1017,
+      "step": 26830
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.1050898537783234e-05,
+      "loss": 0.1997,
+      "step": 26840
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.104754994977118e-05,
+      "loss": 0.2078,
+      "step": 26850
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.1044201361759126e-05,
+      "loss": 0.1966,
+      "step": 26860
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.104085277374707e-05,
+      "loss": 0.111,
+      "step": 26870
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.1037504185735017e-05,
+      "loss": 0.1695,
+      "step": 26880
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.103415559772296e-05,
+      "loss": 0.1949,
+      "step": 26890
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.1030807009710905e-05,
+      "loss": 0.1888,
+      "step": 26900
+    },
+    {
+      "epoch": 3.0,
+      "eval_loss": 0.3188088834285736,
+      "eval_runtime": 251.2742,
+      "eval_samples_per_second": 7.936,
+      "eval_steps_per_second": 3.968,
+      "eval_wer": 0.13169486891271986,
+      "step": 26907
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.102745842169885e-05,
+      "loss": 0.1339,
+      "step": 26910
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.1024109833686797e-05,
+      "loss": 0.1373,
+      "step": 26920
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.102076124567474e-05,
+      "loss": 0.1374,
+      "step": 26930
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.1017412657662685e-05,
+      "loss": 0.1768,
+      "step": 26940
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.1014064069650632e-05,
+      "loss": 0.4,
+      "step": 26950
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.1010715481638576e-05,
+      "loss": 0.318,
+      "step": 26960
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.100736689362652e-05,
+      "loss": 0.0908,
+      "step": 26970
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.1004018305614468e-05,
+      "loss": 0.0875,
+      "step": 26980
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.1000669717602412e-05,
+      "loss": 0.1689,
+      "step": 26990
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.0997321129590356e-05,
+      "loss": 0.1132,
+      "step": 27000
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.09939725415783e-05,
+      "loss": 0.2781,
+      "step": 27010
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.0990623953566248e-05,
+      "loss": 0.1432,
+      "step": 27020
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.0987275365554192e-05,
+      "loss": 0.1347,
+      "step": 27030
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.0983926777542136e-05,
+      "loss": 0.1426,
+      "step": 27040
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.0980578189530083e-05,
+      "loss": 0.086,
+      "step": 27050
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.0977229601518027e-05,
+      "loss": 0.1678,
+      "step": 27060
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.097388101350597e-05,
+      "loss": 0.0734,
+      "step": 27070
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.0970532425493915e-05,
+      "loss": 0.139,
+      "step": 27080
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.0967183837481863e-05,
+      "loss": 0.1456,
+      "step": 27090
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.0963835249469807e-05,
+      "loss": 0.1254,
+      "step": 27100
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.0960486661457754e-05,
+      "loss": 0.2568,
+      "step": 27110
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.09571380734457e-05,
+      "loss": 0.2235,
+      "step": 27120
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.0953789485433643e-05,
+      "loss": 0.1723,
+      "step": 27130
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.0950440897421587e-05,
+      "loss": 0.147,
+      "step": 27140
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.094709230940953e-05,
+      "loss": 0.1458,
+      "step": 27150
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.0943743721397478e-05,
+      "loss": 0.2285,
+      "step": 27160
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.0940395133385422e-05,
+      "loss": 0.161,
+      "step": 27170
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.093704654537337e-05,
+      "loss": 0.1223,
+      "step": 27180
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.0933697957361314e-05,
+      "loss": 0.0889,
+      "step": 27190
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.0930349369349258e-05,
+      "loss": 0.4228,
+      "step": 27200
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.0927000781337202e-05,
+      "loss": 0.2424,
+      "step": 27210
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.0923652193325146e-05,
+      "loss": 0.1368,
+      "step": 27220
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.0920303605313094e-05,
+      "loss": 0.0819,
+      "step": 27230
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.0916955017301038e-05,
+      "loss": 0.1603,
+      "step": 27240
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.0913606429288985e-05,
+      "loss": 0.1289,
+      "step": 27250
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.091025784127693e-05,
+      "loss": 0.2089,
+      "step": 27260
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.0906909253264873e-05,
+      "loss": 0.2061,
+      "step": 27270
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.0903560665252817e-05,
+      "loss": 0.5325,
+      "step": 27280
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.0900212077240765e-05,
+      "loss": 0.1256,
+      "step": 27290
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.089686348922871e-05,
+      "loss": 0.111,
+      "step": 27300
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.0893514901216656e-05,
+      "loss": 0.168,
+      "step": 27310
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.08901663132046e-05,
+      "loss": 0.154,
+      "step": 27320
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.0886817725192544e-05,
+      "loss": 0.1958,
+      "step": 27330
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.088346913718049e-05,
+      "loss": 0.187,
+      "step": 27340
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.0880120549168433e-05,
+      "loss": 0.1106,
+      "step": 27350
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.087677196115638e-05,
+      "loss": 0.1601,
+      "step": 27360
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.0873423373144324e-05,
+      "loss": 0.1238,
+      "step": 27370
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.087007478513227e-05,
+      "loss": 0.1413,
+      "step": 27380
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.0866726197120216e-05,
+      "loss": 0.1194,
+      "step": 27390
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.086337760910816e-05,
+      "loss": 0.2071,
+      "step": 27400
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.0860029021096104e-05,
+      "loss": 0.1568,
+      "step": 27410
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.0856680433084048e-05,
+      "loss": 0.1876,
+      "step": 27420
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.0853331845071995e-05,
+      "loss": 0.1116,
+      "step": 27430
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.0849983257059943e-05,
+      "loss": 0.1218,
+      "step": 27440
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.0846634669047887e-05,
+      "loss": 0.1099,
+      "step": 27450
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.084328608103583e-05,
+      "loss": 0.1369,
+      "step": 27460
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.0839937493023775e-05,
+      "loss": 0.1044,
+      "step": 27470
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.083658890501172e-05,
+      "loss": 0.1588,
+      "step": 27480
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.0833240316999663e-05,
+      "loss": 0.1981,
+      "step": 27490
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.082989172898761e-05,
+      "loss": 0.1275,
+      "step": 27500
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.0826543140975558e-05,
+      "loss": 0.175,
+      "step": 27510
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.0823194552963502e-05,
+      "loss": 0.0617,
+      "step": 27520
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.0819845964951446e-05,
+      "loss": 0.2889,
+      "step": 27530
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.081649737693939e-05,
+      "loss": 0.1333,
+      "step": 27540
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.0813148788927334e-05,
+      "loss": 0.2069,
+      "step": 27550
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.080980020091528e-05,
+      "loss": 0.1543,
+      "step": 27560
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.080645161290323e-05,
+      "loss": 0.1231,
+      "step": 27570
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.0803103024891173e-05,
+      "loss": 0.2006,
+      "step": 27580
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.0799754436879117e-05,
+      "loss": 0.1136,
+      "step": 27590
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.079640584886706e-05,
+      "loss": 0.1565,
+      "step": 27600
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.0793057260855006e-05,
+      "loss": 0.2312,
+      "step": 27610
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.078970867284295e-05,
+      "loss": 0.2134,
+      "step": 27620
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.0786360084830894e-05,
+      "loss": 0.0643,
+      "step": 27630
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.0783011496818845e-05,
+      "loss": 0.0985,
+      "step": 27640
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.077966290880679e-05,
+      "loss": 0.1546,
+      "step": 27650
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.0776314320794733e-05,
+      "loss": 0.2978,
+      "step": 27660
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.0772965732782677e-05,
+      "loss": 0.208,
+      "step": 27670
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.076961714477062e-05,
+      "loss": 0.1466,
+      "step": 27680
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.0766268556758565e-05,
+      "loss": 0.0907,
+      "step": 27690
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.0762919968746512e-05,
+      "loss": 0.1317,
+      "step": 27700
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.075957138073446e-05,
+      "loss": 0.2002,
+      "step": 27710
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.0756222792722404e-05,
+      "loss": 0.1234,
+      "step": 27720
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.0752874204710348e-05,
+      "loss": 0.2674,
+      "step": 27730
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.0749525616698292e-05,
+      "loss": 0.1344,
+      "step": 27740
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.0746177028686236e-05,
+      "loss": 0.0885,
+      "step": 27750
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.074282844067418e-05,
+      "loss": 0.1551,
+      "step": 27760
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.0739479852662128e-05,
+      "loss": 0.0959,
+      "step": 27770
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.0736131264650075e-05,
+      "loss": 0.1044,
+      "step": 27780
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.073278267663802e-05,
+      "loss": 0.1452,
+      "step": 27790
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.0729434088625963e-05,
+      "loss": 0.1135,
+      "step": 27800
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.0726085500613907e-05,
+      "loss": 0.2589,
+      "step": 27810
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.072273691260185e-05,
+      "loss": 0.0978,
+      "step": 27820
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.07193883245898e-05,
+      "loss": 0.3705,
+      "step": 27830
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.0716039736577743e-05,
+      "loss": 0.091,
+      "step": 27840
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.071269114856569e-05,
+      "loss": 0.2245,
+      "step": 27850
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.0709342560553635e-05,
+      "loss": 0.252,
+      "step": 27860
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.070599397254158e-05,
+      "loss": 0.133,
+      "step": 27870
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.0702645384529523e-05,
+      "loss": 0.1227,
+      "step": 27880
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.0699296796517467e-05,
+      "loss": 0.11,
+      "step": 27890
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.0695948208505414e-05,
+      "loss": 0.1314,
+      "step": 27900
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.069259962049336e-05,
+      "loss": 0.1643,
+      "step": 27910
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.0689251032481306e-05,
+      "loss": 0.0934,
+      "step": 27920
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.068590244446925e-05,
+      "loss": 0.1108,
+      "step": 27930
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.0682553856457194e-05,
+      "loss": 0.1976,
+      "step": 27940
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.0679205268445138e-05,
+      "loss": 0.0938,
+      "step": 27950
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.0675856680433085e-05,
+      "loss": 0.2034,
+      "step": 27960
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.067250809242103e-05,
+      "loss": 0.284,
+      "step": 27970
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.0669159504408974e-05,
+      "loss": 0.1616,
+      "step": 27980
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.066581091639692e-05,
+      "loss": 0.1253,
+      "step": 27990
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.0662462328384865e-05,
+      "loss": 0.1331,
+      "step": 28000
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.065911374037281e-05,
+      "loss": 0.1568,
+      "step": 28010
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.0655765152360753e-05,
+      "loss": 0.1467,
+      "step": 28020
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.06524165643487e-05,
+      "loss": 0.1599,
+      "step": 28030
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.0649067976336645e-05,
+      "loss": 0.2492,
+      "step": 28040
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.064571938832459e-05,
+      "loss": 0.1199,
+      "step": 28050
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.0642370800312536e-05,
+      "loss": 0.1284,
+      "step": 28060
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.063902221230048e-05,
+      "loss": 0.0941,
+      "step": 28070
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.0635673624288425e-05,
+      "loss": 0.1549,
+      "step": 28080
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.0632325036276372e-05,
+      "loss": 0.1766,
+      "step": 28090
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.0628976448264316e-05,
+      "loss": 0.1702,
+      "step": 28100
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.062562786025226e-05,
+      "loss": 0.118,
+      "step": 28110
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.0622279272240204e-05,
+      "loss": 0.1551,
+      "step": 28120
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.0618930684228152e-05,
+      "loss": 0.0685,
+      "step": 28130
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.0615582096216096e-05,
+      "loss": 0.0778,
+      "step": 28140
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.061223350820404e-05,
+      "loss": 0.2843,
+      "step": 28150
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.0608884920191987e-05,
+      "loss": 0.1885,
+      "step": 28160
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.060553633217993e-05,
+      "loss": 0.1935,
+      "step": 28170
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.0602187744167875e-05,
+      "loss": 0.1093,
+      "step": 28180
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.0598839156155823e-05,
+      "loss": 0.146,
+      "step": 28190
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.0595490568143767e-05,
+      "loss": 0.1356,
+      "step": 28200
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.059214198013171e-05,
+      "loss": 0.148,
+      "step": 28210
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.058879339211966e-05,
+      "loss": 0.0884,
+      "step": 28220
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.0585444804107603e-05,
+      "loss": 0.1463,
+      "step": 28230
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.0582096216095547e-05,
+      "loss": 0.1761,
+      "step": 28240
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.057874762808349e-05,
+      "loss": 0.2322,
+      "step": 28250
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.0575399040071438e-05,
+      "loss": 0.3744,
+      "step": 28260
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.0572050452059382e-05,
+      "loss": 0.1312,
+      "step": 28270
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.0568701864047326e-05,
+      "loss": 0.0667,
+      "step": 28280
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.0565353276035274e-05,
+      "loss": 0.1499,
+      "step": 28290
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.0562004688023218e-05,
+      "loss": 0.1114,
+      "step": 28300
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.0558656100011162e-05,
+      "loss": 0.2045,
+      "step": 28310
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.0555307511999106e-05,
+      "loss": 0.1264,
+      "step": 28320
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.0551958923987054e-05,
+      "loss": 0.1669,
+      "step": 28330
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.0548610335974998e-05,
+      "loss": 0.1867,
+      "step": 28340
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.0545261747962945e-05,
+      "loss": 0.3523,
+      "step": 28350
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.054191315995089e-05,
+      "loss": 0.2239,
+      "step": 28360
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.0538564571938833e-05,
+      "loss": 0.1261,
+      "step": 28370
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.0535215983926777e-05,
+      "loss": 0.2141,
+      "step": 28380
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.053186739591472e-05,
+      "loss": 0.2099,
+      "step": 28390
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.052851880790267e-05,
+      "loss": 0.192,
+      "step": 28400
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.0525170219890613e-05,
+      "loss": 0.2126,
+      "step": 28410
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.052182163187856e-05,
+      "loss": 0.0788,
+      "step": 28420
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.0518473043866504e-05,
+      "loss": 0.1467,
+      "step": 28430
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.051512445585445e-05,
+      "loss": 0.2266,
+      "step": 28440
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.0511775867842393e-05,
+      "loss": 0.1236,
+      "step": 28450
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.0508427279830337e-05,
+      "loss": 0.2523,
+      "step": 28460
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.0505078691818284e-05,
+      "loss": 0.2565,
+      "step": 28470
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.050173010380623e-05,
+      "loss": 0.1897,
+      "step": 28480
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.0498381515794176e-05,
+      "loss": 0.151,
+      "step": 28490
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.049503292778212e-05,
+      "loss": 0.112,
+      "step": 28500
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.0491684339770064e-05,
+      "loss": 0.2862,
+      "step": 28510
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.0488335751758008e-05,
+      "loss": 0.2074,
+      "step": 28520
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.0484987163745952e-05,
+      "loss": 0.1077,
+      "step": 28530
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.04816385757339e-05,
+      "loss": 0.125,
+      "step": 28540
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.0478289987721847e-05,
+      "loss": 0.187,
+      "step": 28550
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.047494139970979e-05,
+      "loss": 0.1296,
+      "step": 28560
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.0471592811697735e-05,
+      "loss": 0.1291,
+      "step": 28570
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.046824422368568e-05,
+      "loss": 0.1859,
+      "step": 28580
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.0464895635673623e-05,
+      "loss": 0.1552,
+      "step": 28590
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.0461547047661567e-05,
+      "loss": 0.1307,
+      "step": 28600
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.0458198459649518e-05,
+      "loss": 0.1641,
+      "step": 28610
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.0454849871637462e-05,
+      "loss": 0.1052,
+      "step": 28620
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.0451501283625406e-05,
+      "loss": 0.15,
+      "step": 28630
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.044815269561335e-05,
+      "loss": 0.1132,
+      "step": 28640
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.0444804107601294e-05,
+      "loss": 0.1246,
+      "step": 28650
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.044145551958924e-05,
+      "loss": 0.1361,
+      "step": 28660
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.0438106931577183e-05,
+      "loss": 0.1326,
+      "step": 28670
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.0434758343565133e-05,
+      "loss": 0.1157,
+      "step": 28680
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.0431409755553077e-05,
+      "loss": 0.1619,
+      "step": 28690
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.042806116754102e-05,
+      "loss": 0.208,
+      "step": 28700
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.0424712579528966e-05,
+      "loss": 0.1161,
+      "step": 28710
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.042136399151691e-05,
+      "loss": 0.111,
+      "step": 28720
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.0418015403504854e-05,
+      "loss": 0.0877,
+      "step": 28730
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.04146668154928e-05,
+      "loss": 0.1277,
+      "step": 28740
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.041131822748075e-05,
+      "loss": 0.2061,
+      "step": 28750
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.0407969639468693e-05,
+      "loss": 0.1828,
+      "step": 28760
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.0404621051456637e-05,
+      "loss": 0.2276,
+      "step": 28770
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.0401607322245784e-05,
+      "loss": 0.2052,
+      "step": 28780
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.0398258734233735e-05,
+      "loss": 0.105,
+      "step": 28790
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.039491014622168e-05,
+      "loss": 0.1126,
+      "step": 28800
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.0391561558209623e-05,
+      "loss": 0.2004,
+      "step": 28810
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.0388212970197567e-05,
+      "loss": 0.1727,
+      "step": 28820
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.038486438218551e-05,
+      "loss": 0.1387,
+      "step": 28830
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.0381515794173455e-05,
+      "loss": 0.128,
+      "step": 28840
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.03781672061614e-05,
+      "loss": 0.1276,
+      "step": 28850
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.037481861814935e-05,
+      "loss": 0.0997,
+      "step": 28860
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.0371470030137294e-05,
+      "loss": 0.2291,
+      "step": 28870
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.036812144212524e-05,
+      "loss": 0.1789,
+      "step": 28880
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.0364772854113182e-05,
+      "loss": 0.1487,
+      "step": 28890
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.0361424266101127e-05,
+      "loss": 0.1422,
+      "step": 28900
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.035807567808907e-05,
+      "loss": 0.1469,
+      "step": 28910
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.0354727090077018e-05,
+      "loss": 0.1097,
+      "step": 28920
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.0351378502064966e-05,
+      "loss": 0.1376,
+      "step": 28930
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.034802991405291e-05,
+      "loss": 0.1069,
+      "step": 28940
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.0344681326040854e-05,
+      "loss": 0.077,
+      "step": 28950
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.0341332738028798e-05,
+      "loss": 0.2882,
+      "step": 28960
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.0337984150016742e-05,
+      "loss": 0.2016,
+      "step": 28970
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.0334635562004686e-05,
+      "loss": 0.1523,
+      "step": 28980
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.0331286973992633e-05,
+      "loss": 0.1044,
+      "step": 28990
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.032793838598058e-05,
+      "loss": 0.1399,
+      "step": 29000
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.0324589797968525e-05,
+      "loss": 0.1887,
+      "step": 29010
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.032124120995647e-05,
+      "loss": 0.1078,
+      "step": 29020
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.0317892621944413e-05,
+      "loss": 0.1744,
+      "step": 29030
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.0314544033932357e-05,
+      "loss": 0.1909,
+      "step": 29040
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.0311195445920305e-05,
+      "loss": 0.0792,
+      "step": 29050
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.030784685790825e-05,
+      "loss": 0.1607,
+      "step": 29060
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.0304498269896196e-05,
+      "loss": 0.1079,
+      "step": 29070
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.030114968188414e-05,
+      "loss": 0.1838,
+      "step": 29080
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.0297801093872084e-05,
+      "loss": 0.1082,
+      "step": 29090
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.029445250586003e-05,
+      "loss": 0.1809,
+      "step": 29100
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.0291103917847972e-05,
+      "loss": 0.1675,
+      "step": 29110
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.028775532983592e-05,
+      "loss": 0.122,
+      "step": 29120
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.0284406741823864e-05,
+      "loss": 0.1081,
+      "step": 29130
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.028105815381181e-05,
+      "loss": 0.1414,
+      "step": 29140
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.0277709565799756e-05,
+      "loss": 0.2213,
+      "step": 29150
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.02743609777877e-05,
+      "loss": 0.1797,
+      "step": 29160
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.0271012389775644e-05,
+      "loss": 0.2934,
+      "step": 29170
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.026766380176359e-05,
+      "loss": 0.1565,
+      "step": 29180
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.0264315213751535e-05,
+      "loss": 0.1037,
+      "step": 29190
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.026096662573948e-05,
+      "loss": 0.1586,
+      "step": 29200
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.0257618037727427e-05,
+      "loss": 0.173,
+      "step": 29210
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.025426944971537e-05,
+      "loss": 0.1379,
+      "step": 29220
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.0250920861703315e-05,
+      "loss": 0.0628,
+      "step": 29230
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.024757227369126e-05,
+      "loss": 0.1261,
+      "step": 29240
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.0244223685679206e-05,
+      "loss": 0.1139,
+      "step": 29250
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.024087509766715e-05,
+      "loss": 0.1566,
+      "step": 29260
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.0237526509655095e-05,
+      "loss": 0.1394,
+      "step": 29270
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.0234177921643042e-05,
+      "loss": 0.172,
+      "step": 29280
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.0230829333630986e-05,
+      "loss": 0.1758,
+      "step": 29290
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.022748074561893e-05,
+      "loss": 0.3367,
+      "step": 29300
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.0224132157606878e-05,
+      "loss": 0.1692,
+      "step": 29310
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.0220783569594822e-05,
+      "loss": 0.2605,
+      "step": 29320
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.0217434981582766e-05,
+      "loss": 0.1417,
+      "step": 29330
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.021408639357071e-05,
+      "loss": 0.2018,
+      "step": 29340
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.0210737805558657e-05,
+      "loss": 0.1123,
+      "step": 29350
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.02073892175466e-05,
+      "loss": 0.235,
+      "step": 29360
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.0204040629534545e-05,
+      "loss": 0.1533,
+      "step": 29370
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.0200692041522493e-05,
+      "loss": 0.107,
+      "step": 29380
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.0197343453510437e-05,
+      "loss": 0.1639,
+      "step": 29390
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.019399486549838e-05,
+      "loss": 0.1657,
+      "step": 29400
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.019064627748633e-05,
+      "loss": 0.2175,
+      "step": 29410
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.0187297689474273e-05,
+      "loss": 0.106,
+      "step": 29420
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.0183949101462217e-05,
+      "loss": 0.1449,
+      "step": 29430
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.0180600513450164e-05,
+      "loss": 0.1222,
+      "step": 29440
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.0177251925438108e-05,
+      "loss": 0.1589,
+      "step": 29450
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.0173903337426052e-05,
+      "loss": 0.1442,
+      "step": 29460
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2.0170554749413996e-05,
+      "loss": 0.1126,
+      "step": 29470
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2.0167206161401944e-05,
+      "loss": 0.0789,
+      "step": 29480
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2.0163857573389888e-05,
+      "loss": 0.1311,
+      "step": 29490
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2.0160508985377832e-05,
+      "loss": 0.1079,
+      "step": 29500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2.015716039736578e-05,
+      "loss": 0.2888,
+      "step": 29510
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2.0153811809353724e-05,
+      "loss": 0.1572,
+      "step": 29520
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2.0150463221341668e-05,
+      "loss": 0.1495,
+      "step": 29530
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2.014711463332961e-05,
+      "loss": 0.219,
+      "step": 29540
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2.014376604531756e-05,
+      "loss": 0.2034,
+      "step": 29550
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2.0140417457305503e-05,
+      "loss": 0.1917,
+      "step": 29560
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2.013706886929345e-05,
+      "loss": 0.0974,
+      "step": 29570
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2.0133720281281395e-05,
+      "loss": 0.0976,
+      "step": 29580
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2.013037169326934e-05,
+      "loss": 0.2132,
+      "step": 29590
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2.0127023105257283e-05,
+      "loss": 0.2294,
+      "step": 29600
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2.0123674517245227e-05,
+      "loss": 0.178,
+      "step": 29610
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2.0120325929233174e-05,
+      "loss": 0.1425,
+      "step": 29620
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2.011697734122112e-05,
+      "loss": 0.1817,
+      "step": 29630
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2.0113628753209066e-05,
+      "loss": 0.1453,
+      "step": 29640
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2.011028016519701e-05,
+      "loss": 0.2023,
+      "step": 29650
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2.0106931577184954e-05,
+      "loss": 0.1862,
+      "step": 29660
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2.0103582989172898e-05,
+      "loss": 0.1998,
+      "step": 29670
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2.0100234401160842e-05,
+      "loss": 0.128,
+      "step": 29680
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2.009688581314879e-05,
+      "loss": 0.208,
+      "step": 29690
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2.0093537225136737e-05,
+      "loss": 0.0944,
+      "step": 29700
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2.009018863712468e-05,
+      "loss": 0.2334,
+      "step": 29710
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2.0086840049112625e-05,
+      "loss": 0.1063,
+      "step": 29720
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2.008349146110057e-05,
+      "loss": 0.1761,
+      "step": 29730
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2.0080142873088513e-05,
+      "loss": 0.129,
+      "step": 29740
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2.0076794285076458e-05,
+      "loss": 0.1485,
+      "step": 29750
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2.0073445697064405e-05,
+      "loss": 0.2183,
+      "step": 29760
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2.0070097109052352e-05,
+      "loss": 0.0972,
+      "step": 29770
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2.0066748521040297e-05,
+      "loss": 0.3114,
+      "step": 29780
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2.006339993302824e-05,
+      "loss": 0.1115,
+      "step": 29790
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2.0060051345016185e-05,
+      "loss": 0.1039,
+      "step": 29800
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2.005670275700413e-05,
+      "loss": 0.164,
+      "step": 29810
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2.0053354168992073e-05,
+      "loss": 0.1218,
+      "step": 29820
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2.0050005580980024e-05,
+      "loss": 0.2339,
+      "step": 29830
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2.0046656992967968e-05,
+      "loss": 0.1232,
+      "step": 29840
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2.0043308404955912e-05,
+      "loss": 0.0812,
+      "step": 29850
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2.0039959816943856e-05,
+      "loss": 0.2063,
+      "step": 29860
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2.00366112289318e-05,
+      "loss": 0.0994,
+      "step": 29870
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2.0033262640919744e-05,
+      "loss": 0.1408,
+      "step": 29880
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2.0029914052907688e-05,
+      "loss": 0.2861,
+      "step": 29890
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2.002656546489564e-05,
+      "loss": 0.1324,
+      "step": 29900
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2.0023216876883583e-05,
+      "loss": 0.2432,
+      "step": 29910
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2.0019868288871527e-05,
+      "loss": 0.1733,
+      "step": 29920
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2.001651970085947e-05,
+      "loss": 0.1256,
+      "step": 29930
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2.0013171112847415e-05,
+      "loss": 0.1675,
+      "step": 29940
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2.000982252483536e-05,
+      "loss": 0.1501,
+      "step": 29950
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2.0006473936823307e-05,
+      "loss": 0.1663,
+      "step": 29960
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2.0003125348811254e-05,
+      "loss": 0.1417,
+      "step": 29970
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.99997767607992e-05,
+      "loss": 0.1246,
+      "step": 29980
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.9996428172787142e-05,
+      "loss": 0.2153,
+      "step": 29990
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.9993079584775087e-05,
+      "loss": 0.1909,
+      "step": 30000
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.998973099676303e-05,
+      "loss": 0.1585,
+      "step": 30010
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.9986382408750975e-05,
+      "loss": 0.147,
+      "step": 30020
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.9983033820738922e-05,
+      "loss": 0.2469,
+      "step": 30030
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.997968523272687e-05,
+      "loss": 0.177,
+      "step": 30040
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.9976336644714814e-05,
+      "loss": 0.1689,
+      "step": 30050
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.9972988056702758e-05,
+      "loss": 0.207,
+      "step": 30060
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.9969639468690702e-05,
+      "loss": 0.1553,
+      "step": 30070
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.9966290880678646e-05,
+      "loss": 0.1713,
+      "step": 30080
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.9962942292666593e-05,
+      "loss": 0.1159,
+      "step": 30090
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.9959593704654537e-05,
+      "loss": 0.0888,
+      "step": 30100
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.9956245116642485e-05,
+      "loss": 0.1844,
+      "step": 30110
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.995289652863043e-05,
+      "loss": 0.1844,
+      "step": 30120
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.9949547940618373e-05,
+      "loss": 0.1261,
+      "step": 30130
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.9946199352606317e-05,
+      "loss": 0.1181,
+      "step": 30140
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.994285076459426e-05,
+      "loss": 0.1662,
+      "step": 30150
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.993950217658221e-05,
+      "loss": 0.131,
+      "step": 30160
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.9936153588570153e-05,
+      "loss": 0.1482,
+      "step": 30170
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.99328050005581e-05,
+      "loss": 0.1103,
+      "step": 30180
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.9929456412546044e-05,
+      "loss": 0.1293,
+      "step": 30190
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.992610782453399e-05,
+      "loss": 0.1554,
+      "step": 30200
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.9922759236521932e-05,
+      "loss": 0.4222,
+      "step": 30210
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.9919410648509876e-05,
+      "loss": 0.1207,
+      "step": 30220
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.9916062060497824e-05,
+      "loss": 0.1461,
+      "step": 30230
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.9912713472485768e-05,
+      "loss": 0.1552,
+      "step": 30240
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.9909364884473715e-05,
+      "loss": 0.1789,
+      "step": 30250
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.990601629646166e-05,
+      "loss": 0.1488,
+      "step": 30260
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.9902667708449604e-05,
+      "loss": 0.1015,
+      "step": 30270
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.9899319120437548e-05,
+      "loss": 0.1142,
+      "step": 30280
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.9895970532425495e-05,
+      "loss": 0.1626,
+      "step": 30290
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.989262194441344e-05,
+      "loss": 0.1668,
+      "step": 30300
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.9889273356401387e-05,
+      "loss": 0.1334,
+      "step": 30310
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.988592476838933e-05,
+      "loss": 0.1591,
+      "step": 30320
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.9882576180377275e-05,
+      "loss": 0.1407,
+      "step": 30330
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.987922759236522e-05,
+      "loss": 0.0893,
+      "step": 30340
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.9875879004353163e-05,
+      "loss": 0.0845,
+      "step": 30350
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.987253041634111e-05,
+      "loss": 0.2419,
+      "step": 30360
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.9869181828329055e-05,
+      "loss": 0.1066,
+      "step": 30370
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.9865833240317002e-05,
+      "loss": 0.1924,
+      "step": 30380
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.9862484652304946e-05,
+      "loss": 0.0691,
+      "step": 30390
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.985913606429289e-05,
+      "loss": 0.1231,
+      "step": 30400
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.9855787476280834e-05,
+      "loss": 0.2197,
+      "step": 30410
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.985243888826878e-05,
+      "loss": 0.169,
+      "step": 30420
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.9849090300256726e-05,
+      "loss": 0.1459,
+      "step": 30430
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.984574171224467e-05,
+      "loss": 0.1109,
+      "step": 30440
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.9842393124232617e-05,
+      "loss": 0.0708,
+      "step": 30450
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.983904453622056e-05,
+      "loss": 0.1945,
+      "step": 30460
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.9835695948208505e-05,
+      "loss": 0.2726,
+      "step": 30470
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.983234736019645e-05,
+      "loss": 0.0817,
+      "step": 30480
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.9828998772184397e-05,
+      "loss": 0.1627,
+      "step": 30490
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.982565018417234e-05,
+      "loss": 0.0625,
+      "step": 30500
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.9822301596160285e-05,
+      "loss": 0.197,
+      "step": 30510
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.9818953008148233e-05,
+      "loss": 0.1362,
+      "step": 30520
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.9815604420136177e-05,
+      "loss": 0.1663,
+      "step": 30530
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.981225583212412e-05,
+      "loss": 0.1451,
+      "step": 30540
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.9808907244112068e-05,
+      "loss": 0.3195,
+      "step": 30550
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.9805558656100012e-05,
+      "loss": 0.1832,
+      "step": 30560
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.9802210068087956e-05,
+      "loss": 0.1794,
+      "step": 30570
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.97988614800759e-05,
+      "loss": 0.1158,
+      "step": 30580
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.9795512892063848e-05,
+      "loss": 0.1322,
+      "step": 30590
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.9792164304051792e-05,
+      "loss": 0.1699,
+      "step": 30600
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.9788815716039736e-05,
+      "loss": 0.1778,
+      "step": 30610
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.9785467128027683e-05,
+      "loss": 0.1656,
+      "step": 30620
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.9782118540015628e-05,
+      "loss": 0.1919,
+      "step": 30630
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.977876995200357e-05,
+      "loss": 0.1886,
+      "step": 30640
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.9775421363991516e-05,
+      "loss": 0.1059,
+      "step": 30650
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.9772072775979463e-05,
+      "loss": 0.273,
+      "step": 30660
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.9768724187967407e-05,
+      "loss": 0.2237,
+      "step": 30670
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.9765375599955355e-05,
+      "loss": 0.1034,
+      "step": 30680
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.97620270119433e-05,
+      "loss": 0.1424,
+      "step": 30690
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.9758678423931243e-05,
+      "loss": 0.1079,
+      "step": 30700
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.9755329835919187e-05,
+      "loss": 0.0977,
+      "step": 30710
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.975198124790713e-05,
+      "loss": 0.1943,
+      "step": 30720
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.974863265989508e-05,
+      "loss": 0.0832,
+      "step": 30730
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.9745284071883023e-05,
+      "loss": 0.1323,
+      "step": 30740
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.974193548387097e-05,
+      "loss": 0.0818,
+      "step": 30750
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.9738586895858914e-05,
+      "loss": 0.2157,
+      "step": 30760
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.9735238307846858e-05,
+      "loss": 0.1639,
+      "step": 30770
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.9731889719834802e-05,
+      "loss": 0.1025,
+      "step": 30780
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.9728541131822746e-05,
+      "loss": 0.2064,
+      "step": 30790
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.9725192543810694e-05,
+      "loss": 0.1291,
+      "step": 30800
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.972184395579864e-05,
+      "loss": 0.1282,
+      "step": 30810
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.9718495367786585e-05,
+      "loss": 0.3137,
+      "step": 30820
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.971514677977453e-05,
+      "loss": 0.0928,
+      "step": 30830
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.9711798191762473e-05,
+      "loss": 0.1626,
+      "step": 30840
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.9708449603750418e-05,
+      "loss": 0.2284,
+      "step": 30850
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.970510101573836e-05,
+      "loss": 0.0668,
+      "step": 30860
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.970175242772631e-05,
+      "loss": 0.4469,
+      "step": 30870
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.9698403839714257e-05,
+      "loss": 0.1379,
+      "step": 30880
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.96950552517022e-05,
+      "loss": 0.1856,
+      "step": 30890
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.9691706663690145e-05,
+      "loss": 0.2534,
+      "step": 30900
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.968835807567809e-05,
+      "loss": 0.2058,
+      "step": 30910
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.9685009487666033e-05,
+      "loss": 0.1254,
+      "step": 30920
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.9681660899653977e-05,
+      "loss": 0.1136,
+      "step": 30930
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.9678312311641928e-05,
+      "loss": 0.1084,
+      "step": 30940
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.9674963723629872e-05,
+      "loss": 0.0924,
+      "step": 30950
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.9671615135617816e-05,
+      "loss": 0.1653,
+      "step": 30960
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.966826654760576e-05,
+      "loss": 0.2012,
+      "step": 30970
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.9664917959593704e-05,
+      "loss": 0.1512,
+      "step": 30980
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.9661569371581648e-05,
+      "loss": 0.1231,
+      "step": 30990
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.9658220783569592e-05,
+      "loss": 0.0851,
+      "step": 31000
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.9654872195557543e-05,
+      "loss": 0.1842,
+      "step": 31010
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.9651523607545487e-05,
+      "loss": 0.1197,
+      "step": 31020
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.964817501953343e-05,
+      "loss": 0.1313,
+      "step": 31030
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.9644826431521375e-05,
+      "loss": 0.1749,
+      "step": 31040
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.964147784350932e-05,
+      "loss": 0.1807,
+      "step": 31050
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.9638129255497263e-05,
+      "loss": 0.132,
+      "step": 31060
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.963478066748521e-05,
+      "loss": 0.1191,
+      "step": 31070
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.963143207947316e-05,
+      "loss": 0.1958,
+      "step": 31080
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.9628083491461102e-05,
+      "loss": 0.223,
+      "step": 31090
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.9624734903449046e-05,
+      "loss": 0.1424,
+      "step": 31100
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.962138631543699e-05,
+      "loss": 0.2055,
+      "step": 31110
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.9618037727424935e-05,
+      "loss": 0.1029,
+      "step": 31120
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.961468913941288e-05,
+      "loss": 0.2032,
+      "step": 31130
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.9611340551400826e-05,
+      "loss": 0.1261,
+      "step": 31140
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.9607991963388774e-05,
+      "loss": 0.147,
+      "step": 31150
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.9604643375376718e-05,
+      "loss": 0.1452,
+      "step": 31160
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.9601294787364662e-05,
+      "loss": 0.1482,
+      "step": 31170
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.9597946199352606e-05,
+      "loss": 0.1258,
+      "step": 31180
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.959459761134055e-05,
+      "loss": 0.2102,
+      "step": 31190
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.9591249023328497e-05,
+      "loss": 0.1319,
+      "step": 31200
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.9587900435316445e-05,
+      "loss": 0.3075,
+      "step": 31210
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.958455184730439e-05,
+      "loss": 0.1764,
+      "step": 31220
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.9581203259292333e-05,
+      "loss": 0.1895,
+      "step": 31230
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.9577854671280277e-05,
+      "loss": 0.1786,
+      "step": 31240
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.957450608326822e-05,
+      "loss": 0.1172,
+      "step": 31250
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.9571157495256165e-05,
+      "loss": 0.2787,
+      "step": 31260
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.9567808907244113e-05,
+      "loss": 0.2582,
+      "step": 31270
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.956446031923206e-05,
+      "loss": 0.1208,
+      "step": 31280
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.9561111731220004e-05,
+      "loss": 0.1242,
+      "step": 31290
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.9557763143207948e-05,
+      "loss": 0.1707,
+      "step": 31300
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.9554414555195892e-05,
+      "loss": 0.2198,
+      "step": 31310
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.9551065967183836e-05,
+      "loss": 0.2939,
+      "step": 31320
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.9547717379171784e-05,
+      "loss": 0.166,
+      "step": 31330
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.9544368791159728e-05,
+      "loss": 0.0664,
+      "step": 31340
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.9541020203147675e-05,
+      "loss": 0.111,
+      "step": 31350
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.953767161513562e-05,
+      "loss": 0.1654,
+      "step": 31360
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.9534323027123564e-05,
+      "loss": 0.2264,
+      "step": 31370
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.9530974439111508e-05,
+      "loss": 0.2857,
+      "step": 31380
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.9527625851099452e-05,
+      "loss": 0.163,
+      "step": 31390
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.95242772630874e-05,
+      "loss": 0.1378,
+      "step": 31400
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.9520928675075343e-05,
+      "loss": 0.1496,
+      "step": 31410
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.951758008706329e-05,
+      "loss": 0.1814,
+      "step": 31420
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.9514231499051235e-05,
+      "loss": 0.1563,
+      "step": 31430
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.951088291103918e-05,
+      "loss": 0.1124,
+      "step": 31440
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.9507534323027123e-05,
+      "loss": 0.1074,
+      "step": 31450
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.950418573501507e-05,
+      "loss": 0.1685,
+      "step": 31460
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.9500837147003014e-05,
+      "loss": 0.48,
+      "step": 31470
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.949748855899096e-05,
+      "loss": 0.1835,
+      "step": 31480
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.9494139970978906e-05,
+      "loss": 0.1749,
+      "step": 31490
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.949079138296685e-05,
+      "loss": 0.1435,
+      "step": 31500
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.9487442794954794e-05,
+      "loss": 0.1537,
+      "step": 31510
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.9484094206942738e-05,
+      "loss": 0.1533,
+      "step": 31520
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.9480745618930686e-05,
+      "loss": 0.2387,
+      "step": 31530
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.947739703091863e-05,
+      "loss": 0.1654,
+      "step": 31540
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.9474048442906574e-05,
+      "loss": 0.2026,
+      "step": 31550
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.947069985489452e-05,
+      "loss": 0.1401,
+      "step": 31560
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.9467351266882465e-05,
+      "loss": 0.2072,
+      "step": 31570
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.946400267887041e-05,
+      "loss": 0.1731,
+      "step": 31580
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.9460654090858357e-05,
+      "loss": 0.135,
+      "step": 31590
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.94573055028463e-05,
+      "loss": 0.1241,
+      "step": 31600
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.9453956914834245e-05,
+      "loss": 0.1634,
+      "step": 31610
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.945060832682219e-05,
+      "loss": 0.1369,
+      "step": 31620
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.9447259738810137e-05,
+      "loss": 0.0505,
+      "step": 31630
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.944391115079808e-05,
+      "loss": 0.1847,
+      "step": 31640
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.9440562562786025e-05,
+      "loss": 0.1696,
+      "step": 31650
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.9437213974773972e-05,
+      "loss": 0.2476,
+      "step": 31660
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.9433865386761916e-05,
+      "loss": 0.1201,
+      "step": 31670
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.943051679874986e-05,
+      "loss": 0.1531,
+      "step": 31680
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.9427168210737804e-05,
+      "loss": 0.1346,
+      "step": 31690
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.9423819622725752e-05,
+      "loss": 0.2291,
+      "step": 31700
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.9420471034713696e-05,
+      "loss": 0.1608,
+      "step": 31710
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.941712244670164e-05,
+      "loss": 0.0874,
+      "step": 31720
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.9413773858689588e-05,
+      "loss": 0.1783,
+      "step": 31730
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.941042527067753e-05,
+      "loss": 0.2313,
+      "step": 31740
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.9407076682665476e-05,
+      "loss": 0.1577,
+      "step": 31750
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.940372809465342e-05,
+      "loss": 0.1612,
+      "step": 31760
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.9400379506641367e-05,
+      "loss": 0.1299,
+      "step": 31770
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.939703091862931e-05,
+      "loss": 0.178,
+      "step": 31780
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.939368233061726e-05,
+      "loss": 0.2642,
+      "step": 31790
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.9390333742605203e-05,
+      "loss": 0.0961,
+      "step": 31800
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.9386985154593147e-05,
+      "loss": 0.2221,
+      "step": 31810
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.938363656658109e-05,
+      "loss": 0.1689,
+      "step": 31820
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.9380287978569035e-05,
+      "loss": 0.1354,
+      "step": 31830
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.9376939390556982e-05,
+      "loss": 0.1152,
+      "step": 31840
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.9373590802544927e-05,
+      "loss": 0.1347,
+      "step": 31850
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.9370242214532874e-05,
+      "loss": 0.2427,
+      "step": 31860
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.9366893626520818e-05,
+      "loss": 0.1128,
+      "step": 31870
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.9363545038508762e-05,
+      "loss": 0.2024,
+      "step": 31880
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.9360196450496706e-05,
+      "loss": 0.1088,
+      "step": 31890
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.935684786248465e-05,
+      "loss": 0.1541,
+      "step": 31900
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.9353499274472598e-05,
+      "loss": 0.2418,
+      "step": 31910
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.9350150686460545e-05,
+      "loss": 0.242,
+      "step": 31920
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.934680209844849e-05,
+      "loss": 0.0839,
+      "step": 31930
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.9343453510436433e-05,
+      "loss": 0.1356,
+      "step": 31940
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.9340104922424377e-05,
+      "loss": 0.3457,
+      "step": 31950
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.933675633441232e-05,
+      "loss": 0.1951,
+      "step": 31960
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.9333407746400266e-05,
+      "loss": 0.1427,
+      "step": 31970
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.9330059158388213e-05,
+      "loss": 0.1953,
+      "step": 31980
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.932671057037616e-05,
+      "loss": 0.1188,
+      "step": 31990
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.9323361982364105e-05,
+      "loss": 0.1333,
+      "step": 32000
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.932001339435205e-05,
+      "loss": 0.1732,
+      "step": 32010
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.9316664806339993e-05,
+      "loss": 0.2076,
+      "step": 32020
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.9313316218327937e-05,
+      "loss": 0.252,
+      "step": 32030
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.930996763031588e-05,
+      "loss": 0.1241,
+      "step": 32040
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.9306619042303832e-05,
+      "loss": 0.2897,
+      "step": 32050
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.9303270454291776e-05,
+      "loss": 0.1268,
+      "step": 32060
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.929992186627972e-05,
+      "loss": 0.3232,
+      "step": 32070
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.9296573278267664e-05,
+      "loss": 0.1683,
+      "step": 32080
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.9293224690255608e-05,
+      "loss": 0.0938,
+      "step": 32090
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.9289876102243552e-05,
+      "loss": 0.3025,
+      "step": 32100
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.92865275142315e-05,
+      "loss": 0.1142,
+      "step": 32110
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.9283178926219447e-05,
+      "loss": 0.1683,
+      "step": 32120
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.927983033820739e-05,
+      "loss": 0.0853,
+      "step": 32130
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.9276481750195335e-05,
+      "loss": 0.108,
+      "step": 32140
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.927313316218328e-05,
+      "loss": 0.1569,
+      "step": 32150
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.9269784574171223e-05,
+      "loss": 0.3315,
+      "step": 32160
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.9266435986159167e-05,
+      "loss": 0.1694,
+      "step": 32170
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.9263087398147118e-05,
+      "loss": 0.1828,
+      "step": 32180
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.9259738810135062e-05,
+      "loss": 0.129,
+      "step": 32190
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.9256390222123006e-05,
+      "loss": 0.2167,
+      "step": 32200
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.925304163411095e-05,
+      "loss": 0.0554,
+      "step": 32210
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.9249693046098895e-05,
+      "loss": 0.1148,
+      "step": 32220
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.924634445808684e-05,
+      "loss": 0.0908,
+      "step": 32230
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.9242995870074783e-05,
+      "loss": 0.1455,
+      "step": 32240
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.9239647282062734e-05,
+      "loss": 0.1615,
+      "step": 32250
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.9236298694050678e-05,
+      "loss": 0.1423,
+      "step": 32260
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.9232950106038622e-05,
+      "loss": 0.1188,
+      "step": 32270
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.9229601518026566e-05,
+      "loss": 0.0679,
+      "step": 32280
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.922625293001451e-05,
+      "loss": 0.1983,
+      "step": 32290
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.9222904342002454e-05,
+      "loss": 0.1008,
+      "step": 32300
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.92195557539904e-05,
+      "loss": 0.2266,
+      "step": 32310
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.921620716597835e-05,
+      "loss": 0.1015,
+      "step": 32320
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.9212858577966293e-05,
+      "loss": 0.1162,
+      "step": 32330
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.9209509989954237e-05,
+      "loss": 0.0788,
+      "step": 32340
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.920616140194218e-05,
+      "loss": 0.1852,
+      "step": 32350
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.9202812813930125e-05,
+      "loss": 0.1605,
+      "step": 32360
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.919946422591807e-05,
+      "loss": 0.1878,
+      "step": 32370
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.9196115637906017e-05,
+      "loss": 0.1068,
+      "step": 32380
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.9192767049893964e-05,
+      "loss": 0.2178,
+      "step": 32390
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.9189418461881908e-05,
+      "loss": 0.1593,
+      "step": 32400
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.9186069873869852e-05,
+      "loss": 0.2669,
+      "step": 32410
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.9182721285857796e-05,
+      "loss": 0.4707,
+      "step": 32420
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.917937269784574e-05,
+      "loss": 0.1833,
+      "step": 32430
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.9176024109833688e-05,
+      "loss": 0.2116,
+      "step": 32440
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.9172675521821632e-05,
+      "loss": 0.1801,
+      "step": 32450
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.916932693380958e-05,
+      "loss": 0.1291,
+      "step": 32460
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.9165978345797524e-05,
+      "loss": 0.1412,
+      "step": 32470
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.9162629757785468e-05,
+      "loss": 0.12,
+      "step": 32480
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.915928116977341e-05,
+      "loss": 0.1205,
+      "step": 32490
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.9155932581761356e-05,
+      "loss": 0.2174,
+      "step": 32500
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.9152583993749303e-05,
+      "loss": 0.1544,
+      "step": 32510
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.9149235405737247e-05,
+      "loss": 0.1169,
+      "step": 32520
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.9145886817725195e-05,
+      "loss": 0.148,
+      "step": 32530
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.914253822971314e-05,
+      "loss": 0.2223,
+      "step": 32540
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.9139189641701083e-05,
+      "loss": 0.0815,
+      "step": 32550
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.9135841053689027e-05,
+      "loss": 0.3072,
+      "step": 32560
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.9132492465676974e-05,
+      "loss": 0.1576,
+      "step": 32570
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.912914387766492e-05,
+      "loss": 0.1859,
+      "step": 32580
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.9125795289652863e-05,
+      "loss": 0.1033,
+      "step": 32590
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.912244670164081e-05,
+      "loss": 0.2071,
+      "step": 32600
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.9119098113628754e-05,
+      "loss": 0.2497,
+      "step": 32610
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.9115749525616698e-05,
+      "loss": 0.0844,
+      "step": 32620
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.9112400937604642e-05,
+      "loss": 0.1136,
+      "step": 32630
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.910905234959259e-05,
+      "loss": 0.1715,
+      "step": 32640
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.9105703761580534e-05,
+      "loss": 0.1258,
+      "step": 32650
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.9102355173568478e-05,
+      "loss": 0.1685,
+      "step": 32660
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.9099006585556425e-05,
+      "loss": 0.0963,
+      "step": 32670
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.909565799754437e-05,
+      "loss": 0.1409,
+      "step": 32680
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.9092309409532313e-05,
+      "loss": 0.0949,
+      "step": 32690
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.908896082152026e-05,
+      "loss": 0.1437,
+      "step": 32700
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.9085612233508205e-05,
+      "loss": 0.2812,
+      "step": 32710
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.908226364549615e-05,
+      "loss": 0.2862,
+      "step": 32720
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.9078915057484093e-05,
+      "loss": 0.1324,
+      "step": 32730
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.907556646947204e-05,
+      "loss": 0.1029,
+      "step": 32740
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.9072217881459985e-05,
+      "loss": 0.1403,
+      "step": 32750
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.906886929344793e-05,
+      "loss": 0.186,
+      "step": 32760
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.9065520705435876e-05,
+      "loss": 0.1608,
+      "step": 32770
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.906217211742382e-05,
+      "loss": 0.1888,
+      "step": 32780
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.9058823529411764e-05,
+      "loss": 0.1697,
+      "step": 32790
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.905547494139971e-05,
+      "loss": 0.0825,
+      "step": 32800
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.9052126353387656e-05,
+      "loss": 0.1468,
+      "step": 32810
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.90487777653756e-05,
+      "loss": 0.2262,
+      "step": 32820
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.9045429177363547e-05,
+      "loss": 0.1798,
+      "step": 32830
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.904208058935149e-05,
+      "loss": 0.0986,
+      "step": 32840
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.9038732001339436e-05,
+      "loss": 0.2637,
+      "step": 32850
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.903538341332738e-05,
+      "loss": 0.2434,
+      "step": 32860
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.9032034825315324e-05,
+      "loss": 0.0652,
+      "step": 32870
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.902868623730327e-05,
+      "loss": 0.1413,
+      "step": 32880
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.9025337649291215e-05,
+      "loss": 0.1813,
+      "step": 32890
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.9021989061279163e-05,
+      "loss": 0.1091,
+      "step": 32900
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.9018640473267107e-05,
+      "loss": 0.1589,
+      "step": 32910
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.901529188525505e-05,
+      "loss": 0.2023,
+      "step": 32920
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.9011943297242995e-05,
+      "loss": 0.0841,
+      "step": 32930
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.900859470923094e-05,
+      "loss": 0.2121,
+      "step": 32940
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.9005246121218887e-05,
+      "loss": 0.0814,
+      "step": 32950
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.9001897533206834e-05,
+      "loss": 0.2085,
+      "step": 32960
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.8998548945194778e-05,
+      "loss": 0.1662,
+      "step": 32970
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.8995200357182722e-05,
+      "loss": 0.0972,
+      "step": 32980
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.8991851769170666e-05,
+      "loss": 0.1311,
+      "step": 32990
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.898850318115861e-05,
+      "loss": 0.0987,
+      "step": 33000
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.8985154593146558e-05,
+      "loss": 0.3535,
+      "step": 33010
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.8981806005134502e-05,
+      "loss": 0.1378,
+      "step": 33020
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.897845741712245e-05,
+      "loss": 0.1304,
+      "step": 33030
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.8975108829110393e-05,
+      "loss": 0.1555,
+      "step": 33040
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.897209509989954e-05,
+      "loss": 0.3663,
+      "step": 33050
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.8968746511887488e-05,
+      "loss": 0.1891,
+      "step": 33060
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.8965397923875432e-05,
+      "loss": 0.1322,
+      "step": 33070
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.896204933586338e-05,
+      "loss": 0.1245,
+      "step": 33080
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.8958700747851324e-05,
+      "loss": 0.1778,
+      "step": 33090
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.8955352159839268e-05,
+      "loss": 0.2142,
+      "step": 33100
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.8952003571827212e-05,
+      "loss": 0.1588,
+      "step": 33110
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.8948654983815156e-05,
+      "loss": 0.2808,
+      "step": 33120
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.8945306395803103e-05,
+      "loss": 0.1944,
+      "step": 33130
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.894195780779105e-05,
+      "loss": 0.1486,
+      "step": 33140
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.8938609219778995e-05,
+      "loss": 0.0772,
+      "step": 33150
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.893526063176694e-05,
+      "loss": 0.2276,
+      "step": 33160
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.8931912043754883e-05,
+      "loss": 0.1133,
+      "step": 33170
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.8928563455742827e-05,
+      "loss": 0.1364,
+      "step": 33180
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.892521486773077e-05,
+      "loss": 0.0542,
+      "step": 33190
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.892186627971872e-05,
+      "loss": 0.092,
+      "step": 33200
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.8918517691706666e-05,
+      "loss": 0.243,
+      "step": 33210
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.891516910369461e-05,
+      "loss": 0.1109,
+      "step": 33220
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.8911820515682554e-05,
+      "loss": 0.0645,
+      "step": 33230
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.89084719276705e-05,
+      "loss": 0.108,
+      "step": 33240
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.8905123339658442e-05,
+      "loss": 0.2617,
+      "step": 33250
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.890177475164639e-05,
+      "loss": 0.2428,
+      "step": 33260
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.8898426163634337e-05,
+      "loss": 0.2323,
+      "step": 33270
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.889507757562228e-05,
+      "loss": 0.2016,
+      "step": 33280
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.8891728987610226e-05,
+      "loss": 0.086,
+      "step": 33290
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.888838039959817e-05,
+      "loss": 0.0685,
+      "step": 33300
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.8885031811586114e-05,
+      "loss": 0.2164,
+      "step": 33310
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.8881683223574058e-05,
+      "loss": 0.1881,
+      "step": 33320
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.8878334635562005e-05,
+      "loss": 0.2313,
+      "step": 33330
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.8874986047549953e-05,
+      "loss": 0.0873,
+      "step": 33340
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.8871637459537897e-05,
+      "loss": 0.1189,
+      "step": 33350
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.886828887152584e-05,
+      "loss": 0.1894,
+      "step": 33360
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.8864940283513785e-05,
+      "loss": 0.1559,
+      "step": 33370
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.886159169550173e-05,
+      "loss": 0.1055,
+      "step": 33380
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.8858243107489673e-05,
+      "loss": 0.1132,
+      "step": 33390
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.8854894519477624e-05,
+      "loss": 0.0703,
+      "step": 33400
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.8851545931465568e-05,
+      "loss": 0.2246,
+      "step": 33410
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.8848197343453512e-05,
+      "loss": 0.275,
+      "step": 33420
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.8844848755441456e-05,
+      "loss": 0.1477,
+      "step": 33430
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.88415001674294e-05,
+      "loss": 0.1229,
+      "step": 33440
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.8838151579417344e-05,
+      "loss": 0.0963,
+      "step": 33450
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.883480299140529e-05,
+      "loss": 0.1866,
+      "step": 33460
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.883145440339324e-05,
+      "loss": 0.1369,
+      "step": 33470
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.8828105815381183e-05,
+      "loss": 0.1021,
+      "step": 33480
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.8824757227369127e-05,
+      "loss": 0.167,
+      "step": 33490
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.882140863935707e-05,
+      "loss": 0.1612,
+      "step": 33500
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.8818060051345015e-05,
+      "loss": 0.2444,
+      "step": 33510
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.881471146333296e-05,
+      "loss": 0.1242,
+      "step": 33520
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.8811362875320907e-05,
+      "loss": 0.1626,
+      "step": 33530
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.8808014287308854e-05,
+      "loss": 0.1269,
+      "step": 33540
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.88046656992968e-05,
+      "loss": 0.1427,
+      "step": 33550
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.8801317111284743e-05,
+      "loss": 0.3592,
+      "step": 33560
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.8797968523272687e-05,
+      "loss": 0.1333,
+      "step": 33570
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.879461993526063e-05,
+      "loss": 0.1104,
+      "step": 33580
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.8791271347248575e-05,
+      "loss": 0.0794,
+      "step": 33590
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.8787922759236522e-05,
+      "loss": 0.2518,
+      "step": 33600
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.878457417122447e-05,
+      "loss": 0.2391,
+      "step": 33610
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.8781225583212414e-05,
+      "loss": 0.2225,
+      "step": 33620
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.8777876995200358e-05,
+      "loss": 0.1939,
+      "step": 33630
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.8774528407188302e-05,
+      "loss": 0.1171,
+      "step": 33640
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.8771179819176246e-05,
+      "loss": 0.1126,
+      "step": 33650
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.8767831231164194e-05,
+      "loss": 0.147,
+      "step": 33660
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.8764482643152138e-05,
+      "loss": 0.0966,
+      "step": 33670
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.8761134055140085e-05,
+      "loss": 0.1468,
+      "step": 33680
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.875778546712803e-05,
+      "loss": 0.1596,
+      "step": 33690
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.8754436879115973e-05,
+      "loss": 0.0994,
+      "step": 33700
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.8751088291103917e-05,
+      "loss": 0.1717,
+      "step": 33710
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.874773970309186e-05,
+      "loss": 0.0981,
+      "step": 33720
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.874439111507981e-05,
+      "loss": 0.1024,
+      "step": 33730
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.8741042527067753e-05,
+      "loss": 0.0866,
+      "step": 33740
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.87376939390557e-05,
+      "loss": 0.2015,
+      "step": 33750
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.8734345351043644e-05,
+      "loss": 0.2841,
+      "step": 33760
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.873099676303159e-05,
+      "loss": 0.1472,
+      "step": 33770
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.8727648175019533e-05,
+      "loss": 0.1435,
+      "step": 33780
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.872429958700748e-05,
+      "loss": 0.1117,
+      "step": 33790
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.8720950998995424e-05,
+      "loss": 0.2315,
+      "step": 33800
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.8717602410983368e-05,
+      "loss": 0.2381,
+      "step": 33810
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.8714253822971316e-05,
+      "loss": 0.1385,
+      "step": 33820
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.871090523495926e-05,
+      "loss": 0.2039,
+      "step": 33830
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.8707556646947204e-05,
+      "loss": 0.1602,
+      "step": 33840
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.8704208058935148e-05,
+      "loss": 0.1803,
+      "step": 33850
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.8700859470923095e-05,
+      "loss": 0.1511,
+      "step": 33860
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.869751088291104e-05,
+      "loss": 0.1756,
+      "step": 33870
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.8694162294898983e-05,
+      "loss": 0.2044,
+      "step": 33880
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.869081370688693e-05,
+      "loss": 0.1637,
+      "step": 33890
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.8687465118874875e-05,
+      "loss": 0.0858,
+      "step": 33900
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.868411653086282e-05,
+      "loss": 0.1812,
+      "step": 33910
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.8680767942850767e-05,
+      "loss": 0.155,
+      "step": 33920
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.867741935483871e-05,
+      "loss": 0.1265,
+      "step": 33930
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.8674070766826655e-05,
+      "loss": 0.2191,
+      "step": 33940
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.86707221788146e-05,
+      "loss": 0.0639,
+      "step": 33950
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.8667373590802546e-05,
+      "loss": 0.198,
+      "step": 33960
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.866402500279049e-05,
+      "loss": 0.2784,
+      "step": 33970
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.8660676414778434e-05,
+      "loss": 0.0772,
+      "step": 33980
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.8657327826766382e-05,
+      "loss": 0.4441,
+      "step": 33990
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.8653979238754326e-05,
+      "loss": 0.1914,
+      "step": 34000
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.865063065074227e-05,
+      "loss": 0.2069,
+      "step": 34010
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.8647282062730214e-05,
+      "loss": 0.2555,
+      "step": 34020
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.864393347471816e-05,
+      "loss": 0.2692,
+      "step": 34030
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.8640584886706106e-05,
+      "loss": 0.093,
+      "step": 34040
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.8637236298694053e-05,
+      "loss": 0.1561,
+      "step": 34050
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.8633887710681997e-05,
+      "loss": 0.1944,
+      "step": 34060
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.863053912266994e-05,
+      "loss": 0.1386,
+      "step": 34070
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.8627190534657885e-05,
+      "loss": 0.1258,
+      "step": 34080
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.862384194664583e-05,
+      "loss": 0.1303,
+      "step": 34090
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.8620493358633777e-05,
+      "loss": 0.1085,
+      "step": 34100
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.861714477062172e-05,
+      "loss": 0.1963,
+      "step": 34110
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.861379618260967e-05,
+      "loss": 0.0779,
+      "step": 34120
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.8610447594597612e-05,
+      "loss": 0.1105,
+      "step": 34130
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.8607099006585557e-05,
+      "loss": 0.1266,
+      "step": 34140
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.86037504185735e-05,
+      "loss": 0.1571,
+      "step": 34150
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.8600401830561448e-05,
+      "loss": 0.2917,
+      "step": 34160
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.8597053242549392e-05,
+      "loss": 0.2429,
+      "step": 34170
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.859370465453734e-05,
+      "loss": 0.1954,
+      "step": 34180
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.8590356066525284e-05,
+      "loss": 0.1379,
+      "step": 34190
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.8587007478513228e-05,
+      "loss": 0.1086,
+      "step": 34200
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.8583658890501172e-05,
+      "loss": 0.1629,
+      "step": 34210
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.8580310302489116e-05,
+      "loss": 0.1292,
+      "step": 34220
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.8576961714477063e-05,
+      "loss": 0.1188,
+      "step": 34230
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.8573613126465007e-05,
+      "loss": 0.1656,
+      "step": 34240
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.8570264538452955e-05,
+      "loss": 0.081,
+      "step": 34250
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.85669159504409e-05,
+      "loss": 0.0965,
+      "step": 34260
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.8563567362428843e-05,
+      "loss": 0.1475,
+      "step": 34270
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.8560218774416787e-05,
+      "loss": 0.1656,
+      "step": 34280
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.855687018640473e-05,
+      "loss": 0.0706,
+      "step": 34290
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.855352159839268e-05,
+      "loss": 0.2607,
+      "step": 34300
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.8550173010380623e-05,
+      "loss": 0.1398,
+      "step": 34310
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.854682442236857e-05,
+      "loss": 0.3616,
+      "step": 34320
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.8543475834356514e-05,
+      "loss": 0.2791,
+      "step": 34330
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.854012724634446e-05,
+      "loss": 0.1473,
+      "step": 34340
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.8536778658332402e-05,
+      "loss": 0.2024,
+      "step": 34350
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.8533430070320346e-05,
+      "loss": 0.2011,
+      "step": 34360
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.8530081482308294e-05,
+      "loss": 0.103,
+      "step": 34370
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.852673289429624e-05,
+      "loss": 0.1186,
+      "step": 34380
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.8523384306284185e-05,
+      "loss": 0.1986,
+      "step": 34390
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.852003571827213e-05,
+      "loss": 0.1552,
+      "step": 34400
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.8516687130260074e-05,
+      "loss": 0.1482,
+      "step": 34410
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.8513338542248018e-05,
+      "loss": 0.2201,
+      "step": 34420
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.8509989954235962e-05,
+      "loss": 0.1377,
+      "step": 34430
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.850664136622391e-05,
+      "loss": 0.1013,
+      "step": 34440
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.8503292778211857e-05,
+      "loss": 0.1272,
+      "step": 34450
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.84999441901998e-05,
+      "loss": 0.1184,
+      "step": 34460
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.8496595602187745e-05,
+      "loss": 0.5988,
+      "step": 34470
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.849324701417569e-05,
+      "loss": 0.1284,
+      "step": 34480
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.8489898426163633e-05,
+      "loss": 0.2721,
+      "step": 34490
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.8486549838151577e-05,
+      "loss": 0.1042,
+      "step": 34500
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.848353610894073e-05,
+      "loss": 0.2477,
+      "step": 34510
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.8480187520928675e-05,
+      "loss": 0.1098,
+      "step": 34520
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.847683893291662e-05,
+      "loss": 0.168,
+      "step": 34530
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.8473490344904563e-05,
+      "loss": 0.1764,
+      "step": 34540
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.847014175689251e-05,
+      "loss": 0.0552,
+      "step": 34550
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.8466793168880458e-05,
+      "loss": 0.2287,
+      "step": 34560
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.8463444580868402e-05,
+      "loss": 0.0896,
+      "step": 34570
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.8460095992856346e-05,
+      "loss": 0.0945,
+      "step": 34580
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.845674740484429e-05,
+      "loss": 0.1449,
+      "step": 34590
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.8453398816832235e-05,
+      "loss": 0.1305,
+      "step": 34600
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.845005022882018e-05,
+      "loss": 0.194,
+      "step": 34610
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.844670164080813e-05,
+      "loss": 0.199,
+      "step": 34620
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.8443353052796074e-05,
+      "loss": 0.1581,
+      "step": 34630
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.8440004464784018e-05,
+      "loss": 0.1524,
+      "step": 34640
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.8436655876771962e-05,
+      "loss": 0.249,
+      "step": 34650
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.8433307288759906e-05,
+      "loss": 0.0918,
+      "step": 34660
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.842995870074785e-05,
+      "loss": 0.0843,
+      "step": 34670
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.8426610112735794e-05,
+      "loss": 0.1472,
+      "step": 34680
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.8423261524723745e-05,
+      "loss": 0.2498,
+      "step": 34690
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.841991293671169e-05,
+      "loss": 0.1835,
+      "step": 34700
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.8416564348699633e-05,
+      "loss": 0.1009,
+      "step": 34710
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.8413215760687577e-05,
+      "loss": 0.1036,
+      "step": 34720
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.840986717267552e-05,
+      "loss": 0.1114,
+      "step": 34730
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.8406518584663465e-05,
+      "loss": 0.08,
+      "step": 34740
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.8403169996651413e-05,
+      "loss": 0.0993,
+      "step": 34750
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.839982140863936e-05,
+      "loss": 0.1519,
+      "step": 34760
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.8396472820627304e-05,
+      "loss": 0.1286,
+      "step": 34770
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.8393124232615248e-05,
+      "loss": 0.1204,
+      "step": 34780
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.8389775644603192e-05,
+      "loss": 0.1052,
+      "step": 34790
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.8386427056591136e-05,
+      "loss": 0.1616,
+      "step": 34800
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.838307846857908e-05,
+      "loss": 0.147,
+      "step": 34810
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.8379729880567028e-05,
+      "loss": 0.2437,
+      "step": 34820
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.8376381292554975e-05,
+      "loss": 0.1457,
+      "step": 34830
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.837303270454292e-05,
+      "loss": 0.0911,
+      "step": 34840
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.8369684116530864e-05,
+      "loss": 0.1058,
+      "step": 34850
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.8366335528518808e-05,
+      "loss": 0.1483,
+      "step": 34860
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.8362986940506752e-05,
+      "loss": 0.314,
+      "step": 34870
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.83596383524947e-05,
+      "loss": 0.157,
+      "step": 34880
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.8356289764482643e-05,
+      "loss": 0.1012,
+      "step": 34890
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.835294117647059e-05,
+      "loss": 0.0754,
+      "step": 34900
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.8349592588458535e-05,
+      "loss": 0.2004,
+      "step": 34910
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.834624400044648e-05,
+      "loss": 0.1655,
+      "step": 34920
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.8342895412434423e-05,
+      "loss": 0.1793,
+      "step": 34930
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.8339546824422367e-05,
+      "loss": 0.1518,
+      "step": 34940
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.8336198236410314e-05,
+      "loss": 0.099,
+      "step": 34950
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.833284964839826e-05,
+      "loss": 0.2023,
+      "step": 34960
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.8329501060386206e-05,
+      "loss": 0.154,
+      "step": 34970
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.832615247237415e-05,
+      "loss": 0.2089,
+      "step": 34980
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.8322803884362094e-05,
+      "loss": 0.1377,
+      "step": 34990
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.8319455296350038e-05,
+      "loss": 0.2136,
+      "step": 35000
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.8316106708337986e-05,
+      "loss": 0.1618,
+      "step": 35010
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.831275812032593e-05,
+      "loss": 0.1516,
+      "step": 35020
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.8309409532313874e-05,
+      "loss": 0.1746,
+      "step": 35030
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.830606094430182e-05,
+      "loss": 0.1344,
+      "step": 35040
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.8302712356289765e-05,
+      "loss": 0.1219,
+      "step": 35050
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.829936376827771e-05,
+      "loss": 0.1601,
+      "step": 35060
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.8296015180265654e-05,
+      "loss": 0.2317,
+      "step": 35070
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.82926665922536e-05,
+      "loss": 0.1088,
+      "step": 35080
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.8289318004241545e-05,
+      "loss": 0.1579,
+      "step": 35090
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.828596941622949e-05,
+      "loss": 0.2061,
+      "step": 35100
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.8282620828217437e-05,
+      "loss": 0.1051,
+      "step": 35110
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.827927224020538e-05,
+      "loss": 0.0733,
+      "step": 35120
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.8275923652193325e-05,
+      "loss": 0.1492,
+      "step": 35130
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.8272575064181272e-05,
+      "loss": 0.171,
+      "step": 35140
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.8269226476169216e-05,
+      "loss": 0.1333,
+      "step": 35150
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.826587788815716e-05,
+      "loss": 0.1586,
+      "step": 35160
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.8262529300145104e-05,
+      "loss": 0.1672,
+      "step": 35170
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.8259180712133052e-05,
+      "loss": 0.1802,
+      "step": 35180
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.8255832124120996e-05,
+      "loss": 0.1369,
+      "step": 35190
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.825248353610894e-05,
+      "loss": 0.1912,
+      "step": 35200
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.8249134948096887e-05,
+      "loss": 0.3164,
+      "step": 35210
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.824578636008483e-05,
+      "loss": 0.1503,
+      "step": 35220
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.8242437772072776e-05,
+      "loss": 0.0992,
+      "step": 35230
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.823908918406072e-05,
+      "loss": 0.1274,
+      "step": 35240
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.8235740596048667e-05,
+      "loss": 0.1521,
+      "step": 35250
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.823239200803661e-05,
+      "loss": 0.2091,
+      "step": 35260
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.822904342002456e-05,
+      "loss": 0.1053,
+      "step": 35270
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.8225694832012503e-05,
+      "loss": 0.1428,
+      "step": 35280
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.8222346244000447e-05,
+      "loss": 0.1648,
+      "step": 35290
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.821899765598839e-05,
+      "loss": 0.1396,
+      "step": 35300
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.8215649067976335e-05,
+      "loss": 0.2294,
+      "step": 35310
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.8212300479964282e-05,
+      "loss": 0.1202,
+      "step": 35320
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.8208951891952227e-05,
+      "loss": 0.198,
+      "step": 35330
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.8205603303940174e-05,
+      "loss": 0.1483,
+      "step": 35340
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.8202254715928118e-05,
+      "loss": 0.1138,
+      "step": 35350
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.8198906127916062e-05,
+      "loss": 0.1851,
+      "step": 35360
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.8195557539904006e-05,
+      "loss": 0.0776,
+      "step": 35370
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.8192208951891954e-05,
+      "loss": 0.1074,
+      "step": 35380
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.8188860363879898e-05,
+      "loss": 0.1735,
+      "step": 35390
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.8185511775867845e-05,
+      "loss": 0.0645,
+      "step": 35400
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.818216318785579e-05,
+      "loss": 0.2082,
+      "step": 35410
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.8178814599843733e-05,
+      "loss": 0.1071,
+      "step": 35420
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.8175466011831677e-05,
+      "loss": 0.325,
+      "step": 35430
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.817211742381962e-05,
+      "loss": 0.2271,
+      "step": 35440
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.816876883580757e-05,
+      "loss": 0.0666,
+      "step": 35450
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.8165420247795513e-05,
+      "loss": 0.2402,
+      "step": 35460
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.816207165978346e-05,
+      "loss": 0.1139,
+      "step": 35470
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.8158723071771405e-05,
+      "loss": 0.2081,
+      "step": 35480
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.815537448375935e-05,
+      "loss": 0.2177,
+      "step": 35490
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.8152025895747293e-05,
+      "loss": 0.1936,
+      "step": 35500
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.8148677307735237e-05,
+      "loss": 0.2166,
+      "step": 35510
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.8145328719723184e-05,
+      "loss": 0.0983,
+      "step": 35520
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.814198013171113e-05,
+      "loss": 0.1192,
+      "step": 35530
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.8138631543699076e-05,
+      "loss": 0.1421,
+      "step": 35540
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.813528295568702e-05,
+      "loss": 0.1231,
+      "step": 35550
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.8131934367674964e-05,
+      "loss": 0.1151,
+      "step": 35560
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.8128585779662908e-05,
+      "loss": 0.1155,
+      "step": 35570
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.8125237191650852e-05,
+      "loss": 0.1285,
+      "step": 35580
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.81218886036388e-05,
+      "loss": 0.1416,
+      "step": 35590
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.8118540015626747e-05,
+      "loss": 0.1636,
+      "step": 35600
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.811519142761469e-05,
+      "loss": 0.1441,
+      "step": 35610
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.8111842839602635e-05,
+      "loss": 0.2076,
+      "step": 35620
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.810849425159058e-05,
+      "loss": 0.1004,
+      "step": 35630
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.8105145663578523e-05,
+      "loss": 0.0902,
+      "step": 35640
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.8101797075566467e-05,
+      "loss": 0.0919,
+      "step": 35650
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.8098448487554415e-05,
+      "loss": 0.1533,
+      "step": 35660
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.8095099899542362e-05,
+      "loss": 0.145,
+      "step": 35670
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.8091751311530306e-05,
+      "loss": 0.138,
+      "step": 35680
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.808840272351825e-05,
+      "loss": 0.1097,
+      "step": 35690
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.8085054135506195e-05,
+      "loss": 0.4397,
+      "step": 35700
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.808170554749414e-05,
+      "loss": 0.1791,
+      "step": 35710
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.8078356959482083e-05,
+      "loss": 0.081,
+      "step": 35720
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.8075008371470034e-05,
+      "loss": 0.1213,
+      "step": 35730
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.8071659783457978e-05,
+      "loss": 0.0835,
+      "step": 35740
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.806831119544592e-05,
+      "loss": 0.1833,
+      "step": 35750
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.8064962607433866e-05,
+      "loss": 0.0788,
+      "step": 35760
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.806161401942181e-05,
+      "loss": 0.3599,
+      "step": 35770
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.8058265431409754e-05,
+      "loss": 0.1772,
+      "step": 35780
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.8054916843397698e-05,
+      "loss": 0.2415,
+      "step": 35790
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.805156825538565e-05,
+      "loss": 0.226,
+      "step": 35800
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.8048219667373593e-05,
+      "loss": 0.1353,
+      "step": 35810
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.8044871079361537e-05,
+      "loss": 0.0801,
+      "step": 35820
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.804152249134948e-05,
+      "loss": 0.1509,
+      "step": 35830
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.8038173903337425e-05,
+      "loss": 0.0813,
+      "step": 35840
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.803482531532537e-05,
+      "loss": 0.173,
+      "step": 35850
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.8031476727313317e-05,
+      "loss": 0.1909,
+      "step": 35860
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.8028128139301264e-05,
+      "loss": 0.1774,
+      "step": 35870
+    },
+    {
+      "epoch": 4.0,
+      "eval_loss": 0.28847813606262207,
+      "eval_runtime": 251.213,
+      "eval_samples_per_second": 7.937,
+      "eval_steps_per_second": 3.969,
+      "eval_wer": 0.11948233435610779,
+      "step": 35876
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.8024779551289208e-05,
+      "loss": 0.122,
+      "step": 35880
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.8021430963277152e-05,
+      "loss": 0.1356,
+      "step": 35890
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.8018082375265096e-05,
+      "loss": 0.2876,
+      "step": 35900
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.801473378725304e-05,
+      "loss": 0.1035,
+      "step": 35910
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.8011385199240985e-05,
+      "loss": 0.2213,
+      "step": 35920
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.8008036611228932e-05,
+      "loss": 0.1211,
+      "step": 35930
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.800468802321688e-05,
+      "loss": 0.0948,
+      "step": 35940
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.8001339435204824e-05,
+      "loss": 0.1619,
+      "step": 35950
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.7997990847192768e-05,
+      "loss": 0.1557,
+      "step": 35960
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.799464225918071e-05,
+      "loss": 0.0613,
+      "step": 35970
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.7991293671168656e-05,
+      "loss": 0.0825,
+      "step": 35980
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.7987945083156603e-05,
+      "loss": 0.106,
+      "step": 35990
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.7984596495144547e-05,
+      "loss": 0.1104,
+      "step": 36000
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.7981247907132495e-05,
+      "loss": 0.1072,
+      "step": 36010
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.797789931912044e-05,
+      "loss": 0.08,
+      "step": 36020
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.7974550731108383e-05,
+      "loss": 0.273,
+      "step": 36030
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.7971202143096327e-05,
+      "loss": 0.0833,
+      "step": 36040
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.796785355508427e-05,
+      "loss": 0.0702,
+      "step": 36050
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.796450496707222e-05,
+      "loss": 0.0463,
+      "step": 36060
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.7961156379060163e-05,
+      "loss": 0.085,
+      "step": 36070
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.795780779104811e-05,
+      "loss": 0.1203,
+      "step": 36080
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.7954459203036054e-05,
+      "loss": 0.168,
+      "step": 36090
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.7951110615023998e-05,
+      "loss": 0.2556,
+      "step": 36100
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.7947762027011942e-05,
+      "loss": 0.1456,
+      "step": 36110
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.794441343899989e-05,
+      "loss": 0.0649,
+      "step": 36120
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.7941064850987834e-05,
+      "loss": 0.074,
+      "step": 36130
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.7937716262975778e-05,
+      "loss": 0.0697,
+      "step": 36140
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.7934367674963725e-05,
+      "loss": 0.0796,
+      "step": 36150
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.793101908695167e-05,
+      "loss": 0.0885,
+      "step": 36160
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.7927670498939613e-05,
+      "loss": 0.0933,
+      "step": 36170
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.7924321910927558e-05,
+      "loss": 0.1005,
+      "step": 36180
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.7920973322915505e-05,
+      "loss": 0.1483,
+      "step": 36190
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.791762473490345e-05,
+      "loss": 0.1345,
+      "step": 36200
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.7914276146891393e-05,
+      "loss": 0.1235,
+      "step": 36210
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.791092755887934e-05,
+      "loss": 0.142,
+      "step": 36220
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.7907578970867285e-05,
+      "loss": 0.2106,
+      "step": 36230
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.790423038285523e-05,
+      "loss": 0.1015,
+      "step": 36240
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.7900881794843176e-05,
+      "loss": 0.2125,
+      "step": 36250
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.789753320683112e-05,
+      "loss": 0.084,
+      "step": 36260
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.7894184618819064e-05,
+      "loss": 0.0522,
+      "step": 36270
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.7890836030807012e-05,
+      "loss": 0.1755,
+      "step": 36280
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.7887487442794956e-05,
+      "loss": 0.2254,
+      "step": 36290
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.78841388547829e-05,
+      "loss": 0.1357,
+      "step": 36300
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.7880790266770844e-05,
+      "loss": 0.0867,
+      "step": 36310
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.787744167875879e-05,
+      "loss": 0.1159,
+      "step": 36320
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.7874093090746736e-05,
+      "loss": 0.1389,
+      "step": 36330
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.787074450273468e-05,
+      "loss": 0.1946,
+      "step": 36340
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.7867395914722627e-05,
+      "loss": 0.1111,
+      "step": 36350
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.786404732671057e-05,
+      "loss": 0.1748,
+      "step": 36360
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7860698738698515e-05,
+      "loss": 0.1516,
+      "step": 36370
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7857350150686463e-05,
+      "loss": 0.0675,
+      "step": 36380
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7854001562674407e-05,
+      "loss": 0.0903,
+      "step": 36390
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.785065297466235e-05,
+      "loss": 0.1791,
+      "step": 36400
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7847304386650295e-05,
+      "loss": 0.1277,
+      "step": 36410
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7843955798638242e-05,
+      "loss": 0.0847,
+      "step": 36420
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7840607210626186e-05,
+      "loss": 0.1008,
+      "step": 36430
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.783725862261413e-05,
+      "loss": 0.0905,
+      "step": 36440
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7833910034602078e-05,
+      "loss": 0.1194,
+      "step": 36450
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7830561446590022e-05,
+      "loss": 0.0838,
+      "step": 36460
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7827212858577966e-05,
+      "loss": 0.084,
+      "step": 36470
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.782386427056591e-05,
+      "loss": 0.1155,
+      "step": 36480
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7820515682553858e-05,
+      "loss": 0.1163,
+      "step": 36490
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7817167094541802e-05,
+      "loss": 0.1201,
+      "step": 36500
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.781381850652975e-05,
+      "loss": 0.0962,
+      "step": 36510
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7810469918517693e-05,
+      "loss": 0.0539,
+      "step": 36520
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7807121330505637e-05,
+      "loss": 0.2177,
+      "step": 36530
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.780377274249358e-05,
+      "loss": 0.075,
+      "step": 36540
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7800424154481526e-05,
+      "loss": 0.1615,
+      "step": 36550
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7797075566469473e-05,
+      "loss": 0.142,
+      "step": 36560
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7793726978457417e-05,
+      "loss": 0.0803,
+      "step": 36570
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7790378390445365e-05,
+      "loss": 0.1068,
+      "step": 36580
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.778702980243331e-05,
+      "loss": 0.1731,
+      "step": 36590
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7783681214421253e-05,
+      "loss": 0.1217,
+      "step": 36600
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7780332626409197e-05,
+      "loss": 0.1329,
+      "step": 36610
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.777698403839714e-05,
+      "loss": 0.1384,
+      "step": 36620
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7773635450385088e-05,
+      "loss": 0.1355,
+      "step": 36630
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7770286862373036e-05,
+      "loss": 0.0664,
+      "step": 36640
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.776693827436098e-05,
+      "loss": 0.0906,
+      "step": 36650
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7763589686348924e-05,
+      "loss": 0.176,
+      "step": 36660
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7760241098336868e-05,
+      "loss": 0.1767,
+      "step": 36670
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7756892510324812e-05,
+      "loss": 0.134,
+      "step": 36680
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7753543922312756e-05,
+      "loss": 0.0903,
+      "step": 36690
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7750195334300704e-05,
+      "loss": 0.0755,
+      "step": 36700
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.774684674628865e-05,
+      "loss": 0.128,
+      "step": 36710
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7743498158276595e-05,
+      "loss": 0.177,
+      "step": 36720
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.774014957026454e-05,
+      "loss": 0.0724,
+      "step": 36730
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.7736800982252483e-05,
+      "loss": 0.0917,
+      "step": 36740
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.7733452394240427e-05,
+      "loss": 0.0888,
+      "step": 36750
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.773010380622837e-05,
+      "loss": 0.0902,
+      "step": 36760
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.7726755218216322e-05,
+      "loss": 0.2507,
+      "step": 36770
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.7723406630204266e-05,
+      "loss": 0.1332,
+      "step": 36780
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.772005804219221e-05,
+      "loss": 0.1007,
+      "step": 36790
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.7716709454180155e-05,
+      "loss": 0.1196,
+      "step": 36800
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.77133608661681e-05,
+      "loss": 0.2155,
+      "step": 36810
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.7710012278156043e-05,
+      "loss": 0.1011,
+      "step": 36820
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.7706663690143987e-05,
+      "loss": 0.2034,
+      "step": 36830
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.7703315102131938e-05,
+      "loss": 0.1606,
+      "step": 36840
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.769996651411988e-05,
+      "loss": 0.1521,
+      "step": 36850
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.7696617926107826e-05,
+      "loss": 0.1092,
+      "step": 36860
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.769326933809577e-05,
+      "loss": 0.1365,
+      "step": 36870
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.7689920750083714e-05,
+      "loss": 0.2084,
+      "step": 36880
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.7686572162071658e-05,
+      "loss": 0.1248,
+      "step": 36890
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.7683223574059605e-05,
+      "loss": 0.173,
+      "step": 36900
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.7679874986047553e-05,
+      "loss": 0.1217,
+      "step": 36910
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.7676526398035497e-05,
+      "loss": 0.0729,
+      "step": 36920
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.767317781002344e-05,
+      "loss": 0.1959,
+      "step": 36930
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.7669829222011385e-05,
+      "loss": 0.1357,
+      "step": 36940
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.766648063399933e-05,
+      "loss": 0.1452,
+      "step": 36950
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.7663132045987273e-05,
+      "loss": 0.1267,
+      "step": 36960
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.765978345797522e-05,
+      "loss": 0.2213,
+      "step": 36970
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.7656434869963168e-05,
+      "loss": 0.2472,
+      "step": 36980
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.7653086281951112e-05,
+      "loss": 0.2294,
+      "step": 36990
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.7649737693939056e-05,
+      "loss": 0.1391,
+      "step": 37000
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.7646389105927e-05,
+      "loss": 0.0947,
+      "step": 37010
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.7643040517914944e-05,
+      "loss": 0.1393,
+      "step": 37020
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.7639691929902892e-05,
+      "loss": 0.1444,
+      "step": 37030
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.7636343341890836e-05,
+      "loss": 0.0906,
+      "step": 37040
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.7632994753878783e-05,
+      "loss": 0.1053,
+      "step": 37050
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.7629646165866728e-05,
+      "loss": 0.0872,
+      "step": 37060
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.762629757785467e-05,
+      "loss": 0.0683,
+      "step": 37070
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.7622948989842616e-05,
+      "loss": 0.2728,
+      "step": 37080
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.761960040183056e-05,
+      "loss": 0.1214,
+      "step": 37090
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.7616251813818507e-05,
+      "loss": 0.1197,
+      "step": 37100
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.761290322580645e-05,
+      "loss": 0.0887,
+      "step": 37110
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.76095546377944e-05,
+      "loss": 0.0877,
+      "step": 37120
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.7606206049782343e-05,
+      "loss": 0.0934,
+      "step": 37130
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.7602857461770287e-05,
+      "loss": 0.3071,
+      "step": 37140
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.759950887375823e-05,
+      "loss": 0.1627,
+      "step": 37150
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.7596160285746175e-05,
+      "loss": 0.0702,
+      "step": 37160
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.7592811697734123e-05,
+      "loss": 0.0337,
+      "step": 37170
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.758946310972207e-05,
+      "loss": 0.0817,
+      "step": 37180
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.7586114521710014e-05,
+      "loss": 0.0736,
+      "step": 37190
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.7582765933697958e-05,
+      "loss": 0.1363,
+      "step": 37200
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.7579417345685902e-05,
+      "loss": 0.1923,
+      "step": 37210
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.7576068757673846e-05,
+      "loss": 0.0988,
+      "step": 37220
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.7572720169661794e-05,
+      "loss": 0.2035,
+      "step": 37230
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.7569371581649738e-05,
+      "loss": 0.1652,
+      "step": 37240
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.7566022993637685e-05,
+      "loss": 0.1231,
+      "step": 37250
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.756267440562563e-05,
+      "loss": 0.1181,
+      "step": 37260
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.7559325817613573e-05,
+      "loss": 0.1136,
+      "step": 37270
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.7555977229601517e-05,
+      "loss": 0.2384,
+      "step": 37280
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.755262864158946e-05,
+      "loss": 0.1173,
+      "step": 37290
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.754928005357741e-05,
+      "loss": 0.1729,
+      "step": 37300
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.7545931465565353e-05,
+      "loss": 0.1295,
+      "step": 37310
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.75425828775533e-05,
+      "loss": 0.11,
+      "step": 37320
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.7539234289541245e-05,
+      "loss": 0.1213,
+      "step": 37330
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.753588570152919e-05,
+      "loss": 0.0912,
+      "step": 37340
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.7532537113517133e-05,
+      "loss": 0.0551,
+      "step": 37350
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.752918852550508e-05,
+      "loss": 0.1003,
+      "step": 37360
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.7525839937493024e-05,
+      "loss": 0.1051,
+      "step": 37370
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.752249134948097e-05,
+      "loss": 0.1782,
+      "step": 37380
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.7519142761468916e-05,
+      "loss": 0.0976,
+      "step": 37390
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.751579417345686e-05,
+      "loss": 0.1051,
+      "step": 37400
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.7512445585444804e-05,
+      "loss": 0.1044,
+      "step": 37410
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.7509096997432748e-05,
+      "loss": 0.2285,
+      "step": 37420
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.7505748409420696e-05,
+      "loss": 0.2352,
+      "step": 37430
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.750239982140864e-05,
+      "loss": 0.0809,
+      "step": 37440
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.7499051233396584e-05,
+      "loss": 0.1398,
+      "step": 37450
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.749570264538453e-05,
+      "loss": 0.1412,
+      "step": 37460
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.7492354057372475e-05,
+      "loss": 0.1186,
+      "step": 37470
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.748900546936042e-05,
+      "loss": 0.1345,
+      "step": 37480
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.7485656881348367e-05,
+      "loss": 0.1566,
+      "step": 37490
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.748230829333631e-05,
+      "loss": 0.1365,
+      "step": 37500
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.7478959705324255e-05,
+      "loss": 0.1031,
+      "step": 37510
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.74756111173122e-05,
+      "loss": 0.0628,
+      "step": 37520
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.7472262529300146e-05,
+      "loss": 0.1546,
+      "step": 37530
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.746891394128809e-05,
+      "loss": 0.1831,
+      "step": 37540
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.7465565353276035e-05,
+      "loss": 0.069,
+      "step": 37550
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.7462216765263982e-05,
+      "loss": 0.079,
+      "step": 37560
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.7458868177251926e-05,
+      "loss": 0.1364,
+      "step": 37570
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.745551958923987e-05,
+      "loss": 0.3023,
+      "step": 37580
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.7452171001227814e-05,
+      "loss": 0.0648,
+      "step": 37590
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.7448822413215762e-05,
+      "loss": 0.1264,
+      "step": 37600
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.7445473825203706e-05,
+      "loss": 0.1523,
+      "step": 37610
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.7442125237191653e-05,
+      "loss": 0.0641,
+      "step": 37620
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.7438776649179597e-05,
+      "loss": 0.2904,
+      "step": 37630
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.743542806116754e-05,
+      "loss": 0.0492,
+      "step": 37640
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.7432079473155486e-05,
+      "loss": 0.0662,
+      "step": 37650
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.742873088514343e-05,
+      "loss": 0.145,
+      "step": 37660
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.7425382297131377e-05,
+      "loss": 0.1046,
+      "step": 37670
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.742203370911932e-05,
+      "loss": 0.2207,
+      "step": 37680
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.741868512110727e-05,
+      "loss": 0.1438,
+      "step": 37690
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.7415336533095213e-05,
+      "loss": 0.1405,
+      "step": 37700
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.7411987945083157e-05,
+      "loss": 0.0393,
+      "step": 37710
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.74086393570711e-05,
+      "loss": 0.0812,
+      "step": 37720
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.7405290769059045e-05,
+      "loss": 0.2095,
+      "step": 37730
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.7401942181046992e-05,
+      "loss": 0.167,
+      "step": 37740
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.739859359303494e-05,
+      "loss": 0.0653,
+      "step": 37750
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.7395245005022884e-05,
+      "loss": 0.178,
+      "step": 37760
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.7391896417010828e-05,
+      "loss": 0.111,
+      "step": 37770
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.7388547828998772e-05,
+      "loss": 0.2042,
+      "step": 37780
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.7385199240986716e-05,
+      "loss": 0.1491,
+      "step": 37790
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.738185065297466e-05,
+      "loss": 0.1554,
+      "step": 37800
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7378502064962608e-05,
+      "loss": 0.1187,
+      "step": 37810
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7375153476950555e-05,
+      "loss": 0.0619,
+      "step": 37820
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.73718048889385e-05,
+      "loss": 0.158,
+      "step": 37830
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7368456300926443e-05,
+      "loss": 0.206,
+      "step": 37840
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7365107712914387e-05,
+      "loss": 0.0908,
+      "step": 37850
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.736175912490233e-05,
+      "loss": 0.0824,
+      "step": 37860
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7358410536890275e-05,
+      "loss": 0.1088,
+      "step": 37870
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7355061948878226e-05,
+      "loss": 0.0845,
+      "step": 37880
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.735171336086617e-05,
+      "loss": 0.091,
+      "step": 37890
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7348364772854114e-05,
+      "loss": 0.1883,
+      "step": 37900
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.734501618484206e-05,
+      "loss": 0.1276,
+      "step": 37910
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7341667596830003e-05,
+      "loss": 0.0894,
+      "step": 37920
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7338319008817947e-05,
+      "loss": 0.153,
+      "step": 37930
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.733497042080589e-05,
+      "loss": 0.0637,
+      "step": 37940
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.733162183279384e-05,
+      "loss": 0.2292,
+      "step": 37950
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7328273244781786e-05,
+      "loss": 0.0487,
+      "step": 37960
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.732492465676973e-05,
+      "loss": 0.1883,
+      "step": 37970
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7321576068757674e-05,
+      "loss": 0.0949,
+      "step": 37980
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7318227480745618e-05,
+      "loss": 0.1261,
+      "step": 37990
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7314878892733562e-05,
+      "loss": 0.0921,
+      "step": 38000
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.731153030472151e-05,
+      "loss": 0.1638,
+      "step": 38010
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7308181716709457e-05,
+      "loss": 0.1268,
+      "step": 38020
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.73048331286974e-05,
+      "loss": 0.1376,
+      "step": 38030
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7301484540685345e-05,
+      "loss": 0.0745,
+      "step": 38040
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.729813595267329e-05,
+      "loss": 0.1084,
+      "step": 38050
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7294787364661233e-05,
+      "loss": 0.1065,
+      "step": 38060
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7291438776649177e-05,
+      "loss": 0.0525,
+      "step": 38070
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.7288090188637128e-05,
+      "loss": 0.1722,
+      "step": 38080
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.7284741600625072e-05,
+      "loss": 0.1466,
+      "step": 38090
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.7281393012613016e-05,
+      "loss": 0.2343,
+      "step": 38100
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.727804442460096e-05,
+      "loss": 0.1592,
+      "step": 38110
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.7274695836588904e-05,
+      "loss": 0.2043,
+      "step": 38120
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.727134724857685e-05,
+      "loss": 0.1609,
+      "step": 38130
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.7267998660564796e-05,
+      "loss": 0.0856,
+      "step": 38140
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.7264650072552743e-05,
+      "loss": 0.1048,
+      "step": 38150
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.7261301484540687e-05,
+      "loss": 0.1045,
+      "step": 38160
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.725795289652863e-05,
+      "loss": 0.1477,
+      "step": 38170
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.7254604308516576e-05,
+      "loss": 0.2484,
+      "step": 38180
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.725125572050452e-05,
+      "loss": 0.1225,
+      "step": 38190
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.7247907132492464e-05,
+      "loss": 0.1564,
+      "step": 38200
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.724455854448041e-05,
+      "loss": 0.1557,
+      "step": 38210
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.724120995646836e-05,
+      "loss": 0.1141,
+      "step": 38220
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.7237861368456303e-05,
+      "loss": 0.0942,
+      "step": 38230
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.7234512780444247e-05,
+      "loss": 0.1483,
+      "step": 38240
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.723116419243219e-05,
+      "loss": 0.2396,
+      "step": 38250
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.7227815604420135e-05,
+      "loss": 0.1611,
+      "step": 38260
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.7224467016408082e-05,
+      "loss": 0.1037,
+      "step": 38270
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.7221118428396027e-05,
+      "loss": 0.2479,
+      "step": 38280
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.7217769840383974e-05,
+      "loss": 0.1084,
+      "step": 38290
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.7214421252371918e-05,
+      "loss": 0.0764,
+      "step": 38300
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.7211072664359862e-05,
+      "loss": 0.1116,
+      "step": 38310
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.7207724076347806e-05,
+      "loss": 0.0873,
+      "step": 38320
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.720437548833575e-05,
+      "loss": 0.1629,
+      "step": 38330
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.7201026900323698e-05,
+      "loss": 0.0731,
+      "step": 38340
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.7197678312311642e-05,
+      "loss": 0.16,
+      "step": 38350
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.719432972429959e-05,
+      "loss": 0.0918,
+      "step": 38360
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.7190981136287533e-05,
+      "loss": 0.0765,
+      "step": 38370
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.7187632548275477e-05,
+      "loss": 0.2476,
+      "step": 38380
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.718428396026342e-05,
+      "loss": 0.1295,
+      "step": 38390
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.718093537225137e-05,
+      "loss": 0.1528,
+      "step": 38400
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.7177586784239313e-05,
+      "loss": 0.0952,
+      "step": 38410
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.7174238196227257e-05,
+      "loss": 0.0686,
+      "step": 38420
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.7170889608215205e-05,
+      "loss": 0.1723,
+      "step": 38430
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.716754102020315e-05,
+      "loss": 0.166,
+      "step": 38440
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.7164192432191093e-05,
+      "loss": 0.0766,
+      "step": 38450
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.7160843844179037e-05,
+      "loss": 0.1304,
+      "step": 38460
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.7157495256166984e-05,
+      "loss": 0.1377,
+      "step": 38470
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.715414666815493e-05,
+      "loss": 0.0739,
+      "step": 38480
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.7150798080142872e-05,
+      "loss": 0.1198,
+      "step": 38490
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.714744949213082e-05,
+      "loss": 0.1052,
+      "step": 38500
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.7144100904118764e-05,
+      "loss": 0.0723,
+      "step": 38510
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.7140752316106708e-05,
+      "loss": 0.1436,
+      "step": 38520
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.7137403728094656e-05,
+      "loss": 0.164,
+      "step": 38530
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.71340551400826e-05,
+      "loss": 0.1254,
+      "step": 38540
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.7130706552070544e-05,
+      "loss": 0.1918,
+      "step": 38550
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.7127357964058488e-05,
+      "loss": 0.1407,
+      "step": 38560
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.7124009376046435e-05,
+      "loss": 0.1514,
+      "step": 38570
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.712066078803438e-05,
+      "loss": 0.2435,
+      "step": 38580
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.7117312200022323e-05,
+      "loss": 0.08,
+      "step": 38590
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.711396361201027e-05,
+      "loss": 0.1431,
+      "step": 38600
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.7110615023998215e-05,
+      "loss": 0.1305,
+      "step": 38610
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.710726643598616e-05,
+      "loss": 0.0641,
+      "step": 38620
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.7103917847974103e-05,
+      "loss": 0.2847,
+      "step": 38630
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.710056925996205e-05,
+      "loss": 0.1401,
+      "step": 38640
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.7097220671949995e-05,
+      "loss": 0.0806,
+      "step": 38650
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.709387208393794e-05,
+      "loss": 0.1556,
+      "step": 38660
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.7090523495925886e-05,
+      "loss": 0.0635,
+      "step": 38670
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.708717490791383e-05,
+      "loss": 0.1511,
+      "step": 38680
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.7083826319901774e-05,
+      "loss": 0.0679,
+      "step": 38690
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.708047773188972e-05,
+      "loss": 0.0842,
+      "step": 38700
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.7077129143877666e-05,
+      "loss": 0.1363,
+      "step": 38710
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.707378055586561e-05,
+      "loss": 0.1077,
+      "step": 38720
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.7070431967853557e-05,
+      "loss": 0.1106,
+      "step": 38730
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.70670833798415e-05,
+      "loss": 0.0917,
+      "step": 38740
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.7063734791829445e-05,
+      "loss": 0.1316,
+      "step": 38750
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.706038620381739e-05,
+      "loss": 0.1386,
+      "step": 38760
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.7057037615805334e-05,
+      "loss": 0.1327,
+      "step": 38770
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.705368902779328e-05,
+      "loss": 0.0819,
+      "step": 38780
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.7050340439781225e-05,
+      "loss": 0.1915,
+      "step": 38790
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.7046991851769173e-05,
+      "loss": 0.0535,
+      "step": 38800
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.7043643263757117e-05,
+      "loss": 0.1874,
+      "step": 38810
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.704029467574506e-05,
+      "loss": 0.1466,
+      "step": 38820
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.7036946087733005e-05,
+      "loss": 0.127,
+      "step": 38830
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.703359749972095e-05,
+      "loss": 0.1762,
+      "step": 38840
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.7030583770510103e-05,
+      "loss": 0.1587,
+      "step": 38850
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.7027235182498047e-05,
+      "loss": 0.0748,
+      "step": 38860
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.702388659448599e-05,
+      "loss": 0.145,
+      "step": 38870
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.7020538006473935e-05,
+      "loss": 0.1031,
+      "step": 38880
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.7017189418461883e-05,
+      "loss": 0.1817,
+      "step": 38890
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.7013840830449827e-05,
+      "loss": 0.0681,
+      "step": 38900
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.7010492242437774e-05,
+      "loss": 0.1975,
+      "step": 38910
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.7007143654425718e-05,
+      "loss": 0.1466,
+      "step": 38920
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.7003795066413662e-05,
+      "loss": 0.1089,
+      "step": 38930
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.7000446478401606e-05,
+      "loss": 0.1276,
+      "step": 38940
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.699709789038955e-05,
+      "loss": 0.0988,
+      "step": 38950
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.6993749302377498e-05,
+      "loss": 0.1137,
+      "step": 38960
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.6990400714365445e-05,
+      "loss": 0.0571,
+      "step": 38970
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.698705212635339e-05,
+      "loss": 0.1285,
+      "step": 38980
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.6983703538341334e-05,
+      "loss": 0.0707,
+      "step": 38990
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.6980354950329278e-05,
+      "loss": 0.0907,
+      "step": 39000
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.6977006362317222e-05,
+      "loss": 0.1196,
+      "step": 39010
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.6973657774305166e-05,
+      "loss": 0.2067,
+      "step": 39020
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.6970309186293113e-05,
+      "loss": 0.1479,
+      "step": 39030
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.696696059828106e-05,
+      "loss": 0.1057,
+      "step": 39040
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.6963612010269005e-05,
+      "loss": 0.1054,
+      "step": 39050
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.696026342225695e-05,
+      "loss": 0.1078,
+      "step": 39060
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.6956914834244893e-05,
+      "loss": 0.1616,
+      "step": 39070
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.6953566246232837e-05,
+      "loss": 0.096,
+      "step": 39080
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.695021765822078e-05,
+      "loss": 0.0535,
+      "step": 39090
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.6946869070208732e-05,
+      "loss": 0.1383,
+      "step": 39100
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.6943520482196676e-05,
+      "loss": 0.1461,
+      "step": 39110
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.694017189418462e-05,
+      "loss": 0.0656,
+      "step": 39120
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.6936823306172564e-05,
+      "loss": 0.1289,
+      "step": 39130
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.6933474718160508e-05,
+      "loss": 0.0592,
+      "step": 39140
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.6930126130148452e-05,
+      "loss": 0.0986,
+      "step": 39150
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.6926777542136396e-05,
+      "loss": 0.1043,
+      "step": 39160
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.6923428954124347e-05,
+      "loss": 0.0611,
+      "step": 39170
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.692008036611229e-05,
+      "loss": 0.1431,
+      "step": 39180
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.6916731778100235e-05,
+      "loss": 0.1326,
+      "step": 39190
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.691338319008818e-05,
+      "loss": 0.1066,
+      "step": 39200
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.6910034602076124e-05,
+      "loss": 0.1659,
+      "step": 39210
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.6906686014064068e-05,
+      "loss": 0.1471,
+      "step": 39220
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.690333742605202e-05,
+      "loss": 0.2086,
+      "step": 39230
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.6899988838039963e-05,
+      "loss": 0.0728,
+      "step": 39240
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.6896640250027907e-05,
+      "loss": 0.0527,
+      "step": 39250
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.689329166201585e-05,
+      "loss": 0.0939,
+      "step": 39260
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.6889943074003795e-05,
+      "loss": 0.1321,
+      "step": 39270
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.688659448599174e-05,
+      "loss": 0.1561,
+      "step": 39280
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.6883245897979683e-05,
+      "loss": 0.0989,
+      "step": 39290
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.6879897309967634e-05,
+      "loss": 0.1442,
+      "step": 39300
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.6876548721955578e-05,
+      "loss": 0.1584,
+      "step": 39310
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.6873200133943522e-05,
+      "loss": 0.113,
+      "step": 39320
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.6869851545931466e-05,
+      "loss": 0.075,
+      "step": 39330
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.686650295791941e-05,
+      "loss": 0.0847,
+      "step": 39340
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.6863154369907354e-05,
+      "loss": 0.0901,
+      "step": 39350
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.68598057818953e-05,
+      "loss": 0.1492,
+      "step": 39360
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.685645719388325e-05,
+      "loss": 0.0872,
+      "step": 39370
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.6853108605871193e-05,
+      "loss": 0.1673,
+      "step": 39380
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.6849760017859137e-05,
+      "loss": 0.1745,
+      "step": 39390
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.684641142984708e-05,
+      "loss": 0.0904,
+      "step": 39400
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.6843062841835025e-05,
+      "loss": 0.1453,
+      "step": 39410
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.683971425382297e-05,
+      "loss": 0.2012,
+      "step": 39420
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.6836365665810917e-05,
+      "loss": 0.1081,
+      "step": 39430
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.6833017077798864e-05,
+      "loss": 0.0751,
+      "step": 39440
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.682966848978681e-05,
+      "loss": 0.0934,
+      "step": 39450
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.6826319901774752e-05,
+      "loss": 0.11,
+      "step": 39460
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.6822971313762697e-05,
+      "loss": 0.082,
+      "step": 39470
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.681962272575064e-05,
+      "loss": 0.2671,
+      "step": 39480
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.6816274137738588e-05,
+      "loss": 0.0947,
+      "step": 39490
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.6812925549726532e-05,
+      "loss": 0.1204,
+      "step": 39500
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.680957696171448e-05,
+      "loss": 0.1158,
+      "step": 39510
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.6806228373702424e-05,
+      "loss": 0.1359,
+      "step": 39520
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.6802879785690368e-05,
+      "loss": 0.141,
+      "step": 39530
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.6799531197678312e-05,
+      "loss": 0.1385,
+      "step": 39540
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.6796182609666256e-05,
+      "loss": 0.1943,
+      "step": 39550
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.6792834021654203e-05,
+      "loss": 0.0692,
+      "step": 39560
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.6789485433642147e-05,
+      "loss": 0.2854,
+      "step": 39570
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.6786136845630095e-05,
+      "loss": 0.1039,
+      "step": 39580
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.678278825761804e-05,
+      "loss": 0.0944,
+      "step": 39590
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.6779439669605983e-05,
+      "loss": 0.1011,
+      "step": 39600
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.6776091081593927e-05,
+      "loss": 0.0677,
+      "step": 39610
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.6772742493581875e-05,
+      "loss": 0.0939,
+      "step": 39620
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.676939390556982e-05,
+      "loss": 0.1265,
+      "step": 39630
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.6766045317557763e-05,
+      "loss": 0.0705,
+      "step": 39640
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.676269672954571e-05,
+      "loss": 0.1426,
+      "step": 39650
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.6759348141533654e-05,
+      "loss": 0.1197,
+      "step": 39660
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.67559995535216e-05,
+      "loss": 0.1242,
+      "step": 39670
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.6752650965509542e-05,
+      "loss": 0.2101,
+      "step": 39680
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.674930237749749e-05,
+      "loss": 0.1093,
+      "step": 39690
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6745953789485434e-05,
+      "loss": 0.1432,
+      "step": 39700
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6742605201473378e-05,
+      "loss": 0.1083,
+      "step": 39710
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6739256613461326e-05,
+      "loss": 0.0802,
+      "step": 39720
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.673590802544927e-05,
+      "loss": 0.1272,
+      "step": 39730
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6732559437437214e-05,
+      "loss": 0.1924,
+      "step": 39740
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.672921084942516e-05,
+      "loss": 0.2734,
+      "step": 39750
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6725862261413105e-05,
+      "loss": 0.14,
+      "step": 39760
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.672251367340105e-05,
+      "loss": 0.1289,
+      "step": 39770
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6719165085388993e-05,
+      "loss": 0.1755,
+      "step": 39780
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.671581649737694e-05,
+      "loss": 0.1002,
+      "step": 39790
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6712467909364885e-05,
+      "loss": 0.1627,
+      "step": 39800
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.670911932135283e-05,
+      "loss": 0.0767,
+      "step": 39810
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6705770733340776e-05,
+      "loss": 0.2087,
+      "step": 39820
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.670242214532872e-05,
+      "loss": 0.1997,
+      "step": 39830
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6699073557316665e-05,
+      "loss": 0.1072,
+      "step": 39840
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.669572496930461e-05,
+      "loss": 0.0935,
+      "step": 39850
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6692376381292556e-05,
+      "loss": 0.1737,
+      "step": 39860
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.66890277932805e-05,
+      "loss": 0.1242,
+      "step": 39870
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.6685679205268444e-05,
+      "loss": 0.1579,
+      "step": 39880
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.6682330617256392e-05,
+      "loss": 0.0899,
+      "step": 39890
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.6678982029244336e-05,
+      "loss": 0.079,
+      "step": 39900
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.667563344123228e-05,
+      "loss": 0.0401,
+      "step": 39910
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.6672284853220224e-05,
+      "loss": 0.117,
+      "step": 39920
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.666893626520817e-05,
+      "loss": 0.1954,
+      "step": 39930
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.6665587677196115e-05,
+      "loss": 0.0763,
+      "step": 39940
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.6662239089184063e-05,
+      "loss": 0.1701,
+      "step": 39950
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.6658890501172007e-05,
+      "loss": 0.0786,
+      "step": 39960
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.665554191315995e-05,
+      "loss": 0.0913,
+      "step": 39970
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.6652193325147895e-05,
+      "loss": 0.169,
+      "step": 39980
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.664884473713584e-05,
+      "loss": 0.1042,
+      "step": 39990
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.6645496149123787e-05,
+      "loss": 0.0766,
+      "step": 40000
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.664214756111173e-05,
+      "loss": 0.0919,
+      "step": 40010
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.6638798973099678e-05,
+      "loss": 0.0943,
+      "step": 40020
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.6635450385087622e-05,
+      "loss": 0.1749,
+      "step": 40030
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.6632101797075566e-05,
+      "loss": 0.0877,
+      "step": 40040
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.662875320906351e-05,
+      "loss": 0.0662,
+      "step": 40050
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.6625404621051455e-05,
+      "loss": 0.1017,
+      "step": 40060
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.6622056033039402e-05,
+      "loss": 0.1682,
+      "step": 40070
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.661870744502735e-05,
+      "loss": 0.2449,
+      "step": 40080
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.6615358857015294e-05,
+      "loss": 0.0672,
+      "step": 40090
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.6612010269003238e-05,
+      "loss": 0.0404,
+      "step": 40100
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.660866168099118e-05,
+      "loss": 0.0766,
+      "step": 40110
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.6605313092979126e-05,
+      "loss": 0.2149,
+      "step": 40120
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.6601964504967073e-05,
+      "loss": 0.1668,
+      "step": 40130
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.6598615916955017e-05,
+      "loss": 0.1397,
+      "step": 40140
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.6595267328942965e-05,
+      "loss": 0.1158,
+      "step": 40150
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.659191874093091e-05,
+      "loss": 0.153,
+      "step": 40160
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.6588570152918853e-05,
+      "loss": 0.1687,
+      "step": 40170
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.6585221564906797e-05,
+      "loss": 0.186,
+      "step": 40180
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.658187297689474e-05,
+      "loss": 0.1261,
+      "step": 40190
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.657852438888269e-05,
+      "loss": 0.0948,
+      "step": 40200
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.6575175800870636e-05,
+      "loss": 0.1121,
+      "step": 40210
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.657182721285858e-05,
+      "loss": 0.1104,
+      "step": 40220
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.6568478624846524e-05,
+      "loss": 0.18,
+      "step": 40230
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.6565130036834468e-05,
+      "loss": 0.1255,
+      "step": 40240
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.6561781448822412e-05,
+      "loss": 0.112,
+      "step": 40250
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.6558432860810356e-05,
+      "loss": 0.1669,
+      "step": 40260
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.6555084272798304e-05,
+      "loss": 0.0938,
+      "step": 40270
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.655173568478625e-05,
+      "loss": 0.0741,
+      "step": 40280
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.6548387096774195e-05,
+      "loss": 0.0763,
+      "step": 40290
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.654503850876214e-05,
+      "loss": 0.0899,
+      "step": 40300
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.6541689920750083e-05,
+      "loss": 0.094,
+      "step": 40310
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.6538341332738028e-05,
+      "loss": 0.1245,
+      "step": 40320
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.653499274472597e-05,
+      "loss": 0.0971,
+      "step": 40330
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.6531644156713922e-05,
+      "loss": 0.1378,
+      "step": 40340
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.6528295568701867e-05,
+      "loss": 0.1213,
+      "step": 40350
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.652494698068981e-05,
+      "loss": 0.1679,
+      "step": 40360
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.6521598392677755e-05,
+      "loss": 0.0901,
+      "step": 40370
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.65182498046657e-05,
+      "loss": 0.0762,
+      "step": 40380
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.6514901216653643e-05,
+      "loss": 0.135,
+      "step": 40390
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.6511552628641587e-05,
+      "loss": 0.0677,
+      "step": 40400
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.6508204040629538e-05,
+      "loss": 0.0968,
+      "step": 40410
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.6504855452617482e-05,
+      "loss": 0.1136,
+      "step": 40420
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.6501506864605426e-05,
+      "loss": 0.1684,
+      "step": 40430
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.649815827659337e-05,
+      "loss": 0.095,
+      "step": 40440
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.6494809688581314e-05,
+      "loss": 0.1076,
+      "step": 40450
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.6491461100569258e-05,
+      "loss": 0.1901,
+      "step": 40460
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.6488112512557206e-05,
+      "loss": 0.06,
+      "step": 40470
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.6484763924545153e-05,
+      "loss": 0.1351,
+      "step": 40480
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.6481415336533097e-05,
+      "loss": 0.0686,
+      "step": 40490
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.647806674852104e-05,
+      "loss": 0.0783,
+      "step": 40500
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.6474718160508985e-05,
+      "loss": 0.1064,
+      "step": 40510
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.647136957249693e-05,
+      "loss": 0.1618,
+      "step": 40520
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.6468020984484873e-05,
+      "loss": 0.1277,
+      "step": 40530
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.646467239647282e-05,
+      "loss": 0.0828,
+      "step": 40540
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.646132380846077e-05,
+      "loss": 0.1271,
+      "step": 40550
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.6457975220448712e-05,
+      "loss": 0.1675,
+      "step": 40560
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.6454626632436657e-05,
+      "loss": 0.0877,
+      "step": 40570
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.64512780444246e-05,
+      "loss": 0.162,
+      "step": 40580
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.6447929456412545e-05,
+      "loss": 0.0824,
+      "step": 40590
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.6444580868400492e-05,
+      "loss": 0.1314,
+      "step": 40600
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.6441232280388436e-05,
+      "loss": 0.0957,
+      "step": 40610
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.6437883692376384e-05,
+      "loss": 0.141,
+      "step": 40620
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.6434535104364328e-05,
+      "loss": 0.1329,
+      "step": 40630
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.6431186516352272e-05,
+      "loss": 0.1564,
+      "step": 40640
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.6427837928340216e-05,
+      "loss": 0.1524,
+      "step": 40650
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.642448934032816e-05,
+      "loss": 0.1106,
+      "step": 40660
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.6421140752316107e-05,
+      "loss": 0.0593,
+      "step": 40670
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.641779216430405e-05,
+      "loss": 0.1249,
+      "step": 40680
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.6414443576292e-05,
+      "loss": 0.1192,
+      "step": 40690
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.6411094988279943e-05,
+      "loss": 0.0838,
+      "step": 40700
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.6407746400267887e-05,
+      "loss": 0.1162,
+      "step": 40710
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.640439781225583e-05,
+      "loss": 0.0515,
+      "step": 40720
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.640104922424378e-05,
+      "loss": 0.105,
+      "step": 40730
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.6397700636231723e-05,
+      "loss": 0.1124,
+      "step": 40740
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.6394352048219667e-05,
+      "loss": 0.1763,
+      "step": 40750
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.6391003460207614e-05,
+      "loss": 0.3579,
+      "step": 40760
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.638765487219556e-05,
+      "loss": 0.095,
+      "step": 40770
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.6384306284183502e-05,
+      "loss": 0.0822,
+      "step": 40780
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.6380957696171446e-05,
+      "loss": 0.1068,
+      "step": 40790
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.6377609108159394e-05,
+      "loss": 0.1556,
+      "step": 40800
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.6374260520147338e-05,
+      "loss": 0.0998,
+      "step": 40810
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.6370911932135282e-05,
+      "loss": 0.066,
+      "step": 40820
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.636756334412323e-05,
+      "loss": 0.1582,
+      "step": 40830
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.6364214756111174e-05,
+      "loss": 0.115,
+      "step": 40840
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.6360866168099118e-05,
+      "loss": 0.1696,
+      "step": 40850
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.6357517580087065e-05,
+      "loss": 0.114,
+      "step": 40860
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.635416899207501e-05,
+      "loss": 0.1042,
+      "step": 40870
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.6350820404062953e-05,
+      "loss": 0.1849,
+      "step": 40880
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.6347471816050897e-05,
+      "loss": 0.0857,
+      "step": 40890
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.6344123228038845e-05,
+      "loss": 0.1079,
+      "step": 40900
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.634077464002679e-05,
+      "loss": 0.0829,
+      "step": 40910
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.6337426052014733e-05,
+      "loss": 0.0684,
+      "step": 40920
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.633407746400268e-05,
+      "loss": 0.2094,
+      "step": 40930
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.6330728875990625e-05,
+      "loss": 0.101,
+      "step": 40940
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.632738028797857e-05,
+      "loss": 0.0893,
+      "step": 40950
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.6324031699966513e-05,
+      "loss": 0.0765,
+      "step": 40960
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.632068311195446e-05,
+      "loss": 0.0754,
+      "step": 40970
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.6317334523942404e-05,
+      "loss": 0.1022,
+      "step": 40980
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.631398593593035e-05,
+      "loss": 0.0864,
+      "step": 40990
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.6310637347918296e-05,
+      "loss": 0.1382,
+      "step": 41000
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.630728875990624e-05,
+      "loss": 0.1012,
+      "step": 41010
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.6303940171894184e-05,
+      "loss": 0.142,
+      "step": 41020
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.630059158388213e-05,
+      "loss": 0.5725,
+      "step": 41030
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.6297242995870075e-05,
+      "loss": 0.0957,
+      "step": 41040
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.629389440785802e-05,
+      "loss": 0.1256,
+      "step": 41050
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.6290545819845967e-05,
+      "loss": 0.3441,
+      "step": 41060
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.628719723183391e-05,
+      "loss": 0.078,
+      "step": 41070
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.6283848643821855e-05,
+      "loss": 0.1596,
+      "step": 41080
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.62805000558098e-05,
+      "loss": 0.1463,
+      "step": 41090
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.6277151467797747e-05,
+      "loss": 0.0661,
+      "step": 41100
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.627380287978569e-05,
+      "loss": 0.1988,
+      "step": 41110
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.6270454291773638e-05,
+      "loss": 0.0881,
+      "step": 41120
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.6267105703761582e-05,
+      "loss": 0.1208,
+      "step": 41130
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.6263757115749526e-05,
+      "loss": 0.1156,
+      "step": 41140
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.626040852773747e-05,
+      "loss": 0.0904,
+      "step": 41150
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.6257059939725414e-05,
+      "loss": 0.0647,
+      "step": 41160
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.6253711351713362e-05,
+      "loss": 0.1039,
+      "step": 41170
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.6250362763701306e-05,
+      "loss": 0.1664,
+      "step": 41180
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.6247014175689253e-05,
+      "loss": 0.0744,
+      "step": 41190
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.6243665587677198e-05,
+      "loss": 0.1588,
+      "step": 41200
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.624031699966514e-05,
+      "loss": 0.1066,
+      "step": 41210
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.6236968411653086e-05,
+      "loss": 0.1495,
+      "step": 41220
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.623361982364103e-05,
+      "loss": 0.207,
+      "step": 41230
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.6230271235628977e-05,
+      "loss": 0.0864,
+      "step": 41240
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.6226922647616925e-05,
+      "loss": 0.0663,
+      "step": 41250
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.622357405960487e-05,
+      "loss": 0.0888,
+      "step": 41260
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.6220225471592813e-05,
+      "loss": 0.0933,
+      "step": 41270
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.6216876883580757e-05,
+      "loss": 0.0917,
+      "step": 41280
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.62135282955687e-05,
+      "loss": 0.2214,
+      "step": 41290
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.6210179707556645e-05,
+      "loss": 0.0879,
+      "step": 41300
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.6206831119544593e-05,
+      "loss": 0.1454,
+      "step": 41310
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.620348253153254e-05,
+      "loss": 0.0811,
+      "step": 41320
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.6200133943520484e-05,
+      "loss": 0.1401,
+      "step": 41330
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.6196785355508428e-05,
+      "loss": 0.2068,
+      "step": 41340
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.6193436767496372e-05,
+      "loss": 0.1244,
+      "step": 41350
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.6190088179484316e-05,
+      "loss": 0.1229,
+      "step": 41360
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.618673959147226e-05,
+      "loss": 0.0936,
+      "step": 41370
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.6183391003460208e-05,
+      "loss": 0.1027,
+      "step": 41380
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.6180042415448155e-05,
+      "loss": 0.0988,
+      "step": 41390
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.61766938274361e-05,
+      "loss": 0.0511,
+      "step": 41400
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.6173345239424043e-05,
+      "loss": 0.1319,
+      "step": 41410
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.6169996651411988e-05,
+      "loss": 0.0757,
+      "step": 41420
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.616664806339993e-05,
+      "loss": 0.1482,
+      "step": 41430
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.6163299475387876e-05,
+      "loss": 0.1261,
+      "step": 41440
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.6159950887375827e-05,
+      "loss": 0.1648,
+      "step": 41450
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.615660229936377e-05,
+      "loss": 0.0888,
+      "step": 41460
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.6153253711351715e-05,
+      "loss": 0.1322,
+      "step": 41470
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.614990512333966e-05,
+      "loss": 0.1601,
+      "step": 41480
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.6146556535327603e-05,
+      "loss": 0.155,
+      "step": 41490
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.6143207947315547e-05,
+      "loss": 0.1518,
+      "step": 41500
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.6139859359303494e-05,
+      "loss": 0.1329,
+      "step": 41510
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.6136510771291442e-05,
+      "loss": 0.0989,
+      "step": 41520
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.6133162183279386e-05,
+      "loss": 0.1325,
+      "step": 41530
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.612981359526733e-05,
+      "loss": 0.1199,
+      "step": 41540
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.6126465007255274e-05,
+      "loss": 0.1134,
+      "step": 41550
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.6123116419243218e-05,
+      "loss": 0.1229,
+      "step": 41560
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.6119767831231162e-05,
+      "loss": 0.1248,
+      "step": 41570
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.611641924321911e-05,
+      "loss": 0.297,
+      "step": 41580
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.6113070655207057e-05,
+      "loss": 0.1571,
+      "step": 41590
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.6109722067195e-05,
+      "loss": 0.043,
+      "step": 41600
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.6106373479182945e-05,
+      "loss": 0.0834,
+      "step": 41610
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.610302489117089e-05,
+      "loss": 0.0692,
+      "step": 41620
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.6099676303158833e-05,
+      "loss": 0.1849,
+      "step": 41630
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.6096327715146777e-05,
+      "loss": 0.1223,
+      "step": 41640
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.6092979127134725e-05,
+      "loss": 0.1581,
+      "step": 41650
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.6089630539122672e-05,
+      "loss": 0.1083,
+      "step": 41660
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.6086281951110616e-05,
+      "loss": 0.0934,
+      "step": 41670
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.608293336309856e-05,
+      "loss": 0.1219,
+      "step": 41680
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.6079584775086505e-05,
+      "loss": 0.2312,
+      "step": 41690
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.607623618707445e-05,
+      "loss": 0.1307,
+      "step": 41700
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.6072887599062396e-05,
+      "loss": 0.2023,
+      "step": 41710
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.606953901105034e-05,
+      "loss": 0.112,
+      "step": 41720
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.6066190423038288e-05,
+      "loss": 0.1739,
+      "step": 41730
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.6062841835026232e-05,
+      "loss": 0.188,
+      "step": 41740
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.6059493247014176e-05,
+      "loss": 0.1139,
+      "step": 41750
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.605614465900212e-05,
+      "loss": 0.1399,
+      "step": 41760
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.6052796070990064e-05,
+      "loss": 0.1145,
+      "step": 41770
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.604944748297801e-05,
+      "loss": 0.2547,
+      "step": 41780
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.6046098894965956e-05,
+      "loss": 0.065,
+      "step": 41790
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.6042750306953903e-05,
+      "loss": 0.068,
+      "step": 41800
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.6039401718941847e-05,
+      "loss": 0.0523,
+      "step": 41810
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.603605313092979e-05,
+      "loss": 0.0667,
+      "step": 41820
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.6032704542917735e-05,
+      "loss": 0.1588,
+      "step": 41830
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.6029355954905683e-05,
+      "loss": 0.1114,
+      "step": 41840
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.6026007366893627e-05,
+      "loss": 0.2067,
+      "step": 41850
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.602265877888157e-05,
+      "loss": 0.1095,
+      "step": 41860
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.6019310190869518e-05,
+      "loss": 0.1063,
+      "step": 41870
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.6015961602857462e-05,
+      "loss": 0.1783,
+      "step": 41880
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.6012613014845406e-05,
+      "loss": 0.1105,
+      "step": 41890
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.600926442683335e-05,
+      "loss": 0.0558,
+      "step": 41900
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.6005915838821298e-05,
+      "loss": 0.123,
+      "step": 41910
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.6002567250809242e-05,
+      "loss": 0.1139,
+      "step": 41920
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.599921866279719e-05,
+      "loss": 0.1541,
+      "step": 41930
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.5995870074785134e-05,
+      "loss": 0.0993,
+      "step": 41940
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.5992521486773078e-05,
+      "loss": 0.1165,
+      "step": 41950
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.5989172898761022e-05,
+      "loss": 0.1276,
+      "step": 41960
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.598582431074897e-05,
+      "loss": 0.1252,
+      "step": 41970
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.5982475722736913e-05,
+      "loss": 0.116,
+      "step": 41980
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.5979127134724857e-05,
+      "loss": 0.1054,
+      "step": 41990
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.5975778546712805e-05,
+      "loss": 0.0844,
+      "step": 42000
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.597242995870075e-05,
+      "loss": 0.1278,
+      "step": 42010
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.5969081370688693e-05,
+      "loss": 0.115,
+      "step": 42020
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.5965732782676637e-05,
+      "loss": 0.1337,
+      "step": 42030
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.5962384194664584e-05,
+      "loss": 0.1146,
+      "step": 42040
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.595903560665253e-05,
+      "loss": 0.1058,
+      "step": 42050
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.5955687018640473e-05,
+      "loss": 0.065,
+      "step": 42060
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.595233843062842e-05,
+      "loss": 0.068,
+      "step": 42070
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.5948989842616364e-05,
+      "loss": 0.0754,
+      "step": 42080
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.5945641254604308e-05,
+      "loss": 0.2029,
+      "step": 42090
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.5942292666592256e-05,
+      "loss": 0.1443,
+      "step": 42100
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.59389440785802e-05,
+      "loss": 0.2062,
+      "step": 42110
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.5935595490568144e-05,
+      "loss": 0.1235,
+      "step": 42120
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.5932246902556088e-05,
+      "loss": 0.2118,
+      "step": 42130
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.5928898314544035e-05,
+      "loss": 0.1316,
+      "step": 42140
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.592554972653198e-05,
+      "loss": 0.0669,
+      "step": 42150
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.5922201138519924e-05,
+      "loss": 0.2194,
+      "step": 42160
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.591885255050787e-05,
+      "loss": 0.0434,
+      "step": 42170
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.5915503962495815e-05,
+      "loss": 0.1952,
+      "step": 42180
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.591215537448376e-05,
+      "loss": 0.1321,
+      "step": 42190
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.5908806786471703e-05,
+      "loss": 0.1439,
+      "step": 42200
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.590545819845965e-05,
+      "loss": 0.089,
+      "step": 42210
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.5902109610447595e-05,
+      "loss": 0.1277,
+      "step": 42220
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.5898761022435542e-05,
+      "loss": 0.1268,
+      "step": 42230
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.5895412434423486e-05,
+      "loss": 0.1889,
+      "step": 42240
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.589206384641143e-05,
+      "loss": 0.1256,
+      "step": 42250
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.5888715258399374e-05,
+      "loss": 0.1726,
+      "step": 42260
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.588536667038732e-05,
+      "loss": 0.0411,
+      "step": 42270
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.5882018082375266e-05,
+      "loss": 0.1948,
+      "step": 42280
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.587866949436321e-05,
+      "loss": 0.1156,
+      "step": 42290
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.5875320906351158e-05,
+      "loss": 0.0866,
+      "step": 42300
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.58719723183391e-05,
+      "loss": 0.0846,
+      "step": 42310
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.5868623730327046e-05,
+      "loss": 0.079,
+      "step": 42320
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.586527514231499e-05,
+      "loss": 0.1862,
+      "step": 42330
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.5861926554302934e-05,
+      "loss": 0.1048,
+      "step": 42340
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.585857796629088e-05,
+      "loss": 0.086,
+      "step": 42350
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.585522937827883e-05,
+      "loss": 0.0594,
+      "step": 42360
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.5851880790266773e-05,
+      "loss": 0.2437,
+      "step": 42370
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5848532202254717e-05,
+      "loss": 0.1148,
+      "step": 42380
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.584518361424266e-05,
+      "loss": 0.0721,
+      "step": 42390
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5841835026230605e-05,
+      "loss": 0.0787,
+      "step": 42400
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.583848643821855e-05,
+      "loss": 0.1437,
+      "step": 42410
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5835137850206497e-05,
+      "loss": 0.1122,
+      "step": 42420
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5831789262194444e-05,
+      "loss": 0.1017,
+      "step": 42430
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5828440674182388e-05,
+      "loss": 0.0999,
+      "step": 42440
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5825092086170332e-05,
+      "loss": 0.177,
+      "step": 42450
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5821743498158276e-05,
+      "loss": 0.1381,
+      "step": 42460
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.581839491014622e-05,
+      "loss": 0.1419,
+      "step": 42470
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.5815046322134164e-05,
+      "loss": 0.1025,
+      "step": 42480
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.5811697734122115e-05,
+      "loss": 0.1237,
+      "step": 42490
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.580834914611006e-05,
+      "loss": 0.0591,
+      "step": 42500
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.5805000558098003e-05,
+      "loss": 0.0716,
+      "step": 42510
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.5801651970085947e-05,
+      "loss": 0.057,
+      "step": 42520
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.579830338207389e-05,
+      "loss": 0.0922,
+      "step": 42530
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.5794954794061836e-05,
+      "loss": 0.1041,
+      "step": 42540
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.579160620604978e-05,
+      "loss": 0.1744,
+      "step": 42550
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.578825761803773e-05,
+      "loss": 0.0799,
+      "step": 42560
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.5784909030025675e-05,
+      "loss": 0.1341,
+      "step": 42570
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.578156044201362e-05,
+      "loss": 0.1021,
+      "step": 42580
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.5778211854001563e-05,
+      "loss": 0.0681,
+      "step": 42590
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.5774863265989507e-05,
+      "loss": 0.069,
+      "step": 42600
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.577151467797745e-05,
+      "loss": 0.1555,
+      "step": 42610
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.57681660899654e-05,
+      "loss": 0.0695,
+      "step": 42620
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.5764817501953346e-05,
+      "loss": 0.1127,
+      "step": 42630
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.576146891394129e-05,
+      "loss": 0.1114,
+      "step": 42640
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.5758120325929234e-05,
+      "loss": 0.146,
+      "step": 42650
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.5754771737917178e-05,
+      "loss": 0.1114,
+      "step": 42660
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.5751423149905122e-05,
+      "loss": 0.1472,
+      "step": 42670
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.5748074561893066e-05,
+      "loss": 0.1506,
+      "step": 42680
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.574506083268222e-05,
+      "loss": 0.3615,
+      "step": 42690
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.5741712244670164e-05,
+      "loss": 0.1172,
+      "step": 42700
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.573836365665811e-05,
+      "loss": 0.1518,
+      "step": 42710
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.5735015068646053e-05,
+      "loss": 0.1131,
+      "step": 42720
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.5731666480633997e-05,
+      "loss": 0.2314,
+      "step": 42730
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5728317892621947e-05,
+      "loss": 0.13,
+      "step": 42740
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.572496930460989e-05,
+      "loss": 0.1374,
+      "step": 42750
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5721620716597836e-05,
+      "loss": 0.048,
+      "step": 42760
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.571827212858578e-05,
+      "loss": 0.1341,
+      "step": 42770
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5714923540573724e-05,
+      "loss": 0.1035,
+      "step": 42780
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5711574952561668e-05,
+      "loss": 0.0749,
+      "step": 42790
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5708226364549615e-05,
+      "loss": 0.0493,
+      "step": 42800
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5704877776537563e-05,
+      "loss": 0.074,
+      "step": 42810
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5701529188525507e-05,
+      "loss": 0.1385,
+      "step": 42820
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.569818060051345e-05,
+      "loss": 0.1628,
+      "step": 42830
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.5694832012501395e-05,
+      "loss": 0.1,
+      "step": 42840
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.569148342448934e-05,
+      "loss": 0.0919,
+      "step": 42850
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.5688134836477283e-05,
+      "loss": 0.2165,
+      "step": 42860
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.568478624846523e-05,
+      "loss": 0.315,
+      "step": 42870
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.5681437660453178e-05,
+      "loss": 0.0686,
+      "step": 42880
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.5678089072441122e-05,
+      "loss": 0.0614,
+      "step": 42890
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.5674740484429066e-05,
+      "loss": 0.0746,
+      "step": 42900
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.567139189641701e-05,
+      "loss": 0.1724,
+      "step": 42910
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.5668043308404954e-05,
+      "loss": 0.2103,
+      "step": 42920
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.5664694720392902e-05,
+      "loss": 0.0973,
+      "step": 42930
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.5661346132380846e-05,
+      "loss": 0.1838,
+      "step": 42940
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.5657997544368793e-05,
+      "loss": 0.0727,
+      "step": 42950
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.5654648956356737e-05,
+      "loss": 0.2311,
+      "step": 42960
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.565130036834468e-05,
+      "loss": 0.063,
+      "step": 42970
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.5647951780332626e-05,
+      "loss": 0.1493,
+      "step": 42980
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.564460319232057e-05,
+      "loss": 0.0724,
+      "step": 42990
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.5641254604308517e-05,
+      "loss": 0.1302,
+      "step": 43000
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.563790601629646e-05,
+      "loss": 0.0708,
+      "step": 43010
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.563455742828441e-05,
+      "loss": 0.1263,
+      "step": 43020
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.5631208840272353e-05,
+      "loss": 0.1328,
+      "step": 43030
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.5627860252260297e-05,
+      "loss": 0.0631,
+      "step": 43040
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.562451166424824e-05,
+      "loss": 0.0834,
+      "step": 43050
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.5621163076236188e-05,
+      "loss": 0.1476,
+      "step": 43060
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.5617814488224132e-05,
+      "loss": 0.1507,
+      "step": 43070
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.5614465900212076e-05,
+      "loss": 0.1678,
+      "step": 43080
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.5611117312200024e-05,
+      "loss": 0.0983,
+      "step": 43090
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.5607768724187968e-05,
+      "loss": 0.1166,
+      "step": 43100
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.5604420136175912e-05,
+      "loss": 0.0673,
+      "step": 43110
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.5601071548163856e-05,
+      "loss": 0.1247,
+      "step": 43120
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.5597722960151804e-05,
+      "loss": 0.2116,
+      "step": 43130
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.5594374372139748e-05,
+      "loss": 0.0834,
+      "step": 43140
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.5591025784127695e-05,
+      "loss": 0.2353,
+      "step": 43150
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.558767719611564e-05,
+      "loss": 0.0763,
+      "step": 43160
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.5584328608103583e-05,
+      "loss": 0.1262,
+      "step": 43170
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.5580980020091527e-05,
+      "loss": 0.2058,
+      "step": 43180
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.5577631432079475e-05,
+      "loss": 0.1472,
+      "step": 43190
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.557428284406742e-05,
+      "loss": 0.0925,
+      "step": 43200
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.5570934256055363e-05,
+      "loss": 0.1051,
+      "step": 43210
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.556758566804331e-05,
+      "loss": 0.0757,
+      "step": 43220
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.5564237080031254e-05,
+      "loss": 0.1257,
+      "step": 43230
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.55608884920192e-05,
+      "loss": 0.0902,
+      "step": 43240
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.5557539904007143e-05,
+      "loss": 0.0906,
+      "step": 43250
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.555419131599509e-05,
+      "loss": 0.1028,
+      "step": 43260
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.5550842727983034e-05,
+      "loss": 0.0728,
+      "step": 43270
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.5547494139970978e-05,
+      "loss": 0.2298,
+      "step": 43280
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.5544145551958926e-05,
+      "loss": 0.1042,
+      "step": 43290
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.554079696394687e-05,
+      "loss": 0.0656,
+      "step": 43300
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.5537448375934814e-05,
+      "loss": 0.2347,
+      "step": 43310
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.553409978792276e-05,
+      "loss": 0.065,
+      "step": 43320
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.5530751199910705e-05,
+      "loss": 0.3419,
+      "step": 43330
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.552740261189865e-05,
+      "loss": 0.0941,
+      "step": 43340
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.5524054023886594e-05,
+      "loss": 0.2189,
+      "step": 43350
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.552070543587454e-05,
+      "loss": 0.0968,
+      "step": 43360
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.5517356847862485e-05,
+      "loss": 0.1676,
+      "step": 43370
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.551400825985043e-05,
+      "loss": 0.1315,
+      "step": 43380
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.5510659671838377e-05,
+      "loss": 0.0772,
+      "step": 43390
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.550731108382632e-05,
+      "loss": 0.148,
+      "step": 43400
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.5503962495814265e-05,
+      "loss": 0.1653,
+      "step": 43410
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.550061390780221e-05,
+      "loss": 0.1541,
+      "step": 43420
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.5497265319790156e-05,
+      "loss": 0.1886,
+      "step": 43430
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.54939167317781e-05,
+      "loss": 0.1411,
+      "step": 43440
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.5490568143766048e-05,
+      "loss": 0.0548,
+      "step": 43450
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.5487219555753992e-05,
+      "loss": 0.1323,
+      "step": 43460
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.5483870967741936e-05,
+      "loss": 0.1125,
+      "step": 43470
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.548052237972988e-05,
+      "loss": 0.144,
+      "step": 43480
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.5477173791717824e-05,
+      "loss": 0.0995,
+      "step": 43490
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.547382520370577e-05,
+      "loss": 0.0644,
+      "step": 43500
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.5470476615693716e-05,
+      "loss": 0.1391,
+      "step": 43510
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.5467128027681663e-05,
+      "loss": 0.0621,
+      "step": 43520
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.5463779439669607e-05,
+      "loss": 0.2769,
+      "step": 43530
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.546043085165755e-05,
+      "loss": 0.1299,
+      "step": 43540
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.5457082263645495e-05,
+      "loss": 0.1127,
+      "step": 43550
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.545373367563344e-05,
+      "loss": 0.0863,
+      "step": 43560
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.5450385087621387e-05,
+      "loss": 0.0923,
+      "step": 43570
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.5447036499609334e-05,
+      "loss": 0.1579,
+      "step": 43580
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.544368791159728e-05,
+      "loss": 0.1004,
+      "step": 43590
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.5440339323585223e-05,
+      "loss": 0.1204,
+      "step": 43600
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.5436990735573167e-05,
+      "loss": 0.1383,
+      "step": 43610
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.543364214756111e-05,
+      "loss": 0.1457,
+      "step": 43620
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.5430293559549055e-05,
+      "loss": 0.1201,
+      "step": 43630
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.5426944971537002e-05,
+      "loss": 0.0849,
+      "step": 43640
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.542359638352495e-05,
+      "loss": 0.0469,
+      "step": 43650
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.5420247795512894e-05,
+      "loss": 0.0711,
+      "step": 43660
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.5416899207500838e-05,
+      "loss": 0.0466,
+      "step": 43670
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.5413550619488782e-05,
+      "loss": 0.1841,
+      "step": 43680
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.5410202031476726e-05,
+      "loss": 0.2027,
+      "step": 43690
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.540685344346467e-05,
+      "loss": 0.1507,
+      "step": 43700
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.540350485545262e-05,
+      "loss": 0.0845,
+      "step": 43710
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.5400156267440565e-05,
+      "loss": 0.0837,
+      "step": 43720
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.539680767942851e-05,
+      "loss": 0.2056,
+      "step": 43730
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.5393459091416453e-05,
+      "loss": 0.1751,
+      "step": 43740
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.5390110503404397e-05,
+      "loss": 0.0651,
+      "step": 43750
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.538676191539234e-05,
+      "loss": 0.0839,
+      "step": 43760
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.5383413327380285e-05,
+      "loss": 0.1836,
+      "step": 43770
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.5380064739368236e-05,
+      "loss": 0.0862,
+      "step": 43780
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.537671615135618e-05,
+      "loss": 0.1716,
+      "step": 43790
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.5373367563344124e-05,
+      "loss": 0.2688,
+      "step": 43800
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.537001897533207e-05,
+      "loss": 0.0968,
+      "step": 43810
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.5366670387320012e-05,
+      "loss": 0.084,
+      "step": 43820
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.5363656658109167e-05,
+      "loss": 0.1685,
+      "step": 43830
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.536030807009711e-05,
+      "loss": 0.1304,
+      "step": 43840
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.5356959482085055e-05,
+      "loss": 0.2459,
+      "step": 43850
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.5353610894073e-05,
+      "loss": 0.1021,
+      "step": 43860
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.5350262306060943e-05,
+      "loss": 0.0673,
+      "step": 43870
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.5346913718048887e-05,
+      "loss": 0.1242,
+      "step": 43880
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.5343565130036838e-05,
+      "loss": 0.1352,
+      "step": 43890
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.5340216542024782e-05,
+      "loss": 0.1277,
+      "step": 43900
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.5336867954012726e-05,
+      "loss": 0.0637,
+      "step": 43910
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.533351936600067e-05,
+      "loss": 0.1256,
+      "step": 43920
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.5330170777988614e-05,
+      "loss": 0.2435,
+      "step": 43930
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.5326822189976558e-05,
+      "loss": 0.1486,
+      "step": 43940
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.5323473601964502e-05,
+      "loss": 0.1647,
+      "step": 43950
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.5320125013952453e-05,
+      "loss": 0.0788,
+      "step": 43960
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.5316776425940397e-05,
+      "loss": 0.0956,
+      "step": 43970
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.531342783792834e-05,
+      "loss": 0.0956,
+      "step": 43980
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.5310079249916285e-05,
+      "loss": 0.1367,
+      "step": 43990
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.530673066190423e-05,
+      "loss": 0.0953,
+      "step": 44000
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5303382073892173e-05,
+      "loss": 0.0872,
+      "step": 44010
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.530003348588012e-05,
+      "loss": 0.1204,
+      "step": 44020
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.529668489786807e-05,
+      "loss": 0.1983,
+      "step": 44030
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5293336309856012e-05,
+      "loss": 0.086,
+      "step": 44040
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5289987721843956e-05,
+      "loss": 0.1287,
+      "step": 44050
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.52866391338319e-05,
+      "loss": 0.0715,
+      "step": 44060
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5283290545819845e-05,
+      "loss": 0.0845,
+      "step": 44070
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.527994195780779e-05,
+      "loss": 0.16,
+      "step": 44080
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.5276593369795736e-05,
+      "loss": 0.1088,
+      "step": 44090
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.5273244781783684e-05,
+      "loss": 0.1412,
+      "step": 44100
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.5269896193771628e-05,
+      "loss": 0.0455,
+      "step": 44110
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.5266547605759572e-05,
+      "loss": 0.095,
+      "step": 44120
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.5263199017747516e-05,
+      "loss": 0.1762,
+      "step": 44130
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.525985042973546e-05,
+      "loss": 0.1694,
+      "step": 44140
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.5256501841723406e-05,
+      "loss": 0.0918,
+      "step": 44150
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.525315325371135e-05,
+      "loss": 0.1249,
+      "step": 44160
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.5249804665699299e-05,
+      "loss": 0.0986,
+      "step": 44170
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.5246456077687243e-05,
+      "loss": 0.2919,
+      "step": 44180
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.5243107489675187e-05,
+      "loss": 0.1575,
+      "step": 44190
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.5239758901663133e-05,
+      "loss": 0.0863,
+      "step": 44200
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.5236410313651077e-05,
+      "loss": 0.1226,
+      "step": 44210
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.5233061725639021e-05,
+      "loss": 0.1385,
+      "step": 44220
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.5229713137626967e-05,
+      "loss": 0.1092,
+      "step": 44230
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.5226364549614914e-05,
+      "loss": 0.1126,
+      "step": 44240
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.5223015961602858e-05,
+      "loss": 0.0946,
+      "step": 44250
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.5219667373590802e-05,
+      "loss": 0.0501,
+      "step": 44260
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.5216318785578748e-05,
+      "loss": 0.0976,
+      "step": 44270
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.5212970197566692e-05,
+      "loss": 0.1318,
+      "step": 44280
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.5209621609554636e-05,
+      "loss": 0.0431,
+      "step": 44290
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.5206273021542585e-05,
+      "loss": 0.1314,
+      "step": 44300
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.520292443353053e-05,
+      "loss": 0.0675,
+      "step": 44310
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.5199575845518474e-05,
+      "loss": 0.0733,
+      "step": 44320
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.519622725750642e-05,
+      "loss": 0.1856,
+      "step": 44330
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.5192878669494363e-05,
+      "loss": 0.0713,
+      "step": 44340
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.5189530081482308e-05,
+      "loss": 0.1195,
+      "step": 44350
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5186181493470253e-05,
+      "loss": 0.1113,
+      "step": 44360
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.51828329054582e-05,
+      "loss": 0.4631,
+      "step": 44370
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5179484317446145e-05,
+      "loss": 0.2297,
+      "step": 44380
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5176135729434089e-05,
+      "loss": 0.0557,
+      "step": 44390
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5172787141422035e-05,
+      "loss": 0.1444,
+      "step": 44400
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5169438553409979e-05,
+      "loss": 0.1085,
+      "step": 44410
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5166089965397923e-05,
+      "loss": 0.0863,
+      "step": 44420
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5162741377385869e-05,
+      "loss": 0.1325,
+      "step": 44430
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5159392789373816e-05,
+      "loss": 0.1724,
+      "step": 44440
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.515604420136176e-05,
+      "loss": 0.1347,
+      "step": 44450
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5152695613349706e-05,
+      "loss": 0.1208,
+      "step": 44460
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.514934702533765e-05,
+      "loss": 0.1208,
+      "step": 44470
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5145998437325594e-05,
+      "loss": 0.1508,
+      "step": 44480
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.514264984931354e-05,
+      "loss": 0.1421,
+      "step": 44490
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5139301261301484e-05,
+      "loss": 0.1244,
+      "step": 44500
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5135952673289431e-05,
+      "loss": 0.146,
+      "step": 44510
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5132604085277375e-05,
+      "loss": 0.1161,
+      "step": 44520
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5129255497265321e-05,
+      "loss": 0.1844,
+      "step": 44530
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5125906909253265e-05,
+      "loss": 0.135,
+      "step": 44540
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.512255832124121e-05,
+      "loss": 0.1055,
+      "step": 44550
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5119209733229155e-05,
+      "loss": 0.1499,
+      "step": 44560
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.51158611452171e-05,
+      "loss": 0.0581,
+      "step": 44570
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5112512557205047e-05,
+      "loss": 0.0625,
+      "step": 44580
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5109163969192992e-05,
+      "loss": 0.08,
+      "step": 44590
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5105815381180936e-05,
+      "loss": 0.0859,
+      "step": 44600
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.510246679316888e-05,
+      "loss": 0.201,
+      "step": 44610
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5099118205156826e-05,
+      "loss": 0.0746,
+      "step": 44620
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.509576961714477e-05,
+      "loss": 0.0793,
+      "step": 44630
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.5092421029132714e-05,
+      "loss": 0.1239,
+      "step": 44640
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.5089072441120662e-05,
+      "loss": 0.0693,
+      "step": 44650
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.5085723853108608e-05,
+      "loss": 0.0888,
+      "step": 44660
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.5082375265096552e-05,
+      "loss": 0.1158,
+      "step": 44670
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.5079026677084496e-05,
+      "loss": 0.1478,
+      "step": 44680
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.5075678089072442e-05,
+      "loss": 0.0968,
+      "step": 44690
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.5072329501060386e-05,
+      "loss": 0.1384,
+      "step": 44700
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.506898091304833e-05,
+      "loss": 0.0723,
+      "step": 44710
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.5065632325036279e-05,
+      "loss": 0.1695,
+      "step": 44720
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.5062283737024223e-05,
+      "loss": 0.0766,
+      "step": 44730
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.5058935149012167e-05,
+      "loss": 0.0813,
+      "step": 44740
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.5055586561000113e-05,
+      "loss": 0.1149,
+      "step": 44750
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.5052237972988057e-05,
+      "loss": 0.1157,
+      "step": 44760
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.5048889384976001e-05,
+      "loss": 0.1431,
+      "step": 44770
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.5045540796963945e-05,
+      "loss": 0.124,
+      "step": 44780
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.5042192208951894e-05,
+      "loss": 0.2089,
+      "step": 44790
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.5038843620939838e-05,
+      "loss": 0.0878,
+      "step": 44800
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.5035495032927782e-05,
+      "loss": 0.0528,
+      "step": 44810
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.5032146444915728e-05,
+      "loss": 0.1279,
+      "step": 44820
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.5028797856903672e-05,
+      "loss": 0.0706,
+      "step": 44830
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.5025449268891616e-05,
+      "loss": 0.0696,
+      "step": 44840
+    },
+    {
+      "epoch": 5.0,
+      "eval_loss": 0.27031269669532776,
+      "eval_runtime": 250.5688,
+      "eval_samples_per_second": 7.958,
+      "eval_steps_per_second": 3.979,
+      "eval_wer": 0.11049002035422426,
+      "step": 44845
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.5022100680879562e-05,
+      "loss": 0.1633,
+      "step": 44850
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.501875209286751e-05,
+      "loss": 0.1198,
+      "step": 44860
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.5015403504855454e-05,
+      "loss": 0.0816,
+      "step": 44870
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.5012054916843398e-05,
+      "loss": 0.1644,
+      "step": 44880
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.5008706328831343e-05,
+      "loss": 0.0893,
+      "step": 44890
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.5005357740819287e-05,
+      "loss": 0.1992,
+      "step": 44900
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.5002009152807232e-05,
+      "loss": 0.091,
+      "step": 44910
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4998660564795179e-05,
+      "loss": 0.0909,
+      "step": 44920
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4995311976783123e-05,
+      "loss": 0.1453,
+      "step": 44930
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4991963388771067e-05,
+      "loss": 0.0828,
+      "step": 44940
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4988614800759015e-05,
+      "loss": 0.0495,
+      "step": 44950
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4985266212746959e-05,
+      "loss": 0.0893,
+      "step": 44960
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4981917624734903e-05,
+      "loss": 0.0518,
+      "step": 44970
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.497856903672285e-05,
+      "loss": 0.0812,
+      "step": 44980
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4975220448710794e-05,
+      "loss": 0.0724,
+      "step": 44990
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4971871860698738e-05,
+      "loss": 0.0627,
+      "step": 45000
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4968523272686684e-05,
+      "loss": 0.067,
+      "step": 45010
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.496517468467463e-05,
+      "loss": 0.2007,
+      "step": 45020
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4961826096662574e-05,
+      "loss": 0.036,
+      "step": 45030
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4958477508650518e-05,
+      "loss": 0.0554,
+      "step": 45040
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.4955128920638466e-05,
+      "loss": 0.2509,
+      "step": 45050
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.495178033262641e-05,
+      "loss": 0.0574,
+      "step": 45060
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4948431744614354e-05,
+      "loss": 0.2887,
+      "step": 45070
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4945083156602301e-05,
+      "loss": 0.0677,
+      "step": 45080
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4941734568590245e-05,
+      "loss": 0.087,
+      "step": 45090
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.493838598057819e-05,
+      "loss": 0.0731,
+      "step": 45100
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4935037392566135e-05,
+      "loss": 0.0755,
+      "step": 45110
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.493168880455408e-05,
+      "loss": 0.0931,
+      "step": 45120
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4928340216542025e-05,
+      "loss": 0.081,
+      "step": 45130
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4924991628529969e-05,
+      "loss": 0.0465,
+      "step": 45140
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.4921643040517916e-05,
+      "loss": 0.1622,
+      "step": 45150
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.491829445250586e-05,
+      "loss": 0.0713,
+      "step": 45160
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.4914945864493805e-05,
+      "loss": 0.1602,
+      "step": 45170
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.491159727648175e-05,
+      "loss": 0.0966,
+      "step": 45180
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.4908248688469696e-05,
+      "loss": 0.1414,
+      "step": 45190
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.490490010045764e-05,
+      "loss": 0.0842,
+      "step": 45200
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.4901551512445586e-05,
+      "loss": 0.076,
+      "step": 45210
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.4898202924433532e-05,
+      "loss": 0.1117,
+      "step": 45220
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.4894854336421476e-05,
+      "loss": 0.0663,
+      "step": 45230
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.4891505748409422e-05,
+      "loss": 0.1113,
+      "step": 45240
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.4888157160397366e-05,
+      "loss": 0.1077,
+      "step": 45250
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.4884808572385311e-05,
+      "loss": 0.0591,
+      "step": 45260
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.4881459984373256e-05,
+      "loss": 0.5859,
+      "step": 45270
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.4878111396361201e-05,
+      "loss": 0.0476,
+      "step": 45280
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.4874762808349147e-05,
+      "loss": 0.092,
+      "step": 45290
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.4871414220337091e-05,
+      "loss": 0.091,
+      "step": 45300
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.4868065632325037e-05,
+      "loss": 0.2228,
+      "step": 45310
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.4864717044312981e-05,
+      "loss": 0.0727,
+      "step": 45320
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.4861368456300927e-05,
+      "loss": 0.1429,
+      "step": 45330
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.4858019868288872e-05,
+      "loss": 0.1982,
+      "step": 45340
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.4854671280276817e-05,
+      "loss": 0.1485,
+      "step": 45350
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.4851322692264762e-05,
+      "loss": 0.1831,
+      "step": 45360
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.4847974104252708e-05,
+      "loss": 0.1199,
+      "step": 45370
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.4844625516240652e-05,
+      "loss": 0.0837,
+      "step": 45380
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.4841276928228596e-05,
+      "loss": 0.0969,
+      "step": 45390
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.4837928340216542e-05,
+      "loss": 0.1148,
+      "step": 45400
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.4834579752204488e-05,
+      "loss": 0.1243,
+      "step": 45410
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.4831231164192432e-05,
+      "loss": 0.0589,
+      "step": 45420
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.4827882576180378e-05,
+      "loss": 0.0492,
+      "step": 45430
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.4824533988168323e-05,
+      "loss": 0.1326,
+      "step": 45440
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.4821185400156267e-05,
+      "loss": 0.1026,
+      "step": 45450
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.4817836812144213e-05,
+      "loss": 0.0656,
+      "step": 45460
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.4814488224132159e-05,
+      "loss": 0.0876,
+      "step": 45470
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.4811139636120103e-05,
+      "loss": 0.1001,
+      "step": 45480
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.4807791048108047e-05,
+      "loss": 0.0515,
+      "step": 45490
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.4804442460095995e-05,
+      "loss": 0.1039,
+      "step": 45500
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.4801093872083939e-05,
+      "loss": 0.1068,
+      "step": 45510
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.4797745284071883e-05,
+      "loss": 0.1516,
+      "step": 45520
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.4794396696059829e-05,
+      "loss": 0.0606,
+      "step": 45530
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.4791048108047774e-05,
+      "loss": 0.0753,
+      "step": 45540
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.4787699520035718e-05,
+      "loss": 0.063,
+      "step": 45550
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.4784350932023662e-05,
+      "loss": 0.3339,
+      "step": 45560
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.478100234401161e-05,
+      "loss": 0.1017,
+      "step": 45570
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.4777653755999554e-05,
+      "loss": 0.1087,
+      "step": 45580
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.4774305167987498e-05,
+      "loss": 0.0745,
+      "step": 45590
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.4770956579975446e-05,
+      "loss": 0.0711,
+      "step": 45600
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.476760799196339e-05,
+      "loss": 0.1083,
+      "step": 45610
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.4764259403951334e-05,
+      "loss": 0.0727,
+      "step": 45620
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.476091081593928e-05,
+      "loss": 0.0476,
+      "step": 45630
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.4757562227927225e-05,
+      "loss": 0.0847,
+      "step": 45640
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.475421363991517e-05,
+      "loss": 0.1263,
+      "step": 45650
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.4750865051903113e-05,
+      "loss": 0.0407,
+      "step": 45660
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.474751646389106e-05,
+      "loss": 0.0986,
+      "step": 45670
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.4744167875879005e-05,
+      "loss": 0.0949,
+      "step": 45680
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.4740819287866949e-05,
+      "loss": 0.1476,
+      "step": 45690
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.4737470699854895e-05,
+      "loss": 0.0922,
+      "step": 45700
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.473412211184284e-05,
+      "loss": 0.1247,
+      "step": 45710
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.4730773523830785e-05,
+      "loss": 0.0613,
+      "step": 45720
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.472742493581873e-05,
+      "loss": 0.0633,
+      "step": 45730
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.4724076347806676e-05,
+      "loss": 0.0953,
+      "step": 45740
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.472072775979462e-05,
+      "loss": 0.0654,
+      "step": 45750
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.4717379171782564e-05,
+      "loss": 0.0463,
+      "step": 45760
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.471403058377051e-05,
+      "loss": 0.1236,
+      "step": 45770
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.4710681995758456e-05,
+      "loss": 0.0747,
+      "step": 45780
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.47073334077464e-05,
+      "loss": 0.1745,
+      "step": 45790
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.4703984819734346e-05,
+      "loss": 0.1552,
+      "step": 45800
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.4700636231722291e-05,
+      "loss": 0.1263,
+      "step": 45810
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.4697287643710235e-05,
+      "loss": 0.1444,
+      "step": 45820
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.4693939055698181e-05,
+      "loss": 0.087,
+      "step": 45830
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.4690590467686125e-05,
+      "loss": 0.1164,
+      "step": 45840
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.4687241879674071e-05,
+      "loss": 0.1227,
+      "step": 45850
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.4683893291662017e-05,
+      "loss": 0.0987,
+      "step": 45860
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.4680544703649961e-05,
+      "loss": 0.0713,
+      "step": 45870
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.4677196115637907e-05,
+      "loss": 0.0412,
+      "step": 45880
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.467384752762585e-05,
+      "loss": 0.0356,
+      "step": 45890
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.4670498939613797e-05,
+      "loss": 0.1602,
+      "step": 45900
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.4667150351601742e-05,
+      "loss": 0.1379,
+      "step": 45910
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.4663801763589686e-05,
+      "loss": 0.0495,
+      "step": 45920
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.4660453175577632e-05,
+      "loss": 0.0912,
+      "step": 45930
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.4657104587565576e-05,
+      "loss": 0.0754,
+      "step": 45940
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.4653755999553522e-05,
+      "loss": 0.1254,
+      "step": 45950
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.4650407411541468e-05,
+      "loss": 0.0613,
+      "step": 45960
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.4647058823529412e-05,
+      "loss": 0.0561,
+      "step": 45970
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.4643710235517358e-05,
+      "loss": 0.0828,
+      "step": 45980
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.4640361647505303e-05,
+      "loss": 0.0797,
+      "step": 45990
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.4637013059493247e-05,
+      "loss": 0.0856,
+      "step": 46000
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.4633664471481192e-05,
+      "loss": 0.0577,
+      "step": 46010
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.4630315883469137e-05,
+      "loss": 0.1,
+      "step": 46020
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.4626967295457083e-05,
+      "loss": 0.1343,
+      "step": 46030
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.4623618707445027e-05,
+      "loss": 0.0823,
+      "step": 46040
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.4620270119432973e-05,
+      "loss": 0.071,
+      "step": 46050
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.4616921531420919e-05,
+      "loss": 0.124,
+      "step": 46060
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.4613572943408863e-05,
+      "loss": 0.0894,
+      "step": 46070
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.4610224355396807e-05,
+      "loss": 0.1966,
+      "step": 46080
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.4606875767384754e-05,
+      "loss": 0.1152,
+      "step": 46090
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.4603527179372698e-05,
+      "loss": 0.0409,
+      "step": 46100
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.4600178591360642e-05,
+      "loss": 0.0796,
+      "step": 46110
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.459683000334859e-05,
+      "loss": 0.106,
+      "step": 46120
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.4593481415336534e-05,
+      "loss": 0.069,
+      "step": 46130
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.4590132827324478e-05,
+      "loss": 0.1398,
+      "step": 46140
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.4586784239312422e-05,
+      "loss": 0.0901,
+      "step": 46150
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.458343565130037e-05,
+      "loss": 0.1097,
+      "step": 46160
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.4580087063288314e-05,
+      "loss": 0.1327,
+      "step": 46170
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.4576738475276258e-05,
+      "loss": 0.1609,
+      "step": 46180
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.4573389887264205e-05,
+      "loss": 0.0718,
+      "step": 46190
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.457004129925215e-05,
+      "loss": 0.0903,
+      "step": 46200
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.4566692711240093e-05,
+      "loss": 0.0861,
+      "step": 46210
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.4563344123228039e-05,
+      "loss": 0.0548,
+      "step": 46220
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.4559995535215985e-05,
+      "loss": 0.0787,
+      "step": 46230
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.4556646947203929e-05,
+      "loss": 0.0593,
+      "step": 46240
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.4553298359191875e-05,
+      "loss": 0.078,
+      "step": 46250
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.454994977117982e-05,
+      "loss": 0.0413,
+      "step": 46260
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.4546601183167765e-05,
+      "loss": 0.0999,
+      "step": 46270
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.4543252595155709e-05,
+      "loss": 0.1274,
+      "step": 46280
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.4539904007143654e-05,
+      "loss": 0.0231,
+      "step": 46290
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.45365554191316e-05,
+      "loss": 0.0573,
+      "step": 46300
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.4533206831119544e-05,
+      "loss": 0.0892,
+      "step": 46310
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.452985824310749e-05,
+      "loss": 0.1469,
+      "step": 46320
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.4526509655095436e-05,
+      "loss": 0.1104,
+      "step": 46330
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.452316106708338e-05,
+      "loss": 0.1619,
+      "step": 46340
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.4519812479071326e-05,
+      "loss": 0.264,
+      "step": 46350
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.4516463891059271e-05,
+      "loss": 0.056,
+      "step": 46360
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.4513115303047215e-05,
+      "loss": 0.0789,
+      "step": 46370
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.4509766715035161e-05,
+      "loss": 0.0999,
+      "step": 46380
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.4506418127023105e-05,
+      "loss": 0.0563,
+      "step": 46390
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.4503069539011051e-05,
+      "loss": 0.0994,
+      "step": 46400
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.4499720950998995e-05,
+      "loss": 0.1035,
+      "step": 46410
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.4496372362986941e-05,
+      "loss": 0.0421,
+      "step": 46420
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.4493023774974887e-05,
+      "loss": 0.1937,
+      "step": 46430
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.448967518696283e-05,
+      "loss": 0.0878,
+      "step": 46440
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.4486326598950777e-05,
+      "loss": 0.1405,
+      "step": 46450
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.448297801093872e-05,
+      "loss": 0.0753,
+      "step": 46460
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.4479629422926666e-05,
+      "loss": 0.0802,
+      "step": 46470
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.4476280834914612e-05,
+      "loss": 0.0447,
+      "step": 46480
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.4472932246902556e-05,
+      "loss": 0.0991,
+      "step": 46490
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.4469583658890502e-05,
+      "loss": 0.1054,
+      "step": 46500
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.4466235070878446e-05,
+      "loss": 0.076,
+      "step": 46510
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.4462886482866392e-05,
+      "loss": 0.0975,
+      "step": 46520
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.4459537894854336e-05,
+      "loss": 0.0979,
+      "step": 46530
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.4456189306842282e-05,
+      "loss": 0.1037,
+      "step": 46540
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.4452840718830227e-05,
+      "loss": 0.1474,
+      "step": 46550
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.4449492130818171e-05,
+      "loss": 0.1164,
+      "step": 46560
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.4446143542806117e-05,
+      "loss": 0.0703,
+      "step": 46570
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.4442794954794063e-05,
+      "loss": 0.1272,
+      "step": 46580
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.4439446366782007e-05,
+      "loss": 0.1207,
+      "step": 46590
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1.4436097778769951e-05,
+      "loss": 0.1055,
+      "step": 46600
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1.4432749190757899e-05,
+      "loss": 0.4005,
+      "step": 46610
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1.4429400602745843e-05,
+      "loss": 0.1021,
+      "step": 46620
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1.4426052014733787e-05,
+      "loss": 0.0864,
+      "step": 46630
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1.4422703426721733e-05,
+      "loss": 0.1585,
+      "step": 46640
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1.4419354838709678e-05,
+      "loss": 0.159,
+      "step": 46650
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1.4416006250697622e-05,
+      "loss": 0.0919,
+      "step": 46660
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1.4412657662685566e-05,
+      "loss": 0.0655,
+      "step": 46670
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1.4409309074673514e-05,
+      "loss": 0.0587,
+      "step": 46680
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.4405960486661458e-05,
+      "loss": 0.0904,
+      "step": 46690
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.4402611898649402e-05,
+      "loss": 0.1415,
+      "step": 46700
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.439926331063735e-05,
+      "loss": 0.0702,
+      "step": 46710
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.4395914722625294e-05,
+      "loss": 0.0699,
+      "step": 46720
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.4392566134613238e-05,
+      "loss": 0.1182,
+      "step": 46730
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.4389217546601183e-05,
+      "loss": 0.0953,
+      "step": 46740
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.438586895858913e-05,
+      "loss": 0.1296,
+      "step": 46750
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.4382520370577073e-05,
+      "loss": 0.0854,
+      "step": 46760
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.4379171782565017e-05,
+      "loss": 0.1065,
+      "step": 46770
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1.4375823194552965e-05,
+      "loss": 0.1101,
+      "step": 46780
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1.4372474606540909e-05,
+      "loss": 0.1133,
+      "step": 46790
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1.4369126018528853e-05,
+      "loss": 0.1338,
+      "step": 46800
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1.43657774305168e-05,
+      "loss": 0.055,
+      "step": 46810
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1.4362428842504745e-05,
+      "loss": 0.1488,
+      "step": 46820
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1.4359080254492689e-05,
+      "loss": 0.1053,
+      "step": 46830
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1.4355731666480634e-05,
+      "loss": 0.0977,
+      "step": 46840
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1.435238307846858e-05,
+      "loss": 0.1329,
+      "step": 46850
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1.4349034490456524e-05,
+      "loss": 0.0935,
+      "step": 46860
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.434568590244447e-05,
+      "loss": 0.0845,
+      "step": 46870
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.4342337314432416e-05,
+      "loss": 0.1062,
+      "step": 46880
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.433898872642036e-05,
+      "loss": 0.0773,
+      "step": 46890
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.4335640138408304e-05,
+      "loss": 0.2249,
+      "step": 46900
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.433229155039625e-05,
+      "loss": 0.0949,
+      "step": 46910
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.4328942962384195e-05,
+      "loss": 0.1107,
+      "step": 46920
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.432559437437214e-05,
+      "loss": 0.1002,
+      "step": 46930
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.4322245786360085e-05,
+      "loss": 0.0763,
+      "step": 46940
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.4318897198348031e-05,
+      "loss": 0.2955,
+      "step": 46950
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.4315548610335975e-05,
+      "loss": 0.0887,
+      "step": 46960
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.4312200022323921e-05,
+      "loss": 0.1043,
+      "step": 46970
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.4308851434311865e-05,
+      "loss": 0.118,
+      "step": 46980
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.430550284629981e-05,
+      "loss": 0.2698,
+      "step": 46990
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.4302154258287756e-05,
+      "loss": 0.1135,
+      "step": 47000
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.42988056702757e-05,
+      "loss": 0.0902,
+      "step": 47010
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.4295457082263646e-05,
+      "loss": 0.0423,
+      "step": 47020
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.429210849425159e-05,
+      "loss": 0.1026,
+      "step": 47030
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.4288759906239536e-05,
+      "loss": 0.1947,
+      "step": 47040
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.428541131822748e-05,
+      "loss": 0.0854,
+      "step": 47050
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.4282062730215426e-05,
+      "loss": 0.0472,
+      "step": 47060
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.4278714142203372e-05,
+      "loss": 0.0808,
+      "step": 47070
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.4275365554191316e-05,
+      "loss": 0.1066,
+      "step": 47080
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.4272016966179262e-05,
+      "loss": 0.0678,
+      "step": 47090
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.4268668378167207e-05,
+      "loss": 0.1484,
+      "step": 47100
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.4265319790155151e-05,
+      "loss": 0.1459,
+      "step": 47110
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.4261971202143096e-05,
+      "loss": 0.0646,
+      "step": 47120
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.4258622614131043e-05,
+      "loss": 0.0855,
+      "step": 47130
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.4255274026118987e-05,
+      "loss": 0.157,
+      "step": 47140
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.4251925438106931e-05,
+      "loss": 0.0888,
+      "step": 47150
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.4248576850094877e-05,
+      "loss": 0.1208,
+      "step": 47160
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.4245228262082823e-05,
+      "loss": 0.1107,
+      "step": 47170
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.4241879674070767e-05,
+      "loss": 0.0499,
+      "step": 47180
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.4238531086058711e-05,
+      "loss": 0.0791,
+      "step": 47190
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.4235182498046658e-05,
+      "loss": 0.1054,
+      "step": 47200
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.4231833910034602e-05,
+      "loss": 0.0897,
+      "step": 47210
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.4228485322022546e-05,
+      "loss": 0.122,
+      "step": 47220
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.4225136734010494e-05,
+      "loss": 0.0551,
+      "step": 47230
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.4221788145998438e-05,
+      "loss": 0.1165,
+      "step": 47240
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.4218439557986382e-05,
+      "loss": 0.1293,
+      "step": 47250
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.4215090969974328e-05,
+      "loss": 0.0871,
+      "step": 47260
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.4211742381962274e-05,
+      "loss": 0.0772,
+      "step": 47270
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.4208393793950218e-05,
+      "loss": 0.1299,
+      "step": 47280
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.4205045205938162e-05,
+      "loss": 0.0715,
+      "step": 47290
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.420169661792611e-05,
+      "loss": 0.2372,
+      "step": 47300
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.4198348029914053e-05,
+      "loss": 0.1314,
+      "step": 47310
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.4194999441901997e-05,
+      "loss": 0.0741,
+      "step": 47320
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.4191650853889945e-05,
+      "loss": 0.0774,
+      "step": 47330
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.4188302265877889e-05,
+      "loss": 0.2174,
+      "step": 47340
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.4184953677865833e-05,
+      "loss": 0.1371,
+      "step": 47350
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.4181605089853779e-05,
+      "loss": 0.1024,
+      "step": 47360
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.4178256501841725e-05,
+      "loss": 0.0723,
+      "step": 47370
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.4174907913829669e-05,
+      "loss": 0.1043,
+      "step": 47380
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.4171559325817613e-05,
+      "loss": 0.058,
+      "step": 47390
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.416821073780556e-05,
+      "loss": 0.1186,
+      "step": 47400
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.4164862149793504e-05,
+      "loss": 0.0778,
+      "step": 47410
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.4161513561781448e-05,
+      "loss": 0.1056,
+      "step": 47420
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.4158164973769394e-05,
+      "loss": 0.0297,
+      "step": 47430
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.415481638575734e-05,
+      "loss": 0.1027,
+      "step": 47440
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.4151467797745284e-05,
+      "loss": 0.044,
+      "step": 47450
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.414811920973323e-05,
+      "loss": 0.0466,
+      "step": 47460
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.4144770621721175e-05,
+      "loss": 0.1434,
+      "step": 47470
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.414142203370912e-05,
+      "loss": 0.0706,
+      "step": 47480
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.4138073445697065e-05,
+      "loss": 0.0508,
+      "step": 47490
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.413472485768501e-05,
+      "loss": 0.2215,
+      "step": 47500
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.4131376269672955e-05,
+      "loss": 0.0756,
+      "step": 47510
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.41280276816609e-05,
+      "loss": 0.1238,
+      "step": 47520
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.4124679093648845e-05,
+      "loss": 0.1452,
+      "step": 47530
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.412133050563679e-05,
+      "loss": 0.073,
+      "step": 47540
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.4117981917624735e-05,
+      "loss": 0.0967,
+      "step": 47550
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.411463332961268e-05,
+      "loss": 0.0639,
+      "step": 47560
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.4111284741600625e-05,
+      "loss": 0.1235,
+      "step": 47570
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.410793615358857e-05,
+      "loss": 0.0378,
+      "step": 47580
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.4104587565576516e-05,
+      "loss": 0.1144,
+      "step": 47590
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.410123897756446e-05,
+      "loss": 0.0693,
+      "step": 47600
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.4097890389552406e-05,
+      "loss": 0.0552,
+      "step": 47610
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.4094541801540352e-05,
+      "loss": 0.0947,
+      "step": 47620
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.4091193213528296e-05,
+      "loss": 0.1701,
+      "step": 47630
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.408784462551624e-05,
+      "loss": 0.158,
+      "step": 47640
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.4084496037504186e-05,
+      "loss": 0.0738,
+      "step": 47650
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.4081147449492131e-05,
+      "loss": 0.0994,
+      "step": 47660
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.4077798861480076e-05,
+      "loss": 0.1298,
+      "step": 47670
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1.4074450273468021e-05,
+      "loss": 0.1055,
+      "step": 47680
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1.4071101685455967e-05,
+      "loss": 0.126,
+      "step": 47690
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1.4067753097443911e-05,
+      "loss": 0.1435,
+      "step": 47700
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1.4064404509431857e-05,
+      "loss": 0.0755,
+      "step": 47710
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1.4061055921419803e-05,
+      "loss": 0.0834,
+      "step": 47720
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1.4057707333407747e-05,
+      "loss": 0.1184,
+      "step": 47730
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1.405435874539569e-05,
+      "loss": 0.0897,
+      "step": 47740
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1.4051010157383638e-05,
+      "loss": 0.1096,
+      "step": 47750
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.4047661569371582e-05,
+      "loss": 0.1228,
+      "step": 47760
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.4044312981359526e-05,
+      "loss": 0.0736,
+      "step": 47770
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.4040964393347472e-05,
+      "loss": 0.1716,
+      "step": 47780
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.4037615805335418e-05,
+      "loss": 0.0398,
+      "step": 47790
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.4034267217323362e-05,
+      "loss": 0.0735,
+      "step": 47800
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.4030918629311306e-05,
+      "loss": 0.0552,
+      "step": 47810
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.4027570041299254e-05,
+      "loss": 0.1514,
+      "step": 47820
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.4024221453287198e-05,
+      "loss": 0.1679,
+      "step": 47830
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.4020872865275142e-05,
+      "loss": 0.1752,
+      "step": 47840
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.401752427726309e-05,
+      "loss": 0.381,
+      "step": 47850
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.4014175689251033e-05,
+      "loss": 0.0637,
+      "step": 47860
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.4010827101238977e-05,
+      "loss": 0.1135,
+      "step": 47870
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.4007478513226923e-05,
+      "loss": 0.1722,
+      "step": 47880
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.4004129925214869e-05,
+      "loss": 0.099,
+      "step": 47890
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.4000781337202813e-05,
+      "loss": 0.0841,
+      "step": 47900
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.3997432749190757e-05,
+      "loss": 0.0617,
+      "step": 47910
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.3994084161178704e-05,
+      "loss": 0.1549,
+      "step": 47920
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.3990735573166649e-05,
+      "loss": 0.0591,
+      "step": 47930
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.3987386985154593e-05,
+      "loss": 0.0787,
+      "step": 47940
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.3984038397142538e-05,
+      "loss": 0.1282,
+      "step": 47950
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.3980689809130484e-05,
+      "loss": 0.0917,
+      "step": 47960
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.3977341221118428e-05,
+      "loss": 0.0374,
+      "step": 47970
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.3973992633106374e-05,
+      "loss": 0.0544,
+      "step": 47980
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.397064404509432e-05,
+      "loss": 0.09,
+      "step": 47990
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.3967295457082264e-05,
+      "loss": 0.134,
+      "step": 48000
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.396394686907021e-05,
+      "loss": 0.0727,
+      "step": 48010
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.3960598281058154e-05,
+      "loss": 0.0971,
+      "step": 48020
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.39572496930461e-05,
+      "loss": 0.1387,
+      "step": 48030
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.3953901105034044e-05,
+      "loss": 0.08,
+      "step": 48040
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.395055251702199e-05,
+      "loss": 0.1275,
+      "step": 48050
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.394753878781114e-05,
+      "loss": 0.1249,
+      "step": 48060
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.3944190199799086e-05,
+      "loss": 0.1211,
+      "step": 48070
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.394084161178703e-05,
+      "loss": 0.0718,
+      "step": 48080
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.3937493023774976e-05,
+      "loss": 0.0706,
+      "step": 48090
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.3934144435762921e-05,
+      "loss": 0.0991,
+      "step": 48100
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.3930795847750865e-05,
+      "loss": 0.0417,
+      "step": 48110
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.392744725973881e-05,
+      "loss": 0.0313,
+      "step": 48120
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.3924098671726755e-05,
+      "loss": 0.0932,
+      "step": 48130
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.3920750083714701e-05,
+      "loss": 0.0989,
+      "step": 48140
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.3917401495702645e-05,
+      "loss": 0.1379,
+      "step": 48150
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.3914052907690591e-05,
+      "loss": 0.09,
+      "step": 48160
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.3910704319678537e-05,
+      "loss": 0.1372,
+      "step": 48170
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.390735573166648e-05,
+      "loss": 0.0993,
+      "step": 48180
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.3904007143654427e-05,
+      "loss": 0.0502,
+      "step": 48190
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.390065855564237e-05,
+      "loss": 0.1157,
+      "step": 48200
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1.3897309967630316e-05,
+      "loss": 0.0719,
+      "step": 48210
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1.3893961379618262e-05,
+      "loss": 0.1455,
+      "step": 48220
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1.3890612791606206e-05,
+      "loss": 0.1061,
+      "step": 48230
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1.3887264203594152e-05,
+      "loss": 0.3486,
+      "step": 48240
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1.3883915615582096e-05,
+      "loss": 0.0935,
+      "step": 48250
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1.3880567027570042e-05,
+      "loss": 0.0695,
+      "step": 48260
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1.3877218439557986e-05,
+      "loss": 0.1135,
+      "step": 48270
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1.3873869851545932e-05,
+      "loss": 0.1332,
+      "step": 48280
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1.3870521263533877e-05,
+      "loss": 0.0943,
+      "step": 48290
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.3867172675521821e-05,
+      "loss": 0.0481,
+      "step": 48300
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.3863824087509767e-05,
+      "loss": 0.1407,
+      "step": 48310
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.3860475499497713e-05,
+      "loss": 0.162,
+      "step": 48320
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.3857126911485657e-05,
+      "loss": 0.2439,
+      "step": 48330
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.3853778323473601e-05,
+      "loss": 0.0292,
+      "step": 48340
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.3850429735461547e-05,
+      "loss": 0.1219,
+      "step": 48350
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.3847081147449493e-05,
+      "loss": 0.0558,
+      "step": 48360
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.3843732559437437e-05,
+      "loss": 0.1344,
+      "step": 48370
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.3840383971425383e-05,
+      "loss": 0.2018,
+      "step": 48380
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.3837035383413328e-05,
+      "loss": 0.0812,
+      "step": 48390
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.3833686795401272e-05,
+      "loss": 0.1327,
+      "step": 48400
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.3830338207389218e-05,
+      "loss": 0.095,
+      "step": 48410
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.3826989619377164e-05,
+      "loss": 0.1201,
+      "step": 48420
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.3823641031365108e-05,
+      "loss": 0.0541,
+      "step": 48430
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.3820292443353052e-05,
+      "loss": 0.0732,
+      "step": 48440
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.3816943855341e-05,
+      "loss": 0.0908,
+      "step": 48450
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.3813595267328944e-05,
+      "loss": 0.1817,
+      "step": 48460
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.3810246679316888e-05,
+      "loss": 0.102,
+      "step": 48470
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.3806898091304833e-05,
+      "loss": 0.1186,
+      "step": 48480
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.380354950329278e-05,
+      "loss": 0.1163,
+      "step": 48490
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.3800200915280723e-05,
+      "loss": 0.0827,
+      "step": 48500
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.3796852327268667e-05,
+      "loss": 0.0782,
+      "step": 48510
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.3793503739256615e-05,
+      "loss": 0.1369,
+      "step": 48520
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.3790155151244559e-05,
+      "loss": 0.2161,
+      "step": 48530
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.3786806563232503e-05,
+      "loss": 0.0879,
+      "step": 48540
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.378345797522045e-05,
+      "loss": 0.18,
+      "step": 48550
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.3780109387208395e-05,
+      "loss": 0.0767,
+      "step": 48560
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.3776760799196339e-05,
+      "loss": 0.0289,
+      "step": 48570
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.3773412211184284e-05,
+      "loss": 0.0692,
+      "step": 48580
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.377006362317223e-05,
+      "loss": 0.1002,
+      "step": 48590
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.3766715035160174e-05,
+      "loss": 0.1339,
+      "step": 48600
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.3763366447148118e-05,
+      "loss": 0.2227,
+      "step": 48610
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.3760017859136066e-05,
+      "loss": 0.0463,
+      "step": 48620
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.375666927112401e-05,
+      "loss": 0.1465,
+      "step": 48630
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.3753320683111954e-05,
+      "loss": 0.0689,
+      "step": 48640
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.37499720950999e-05,
+      "loss": 0.1037,
+      "step": 48650
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.3746623507087845e-05,
+      "loss": 0.0916,
+      "step": 48660
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.374327491907579e-05,
+      "loss": 0.1481,
+      "step": 48670
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.3739926331063735e-05,
+      "loss": 0.1205,
+      "step": 48680
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.3736577743051681e-05,
+      "loss": 0.2414,
+      "step": 48690
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.3733229155039625e-05,
+      "loss": 0.1527,
+      "step": 48700
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.3729880567027571e-05,
+      "loss": 0.0752,
+      "step": 48710
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.3726531979015515e-05,
+      "loss": 0.1265,
+      "step": 48720
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.372318339100346e-05,
+      "loss": 0.1169,
+      "step": 48730
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.3719834802991405e-05,
+      "loss": 0.0623,
+      "step": 48740
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.371648621497935e-05,
+      "loss": 0.0972,
+      "step": 48750
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.3713137626967296e-05,
+      "loss": 0.0623,
+      "step": 48760
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.370978903895524e-05,
+      "loss": 0.0397,
+      "step": 48770
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.3706440450943186e-05,
+      "loss": 0.0464,
+      "step": 48780
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.370309186293113e-05,
+      "loss": 0.1015,
+      "step": 48790
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.3699743274919076e-05,
+      "loss": 0.155,
+      "step": 48800
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.3696394686907022e-05,
+      "loss": 0.1564,
+      "step": 48810
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.3693046098894966e-05,
+      "loss": 0.1191,
+      "step": 48820
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.3689697510882912e-05,
+      "loss": 0.13,
+      "step": 48830
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.3686348922870857e-05,
+      "loss": 0.0765,
+      "step": 48840
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.3683000334858801e-05,
+      "loss": 0.0935,
+      "step": 48850
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.3679651746846747e-05,
+      "loss": 0.0744,
+      "step": 48860
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.3676303158834691e-05,
+      "loss": 0.0594,
+      "step": 48870
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.3672954570822637e-05,
+      "loss": 0.041,
+      "step": 48880
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.3669605982810581e-05,
+      "loss": 0.1362,
+      "step": 48890
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.3666257394798527e-05,
+      "loss": 0.1042,
+      "step": 48900
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.3662908806786473e-05,
+      "loss": 0.0511,
+      "step": 48910
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.3659560218774417e-05,
+      "loss": 0.0919,
+      "step": 48920
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1.3656211630762363e-05,
+      "loss": 0.1263,
+      "step": 48930
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1.3652863042750308e-05,
+      "loss": 0.0913,
+      "step": 48940
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1.3649514454738252e-05,
+      "loss": 0.1287,
+      "step": 48950
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1.3646165866726196e-05,
+      "loss": 0.1371,
+      "step": 48960
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1.3642817278714144e-05,
+      "loss": 0.0468,
+      "step": 48970
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1.3639468690702088e-05,
+      "loss": 0.0926,
+      "step": 48980
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1.3636120102690032e-05,
+      "loss": 0.121,
+      "step": 48990
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1.3632771514677978e-05,
+      "loss": 0.1847,
+      "step": 49000
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1.3629422926665924e-05,
+      "loss": 0.1015,
+      "step": 49010
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.3626074338653868e-05,
+      "loss": 0.0939,
+      "step": 49020
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.3622725750641812e-05,
+      "loss": 0.0827,
+      "step": 49030
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.361937716262976e-05,
+      "loss": 0.0877,
+      "step": 49040
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.3616028574617703e-05,
+      "loss": 0.115,
+      "step": 49050
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.3612679986605647e-05,
+      "loss": 0.0784,
+      "step": 49060
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.3609331398593595e-05,
+      "loss": 0.1534,
+      "step": 49070
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.3605982810581539e-05,
+      "loss": 0.11,
+      "step": 49080
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.3602634222569483e-05,
+      "loss": 0.1313,
+      "step": 49090
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.3599285634557429e-05,
+      "loss": 0.1157,
+      "step": 49100
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1.3595937046545374e-05,
+      "loss": 0.1329,
+      "step": 49110
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1.3592588458533319e-05,
+      "loss": 0.1169,
+      "step": 49120
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1.3589239870521263e-05,
+      "loss": 0.1122,
+      "step": 49130
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1.358589128250921e-05,
+      "loss": 0.0905,
+      "step": 49140
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1.3582542694497154e-05,
+      "loss": 0.1308,
+      "step": 49150
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1.3579194106485098e-05,
+      "loss": 0.0596,
+      "step": 49160
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1.3575845518473044e-05,
+      "loss": 0.0545,
+      "step": 49170
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1.357249693046099e-05,
+      "loss": 0.0936,
+      "step": 49180
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1.3569148342448934e-05,
+      "loss": 0.1488,
+      "step": 49190
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1.356579975443688e-05,
+      "loss": 0.1255,
+      "step": 49200
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1.3562451166424825e-05,
+      "loss": 0.0272,
+      "step": 49210
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1.355910257841277e-05,
+      "loss": 0.083,
+      "step": 49220
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1.3555753990400714e-05,
+      "loss": 0.1218,
+      "step": 49230
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1.355240540238866e-05,
+      "loss": 0.0391,
+      "step": 49240
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1.3549056814376605e-05,
+      "loss": 0.1374,
+      "step": 49250
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1.354570822636455e-05,
+      "loss": 0.086,
+      "step": 49260
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1.3542359638352495e-05,
+      "loss": 0.0513,
+      "step": 49270
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1.353901105034044e-05,
+      "loss": 0.0471,
+      "step": 49280
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.3535662462328385e-05,
+      "loss": 0.0937,
+      "step": 49290
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.353231387431633e-05,
+      "loss": 0.0894,
+      "step": 49300
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.3528965286304276e-05,
+      "loss": 0.0708,
+      "step": 49310
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.352561669829222e-05,
+      "loss": 0.1078,
+      "step": 49320
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.3522268110280166e-05,
+      "loss": 0.0774,
+      "step": 49330
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.351891952226811e-05,
+      "loss": 0.0944,
+      "step": 49340
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.3515570934256056e-05,
+      "loss": 0.1083,
+      "step": 49350
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.3512222346244e-05,
+      "loss": 0.0784,
+      "step": 49360
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.3508873758231946e-05,
+      "loss": 0.1463,
+      "step": 49370
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.3505525170219892e-05,
+      "loss": 0.0539,
+      "step": 49380
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.3502176582207836e-05,
+      "loss": 0.1058,
+      "step": 49390
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.3498827994195781e-05,
+      "loss": 0.0817,
+      "step": 49400
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.3495479406183726e-05,
+      "loss": 0.0584,
+      "step": 49410
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.3492130818171671e-05,
+      "loss": 0.0749,
+      "step": 49420
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.3488782230159617e-05,
+      "loss": 0.0795,
+      "step": 49430
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.3485433642147561e-05,
+      "loss": 0.1436,
+      "step": 49440
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.3482085054135507e-05,
+      "loss": 0.1616,
+      "step": 49450
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.3478736466123453e-05,
+      "loss": 0.058,
+      "step": 49460
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.3475387878111397e-05,
+      "loss": 0.1286,
+      "step": 49470
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.347203929009934e-05,
+      "loss": 0.0574,
+      "step": 49480
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.3468690702087287e-05,
+      "loss": 0.1287,
+      "step": 49490
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.3465342114075232e-05,
+      "loss": 0.135,
+      "step": 49500
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.3461993526063176e-05,
+      "loss": 0.0957,
+      "step": 49510
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.3458644938051122e-05,
+      "loss": 0.1224,
+      "step": 49520
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.3455296350039068e-05,
+      "loss": 0.0842,
+      "step": 49530
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.3451947762027012e-05,
+      "loss": 0.0787,
+      "step": 49540
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.3448599174014956e-05,
+      "loss": 0.1276,
+      "step": 49550
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.3445250586002904e-05,
+      "loss": 0.0883,
+      "step": 49560
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.3441901997990848e-05,
+      "loss": 0.0753,
+      "step": 49570
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.3438553409978792e-05,
+      "loss": 0.0168,
+      "step": 49580
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.343520482196674e-05,
+      "loss": 0.0719,
+      "step": 49590
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.3431856233954683e-05,
+      "loss": 0.1018,
+      "step": 49600
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.3428507645942627e-05,
+      "loss": 0.1008,
+      "step": 49610
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.3425159057930571e-05,
+      "loss": 0.0536,
+      "step": 49620
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.3421810469918519e-05,
+      "loss": 0.0735,
+      "step": 49630
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.3418461881906463e-05,
+      "loss": 0.0791,
+      "step": 49640
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.3415113293894407e-05,
+      "loss": 0.0772,
+      "step": 49650
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.3411764705882354e-05,
+      "loss": 0.1384,
+      "step": 49660
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.3408416117870299e-05,
+      "loss": 0.1113,
+      "step": 49670
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.3405067529858243e-05,
+      "loss": 0.0558,
+      "step": 49680
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.3401718941846188e-05,
+      "loss": 0.0739,
+      "step": 49690
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.3398370353834134e-05,
+      "loss": 0.09,
+      "step": 49700
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.3395021765822078e-05,
+      "loss": 0.0352,
+      "step": 49710
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.3391673177810024e-05,
+      "loss": 0.1288,
+      "step": 49720
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.338832458979797e-05,
+      "loss": 0.1065,
+      "step": 49730
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1.3384976001785914e-05,
+      "loss": 0.0811,
+      "step": 49740
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1.3381627413773858e-05,
+      "loss": 0.1158,
+      "step": 49750
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1.3378278825761805e-05,
+      "loss": 0.0523,
+      "step": 49760
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1.337493023774975e-05,
+      "loss": 0.0476,
+      "step": 49770
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1.3371581649737694e-05,
+      "loss": 0.1259,
+      "step": 49780
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1.336823306172564e-05,
+      "loss": 0.1922,
+      "step": 49790
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1.3364884473713585e-05,
+      "loss": 0.1206,
+      "step": 49800
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1.3361535885701529e-05,
+      "loss": 0.0176,
+      "step": 49810
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1.3358187297689475e-05,
+      "loss": 0.0476,
+      "step": 49820
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1.335483870967742e-05,
+      "loss": 0.0684,
+      "step": 49830
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1.3351490121665365e-05,
+      "loss": 0.0511,
+      "step": 49840
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1.334814153365331e-05,
+      "loss": 0.0599,
+      "step": 49850
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1.3344792945641255e-05,
+      "loss": 0.132,
+      "step": 49860
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1.33414443576292e-05,
+      "loss": 0.145,
+      "step": 49870
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1.3338095769617144e-05,
+      "loss": 0.1185,
+      "step": 49880
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1.333474718160509e-05,
+      "loss": 0.0423,
+      "step": 49890
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1.3331398593593036e-05,
+      "loss": 0.0836,
+      "step": 49900
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1.332805000558098e-05,
+      "loss": 0.0865,
+      "step": 49910
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.3324701417568926e-05,
+      "loss": 0.0607,
+      "step": 49920
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.332135282955687e-05,
+      "loss": 0.1082,
+      "step": 49930
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.3318004241544816e-05,
+      "loss": 0.0688,
+      "step": 49940
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.3314655653532761e-05,
+      "loss": 0.1214,
+      "step": 49950
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.3311307065520705e-05,
+      "loss": 0.1071,
+      "step": 49960
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.3307958477508651e-05,
+      "loss": 0.054,
+      "step": 49970
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.3304609889496595e-05,
+      "loss": 0.0453,
+      "step": 49980
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.3301261301484541e-05,
+      "loss": 0.099,
+      "step": 49990
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.3297912713472485e-05,
+      "loss": 0.299,
+      "step": 50000
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.3294564125460431e-05,
+      "loss": 0.0944,
+      "step": 50010
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.3291215537448377e-05,
+      "loss": 0.0476,
+      "step": 50020
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.328786694943632e-05,
+      "loss": 0.0534,
+      "step": 50030
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.3284518361424267e-05,
+      "loss": 0.1208,
+      "step": 50040
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.3281169773412212e-05,
+      "loss": 0.1671,
+      "step": 50050
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.3277821185400156e-05,
+      "loss": 0.0824,
+      "step": 50060
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.32744725973881e-05,
+      "loss": 0.0568,
+      "step": 50070
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.3271124009376048e-05,
+      "loss": 0.0908,
+      "step": 50080
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.3267775421363992e-05,
+      "loss": 0.1092,
+      "step": 50090
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.3264426833351936e-05,
+      "loss": 0.1775,
+      "step": 50100
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.3261078245339882e-05,
+      "loss": 0.1031,
+      "step": 50110
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.3257729657327828e-05,
+      "loss": 0.0648,
+      "step": 50120
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.3254381069315772e-05,
+      "loss": 0.0511,
+      "step": 50130
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.3251032481303716e-05,
+      "loss": 0.141,
+      "step": 50140
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.3247683893291663e-05,
+      "loss": 0.0657,
+      "step": 50150
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.3244335305279607e-05,
+      "loss": 0.0794,
+      "step": 50160
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.3240986717267551e-05,
+      "loss": 0.1203,
+      "step": 50170
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.3237638129255499e-05,
+      "loss": 0.0768,
+      "step": 50180
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.3234289541243443e-05,
+      "loss": 0.2054,
+      "step": 50190
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.3230940953231387e-05,
+      "loss": 0.144,
+      "step": 50200
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.3227592365219333e-05,
+      "loss": 0.0992,
+      "step": 50210
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.3224243777207279e-05,
+      "loss": 0.0816,
+      "step": 50220
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.3220895189195223e-05,
+      "loss": 0.094,
+      "step": 50230
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.3217546601183167e-05,
+      "loss": 0.0934,
+      "step": 50240
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.3214198013171114e-05,
+      "loss": 0.0787,
+      "step": 50250
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.3210849425159058e-05,
+      "loss": 0.0814,
+      "step": 50260
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.3207500837147002e-05,
+      "loss": 0.0914,
+      "step": 50270
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.320415224913495e-05,
+      "loss": 0.1229,
+      "step": 50280
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.3200803661122894e-05,
+      "loss": 0.1083,
+      "step": 50290
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.3197455073110838e-05,
+      "loss": 0.1047,
+      "step": 50300
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.3194106485098784e-05,
+      "loss": 0.071,
+      "step": 50310
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.319075789708673e-05,
+      "loss": 0.0717,
+      "step": 50320
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.3187409309074674e-05,
+      "loss": 0.1497,
+      "step": 50330
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.318406072106262e-05,
+      "loss": 0.0815,
+      "step": 50340
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.3180712133050565e-05,
+      "loss": 0.0916,
+      "step": 50350
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.3177363545038509e-05,
+      "loss": 0.1439,
+      "step": 50360
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.3174014957026453e-05,
+      "loss": 0.1572,
+      "step": 50370
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.3170666369014399e-05,
+      "loss": 0.0764,
+      "step": 50380
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.3167317781002345e-05,
+      "loss": 0.0644,
+      "step": 50390
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.3163969192990289e-05,
+      "loss": 0.2301,
+      "step": 50400
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.3160620604978235e-05,
+      "loss": 0.0703,
+      "step": 50410
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.315727201696618e-05,
+      "loss": 0.0845,
+      "step": 50420
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.3153923428954124e-05,
+      "loss": 0.0518,
+      "step": 50430
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.315057484094207e-05,
+      "loss": 0.1138,
+      "step": 50440
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.3147226252930014e-05,
+      "loss": 0.0872,
+      "step": 50450
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.314387766491796e-05,
+      "loss": 0.1502,
+      "step": 50460
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.3140529076905906e-05,
+      "loss": 0.0874,
+      "step": 50470
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.313718048889385e-05,
+      "loss": 0.0552,
+      "step": 50480
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.3133831900881796e-05,
+      "loss": 0.0599,
+      "step": 50490
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.313048331286974e-05,
+      "loss": 0.117,
+      "step": 50500
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.3127134724857685e-05,
+      "loss": 0.1036,
+      "step": 50510
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.312378613684563e-05,
+      "loss": 0.0665,
+      "step": 50520
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.3120437548833575e-05,
+      "loss": 0.1061,
+      "step": 50530
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.3117088960821521e-05,
+      "loss": 0.1561,
+      "step": 50540
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.3113740372809465e-05,
+      "loss": 0.1008,
+      "step": 50550
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.3110391784797411e-05,
+      "loss": 0.1101,
+      "step": 50560
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.3107043196785357e-05,
+      "loss": 0.0374,
+      "step": 50570
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.31036946087733e-05,
+      "loss": 0.0459,
+      "step": 50580
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.3100346020761245e-05,
+      "loss": 0.0613,
+      "step": 50590
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.3096997432749192e-05,
+      "loss": 0.1191,
+      "step": 50600
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.3093648844737136e-05,
+      "loss": 0.0407,
+      "step": 50610
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.309030025672508e-05,
+      "loss": 0.0711,
+      "step": 50620
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.3086951668713026e-05,
+      "loss": 0.2486,
+      "step": 50630
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.3083603080700972e-05,
+      "loss": 0.0719,
+      "step": 50640
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.3080254492688916e-05,
+      "loss": 0.0691,
+      "step": 50650
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.307690590467686e-05,
+      "loss": 0.1863,
+      "step": 50660
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.3073557316664808e-05,
+      "loss": 0.0322,
+      "step": 50670
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.3070208728652752e-05,
+      "loss": 0.0961,
+      "step": 50680
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.3066860140640696e-05,
+      "loss": 0.077,
+      "step": 50690
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.3063511552628643e-05,
+      "loss": 0.1206,
+      "step": 50700
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.3060162964616587e-05,
+      "loss": 0.167,
+      "step": 50710
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.3056814376604531e-05,
+      "loss": 0.0545,
+      "step": 50720
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.3053465788592477e-05,
+      "loss": 0.0536,
+      "step": 50730
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.3050117200580423e-05,
+      "loss": 0.1299,
+      "step": 50740
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.3046768612568367e-05,
+      "loss": 0.1694,
+      "step": 50750
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.3043420024556311e-05,
+      "loss": 0.0344,
+      "step": 50760
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.3040071436544259e-05,
+      "loss": 0.0503,
+      "step": 50770
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.3036722848532203e-05,
+      "loss": 0.0844,
+      "step": 50780
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.3033374260520147e-05,
+      "loss": 0.0944,
+      "step": 50790
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.3030025672508094e-05,
+      "loss": 0.1582,
+      "step": 50800
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.3027011943297245e-05,
+      "loss": 0.1435,
+      "step": 50810
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.3023663355285189e-05,
+      "loss": 0.155,
+      "step": 50820
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.3020314767273133e-05,
+      "loss": 0.1289,
+      "step": 50830
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.3016966179261077e-05,
+      "loss": 0.0466,
+      "step": 50840
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.3013617591249024e-05,
+      "loss": 0.1198,
+      "step": 50850
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.3010269003236969e-05,
+      "loss": 0.1149,
+      "step": 50860
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.3006920415224913e-05,
+      "loss": 0.0908,
+      "step": 50870
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.300357182721286e-05,
+      "loss": 0.1127,
+      "step": 50880
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.3000223239200804e-05,
+      "loss": 0.0986,
+      "step": 50890
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.2996874651188748e-05,
+      "loss": 0.1251,
+      "step": 50900
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.2993526063176694e-05,
+      "loss": 0.0952,
+      "step": 50910
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.299017747516464e-05,
+      "loss": 0.1174,
+      "step": 50920
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.2986828887152584e-05,
+      "loss": 0.0312,
+      "step": 50930
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.298348029914053e-05,
+      "loss": 0.0398,
+      "step": 50940
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.2980131711128475e-05,
+      "loss": 0.048,
+      "step": 50950
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.297678312311642e-05,
+      "loss": 0.048,
+      "step": 50960
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.2973434535104364e-05,
+      "loss": 0.0837,
+      "step": 50970
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.2970085947092311e-05,
+      "loss": 0.0726,
+      "step": 50980
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.2966737359080255e-05,
+      "loss": 0.0834,
+      "step": 50990
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.2963388771068199e-05,
+      "loss": 0.0974,
+      "step": 51000
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.2960040183056145e-05,
+      "loss": 0.0791,
+      "step": 51010
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.295669159504409e-05,
+      "loss": 0.0545,
+      "step": 51020
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.2953343007032035e-05,
+      "loss": 0.1442,
+      "step": 51030
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.294999441901998e-05,
+      "loss": 0.0541,
+      "step": 51040
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.2946645831007926e-05,
+      "loss": 0.0915,
+      "step": 51050
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.294329724299587e-05,
+      "loss": 0.1347,
+      "step": 51060
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.2939948654983816e-05,
+      "loss": 0.0653,
+      "step": 51070
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.293660006697176e-05,
+      "loss": 0.0648,
+      "step": 51080
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.2933251478959706e-05,
+      "loss": 0.1143,
+      "step": 51090
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.292990289094765e-05,
+      "loss": 0.1293,
+      "step": 51100
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.2926554302935596e-05,
+      "loss": 0.0553,
+      "step": 51110
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.2923205714923542e-05,
+      "loss": 0.0548,
+      "step": 51120
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.2919857126911486e-05,
+      "loss": 0.1101,
+      "step": 51130
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.2916508538899431e-05,
+      "loss": 0.1967,
+      "step": 51140
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.2913159950887376e-05,
+      "loss": 0.1272,
+      "step": 51150
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.2909811362875321e-05,
+      "loss": 0.0828,
+      "step": 51160
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.2906462774863267e-05,
+      "loss": 0.0821,
+      "step": 51170
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.2903114186851211e-05,
+      "loss": 0.0941,
+      "step": 51180
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.2899765598839157e-05,
+      "loss": 0.1084,
+      "step": 51190
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.2896417010827101e-05,
+      "loss": 0.0893,
+      "step": 51200
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.2893068422815047e-05,
+      "loss": 0.1316,
+      "step": 51210
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.288971983480299e-05,
+      "loss": 0.1468,
+      "step": 51220
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.2886371246790937e-05,
+      "loss": 0.1416,
+      "step": 51230
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.2883022658778882e-05,
+      "loss": 0.1075,
+      "step": 51240
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.2879674070766826e-05,
+      "loss": 0.1429,
+      "step": 51250
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.2876325482754772e-05,
+      "loss": 0.1568,
+      "step": 51260
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.2872976894742718e-05,
+      "loss": 0.1454,
+      "step": 51270
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.2869628306730662e-05,
+      "loss": 0.0454,
+      "step": 51280
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.2866279718718606e-05,
+      "loss": 0.0294,
+      "step": 51290
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.2862931130706554e-05,
+      "loss": 0.1004,
+      "step": 51300
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.2859582542694498e-05,
+      "loss": 0.1255,
+      "step": 51310
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.2856233954682442e-05,
+      "loss": 0.1062,
+      "step": 51320
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.2852885366670387e-05,
+      "loss": 0.0533,
+      "step": 51330
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.2849536778658333e-05,
+      "loss": 0.1116,
+      "step": 51340
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.2846188190646277e-05,
+      "loss": 0.1041,
+      "step": 51350
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.2842839602634221e-05,
+      "loss": 0.081,
+      "step": 51360
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.2839491014622169e-05,
+      "loss": 0.0941,
+      "step": 51370
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.2836142426610113e-05,
+      "loss": 0.0657,
+      "step": 51380
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.2832793838598057e-05,
+      "loss": 0.076,
+      "step": 51390
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.2829445250586004e-05,
+      "loss": 0.1504,
+      "step": 51400
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.2826096662573949e-05,
+      "loss": 0.0947,
+      "step": 51410
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.2822748074561893e-05,
+      "loss": 0.1076,
+      "step": 51420
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.281939948654984e-05,
+      "loss": 0.0511,
+      "step": 51430
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.2816050898537784e-05,
+      "loss": 0.0881,
+      "step": 51440
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.2812702310525728e-05,
+      "loss": 0.0857,
+      "step": 51450
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.2809353722513672e-05,
+      "loss": 0.1599,
+      "step": 51460
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.280600513450162e-05,
+      "loss": 0.0545,
+      "step": 51470
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.2802656546489564e-05,
+      "loss": 0.0736,
+      "step": 51480
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.2799307958477508e-05,
+      "loss": 0.1403,
+      "step": 51490
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.2795959370465455e-05,
+      "loss": 0.1223,
+      "step": 51500
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.27926107824534e-05,
+      "loss": 0.0666,
+      "step": 51510
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.2789262194441344e-05,
+      "loss": 0.2409,
+      "step": 51520
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.278591360642929e-05,
+      "loss": 0.1252,
+      "step": 51530
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.2782565018417235e-05,
+      "loss": 0.0991,
+      "step": 51540
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.2779216430405179e-05,
+      "loss": 0.1095,
+      "step": 51550
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.2775867842393125e-05,
+      "loss": 0.0838,
+      "step": 51560
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.277251925438107e-05,
+      "loss": 0.1614,
+      "step": 51570
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.2769170666369015e-05,
+      "loss": 0.2143,
+      "step": 51580
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.2765822078356959e-05,
+      "loss": 0.0647,
+      "step": 51590
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.2762473490344905e-05,
+      "loss": 0.1012,
+      "step": 51600
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.275912490233285e-05,
+      "loss": 0.0949,
+      "step": 51610
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.2755776314320794e-05,
+      "loss": 0.128,
+      "step": 51620
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.275242772630874e-05,
+      "loss": 0.0896,
+      "step": 51630
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.2749079138296686e-05,
+      "loss": 0.0561,
+      "step": 51640
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.274573055028463e-05,
+      "loss": 0.0929,
+      "step": 51650
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.2742381962272576e-05,
+      "loss": 0.0692,
+      "step": 51660
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.273903337426052e-05,
+      "loss": 0.0668,
+      "step": 51670
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.2735684786248466e-05,
+      "loss": 0.0866,
+      "step": 51680
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.2732336198236411e-05,
+      "loss": 0.1047,
+      "step": 51690
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.2728987610224355e-05,
+      "loss": 0.1247,
+      "step": 51700
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.2725639022212301e-05,
+      "loss": 0.1584,
+      "step": 51710
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.2722290434200245e-05,
+      "loss": 0.0904,
+      "step": 51720
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.2718941846188191e-05,
+      "loss": 0.0583,
+      "step": 51730
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.2715593258176135e-05,
+      "loss": 0.1646,
+      "step": 51740
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.2712244670164081e-05,
+      "loss": 0.0712,
+      "step": 51750
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.2708896082152027e-05,
+      "loss": 0.1565,
+      "step": 51760
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.270554749413997e-05,
+      "loss": 0.0679,
+      "step": 51770
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.2702198906127917e-05,
+      "loss": 0.0652,
+      "step": 51780
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.2698850318115862e-05,
+      "loss": 0.0581,
+      "step": 51790
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.2695501730103806e-05,
+      "loss": 0.1754,
+      "step": 51800
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.269215314209175e-05,
+      "loss": 0.0194,
+      "step": 51810
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.2688804554079698e-05,
+      "loss": 0.1061,
+      "step": 51820
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.2685455966067642e-05,
+      "loss": 0.0688,
+      "step": 51830
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.2682107378055586e-05,
+      "loss": 0.0386,
+      "step": 51840
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.2678758790043532e-05,
+      "loss": 0.0748,
+      "step": 51850
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.2675410202031478e-05,
+      "loss": 0.077,
+      "step": 51860
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.2672061614019422e-05,
+      "loss": 0.102,
+      "step": 51870
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.2668713026007367e-05,
+      "loss": 0.1255,
+      "step": 51880
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.2665364437995313e-05,
+      "loss": 0.0783,
+      "step": 51890
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.2662015849983257e-05,
+      "loss": 0.0966,
+      "step": 51900
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.2658667261971201e-05,
+      "loss": 0.1092,
+      "step": 51910
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.2655318673959149e-05,
+      "loss": 0.1791,
+      "step": 51920
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.2651970085947093e-05,
+      "loss": 0.0607,
+      "step": 51930
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.2648621497935037e-05,
+      "loss": 0.1783,
+      "step": 51940
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.2645272909922983e-05,
+      "loss": 0.118,
+      "step": 51950
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.2641924321910929e-05,
+      "loss": 0.0697,
+      "step": 51960
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.2638575733898873e-05,
+      "loss": 0.0692,
+      "step": 51970
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.2635227145886817e-05,
+      "loss": 0.0639,
+      "step": 51980
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.2631878557874764e-05,
+      "loss": 0.2394,
+      "step": 51990
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.2628529969862708e-05,
+      "loss": 0.0521,
+      "step": 52000
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.2625181381850652e-05,
+      "loss": 0.0901,
+      "step": 52010
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.26218327938386e-05,
+      "loss": 0.0851,
+      "step": 52020
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.2618484205826544e-05,
+      "loss": 0.032,
+      "step": 52030
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.2615135617814488e-05,
+      "loss": 0.1072,
+      "step": 52040
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.2611787029802434e-05,
+      "loss": 0.0554,
+      "step": 52050
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.260843844179038e-05,
+      "loss": 0.0866,
+      "step": 52060
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.2605089853778323e-05,
+      "loss": 0.0525,
+      "step": 52070
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.2601741265766268e-05,
+      "loss": 0.1048,
+      "step": 52080
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.2598392677754215e-05,
+      "loss": 0.0707,
+      "step": 52090
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.2595044089742159e-05,
+      "loss": 0.1583,
+      "step": 52100
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.2591695501730103e-05,
+      "loss": 0.0634,
+      "step": 52110
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.2588346913718049e-05,
+      "loss": 0.0986,
+      "step": 52120
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.2584998325705995e-05,
+      "loss": 0.0544,
+      "step": 52130
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.2581649737693939e-05,
+      "loss": 0.0709,
+      "step": 52140
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.2578301149681885e-05,
+      "loss": 0.1089,
+      "step": 52150
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.257495256166983e-05,
+      "loss": 0.0736,
+      "step": 52160
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.2571603973657774e-05,
+      "loss": 0.2019,
+      "step": 52170
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.256825538564572e-05,
+      "loss": 0.061,
+      "step": 52180
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.2564906797633664e-05,
+      "loss": 0.0408,
+      "step": 52190
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.256155820962161e-05,
+      "loss": 0.1413,
+      "step": 52200
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.2558209621609554e-05,
+      "loss": 0.093,
+      "step": 52210
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.25548610335975e-05,
+      "loss": 0.0375,
+      "step": 52220
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.2551512445585446e-05,
+      "loss": 0.0298,
+      "step": 52230
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.254816385757339e-05,
+      "loss": 0.0449,
+      "step": 52240
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.2544815269561335e-05,
+      "loss": 0.1083,
+      "step": 52250
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.254146668154928e-05,
+      "loss": 0.0525,
+      "step": 52260
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.2538118093537225e-05,
+      "loss": 0.118,
+      "step": 52270
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.2534769505525171e-05,
+      "loss": 0.0901,
+      "step": 52280
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.2531420917513115e-05,
+      "loss": 0.0982,
+      "step": 52290
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.2528072329501061e-05,
+      "loss": 0.1132,
+      "step": 52300
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.2524723741489007e-05,
+      "loss": 0.1738,
+      "step": 52310
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.252137515347695e-05,
+      "loss": 0.0735,
+      "step": 52320
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.2518026565464897e-05,
+      "loss": 0.0671,
+      "step": 52330
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.251467797745284e-05,
+      "loss": 0.0543,
+      "step": 52340
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.2511329389440786e-05,
+      "loss": 0.2866,
+      "step": 52350
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.250798080142873e-05,
+      "loss": 0.0982,
+      "step": 52360
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.2504632213416676e-05,
+      "loss": 0.0495,
+      "step": 52370
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.2501283625404622e-05,
+      "loss": 0.0957,
+      "step": 52380
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.2497935037392566e-05,
+      "loss": 0.0986,
+      "step": 52390
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.2494586449380512e-05,
+      "loss": 0.1379,
+      "step": 52400
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.2491237861368458e-05,
+      "loss": 0.1216,
+      "step": 52410
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.2487889273356402e-05,
+      "loss": 0.1881,
+      "step": 52420
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.2484540685344346e-05,
+      "loss": 0.0493,
+      "step": 52430
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.2481192097332293e-05,
+      "loss": 0.1062,
+      "step": 52440
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.2477843509320237e-05,
+      "loss": 0.1388,
+      "step": 52450
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.2474494921308181e-05,
+      "loss": 0.1157,
+      "step": 52460
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.2471146333296127e-05,
+      "loss": 0.0851,
+      "step": 52470
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.2467797745284073e-05,
+      "loss": 0.0481,
+      "step": 52480
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.2464449157272017e-05,
+      "loss": 0.0866,
+      "step": 52490
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.2461100569259961e-05,
+      "loss": 0.0633,
+      "step": 52500
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.2457751981247908e-05,
+      "loss": 0.0435,
+      "step": 52510
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.2454403393235853e-05,
+      "loss": 0.058,
+      "step": 52520
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.2451054805223797e-05,
+      "loss": 0.1375,
+      "step": 52530
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.2447706217211744e-05,
+      "loss": 0.0673,
+      "step": 52540
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.2444357629199688e-05,
+      "loss": 0.1051,
+      "step": 52550
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.2441009041187632e-05,
+      "loss": 0.1088,
+      "step": 52560
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.2437660453175578e-05,
+      "loss": 0.2811,
+      "step": 52570
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.2434311865163524e-05,
+      "loss": 0.0636,
+      "step": 52580
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.2430963277151468e-05,
+      "loss": 0.0719,
+      "step": 52590
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.2427614689139412e-05,
+      "loss": 0.27,
+      "step": 52600
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1.242426610112736e-05,
+      "loss": 0.0778,
+      "step": 52610
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1.2420917513115303e-05,
+      "loss": 0.1304,
+      "step": 52620
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1.2417568925103248e-05,
+      "loss": 0.043,
+      "step": 52630
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1.2414220337091193e-05,
+      "loss": 0.0894,
+      "step": 52640
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1.2410871749079139e-05,
+      "loss": 0.0734,
+      "step": 52650
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1.2407523161067083e-05,
+      "loss": 0.1323,
+      "step": 52660
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1.2404174573055029e-05,
+      "loss": 0.1057,
+      "step": 52670
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1.2400825985042975e-05,
+      "loss": 0.0719,
+      "step": 52680
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1.2397477397030919e-05,
+      "loss": 0.2849,
+      "step": 52690
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.2394128809018865e-05,
+      "loss": 0.1994,
+      "step": 52700
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.2390780221006809e-05,
+      "loss": 0.0559,
+      "step": 52710
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.2387431632994754e-05,
+      "loss": 0.0784,
+      "step": 52720
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.2384083044982698e-05,
+      "loss": 0.1094,
+      "step": 52730
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.2380734456970644e-05,
+      "loss": 0.2327,
+      "step": 52740
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.237738586895859e-05,
+      "loss": 0.0839,
+      "step": 52750
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.2374037280946534e-05,
+      "loss": 0.1222,
+      "step": 52760
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.237068869293448e-05,
+      "loss": 0.0501,
+      "step": 52770
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.2367340104922426e-05,
+      "loss": 0.076,
+      "step": 52780
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.236399151691037e-05,
+      "loss": 0.5618,
+      "step": 52790
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.2360642928898315e-05,
+      "loss": 0.1162,
+      "step": 52800
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.235729434088626e-05,
+      "loss": 0.1004,
+      "step": 52810
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.2353945752874205e-05,
+      "loss": 0.0822,
+      "step": 52820
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.235059716486215e-05,
+      "loss": 0.1571,
+      "step": 52830
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.2347248576850095e-05,
+      "loss": 0.0591,
+      "step": 52840
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.2343899988838041e-05,
+      "loss": 0.1668,
+      "step": 52850
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.2340551400825985e-05,
+      "loss": 0.0519,
+      "step": 52860
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.233720281281393e-05,
+      "loss": 0.0891,
+      "step": 52870
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.2333854224801875e-05,
+      "loss": 0.0569,
+      "step": 52880
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.233050563678982e-05,
+      "loss": 0.0877,
+      "step": 52890
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.2327157048777766e-05,
+      "loss": 0.188,
+      "step": 52900
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.232380846076571e-05,
+      "loss": 0.0824,
+      "step": 52910
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.2320459872753656e-05,
+      "loss": 0.0626,
+      "step": 52920
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.2317111284741602e-05,
+      "loss": 0.1336,
+      "step": 52930
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.2313762696729546e-05,
+      "loss": 0.1229,
+      "step": 52940
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.231041410871749e-05,
+      "loss": 0.127,
+      "step": 52950
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.2307065520705436e-05,
+      "loss": 0.0557,
+      "step": 52960
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.2303716932693382e-05,
+      "loss": 0.0801,
+      "step": 52970
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.2300368344681326e-05,
+      "loss": 0.0669,
+      "step": 52980
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.2297019756669271e-05,
+      "loss": 0.1926,
+      "step": 52990
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.2293671168657217e-05,
+      "loss": 0.1446,
+      "step": 53000
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.2290322580645161e-05,
+      "loss": 0.0885,
+      "step": 53010
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.2286973992633105e-05,
+      "loss": 0.0787,
+      "step": 53020
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.2283625404621053e-05,
+      "loss": 0.0889,
+      "step": 53030
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.2280276816608997e-05,
+      "loss": 0.0858,
+      "step": 53040
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.2276928228596941e-05,
+      "loss": 0.0838,
+      "step": 53050
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.2273579640584888e-05,
+      "loss": 0.0811,
+      "step": 53060
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.2270231052572833e-05,
+      "loss": 0.1672,
+      "step": 53070
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.2266882464560777e-05,
+      "loss": 0.084,
+      "step": 53080
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.226353387654872e-05,
+      "loss": 0.0849,
+      "step": 53090
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.2260185288536668e-05,
+      "loss": 0.0957,
+      "step": 53100
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.2256836700524612e-05,
+      "loss": 0.0707,
+      "step": 53110
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.2253488112512556e-05,
+      "loss": 0.1134,
+      "step": 53120
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.2250139524500504e-05,
+      "loss": 0.0297,
+      "step": 53130
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.2246790936488448e-05,
+      "loss": 0.0572,
+      "step": 53140
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.2243442348476392e-05,
+      "loss": 0.2153,
+      "step": 53150
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.2240093760464338e-05,
+      "loss": 0.0712,
+      "step": 53160
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.2236745172452283e-05,
+      "loss": 0.0962,
+      "step": 53170
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.2233396584440228e-05,
+      "loss": 0.0545,
+      "step": 53180
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.2230047996428173e-05,
+      "loss": 0.0569,
+      "step": 53190
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.2226699408416119e-05,
+      "loss": 0.0981,
+      "step": 53200
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.2223350820404063e-05,
+      "loss": 0.1038,
+      "step": 53210
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.2220002232392007e-05,
+      "loss": 0.0774,
+      "step": 53220
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.2216653644379955e-05,
+      "loss": 0.1601,
+      "step": 53230
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.2213305056367899e-05,
+      "loss": 0.0786,
+      "step": 53240
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.2209956468355843e-05,
+      "loss": 0.0889,
+      "step": 53250
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.2206607880343789e-05,
+      "loss": 0.0817,
+      "step": 53260
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.2203259292331734e-05,
+      "loss": 0.1271,
+      "step": 53270
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.2199910704319678e-05,
+      "loss": 0.0711,
+      "step": 53280
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.2196562116307624e-05,
+      "loss": 0.0282,
+      "step": 53290
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.219321352829557e-05,
+      "loss": 0.2228,
+      "step": 53300
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.2189864940283514e-05,
+      "loss": 0.0674,
+      "step": 53310
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.218651635227146e-05,
+      "loss": 0.144,
+      "step": 53320
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1.2183167764259404e-05,
+      "loss": 0.0968,
+      "step": 53330
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1.217981917624735e-05,
+      "loss": 0.1083,
+      "step": 53340
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1.2176470588235294e-05,
+      "loss": 0.1298,
+      "step": 53350
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1.217312200022324e-05,
+      "loss": 0.1454,
+      "step": 53360
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1.2169773412211185e-05,
+      "loss": 0.0582,
+      "step": 53370
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1.216642482419913e-05,
+      "loss": 0.1526,
+      "step": 53380
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1.2163076236187075e-05,
+      "loss": 0.0471,
+      "step": 53390
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1.215972764817502e-05,
+      "loss": 0.1457,
+      "step": 53400
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1.2156379060162965e-05,
+      "loss": 0.0802,
+      "step": 53410
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.215303047215091e-05,
+      "loss": 0.0629,
+      "step": 53420
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.2149681884138855e-05,
+      "loss": 0.0522,
+      "step": 53430
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.21463332961268e-05,
+      "loss": 0.1443,
+      "step": 53440
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.2142984708114746e-05,
+      "loss": 0.0663,
+      "step": 53450
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.213963612010269e-05,
+      "loss": 0.0616,
+      "step": 53460
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.2136287532090634e-05,
+      "loss": 0.1107,
+      "step": 53470
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.213293894407858e-05,
+      "loss": 0.0827,
+      "step": 53480
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.2129590356066526e-05,
+      "loss": 0.0569,
+      "step": 53490
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.212624176805447e-05,
+      "loss": 0.1087,
+      "step": 53500
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.2122893180042416e-05,
+      "loss": 0.1308,
+      "step": 53510
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.2119544592030362e-05,
+      "loss": 0.0284,
+      "step": 53520
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.2116196004018306e-05,
+      "loss": 0.0882,
+      "step": 53530
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.211284741600625e-05,
+      "loss": 0.0648,
+      "step": 53540
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.2109498827994197e-05,
+      "loss": 0.1403,
+      "step": 53550
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.2106150239982141e-05,
+      "loss": 0.0568,
+      "step": 53560
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.2102801651970085e-05,
+      "loss": 0.0362,
+      "step": 53570
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.2099453063958031e-05,
+      "loss": 0.033,
+      "step": 53580
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.2096104475945977e-05,
+      "loss": 0.0816,
+      "step": 53590
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.2092755887933921e-05,
+      "loss": 0.1588,
+      "step": 53600
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.2089407299921865e-05,
+      "loss": 0.0964,
+      "step": 53610
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.2086058711909813e-05,
+      "loss": 0.0751,
+      "step": 53620
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.2082710123897757e-05,
+      "loss": 0.0743,
+      "step": 53630
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.20793615358857e-05,
+      "loss": 0.0196,
+      "step": 53640
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.2076012947873648e-05,
+      "loss": 0.0522,
+      "step": 53650
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.2072664359861592e-05,
+      "loss": 0.1488,
+      "step": 53660
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.2069315771849536e-05,
+      "loss": 0.0938,
+      "step": 53670
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.2065967183837484e-05,
+      "loss": 0.0794,
+      "step": 53680
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.2062618595825428e-05,
+      "loss": 0.1179,
+      "step": 53690
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.2059270007813372e-05,
+      "loss": 0.0994,
+      "step": 53700
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.2055921419801316e-05,
+      "loss": 0.0427,
+      "step": 53710
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.2052572831789263e-05,
+      "loss": 0.1125,
+      "step": 53720
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.2049224243777208e-05,
+      "loss": 0.0502,
+      "step": 53730
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.2045875655765152e-05,
+      "loss": 0.0701,
+      "step": 53740
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.2042527067753099e-05,
+      "loss": 0.0469,
+      "step": 53750
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.2039178479741043e-05,
+      "loss": 0.0615,
+      "step": 53760
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.2035829891728987e-05,
+      "loss": 0.0873,
+      "step": 53770
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.2032481303716933e-05,
+      "loss": 0.0589,
+      "step": 53780
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.2029132715704879e-05,
+      "loss": 0.058,
+      "step": 53790
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.2025784127692823e-05,
+      "loss": 0.3415,
+      "step": 53800
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.2022435539680769e-05,
+      "loss": 0.254,
+      "step": 53810
+    },
+    {
+      "epoch": 6.0,
+      "eval_loss": 0.2817317247390747,
+      "eval_runtime": 251.5149,
+      "eval_samples_per_second": 7.928,
+      "eval_steps_per_second": 3.964,
+      "eval_wer": 0.09715344654737673,
+      "step": 53814
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.2019086951668714e-05,
+      "loss": 0.1141,
+      "step": 53820
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.2015738363656658e-05,
+      "loss": 0.1025,
+      "step": 53830
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.2012389775644602e-05,
+      "loss": 0.0543,
+      "step": 53840
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.2009041187632548e-05,
+      "loss": 0.0674,
+      "step": 53850
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.2005692599620494e-05,
+      "loss": 0.0513,
+      "step": 53860
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.2002344011608438e-05,
+      "loss": 0.0539,
+      "step": 53870
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.1998995423596384e-05,
+      "loss": 0.0455,
+      "step": 53880
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.199564683558433e-05,
+      "loss": 0.0608,
+      "step": 53890
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.1992298247572274e-05,
+      "loss": 0.0352,
+      "step": 53900
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.198894965956022e-05,
+      "loss": 0.0437,
+      "step": 53910
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.1985601071548164e-05,
+      "loss": 0.0599,
+      "step": 53920
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.198225248353611e-05,
+      "loss": 0.0636,
+      "step": 53930
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.1978903895524055e-05,
+      "loss": 0.0905,
+      "step": 53940
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.1975555307512e-05,
+      "loss": 0.0479,
+      "step": 53950
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.1972206719499945e-05,
+      "loss": 0.1234,
+      "step": 53960
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.1968858131487889e-05,
+      "loss": 0.1287,
+      "step": 53970
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.1965509543475835e-05,
+      "loss": 0.0468,
+      "step": 53980
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.1962160955463779e-05,
+      "loss": 0.1147,
+      "step": 53990
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.1958812367451725e-05,
+      "loss": 0.0545,
+      "step": 54000
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.195546377943967e-05,
+      "loss": 0.0608,
+      "step": 54010
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.1952115191427614e-05,
+      "loss": 0.1188,
+      "step": 54020
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.194876660341556e-05,
+      "loss": 0.0552,
+      "step": 54030
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.1945418015403506e-05,
+      "loss": 0.2718,
+      "step": 54040
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.194206942739145e-05,
+      "loss": 0.0403,
+      "step": 54050
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.1938720839379394e-05,
+      "loss": 0.0329,
+      "step": 54060
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.1935372251367342e-05,
+      "loss": 0.0716,
+      "step": 54070
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.1932023663355286e-05,
+      "loss": 0.0475,
+      "step": 54080
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.192867507534323e-05,
+      "loss": 0.0773,
+      "step": 54090
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.1925326487331176e-05,
+      "loss": 0.0361,
+      "step": 54100
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.1921977899319121e-05,
+      "loss": 0.0503,
+      "step": 54110
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.1918629311307065e-05,
+      "loss": 0.0886,
+      "step": 54120
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.1915280723295011e-05,
+      "loss": 0.1362,
+      "step": 54130
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.1911932135282957e-05,
+      "loss": 0.2495,
+      "step": 54140
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.1908583547270901e-05,
+      "loss": 0.0489,
+      "step": 54150
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.1905234959258845e-05,
+      "loss": 0.0647,
+      "step": 54160
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.1901886371246793e-05,
+      "loss": 0.0783,
+      "step": 54170
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.1898537783234737e-05,
+      "loss": 0.0689,
+      "step": 54180
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.189518919522268e-05,
+      "loss": 0.0126,
+      "step": 54190
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.1891840607210626e-05,
+      "loss": 0.0461,
+      "step": 54200
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.1888492019198572e-05,
+      "loss": 0.0462,
+      "step": 54210
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.1885143431186516e-05,
+      "loss": 0.0738,
+      "step": 54220
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.188179484317446e-05,
+      "loss": 0.0359,
+      "step": 54230
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.1878446255162408e-05,
+      "loss": 0.1215,
+      "step": 54240
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.1875097667150352e-05,
+      "loss": 0.0465,
+      "step": 54250
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.1871749079138296e-05,
+      "loss": 0.019,
+      "step": 54260
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.1868400491126243e-05,
+      "loss": 0.0855,
+      "step": 54270
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.1865051903114187e-05,
+      "loss": 0.116,
+      "step": 54280
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.1861703315102132e-05,
+      "loss": 0.0692,
+      "step": 54290
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.1858354727090077e-05,
+      "loss": 0.0646,
+      "step": 54300
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.1855006139078023e-05,
+      "loss": 0.0822,
+      "step": 54310
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.1851657551065967e-05,
+      "loss": 0.1592,
+      "step": 54320
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.1848308963053911e-05,
+      "loss": 0.0752,
+      "step": 54330
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.1844960375041859e-05,
+      "loss": 0.0751,
+      "step": 54340
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.1841611787029803e-05,
+      "loss": 0.0723,
+      "step": 54350
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.1838263199017747e-05,
+      "loss": 0.0467,
+      "step": 54360
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.1834914611005693e-05,
+      "loss": 0.0899,
+      "step": 54370
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.1831566022993638e-05,
+      "loss": 0.1118,
+      "step": 54380
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.1828217434981582e-05,
+      "loss": 0.0618,
+      "step": 54390
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.1824868846969528e-05,
+      "loss": 0.1778,
+      "step": 54400
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.1821520258957474e-05,
+      "loss": 0.0629,
+      "step": 54410
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.1818171670945418e-05,
+      "loss": 0.0733,
+      "step": 54420
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.1814823082933364e-05,
+      "loss": 0.0622,
+      "step": 54430
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.1811474494921308e-05,
+      "loss": 0.0309,
+      "step": 54440
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.1808125906909254e-05,
+      "loss": 0.0239,
+      "step": 54450
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.1804777318897198e-05,
+      "loss": 0.051,
+      "step": 54460
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.1801428730885144e-05,
+      "loss": 0.0665,
+      "step": 54470
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.179808014287309e-05,
+      "loss": 0.0991,
+      "step": 54480
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.1794731554861033e-05,
+      "loss": 0.0365,
+      "step": 54490
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.1791382966848979e-05,
+      "loss": 0.0381,
+      "step": 54500
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.1788034378836923e-05,
+      "loss": 0.1032,
+      "step": 54510
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.1784685790824869e-05,
+      "loss": 0.1274,
+      "step": 54520
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.178167206161402e-05,
+      "loss": 0.0846,
+      "step": 54530
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.1778323473601965e-05,
+      "loss": 0.0773,
+      "step": 54540
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.177497488558991e-05,
+      "loss": 0.0783,
+      "step": 54550
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.1771626297577855e-05,
+      "loss": 0.0758,
+      "step": 54560
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.17682777095658e-05,
+      "loss": 0.0343,
+      "step": 54570
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.1764929121553745e-05,
+      "loss": 0.0386,
+      "step": 54580
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.1761580533541691e-05,
+      "loss": 0.0378,
+      "step": 54590
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.1758231945529635e-05,
+      "loss": 0.1309,
+      "step": 54600
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.175488335751758e-05,
+      "loss": 0.0731,
+      "step": 54610
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.1751534769505525e-05,
+      "loss": 0.1165,
+      "step": 54620
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.174818618149347e-05,
+      "loss": 0.0923,
+      "step": 54630
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.1744837593481416e-05,
+      "loss": 0.0532,
+      "step": 54640
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.174148900546936e-05,
+      "loss": 0.0826,
+      "step": 54650
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.1738140417457306e-05,
+      "loss": 0.0719,
+      "step": 54660
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.173479182944525e-05,
+      "loss": 0.0491,
+      "step": 54670
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.1731443241433196e-05,
+      "loss": 0.063,
+      "step": 54680
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.172809465342114e-05,
+      "loss": 0.108,
+      "step": 54690
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.1724746065409086e-05,
+      "loss": 0.0693,
+      "step": 54700
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.1721397477397032e-05,
+      "loss": 0.0531,
+      "step": 54710
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.1718048889384976e-05,
+      "loss": 0.0477,
+      "step": 54720
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.1714700301372921e-05,
+      "loss": 0.0674,
+      "step": 54730
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.1711351713360867e-05,
+      "loss": 0.1153,
+      "step": 54740
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.1708003125348811e-05,
+      "loss": 0.0316,
+      "step": 54750
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.1704654537336755e-05,
+      "loss": 0.17,
+      "step": 54760
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.1701305949324703e-05,
+      "loss": 0.0812,
+      "step": 54770
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.1697957361312647e-05,
+      "loss": 0.0773,
+      "step": 54780
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.1694608773300591e-05,
+      "loss": 0.0683,
+      "step": 54790
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.1691260185288537e-05,
+      "loss": 0.0605,
+      "step": 54800
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.1687911597276483e-05,
+      "loss": 0.0927,
+      "step": 54810
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.1684563009264427e-05,
+      "loss": 0.0972,
+      "step": 54820
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.1681214421252372e-05,
+      "loss": 0.0922,
+      "step": 54830
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.1677865833240318e-05,
+      "loss": 0.019,
+      "step": 54840
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.1674517245228262e-05,
+      "loss": 0.0284,
+      "step": 54850
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.1671168657216206e-05,
+      "loss": 0.0758,
+      "step": 54860
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.1667820069204154e-05,
+      "loss": 0.0802,
+      "step": 54870
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.1664471481192098e-05,
+      "loss": 0.1025,
+      "step": 54880
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.1661122893180042e-05,
+      "loss": 0.0219,
+      "step": 54890
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.165777430516799e-05,
+      "loss": 0.1329,
+      "step": 54900
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.1654425717155933e-05,
+      "loss": 0.0833,
+      "step": 54910
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.1651077129143878e-05,
+      "loss": 0.0704,
+      "step": 54920
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.1647728541131822e-05,
+      "loss": 0.0759,
+      "step": 54930
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.1644379953119769e-05,
+      "loss": 0.0172,
+      "step": 54940
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.1641031365107713e-05,
+      "loss": 0.0811,
+      "step": 54950
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.1637682777095657e-05,
+      "loss": 0.0402,
+      "step": 54960
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.1634334189083605e-05,
+      "loss": 0.0996,
+      "step": 54970
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.1630985601071549e-05,
+      "loss": 0.139,
+      "step": 54980
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.1627637013059493e-05,
+      "loss": 0.0543,
+      "step": 54990
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.1624288425047439e-05,
+      "loss": 0.0581,
+      "step": 55000
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.1620939837035384e-05,
+      "loss": 0.0427,
+      "step": 55010
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.1617591249023328e-05,
+      "loss": 0.1228,
+      "step": 55020
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.1614242661011274e-05,
+      "loss": 0.0123,
+      "step": 55030
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.161089407299922e-05,
+      "loss": 0.0369,
+      "step": 55040
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.1607545484987164e-05,
+      "loss": 0.0468,
+      "step": 55050
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.1604196896975108e-05,
+      "loss": 0.1636,
+      "step": 55060
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.1600848308963054e-05,
+      "loss": 0.1582,
+      "step": 55070
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.1597499720951e-05,
+      "loss": 0.0306,
+      "step": 55080
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.1594151132938944e-05,
+      "loss": 0.0935,
+      "step": 55090
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.159080254492689e-05,
+      "loss": 0.0433,
+      "step": 55100
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.1587453956914835e-05,
+      "loss": 0.0542,
+      "step": 55110
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.158410536890278e-05,
+      "loss": 0.105,
+      "step": 55120
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.1580756780890725e-05,
+      "loss": 0.074,
+      "step": 55130
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.157740819287867e-05,
+      "loss": 0.0518,
+      "step": 55140
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.1574059604866615e-05,
+      "loss": 0.1006,
+      "step": 55150
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.157071101685456e-05,
+      "loss": 0.1056,
+      "step": 55160
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.1567362428842505e-05,
+      "loss": 0.0704,
+      "step": 55170
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.156401384083045e-05,
+      "loss": 0.0851,
+      "step": 55180
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.1560665252818395e-05,
+      "loss": 0.0542,
+      "step": 55190
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.155731666480634e-05,
+      "loss": 0.0531,
+      "step": 55200
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.1553968076794284e-05,
+      "loss": 0.0819,
+      "step": 55210
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.155061948878223e-05,
+      "loss": 0.0468,
+      "step": 55220
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.1547270900770176e-05,
+      "loss": 0.0833,
+      "step": 55230
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.154392231275812e-05,
+      "loss": 0.0679,
+      "step": 55240
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.1540573724746066e-05,
+      "loss": 0.0257,
+      "step": 55250
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.1537225136734012e-05,
+      "loss": 0.0926,
+      "step": 55260
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.1533876548721956e-05,
+      "loss": 0.1126,
+      "step": 55270
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.15305279607099e-05,
+      "loss": 0.0959,
+      "step": 55280
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.1527179372697847e-05,
+      "loss": 0.051,
+      "step": 55290
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.1523830784685791e-05,
+      "loss": 0.0321,
+      "step": 55300
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.1520482196673735e-05,
+      "loss": 0.0994,
+      "step": 55310
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.1517133608661681e-05,
+      "loss": 0.0848,
+      "step": 55320
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.1513785020649627e-05,
+      "loss": 0.0801,
+      "step": 55330
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.1510436432637571e-05,
+      "loss": 0.0978,
+      "step": 55340
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.1507087844625517e-05,
+      "loss": 0.1239,
+      "step": 55350
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.1503739256613463e-05,
+      "loss": 0.0823,
+      "step": 55360
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.1500390668601407e-05,
+      "loss": 0.1014,
+      "step": 55370
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.149704208058935e-05,
+      "loss": 0.115,
+      "step": 55380
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.1493693492577298e-05,
+      "loss": 0.093,
+      "step": 55390
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.1490344904565242e-05,
+      "loss": 0.0518,
+      "step": 55400
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.1486996316553186e-05,
+      "loss": 0.0618,
+      "step": 55410
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.1483647728541132e-05,
+      "loss": 0.1154,
+      "step": 55420
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.1480299140529078e-05,
+      "loss": 0.0486,
+      "step": 55430
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.1476950552517022e-05,
+      "loss": 0.0633,
+      "step": 55440
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.1473601964504966e-05,
+      "loss": 0.0412,
+      "step": 55450
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.1470253376492913e-05,
+      "loss": 0.0921,
+      "step": 55460
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.1466904788480857e-05,
+      "loss": 0.0875,
+      "step": 55470
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.1463556200468802e-05,
+      "loss": 0.1552,
+      "step": 55480
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.1460207612456749e-05,
+      "loss": 0.0923,
+      "step": 55490
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.1456859024444693e-05,
+      "loss": 0.0628,
+      "step": 55500
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.1453510436432637e-05,
+      "loss": 0.0556,
+      "step": 55510
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.1450161848420583e-05,
+      "loss": 0.1636,
+      "step": 55520
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.1446813260408529e-05,
+      "loss": 0.0676,
+      "step": 55530
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.1443464672396473e-05,
+      "loss": 0.055,
+      "step": 55540
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.1440116084384417e-05,
+      "loss": 0.0593,
+      "step": 55550
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.1436767496372364e-05,
+      "loss": 0.0772,
+      "step": 55560
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.1433418908360308e-05,
+      "loss": 0.0762,
+      "step": 55570
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.1430070320348252e-05,
+      "loss": 0.0595,
+      "step": 55580
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.1426721732336198e-05,
+      "loss": 0.0451,
+      "step": 55590
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.1423373144324144e-05,
+      "loss": 0.052,
+      "step": 55600
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.1420024556312088e-05,
+      "loss": 0.0903,
+      "step": 55610
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.1416675968300034e-05,
+      "loss": 0.1853,
+      "step": 55620
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.141332738028798e-05,
+      "loss": 0.0443,
+      "step": 55630
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.1409978792275924e-05,
+      "loss": 0.1112,
+      "step": 55640
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.140663020426387e-05,
+      "loss": 0.0654,
+      "step": 55650
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.1403281616251814e-05,
+      "loss": 0.0587,
+      "step": 55660
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.139993302823976e-05,
+      "loss": 0.1441,
+      "step": 55670
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.1396584440227703e-05,
+      "loss": 0.1033,
+      "step": 55680
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.1393235852215649e-05,
+      "loss": 0.0822,
+      "step": 55690
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.1389887264203595e-05,
+      "loss": 0.1128,
+      "step": 55700
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.1386538676191539e-05,
+      "loss": 0.0328,
+      "step": 55710
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.1383190088179485e-05,
+      "loss": 0.0766,
+      "step": 55720
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.1379841500167429e-05,
+      "loss": 0.068,
+      "step": 55730
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.1376492912155375e-05,
+      "loss": 0.0736,
+      "step": 55740
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.137314432414332e-05,
+      "loss": 0.0491,
+      "step": 55750
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.1369795736131264e-05,
+      "loss": 0.0931,
+      "step": 55760
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.136644714811921e-05,
+      "loss": 0.101,
+      "step": 55770
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.1363098560107156e-05,
+      "loss": 0.1205,
+      "step": 55780
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.13597499720951e-05,
+      "loss": 0.0747,
+      "step": 55790
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.1356401384083046e-05,
+      "loss": 0.0281,
+      "step": 55800
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.135305279607099e-05,
+      "loss": 0.057,
+      "step": 55810
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.1349704208058936e-05,
+      "loss": 0.1665,
+      "step": 55820
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.134635562004688e-05,
+      "loss": 0.0554,
+      "step": 55830
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.1343007032034826e-05,
+      "loss": 0.0618,
+      "step": 55840
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.1339658444022771e-05,
+      "loss": 0.05,
+      "step": 55850
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.1336309856010715e-05,
+      "loss": 0.06,
+      "step": 55860
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.1332961267998661e-05,
+      "loss": 0.0601,
+      "step": 55870
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.1329612679986607e-05,
+      "loss": 0.0641,
+      "step": 55880
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.1326264091974551e-05,
+      "loss": 0.0516,
+      "step": 55890
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.1322915503962495e-05,
+      "loss": 0.0614,
+      "step": 55900
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.1319566915950442e-05,
+      "loss": 0.0447,
+      "step": 55910
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.1316218327938387e-05,
+      "loss": 0.0867,
+      "step": 55920
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.131286973992633e-05,
+      "loss": 0.0687,
+      "step": 55930
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.1309521151914276e-05,
+      "loss": 0.0752,
+      "step": 55940
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.1306172563902222e-05,
+      "loss": 0.0693,
+      "step": 55950
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.1302823975890166e-05,
+      "loss": 0.1236,
+      "step": 55960
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.129947538787811e-05,
+      "loss": 0.0714,
+      "step": 55970
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.1296126799866058e-05,
+      "loss": 0.0347,
+      "step": 55980
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.1292778211854002e-05,
+      "loss": 0.0656,
+      "step": 55990
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.1289429623841946e-05,
+      "loss": 0.0767,
+      "step": 56000
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.1286081035829893e-05,
+      "loss": 0.1439,
+      "step": 56010
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.1282732447817837e-05,
+      "loss": 0.0906,
+      "step": 56020
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.1279383859805782e-05,
+      "loss": 0.041,
+      "step": 56030
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.1276035271793727e-05,
+      "loss": 0.0611,
+      "step": 56040
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.1272686683781673e-05,
+      "loss": 0.114,
+      "step": 56050
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.1269338095769617e-05,
+      "loss": 0.1078,
+      "step": 56060
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.1265989507757561e-05,
+      "loss": 0.064,
+      "step": 56070
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.1262640919745509e-05,
+      "loss": 0.1556,
+      "step": 56080
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.1259292331733453e-05,
+      "loss": 0.1289,
+      "step": 56090
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.1255943743721397e-05,
+      "loss": 0.0508,
+      "step": 56100
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.1252595155709343e-05,
+      "loss": 0.1272,
+      "step": 56110
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.1249246567697288e-05,
+      "loss": 0.0567,
+      "step": 56120
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.1245897979685232e-05,
+      "loss": 0.0733,
+      "step": 56130
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.1242549391673178e-05,
+      "loss": 0.0872,
+      "step": 56140
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.1239200803661124e-05,
+      "loss": 0.2223,
+      "step": 56150
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.1235852215649068e-05,
+      "loss": 0.1249,
+      "step": 56160
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.1232503627637014e-05,
+      "loss": 0.0338,
+      "step": 56170
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.1229155039624958e-05,
+      "loss": 0.0366,
+      "step": 56180
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.1225806451612904e-05,
+      "loss": 0.0387,
+      "step": 56190
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.1222457863600848e-05,
+      "loss": 0.0411,
+      "step": 56200
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.1219109275588794e-05,
+      "loss": 0.0655,
+      "step": 56210
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.121576068757674e-05,
+      "loss": 0.108,
+      "step": 56220
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.1212412099564683e-05,
+      "loss": 0.051,
+      "step": 56230
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.1209063511552629e-05,
+      "loss": 0.0645,
+      "step": 56240
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.1205714923540575e-05,
+      "loss": 0.0711,
+      "step": 56250
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.1202366335528519e-05,
+      "loss": 0.0679,
+      "step": 56260
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.1199017747516465e-05,
+      "loss": 0.0555,
+      "step": 56270
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.1195669159504409e-05,
+      "loss": 0.0844,
+      "step": 56280
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.1192320571492355e-05,
+      "loss": 0.1092,
+      "step": 56290
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.1188971983480299e-05,
+      "loss": 0.0536,
+      "step": 56300
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.1185623395468244e-05,
+      "loss": 0.0476,
+      "step": 56310
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.118227480745619e-05,
+      "loss": 0.0702,
+      "step": 56320
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.1178926219444134e-05,
+      "loss": 0.0364,
+      "step": 56330
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.117557763143208e-05,
+      "loss": 0.072,
+      "step": 56340
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.1172229043420024e-05,
+      "loss": 0.0684,
+      "step": 56350
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.116888045540797e-05,
+      "loss": 0.0987,
+      "step": 56360
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.1165531867395916e-05,
+      "loss": 0.0962,
+      "step": 56370
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.116218327938386e-05,
+      "loss": 0.0971,
+      "step": 56380
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.1158834691371805e-05,
+      "loss": 0.079,
+      "step": 56390
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.1155486103359751e-05,
+      "loss": 0.034,
+      "step": 56400
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.1152137515347695e-05,
+      "loss": 0.0979,
+      "step": 56410
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.114878892733564e-05,
+      "loss": 0.0909,
+      "step": 56420
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.1145440339323585e-05,
+      "loss": 0.1255,
+      "step": 56430
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.1142091751311531e-05,
+      "loss": 0.0795,
+      "step": 56440
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.1138743163299475e-05,
+      "loss": 0.0785,
+      "step": 56450
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.113539457528742e-05,
+      "loss": 0.1017,
+      "step": 56460
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.1132045987275367e-05,
+      "loss": 0.0901,
+      "step": 56470
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.112869739926331e-05,
+      "loss": 0.0615,
+      "step": 56480
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.1125348811251255e-05,
+      "loss": 0.1114,
+      "step": 56490
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.1122000223239202e-05,
+      "loss": 0.0466,
+      "step": 56500
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.1118651635227146e-05,
+      "loss": 0.0827,
+      "step": 56510
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.111530304721509e-05,
+      "loss": 0.0736,
+      "step": 56520
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.1111954459203038e-05,
+      "loss": 0.025,
+      "step": 56530
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.1108605871190982e-05,
+      "loss": 0.0608,
+      "step": 56540
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.1105257283178926e-05,
+      "loss": 0.0268,
+      "step": 56550
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.110190869516687e-05,
+      "loss": 0.0943,
+      "step": 56560
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.1098560107154817e-05,
+      "loss": 0.0492,
+      "step": 56570
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.1095211519142762e-05,
+      "loss": 0.0553,
+      "step": 56580
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.1091862931130706e-05,
+      "loss": 0.0572,
+      "step": 56590
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.1088514343118653e-05,
+      "loss": 0.0901,
+      "step": 56600
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.1085165755106597e-05,
+      "loss": 0.1149,
+      "step": 56610
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.1081817167094541e-05,
+      "loss": 0.1992,
+      "step": 56620
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.1078468579082487e-05,
+      "loss": 0.0689,
+      "step": 56630
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.1075119991070433e-05,
+      "loss": 0.0661,
+      "step": 56640
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.1071771403058377e-05,
+      "loss": 0.0431,
+      "step": 56650
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.1068422815046323e-05,
+      "loss": 0.0214,
+      "step": 56660
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.1065074227034268e-05,
+      "loss": 0.051,
+      "step": 56670
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.1061725639022212e-05,
+      "loss": 0.1725,
+      "step": 56680
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.1058377051010157e-05,
+      "loss": 0.0817,
+      "step": 56690
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.1055028462998104e-05,
+      "loss": 0.105,
+      "step": 56700
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.1051679874986048e-05,
+      "loss": 0.1115,
+      "step": 56710
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.1048331286973992e-05,
+      "loss": 0.1249,
+      "step": 56720
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.1044982698961938e-05,
+      "loss": 0.0565,
+      "step": 56730
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.1041634110949884e-05,
+      "loss": 0.0848,
+      "step": 56740
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.1038285522937828e-05,
+      "loss": 0.0493,
+      "step": 56750
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.1034936934925773e-05,
+      "loss": 0.0846,
+      "step": 56760
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.103158834691372e-05,
+      "loss": 0.1257,
+      "step": 56770
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.1028239758901663e-05,
+      "loss": 0.0595,
+      "step": 56780
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.1024891170889609e-05,
+      "loss": 0.0717,
+      "step": 56790
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.1021542582877553e-05,
+      "loss": 0.0521,
+      "step": 56800
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.1018193994865499e-05,
+      "loss": 0.0578,
+      "step": 56810
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.1014845406853443e-05,
+      "loss": 0.1236,
+      "step": 56820
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.1011496818841389e-05,
+      "loss": 0.1296,
+      "step": 56830
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.1008148230829335e-05,
+      "loss": 0.0657,
+      "step": 56840
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.1004799642817279e-05,
+      "loss": 0.055,
+      "step": 56850
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.1001451054805224e-05,
+      "loss": 0.079,
+      "step": 56860
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.0998102466793168e-05,
+      "loss": 0.0294,
+      "step": 56870
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.0994753878781114e-05,
+      "loss": 0.0576,
+      "step": 56880
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.099140529076906e-05,
+      "loss": 0.0528,
+      "step": 56890
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.0988056702757004e-05,
+      "loss": 0.0784,
+      "step": 56900
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.098470811474495e-05,
+      "loss": 0.0874,
+      "step": 56910
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.0981359526732896e-05,
+      "loss": 0.0436,
+      "step": 56920
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.097801093872084e-05,
+      "loss": 0.044,
+      "step": 56930
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.0974662350708784e-05,
+      "loss": 0.03,
+      "step": 56940
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.097131376269673e-05,
+      "loss": 0.0716,
+      "step": 56950
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.0967965174684675e-05,
+      "loss": 0.0705,
+      "step": 56960
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.096461658667262e-05,
+      "loss": 0.0796,
+      "step": 56970
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.0961267998660565e-05,
+      "loss": 0.1128,
+      "step": 56980
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.0957919410648511e-05,
+      "loss": 0.0354,
+      "step": 56990
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.0954570822636455e-05,
+      "loss": 0.1542,
+      "step": 57000
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.0951222234624399e-05,
+      "loss": 0.1401,
+      "step": 57010
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.0947873646612347e-05,
+      "loss": 0.0478,
+      "step": 57020
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.094452505860029e-05,
+      "loss": 0.1379,
+      "step": 57030
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.0941176470588235e-05,
+      "loss": 0.0898,
+      "step": 57040
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.093782788257618e-05,
+      "loss": 0.0724,
+      "step": 57050
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.0934479294564126e-05,
+      "loss": 0.1642,
+      "step": 57060
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.093113070655207e-05,
+      "loss": 0.1028,
+      "step": 57070
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.0927782118540014e-05,
+      "loss": 0.0978,
+      "step": 57080
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.0924433530527962e-05,
+      "loss": 0.0588,
+      "step": 57090
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.0921084942515906e-05,
+      "loss": 0.0248,
+      "step": 57100
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.091773635450385e-05,
+      "loss": 0.0812,
+      "step": 57110
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.0914387766491797e-05,
+      "loss": 0.0383,
+      "step": 57120
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.0911039178479742e-05,
+      "loss": 0.0532,
+      "step": 57130
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.0907690590467686e-05,
+      "loss": 0.0906,
+      "step": 57140
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.0904342002455633e-05,
+      "loss": 0.1349,
+      "step": 57150
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.0900993414443577e-05,
+      "loss": 0.0494,
+      "step": 57160
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.0897644826431521e-05,
+      "loss": 0.0822,
+      "step": 57170
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.0894296238419465e-05,
+      "loss": 0.0227,
+      "step": 57180
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.0890947650407413e-05,
+      "loss": 0.0469,
+      "step": 57190
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.0887599062395357e-05,
+      "loss": 0.0517,
+      "step": 57200
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.0884250474383301e-05,
+      "loss": 0.0852,
+      "step": 57210
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.0880901886371248e-05,
+      "loss": 0.1293,
+      "step": 57220
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.0877553298359192e-05,
+      "loss": 0.0637,
+      "step": 57230
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.0874204710347136e-05,
+      "loss": 0.121,
+      "step": 57240
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.0870856122335082e-05,
+      "loss": 0.0722,
+      "step": 57250
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.0867507534323028e-05,
+      "loss": 0.0736,
+      "step": 57260
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.0864158946310972e-05,
+      "loss": 0.0556,
+      "step": 57270
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.0860810358298918e-05,
+      "loss": 0.0425,
+      "step": 57280
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.0857461770286864e-05,
+      "loss": 0.0173,
+      "step": 57290
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.0854448041076014e-05,
+      "loss": 0.0363,
+      "step": 57300
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.0851099453063958e-05,
+      "loss": 0.0896,
+      "step": 57310
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.0847750865051902e-05,
+      "loss": 0.1063,
+      "step": 57320
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.0844402277039848e-05,
+      "loss": 0.0361,
+      "step": 57330
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.0841053689027794e-05,
+      "loss": 0.0461,
+      "step": 57340
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.0837705101015738e-05,
+      "loss": 0.0843,
+      "step": 57350
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.0834356513003684e-05,
+      "loss": 0.0814,
+      "step": 57360
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.083100792499163e-05,
+      "loss": 0.0864,
+      "step": 57370
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.0827659336979574e-05,
+      "loss": 0.0874,
+      "step": 57380
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.0824310748967518e-05,
+      "loss": 0.1099,
+      "step": 57390
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.0820962160955465e-05,
+      "loss": 0.0448,
+      "step": 57400
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.081761357294341e-05,
+      "loss": 0.1007,
+      "step": 57410
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.0814264984931353e-05,
+      "loss": 0.131,
+      "step": 57420
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.0810916396919299e-05,
+      "loss": 0.0275,
+      "step": 57430
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.0807567808907245e-05,
+      "loss": 0.0864,
+      "step": 57440
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.0804219220895189e-05,
+      "loss": 0.0146,
+      "step": 57450
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.0800870632883135e-05,
+      "loss": 0.0851,
+      "step": 57460
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.079752204487108e-05,
+      "loss": 0.0925,
+      "step": 57470
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.0794173456859025e-05,
+      "loss": 0.0488,
+      "step": 57480
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.079082486884697e-05,
+      "loss": 0.0578,
+      "step": 57490
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.0787476280834914e-05,
+      "loss": 0.0269,
+      "step": 57500
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.078412769282286e-05,
+      "loss": 0.1623,
+      "step": 57510
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.0780779104810804e-05,
+      "loss": 0.1025,
+      "step": 57520
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.077743051679875e-05,
+      "loss": 0.1227,
+      "step": 57530
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.0774081928786696e-05,
+      "loss": 0.1841,
+      "step": 57540
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.077073334077464e-05,
+      "loss": 0.0919,
+      "step": 57550
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.0767384752762586e-05,
+      "loss": 0.0808,
+      "step": 57560
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.076403616475053e-05,
+      "loss": 0.0548,
+      "step": 57570
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.0760687576738475e-05,
+      "loss": 0.0233,
+      "step": 57580
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.0757338988726421e-05,
+      "loss": 0.0492,
+      "step": 57590
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.0753990400714365e-05,
+      "loss": 0.0777,
+      "step": 57600
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.0750641812702311e-05,
+      "loss": 0.1454,
+      "step": 57610
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.0747293224690257e-05,
+      "loss": 0.0429,
+      "step": 57620
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.0743944636678201e-05,
+      "loss": 0.0737,
+      "step": 57630
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.0740596048666145e-05,
+      "loss": 0.1297,
+      "step": 57640
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.073724746065409e-05,
+      "loss": 0.0897,
+      "step": 57650
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.0733898872642037e-05,
+      "loss": 0.1127,
+      "step": 57660
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.073055028462998e-05,
+      "loss": 0.0439,
+      "step": 57670
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.0727201696617926e-05,
+      "loss": 0.1001,
+      "step": 57680
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.0723853108605872e-05,
+      "loss": 0.0249,
+      "step": 57690
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.0720504520593816e-05,
+      "loss": 0.127,
+      "step": 57700
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.071715593258176e-05,
+      "loss": 0.0461,
+      "step": 57710
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.0713807344569708e-05,
+      "loss": 0.026,
+      "step": 57720
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.0710458756557652e-05,
+      "loss": 0.048,
+      "step": 57730
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.0707110168545596e-05,
+      "loss": 0.0761,
+      "step": 57740
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.0703761580533543e-05,
+      "loss": 0.0248,
+      "step": 57750
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.0700412992521487e-05,
+      "loss": 0.0454,
+      "step": 57760
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.0697064404509432e-05,
+      "loss": 0.127,
+      "step": 57770
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.0693715816497376e-05,
+      "loss": 0.1391,
+      "step": 57780
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.0690367228485323e-05,
+      "loss": 0.0511,
+      "step": 57790
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.0687018640473267e-05,
+      "loss": 0.0644,
+      "step": 57800
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.0683670052461211e-05,
+      "loss": 0.0861,
+      "step": 57810
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.0680321464449159e-05,
+      "loss": 0.0519,
+      "step": 57820
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.0676972876437103e-05,
+      "loss": 0.07,
+      "step": 57830
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.0673624288425047e-05,
+      "loss": 0.0583,
+      "step": 57840
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.0670275700412994e-05,
+      "loss": 0.0682,
+      "step": 57850
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.0666927112400938e-05,
+      "loss": 0.1137,
+      "step": 57860
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.0663578524388882e-05,
+      "loss": 0.0742,
+      "step": 57870
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.0660229936376828e-05,
+      "loss": 0.104,
+      "step": 57880
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.0656881348364774e-05,
+      "loss": 0.1105,
+      "step": 57890
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.0653532760352718e-05,
+      "loss": 0.0756,
+      "step": 57900
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.0650184172340662e-05,
+      "loss": 0.0821,
+      "step": 57910
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.064683558432861e-05,
+      "loss": 0.0925,
+      "step": 57920
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.0643486996316554e-05,
+      "loss": 0.0523,
+      "step": 57930
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.0640138408304498e-05,
+      "loss": 0.1184,
+      "step": 57940
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.0636789820292444e-05,
+      "loss": 0.0733,
+      "step": 57950
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.063344123228039e-05,
+      "loss": 0.0851,
+      "step": 57960
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.0630092644268333e-05,
+      "loss": 0.1585,
+      "step": 57970
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.0627078915057484e-05,
+      "loss": 0.0497,
+      "step": 57980
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.0623730327045428e-05,
+      "loss": 0.0375,
+      "step": 57990
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.0620381739033376e-05,
+      "loss": 0.0468,
+      "step": 58000
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.061703315102132e-05,
+      "loss": 0.1208,
+      "step": 58010
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.0613684563009264e-05,
+      "loss": 0.097,
+      "step": 58020
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.061033597499721e-05,
+      "loss": 0.0792,
+      "step": 58030
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.0606987386985155e-05,
+      "loss": 0.1158,
+      "step": 58040
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.06036387989731e-05,
+      "loss": 0.0939,
+      "step": 58050
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.0600290210961045e-05,
+      "loss": 0.0417,
+      "step": 58060
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.0596941622948991e-05,
+      "loss": 0.131,
+      "step": 58070
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.0593593034936935e-05,
+      "loss": 0.0565,
+      "step": 58080
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.059024444692488e-05,
+      "loss": 0.0653,
+      "step": 58090
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.0586895858912826e-05,
+      "loss": 0.0427,
+      "step": 58100
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.058354727090077e-05,
+      "loss": 0.0403,
+      "step": 58110
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.0580198682888715e-05,
+      "loss": 0.112,
+      "step": 58120
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.057685009487666e-05,
+      "loss": 0.1061,
+      "step": 58130
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.0573501506864606e-05,
+      "loss": 0.0451,
+      "step": 58140
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.057015291885255e-05,
+      "loss": 0.0218,
+      "step": 58150
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.0566804330840496e-05,
+      "loss": 0.0456,
+      "step": 58160
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.0563455742828442e-05,
+      "loss": 0.091,
+      "step": 58170
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.0560107154816386e-05,
+      "loss": 0.0566,
+      "step": 58180
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.0556758566804332e-05,
+      "loss": 0.0706,
+      "step": 58190
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.0553409978792276e-05,
+      "loss": 0.0449,
+      "step": 58200
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.0550061390780221e-05,
+      "loss": 0.0535,
+      "step": 58210
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.0546712802768167e-05,
+      "loss": 0.0582,
+      "step": 58220
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.0543364214756111e-05,
+      "loss": 0.0713,
+      "step": 58230
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.0540015626744057e-05,
+      "loss": 0.0579,
+      "step": 58240
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.0536667038732001e-05,
+      "loss": 0.0613,
+      "step": 58250
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.0533318450719947e-05,
+      "loss": 0.044,
+      "step": 58260
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.0529969862707891e-05,
+      "loss": 0.1459,
+      "step": 58270
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.0526621274695837e-05,
+      "loss": 0.0825,
+      "step": 58280
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.0523272686683783e-05,
+      "loss": 0.0428,
+      "step": 58290
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.0519924098671727e-05,
+      "loss": 0.102,
+      "step": 58300
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.0516575510659672e-05,
+      "loss": 0.0411,
+      "step": 58310
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.0513226922647618e-05,
+      "loss": 0.123,
+      "step": 58320
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.0509878334635562e-05,
+      "loss": 0.0668,
+      "step": 58330
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.0506529746623506e-05,
+      "loss": 0.1614,
+      "step": 58340
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.0503181158611454e-05,
+      "loss": 0.0481,
+      "step": 58350
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.0499832570599398e-05,
+      "loss": 0.121,
+      "step": 58360
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.0496483982587342e-05,
+      "loss": 0.0563,
+      "step": 58370
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.0493135394575288e-05,
+      "loss": 0.0662,
+      "step": 58380
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.0489786806563233e-05,
+      "loss": 0.0985,
+      "step": 58390
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.0486438218551177e-05,
+      "loss": 0.1849,
+      "step": 58400
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.0483089630539122e-05,
+      "loss": 0.0514,
+      "step": 58410
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.0479741042527069e-05,
+      "loss": 0.1077,
+      "step": 58420
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.0476392454515013e-05,
+      "loss": 0.0526,
+      "step": 58430
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.0473043866502957e-05,
+      "loss": 0.1193,
+      "step": 58440
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.0469695278490905e-05,
+      "loss": 0.0731,
+      "step": 58450
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.0466346690478849e-05,
+      "loss": 0.0632,
+      "step": 58460
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.0462998102466793e-05,
+      "loss": 0.036,
+      "step": 58470
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.0459649514454739e-05,
+      "loss": 0.0734,
+      "step": 58480
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.0456300926442684e-05,
+      "loss": 0.1366,
+      "step": 58490
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.0452952338430628e-05,
+      "loss": 0.0341,
+      "step": 58500
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.0449603750418572e-05,
+      "loss": 0.0293,
+      "step": 58510
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.044625516240652e-05,
+      "loss": 0.1374,
+      "step": 58520
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.0442906574394464e-05,
+      "loss": 0.0349,
+      "step": 58530
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.0439557986382408e-05,
+      "loss": 0.0648,
+      "step": 58540
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.0436209398370356e-05,
+      "loss": 0.1181,
+      "step": 58550
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.04328608103583e-05,
+      "loss": 0.0781,
+      "step": 58560
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.0429512222346244e-05,
+      "loss": 0.1113,
+      "step": 58570
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.042616363433419e-05,
+      "loss": 0.0945,
+      "step": 58580
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.0422815046322135e-05,
+      "loss": 0.0417,
+      "step": 58590
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.041946645831008e-05,
+      "loss": 0.0595,
+      "step": 58600
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.0416117870298023e-05,
+      "loss": 0.0499,
+      "step": 58610
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.041276928228597e-05,
+      "loss": 0.1312,
+      "step": 58620
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.0409420694273915e-05,
+      "loss": 0.1934,
+      "step": 58630
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.0406072106261859e-05,
+      "loss": 0.0569,
+      "step": 58640
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.0402723518249805e-05,
+      "loss": 0.0647,
+      "step": 58650
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.039937493023775e-05,
+      "loss": 0.0721,
+      "step": 58660
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.0396026342225695e-05,
+      "loss": 0.113,
+      "step": 58670
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.039267775421364e-05,
+      "loss": 0.1586,
+      "step": 58680
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.0389329166201586e-05,
+      "loss": 0.0418,
+      "step": 58690
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.038598057818953e-05,
+      "loss": 0.0797,
+      "step": 58700
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.0382631990177476e-05,
+      "loss": 0.075,
+      "step": 58710
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.037928340216542e-05,
+      "loss": 0.03,
+      "step": 58720
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.0375934814153366e-05,
+      "loss": 0.1696,
+      "step": 58730
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.037258622614131e-05,
+      "loss": 0.0775,
+      "step": 58740
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.0369237638129256e-05,
+      "loss": 0.1142,
+      "step": 58750
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.0365889050117201e-05,
+      "loss": 0.0553,
+      "step": 58760
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.0362540462105146e-05,
+      "loss": 0.1828,
+      "step": 58770
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.0359191874093091e-05,
+      "loss": 0.0534,
+      "step": 58780
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.0355843286081035e-05,
+      "loss": 0.0625,
+      "step": 58790
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.0352494698068981e-05,
+      "loss": 0.0535,
+      "step": 58800
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.0349146110056927e-05,
+      "loss": 0.0897,
+      "step": 58810
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.0345797522044871e-05,
+      "loss": 0.1138,
+      "step": 58820
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.0342448934032817e-05,
+      "loss": 0.0598,
+      "step": 58830
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.0339100346020762e-05,
+      "loss": 0.0548,
+      "step": 58840
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.0335751758008707e-05,
+      "loss": 0.1195,
+      "step": 58850
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.033240316999665e-05,
+      "loss": 0.1322,
+      "step": 58860
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.0329054581984596e-05,
+      "loss": 0.1079,
+      "step": 58870
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.0325705993972542e-05,
+      "loss": 0.0419,
+      "step": 58880
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.0322357405960486e-05,
+      "loss": 0.069,
+      "step": 58890
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.0319008817948432e-05,
+      "loss": 0.0725,
+      "step": 58900
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.0315660229936378e-05,
+      "loss": 0.1007,
+      "step": 58910
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.0312311641924322e-05,
+      "loss": 0.046,
+      "step": 58920
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.0308963053912266e-05,
+      "loss": 0.0517,
+      "step": 58930
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.0305614465900213e-05,
+      "loss": 0.0623,
+      "step": 58940
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.0302265877888157e-05,
+      "loss": 0.0451,
+      "step": 58950
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.0298917289876102e-05,
+      "loss": 0.1061,
+      "step": 58960
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.0295568701864049e-05,
+      "loss": 0.1384,
+      "step": 58970
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.0292220113851993e-05,
+      "loss": 0.0309,
+      "step": 58980
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.0288871525839937e-05,
+      "loss": 0.0596,
+      "step": 58990
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.0285522937827883e-05,
+      "loss": 0.0964,
+      "step": 59000
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.0282174349815829e-05,
+      "loss": 0.0489,
+      "step": 59010
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.0278825761803773e-05,
+      "loss": 0.124,
+      "step": 59020
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.0275477173791717e-05,
+      "loss": 0.0751,
+      "step": 59030
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.0272128585779664e-05,
+      "loss": 0.0888,
+      "step": 59040
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.0268779997767608e-05,
+      "loss": 0.073,
+      "step": 59050
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.0265431409755552e-05,
+      "loss": 0.0503,
+      "step": 59060
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.02620828217435e-05,
+      "loss": 0.0537,
+      "step": 59070
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.0258734233731444e-05,
+      "loss": 0.0448,
+      "step": 59080
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.0255385645719388e-05,
+      "loss": 0.0375,
+      "step": 59090
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.0252037057707334e-05,
+      "loss": 0.0459,
+      "step": 59100
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.024868846969528e-05,
+      "loss": 0.0522,
+      "step": 59110
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.0245339881683224e-05,
+      "loss": 0.1171,
+      "step": 59120
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.0241991293671168e-05,
+      "loss": 0.122,
+      "step": 59130
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.0238642705659115e-05,
+      "loss": 0.0419,
+      "step": 59140
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.023529411764706e-05,
+      "loss": 0.0426,
+      "step": 59150
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.0231945529635003e-05,
+      "loss": 0.0989,
+      "step": 59160
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.0228596941622949e-05,
+      "loss": 0.0443,
+      "step": 59170
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.0225248353610895e-05,
+      "loss": 0.3222,
+      "step": 59180
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.0221899765598839e-05,
+      "loss": 0.0833,
+      "step": 59190
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.0218551177586785e-05,
+      "loss": 0.0498,
+      "step": 59200
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.021520258957473e-05,
+      "loss": 0.0697,
+      "step": 59210
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.0211854001562675e-05,
+      "loss": 0.0505,
+      "step": 59220
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.020850541355062e-05,
+      "loss": 0.1023,
+      "step": 59230
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.0205156825538564e-05,
+      "loss": 0.0498,
+      "step": 59240
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.020180823752651e-05,
+      "loss": 0.0433,
+      "step": 59250
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.0198459649514454e-05,
+      "loss": 0.1255,
+      "step": 59260
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.01951110615024e-05,
+      "loss": 0.095,
+      "step": 59270
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.0191762473490346e-05,
+      "loss": 0.1839,
+      "step": 59280
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.018841388547829e-05,
+      "loss": 0.1581,
+      "step": 59290
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.0185065297466236e-05,
+      "loss": 0.1158,
+      "step": 59300
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.018171670945418e-05,
+      "loss": 0.0742,
+      "step": 59310
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.0178368121442125e-05,
+      "loss": 0.29,
+      "step": 59320
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.0175019533430071e-05,
+      "loss": 0.0277,
+      "step": 59330
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.0171670945418015e-05,
+      "loss": 0.1224,
+      "step": 59340
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.0168322357405961e-05,
+      "loss": 0.0663,
+      "step": 59350
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.0164973769393905e-05,
+      "loss": 0.0344,
+      "step": 59360
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.0161625181381851e-05,
+      "loss": 0.0609,
+      "step": 59370
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.0158276593369795e-05,
+      "loss": 0.0829,
+      "step": 59380
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.015492800535774e-05,
+      "loss": 0.0893,
+      "step": 59390
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.0151579417345687e-05,
+      "loss": 0.1087,
+      "step": 59400
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.014823082933363e-05,
+      "loss": 0.0312,
+      "step": 59410
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.0144882241321576e-05,
+      "loss": 0.0689,
+      "step": 59420
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.0141533653309522e-05,
+      "loss": 0.1066,
+      "step": 59430
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.0138185065297466e-05,
+      "loss": 0.0986,
+      "step": 59440
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.0134836477285412e-05,
+      "loss": 0.0859,
+      "step": 59450
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.0131487889273358e-05,
+      "loss": 0.0821,
+      "step": 59460
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.0128139301261302e-05,
+      "loss": 0.1305,
+      "step": 59470
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.0124790713249246e-05,
+      "loss": 0.0503,
+      "step": 59480
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.0121442125237192e-05,
+      "loss": 0.2804,
+      "step": 59490
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.0118093537225137e-05,
+      "loss": 0.0997,
+      "step": 59500
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.0114744949213082e-05,
+      "loss": 0.1509,
+      "step": 59510
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.0111396361201027e-05,
+      "loss": 0.0916,
+      "step": 59520
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.0108047773188973e-05,
+      "loss": 0.0603,
+      "step": 59530
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.0104699185176917e-05,
+      "loss": 0.1426,
+      "step": 59540
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.0101350597164861e-05,
+      "loss": 0.0958,
+      "step": 59550
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.0098002009152809e-05,
+      "loss": 0.0796,
+      "step": 59560
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.0094653421140753e-05,
+      "loss": 0.106,
+      "step": 59570
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.0091304833128697e-05,
+      "loss": 0.0666,
+      "step": 59580
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.0087956245116644e-05,
+      "loss": 0.0803,
+      "step": 59590
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.0084607657104588e-05,
+      "loss": 0.0824,
+      "step": 59600
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.0081259069092532e-05,
+      "loss": 0.0536,
+      "step": 59610
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.0077910481080477e-05,
+      "loss": 0.1145,
+      "step": 59620
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.0074561893068424e-05,
+      "loss": 0.0605,
+      "step": 59630
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.0071213305056368e-05,
+      "loss": 0.0892,
+      "step": 59640
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.0067864717044312e-05,
+      "loss": 0.0563,
+      "step": 59650
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.006451612903226e-05,
+      "loss": 0.2871,
+      "step": 59660
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.0061167541020204e-05,
+      "loss": 0.1113,
+      "step": 59670
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.0057818953008148e-05,
+      "loss": 0.047,
+      "step": 59680
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.0054470364996093e-05,
+      "loss": 0.1396,
+      "step": 59690
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.005112177698404e-05,
+      "loss": 0.0294,
+      "step": 59700
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.0047773188971983e-05,
+      "loss": 0.0406,
+      "step": 59710
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.0044424600959929e-05,
+      "loss": 0.1161,
+      "step": 59720
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.0041076012947875e-05,
+      "loss": 0.0288,
+      "step": 59730
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.0037727424935819e-05,
+      "loss": 0.0497,
+      "step": 59740
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.0034378836923763e-05,
+      "loss": 0.1615,
+      "step": 59750
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.0031030248911709e-05,
+      "loss": 0.0706,
+      "step": 59760
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.0027681660899655e-05,
+      "loss": 0.1165,
+      "step": 59770
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.0024333072887599e-05,
+      "loss": 0.0973,
+      "step": 59780
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.0020984484875544e-05,
+      "loss": 0.1075,
+      "step": 59790
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.001763589686349e-05,
+      "loss": 0.071,
+      "step": 59800
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.0014287308851434e-05,
+      "loss": 0.0667,
+      "step": 59810
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.001093872083938e-05,
+      "loss": 0.1013,
+      "step": 59820
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.0007590132827324e-05,
+      "loss": 0.1103,
+      "step": 59830
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.000424154481527e-05,
+      "loss": 0.133,
+      "step": 59840
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.0000892956803216e-05,
+      "loss": 0.033,
+      "step": 59850
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 9.99754436879116e-06,
+      "loss": 0.0441,
+      "step": 59860
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 9.994195780779105e-06,
+      "loss": 0.0669,
+      "step": 59870
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 9.99084719276705e-06,
+      "loss": 0.0371,
+      "step": 59880
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 9.987498604754995e-06,
+      "loss": 0.0946,
+      "step": 59890
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 9.984150016742941e-06,
+      "loss": 0.052,
+      "step": 59900
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 9.980801428730885e-06,
+      "loss": 0.1893,
+      "step": 59910
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 9.977452840718831e-06,
+      "loss": 0.0924,
+      "step": 59920
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 9.974104252706775e-06,
+      "loss": 0.0391,
+      "step": 59930
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 9.97075566469472e-06,
+      "loss": 0.1758,
+      "step": 59940
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 9.967407076682667e-06,
+      "loss": 0.0741,
+      "step": 59950
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 9.96405848867061e-06,
+      "loss": 0.0612,
+      "step": 59960
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 9.960709900658556e-06,
+      "loss": 0.0786,
+      "step": 59970
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 9.957361312646502e-06,
+      "loss": 0.0822,
+      "step": 59980
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 9.954012724634446e-06,
+      "loss": 0.0687,
+      "step": 59990
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 9.95066413662239e-06,
+      "loss": 0.0142,
+      "step": 60000
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 9.947315548610336e-06,
+      "loss": 0.1316,
+      "step": 60010
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 9.943966960598282e-06,
+      "loss": 0.0583,
+      "step": 60020
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 9.940618372586226e-06,
+      "loss": 0.069,
+      "step": 60030
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 9.937269784574172e-06,
+      "loss": 0.0145,
+      "step": 60040
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 9.933921196562117e-06,
+      "loss": 0.0639,
+      "step": 60050
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 9.930572608550062e-06,
+      "loss": 0.0483,
+      "step": 60060
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 9.927224020538006e-06,
+      "loss": 0.1245,
+      "step": 60070
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 9.923875432525953e-06,
+      "loss": 0.074,
+      "step": 60080
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 9.920526844513897e-06,
+      "loss": 0.0815,
+      "step": 60090
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 9.917178256501841e-06,
+      "loss": 0.1323,
+      "step": 60100
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 9.913829668489787e-06,
+      "loss": 0.0768,
+      "step": 60110
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 9.910481080477733e-06,
+      "loss": 0.123,
+      "step": 60120
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 9.907132492465677e-06,
+      "loss": 0.1515,
+      "step": 60130
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 9.903783904453621e-06,
+      "loss": 0.0644,
+      "step": 60140
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 9.900435316441568e-06,
+      "loss": 0.1204,
+      "step": 60150
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 9.897086728429512e-06,
+      "loss": 0.1019,
+      "step": 60160
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 9.893738140417456e-06,
+      "loss": 0.0687,
+      "step": 60170
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 9.890389552405404e-06,
+      "loss": 0.0235,
+      "step": 60180
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 9.887040964393348e-06,
+      "loss": 0.0698,
+      "step": 60190
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 9.883692376381292e-06,
+      "loss": 0.1181,
+      "step": 60200
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 9.880343788369238e-06,
+      "loss": 0.0437,
+      "step": 60210
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 9.876995200357184e-06,
+      "loss": 0.0569,
+      "step": 60220
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 9.873646612345128e-06,
+      "loss": 0.3347,
+      "step": 60230
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 9.870298024333072e-06,
+      "loss": 0.1221,
+      "step": 60240
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 9.86694943632102e-06,
+      "loss": 0.0497,
+      "step": 60250
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 9.863600848308963e-06,
+      "loss": 0.1742,
+      "step": 60260
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 9.860252260296907e-06,
+      "loss": 0.1026,
+      "step": 60270
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 9.856903672284853e-06,
+      "loss": 0.1058,
+      "step": 60280
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 9.853555084272799e-06,
+      "loss": 0.0718,
+      "step": 60290
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 9.850206496260743e-06,
+      "loss": 0.0364,
+      "step": 60300
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 9.846857908248689e-06,
+      "loss": 0.0234,
+      "step": 60310
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 9.843509320236635e-06,
+      "loss": 0.136,
+      "step": 60320
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 9.840160732224579e-06,
+      "loss": 0.0836,
+      "step": 60330
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 9.836812144212524e-06,
+      "loss": 0.0306,
+      "step": 60340
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 9.833463556200468e-06,
+      "loss": 0.051,
+      "step": 60350
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 9.830114968188414e-06,
+      "loss": 0.0602,
+      "step": 60360
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 9.826766380176358e-06,
+      "loss": 0.0714,
+      "step": 60370
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 9.823417792164304e-06,
+      "loss": 0.0651,
+      "step": 60380
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 9.82006920415225e-06,
+      "loss": 0.2184,
+      "step": 60390
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 9.816720616140194e-06,
+      "loss": 0.0474,
+      "step": 60400
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 9.81337202812814e-06,
+      "loss": 0.0816,
+      "step": 60410
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 9.810023440116085e-06,
+      "loss": 0.0758,
+      "step": 60420
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 9.80667485210403e-06,
+      "loss": 0.0695,
+      "step": 60430
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 9.803326264091975e-06,
+      "loss": 0.1688,
+      "step": 60440
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 9.79997767607992e-06,
+      "loss": 0.0878,
+      "step": 60450
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 9.796629088067865e-06,
+      "loss": 0.135,
+      "step": 60460
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 9.793280500055811e-06,
+      "loss": 0.0652,
+      "step": 60470
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 9.789931912043755e-06,
+      "loss": 0.0902,
+      "step": 60480
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 9.7865833240317e-06,
+      "loss": 0.0957,
+      "step": 60490
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 9.783234736019645e-06,
+      "loss": 0.0627,
+      "step": 60500
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 9.77988614800759e-06,
+      "loss": 0.0964,
+      "step": 60510
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 9.776537559995535e-06,
+      "loss": 0.0719,
+      "step": 60520
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 9.77318897198348e-06,
+      "loss": 0.0708,
+      "step": 60530
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 9.769840383971426e-06,
+      "loss": 0.0963,
+      "step": 60540
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 9.76649179595937e-06,
+      "loss": 0.0515,
+      "step": 60550
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 9.763143207947316e-06,
+      "loss": 0.0661,
+      "step": 60560
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 9.759794619935262e-06,
+      "loss": 0.1804,
+      "step": 60570
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 9.756446031923206e-06,
+      "loss": 0.0491,
+      "step": 60580
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 9.75309744391115e-06,
+      "loss": 0.1254,
+      "step": 60590
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 9.749748855899097e-06,
+      "loss": 0.0666,
+      "step": 60600
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 9.746400267887041e-06,
+      "loss": 0.0979,
+      "step": 60610
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 9.743051679874986e-06,
+      "loss": 0.0743,
+      "step": 60620
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 9.740037950664136e-06,
+      "loss": 0.0688,
+      "step": 60630
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 9.736689362652082e-06,
+      "loss": 0.0857,
+      "step": 60640
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 9.733340774640028e-06,
+      "loss": 0.0524,
+      "step": 60650
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 9.729992186627972e-06,
+      "loss": 0.0328,
+      "step": 60660
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 9.726643598615918e-06,
+      "loss": 0.1326,
+      "step": 60670
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 9.723295010603863e-06,
+      "loss": 0.1556,
+      "step": 60680
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 9.719946422591807e-06,
+      "loss": 0.0655,
+      "step": 60690
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 9.716597834579752e-06,
+      "loss": 0.0955,
+      "step": 60700
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 9.713249246567697e-06,
+      "loss": 0.0694,
+      "step": 60710
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 9.709900658555643e-06,
+      "loss": 0.1118,
+      "step": 60720
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 9.706552070543587e-06,
+      "loss": 0.0892,
+      "step": 60730
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 9.703203482531533e-06,
+      "loss": 0.1141,
+      "step": 60740
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 9.699854894519479e-06,
+      "loss": 0.0429,
+      "step": 60750
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 9.696506306507423e-06,
+      "loss": 0.0825,
+      "step": 60760
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 9.693157718495367e-06,
+      "loss": 0.088,
+      "step": 60770
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 9.689809130483314e-06,
+      "loss": 0.0571,
+      "step": 60780
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 9.686460542471258e-06,
+      "loss": 0.066,
+      "step": 60790
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 9.683111954459202e-06,
+      "loss": 0.0937,
+      "step": 60800
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 9.67976336644715e-06,
+      "loss": 0.0755,
+      "step": 60810
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 9.676414778435094e-06,
+      "loss": 0.0774,
+      "step": 60820
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 9.673066190423038e-06,
+      "loss": 0.0493,
+      "step": 60830
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 9.669717602410982e-06,
+      "loss": 0.0431,
+      "step": 60840
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 9.66636901439893e-06,
+      "loss": 0.0751,
+      "step": 60850
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 9.663020426386874e-06,
+      "loss": 0.0562,
+      "step": 60860
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 9.659671838374818e-06,
+      "loss": 0.2013,
+      "step": 60870
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 9.656323250362765e-06,
+      "loss": 0.0418,
+      "step": 60880
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 9.65297466235071e-06,
+      "loss": 0.0913,
+      "step": 60890
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 9.649626074338653e-06,
+      "loss": 0.0308,
+      "step": 60900
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 9.646277486326599e-06,
+      "loss": 0.1197,
+      "step": 60910
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 9.642928898314545e-06,
+      "loss": 0.0634,
+      "step": 60920
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 9.639580310302489e-06,
+      "loss": 0.0761,
+      "step": 60930
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 9.636231722290435e-06,
+      "loss": 0.0938,
+      "step": 60940
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 9.63288313427838e-06,
+      "loss": 0.0952,
+      "step": 60950
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 9.629534546266325e-06,
+      "loss": 0.0474,
+      "step": 60960
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 9.626185958254269e-06,
+      "loss": 0.0722,
+      "step": 60970
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 9.622837370242214e-06,
+      "loss": 0.0293,
+      "step": 60980
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 9.61948878223016e-06,
+      "loss": 0.2301,
+      "step": 60990
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 9.616140194218104e-06,
+      "loss": 0.0467,
+      "step": 61000
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 9.61279160620605e-06,
+      "loss": 0.1107,
+      "step": 61010
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 9.609443018193996e-06,
+      "loss": 0.0805,
+      "step": 61020
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 9.60609443018194e-06,
+      "loss": 0.0622,
+      "step": 61030
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 9.602745842169886e-06,
+      "loss": 0.0569,
+      "step": 61040
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 9.59939725415783e-06,
+      "loss": 0.1016,
+      "step": 61050
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 9.596048666145775e-06,
+      "loss": 0.0424,
+      "step": 61060
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 9.592700078133721e-06,
+      "loss": 0.1066,
+      "step": 61070
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 9.589351490121665e-06,
+      "loss": 0.0375,
+      "step": 61080
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 9.586002902109611e-06,
+      "loss": 0.0879,
+      "step": 61090
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 9.582654314097555e-06,
+      "loss": 0.0747,
+      "step": 61100
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 9.579305726085501e-06,
+      "loss": 0.0414,
+      "step": 61110
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 9.575957138073447e-06,
+      "loss": 0.1524,
+      "step": 61120
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 9.57260855006139e-06,
+      "loss": 0.0421,
+      "step": 61130
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 9.569259962049337e-06,
+      "loss": 0.1064,
+      "step": 61140
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 9.56591137403728e-06,
+      "loss": 0.1296,
+      "step": 61150
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 9.562562786025226e-06,
+      "loss": 0.0213,
+      "step": 61160
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 9.559214198013172e-06,
+      "loss": 0.1248,
+      "step": 61170
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 9.555865610001116e-06,
+      "loss": 0.0672,
+      "step": 61180
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 9.552517021989062e-06,
+      "loss": 0.1487,
+      "step": 61190
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 9.549168433977006e-06,
+      "loss": 0.1142,
+      "step": 61200
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 9.545819845964952e-06,
+      "loss": 0.1013,
+      "step": 61210
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 9.542471257952896e-06,
+      "loss": 0.0741,
+      "step": 61220
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 9.539122669940842e-06,
+      "loss": 0.0922,
+      "step": 61230
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 9.535774081928787e-06,
+      "loss": 0.0462,
+      "step": 61240
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 9.532425493916732e-06,
+      "loss": 0.2549,
+      "step": 61250
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 9.529076905904677e-06,
+      "loss": 0.1295,
+      "step": 61260
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 9.525728317892623e-06,
+      "loss": 0.1121,
+      "step": 61270
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 9.522379729880567e-06,
+      "loss": 0.0271,
+      "step": 61280
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 9.519031141868511e-06,
+      "loss": 0.0379,
+      "step": 61290
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 9.515682553856459e-06,
+      "loss": 0.0525,
+      "step": 61300
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 9.512333965844403e-06,
+      "loss": 0.0561,
+      "step": 61310
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 9.508985377832347e-06,
+      "loss": 0.0763,
+      "step": 61320
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 9.505636789820293e-06,
+      "loss": 0.0593,
+      "step": 61330
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 9.502288201808238e-06,
+      "loss": 0.1232,
+      "step": 61340
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 9.498939613796182e-06,
+      "loss": 0.0506,
+      "step": 61350
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 9.495591025784126e-06,
+      "loss": 0.022,
+      "step": 61360
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 9.492242437772074e-06,
+      "loss": 0.1578,
+      "step": 61370
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 9.488893849760018e-06,
+      "loss": 0.0204,
+      "step": 61380
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 9.485545261747962e-06,
+      "loss": 0.0583,
+      "step": 61390
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 9.48219667373591e-06,
+      "loss": 0.0304,
+      "step": 61400
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 9.478848085723854e-06,
+      "loss": 0.0518,
+      "step": 61410
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 9.475499497711798e-06,
+      "loss": 0.1067,
+      "step": 61420
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 9.472150909699743e-06,
+      "loss": 0.1312,
+      "step": 61430
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 9.46880232168769e-06,
+      "loss": 0.1231,
+      "step": 61440
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 9.465453733675633e-06,
+      "loss": 0.1041,
+      "step": 61450
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 9.462105145663577e-06,
+      "loss": 0.1295,
+      "step": 61460
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 9.458756557651525e-06,
+      "loss": 0.0938,
+      "step": 61470
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 9.455407969639469e-06,
+      "loss": 0.0972,
+      "step": 61480
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 9.452059381627413e-06,
+      "loss": 0.0595,
+      "step": 61490
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 9.448710793615359e-06,
+      "loss": 0.0673,
+      "step": 61500
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 9.445362205603305e-06,
+      "loss": 0.0246,
+      "step": 61510
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 9.442013617591249e-06,
+      "loss": 0.0941,
+      "step": 61520
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 9.438665029579194e-06,
+      "loss": 0.0629,
+      "step": 61530
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 9.43531644156714e-06,
+      "loss": 0.1436,
+      "step": 61540
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 9.431967853555084e-06,
+      "loss": 0.0688,
+      "step": 61550
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 9.42861926554303e-06,
+      "loss": 0.101,
+      "step": 61560
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 9.425270677530976e-06,
+      "loss": 0.1802,
+      "step": 61570
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 9.42192208951892e-06,
+      "loss": 0.0531,
+      "step": 61580
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 9.418573501506864e-06,
+      "loss": 0.0331,
+      "step": 61590
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 9.41522491349481e-06,
+      "loss": 0.1009,
+      "step": 61600
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 9.411876325482755e-06,
+      "loss": 0.0718,
+      "step": 61610
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 9.4085277374707e-06,
+      "loss": 0.1523,
+      "step": 61620
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 9.405179149458645e-06,
+      "loss": 0.1847,
+      "step": 61630
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 9.401830561446591e-06,
+      "loss": 0.0619,
+      "step": 61640
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 9.398481973434535e-06,
+      "loss": 0.0599,
+      "step": 61650
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 9.395133385422481e-06,
+      "loss": 0.0748,
+      "step": 61660
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 9.391784797410425e-06,
+      "loss": 0.1077,
+      "step": 61670
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 9.38843620939837e-06,
+      "loss": 0.1081,
+      "step": 61680
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 9.385087621386317e-06,
+      "loss": 0.0602,
+      "step": 61690
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 9.38173903337426e-06,
+      "loss": 0.052,
+      "step": 61700
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 9.378390445362206e-06,
+      "loss": 0.0598,
+      "step": 61710
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 9.37504185735015e-06,
+      "loss": 0.0837,
+      "step": 61720
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 9.371693269338096e-06,
+      "loss": 0.0626,
+      "step": 61730
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 9.36834468132604e-06,
+      "loss": 0.078,
+      "step": 61740
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 9.364996093313986e-06,
+      "loss": 0.0349,
+      "step": 61750
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 9.361647505301932e-06,
+      "loss": 0.0563,
+      "step": 61760
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 9.358298917289876e-06,
+      "loss": 0.1534,
+      "step": 61770
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 9.354950329277822e-06,
+      "loss": 0.0398,
+      "step": 61780
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 9.351601741265767e-06,
+      "loss": 0.0912,
+      "step": 61790
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 9.348253153253711e-06,
+      "loss": 0.0726,
+      "step": 61800
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 9.344904565241656e-06,
+      "loss": 0.0488,
+      "step": 61810
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 9.341555977229603e-06,
+      "loss": 0.2143,
+      "step": 61820
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 9.338207389217547e-06,
+      "loss": 0.0861,
+      "step": 61830
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 9.334858801205491e-06,
+      "loss": 0.0598,
+      "step": 61840
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 9.331510213193437e-06,
+      "loss": 0.0585,
+      "step": 61850
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 9.328161625181383e-06,
+      "loss": 0.1335,
+      "step": 61860
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 9.324813037169327e-06,
+      "loss": 0.0963,
+      "step": 61870
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 9.321464449157271e-06,
+      "loss": 0.0389,
+      "step": 61880
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 9.318115861145218e-06,
+      "loss": 0.2009,
+      "step": 61890
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 9.314767273133162e-06,
+      "loss": 0.4019,
+      "step": 61900
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 9.311418685121106e-06,
+      "loss": 0.0987,
+      "step": 61910
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 9.308070097109054e-06,
+      "loss": 0.1348,
+      "step": 61920
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 9.304721509096998e-06,
+      "loss": 0.0508,
+      "step": 61930
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 9.301372921084942e-06,
+      "loss": 0.0747,
+      "step": 61940
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 9.298024333072888e-06,
+      "loss": 0.0665,
+      "step": 61950
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 9.294675745060834e-06,
+      "loss": 0.0739,
+      "step": 61960
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 9.291327157048778e-06,
+      "loss": 0.0466,
+      "step": 61970
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 9.287978569036722e-06,
+      "loss": 0.1095,
+      "step": 61980
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 9.28462998102467e-06,
+      "loss": 0.0998,
+      "step": 61990
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 9.281281393012613e-06,
+      "loss": 0.0413,
+      "step": 62000
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 9.277932805000557e-06,
+      "loss": 0.0415,
+      "step": 62010
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 9.274584216988505e-06,
+      "loss": 0.057,
+      "step": 62020
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 9.271235628976449e-06,
+      "loss": 0.0553,
+      "step": 62030
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 9.267887040964393e-06,
+      "loss": 0.0809,
+      "step": 62040
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 9.264538452952339e-06,
+      "loss": 0.015,
+      "step": 62050
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 9.261189864940285e-06,
+      "loss": 0.1154,
+      "step": 62060
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 9.257841276928229e-06,
+      "loss": 0.1386,
+      "step": 62070
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 9.254492688916173e-06,
+      "loss": 0.0279,
+      "step": 62080
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 9.25114410090412e-06,
+      "loss": 0.1054,
+      "step": 62090
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 9.247795512892064e-06,
+      "loss": 0.0182,
+      "step": 62100
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 9.244446924880008e-06,
+      "loss": 0.0864,
+      "step": 62110
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 9.241098336867954e-06,
+      "loss": 0.1223,
+      "step": 62120
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 9.2377497488559e-06,
+      "loss": 0.1038,
+      "step": 62130
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 9.234401160843844e-06,
+      "loss": 0.021,
+      "step": 62140
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 9.23105257283179e-06,
+      "loss": 0.0628,
+      "step": 62150
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 9.227703984819735e-06,
+      "loss": 0.1214,
+      "step": 62160
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 9.22435539680768e-06,
+      "loss": 0.1438,
+      "step": 62170
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 9.221006808795625e-06,
+      "loss": 0.0662,
+      "step": 62180
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 9.21765822078357e-06,
+      "loss": 0.0349,
+      "step": 62190
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 9.214309632771515e-06,
+      "loss": 0.1451,
+      "step": 62200
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.21096104475946e-06,
+      "loss": 0.0835,
+      "step": 62210
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.207612456747405e-06,
+      "loss": 0.1092,
+      "step": 62220
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.20426386873535e-06,
+      "loss": 0.0643,
+      "step": 62230
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.200915280723295e-06,
+      "loss": 0.0882,
+      "step": 62240
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.19756669271124e-06,
+      "loss": 0.0279,
+      "step": 62250
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.194218104699185e-06,
+      "loss": 0.1231,
+      "step": 62260
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.19086951668713e-06,
+      "loss": 0.416,
+      "step": 62270
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.187520928675076e-06,
+      "loss": 0.0558,
+      "step": 62280
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.18417234066302e-06,
+      "loss": 0.0471,
+      "step": 62290
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.180823752650966e-06,
+      "loss": 0.0424,
+      "step": 62300
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.177475164638912e-06,
+      "loss": 0.0225,
+      "step": 62310
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.174126576626856e-06,
+      "loss": 0.0881,
+      "step": 62320
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.1707779886148e-06,
+      "loss": 0.0604,
+      "step": 62330
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.167429400602746e-06,
+      "loss": 0.2354,
+      "step": 62340
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.164080812590691e-06,
+      "loss": 0.0753,
+      "step": 62350
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.160732224578636e-06,
+      "loss": 0.131,
+      "step": 62360
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.157383636566581e-06,
+      "loss": 0.0593,
+      "step": 62370
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.154035048554527e-06,
+      "loss": 0.1101,
+      "step": 62380
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.150686460542471e-06,
+      "loss": 0.0353,
+      "step": 62390
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.147337872530415e-06,
+      "loss": 0.0783,
+      "step": 62400
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.143989284518363e-06,
+      "loss": 0.1456,
+      "step": 62410
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.140640696506307e-06,
+      "loss": 0.0893,
+      "step": 62420
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.13729210849425e-06,
+      "loss": 0.1,
+      "step": 62430
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.133943520482198e-06,
+      "loss": 0.0506,
+      "step": 62440
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.130594932470142e-06,
+      "loss": 0.0603,
+      "step": 62450
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.127246344458086e-06,
+      "loss": 0.0363,
+      "step": 62460
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.123897756446032e-06,
+      "loss": 0.06,
+      "step": 62470
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.120549168433978e-06,
+      "loss": 0.0829,
+      "step": 62480
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.117200580421922e-06,
+      "loss": 0.0776,
+      "step": 62490
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.113851992409866e-06,
+      "loss": 0.0995,
+      "step": 62500
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.110503404397814e-06,
+      "loss": 0.151,
+      "step": 62510
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.107154816385758e-06,
+      "loss": 0.0809,
+      "step": 62520
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.103806228373702e-06,
+      "loss": 0.1239,
+      "step": 62530
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.10045764036165e-06,
+      "loss": 0.0604,
+      "step": 62540
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.097109052349593e-06,
+      "loss": 0.0572,
+      "step": 62550
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.093760464337537e-06,
+      "loss": 0.1108,
+      "step": 62560
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.090411876325483e-06,
+      "loss": 0.366,
+      "step": 62570
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.087063288313429e-06,
+      "loss": 0.0953,
+      "step": 62580
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.083714700301373e-06,
+      "loss": 0.0323,
+      "step": 62590
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.080366112289317e-06,
+      "loss": 0.0607,
+      "step": 62600
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.077017524277265e-06,
+      "loss": 0.0906,
+      "step": 62610
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.073668936265209e-06,
+      "loss": 0.1969,
+      "step": 62620
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.070320348253153e-06,
+      "loss": 0.023,
+      "step": 62630
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.066971760241098e-06,
+      "loss": 0.0463,
+      "step": 62640
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.063623172229044e-06,
+      "loss": 0.1083,
+      "step": 62650
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.060274584216988e-06,
+      "loss": 0.0535,
+      "step": 62660
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.056925996204934e-06,
+      "loss": 0.1023,
+      "step": 62670
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.05357740819288e-06,
+      "loss": 0.0784,
+      "step": 62680
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.050228820180824e-06,
+      "loss": 0.0744,
+      "step": 62690
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.04688023216877e-06,
+      "loss": 0.1059,
+      "step": 62700
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.043531644156714e-06,
+      "loss": 0.0489,
+      "step": 62710
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.04018305614466e-06,
+      "loss": 0.0712,
+      "step": 62720
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.036834468132604e-06,
+      "loss": 0.088,
+      "step": 62730
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.03348588012055e-06,
+      "loss": 0.1482,
+      "step": 62740
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.030137292108495e-06,
+      "loss": 0.0573,
+      "step": 62750
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.02678870409644e-06,
+      "loss": 0.0598,
+      "step": 62760
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.023440116084385e-06,
+      "loss": 0.0706,
+      "step": 62770
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.020091528072329e-06,
+      "loss": 0.0464,
+      "step": 62780
+    },
+    {
+      "epoch": 7.0,
+      "eval_loss": 0.269069105386734,
+      "eval_runtime": 250.6126,
+      "eval_samples_per_second": 7.957,
+      "eval_steps_per_second": 3.978,
+      "eval_wer": 0.09104717926907069,
+      "step": 62783
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.016742940060275e-06,
+      "loss": 0.0777,
+      "step": 62790
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.01339435204822e-06,
+      "loss": 0.0666,
+      "step": 62800
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.010045764036165e-06,
+      "loss": 0.0675,
+      "step": 62810
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.00669717602411e-06,
+      "loss": 0.0716,
+      "step": 62820
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.003348588012054e-06,
+      "loss": 0.058,
+      "step": 62830
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9e-06,
+      "loss": 0.0538,
+      "step": 62840
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 8.996651411987944e-06,
+      "loss": 0.0316,
+      "step": 62850
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 8.99330282397589e-06,
+      "loss": 0.0716,
+      "step": 62860
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 8.989954235963836e-06,
+      "loss": 0.065,
+      "step": 62870
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 8.98660564795178e-06,
+      "loss": 0.0723,
+      "step": 62880
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 8.983257059939726e-06,
+      "loss": 0.0778,
+      "step": 62890
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 8.979908471927671e-06,
+      "loss": 0.0615,
+      "step": 62900
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 8.976559883915616e-06,
+      "loss": 0.0474,
+      "step": 62910
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 8.973211295903561e-06,
+      "loss": 0.0826,
+      "step": 62920
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 8.969862707891507e-06,
+      "loss": 0.0639,
+      "step": 62930
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 8.966514119879451e-06,
+      "loss": 0.0787,
+      "step": 62940
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 8.963165531867395e-06,
+      "loss": 0.0263,
+      "step": 62950
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 8.959816943855341e-06,
+      "loss": 0.0226,
+      "step": 62960
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 8.956468355843287e-06,
+      "loss": 0.0544,
+      "step": 62970
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 8.95311976783123e-06,
+      "loss": 0.0736,
+      "step": 62980
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 8.949771179819177e-06,
+      "loss": 0.0381,
+      "step": 62990
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 8.946422591807122e-06,
+      "loss": 0.0514,
+      "step": 63000
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 8.943074003795066e-06,
+      "loss": 0.0298,
+      "step": 63010
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 8.93972541578301e-06,
+      "loss": 0.0679,
+      "step": 63020
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 8.936376827770958e-06,
+      "loss": 0.0602,
+      "step": 63030
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 8.933028239758902e-06,
+      "loss": 0.0696,
+      "step": 63040
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 8.929679651746846e-06,
+      "loss": 0.0342,
+      "step": 63050
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 8.926331063734794e-06,
+      "loss": 0.0766,
+      "step": 63060
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 8.922982475722738e-06,
+      "loss": 0.0693,
+      "step": 63070
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 8.919633887710682e-06,
+      "loss": 0.0688,
+      "step": 63080
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 8.916285299698626e-06,
+      "loss": 0.1069,
+      "step": 63090
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 8.912936711686573e-06,
+      "loss": 0.054,
+      "step": 63100
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 8.909588123674517e-06,
+      "loss": 0.0185,
+      "step": 63110
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 8.906239535662461e-06,
+      "loss": 0.0838,
+      "step": 63120
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 8.902890947650409e-06,
+      "loss": 0.0744,
+      "step": 63130
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 8.899542359638353e-06,
+      "loss": 0.072,
+      "step": 63140
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 8.896193771626297e-06,
+      "loss": 0.0714,
+      "step": 63150
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 8.892845183614243e-06,
+      "loss": 0.0645,
+      "step": 63160
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 8.889496595602189e-06,
+      "loss": 0.0411,
+      "step": 63170
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 8.886148007590133e-06,
+      "loss": 0.066,
+      "step": 63180
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 8.882799419578078e-06,
+      "loss": 0.1144,
+      "step": 63190
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 8.879450831566024e-06,
+      "loss": 0.0362,
+      "step": 63200
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 8.876102243553968e-06,
+      "loss": 0.1056,
+      "step": 63210
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 8.872753655541912e-06,
+      "loss": 0.0308,
+      "step": 63220
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 8.869405067529858e-06,
+      "loss": 0.0579,
+      "step": 63230
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 8.866056479517804e-06,
+      "loss": 0.0664,
+      "step": 63240
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 8.862707891505748e-06,
+      "loss": 0.0323,
+      "step": 63250
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 8.859359303493694e-06,
+      "loss": 0.0868,
+      "step": 63260
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 8.85601071548164e-06,
+      "loss": 0.0745,
+      "step": 63270
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 8.852662127469584e-06,
+      "loss": 0.0275,
+      "step": 63280
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 8.84931353945753e-06,
+      "loss": 0.0904,
+      "step": 63290
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 8.845964951445473e-06,
+      "loss": 0.0444,
+      "step": 63300
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 8.842616363433419e-06,
+      "loss": 0.0801,
+      "step": 63310
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 8.839267775421365e-06,
+      "loss": 0.0798,
+      "step": 63320
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 8.835919187409309e-06,
+      "loss": 0.1211,
+      "step": 63330
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 8.832570599397255e-06,
+      "loss": 0.0555,
+      "step": 63340
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 8.829222011385199e-06,
+      "loss": 0.0969,
+      "step": 63350
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 8.825873423373145e-06,
+      "loss": 0.0511,
+      "step": 63360
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 8.82252483536109e-06,
+      "loss": 0.0557,
+      "step": 63370
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 8.819176247349034e-06,
+      "loss": 0.0549,
+      "step": 63380
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 8.81582765933698e-06,
+      "loss": 0.1492,
+      "step": 63390
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 8.812479071324924e-06,
+      "loss": 0.0414,
+      "step": 63400
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 8.80913048331287e-06,
+      "loss": 0.0485,
+      "step": 63410
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 8.805781895300816e-06,
+      "loss": 0.0442,
+      "step": 63420
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 8.80243330728876e-06,
+      "loss": 0.0649,
+      "step": 63430
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 8.799084719276706e-06,
+      "loss": 0.2194,
+      "step": 63440
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 8.795736131264651e-06,
+      "loss": 0.0471,
+      "step": 63450
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 8.792387543252596e-06,
+      "loss": 0.0793,
+      "step": 63460
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 8.78903895524054e-06,
+      "loss": 0.0451,
+      "step": 63470
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 8.785690367228485e-06,
+      "loss": 0.0555,
+      "step": 63480
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 8.782341779216431e-06,
+      "loss": 0.0739,
+      "step": 63490
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 8.778993191204375e-06,
+      "loss": 0.0876,
+      "step": 63500
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 8.775644603192321e-06,
+      "loss": 0.0388,
+      "step": 63510
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 8.772296015180267e-06,
+      "loss": 0.1003,
+      "step": 63520
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 8.76894742716821e-06,
+      "loss": 0.0927,
+      "step": 63530
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 8.765598839156155e-06,
+      "loss": 0.062,
+      "step": 63540
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 8.762250251144102e-06,
+      "loss": 0.0808,
+      "step": 63550
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 8.758901663132046e-06,
+      "loss": 0.0339,
+      "step": 63560
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 8.75555307511999e-06,
+      "loss": 0.0832,
+      "step": 63570
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 8.752204487107936e-06,
+      "loss": 0.0837,
+      "step": 63580
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 8.748855899095882e-06,
+      "loss": 0.0703,
+      "step": 63590
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 8.745507311083826e-06,
+      "loss": 0.0266,
+      "step": 63600
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 8.74215872307177e-06,
+      "loss": 0.0241,
+      "step": 63610
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 8.738810135059718e-06,
+      "loss": 0.0533,
+      "step": 63620
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 8.735461547047662e-06,
+      "loss": 0.075,
+      "step": 63630
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 8.732112959035606e-06,
+      "loss": 0.1146,
+      "step": 63640
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 8.728764371023553e-06,
+      "loss": 0.0418,
+      "step": 63650
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 8.725415783011497e-06,
+      "loss": 0.0373,
+      "step": 63660
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 8.722067194999441e-06,
+      "loss": 0.0799,
+      "step": 63670
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 8.718718606987387e-06,
+      "loss": 0.0302,
+      "step": 63680
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 8.715370018975333e-06,
+      "loss": 0.0929,
+      "step": 63690
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 8.712021430963277e-06,
+      "loss": 0.1047,
+      "step": 63700
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 8.708672842951221e-06,
+      "loss": 0.0474,
+      "step": 63710
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 8.705324254939169e-06,
+      "loss": 0.0397,
+      "step": 63720
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.701975666927113e-06,
+      "loss": 0.0923,
+      "step": 63730
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.698627078915057e-06,
+      "loss": 0.0529,
+      "step": 63740
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.695278490903002e-06,
+      "loss": 0.0456,
+      "step": 63750
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.691929902890948e-06,
+      "loss": 0.0656,
+      "step": 63760
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.688581314878892e-06,
+      "loss": 0.0785,
+      "step": 63770
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.685232726866838e-06,
+      "loss": 0.1292,
+      "step": 63780
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.681884138854784e-06,
+      "loss": 0.0812,
+      "step": 63790
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.678535550842728e-06,
+      "loss": 0.0627,
+      "step": 63800
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 8.675186962830674e-06,
+      "loss": 0.084,
+      "step": 63810
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.67183837481862e-06,
+      "loss": 0.1412,
+      "step": 63820
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.668489786806564e-06,
+      "loss": 0.0736,
+      "step": 63830
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.665141198794508e-06,
+      "loss": 0.1311,
+      "step": 63840
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.661792610782453e-06,
+      "loss": 0.1093,
+      "step": 63850
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.658444022770399e-06,
+      "loss": 0.0466,
+      "step": 63860
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.655095434758343e-06,
+      "loss": 0.0263,
+      "step": 63870
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.651746846746289e-06,
+      "loss": 0.062,
+      "step": 63880
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.648398258734235e-06,
+      "loss": 0.0566,
+      "step": 63890
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 8.645049670722179e-06,
+      "loss": 0.0439,
+      "step": 63900
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.641701082710125e-06,
+      "loss": 0.0197,
+      "step": 63910
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.638352494698069e-06,
+      "loss": 0.0427,
+      "step": 63920
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.635003906686014e-06,
+      "loss": 0.0224,
+      "step": 63930
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.63165531867396e-06,
+      "loss": 0.0437,
+      "step": 63940
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.628306730661904e-06,
+      "loss": 0.0391,
+      "step": 63950
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.62495814264985e-06,
+      "loss": 0.0771,
+      "step": 63960
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.621609554637794e-06,
+      "loss": 0.1004,
+      "step": 63970
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.61826096662574e-06,
+      "loss": 0.0333,
+      "step": 63980
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 8.614912378613684e-06,
+      "loss": 0.0482,
+      "step": 63990
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.61156379060163e-06,
+      "loss": 0.0527,
+      "step": 64000
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.608215202589575e-06,
+      "loss": 0.0966,
+      "step": 64010
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.60486661457752e-06,
+      "loss": 0.1037,
+      "step": 64020
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.601518026565465e-06,
+      "loss": 0.0334,
+      "step": 64030
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.598169438553411e-06,
+      "loss": 0.0659,
+      "step": 64040
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.594820850541355e-06,
+      "loss": 0.0974,
+      "step": 64050
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.5914722625293e-06,
+      "loss": 0.0438,
+      "step": 64060
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.588123674517247e-06,
+      "loss": 0.0339,
+      "step": 64070
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 8.58477508650519e-06,
+      "loss": 0.1014,
+      "step": 64080
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.581426498493135e-06,
+      "loss": 0.0487,
+      "step": 64090
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.57807791048108e-06,
+      "loss": 0.1009,
+      "step": 64100
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.574729322469026e-06,
+      "loss": 0.0576,
+      "step": 64110
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.57138073445697e-06,
+      "loss": 0.0205,
+      "step": 64120
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.568032146444915e-06,
+      "loss": 0.1295,
+      "step": 64130
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.564683558432862e-06,
+      "loss": 0.0949,
+      "step": 64140
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.561334970420806e-06,
+      "loss": 0.1196,
+      "step": 64150
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.55798638240875e-06,
+      "loss": 0.1029,
+      "step": 64160
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 8.554637794396698e-06,
+      "loss": 0.0491,
+      "step": 64170
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.551289206384642e-06,
+      "loss": 0.0483,
+      "step": 64180
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.547940618372586e-06,
+      "loss": 0.0631,
+      "step": 64190
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.544592030360532e-06,
+      "loss": 0.0181,
+      "step": 64200
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.541243442348477e-06,
+      "loss": 0.0418,
+      "step": 64210
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.537894854336421e-06,
+      "loss": 0.0791,
+      "step": 64220
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.534546266324365e-06,
+      "loss": 0.0741,
+      "step": 64230
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.531197678312313e-06,
+      "loss": 0.0345,
+      "step": 64240
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.527849090300257e-06,
+      "loss": 0.0812,
+      "step": 64250
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 8.524500502288201e-06,
+      "loss": 0.0358,
+      "step": 64260
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.521151914276149e-06,
+      "loss": 0.0962,
+      "step": 64270
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.517803326264093e-06,
+      "loss": 0.1132,
+      "step": 64280
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.514454738252037e-06,
+      "loss": 0.0485,
+      "step": 64290
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.511106150239982e-06,
+      "loss": 0.1012,
+      "step": 64300
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.507757562227928e-06,
+      "loss": 0.078,
+      "step": 64310
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.504408974215872e-06,
+      "loss": 0.0535,
+      "step": 64320
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.501060386203818e-06,
+      "loss": 0.0718,
+      "step": 64330
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.497711798191764e-06,
+      "loss": 0.1039,
+      "step": 64340
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 8.494363210179708e-06,
+      "loss": 0.0451,
+      "step": 64350
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.491014622167652e-06,
+      "loss": 0.036,
+      "step": 64360
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.487666034155598e-06,
+      "loss": 0.0666,
+      "step": 64370
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.484317446143543e-06,
+      "loss": 0.122,
+      "step": 64380
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.480968858131488e-06,
+      "loss": 0.0523,
+      "step": 64390
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.477620270119433e-06,
+      "loss": 0.0403,
+      "step": 64400
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.474271682107379e-06,
+      "loss": 0.0722,
+      "step": 64410
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.470923094095323e-06,
+      "loss": 0.0487,
+      "step": 64420
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.467574506083269e-06,
+      "loss": 0.0926,
+      "step": 64430
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 8.464225918071213e-06,
+      "loss": 0.0897,
+      "step": 64440
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.460877330059159e-06,
+      "loss": 0.1148,
+      "step": 64450
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.457528742047103e-06,
+      "loss": 0.0537,
+      "step": 64460
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.454180154035049e-06,
+      "loss": 0.0791,
+      "step": 64470
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.450831566022994e-06,
+      "loss": 0.1601,
+      "step": 64480
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.447482978010938e-06,
+      "loss": 0.1008,
+      "step": 64490
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.444134389998884e-06,
+      "loss": 0.0215,
+      "step": 64500
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.440785801986828e-06,
+      "loss": 0.0735,
+      "step": 64510
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.437437213974774e-06,
+      "loss": 0.0443,
+      "step": 64520
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 8.43408862596272e-06,
+      "loss": 0.103,
+      "step": 64530
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 8.430740037950664e-06,
+      "loss": 0.0972,
+      "step": 64540
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 8.42739144993861e-06,
+      "loss": 0.0258,
+      "step": 64550
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 8.424042861926555e-06,
+      "loss": 0.3464,
+      "step": 64560
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 8.4206942739145e-06,
+      "loss": 0.0631,
+      "step": 64570
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 8.417345685902444e-06,
+      "loss": 0.0746,
+      "step": 64580
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 8.41399709789039e-06,
+      "loss": 0.0753,
+      "step": 64590
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 8.410648509878335e-06,
+      "loss": 0.0359,
+      "step": 64600
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 8.40729992186628e-06,
+      "loss": 0.0308,
+      "step": 64610
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 8.403951333854225e-06,
+      "loss": 0.0737,
+      "step": 64620
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 8.40060274584217e-06,
+      "loss": 0.0505,
+      "step": 64630
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 8.397254157830115e-06,
+      "loss": 0.0842,
+      "step": 64640
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 8.393905569818059e-06,
+      "loss": 0.0449,
+      "step": 64650
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 8.390556981806006e-06,
+      "loss": 0.0574,
+      "step": 64660
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 8.38720839379395e-06,
+      "loss": 0.0707,
+      "step": 64670
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 8.383859805781895e-06,
+      "loss": 0.0483,
+      "step": 64680
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 8.380511217769842e-06,
+      "loss": 0.0852,
+      "step": 64690
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 8.377162629757786e-06,
+      "loss": 0.0501,
+      "step": 64700
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 8.37381404174573e-06,
+      "loss": 0.1606,
+      "step": 64710
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 8.370465453733676e-06,
+      "loss": 0.0881,
+      "step": 64720
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 8.367116865721622e-06,
+      "loss": 0.0441,
+      "step": 64730
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 8.363768277709566e-06,
+      "loss": 0.077,
+      "step": 64740
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 8.36041968969751e-06,
+      "loss": 0.0101,
+      "step": 64750
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 8.357071101685457e-06,
+      "loss": 0.0397,
+      "step": 64760
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 8.353722513673401e-06,
+      "loss": 0.0661,
+      "step": 64770
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 8.350373925661345e-06,
+      "loss": 0.1154,
+      "step": 64780
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 8.347025337649293e-06,
+      "loss": 0.0646,
+      "step": 64790
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 8.343676749637237e-06,
+      "loss": 0.0455,
+      "step": 64800
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 8.340328161625181e-06,
+      "loss": 0.0321,
+      "step": 64810
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 8.336979573613127e-06,
+      "loss": 0.0666,
+      "step": 64820
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 8.333630985601073e-06,
+      "loss": 0.0424,
+      "step": 64830
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 8.330282397589017e-06,
+      "loss": 0.1114,
+      "step": 64840
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 8.32693380957696e-06,
+      "loss": 0.0329,
+      "step": 64850
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 8.323585221564908e-06,
+      "loss": 0.0366,
+      "step": 64860
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 8.320236633552852e-06,
+      "loss": 0.1053,
+      "step": 64870
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 8.316888045540796e-06,
+      "loss": 0.1078,
+      "step": 64880
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 8.313539457528742e-06,
+      "loss": 0.0948,
+      "step": 64890
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 8.310190869516688e-06,
+      "loss": 0.0456,
+      "step": 64900
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 8.306842281504632e-06,
+      "loss": 0.0249,
+      "step": 64910
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 8.303493693492578e-06,
+      "loss": 0.0722,
+      "step": 64920
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 8.300145105480523e-06,
+      "loss": 0.0732,
+      "step": 64930
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 8.296796517468468e-06,
+      "loss": 0.1283,
+      "step": 64940
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 8.293447929456413e-06,
+      "loss": 0.0372,
+      "step": 64950
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 8.290099341444357e-06,
+      "loss": 0.0445,
+      "step": 64960
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 8.286750753432303e-06,
+      "loss": 0.0764,
+      "step": 64970
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 8.283402165420247e-06,
+      "loss": 0.0585,
+      "step": 64980
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 8.280053577408193e-06,
+      "loss": 0.0419,
+      "step": 64990
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 8.276704989396139e-06,
+      "loss": 0.0727,
+      "step": 65000
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 8.273356401384083e-06,
+      "loss": 0.0913,
+      "step": 65010
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 8.270007813372029e-06,
+      "loss": 0.012,
+      "step": 65020
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 8.266659225359973e-06,
+      "loss": 0.0256,
+      "step": 65030
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 8.263310637347918e-06,
+      "loss": 0.0465,
+      "step": 65040
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 8.259962049335864e-06,
+      "loss": 0.0778,
+      "step": 65050
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 8.256613461323808e-06,
+      "loss": 0.0776,
+      "step": 65060
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 8.253264873311754e-06,
+      "loss": 0.0243,
+      "step": 65070
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 8.2499162852997e-06,
+      "loss": 0.0574,
+      "step": 65080
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 8.246567697287644e-06,
+      "loss": 0.0821,
+      "step": 65090
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 8.243219109275588e-06,
+      "loss": 0.0496,
+      "step": 65100
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 8.239870521263534e-06,
+      "loss": 0.098,
+      "step": 65110
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 8.23652193325148e-06,
+      "loss": 0.0223,
+      "step": 65120
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 8.233173345239424e-06,
+      "loss": 0.0659,
+      "step": 65130
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 8.22982475722737e-06,
+      "loss": 0.0805,
+      "step": 65140
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 8.226476169215315e-06,
+      "loss": 0.1433,
+      "step": 65150
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 8.22312758120326e-06,
+      "loss": 0.0372,
+      "step": 65160
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 8.219778993191205e-06,
+      "loss": 0.0385,
+      "step": 65170
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 8.21643040517915e-06,
+      "loss": 0.0284,
+      "step": 65180
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 8.213081817167095e-06,
+      "loss": 0.0559,
+      "step": 65190
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 8.209733229155039e-06,
+      "loss": 0.0399,
+      "step": 65200
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 8.206384641142985e-06,
+      "loss": 0.0504,
+      "step": 65210
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 8.20303605313093e-06,
+      "loss": 0.0563,
+      "step": 65220
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 8.199687465118874e-06,
+      "loss": 0.0469,
+      "step": 65230
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 8.19633887710682e-06,
+      "loss": 0.0619,
+      "step": 65240
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 8.192990289094766e-06,
+      "loss": 0.1805,
+      "step": 65250
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 8.18964170108271e-06,
+      "loss": 0.0773,
+      "step": 65260
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 8.186293113070654e-06,
+      "loss": 0.2116,
+      "step": 65270
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 8.182944525058602e-06,
+      "loss": 0.147,
+      "step": 65280
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 8.179595937046546e-06,
+      "loss": 0.09,
+      "step": 65290
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 8.17624734903449e-06,
+      "loss": 0.0855,
+      "step": 65300
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 8.172898761022437e-06,
+      "loss": 0.0647,
+      "step": 65310
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 8.169550173010381e-06,
+      "loss": 0.029,
+      "step": 65320
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 8.166201584998325e-06,
+      "loss": 0.071,
+      "step": 65330
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.16285299698627e-06,
+      "loss": 0.0835,
+      "step": 65340
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.159504408974217e-06,
+      "loss": 0.0587,
+      "step": 65350
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.156155820962161e-06,
+      "loss": 0.1219,
+      "step": 65360
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.152807232950105e-06,
+      "loss": 0.0363,
+      "step": 65370
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.149458644938053e-06,
+      "loss": 0.094,
+      "step": 65380
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.146110056925997e-06,
+      "loss": 0.0377,
+      "step": 65390
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.14276146891394e-06,
+      "loss": 0.07,
+      "step": 65400
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.139412880901886e-06,
+      "loss": 0.0587,
+      "step": 65410
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.136064292889832e-06,
+      "loss": 0.1097,
+      "step": 65420
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.132715704877776e-06,
+      "loss": 0.0457,
+      "step": 65430
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.129367116865722e-06,
+      "loss": 0.0679,
+      "step": 65440
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.126018528853668e-06,
+      "loss": 0.0307,
+      "step": 65450
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.122669940841612e-06,
+      "loss": 0.0309,
+      "step": 65460
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.119321352829556e-06,
+      "loss": 0.054,
+      "step": 65470
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.115972764817502e-06,
+      "loss": 0.0732,
+      "step": 65480
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.112624176805448e-06,
+      "loss": 0.1173,
+      "step": 65490
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.109275588793392e-06,
+      "loss": 0.0321,
+      "step": 65500
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.105927000781337e-06,
+      "loss": 0.0534,
+      "step": 65510
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.102578412769283e-06,
+      "loss": 0.0325,
+      "step": 65520
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.099229824757227e-06,
+      "loss": 0.0613,
+      "step": 65530
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.095881236745173e-06,
+      "loss": 0.0952,
+      "step": 65540
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.092532648733117e-06,
+      "loss": 0.0331,
+      "step": 65550
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.089184060721063e-06,
+      "loss": 0.1044,
+      "step": 65560
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.085835472709009e-06,
+      "loss": 0.4136,
+      "step": 65570
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.082486884696953e-06,
+      "loss": 0.0126,
+      "step": 65580
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.079138296684898e-06,
+      "loss": 0.1831,
+      "step": 65590
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.075789708672843e-06,
+      "loss": 0.0685,
+      "step": 65600
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.072441120660788e-06,
+      "loss": 0.0309,
+      "step": 65610
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.069092532648734e-06,
+      "loss": 0.1082,
+      "step": 65620
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.065743944636678e-06,
+      "loss": 0.0684,
+      "step": 65630
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.062395356624624e-06,
+      "loss": 0.0906,
+      "step": 65640
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.059046768612568e-06,
+      "loss": 0.0604,
+      "step": 65650
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.055698180600514e-06,
+      "loss": 0.0674,
+      "step": 65660
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.05234959258846e-06,
+      "loss": 0.0559,
+      "step": 65670
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.049001004576404e-06,
+      "loss": 0.0599,
+      "step": 65680
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.04565241656435e-06,
+      "loss": 0.1849,
+      "step": 65690
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.042303828552295e-06,
+      "loss": 0.0214,
+      "step": 65700
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.03895524054024e-06,
+      "loss": 0.0564,
+      "step": 65710
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.035606652528183e-06,
+      "loss": 0.069,
+      "step": 65720
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.032258064516129e-06,
+      "loss": 0.0368,
+      "step": 65730
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.028909476504075e-06,
+      "loss": 0.1036,
+      "step": 65740
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.025560888492019e-06,
+      "loss": 0.0473,
+      "step": 65750
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.022212300479965e-06,
+      "loss": 0.0806,
+      "step": 65760
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.01886371246791e-06,
+      "loss": 0.0558,
+      "step": 65770
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.015515124455854e-06,
+      "loss": 0.0862,
+      "step": 65780
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.012166536443799e-06,
+      "loss": 0.1107,
+      "step": 65790
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.008817948431746e-06,
+      "loss": 0.0477,
+      "step": 65800
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.00546936041969e-06,
+      "loss": 0.0519,
+      "step": 65810
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.002120772407634e-06,
+      "loss": 0.0563,
+      "step": 65820
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 7.998772184395582e-06,
+      "loss": 0.1274,
+      "step": 65830
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 7.995423596383526e-06,
+      "loss": 0.1214,
+      "step": 65840
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 7.99207500837147e-06,
+      "loss": 0.0489,
+      "step": 65850
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 7.988726420359414e-06,
+      "loss": 0.0433,
+      "step": 65860
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 7.985377832347361e-06,
+      "loss": 0.0291,
+      "step": 65870
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 7.982029244335305e-06,
+      "loss": 0.0357,
+      "step": 65880
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 7.97868065632325e-06,
+      "loss": 0.0429,
+      "step": 65890
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 7.975332068311197e-06,
+      "loss": 0.0448,
+      "step": 65900
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 7.971983480299141e-06,
+      "loss": 0.116,
+      "step": 65910
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 7.968634892287085e-06,
+      "loss": 0.0673,
+      "step": 65920
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 7.96528630427503e-06,
+      "loss": 0.0455,
+      "step": 65930
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 7.961937716262977e-06,
+      "loss": 0.0721,
+      "step": 65940
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 7.95858912825092e-06,
+      "loss": 0.0896,
+      "step": 65950
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 7.955240540238866e-06,
+      "loss": 0.0309,
+      "step": 65960
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 7.951891952226812e-06,
+      "loss": 0.0337,
+      "step": 65970
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 7.948543364214756e-06,
+      "loss": 0.0602,
+      "step": 65980
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 7.9451947762027e-06,
+      "loss": 0.0541,
+      "step": 65990
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 7.941846188190646e-06,
+      "loss": 0.0569,
+      "step": 66000
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 7.938497600178592e-06,
+      "loss": 0.0566,
+      "step": 66010
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 7.935149012166536e-06,
+      "loss": 0.0566,
+      "step": 66020
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 7.931800424154482e-06,
+      "loss": 0.0894,
+      "step": 66030
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 7.928451836142428e-06,
+      "loss": 0.0562,
+      "step": 66040
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 7.925103248130372e-06,
+      "loss": 0.0503,
+      "step": 66050
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 7.921754660118317e-06,
+      "loss": 0.1095,
+      "step": 66060
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 7.918406072106263e-06,
+      "loss": 0.0446,
+      "step": 66070
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 7.915057484094207e-06,
+      "loss": 0.0571,
+      "step": 66080
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 7.911708896082151e-06,
+      "loss": 0.0282,
+      "step": 66090
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 7.908360308070097e-06,
+      "loss": 0.0714,
+      "step": 66100
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 7.905011720058043e-06,
+      "loss": 0.0557,
+      "step": 66110
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 7.901663132045987e-06,
+      "loss": 0.0385,
+      "step": 66120
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 7.898314544033933e-06,
+      "loss": 0.0702,
+      "step": 66130
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 7.894965956021878e-06,
+      "loss": 0.0868,
+      "step": 66140
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 7.891617368009822e-06,
+      "loss": 0.0559,
+      "step": 66150
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 7.888268779997768e-06,
+      "loss": 0.0596,
+      "step": 66160
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 7.884920191985712e-06,
+      "loss": 0.032,
+      "step": 66170
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 7.881571603973658e-06,
+      "loss": 0.0692,
+      "step": 66180
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 7.878223015961604e-06,
+      "loss": 0.0474,
+      "step": 66190
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 7.874874427949548e-06,
+      "loss": 0.0574,
+      "step": 66200
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 7.871525839937494e-06,
+      "loss": 0.0419,
+      "step": 66210
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 7.868177251925438e-06,
+      "loss": 0.0683,
+      "step": 66220
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 7.864828663913384e-06,
+      "loss": 0.0648,
+      "step": 66230
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 7.861480075901328e-06,
+      "loss": 0.0169,
+      "step": 66240
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 7.858131487889273e-06,
+      "loss": 0.0731,
+      "step": 66250
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 7.85478289987722e-06,
+      "loss": 0.0276,
+      "step": 66260
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 7.851434311865163e-06,
+      "loss": 0.0372,
+      "step": 66270
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 7.848085723853109e-06,
+      "loss": 0.0533,
+      "step": 66280
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 7.844737135841055e-06,
+      "loss": 0.0513,
+      "step": 66290
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 7.841388547828999e-06,
+      "loss": 0.0412,
+      "step": 66300
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 7.838039959816943e-06,
+      "loss": 0.0505,
+      "step": 66310
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 7.83469137180489e-06,
+      "loss": 0.0457,
+      "step": 66320
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 7.831342783792834e-06,
+      "loss": 0.0749,
+      "step": 66330
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 7.827994195780779e-06,
+      "loss": 0.1401,
+      "step": 66340
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 7.824645607768724e-06,
+      "loss": 0.0737,
+      "step": 66350
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 7.82129701975667e-06,
+      "loss": 0.0602,
+      "step": 66360
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 7.817948431744614e-06,
+      "loss": 0.0767,
+      "step": 66370
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 7.814599843732558e-06,
+      "loss": 0.0588,
+      "step": 66380
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 7.811251255720506e-06,
+      "loss": 0.0526,
+      "step": 66390
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 7.80790266770845e-06,
+      "loss": 0.0887,
+      "step": 66400
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 7.804554079696394e-06,
+      "loss": 0.0335,
+      "step": 66410
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 7.801205491684341e-06,
+      "loss": 0.0351,
+      "step": 66420
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 7.797856903672285e-06,
+      "loss": 0.0821,
+      "step": 66430
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 7.79450831566023e-06,
+      "loss": 0.0953,
+      "step": 66440
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 7.791159727648175e-06,
+      "loss": 0.1002,
+      "step": 66450
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 7.787811139636121e-06,
+      "loss": 0.0548,
+      "step": 66460
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 7.784462551624065e-06,
+      "loss": 0.0379,
+      "step": 66470
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 7.781113963612009e-06,
+      "loss": 0.0449,
+      "step": 66480
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 7.777765375599957e-06,
+      "loss": 0.1108,
+      "step": 66490
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 7.7744167875879e-06,
+      "loss": 0.0489,
+      "step": 66500
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 7.771068199575845e-06,
+      "loss": 0.073,
+      "step": 66510
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 7.767719611563792e-06,
+      "loss": 0.0378,
+      "step": 66520
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 7.764371023551736e-06,
+      "loss": 0.1164,
+      "step": 66530
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 7.76102243553968e-06,
+      "loss": 0.074,
+      "step": 66540
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 7.757673847527626e-06,
+      "loss": 0.0707,
+      "step": 66550
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 7.754325259515572e-06,
+      "loss": 0.0839,
+      "step": 66560
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 7.750976671503516e-06,
+      "loss": 0.0776,
+      "step": 66570
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 7.747628083491462e-06,
+      "loss": 0.0732,
+      "step": 66580
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 7.744279495479407e-06,
+      "loss": 0.0432,
+      "step": 66590
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 7.740930907467352e-06,
+      "loss": 0.0313,
+      "step": 66600
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 7.737582319455296e-06,
+      "loss": 0.1051,
+      "step": 66610
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 7.734233731443241e-06,
+      "loss": 0.0827,
+      "step": 66620
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 7.730885143431187e-06,
+      "loss": 0.0922,
+      "step": 66630
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 7.727536555419131e-06,
+      "loss": 0.0565,
+      "step": 66640
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 7.724187967407077e-06,
+      "loss": 0.0554,
+      "step": 66650
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 7.720839379395023e-06,
+      "loss": 0.1028,
+      "step": 66660
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 7.717490791382967e-06,
+      "loss": 0.0929,
+      "step": 66670
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 7.714142203370913e-06,
+      "loss": 0.0315,
+      "step": 66680
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 7.710793615358857e-06,
+      "loss": 0.0987,
+      "step": 66690
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 7.707445027346802e-06,
+      "loss": 0.0748,
+      "step": 66700
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 7.704096439334748e-06,
+      "loss": 0.0266,
+      "step": 66710
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 7.700747851322692e-06,
+      "loss": 0.0569,
+      "step": 66720
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 7.697399263310638e-06,
+      "loss": 0.0917,
+      "step": 66730
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 7.694050675298582e-06,
+      "loss": 0.0798,
+      "step": 66740
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 7.690702087286528e-06,
+      "loss": 0.0624,
+      "step": 66750
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 7.687353499274472e-06,
+      "loss": 0.0293,
+      "step": 66760
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 7.684004911262418e-06,
+      "loss": 0.0525,
+      "step": 66770
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 7.680656323250364e-06,
+      "loss": 0.0953,
+      "step": 66780
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 7.677307735238308e-06,
+      "loss": 0.0102,
+      "step": 66790
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 7.673959147226253e-06,
+      "loss": 0.0585,
+      "step": 66800
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 7.670610559214199e-06,
+      "loss": 0.1299,
+      "step": 66810
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 7.667261971202143e-06,
+      "loss": 0.0437,
+      "step": 66820
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 7.663913383190087e-06,
+      "loss": 0.0711,
+      "step": 66830
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 7.660564795178033e-06,
+      "loss": 0.0329,
+      "step": 66840
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 7.657216207165979e-06,
+      "loss": 0.0507,
+      "step": 66850
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 7.653867619153923e-06,
+      "loss": 0.0838,
+      "step": 66860
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 7.650519031141869e-06,
+      "loss": 0.0872,
+      "step": 66870
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 7.647170443129814e-06,
+      "loss": 0.0941,
+      "step": 66880
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 7.643821855117759e-06,
+      "loss": 0.0956,
+      "step": 66890
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 7.640473267105703e-06,
+      "loss": 0.1733,
+      "step": 66900
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 7.63712467909365e-06,
+      "loss": 0.0615,
+      "step": 66910
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 7.633776091081594e-06,
+      "loss": 0.0606,
+      "step": 66920
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 7.630427503069538e-06,
+      "loss": 0.0436,
+      "step": 66930
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 7.627078915057485e-06,
+      "loss": 0.0478,
+      "step": 66940
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 7.6240651858466355e-06,
+      "loss": 0.068,
+      "step": 66950
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 7.6207165978345796e-06,
+      "loss": 0.077,
+      "step": 66960
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 7.6173680098225245e-06,
+      "loss": 0.0716,
+      "step": 66970
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 7.614019421810471e-06,
+      "loss": 0.0556,
+      "step": 66980
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 7.610670833798415e-06,
+      "loss": 0.0632,
+      "step": 66990
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 7.60732224578636e-06,
+      "loss": 0.0865,
+      "step": 67000
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 7.603973657774305e-06,
+      "loss": 0.0395,
+      "step": 67010
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 7.600625069762251e-06,
+      "loss": 0.0466,
+      "step": 67020
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 7.597276481750196e-06,
+      "loss": 0.0387,
+      "step": 67030
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 7.59392789373814e-06,
+      "loss": 0.1425,
+      "step": 67040
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 7.590579305726086e-06,
+      "loss": 0.064,
+      "step": 67050
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 7.587230717714031e-06,
+      "loss": 0.0535,
+      "step": 67060
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 7.583882129701975e-06,
+      "loss": 0.0393,
+      "step": 67070
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 7.58053354168992e-06,
+      "loss": 0.0625,
+      "step": 67080
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 7.577184953677866e-06,
+      "loss": 0.014,
+      "step": 67090
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 7.573836365665811e-06,
+      "loss": 0.0791,
+      "step": 67100
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 7.570487777653756e-06,
+      "loss": 0.0316,
+      "step": 67110
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 7.567139189641702e-06,
+      "loss": 0.0589,
+      "step": 67120
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 7.563790601629647e-06,
+      "loss": 0.0508,
+      "step": 67130
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 7.5604420136175915e-06,
+      "loss": 0.1119,
+      "step": 67140
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 7.557093425605536e-06,
+      "loss": 0.0218,
+      "step": 67150
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 7.553744837593482e-06,
+      "loss": 0.0311,
+      "step": 67160
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 7.550396249581426e-06,
+      "loss": 0.0325,
+      "step": 67170
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 7.547047661569371e-06,
+      "loss": 0.0521,
+      "step": 67180
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 7.543699073557318e-06,
+      "loss": 0.0593,
+      "step": 67190
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 7.540350485545262e-06,
+      "loss": 0.0259,
+      "step": 67200
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 7.537001897533207e-06,
+      "loss": 0.033,
+      "step": 67210
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 7.533653309521151e-06,
+      "loss": 0.0675,
+      "step": 67220
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 7.5303047215090975e-06,
+      "loss": 0.1234,
+      "step": 67230
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 7.5269561334970424e-06,
+      "loss": 0.1484,
+      "step": 67240
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 7.5236075454849865e-06,
+      "loss": 0.0723,
+      "step": 67250
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 7.520258957472933e-06,
+      "loss": 0.06,
+      "step": 67260
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 7.516910369460878e-06,
+      "loss": 0.1104,
+      "step": 67270
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 7.513561781448822e-06,
+      "loss": 0.0417,
+      "step": 67280
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 7.510213193436769e-06,
+      "loss": 0.0583,
+      "step": 67290
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 7.506864605424713e-06,
+      "loss": 0.0421,
+      "step": 67300
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 7.503516017412658e-06,
+      "loss": 0.0119,
+      "step": 67310
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 7.500167429400603e-06,
+      "loss": 0.0518,
+      "step": 67320
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 7.4968188413885476e-06,
+      "loss": 0.0902,
+      "step": 67330
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 7.493470253376493e-06,
+      "loss": 0.069,
+      "step": 67340
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 7.490121665364438e-06,
+      "loss": 0.0517,
+      "step": 67350
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 7.486773077352383e-06,
+      "loss": 0.0597,
+      "step": 67360
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 7.483424489340329e-06,
+      "loss": 0.0331,
+      "step": 67370
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 7.480075901328273e-06,
+      "loss": 0.0555,
+      "step": 67380
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 7.476727313316219e-06,
+      "loss": 0.0517,
+      "step": 67390
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 7.473378725304163e-06,
+      "loss": 0.0102,
+      "step": 67400
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 7.470030137292109e-06,
+      "loss": 0.036,
+      "step": 67410
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 7.466681549280054e-06,
+      "loss": 0.0421,
+      "step": 67420
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 7.4633329612679985e-06,
+      "loss": 0.0257,
+      "step": 67430
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 7.459984373255944e-06,
+      "loss": 0.1085,
+      "step": 67440
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 7.456635785243889e-06,
+      "loss": 0.0529,
+      "step": 67450
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 7.453287197231834e-06,
+      "loss": 0.0629,
+      "step": 67460
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 7.449938609219779e-06,
+      "loss": 0.0592,
+      "step": 67470
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 7.446590021207724e-06,
+      "loss": 0.0794,
+      "step": 67480
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 7.44324143319567e-06,
+      "loss": 0.0508,
+      "step": 67490
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 7.439892845183615e-06,
+      "loss": 0.0604,
+      "step": 67500
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 7.4365442571715595e-06,
+      "loss": 0.0805,
+      "step": 67510
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 7.4331956691595045e-06,
+      "loss": 0.0385,
+      "step": 67520
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 7.429847081147449e-06,
+      "loss": 0.0429,
+      "step": 67530
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 7.4268333519366e-06,
+      "loss": 0.1241,
+      "step": 67540
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 7.423484763924546e-06,
+      "loss": 0.0229,
+      "step": 67550
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 7.42013617591249e-06,
+      "loss": 0.0705,
+      "step": 67560
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 7.416787587900436e-06,
+      "loss": 0.0445,
+      "step": 67570
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 7.413438999888381e-06,
+      "loss": 0.0642,
+      "step": 67580
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 7.4100904118763255e-06,
+      "loss": 0.0378,
+      "step": 67590
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 7.4067418238642704e-06,
+      "loss": 0.0418,
+      "step": 67600
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 7.403393235852215e-06,
+      "loss": 0.0365,
+      "step": 67610
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 7.400044647840161e-06,
+      "loss": 0.065,
+      "step": 67620
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 7.396696059828106e-06,
+      "loss": 0.0425,
+      "step": 67630
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 7.393347471816051e-06,
+      "loss": 0.0226,
+      "step": 67640
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 7.389998883803996e-06,
+      "loss": 0.0546,
+      "step": 67650
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 7.386650295791942e-06,
+      "loss": 0.0499,
+      "step": 67660
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 7.383301707779887e-06,
+      "loss": 0.0617,
+      "step": 67670
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.3799531197678315e-06,
+      "loss": 0.0514,
+      "step": 67680
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.376604531755776e-06,
+      "loss": 0.0648,
+      "step": 67690
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.373255943743721e-06,
+      "loss": 0.0175,
+      "step": 67700
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.369907355731667e-06,
+      "loss": 0.0334,
+      "step": 67710
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.366558767719611e-06,
+      "loss": 0.0383,
+      "step": 67720
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.363210179707557e-06,
+      "loss": 0.0301,
+      "step": 67730
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.359861591695502e-06,
+      "loss": 0.0523,
+      "step": 67740
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.356513003683447e-06,
+      "loss": 0.0271,
+      "step": 67750
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.3531644156713926e-06,
+      "loss": 0.077,
+      "step": 67760
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.349815827659337e-06,
+      "loss": 0.0658,
+      "step": 67770
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.346467239647282e-06,
+      "loss": 0.0625,
+      "step": 67780
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.343118651635227e-06,
+      "loss": 0.1057,
+      "step": 67790
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.339770063623172e-06,
+      "loss": 0.0347,
+      "step": 67800
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.336421475611118e-06,
+      "loss": 0.0324,
+      "step": 67810
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.333072887599062e-06,
+      "loss": 0.0422,
+      "step": 67820
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.329724299587008e-06,
+      "loss": 0.056,
+      "step": 67830
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.326375711574953e-06,
+      "loss": 0.0477,
+      "step": 67840
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.323027123562898e-06,
+      "loss": 0.0917,
+      "step": 67850
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.3196785355508435e-06,
+      "loss": 0.092,
+      "step": 67860
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.3163299475387875e-06,
+      "loss": 0.0193,
+      "step": 67870
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.312981359526733e-06,
+      "loss": 0.0596,
+      "step": 67880
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.309632771514678e-06,
+      "loss": 0.0486,
+      "step": 67890
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.306284183502623e-06,
+      "loss": 0.0584,
+      "step": 67900
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.302935595490568e-06,
+      "loss": 0.0687,
+      "step": 67910
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.299587007478514e-06,
+      "loss": 0.0556,
+      "step": 67920
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.296238419466459e-06,
+      "loss": 0.0458,
+      "step": 67930
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.292889831454404e-06,
+      "loss": 0.1105,
+      "step": 67940
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 7.289541243442349e-06,
+      "loss": 0.0743,
+      "step": 67950
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 7.2861926554302935e-06,
+      "loss": 0.0212,
+      "step": 67960
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 7.282844067418239e-06,
+      "loss": 0.0743,
+      "step": 67970
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 7.279495479406183e-06,
+      "loss": 0.053,
+      "step": 67980
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 7.276146891394129e-06,
+      "loss": 0.0585,
+      "step": 67990
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 7.272798303382074e-06,
+      "loss": 0.0453,
+      "step": 68000
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 7.269449715370019e-06,
+      "loss": 0.0586,
+      "step": 68010
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 7.266101127357965e-06,
+      "loss": 0.0811,
+      "step": 68020
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.262752539345909e-06,
+      "loss": 0.087,
+      "step": 68030
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.259403951333855e-06,
+      "loss": 0.1565,
+      "step": 68040
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.256055363321799e-06,
+      "loss": 0.0566,
+      "step": 68050
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.2527067753097444e-06,
+      "loss": 0.0887,
+      "step": 68060
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.24935818729769e-06,
+      "loss": 0.0366,
+      "step": 68070
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.246009599285634e-06,
+      "loss": 0.0546,
+      "step": 68080
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.24266101127358e-06,
+      "loss": 0.0539,
+      "step": 68090
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.239312423261525e-06,
+      "loss": 0.045,
+      "step": 68100
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.23596383524947e-06,
+      "loss": 0.0256,
+      "step": 68110
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.232615247237416e-06,
+      "loss": 0.0418,
+      "step": 68120
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.22926665922536e-06,
+      "loss": 0.0384,
+      "step": 68130
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.2259180712133055e-06,
+      "loss": 0.0509,
+      "step": 68140
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.22256948320125e-06,
+      "loss": 0.0378,
+      "step": 68150
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.219220895189195e-06,
+      "loss": 0.0308,
+      "step": 68160
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.21587230717714e-06,
+      "loss": 0.0668,
+      "step": 68170
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.212523719165085e-06,
+      "loss": 0.0402,
+      "step": 68180
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.209175131153031e-06,
+      "loss": 0.0511,
+      "step": 68190
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.205826543140976e-06,
+      "loss": 0.1119,
+      "step": 68200
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.202477955128921e-06,
+      "loss": 0.0631,
+      "step": 68210
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.199129367116866e-06,
+      "loss": 0.126,
+      "step": 68220
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.1957807791048115e-06,
+      "loss": 0.0962,
+      "step": 68230
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.1924321910927556e-06,
+      "loss": 0.0449,
+      "step": 68240
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.189083603080701e-06,
+      "loss": 0.0669,
+      "step": 68250
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.185735015068646e-06,
+      "loss": 0.0327,
+      "step": 68260
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.182386427056591e-06,
+      "loss": 0.0084,
+      "step": 68270
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.179037839044537e-06,
+      "loss": 0.09,
+      "step": 68280
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.175689251032481e-06,
+      "loss": 0.0341,
+      "step": 68290
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.172340663020427e-06,
+      "loss": 0.048,
+      "step": 68300
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.168992075008372e-06,
+      "loss": 0.036,
+      "step": 68310
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.165643486996317e-06,
+      "loss": 0.0726,
+      "step": 68320
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.162294898984262e-06,
+      "loss": 0.0547,
+      "step": 68330
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.1589463109722065e-06,
+      "loss": 0.0649,
+      "step": 68340
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.155597722960152e-06,
+      "loss": 0.0675,
+      "step": 68350
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.152249134948097e-06,
+      "loss": 0.0897,
+      "step": 68360
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.148900546936042e-06,
+      "loss": 0.0291,
+      "step": 68370
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.145551958923988e-06,
+      "loss": 0.0874,
+      "step": 68380
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.142203370911932e-06,
+      "loss": 0.0789,
+      "step": 68390
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.138854782899878e-06,
+      "loss": 0.0595,
+      "step": 68400
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.135506194887823e-06,
+      "loss": 0.0299,
+      "step": 68410
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.1321576068757675e-06,
+      "loss": 0.0204,
+      "step": 68420
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.1288090188637124e-06,
+      "loss": 0.0521,
+      "step": 68430
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.125460430851657e-06,
+      "loss": 0.0898,
+      "step": 68440
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.122111842839603e-06,
+      "loss": 0.0428,
+      "step": 68450
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.118763254827548e-06,
+      "loss": 0.0345,
+      "step": 68460
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.115414666815493e-06,
+      "loss": 0.0274,
+      "step": 68470
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.112066078803438e-06,
+      "loss": 0.0353,
+      "step": 68480
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.108717490791383e-06,
+      "loss": 0.1132,
+      "step": 68490
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.105368902779328e-06,
+      "loss": 0.0538,
+      "step": 68500
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.1020203147672735e-06,
+      "loss": 0.0436,
+      "step": 68510
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.0986717267552184e-06,
+      "loss": 0.0332,
+      "step": 68520
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.095323138743163e-06,
+      "loss": 0.043,
+      "step": 68530
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.091974550731109e-06,
+      "loss": 0.0635,
+      "step": 68540
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.088625962719053e-06,
+      "loss": 0.0311,
+      "step": 68550
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.085277374706999e-06,
+      "loss": 0.0775,
+      "step": 68560
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.081928786694944e-06,
+      "loss": 0.0473,
+      "step": 68570
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.078580198682889e-06,
+      "loss": 0.0422,
+      "step": 68580
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.0752316106708346e-06,
+      "loss": 0.1417,
+      "step": 68590
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.071883022658779e-06,
+      "loss": 0.0648,
+      "step": 68600
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.068534434646724e-06,
+      "loss": 0.0962,
+      "step": 68610
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.0651858466346685e-06,
+      "loss": 0.0586,
+      "step": 68620
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.061837258622614e-06,
+      "loss": 0.0347,
+      "step": 68630
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.05848867061056e-06,
+      "loss": 0.0229,
+      "step": 68640
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.055140082598504e-06,
+      "loss": 0.0475,
+      "step": 68650
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.05179149458645e-06,
+      "loss": 0.0582,
+      "step": 68660
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.048442906574395e-06,
+      "loss": 0.0994,
+      "step": 68670
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.04509431856234e-06,
+      "loss": 0.13,
+      "step": 68680
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.041745730550285e-06,
+      "loss": 0.0476,
+      "step": 68690
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.0383971425382295e-06,
+      "loss": 0.0585,
+      "step": 68700
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.035048554526175e-06,
+      "loss": 0.0453,
+      "step": 68710
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.03169996651412e-06,
+      "loss": 0.0201,
+      "step": 68720
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.028351378502065e-06,
+      "loss": 0.0766,
+      "step": 68730
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.02500279049001e-06,
+      "loss": 0.0254,
+      "step": 68740
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.021654202477955e-06,
+      "loss": 0.0792,
+      "step": 68750
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.018305614465901e-06,
+      "loss": 0.0453,
+      "step": 68760
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.014957026453846e-06,
+      "loss": 0.0591,
+      "step": 68770
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.011608438441791e-06,
+      "loss": 0.0506,
+      "step": 68780
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.0082598504297355e-06,
+      "loss": 0.0893,
+      "step": 68790
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.0049112624176805e-06,
+      "loss": 0.0842,
+      "step": 68800
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.001562674405625e-06,
+      "loss": 0.1315,
+      "step": 68810
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 6.998214086393571e-06,
+      "loss": 0.0329,
+      "step": 68820
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 6.994865498381516e-06,
+      "loss": 0.0765,
+      "step": 68830
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 6.991516910369461e-06,
+      "loss": 0.0729,
+      "step": 68840
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 6.988168322357407e-06,
+      "loss": 0.0531,
+      "step": 68850
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 6.984819734345351e-06,
+      "loss": 0.1451,
+      "step": 68860
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 6.981471146333297e-06,
+      "loss": 0.0364,
+      "step": 68870
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 6.978122558321241e-06,
+      "loss": 0.0097,
+      "step": 68880
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 6.9747739703091864e-06,
+      "loss": 0.0968,
+      "step": 68890
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 6.971425382297132e-06,
+      "loss": 0.0691,
+      "step": 68900
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 6.968076794285076e-06,
+      "loss": 0.0674,
+      "step": 68910
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 6.964728206273022e-06,
+      "loss": 0.0508,
+      "step": 68920
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 6.961379618260966e-06,
+      "loss": 0.0965,
+      "step": 68930
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 6.958031030248912e-06,
+      "loss": 0.0476,
+      "step": 68940
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 6.954682442236857e-06,
+      "loss": 0.0385,
+      "step": 68950
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 6.951333854224802e-06,
+      "loss": 0.0138,
+      "step": 68960
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 6.9479852662127475e-06,
+      "loss": 0.0843,
+      "step": 68970
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 6.944636678200692e-06,
+      "loss": 0.1912,
+      "step": 68980
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 6.941288090188637e-06,
+      "loss": 0.1549,
+      "step": 68990
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 6.937939502176582e-06,
+      "loss": 0.0479,
+      "step": 69000
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 6.934590914164527e-06,
+      "loss": 0.1154,
+      "step": 69010
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 6.931242326152473e-06,
+      "loss": 0.0716,
+      "step": 69020
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 6.927893738140418e-06,
+      "loss": 0.0609,
+      "step": 69030
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 6.924545150128363e-06,
+      "loss": 0.0942,
+      "step": 69040
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 6.921196562116308e-06,
+      "loss": 0.0728,
+      "step": 69050
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 6.917847974104253e-06,
+      "loss": 0.0579,
+      "step": 69060
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 6.9144993860921976e-06,
+      "loss": 0.0437,
+      "step": 69070
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 6.911150798080143e-06,
+      "loss": 0.0354,
+      "step": 69080
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 6.907802210068088e-06,
+      "loss": 0.0434,
+      "step": 69090
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 6.904453622056033e-06,
+      "loss": 0.0728,
+      "step": 69100
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 6.901105034043979e-06,
+      "loss": 0.0412,
+      "step": 69110
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 6.897756446031923e-06,
+      "loss": 0.0794,
+      "step": 69120
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 6.894407858019869e-06,
+      "loss": 0.1144,
+      "step": 69130
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 6.891059270007813e-06,
+      "loss": 0.0448,
+      "step": 69140
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 6.887710681995759e-06,
+      "loss": 0.0315,
+      "step": 69150
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 6.884362093983704e-06,
+      "loss": 0.0433,
+      "step": 69160
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 6.8810135059716485e-06,
+      "loss": 0.0968,
+      "step": 69170
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 6.877664917959594e-06,
+      "loss": 0.0377,
+      "step": 69180
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 6.874316329947538e-06,
+      "loss": 0.0803,
+      "step": 69190
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 6.870967741935484e-06,
+      "loss": 0.1229,
+      "step": 69200
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 6.86761915392343e-06,
+      "loss": 0.0845,
+      "step": 69210
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 6.864270565911374e-06,
+      "loss": 0.0508,
+      "step": 69220
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 6.86092197789932e-06,
+      "loss": 0.071,
+      "step": 69230
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 6.857573389887264e-06,
+      "loss": 0.0583,
+      "step": 69240
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 6.8542248018752095e-06,
+      "loss": 0.0545,
+      "step": 69250
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 6.8508762138631544e-06,
+      "loss": 0.0277,
+      "step": 69260
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 6.847527625851099e-06,
+      "loss": 0.081,
+      "step": 69270
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 6.844179037839045e-06,
+      "loss": 0.0364,
+      "step": 69280
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 6.84083044982699e-06,
+      "loss": 0.0863,
+      "step": 69290
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 6.837481861814935e-06,
+      "loss": 0.0629,
+      "step": 69300
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 6.83413327380288e-06,
+      "loss": 0.0277,
+      "step": 69310
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 6.830784685790825e-06,
+      "loss": 0.0402,
+      "step": 69320
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 6.82743609777877e-06,
+      "loss": 0.0323,
+      "step": 69330
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 6.8240875097667155e-06,
+      "loss": 0.0791,
+      "step": 69340
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 6.8207389217546604e-06,
+      "loss": 0.1141,
+      "step": 69350
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 6.817390333742605e-06,
+      "loss": 0.0538,
+      "step": 69360
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 6.81404174573055e-06,
+      "loss": 0.0437,
+      "step": 69370
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 6.810693157718495e-06,
+      "loss": 0.1441,
+      "step": 69380
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 6.807344569706441e-06,
+      "loss": 0.0893,
+      "step": 69390
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 6.803995981694385e-06,
+      "loss": 0.0522,
+      "step": 69400
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 6.800647393682331e-06,
+      "loss": 0.0556,
+      "step": 69410
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 6.7972988056702766e-06,
+      "loss": 0.0526,
+      "step": 69420
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 6.793950217658221e-06,
+      "loss": 0.021,
+      "step": 69430
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 6.790601629646166e-06,
+      "loss": 0.0562,
+      "step": 69440
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 6.7872530416341105e-06,
+      "loss": 0.0959,
+      "step": 69450
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 6.783904453622056e-06,
+      "loss": 0.0306,
+      "step": 69460
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 6.780555865610002e-06,
+      "loss": 0.0301,
+      "step": 69470
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 6.777207277597946e-06,
+      "loss": 0.1072,
+      "step": 69480
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 6.773858689585892e-06,
+      "loss": 0.0627,
+      "step": 69490
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 6.770510101573836e-06,
+      "loss": 0.0401,
+      "step": 69500
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 6.767161513561782e-06,
+      "loss": 0.0562,
+      "step": 69510
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 6.763812925549727e-06,
+      "loss": 0.188,
+      "step": 69520
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 6.7604643375376716e-06,
+      "loss": 0.0248,
+      "step": 69530
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 6.757115749525617e-06,
+      "loss": 0.0553,
+      "step": 69540
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 6.753767161513562e-06,
+      "loss": 0.016,
+      "step": 69550
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 6.750418573501507e-06,
+      "loss": 0.0179,
+      "step": 69560
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 6.747069985489452e-06,
+      "loss": 0.098,
+      "step": 69570
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 6.743721397477397e-06,
+      "loss": 0.0934,
+      "step": 69580
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 6.740372809465342e-06,
+      "loss": 0.0812,
+      "step": 69590
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 6.737024221453288e-06,
+      "loss": 0.0907,
+      "step": 69600
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 6.733675633441233e-06,
+      "loss": 0.0519,
+      "step": 69610
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 6.7303270454291775e-06,
+      "loss": 0.0403,
+      "step": 69620
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 6.7269784574171225e-06,
+      "loss": 0.1292,
+      "step": 69630
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 6.723629869405067e-06,
+      "loss": 0.0592,
+      "step": 69640
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 6.720281281393013e-06,
+      "loss": 0.0614,
+      "step": 69650
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 6.716932693380958e-06,
+      "loss": 0.0817,
+      "step": 69660
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 6.713584105368903e-06,
+      "loss": 0.046,
+      "step": 69670
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 6.710235517356848e-06,
+      "loss": 0.0459,
+      "step": 69680
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 6.706886929344793e-06,
+      "loss": 0.0714,
+      "step": 69690
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 6.703538341332739e-06,
+      "loss": 0.0205,
+      "step": 69700
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 6.700189753320683e-06,
+      "loss": 0.0479,
+      "step": 69710
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 6.6968411653086284e-06,
+      "loss": 0.0796,
+      "step": 69720
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 6.693492577296574e-06,
+      "loss": 0.085,
+      "step": 69730
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 6.690143989284518e-06,
+      "loss": 0.1176,
+      "step": 69740
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 6.686795401272464e-06,
+      "loss": 0.0242,
+      "step": 69750
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 6.683446813260408e-06,
+      "loss": 0.0116,
+      "step": 69760
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 6.680098225248354e-06,
+      "loss": 0.0702,
+      "step": 69770
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 6.676749637236299e-06,
+      "loss": 0.099,
+      "step": 69780
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 6.673401049224244e-06,
+      "loss": 0.0767,
+      "step": 69790
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 6.6700524612121895e-06,
+      "loss": 0.0415,
+      "step": 69800
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 6.666703873200134e-06,
+      "loss": 0.1033,
+      "step": 69810
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 6.663355285188079e-06,
+      "loss": 0.0504,
+      "step": 69820
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 6.660006697176024e-06,
+      "loss": 0.0609,
+      "step": 69830
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 6.656658109163969e-06,
+      "loss": 0.1175,
+      "step": 69840
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 6.653309521151914e-06,
+      "loss": 0.0374,
+      "step": 69850
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 6.64996093313986e-06,
+      "loss": 0.0385,
+      "step": 69860
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 6.646612345127805e-06,
+      "loss": 0.0827,
+      "step": 69870
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 6.64326375711575e-06,
+      "loss": 0.0593,
+      "step": 69880
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 6.639915169103695e-06,
+      "loss": 0.0198,
+      "step": 69890
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 6.6365665810916396e-06,
+      "loss": 0.0673,
+      "step": 69900
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 6.633217993079585e-06,
+      "loss": 0.0426,
+      "step": 69910
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 6.62986940506753e-06,
+      "loss": 0.041,
+      "step": 69920
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 6.626520817055475e-06,
+      "loss": 0.102,
+      "step": 69930
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 6.62317222904342e-06,
+      "loss": 0.0735,
+      "step": 69940
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 6.619823641031365e-06,
+      "loss": 0.0344,
+      "step": 69950
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 6.616475053019311e-06,
+      "loss": 0.0809,
+      "step": 69960
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 6.613126465007255e-06,
+      "loss": 0.0567,
+      "step": 69970
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 6.609777876995201e-06,
+      "loss": 0.1191,
+      "step": 69980
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 6.6064292889831455e-06,
+      "loss": 0.0566,
+      "step": 69990
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 6.6030807009710905e-06,
+      "loss": 0.0771,
+      "step": 70000
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 6.599732112959036e-06,
+      "loss": 0.0836,
+      "step": 70010
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 6.59638352494698e-06,
+      "loss": 0.0896,
+      "step": 70020
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 6.593034936934926e-06,
+      "loss": 0.0866,
+      "step": 70030
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 6.589686348922871e-06,
+      "loss": 0.0577,
+      "step": 70040
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 6.586337760910816e-06,
+      "loss": 0.1056,
+      "step": 70050
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 6.582989172898762e-06,
+      "loss": 0.0326,
+      "step": 70060
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 6.579640584886706e-06,
+      "loss": 0.0242,
+      "step": 70070
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 6.5762919968746515e-06,
+      "loss": 0.0345,
+      "step": 70080
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 6.5729434088625965e-06,
+      "loss": 0.1056,
+      "step": 70090
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 6.569594820850541e-06,
+      "loss": 0.1871,
+      "step": 70100
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 6.566246232838487e-06,
+      "loss": 0.0677,
+      "step": 70110
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 6.562897644826431e-06,
+      "loss": 0.0398,
+      "step": 70120
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 6.559549056814377e-06,
+      "loss": 0.0917,
+      "step": 70130
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 6.556200468802322e-06,
+      "loss": 0.0352,
+      "step": 70140
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 6.552851880790267e-06,
+      "loss": 0.1391,
+      "step": 70150
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 6.549503292778212e-06,
+      "loss": 0.076,
+      "step": 70160
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 6.5461547047661575e-06,
+      "loss": 0.0359,
+      "step": 70170
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 6.5428061167541024e-06,
+      "loss": 0.068,
+      "step": 70180
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 6.539457528742047e-06,
+      "loss": 0.1034,
+      "step": 70190
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 6.536108940729992e-06,
+      "loss": 0.0356,
+      "step": 70200
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 6.532760352717937e-06,
+      "loss": 0.0503,
+      "step": 70210
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 6.529411764705883e-06,
+      "loss": 0.0367,
+      "step": 70220
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 6.526063176693827e-06,
+      "loss": 0.0317,
+      "step": 70230
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 6.522714588681773e-06,
+      "loss": 0.0547,
+      "step": 70240
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 6.519366000669718e-06,
+      "loss": 0.0371,
+      "step": 70250
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 6.516017412657663e-06,
+      "loss": 0.1339,
+      "step": 70260
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 6.512668824645608e-06,
+      "loss": 0.0297,
+      "step": 70270
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 6.5093202366335525e-06,
+      "loss": 0.0164,
+      "step": 70280
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 6.505971648621498e-06,
+      "loss": 0.0638,
+      "step": 70290
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 6.502623060609443e-06,
+      "loss": 0.0228,
+      "step": 70300
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 6.499274472597388e-06,
+      "loss": 0.0215,
+      "step": 70310
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 6.495925884585334e-06,
+      "loss": 0.066,
+      "step": 70320
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 6.492577296573278e-06,
+      "loss": 0.0389,
+      "step": 70330
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 6.489228708561224e-06,
+      "loss": 0.0356,
+      "step": 70340
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 6.485880120549169e-06,
+      "loss": 0.056,
+      "step": 70350
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 6.4825315325371136e-06,
+      "loss": 0.034,
+      "step": 70360
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.479182944525059e-06,
+      "loss": 0.2136,
+      "step": 70370
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.475834356513003e-06,
+      "loss": 0.0899,
+      "step": 70380
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.472485768500949e-06,
+      "loss": 0.0458,
+      "step": 70390
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.469137180488894e-06,
+      "loss": 0.0736,
+      "step": 70400
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.465788592476839e-06,
+      "loss": 0.0332,
+      "step": 70410
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.462440004464784e-06,
+      "loss": 0.0647,
+      "step": 70420
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.459091416452729e-06,
+      "loss": 0.0655,
+      "step": 70430
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.4560776872418795e-06,
+      "loss": 0.1291,
+      "step": 70440
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.452729099229825e-06,
+      "loss": 0.1012,
+      "step": 70450
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.449380511217769e-06,
+      "loss": 0.0511,
+      "step": 70460
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.446031923205715e-06,
+      "loss": 0.0127,
+      "step": 70470
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.44268333519366e-06,
+      "loss": 0.0673,
+      "step": 70480
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.439334747181605e-06,
+      "loss": 0.0652,
+      "step": 70490
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.435986159169551e-06,
+      "loss": 0.0709,
+      "step": 70500
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.432637571157496e-06,
+      "loss": 0.0512,
+      "step": 70510
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.429288983145441e-06,
+      "loss": 0.0552,
+      "step": 70520
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.4259403951333855e-06,
+      "loss": 0.0681,
+      "step": 70530
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.4225918071213304e-06,
+      "loss": 0.0534,
+      "step": 70540
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.419243219109275e-06,
+      "loss": 0.0384,
+      "step": 70550
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.415894631097221e-06,
+      "loss": 0.1733,
+      "step": 70560
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.412546043085166e-06,
+      "loss": 0.0738,
+      "step": 70570
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.409197455073111e-06,
+      "loss": 0.0715,
+      "step": 70580
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.405848867061056e-06,
+      "loss": 0.0848,
+      "step": 70590
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.402500279049001e-06,
+      "loss": 0.0141,
+      "step": 70600
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.399151691036947e-06,
+      "loss": 0.1211,
+      "step": 70610
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.3958031030248915e-06,
+      "loss": 0.0263,
+      "step": 70620
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.3924545150128364e-06,
+      "loss": 0.0264,
+      "step": 70630
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.389105927000782e-06,
+      "loss": 0.0566,
+      "step": 70640
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.385757338988726e-06,
+      "loss": 0.0983,
+      "step": 70650
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.382408750976672e-06,
+      "loss": 0.0846,
+      "step": 70660
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.379060162964616e-06,
+      "loss": 0.0553,
+      "step": 70670
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.375711574952562e-06,
+      "loss": 0.1022,
+      "step": 70680
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.372362986940508e-06,
+      "loss": 0.0534,
+      "step": 70690
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.369014398928452e-06,
+      "loss": 0.0792,
+      "step": 70700
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.3656658109163975e-06,
+      "loss": 0.0631,
+      "step": 70710
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.3623172229043416e-06,
+      "loss": 0.0274,
+      "step": 70720
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.358968634892287e-06,
+      "loss": 0.0876,
+      "step": 70730
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.355620046880232e-06,
+      "loss": 0.0574,
+      "step": 70740
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.352271458868177e-06,
+      "loss": 0.0246,
+      "step": 70750
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.348922870856123e-06,
+      "loss": 0.0438,
+      "step": 70760
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.345574282844067e-06,
+      "loss": 0.0758,
+      "step": 70770
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.342225694832013e-06,
+      "loss": 0.0696,
+      "step": 70780
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.338877106819958e-06,
+      "loss": 0.0702,
+      "step": 70790
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.335528518807903e-06,
+      "loss": 0.0559,
+      "step": 70800
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.3321799307958475e-06,
+      "loss": 0.0483,
+      "step": 70810
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.328831342783793e-06,
+      "loss": 0.032,
+      "step": 70820
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.325482754771738e-06,
+      "loss": 0.08,
+      "step": 70830
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.322134166759683e-06,
+      "loss": 0.0568,
+      "step": 70840
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.318785578747628e-06,
+      "loss": 0.0876,
+      "step": 70850
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.315436990735573e-06,
+      "loss": 0.0712,
+      "step": 70860
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.312088402723519e-06,
+      "loss": 0.041,
+      "step": 70870
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.308739814711464e-06,
+      "loss": 0.0453,
+      "step": 70880
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.305391226699409e-06,
+      "loss": 0.0524,
+      "step": 70890
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.3020426386873535e-06,
+      "loss": 0.0448,
+      "step": 70900
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.2986940506752985e-06,
+      "loss": 0.0564,
+      "step": 70910
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.295345462663244e-06,
+      "loss": 0.0448,
+      "step": 70920
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.291996874651188e-06,
+      "loss": 0.0446,
+      "step": 70930
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.288648286639134e-06,
+      "loss": 0.0399,
+      "step": 70940
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.28529969862708e-06,
+      "loss": 0.0243,
+      "step": 70950
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.281951110615024e-06,
+      "loss": 0.0469,
+      "step": 70960
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.27860252260297e-06,
+      "loss": 0.0324,
+      "step": 70970
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.275253934590914e-06,
+      "loss": 0.0355,
+      "step": 70980
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.2719053465788595e-06,
+      "loss": 0.0649,
+      "step": 70990
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.2685567585668044e-06,
+      "loss": 0.025,
+      "step": 71000
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.265208170554749e-06,
+      "loss": 0.058,
+      "step": 71010
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.261859582542695e-06,
+      "loss": 0.0632,
+      "step": 71020
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.258510994530639e-06,
+      "loss": 0.0298,
+      "step": 71030
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.255162406518585e-06,
+      "loss": 0.0527,
+      "step": 71040
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.25181381850653e-06,
+      "loss": 0.0377,
+      "step": 71050
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.248465230494475e-06,
+      "loss": 0.0432,
+      "step": 71060
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.2451166424824206e-06,
+      "loss": 0.0793,
+      "step": 71070
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.2417680544703655e-06,
+      "loss": 0.0353,
+      "step": 71080
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.23841946645831e-06,
+      "loss": 0.0298,
+      "step": 71090
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.235070878446255e-06,
+      "loss": 0.0331,
+      "step": 71100
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.2317222904342e-06,
+      "loss": 0.0795,
+      "step": 71110
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.228373702422145e-06,
+      "loss": 0.0919,
+      "step": 71120
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.225025114410091e-06,
+      "loss": 0.0727,
+      "step": 71130
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.221676526398036e-06,
+      "loss": 0.0755,
+      "step": 71140
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.218327938385981e-06,
+      "loss": 0.039,
+      "step": 71150
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.214979350373926e-06,
+      "loss": 0.0383,
+      "step": 71160
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.211630762361871e-06,
+      "loss": 0.0118,
+      "step": 71170
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.208282174349816e-06,
+      "loss": 0.0496,
+      "step": 71180
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.2049335863377605e-06,
+      "loss": 0.0716,
+      "step": 71190
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.201584998325706e-06,
+      "loss": 0.1596,
+      "step": 71200
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.198236410313651e-06,
+      "loss": 0.0111,
+      "step": 71210
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.194887822301596e-06,
+      "loss": 0.0587,
+      "step": 71220
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.191539234289542e-06,
+      "loss": 0.0863,
+      "step": 71230
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.188190646277486e-06,
+      "loss": 0.0513,
+      "step": 71240
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.184842058265432e-06,
+      "loss": 0.0296,
+      "step": 71250
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.181493470253377e-06,
+      "loss": 0.0431,
+      "step": 71260
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.1781448822413215e-06,
+      "loss": 0.0389,
+      "step": 71270
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.174796294229267e-06,
+      "loss": 0.1132,
+      "step": 71280
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.171447706217211e-06,
+      "loss": 0.0807,
+      "step": 71290
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.168099118205157e-06,
+      "loss": 0.0598,
+      "step": 71300
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.164750530193102e-06,
+      "loss": 0.0533,
+      "step": 71310
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.161401942181047e-06,
+      "loss": 0.0516,
+      "step": 71320
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.158053354168993e-06,
+      "loss": 0.063,
+      "step": 71330
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.154704766156937e-06,
+      "loss": 0.1042,
+      "step": 71340
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.151356178144883e-06,
+      "loss": 0.0363,
+      "step": 71350
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.1480075901328275e-06,
+      "loss": 0.0839,
+      "step": 71360
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.1446590021207724e-06,
+      "loss": 0.0357,
+      "step": 71370
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.141310414108717e-06,
+      "loss": 0.0405,
+      "step": 71380
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.137961826096663e-06,
+      "loss": 0.0412,
+      "step": 71390
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.134613238084608e-06,
+      "loss": 0.0548,
+      "step": 71400
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.131264650072553e-06,
+      "loss": 0.0366,
+      "step": 71410
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.127916062060498e-06,
+      "loss": 0.0508,
+      "step": 71420
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.124567474048443e-06,
+      "loss": 0.0684,
+      "step": 71430
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.121218886036389e-06,
+      "loss": 0.0604,
+      "step": 71440
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.117870298024333e-06,
+      "loss": 0.1518,
+      "step": 71450
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.1145217100122784e-06,
+      "loss": 0.025,
+      "step": 71460
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.111173122000223e-06,
+      "loss": 0.0384,
+      "step": 71470
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.107824533988168e-06,
+      "loss": 0.0269,
+      "step": 71480
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.104475945976114e-06,
+      "loss": 0.0347,
+      "step": 71490
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.101127357964058e-06,
+      "loss": 0.0393,
+      "step": 71500
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.097778769952004e-06,
+      "loss": 0.0512,
+      "step": 71510
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.094430181939949e-06,
+      "loss": 0.0307,
+      "step": 71520
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.091081593927894e-06,
+      "loss": 0.0411,
+      "step": 71530
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.0877330059158395e-06,
+      "loss": 0.04,
+      "step": 71540
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.0843844179037836e-06,
+      "loss": 0.0345,
+      "step": 71550
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.081035829891729e-06,
+      "loss": 0.1396,
+      "step": 71560
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.077687241879674e-06,
+      "loss": 0.0365,
+      "step": 71570
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.074338653867619e-06,
+      "loss": 0.0403,
+      "step": 71580
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.070990065855565e-06,
+      "loss": 0.1169,
+      "step": 71590
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.067641477843509e-06,
+      "loss": 0.0569,
+      "step": 71600
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.064292889831455e-06,
+      "loss": 0.0311,
+      "step": 71610
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.0609443018194e-06,
+      "loss": 0.0177,
+      "step": 71620
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.057595713807345e-06,
+      "loss": 0.0682,
+      "step": 71630
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.0542471257952896e-06,
+      "loss": 0.0638,
+      "step": 71640
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.0508985377832345e-06,
+      "loss": 0.0548,
+      "step": 71650
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.04754994977118e-06,
+      "loss": 0.0873,
+      "step": 71660
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.044201361759125e-06,
+      "loss": 0.0375,
+      "step": 71670
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.04085277374707e-06,
+      "loss": 0.0821,
+      "step": 71680
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.037504185735015e-06,
+      "loss": 0.0348,
+      "step": 71690
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.034155597722961e-06,
+      "loss": 0.0781,
+      "step": 71700
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.030807009710905e-06,
+      "loss": 0.0276,
+      "step": 71710
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.027458421698851e-06,
+      "loss": 0.0692,
+      "step": 71720
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.0241098336867955e-06,
+      "loss": 0.0113,
+      "step": 71730
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.0207612456747405e-06,
+      "loss": 0.0412,
+      "step": 71740
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.017412657662686e-06,
+      "loss": 0.0426,
+      "step": 71750
+    },
+    {
+      "epoch": 8.0,
+      "eval_loss": 0.3032523989677429,
+      "eval_runtime": 254.2896,
+      "eval_samples_per_second": 7.841,
+      "eval_steps_per_second": 3.921,
+      "eval_wer": 0.0874624054439955,
+      "step": 71752
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.01406406965063e-06,
+      "loss": 0.0251,
+      "step": 71760
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.010715481638576e-06,
+      "loss": 0.0665,
+      "step": 71770
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.007366893626521e-06,
+      "loss": 0.0304,
+      "step": 71780
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.004018305614466e-06,
+      "loss": 0.0468,
+      "step": 71790
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.000669717602412e-06,
+      "loss": 0.026,
+      "step": 71800
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 5.997321129590356e-06,
+      "loss": 0.0553,
+      "step": 71810
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 5.9939725415783015e-06,
+      "loss": 0.0267,
+      "step": 71820
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 5.9906239535662464e-06,
+      "loss": 0.5005,
+      "step": 71830
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 5.987275365554191e-06,
+      "loss": 0.0211,
+      "step": 71840
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 5.983926777542137e-06,
+      "loss": 0.0259,
+      "step": 71850
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 5.980578189530081e-06,
+      "loss": 0.0387,
+      "step": 71860
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 5.977229601518027e-06,
+      "loss": 0.1373,
+      "step": 71870
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 5.973881013505972e-06,
+      "loss": 0.0307,
+      "step": 71880
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 5.970532425493917e-06,
+      "loss": 0.0509,
+      "step": 71890
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 5.967183837481862e-06,
+      "loss": 0.0468,
+      "step": 71900
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 5.963835249469807e-06,
+      "loss": 0.0403,
+      "step": 71910
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 5.960486661457752e-06,
+      "loss": 0.0234,
+      "step": 71920
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 5.957138073445697e-06,
+      "loss": 0.022,
+      "step": 71930
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 5.953789485433642e-06,
+      "loss": 0.0401,
+      "step": 71940
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 5.950440897421587e-06,
+      "loss": 0.0407,
+      "step": 71950
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 5.947092309409532e-06,
+      "loss": 0.0645,
+      "step": 71960
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 5.943743721397478e-06,
+      "loss": 0.0342,
+      "step": 71970
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.940395133385423e-06,
+      "loss": 0.0248,
+      "step": 71980
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.937046545373368e-06,
+      "loss": 0.0535,
+      "step": 71990
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.933697957361313e-06,
+      "loss": 0.0495,
+      "step": 72000
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.930349369349258e-06,
+      "loss": 0.0298,
+      "step": 72010
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.9270007813372025e-06,
+      "loss": 0.0072,
+      "step": 72020
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.923652193325148e-06,
+      "loss": 0.0214,
+      "step": 72030
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.920303605313093e-06,
+      "loss": 0.0338,
+      "step": 72040
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.916955017301038e-06,
+      "loss": 0.0483,
+      "step": 72050
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.913606429288984e-06,
+      "loss": 0.0755,
+      "step": 72060
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.910257841276928e-06,
+      "loss": 0.0405,
+      "step": 72070
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.906909253264874e-06,
+      "loss": 0.0522,
+      "step": 72080
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.903560665252818e-06,
+      "loss": 0.0081,
+      "step": 72090
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.9002120772407635e-06,
+      "loss": 0.0138,
+      "step": 72100
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.896863489228709e-06,
+      "loss": 0.0674,
+      "step": 72110
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.893514901216653e-06,
+      "loss": 0.0387,
+      "step": 72120
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.890166313204599e-06,
+      "loss": 0.0449,
+      "step": 72130
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.886817725192544e-06,
+      "loss": 0.0379,
+      "step": 72140
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.883469137180489e-06,
+      "loss": 0.1031,
+      "step": 72150
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.880120549168434e-06,
+      "loss": 0.0532,
+      "step": 72160
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.876771961156379e-06,
+      "loss": 0.0437,
+      "step": 72170
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.873423373144325e-06,
+      "loss": 0.0131,
+      "step": 72180
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.8700747851322695e-06,
+      "loss": 0.06,
+      "step": 72190
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.8667261971202144e-06,
+      "loss": 0.0209,
+      "step": 72200
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.863377609108159e-06,
+      "loss": 0.0716,
+      "step": 72210
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.860029021096104e-06,
+      "loss": 0.0382,
+      "step": 72220
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.85668043308405e-06,
+      "loss": 0.0524,
+      "step": 72230
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.853331845071995e-06,
+      "loss": 0.0232,
+      "step": 72240
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.84998325705994e-06,
+      "loss": 0.0263,
+      "step": 72250
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.846634669047885e-06,
+      "loss": 0.0169,
+      "step": 72260
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.843286081035831e-06,
+      "loss": 0.0733,
+      "step": 72270
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.839937493023775e-06,
+      "loss": 0.0285,
+      "step": 72280
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.8365889050117204e-06,
+      "loss": 0.065,
+      "step": 72290
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.833240316999665e-06,
+      "loss": 0.0956,
+      "step": 72300
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.82989172898761e-06,
+      "loss": 0.0789,
+      "step": 72310
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.826543140975556e-06,
+      "loss": 0.0492,
+      "step": 72320
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.8231945529635e-06,
+      "loss": 0.0328,
+      "step": 72330
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.819845964951446e-06,
+      "loss": 0.0425,
+      "step": 72340
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.81649737693939e-06,
+      "loss": 0.0402,
+      "step": 72350
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.813148788927336e-06,
+      "loss": 0.0743,
+      "step": 72360
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.8098002009152815e-06,
+      "loss": 0.0368,
+      "step": 72370
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.8064516129032256e-06,
+      "loss": 0.0528,
+      "step": 72380
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.803103024891171e-06,
+      "loss": 0.1011,
+      "step": 72390
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.799754436879115e-06,
+      "loss": 0.0937,
+      "step": 72400
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.796405848867061e-06,
+      "loss": 0.0671,
+      "step": 72410
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.793057260855007e-06,
+      "loss": 0.0078,
+      "step": 72420
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.789708672842951e-06,
+      "loss": 0.0693,
+      "step": 72430
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.786360084830897e-06,
+      "loss": 0.0317,
+      "step": 72440
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.783011496818842e-06,
+      "loss": 0.1261,
+      "step": 72450
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.779662908806787e-06,
+      "loss": 0.0671,
+      "step": 72460
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.7763143207947316e-06,
+      "loss": 0.0353,
+      "step": 72470
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.7729657327826765e-06,
+      "loss": 0.0294,
+      "step": 72480
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.769617144770622e-06,
+      "loss": 0.0438,
+      "step": 72490
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.766268556758567e-06,
+      "loss": 0.0336,
+      "step": 72500
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.762919968746512e-06,
+      "loss": 0.0549,
+      "step": 72510
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.759571380734457e-06,
+      "loss": 0.0497,
+      "step": 72520
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.756222792722402e-06,
+      "loss": 0.0509,
+      "step": 72530
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.752874204710347e-06,
+      "loss": 0.0401,
+      "step": 72540
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.749525616698293e-06,
+      "loss": 0.0886,
+      "step": 72550
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.7461770286862375e-06,
+      "loss": 0.1142,
+      "step": 72560
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.7428284406741825e-06,
+      "loss": 0.0701,
+      "step": 72570
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.739479852662128e-06,
+      "loss": 0.0608,
+      "step": 72580
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.736131264650072e-06,
+      "loss": 0.0433,
+      "step": 72590
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.732782676638018e-06,
+      "loss": 0.0235,
+      "step": 72600
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.729434088625962e-06,
+      "loss": 0.1272,
+      "step": 72610
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.726085500613908e-06,
+      "loss": 0.0754,
+      "step": 72620
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.722736912601854e-06,
+      "loss": 0.1014,
+      "step": 72630
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.719388324589798e-06,
+      "loss": 0.0742,
+      "step": 72640
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.7160397365777435e-06,
+      "loss": 0.0286,
+      "step": 72650
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.712691148565688e-06,
+      "loss": 0.0776,
+      "step": 72660
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.709342560553633e-06,
+      "loss": 0.0196,
+      "step": 72670
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.705993972541579e-06,
+      "loss": 0.1009,
+      "step": 72680
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.702645384529523e-06,
+      "loss": 0.0261,
+      "step": 72690
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.699296796517469e-06,
+      "loss": 0.0231,
+      "step": 72700
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.695948208505414e-06,
+      "loss": 0.0436,
+      "step": 72710
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.692599620493359e-06,
+      "loss": 0.1036,
+      "step": 72720
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.6895858912825095e-06,
+      "loss": 0.0499,
+      "step": 72730
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.686237303270454e-06,
+      "loss": 0.0934,
+      "step": 72740
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.682888715258399e-06,
+      "loss": 0.0461,
+      "step": 72750
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.679540127246345e-06,
+      "loss": 0.0315,
+      "step": 72760
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.676191539234289e-06,
+      "loss": 0.0302,
+      "step": 72770
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.672842951222235e-06,
+      "loss": 0.0938,
+      "step": 72780
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.66949436321018e-06,
+      "loss": 0.0724,
+      "step": 72790
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.666145775198125e-06,
+      "loss": 0.0427,
+      "step": 72800
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.6627971871860706e-06,
+      "loss": 0.0439,
+      "step": 72810
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.659448599174015e-06,
+      "loss": 0.0308,
+      "step": 72820
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.65610001116196e-06,
+      "loss": 0.0307,
+      "step": 72830
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.652751423149905e-06,
+      "loss": 0.0452,
+      "step": 72840
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.64940283513785e-06,
+      "loss": 0.0449,
+      "step": 72850
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.646054247125795e-06,
+      "loss": 0.013,
+      "step": 72860
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.64270565911374e-06,
+      "loss": 0.0621,
+      "step": 72870
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.639357071101686e-06,
+      "loss": 0.0457,
+      "step": 72880
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.636008483089631e-06,
+      "loss": 0.042,
+      "step": 72890
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.632659895077576e-06,
+      "loss": 0.0869,
+      "step": 72900
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.629311307065521e-06,
+      "loss": 0.027,
+      "step": 72910
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.625962719053466e-06,
+      "loss": 0.0782,
+      "step": 72920
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.622614131041411e-06,
+      "loss": 0.0479,
+      "step": 72930
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.619265543029356e-06,
+      "loss": 0.0555,
+      "step": 72940
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.615916955017301e-06,
+      "loss": 0.0283,
+      "step": 72950
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.612568367005246e-06,
+      "loss": 0.1015,
+      "step": 72960
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.609219778993192e-06,
+      "loss": 0.0774,
+      "step": 72970
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.605871190981136e-06,
+      "loss": 0.0185,
+      "step": 72980
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.602522602969082e-06,
+      "loss": 0.0822,
+      "step": 72990
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.599174014957027e-06,
+      "loss": 0.0921,
+      "step": 73000
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.5958254269449715e-06,
+      "loss": 0.0307,
+      "step": 73010
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.592476838932917e-06,
+      "loss": 0.0268,
+      "step": 73020
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.589128250920861e-06,
+      "loss": 0.019,
+      "step": 73030
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.585779662908807e-06,
+      "loss": 0.0184,
+      "step": 73040
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.582431074896752e-06,
+      "loss": 0.1164,
+      "step": 73050
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.579082486884697e-06,
+      "loss": 0.0752,
+      "step": 73060
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.575733898872643e-06,
+      "loss": 0.1286,
+      "step": 73070
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.572385310860587e-06,
+      "loss": 0.0483,
+      "step": 73080
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.569036722848533e-06,
+      "loss": 0.5778,
+      "step": 73090
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.5656881348364775e-06,
+      "loss": 0.0517,
+      "step": 73100
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.5623395468244224e-06,
+      "loss": 0.0435,
+      "step": 73110
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.558990958812367e-06,
+      "loss": 0.07,
+      "step": 73120
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.555642370800312e-06,
+      "loss": 0.0169,
+      "step": 73130
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.552293782788258e-06,
+      "loss": 0.0708,
+      "step": 73140
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.548945194776203e-06,
+      "loss": 0.0544,
+      "step": 73150
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.545596606764148e-06,
+      "loss": 0.059,
+      "step": 73160
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.542248018752093e-06,
+      "loss": 0.0266,
+      "step": 73170
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.538899430740038e-06,
+      "loss": 0.0358,
+      "step": 73180
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.5355508427279835e-06,
+      "loss": 0.0327,
+      "step": 73190
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.532202254715928e-06,
+      "loss": 0.0611,
+      "step": 73200
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.528853666703873e-06,
+      "loss": 0.0389,
+      "step": 73210
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.525505078691818e-06,
+      "loss": 0.0358,
+      "step": 73220
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.522156490679764e-06,
+      "loss": 0.0212,
+      "step": 73230
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.518807902667708e-06,
+      "loss": 0.0262,
+      "step": 73240
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.515459314655654e-06,
+      "loss": 0.0811,
+      "step": 73250
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.512110726643599e-06,
+      "loss": 0.0434,
+      "step": 73260
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.508762138631544e-06,
+      "loss": 0.0543,
+      "step": 73270
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.5054135506194895e-06,
+      "loss": 0.089,
+      "step": 73280
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.5020649626074336e-06,
+      "loss": 0.0487,
+      "step": 73290
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.498716374595379e-06,
+      "loss": 0.1104,
+      "step": 73300
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.495367786583323e-06,
+      "loss": 0.0476,
+      "step": 73310
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.492019198571269e-06,
+      "loss": 0.0146,
+      "step": 73320
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.488670610559215e-06,
+      "loss": 0.0109,
+      "step": 73330
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.485322022547159e-06,
+      "loss": 0.0828,
+      "step": 73340
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.481973434535105e-06,
+      "loss": 0.2911,
+      "step": 73350
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.47862484652305e-06,
+      "loss": 0.0458,
+      "step": 73360
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.475276258510995e-06,
+      "loss": 0.0746,
+      "step": 73370
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.47192767049894e-06,
+      "loss": 0.0648,
+      "step": 73380
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.4685790824868845e-06,
+      "loss": 0.0933,
+      "step": 73390
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.46523049447483e-06,
+      "loss": 0.0427,
+      "step": 73400
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.461881906462775e-06,
+      "loss": 0.0287,
+      "step": 73410
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.45853331845072e-06,
+      "loss": 0.0386,
+      "step": 73420
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.455184730438665e-06,
+      "loss": 0.0185,
+      "step": 73430
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.45183614242661e-06,
+      "loss": 0.0666,
+      "step": 73440
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.448487554414556e-06,
+      "loss": 0.0134,
+      "step": 73450
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.445138966402501e-06,
+      "loss": 0.0977,
+      "step": 73460
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.4417903783904455e-06,
+      "loss": 0.0447,
+      "step": 73470
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.4384417903783904e-06,
+      "loss": 0.0323,
+      "step": 73480
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.435093202366335e-06,
+      "loss": 0.0785,
+      "step": 73490
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.43174461435428e-06,
+      "loss": 0.0252,
+      "step": 73500
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.428396026342226e-06,
+      "loss": 0.0615,
+      "step": 73510
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.425047438330171e-06,
+      "loss": 0.0701,
+      "step": 73520
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.421698850318116e-06,
+      "loss": 0.0142,
+      "step": 73530
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.418350262306062e-06,
+      "loss": 0.0818,
+      "step": 73540
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.415001674294006e-06,
+      "loss": 0.0961,
+      "step": 73550
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.4116530862819515e-06,
+      "loss": 0.0645,
+      "step": 73560
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.408304498269896e-06,
+      "loss": 0.0095,
+      "step": 73570
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.404955910257841e-06,
+      "loss": 0.0683,
+      "step": 73580
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.401607322245787e-06,
+      "loss": 0.0455,
+      "step": 73590
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.398258734233731e-06,
+      "loss": 0.087,
+      "step": 73600
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.394910146221677e-06,
+      "loss": 0.0952,
+      "step": 73610
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.391561558209621e-06,
+      "loss": 0.0498,
+      "step": 73620
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.388212970197567e-06,
+      "loss": 0.0492,
+      "step": 73630
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.3848643821855126e-06,
+      "loss": 0.0324,
+      "step": 73640
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.381515794173457e-06,
+      "loss": 0.0327,
+      "step": 73650
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.378167206161402e-06,
+      "loss": 0.0505,
+      "step": 73660
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.374818618149347e-06,
+      "loss": 0.0537,
+      "step": 73670
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.371470030137292e-06,
+      "loss": 0.0268,
+      "step": 73680
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.368121442125237e-06,
+      "loss": 0.0238,
+      "step": 73690
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.364772854113182e-06,
+      "loss": 0.0498,
+      "step": 73700
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.361424266101128e-06,
+      "loss": 0.0159,
+      "step": 73710
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.358075678089073e-06,
+      "loss": 0.0414,
+      "step": 73720
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.354727090077018e-06,
+      "loss": 0.0287,
+      "step": 73730
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.351378502064963e-06,
+      "loss": 0.0387,
+      "step": 73740
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.3480299140529075e-06,
+      "loss": 0.1461,
+      "step": 73750
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.3446813260408525e-06,
+      "loss": 0.0525,
+      "step": 73760
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.341332738028798e-06,
+      "loss": 0.0469,
+      "step": 73770
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.337984150016743e-06,
+      "loss": 0.0339,
+      "step": 73780
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.334635562004688e-06,
+      "loss": 0.022,
+      "step": 73790
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.331286973992634e-06,
+      "loss": 0.0219,
+      "step": 73800
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.327938385980578e-06,
+      "loss": 0.0535,
+      "step": 73810
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.324589797968524e-06,
+      "loss": 0.0891,
+      "step": 73820
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.321241209956469e-06,
+      "loss": 0.034,
+      "step": 73830
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.3178926219444135e-06,
+      "loss": 0.0372,
+      "step": 73840
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.314544033932359e-06,
+      "loss": 0.053,
+      "step": 73850
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.311195445920303e-06,
+      "loss": 0.0292,
+      "step": 73860
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.307846857908249e-06,
+      "loss": 0.0216,
+      "step": 73870
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.304498269896193e-06,
+      "loss": 0.2596,
+      "step": 73880
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.301149681884139e-06,
+      "loss": 0.0967,
+      "step": 73890
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.297801093872085e-06,
+      "loss": 0.1005,
+      "step": 73900
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.294452505860029e-06,
+      "loss": 0.0312,
+      "step": 73910
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.291103917847975e-06,
+      "loss": 0.0253,
+      "step": 73920
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.287755329835919e-06,
+      "loss": 0.0295,
+      "step": 73930
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.2844067418238644e-06,
+      "loss": 0.0389,
+      "step": 73940
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.281058153811809e-06,
+      "loss": 0.0197,
+      "step": 73950
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.277709565799754e-06,
+      "loss": 0.0714,
+      "step": 73960
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.2743609777877e-06,
+      "loss": 0.1081,
+      "step": 73970
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.271012389775645e-06,
+      "loss": 0.0556,
+      "step": 73980
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.26766380176359e-06,
+      "loss": 0.0599,
+      "step": 73990
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.264315213751535e-06,
+      "loss": 0.0642,
+      "step": 74000
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.26096662573948e-06,
+      "loss": 0.0318,
+      "step": 74010
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.257618037727425e-06,
+      "loss": 0.03,
+      "step": 74020
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.25426944971537e-06,
+      "loss": 0.0265,
+      "step": 74030
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.250920861703315e-06,
+      "loss": 0.0974,
+      "step": 74040
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.24757227369126e-06,
+      "loss": 0.0836,
+      "step": 74050
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.244223685679205e-06,
+      "loss": 0.0562,
+      "step": 74060
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.24087509766715e-06,
+      "loss": 0.0216,
+      "step": 74070
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.237526509655096e-06,
+      "loss": 0.0683,
+      "step": 74080
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.234177921643041e-06,
+      "loss": 0.045,
+      "step": 74090
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.230829333630986e-06,
+      "loss": 0.0436,
+      "step": 74100
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.2274807456189315e-06,
+      "loss": 0.0261,
+      "step": 74110
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.2241321576068756e-06,
+      "loss": 0.0343,
+      "step": 74120
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.220783569594821e-06,
+      "loss": 0.0713,
+      "step": 74130
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.217434981582765e-06,
+      "loss": 0.07,
+      "step": 74140
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.214086393570711e-06,
+      "loss": 0.1259,
+      "step": 74150
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.210737805558657e-06,
+      "loss": 0.0705,
+      "step": 74160
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.207389217546601e-06,
+      "loss": 0.0406,
+      "step": 74170
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.204040629534547e-06,
+      "loss": 0.0325,
+      "step": 74180
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.200692041522491e-06,
+      "loss": 0.0258,
+      "step": 74190
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.197343453510437e-06,
+      "loss": 0.0357,
+      "step": 74200
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.1939948654983815e-06,
+      "loss": 0.0768,
+      "step": 74210
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.1906462774863265e-06,
+      "loss": 0.0177,
+      "step": 74220
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.187297689474272e-06,
+      "loss": 0.0166,
+      "step": 74230
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.183949101462217e-06,
+      "loss": 0.0547,
+      "step": 74240
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.180600513450162e-06,
+      "loss": 0.0932,
+      "step": 74250
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.177251925438107e-06,
+      "loss": 0.1056,
+      "step": 74260
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.173903337426052e-06,
+      "loss": 0.02,
+      "step": 74270
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.170554749413998e-06,
+      "loss": 0.1405,
+      "step": 74280
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.167206161401943e-06,
+      "loss": 0.0535,
+      "step": 74290
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.1638575733898875e-06,
+      "loss": 0.0455,
+      "step": 74300
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.1605089853778324e-06,
+      "loss": 0.1758,
+      "step": 74310
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.157160397365777e-06,
+      "loss": 0.0463,
+      "step": 74320
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.153811809353722e-06,
+      "loss": 0.0347,
+      "step": 74330
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.150463221341668e-06,
+      "loss": 0.133,
+      "step": 74340
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.147114633329613e-06,
+      "loss": 0.0405,
+      "step": 74350
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.143766045317558e-06,
+      "loss": 0.0616,
+      "step": 74360
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.140417457305503e-06,
+      "loss": 0.0101,
+      "step": 74370
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.137068869293448e-06,
+      "loss": 0.0922,
+      "step": 74380
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.1337202812813935e-06,
+      "loss": 0.0183,
+      "step": 74390
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.130371693269338e-06,
+      "loss": 0.1003,
+      "step": 74400
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.127023105257283e-06,
+      "loss": 0.0499,
+      "step": 74410
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.123674517245229e-06,
+      "loss": 0.0215,
+      "step": 74420
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.120325929233173e-06,
+      "loss": 0.0328,
+      "step": 74430
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.116977341221119e-06,
+      "loss": 0.0163,
+      "step": 74440
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.113628753209063e-06,
+      "loss": 0.092,
+      "step": 74450
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.110280165197009e-06,
+      "loss": 0.0592,
+      "step": 74460
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.106931577184954e-06,
+      "loss": 0.0715,
+      "step": 74470
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.103582989172899e-06,
+      "loss": 0.0213,
+      "step": 74480
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.100234401160844e-06,
+      "loss": 0.0272,
+      "step": 74490
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.0968858131487885e-06,
+      "loss": 0.0645,
+      "step": 74500
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.093537225136734e-06,
+      "loss": 0.0844,
+      "step": 74510
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.090188637124679e-06,
+      "loss": 0.0386,
+      "step": 74520
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.086840049112624e-06,
+      "loss": 0.0528,
+      "step": 74530
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.08349146110057e-06,
+      "loss": 0.0604,
+      "step": 74540
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.080142873088515e-06,
+      "loss": 0.0208,
+      "step": 74550
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.07679428507646e-06,
+      "loss": 0.0309,
+      "step": 74560
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.073445697064405e-06,
+      "loss": 0.0755,
+      "step": 74570
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.0700971090523496e-06,
+      "loss": 0.0572,
+      "step": 74580
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.0667485210402945e-06,
+      "loss": 0.0814,
+      "step": 74590
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.06339993302824e-06,
+      "loss": 0.0526,
+      "step": 74600
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.060051345016185e-06,
+      "loss": 0.0415,
+      "step": 74610
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.05670275700413e-06,
+      "loss": 0.0348,
+      "step": 74620
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.053354168992075e-06,
+      "loss": 0.0025,
+      "step": 74630
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.05000558098002e-06,
+      "loss": 0.0377,
+      "step": 74640
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.046656992967966e-06,
+      "loss": 0.0269,
+      "step": 74650
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.04330840495591e-06,
+      "loss": 0.0881,
+      "step": 74660
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.0399598169438555e-06,
+      "loss": 0.019,
+      "step": 74670
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.0366112289318005e-06,
+      "loss": 0.0936,
+      "step": 74680
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.033262640919745e-06,
+      "loss": 0.0481,
+      "step": 74690
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.029914052907691e-06,
+      "loss": 0.0366,
+      "step": 74700
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.026565464895635e-06,
+      "loss": 0.0867,
+      "step": 74710
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.023216876883581e-06,
+      "loss": 0.0306,
+      "step": 74720
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.019868288871527e-06,
+      "loss": 0.0532,
+      "step": 74730
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.016519700859471e-06,
+      "loss": 0.0474,
+      "step": 74740
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.013171112847417e-06,
+      "loss": 0.0367,
+      "step": 74750
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.009822524835361e-06,
+      "loss": 0.0209,
+      "step": 74760
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.0064739368233064e-06,
+      "loss": 0.0151,
+      "step": 74770
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.003125348811251e-06,
+      "loss": 0.1056,
+      "step": 74780
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 4.999776760799196e-06,
+      "loss": 0.027,
+      "step": 74790
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 4.996428172787142e-06,
+      "loss": 0.0388,
+      "step": 74800
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 4.993079584775086e-06,
+      "loss": 0.0638,
+      "step": 74810
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 4.989730996763032e-06,
+      "loss": 0.0548,
+      "step": 74820
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 4.986382408750977e-06,
+      "loss": 0.0529,
+      "step": 74830
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 4.983033820738922e-06,
+      "loss": 0.0428,
+      "step": 74840
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 4.979685232726867e-06,
+      "loss": 0.0547,
+      "step": 74850
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 4.9763366447148124e-06,
+      "loss": 0.0921,
+      "step": 74860
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 4.972988056702757e-06,
+      "loss": 0.0389,
+      "step": 74870
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 4.969639468690702e-06,
+      "loss": 0.0348,
+      "step": 74880
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 4.966290880678647e-06,
+      "loss": 0.0336,
+      "step": 74890
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 4.962942292666592e-06,
+      "loss": 0.0818,
+      "step": 74900
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 4.959593704654538e-06,
+      "loss": 0.0533,
+      "step": 74910
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 4.956245116642482e-06,
+      "loss": 0.0331,
+      "step": 74920
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 4.952896528630428e-06,
+      "loss": 0.0185,
+      "step": 74930
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 4.949547940618373e-06,
+      "loss": 0.0155,
+      "step": 74940
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 4.9461993526063176e-06,
+      "loss": 0.0926,
+      "step": 74950
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 4.942850764594263e-06,
+      "loss": 0.0322,
+      "step": 74960
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 4.939502176582207e-06,
+      "loss": 0.0654,
+      "step": 74970
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 4.936153588570153e-06,
+      "loss": 0.0611,
+      "step": 74980
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 4.932805000558099e-06,
+      "loss": 0.0615,
+      "step": 74990
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 4.929456412546043e-06,
+      "loss": 0.0294,
+      "step": 75000
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 4.926107824533989e-06,
+      "loss": 0.0527,
+      "step": 75010
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 4.922759236521933e-06,
+      "loss": 0.1063,
+      "step": 75020
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 4.919410648509879e-06,
+      "loss": 0.0171,
+      "step": 75030
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 4.9160620604978235e-06,
+      "loss": 0.0648,
+      "step": 75040
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 4.9127134724857685e-06,
+      "loss": 0.0193,
+      "step": 75050
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 4.909364884473714e-06,
+      "loss": 0.0933,
+      "step": 75060
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 4.906016296461658e-06,
+      "loss": 0.0818,
+      "step": 75070
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 4.902667708449604e-06,
+      "loss": 0.0221,
+      "step": 75080
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 4.899319120437549e-06,
+      "loss": 0.0199,
+      "step": 75090
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 4.895970532425494e-06,
+      "loss": 0.0252,
+      "step": 75100
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 4.892621944413439e-06,
+      "loss": 0.0516,
+      "step": 75110
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 4.889273356401384e-06,
+      "loss": 0.0333,
+      "step": 75120
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 4.8859247683893295e-06,
+      "loss": 0.0413,
+      "step": 75130
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 4.8825761803772745e-06,
+      "loss": 0.0124,
+      "step": 75140
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 4.879227592365219e-06,
+      "loss": 0.0753,
+      "step": 75150
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 4.875879004353164e-06,
+      "loss": 0.0132,
+      "step": 75160
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 4.87253041634111e-06,
+      "loss": 0.0402,
+      "step": 75170
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 4.869181828329055e-06,
+      "loss": 0.0612,
+      "step": 75180
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 4.865833240317e-06,
+      "loss": 0.0415,
+      "step": 75190
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 4.862484652304945e-06,
+      "loss": 0.0798,
+      "step": 75200
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 4.85913606429289e-06,
+      "loss": 0.0567,
+      "step": 75210
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 4.8557874762808355e-06,
+      "loss": 0.0656,
+      "step": 75220
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 4.85243888826878e-06,
+      "loss": 0.024,
+      "step": 75230
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 4.849090300256725e-06,
+      "loss": 0.0154,
+      "step": 75240
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 4.84574171224467e-06,
+      "loss": 0.0448,
+      "step": 75250
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 4.842393124232615e-06,
+      "loss": 0.0642,
+      "step": 75260
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 4.839044536220561e-06,
+      "loss": 0.0984,
+      "step": 75270
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 4.835695948208505e-06,
+      "loss": 0.027,
+      "step": 75280
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 4.832347360196451e-06,
+      "loss": 0.0133,
+      "step": 75290
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 4.828998772184396e-06,
+      "loss": 0.0523,
+      "step": 75300
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 4.825650184172341e-06,
+      "loss": 0.0442,
+      "step": 75310
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 4.822301596160286e-06,
+      "loss": 0.063,
+      "step": 75320
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 4.8189530081482305e-06,
+      "loss": 0.0331,
+      "step": 75330
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 4.815604420136176e-06,
+      "loss": 0.066,
+      "step": 75340
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 4.812255832124121e-06,
+      "loss": 0.0723,
+      "step": 75350
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 4.808907244112066e-06,
+      "loss": 0.0415,
+      "step": 75360
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 4.805558656100011e-06,
+      "loss": 0.0376,
+      "step": 75370
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 4.802210068087956e-06,
+      "loss": 0.0442,
+      "step": 75380
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 4.798861480075902e-06,
+      "loss": 0.0319,
+      "step": 75390
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 4.795512892063847e-06,
+      "loss": 0.1101,
+      "step": 75400
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 4.7921643040517916e-06,
+      "loss": 0.0358,
+      "step": 75410
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 4.7888157160397365e-06,
+      "loss": 0.079,
+      "step": 75420
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 4.785467128027682e-06,
+      "loss": 0.0531,
+      "step": 75430
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 4.782118540015627e-06,
+      "loss": 0.0568,
+      "step": 75440
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 4.778769952003572e-06,
+      "loss": 0.0384,
+      "step": 75450
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 4.775421363991517e-06,
+      "loss": 0.1232,
+      "step": 75460
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 4.772072775979462e-06,
+      "loss": 0.0136,
+      "step": 75470
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 4.768724187967408e-06,
+      "loss": 0.0515,
+      "step": 75480
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 4.765375599955352e-06,
+      "loss": 0.0923,
+      "step": 75490
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 4.7620270119432975e-06,
+      "loss": 0.074,
+      "step": 75500
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 4.7586784239312425e-06,
+      "loss": 0.0658,
+      "step": 75510
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 4.755329835919187e-06,
+      "loss": 0.0274,
+      "step": 75520
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 4.751981247907133e-06,
+      "loss": 0.036,
+      "step": 75530
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 4.748632659895077e-06,
+      "loss": 0.0525,
+      "step": 75540
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 4.745284071883023e-06,
+      "loss": 0.0597,
+      "step": 75550
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 4.741935483870967e-06,
+      "loss": 0.071,
+      "step": 75560
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 4.738586895858913e-06,
+      "loss": 0.1261,
+      "step": 75570
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 4.735238307846859e-06,
+      "loss": 0.0543,
+      "step": 75580
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 4.731889719834803e-06,
+      "loss": 0.0642,
+      "step": 75590
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 4.7285411318227484e-06,
+      "loss": 0.058,
+      "step": 75600
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 4.725192543810693e-06,
+      "loss": 0.0225,
+      "step": 75610
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 4.721843955798638e-06,
+      "loss": 0.0625,
+      "step": 75620
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 4.718495367786584e-06,
+      "loss": 0.0202,
+      "step": 75630
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 4.715146779774528e-06,
+      "loss": 0.1149,
+      "step": 75640
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 4.711798191762474e-06,
+      "loss": 0.0477,
+      "step": 75650
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 4.708449603750419e-06,
+      "loss": 0.038,
+      "step": 75660
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 4.705101015738364e-06,
+      "loss": 0.0204,
+      "step": 75670
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 4.701752427726309e-06,
+      "loss": 0.0121,
+      "step": 75680
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 4.698403839714254e-06,
+      "loss": 0.0192,
+      "step": 75690
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 4.695055251702199e-06,
+      "loss": 0.0836,
+      "step": 75700
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 4.691706663690144e-06,
+      "loss": 0.0656,
+      "step": 75710
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 4.688358075678089e-06,
+      "loss": 0.0289,
+      "step": 75720
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 4.685009487666034e-06,
+      "loss": 0.0649,
+      "step": 75730
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 4.68166089965398e-06,
+      "loss": 0.0731,
+      "step": 75740
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 4.678312311641924e-06,
+      "loss": 0.0827,
+      "step": 75750
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 4.67496372362987e-06,
+      "loss": 0.05,
+      "step": 75760
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 4.671615135617815e-06,
+      "loss": 0.029,
+      "step": 75770
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 4.6682665476057596e-06,
+      "loss": 0.0506,
+      "step": 75780
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 4.664917959593705e-06,
+      "loss": 0.0215,
+      "step": 75790
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 4.661569371581649e-06,
+      "loss": 0.0178,
+      "step": 75800
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 4.658220783569595e-06,
+      "loss": 0.0718,
+      "step": 75810
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 4.654872195557539e-06,
+      "loss": 0.0152,
+      "step": 75820
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 4.651523607545485e-06,
+      "loss": 0.0647,
+      "step": 75830
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 4.648175019533431e-06,
+      "loss": 0.0528,
+      "step": 75840
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 4.644826431521375e-06,
+      "loss": 0.0844,
+      "step": 75850
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 4.641477843509321e-06,
+      "loss": 0.1039,
+      "step": 75860
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 4.6381292554972656e-06,
+      "loss": 0.0698,
+      "step": 75870
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 4.6347806674852105e-06,
+      "loss": 0.0353,
+      "step": 75880
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 4.631432079473156e-06,
+      "loss": 0.0307,
+      "step": 75890
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 4.6280834914611e-06,
+      "loss": 0.0769,
+      "step": 75900
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 4.624734903449046e-06,
+      "loss": 0.0343,
+      "step": 75910
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 4.621386315436991e-06,
+      "loss": 0.0808,
+      "step": 75920
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.618037727424936e-06,
+      "loss": 0.0272,
+      "step": 75930
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.614689139412881e-06,
+      "loss": 0.0452,
+      "step": 75940
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.611340551400826e-06,
+      "loss": 0.0598,
+      "step": 75950
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.6079919633887715e-06,
+      "loss": 0.0315,
+      "step": 75960
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.6046433753767165e-06,
+      "loss": 0.1412,
+      "step": 75970
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.601294787364661e-06,
+      "loss": 0.0319,
+      "step": 75980
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.597946199352606e-06,
+      "loss": 0.0805,
+      "step": 75990
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.594597611340551e-06,
+      "loss": 0.0672,
+      "step": 76000
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.591249023328496e-06,
+      "loss": 0.0435,
+      "step": 76010
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.587900435316442e-06,
+      "loss": 0.0429,
+      "step": 76020
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.584551847304387e-06,
+      "loss": 0.0246,
+      "step": 76030
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.581203259292332e-06,
+      "loss": 0.0444,
+      "step": 76040
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.5778546712802775e-06,
+      "loss": 0.0421,
+      "step": 76050
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.574506083268222e-06,
+      "loss": 0.0313,
+      "step": 76060
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.571157495256167e-06,
+      "loss": 0.0753,
+      "step": 76070
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.567808907244112e-06,
+      "loss": 0.1056,
+      "step": 76080
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.564460319232057e-06,
+      "loss": 0.026,
+      "step": 76090
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.561111731220003e-06,
+      "loss": 0.0455,
+      "step": 76100
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.557763143207947e-06,
+      "loss": 0.0915,
+      "step": 76110
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.554414555195893e-06,
+      "loss": 0.0762,
+      "step": 76120
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.551065967183837e-06,
+      "loss": 0.0321,
+      "step": 76130
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.547717379171783e-06,
+      "loss": 0.0635,
+      "step": 76140
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.5443687911597284e-06,
+      "loss": 0.062,
+      "step": 76150
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.5410202031476725e-06,
+      "loss": 0.0367,
+      "step": 76160
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.537671615135618e-06,
+      "loss": 0.047,
+      "step": 76170
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.534323027123563e-06,
+      "loss": 0.0608,
+      "step": 76180
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.530974439111508e-06,
+      "loss": 0.0537,
+      "step": 76190
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.527625851099453e-06,
+      "loss": 0.0298,
+      "step": 76200
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.524277263087398e-06,
+      "loss": 0.0253,
+      "step": 76210
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.520928675075344e-06,
+      "loss": 0.0577,
+      "step": 76220
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.517580087063289e-06,
+      "loss": 0.0718,
+      "step": 76230
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.5142314990512336e-06,
+      "loss": 0.0751,
+      "step": 76240
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.5108829110391785e-06,
+      "loss": 0.0371,
+      "step": 76250
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.507534323027123e-06,
+      "loss": 0.0612,
+      "step": 76260
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.504185735015068e-06,
+      "loss": 0.1205,
+      "step": 76270
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.500837147003014e-06,
+      "loss": 0.0453,
+      "step": 76280
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.497488558990959e-06,
+      "loss": 0.0156,
+      "step": 76290
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.494139970978904e-06,
+      "loss": 0.0751,
+      "step": 76300
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.490791382966849e-06,
+      "loss": 0.0718,
+      "step": 76310
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.487442794954794e-06,
+      "loss": 0.0445,
+      "step": 76320
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.4840942069427395e-06,
+      "loss": 0.0388,
+      "step": 76330
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.4807456189306845e-06,
+      "loss": 0.0527,
+      "step": 76340
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.477397030918629e-06,
+      "loss": 0.0337,
+      "step": 76350
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.474048442906575e-06,
+      "loss": 0.1271,
+      "step": 76360
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.470699854894519e-06,
+      "loss": 0.0717,
+      "step": 76370
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.467351266882465e-06,
+      "loss": 0.0349,
+      "step": 76380
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.464002678870409e-06,
+      "loss": 0.0121,
+      "step": 76390
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.460654090858355e-06,
+      "loss": 0.0568,
+      "step": 76400
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.457305502846301e-06,
+      "loss": 0.129,
+      "step": 76410
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.453956914834245e-06,
+      "loss": 0.0272,
+      "step": 76420
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.4506083268221905e-06,
+      "loss": 0.0545,
+      "step": 76430
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.4472597388101345e-06,
+      "loss": 0.0652,
+      "step": 76440
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.44391115079808e-06,
+      "loss": 0.0427,
+      "step": 76450
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.440562562786025e-06,
+      "loss": 0.0869,
+      "step": 76460
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.43721397477397e-06,
+      "loss": 0.04,
+      "step": 76470
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.433865386761916e-06,
+      "loss": 0.0252,
+      "step": 76480
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.430516798749861e-06,
+      "loss": 0.0282,
+      "step": 76490
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.427168210737806e-06,
+      "loss": 0.066,
+      "step": 76500
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.423819622725751e-06,
+      "loss": 0.0454,
+      "step": 76510
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.420471034713696e-06,
+      "loss": 0.1015,
+      "step": 76520
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.4171224467016405e-06,
+      "loss": 0.0285,
+      "step": 76530
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.413773858689586e-06,
+      "loss": 0.0392,
+      "step": 76540
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.410425270677531e-06,
+      "loss": 0.0447,
+      "step": 76550
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.407076682665476e-06,
+      "loss": 0.0317,
+      "step": 76560
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.403728094653421e-06,
+      "loss": 0.0685,
+      "step": 76570
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.400379506641366e-06,
+      "loss": 0.0201,
+      "step": 76580
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.397030918629312e-06,
+      "loss": 0.0419,
+      "step": 76590
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.393682330617257e-06,
+      "loss": 0.073,
+      "step": 76600
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.3903337426052016e-06,
+      "loss": 0.0219,
+      "step": 76610
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.3869851545931465e-06,
+      "loss": 0.0364,
+      "step": 76620
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.383636566581091e-06,
+      "loss": 0.0389,
+      "step": 76630
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.380287978569037e-06,
+      "loss": 0.0582,
+      "step": 76640
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.376939390556981e-06,
+      "loss": 0.0317,
+      "step": 76650
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.373590802544927e-06,
+      "loss": 0.0377,
+      "step": 76660
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.370242214532873e-06,
+      "loss": 0.0774,
+      "step": 76670
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.366893626520817e-06,
+      "loss": 0.0332,
+      "step": 76680
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.363545038508763e-06,
+      "loss": 0.0753,
+      "step": 76690
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.360196450496707e-06,
+      "loss": 0.1089,
+      "step": 76700
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.3568478624846525e-06,
+      "loss": 0.0196,
+      "step": 76710
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.353499274472597e-06,
+      "loss": 0.0327,
+      "step": 76720
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.350150686460542e-06,
+      "loss": 0.0068,
+      "step": 76730
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.346802098448488e-06,
+      "loss": 0.0355,
+      "step": 76740
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.343453510436432e-06,
+      "loss": 0.0166,
+      "step": 76750
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.340104922424378e-06,
+      "loss": 0.0839,
+      "step": 76760
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.337091193213529e-06,
+      "loss": 0.0297,
+      "step": 76770
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.333742605201473e-06,
+      "loss": 0.0638,
+      "step": 76780
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.3303940171894185e-06,
+      "loss": 0.009,
+      "step": 76790
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.327045429177364e-06,
+      "loss": 0.0825,
+      "step": 76800
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.323696841165308e-06,
+      "loss": 0.0284,
+      "step": 76810
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.320348253153254e-06,
+      "loss": 0.0171,
+      "step": 76820
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.316999665141199e-06,
+      "loss": 0.0417,
+      "step": 76830
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.313651077129144e-06,
+      "loss": 0.0786,
+      "step": 76840
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.31030248911709e-06,
+      "loss": 0.0993,
+      "step": 76850
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.306953901105034e-06,
+      "loss": 0.0913,
+      "step": 76860
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.3036053130929795e-06,
+      "loss": 0.0255,
+      "step": 76870
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.3002567250809244e-06,
+      "loss": 0.044,
+      "step": 76880
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.296908137068869e-06,
+      "loss": 0.0724,
+      "step": 76890
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.293559549056814e-06,
+      "loss": 0.0286,
+      "step": 76900
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.290210961044759e-06,
+      "loss": 0.0361,
+      "step": 76910
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.286862373032705e-06,
+      "loss": 0.0381,
+      "step": 76920
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.28351378502065e-06,
+      "loss": 0.0241,
+      "step": 76930
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.280165197008595e-06,
+      "loss": 0.0874,
+      "step": 76940
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.27681660899654e-06,
+      "loss": 0.0464,
+      "step": 76950
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.273468020984485e-06,
+      "loss": 0.063,
+      "step": 76960
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.27011943297243e-06,
+      "loss": 0.1034,
+      "step": 76970
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.266770844960375e-06,
+      "loss": 0.113,
+      "step": 76980
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.26342225694832e-06,
+      "loss": 0.0578,
+      "step": 76990
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.260073668936265e-06,
+      "loss": 0.0454,
+      "step": 77000
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.256725080924211e-06,
+      "loss": 0.061,
+      "step": 77010
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.253376492912155e-06,
+      "loss": 0.0534,
+      "step": 77020
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.250027904900101e-06,
+      "loss": 0.0526,
+      "step": 77030
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.246679316888046e-06,
+      "loss": 0.0297,
+      "step": 77040
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.243330728875991e-06,
+      "loss": 0.0364,
+      "step": 77050
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.239982140863936e-06,
+      "loss": 0.0135,
+      "step": 77060
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.2366335528518805e-06,
+      "loss": 0.078,
+      "step": 77070
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.233284964839826e-06,
+      "loss": 0.0106,
+      "step": 77080
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.22993637682777e-06,
+      "loss": 0.1018,
+      "step": 77090
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.226587788815716e-06,
+      "loss": 0.0313,
+      "step": 77100
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.223239200803662e-06,
+      "loss": 0.0294,
+      "step": 77110
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.219890612791606e-06,
+      "loss": 0.0657,
+      "step": 77120
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.216542024779552e-06,
+      "loss": 0.0189,
+      "step": 77130
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.213193436767497e-06,
+      "loss": 0.0366,
+      "step": 77140
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.2098448487554415e-06,
+      "loss": 0.0256,
+      "step": 77150
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.2064962607433865e-06,
+      "loss": 0.034,
+      "step": 77160
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.203147672731331e-06,
+      "loss": 0.0543,
+      "step": 77170
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.199799084719277e-06,
+      "loss": 0.0856,
+      "step": 77180
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.196450496707222e-06,
+      "loss": 0.0561,
+      "step": 77190
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.193101908695167e-06,
+      "loss": 0.0513,
+      "step": 77200
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.189753320683112e-06,
+      "loss": 0.0379,
+      "step": 77210
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.186404732671057e-06,
+      "loss": 0.0169,
+      "step": 77220
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.183056144659002e-06,
+      "loss": 0.0271,
+      "step": 77230
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.1797075566469475e-06,
+      "loss": 0.0741,
+      "step": 77240
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.1763589686348925e-06,
+      "loss": 0.0957,
+      "step": 77250
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.173010380622837e-06,
+      "loss": 0.0676,
+      "step": 77260
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.169661792610783e-06,
+      "loss": 0.0603,
+      "step": 77270
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.166313204598727e-06,
+      "loss": 0.0419,
+      "step": 77280
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.162964616586673e-06,
+      "loss": 0.0236,
+      "step": 77290
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.159616028574618e-06,
+      "loss": 0.0459,
+      "step": 77300
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.156267440562563e-06,
+      "loss": 0.1163,
+      "step": 77310
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.152918852550509e-06,
+      "loss": 0.0599,
+      "step": 77320
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.149570264538453e-06,
+      "loss": 0.0781,
+      "step": 77330
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.1462216765263984e-06,
+      "loss": 0.0657,
+      "step": 77340
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.1428730885143425e-06,
+      "loss": 0.0347,
+      "step": 77350
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.139524500502288e-06,
+      "loss": 0.0485,
+      "step": 77360
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.136175912490234e-06,
+      "loss": 0.0744,
+      "step": 77370
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.132827324478178e-06,
+      "loss": 0.062,
+      "step": 77380
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.129478736466124e-06,
+      "loss": 0.05,
+      "step": 77390
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.126130148454068e-06,
+      "loss": 0.0105,
+      "step": 77400
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.122781560442014e-06,
+      "loss": 0.0393,
+      "step": 77410
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.119432972429959e-06,
+      "loss": 0.0117,
+      "step": 77420
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.1160843844179036e-06,
+      "loss": 0.0553,
+      "step": 77430
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.112735796405849e-06,
+      "loss": 0.0433,
+      "step": 77440
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.109387208393794e-06,
+      "loss": 0.0484,
+      "step": 77450
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.106038620381739e-06,
+      "loss": 0.0514,
+      "step": 77460
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.102690032369684e-06,
+      "loss": 0.0165,
+      "step": 77470
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.099341444357629e-06,
+      "loss": 0.0307,
+      "step": 77480
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.095992856345575e-06,
+      "loss": 0.0197,
+      "step": 77490
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.09264426833352e-06,
+      "loss": 0.0313,
+      "step": 77500
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.089295680321465e-06,
+      "loss": 0.048,
+      "step": 77510
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.0859470923094096e-06,
+      "loss": 0.0638,
+      "step": 77520
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.0825985042973545e-06,
+      "loss": 0.0333,
+      "step": 77530
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.079249916285299e-06,
+      "loss": 0.0264,
+      "step": 77540
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.075901328273245e-06,
+      "loss": 0.0323,
+      "step": 77550
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.07255274026119e-06,
+      "loss": 0.0598,
+      "step": 77560
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.069204152249135e-06,
+      "loss": 0.0144,
+      "step": 77570
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.065855564237081e-06,
+      "loss": 0.0197,
+      "step": 77580
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.062506976225025e-06,
+      "loss": 0.0739,
+      "step": 77590
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.059158388212971e-06,
+      "loss": 0.0994,
+      "step": 77600
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.055809800200915e-06,
+      "loss": 0.06,
+      "step": 77610
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.0524612121888605e-06,
+      "loss": 0.0531,
+      "step": 77620
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.049112624176806e-06,
+      "loss": 0.0428,
+      "step": 77630
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.04576403616475e-06,
+      "loss": 0.062,
+      "step": 77640
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.042415448152696e-06,
+      "loss": 0.0856,
+      "step": 77650
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.03906686014064e-06,
+      "loss": 0.0591,
+      "step": 77660
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.035718272128586e-06,
+      "loss": 0.0605,
+      "step": 77670
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.032369684116531e-06,
+      "loss": 0.1055,
+      "step": 77680
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.029021096104476e-06,
+      "loss": 0.0459,
+      "step": 77690
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.0256725080924215e-06,
+      "loss": 0.087,
+      "step": 77700
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.0223239200803664e-06,
+      "loss": 0.0427,
+      "step": 77710
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.018975332068311e-06,
+      "loss": 0.0816,
+      "step": 77720
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.015626744056256e-06,
+      "loss": 0.0365,
+      "step": 77730
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.012278156044201e-06,
+      "loss": 0.0327,
+      "step": 77740
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.008929568032147e-06,
+      "loss": 0.0793,
+      "step": 77750
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.005580980020092e-06,
+      "loss": 0.0474,
+      "step": 77760
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.002232392008037e-06,
+      "loss": 0.0067,
+      "step": 77770
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 3.998883803995982e-06,
+      "loss": 0.0475,
+      "step": 77780
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 3.995535215983927e-06,
+      "loss": 0.1078,
+      "step": 77790
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 3.992186627971872e-06,
+      "loss": 0.0274,
+      "step": 77800
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 3.988838039959817e-06,
+      "loss": 0.0382,
+      "step": 77810
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 3.985489451947762e-06,
+      "loss": 0.0503,
+      "step": 77820
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 3.982140863935707e-06,
+      "loss": 0.0241,
+      "step": 77830
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 3.978792275923652e-06,
+      "loss": 0.0454,
+      "step": 77840
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 3.975443687911597e-06,
+      "loss": 0.1064,
+      "step": 77850
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 3.972095099899543e-06,
+      "loss": 0.0567,
+      "step": 77860
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 3.968746511887487e-06,
+      "loss": 0.0438,
+      "step": 77870
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 3.965397923875433e-06,
+      "loss": 0.0688,
+      "step": 77880
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 3.962049335863378e-06,
+      "loss": 0.0085,
+      "step": 77890
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.9587007478513225e-06,
+      "loss": 0.0564,
+      "step": 77900
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.955352159839268e-06,
+      "loss": 0.0284,
+      "step": 77910
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.952003571827212e-06,
+      "loss": 0.0526,
+      "step": 77920
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.948654983815158e-06,
+      "loss": 0.0288,
+      "step": 77930
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.945306395803104e-06,
+      "loss": 0.0494,
+      "step": 77940
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.941957807791048e-06,
+      "loss": 0.0462,
+      "step": 77950
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.938609219778994e-06,
+      "loss": 0.0937,
+      "step": 77960
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.935260631766938e-06,
+      "loss": 0.094,
+      "step": 77970
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.9319120437548836e-06,
+      "loss": 0.0802,
+      "step": 77980
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.9285634557428285e-06,
+      "loss": 0.0109,
+      "step": 77990
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.925214867730773e-06,
+      "loss": 0.0779,
+      "step": 78000
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.921866279718719e-06,
+      "loss": 0.043,
+      "step": 78010
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.918517691706664e-06,
+      "loss": 0.0632,
+      "step": 78020
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.915169103694609e-06,
+      "loss": 0.0361,
+      "step": 78030
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.911820515682554e-06,
+      "loss": 0.0163,
+      "step": 78040
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.908471927670499e-06,
+      "loss": 0.031,
+      "step": 78050
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.905123339658444e-06,
+      "loss": 0.0364,
+      "step": 78060
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 3.9017747516463895e-06,
+      "loss": 0.0329,
+      "step": 78070
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.8984261636343345e-06,
+      "loss": 0.0179,
+      "step": 78080
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.895077575622279e-06,
+      "loss": 0.0597,
+      "step": 78090
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.891728987610224e-06,
+      "loss": 0.0175,
+      "step": 78100
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.888380399598169e-06,
+      "loss": 0.0649,
+      "step": 78110
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.885031811586115e-06,
+      "loss": 0.1036,
+      "step": 78120
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.881683223574059e-06,
+      "loss": 0.0404,
+      "step": 78130
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.878334635562005e-06,
+      "loss": 0.0285,
+      "step": 78140
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.87498604754995e-06,
+      "loss": 0.052,
+      "step": 78150
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.871637459537895e-06,
+      "loss": 0.2775,
+      "step": 78160
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.8682888715258404e-06,
+      "loss": 0.0181,
+      "step": 78170
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.8649402835137845e-06,
+      "loss": 0.0357,
+      "step": 78180
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.86159169550173e-06,
+      "loss": 0.047,
+      "step": 78190
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.858243107489676e-06,
+      "loss": 0.0245,
+      "step": 78200
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.85489451947762e-06,
+      "loss": 0.0742,
+      "step": 78210
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.851545931465566e-06,
+      "loss": 0.0295,
+      "step": 78220
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.84819734345351e-06,
+      "loss": 0.0762,
+      "step": 78230
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.844848755441456e-06,
+      "loss": 0.0781,
+      "step": 78240
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.841500167429401e-06,
+      "loss": 0.077,
+      "step": 78250
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.838151579417346e-06,
+      "loss": 0.0876,
+      "step": 78260
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.834802991405291e-06,
+      "loss": 0.0497,
+      "step": 78270
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.831454403393235e-06,
+      "loss": 0.0412,
+      "step": 78280
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.828105815381181e-06,
+      "loss": 0.0491,
+      "step": 78290
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.824757227369126e-06,
+      "loss": 0.0586,
+      "step": 78300
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.821408639357071e-06,
+      "loss": 0.0288,
+      "step": 78310
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.818060051345016e-06,
+      "loss": 0.0421,
+      "step": 78320
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.814711463332962e-06,
+      "loss": 0.054,
+      "step": 78330
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.8113628753209066e-06,
+      "loss": 0.035,
+      "step": 78340
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.8080142873088516e-06,
+      "loss": 0.0783,
+      "step": 78350
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.804665699296797e-06,
+      "loss": 0.0776,
+      "step": 78360
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.8013171112847414e-06,
+      "loss": 0.0353,
+      "step": 78370
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.7979685232726867e-06,
+      "loss": 0.0741,
+      "step": 78380
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.794619935260632e-06,
+      "loss": 0.0094,
+      "step": 78390
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.791271347248577e-06,
+      "loss": 0.0406,
+      "step": 78400
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.7879227592365224e-06,
+      "loss": 0.0167,
+      "step": 78410
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.784574171224467e-06,
+      "loss": 0.0796,
+      "step": 78420
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 3.781225583212412e-06,
+      "loss": 0.0512,
+      "step": 78430
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.777876995200357e-06,
+      "loss": 0.0709,
+      "step": 78440
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.7745284071883025e-06,
+      "loss": 0.1041,
+      "step": 78450
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.771179819176248e-06,
+      "loss": 0.1323,
+      "step": 78460
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.7678312311641923e-06,
+      "loss": 0.0384,
+      "step": 78470
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.7644826431521377e-06,
+      "loss": 0.0477,
+      "step": 78480
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.7611340551400826e-06,
+      "loss": 0.0217,
+      "step": 78490
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.757785467128028e-06,
+      "loss": 0.0657,
+      "step": 78500
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.7544368791159724e-06,
+      "loss": 0.0435,
+      "step": 78510
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.7510882911039178e-06,
+      "loss": 0.0195,
+      "step": 78520
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.747739703091863e-06,
+      "loss": 0.0268,
+      "step": 78530
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.744391115079808e-06,
+      "loss": 0.1129,
+      "step": 78540
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.741042527067753e-06,
+      "loss": 0.0316,
+      "step": 78550
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.7376939390556983e-06,
+      "loss": 0.0434,
+      "step": 78560
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.7343453510436436e-06,
+      "loss": 0.0163,
+      "step": 78570
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.7309967630315886e-06,
+      "loss": 0.1504,
+      "step": 78580
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.7276481750195335e-06,
+      "loss": 0.0396,
+      "step": 78590
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.7242995870074784e-06,
+      "loss": 0.0083,
+      "step": 78600
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 3.7209509989954233e-06,
+      "loss": 0.0127,
+      "step": 78610
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.717602410983369e-06,
+      "loss": 0.0437,
+      "step": 78620
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.714253822971314e-06,
+      "loss": 0.0634,
+      "step": 78630
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.710905234959259e-06,
+      "loss": 0.0553,
+      "step": 78640
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.707556646947204e-06,
+      "loss": 0.0289,
+      "step": 78650
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.704208058935149e-06,
+      "loss": 0.0801,
+      "step": 78660
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.700859470923094e-06,
+      "loss": 0.0219,
+      "step": 78670
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.6975108829110395e-06,
+      "loss": 0.0555,
+      "step": 78680
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.6941622948989844e-06,
+      "loss": 0.0365,
+      "step": 78690
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.6908137068869293e-06,
+      "loss": 0.0524,
+      "step": 78700
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.6874651188748747e-06,
+      "loss": 0.0463,
+      "step": 78710
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.6841165308628196e-06,
+      "loss": 0.0193,
+      "step": 78720
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.6807679428507645e-06,
+      "loss": 0.0511,
+      "step": 78730
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.6774193548387094e-06,
+      "loss": 0.0599,
+      "step": 78740
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.674070766826655e-06,
+      "loss": 0.0679,
+      "step": 78750
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.6707221788146e-06,
+      "loss": 0.0227,
+      "step": 78760
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.667373590802545e-06,
+      "loss": 0.0333,
+      "step": 78770
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.66402500279049e-06,
+      "loss": 0.0278,
+      "step": 78780
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.6606764147784353e-06,
+      "loss": 0.0532,
+      "step": 78790
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.6573278267663802e-06,
+      "loss": 0.0247,
+      "step": 78800
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.6539792387543256e-06,
+      "loss": 0.0303,
+      "step": 78810
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.6506306507422705e-06,
+      "loss": 0.0207,
+      "step": 78820
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.6472820627302154e-06,
+      "loss": 0.1133,
+      "step": 78830
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.6439334747181607e-06,
+      "loss": 0.0508,
+      "step": 78840
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.6405848867061057e-06,
+      "loss": 0.062,
+      "step": 78850
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.6372362986940506e-06,
+      "loss": 0.0729,
+      "step": 78860
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.6338877106819955e-06,
+      "loss": 0.0876,
+      "step": 78870
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.6305391226699413e-06,
+      "loss": 0.0278,
+      "step": 78880
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.627190534657886e-06,
+      "loss": 0.0231,
+      "step": 78890
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.623841946645831e-06,
+      "loss": 0.055,
+      "step": 78900
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.620493358633776e-06,
+      "loss": 0.0263,
+      "step": 78910
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.617144770621721e-06,
+      "loss": 0.0309,
+      "step": 78920
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.6137961826096663e-06,
+      "loss": 0.0571,
+      "step": 78930
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.6104475945976116e-06,
+      "loss": 0.007,
+      "step": 78940
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.6070990065855566e-06,
+      "loss": 0.0821,
+      "step": 78950
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.6037504185735015e-06,
+      "loss": 0.0497,
+      "step": 78960
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.600401830561447e-06,
+      "loss": 0.1069,
+      "step": 78970
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.5970532425493918e-06,
+      "loss": 0.0117,
+      "step": 78980
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.5937046545373367e-06,
+      "loss": 0.0223,
+      "step": 78990
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.5903560665252816e-06,
+      "loss": 0.0681,
+      "step": 79000
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.5870074785132274e-06,
+      "loss": 0.0988,
+      "step": 79010
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.5836588905011723e-06,
+      "loss": 0.0552,
+      "step": 79020
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.580310302489117e-06,
+      "loss": 0.0324,
+      "step": 79030
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.576961714477062e-06,
+      "loss": 0.0417,
+      "step": 79040
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.573613126465007e-06,
+      "loss": 0.0885,
+      "step": 79050
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.5702645384529524e-06,
+      "loss": 0.1192,
+      "step": 79060
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.5669159504408977e-06,
+      "loss": 0.0534,
+      "step": 79070
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.5635673624288427e-06,
+      "loss": 0.0154,
+      "step": 79080
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.5602187744167876e-06,
+      "loss": 0.036,
+      "step": 79090
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.556870186404733e-06,
+      "loss": 0.0556,
+      "step": 79100
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.553521598392678e-06,
+      "loss": 0.0435,
+      "step": 79110
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.5501730103806228e-06,
+      "loss": 0.148,
+      "step": 79120
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.546824422368568e-06,
+      "loss": 0.0248,
+      "step": 79130
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.543475834356513e-06,
+      "loss": 0.005,
+      "step": 79140
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.5401272463444584e-06,
+      "loss": 0.0357,
+      "step": 79150
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.5367786583324033e-06,
+      "loss": 0.1241,
+      "step": 79160
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.5334300703203482e-06,
+      "loss": 0.0673,
+      "step": 79170
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.530081482308293e-06,
+      "loss": 0.073,
+      "step": 79180
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.5267328942962385e-06,
+      "loss": 0.036,
+      "step": 79190
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.523384306284184e-06,
+      "loss": 0.0747,
+      "step": 79200
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.5200357182721288e-06,
+      "loss": 0.0245,
+      "step": 79210
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.5166871302600737e-06,
+      "loss": 0.0248,
+      "step": 79220
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.513338542248019e-06,
+      "loss": 0.0659,
+      "step": 79230
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.509989954235964e-06,
+      "loss": 0.0171,
+      "step": 79240
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.506641366223909e-06,
+      "loss": 0.0343,
+      "step": 79250
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.503292778211854e-06,
+      "loss": 0.0909,
+      "step": 79260
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.499944190199799e-06,
+      "loss": 0.0363,
+      "step": 79270
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.4965956021877445e-06,
+      "loss": 0.0472,
+      "step": 79280
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.4932470141756894e-06,
+      "loss": 0.0149,
+      "step": 79290
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.4898984261636343e-06,
+      "loss": 0.0479,
+      "step": 79300
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.4865498381515792e-06,
+      "loss": 0.062,
+      "step": 79310
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.4832012501395246e-06,
+      "loss": 0.0219,
+      "step": 79320
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.47985266212747e-06,
+      "loss": 0.0082,
+      "step": 79330
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.476504074115415e-06,
+      "loss": 0.0543,
+      "step": 79340
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.4731554861033598e-06,
+      "loss": 0.0447,
+      "step": 79350
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.4698068980913047e-06,
+      "loss": 0.1155,
+      "step": 79360
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.46645831007925e-06,
+      "loss": 0.033,
+      "step": 79370
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.463109722067195e-06,
+      "loss": 0.0276,
+      "step": 79380
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.4597611340551403e-06,
+      "loss": 0.0525,
+      "step": 79390
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.4564125460430852e-06,
+      "loss": 0.0264,
+      "step": 79400
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.4530639580310306e-06,
+      "loss": 0.0225,
+      "step": 79410
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.4497153700189755e-06,
+      "loss": 0.039,
+      "step": 79420
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.4463667820069204e-06,
+      "loss": 0.03,
+      "step": 79430
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.4430181939948653e-06,
+      "loss": 0.0536,
+      "step": 79440
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.4396696059828107e-06,
+      "loss": 0.0506,
+      "step": 79450
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.436321017970756e-06,
+      "loss": 0.0525,
+      "step": 79460
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.432972429958701e-06,
+      "loss": 0.0661,
+      "step": 79470
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.429623841946646e-06,
+      "loss": 0.0339,
+      "step": 79480
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.4262752539345908e-06,
+      "loss": 0.0426,
+      "step": 79490
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.422926665922536e-06,
+      "loss": 0.0702,
+      "step": 79500
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.419578077910481e-06,
+      "loss": 0.0801,
+      "step": 79510
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.4162294898984264e-06,
+      "loss": 0.0616,
+      "step": 79520
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.4128809018863713e-06,
+      "loss": 0.0207,
+      "step": 79530
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.4095323138743167e-06,
+      "loss": 0.0637,
+      "step": 79540
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.4061837258622616e-06,
+      "loss": 0.0425,
+      "step": 79550
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.4028351378502065e-06,
+      "loss": 0.0355,
+      "step": 79560
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.3994865498381514e-06,
+      "loss": 0.0678,
+      "step": 79570
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.3961379618260968e-06,
+      "loss": 0.0479,
+      "step": 79580
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.3931242326152474e-06,
+      "loss": 0.0486,
+      "step": 79590
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.3897756446031924e-06,
+      "loss": 0.0942,
+      "step": 79600
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.3864270565911373e-06,
+      "loss": 0.0331,
+      "step": 79610
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.3830784685790826e-06,
+      "loss": 0.0516,
+      "step": 79620
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.3797298805670276e-06,
+      "loss": 0.0404,
+      "step": 79630
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.376381292554973e-06,
+      "loss": 0.011,
+      "step": 79640
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.373032704542918e-06,
+      "loss": 0.0823,
+      "step": 79650
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.369684116530863e-06,
+      "loss": 0.0593,
+      "step": 79660
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.366335528518808e-06,
+      "loss": 0.0754,
+      "step": 79670
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.362986940506753e-06,
+      "loss": 0.0425,
+      "step": 79680
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.359638352494698e-06,
+      "loss": 0.0723,
+      "step": 79690
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.356289764482643e-06,
+      "loss": 0.042,
+      "step": 79700
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.3529411764705886e-06,
+      "loss": 0.0507,
+      "step": 79710
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.3495925884585335e-06,
+      "loss": 0.0764,
+      "step": 79720
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.3462440004464785e-06,
+      "loss": 0.0561,
+      "step": 79730
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.3428954124344234e-06,
+      "loss": 0.0239,
+      "step": 79740
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.3395468244223687e-06,
+      "loss": 0.0531,
+      "step": 79750
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.3361982364103136e-06,
+      "loss": 0.0551,
+      "step": 79760
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.332849648398259e-06,
+      "loss": 0.0438,
+      "step": 79770
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.329501060386204e-06,
+      "loss": 0.0397,
+      "step": 79780
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.326152472374149e-06,
+      "loss": 0.0263,
+      "step": 79790
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.322803884362094e-06,
+      "loss": 0.0917,
+      "step": 79800
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.319455296350039e-06,
+      "loss": 0.0472,
+      "step": 79810
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.316106708337984e-06,
+      "loss": 0.0512,
+      "step": 79820
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.312758120325929e-06,
+      "loss": 0.0459,
+      "step": 79830
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.3094095323138747e-06,
+      "loss": 0.0229,
+      "step": 79840
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.3060609443018196e-06,
+      "loss": 0.0372,
+      "step": 79850
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.3027123562897646e-06,
+      "loss": 0.0549,
+      "step": 79860
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.2993637682777095e-06,
+      "loss": 0.0635,
+      "step": 79870
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.296015180265655e-06,
+      "loss": 0.0361,
+      "step": 79880
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.2926665922535997e-06,
+      "loss": 0.1273,
+      "step": 79890
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.289318004241545e-06,
+      "loss": 0.0143,
+      "step": 79900
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.28596941622949e-06,
+      "loss": 0.0484,
+      "step": 79910
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.282620828217435e-06,
+      "loss": 0.0239,
+      "step": 79920
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.2792722402053803e-06,
+      "loss": 0.056,
+      "step": 79930
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.275923652193325e-06,
+      "loss": 0.0253,
+      "step": 79940
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.27257506418127e-06,
+      "loss": 0.0616,
+      "step": 79950
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.269226476169215e-06,
+      "loss": 0.0619,
+      "step": 79960
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.265877888157161e-06,
+      "loss": 0.0139,
+      "step": 79970
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.2625293001451057e-06,
+      "loss": 0.0245,
+      "step": 79980
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.2591807121330506e-06,
+      "loss": 0.0185,
+      "step": 79990
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.2558321241209956e-06,
+      "loss": 0.0251,
+      "step": 80000
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.2524835361089405e-06,
+      "loss": 0.0166,
+      "step": 80010
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.249134948096886e-06,
+      "loss": 0.0465,
+      "step": 80020
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.245786360084831e-06,
+      "loss": 0.0563,
+      "step": 80030
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.242437772072776e-06,
+      "loss": 0.0574,
+      "step": 80040
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.239089184060721e-06,
+      "loss": 0.0653,
+      "step": 80050
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.2357405960486664e-06,
+      "loss": 0.0312,
+      "step": 80060
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.2323920080366113e-06,
+      "loss": 0.0383,
+      "step": 80070
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.229043420024556e-06,
+      "loss": 0.0388,
+      "step": 80080
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.2256948320125016e-06,
+      "loss": 0.0176,
+      "step": 80090
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.222346244000447e-06,
+      "loss": 0.0176,
+      "step": 80100
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.218997655988392e-06,
+      "loss": 0.0241,
+      "step": 80110
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.2156490679763367e-06,
+      "loss": 0.0116,
+      "step": 80120
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.2123004799642817e-06,
+      "loss": 0.0549,
+      "step": 80130
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.2089518919522266e-06,
+      "loss": 0.0449,
+      "step": 80140
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.205603303940172e-06,
+      "loss": 0.0481,
+      "step": 80150
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.2022547159281173e-06,
+      "loss": 0.0439,
+      "step": 80160
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.198906127916062e-06,
+      "loss": 0.0375,
+      "step": 80170
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.195557539904007e-06,
+      "loss": 0.0178,
+      "step": 80180
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.1922089518919525e-06,
+      "loss": 0.0418,
+      "step": 80190
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.1888603638798974e-06,
+      "loss": 0.0566,
+      "step": 80200
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.1855117758678423e-06,
+      "loss": 0.1238,
+      "step": 80210
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.1821631878557876e-06,
+      "loss": 0.0157,
+      "step": 80220
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.1788145998437326e-06,
+      "loss": 0.0208,
+      "step": 80230
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.175466011831678e-06,
+      "loss": 0.0445,
+      "step": 80240
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.172117423819623e-06,
+      "loss": 0.0167,
+      "step": 80250
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.1687688358075678e-06,
+      "loss": 0.0492,
+      "step": 80260
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.1654202477955127e-06,
+      "loss": 0.0114,
+      "step": 80270
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.162071659783458e-06,
+      "loss": 0.0218,
+      "step": 80280
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.1587230717714034e-06,
+      "loss": 0.0628,
+      "step": 80290
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.1553744837593483e-06,
+      "loss": 0.0283,
+      "step": 80300
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.152025895747293e-06,
+      "loss": 0.0184,
+      "step": 80310
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.1486773077352385e-06,
+      "loss": 0.0873,
+      "step": 80320
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.1453287197231835e-06,
+      "loss": 0.0254,
+      "step": 80330
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.1419801317111284e-06,
+      "loss": 0.1,
+      "step": 80340
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.1386315436990737e-06,
+      "loss": 0.0294,
+      "step": 80350
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.1352829556870187e-06,
+      "loss": 0.0583,
+      "step": 80360
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.131934367674964e-06,
+      "loss": 0.0639,
+      "step": 80370
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.128585779662909e-06,
+      "loss": 0.0468,
+      "step": 80380
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.125237191650854e-06,
+      "loss": 0.0324,
+      "step": 80390
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.1218886036387988e-06,
+      "loss": 0.0522,
+      "step": 80400
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.118540015626744e-06,
+      "loss": 0.0243,
+      "step": 80410
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.1151914276146895e-06,
+      "loss": 0.0239,
+      "step": 80420
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.1118428396026344e-06,
+      "loss": 0.0476,
+      "step": 80430
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.1084942515905793e-06,
+      "loss": 0.0221,
+      "step": 80440
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.1051456635785242e-06,
+      "loss": 0.1198,
+      "step": 80450
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.1017970755664696e-06,
+      "loss": 0.0213,
+      "step": 80460
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.0984484875544145e-06,
+      "loss": 0.0218,
+      "step": 80470
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.09509989954236e-06,
+      "loss": 0.0208,
+      "step": 80480
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.0917513115303047e-06,
+      "loss": 0.0146,
+      "step": 80490
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.08840272351825e-06,
+      "loss": 0.038,
+      "step": 80500
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.085054135506195e-06,
+      "loss": 0.083,
+      "step": 80510
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.08170554749414e-06,
+      "loss": 0.007,
+      "step": 80520
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.078356959482085e-06,
+      "loss": 0.0218,
+      "step": 80530
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.0750083714700306e-06,
+      "loss": 0.0394,
+      "step": 80540
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.0716597834579755e-06,
+      "loss": 0.025,
+      "step": 80550
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.0683111954459205e-06,
+      "loss": 0.0858,
+      "step": 80560
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.0649626074338654e-06,
+      "loss": 0.0751,
+      "step": 80570
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.0616140194218103e-06,
+      "loss": 0.0371,
+      "step": 80580
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.0582654314097557e-06,
+      "loss": 0.0402,
+      "step": 80590
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.0549168433977006e-06,
+      "loss": 0.0477,
+      "step": 80600
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.051568255385646e-06,
+      "loss": 0.1246,
+      "step": 80610
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.048219667373591e-06,
+      "loss": 0.0237,
+      "step": 80620
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.044871079361536e-06,
+      "loss": 0.0288,
+      "step": 80630
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.041522491349481e-06,
+      "loss": 0.0353,
+      "step": 80640
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.038173903337426e-06,
+      "loss": 0.0353,
+      "step": 80650
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.034825315325371e-06,
+      "loss": 0.0868,
+      "step": 80660
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.0314767273133163e-06,
+      "loss": 0.0615,
+      "step": 80670
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.0281281393012616e-06,
+      "loss": 0.0539,
+      "step": 80680
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.0247795512892066e-06,
+      "loss": 0.0399,
+      "step": 80690
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.0214309632771515e-06,
+      "loss": 0.0732,
+      "step": 80700
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.0180823752650964e-06,
+      "loss": 0.0696,
+      "step": 80710
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.0147337872530417e-06,
+      "loss": 0.035,
+      "step": 80720
+    },
+    {
+      "epoch": 9.0,
+      "eval_loss": 0.3149695098400116,
+      "eval_runtime": 251.0688,
+      "eval_samples_per_second": 7.942,
+      "eval_steps_per_second": 3.971,
+      "eval_wer": 0.08409028769328918,
+      "step": 80721
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.0113851992409867e-06,
+      "loss": 0.0428,
+      "step": 80730
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.008036611228932e-06,
+      "loss": 0.0224,
+      "step": 80740
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.004688023216877e-06,
+      "loss": 0.0398,
+      "step": 80750
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.0013394352048223e-06,
+      "loss": 0.0135,
+      "step": 80760
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.997990847192767e-06,
+      "loss": 0.0329,
+      "step": 80770
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.994642259180712e-06,
+      "loss": 0.0427,
+      "step": 80780
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.991293671168657e-06,
+      "loss": 0.0586,
+      "step": 80790
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.9879450831566024e-06,
+      "loss": 0.0881,
+      "step": 80800
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.9845964951445477e-06,
+      "loss": 0.0502,
+      "step": 80810
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.9812479071324927e-06,
+      "loss": 0.027,
+      "step": 80820
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.9778993191204376e-06,
+      "loss": 0.0282,
+      "step": 80830
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.9745507311083825e-06,
+      "loss": 0.061,
+      "step": 80840
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.971202143096328e-06,
+      "loss": 0.0253,
+      "step": 80850
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.9678535550842728e-06,
+      "loss": 0.0127,
+      "step": 80860
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.964504967072218e-06,
+      "loss": 0.1232,
+      "step": 80870
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.961156379060163e-06,
+      "loss": 0.0306,
+      "step": 80880
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.957807791048108e-06,
+      "loss": 0.0111,
+      "step": 80890
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.9547940618372586e-06,
+      "loss": 0.0795,
+      "step": 80900
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.9514454738252035e-06,
+      "loss": 0.0488,
+      "step": 80910
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.948096885813149e-06,
+      "loss": 0.0665,
+      "step": 80920
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.9447482978010942e-06,
+      "loss": 0.0197,
+      "step": 80930
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.941399709789039e-06,
+      "loss": 0.0708,
+      "step": 80940
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.938051121776984e-06,
+      "loss": 0.0592,
+      "step": 80950
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.934702533764929e-06,
+      "loss": 0.0331,
+      "step": 80960
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.9313539457528743e-06,
+      "loss": 0.0473,
+      "step": 80970
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.9280053577408193e-06,
+      "loss": 0.0689,
+      "step": 80980
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.9246567697287646e-06,
+      "loss": 0.0261,
+      "step": 80990
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.9213081817167095e-06,
+      "loss": 0.0206,
+      "step": 81000
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.9179595937046545e-06,
+      "loss": 0.0163,
+      "step": 81010
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.9146110056926e-06,
+      "loss": 0.0504,
+      "step": 81020
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.9112624176805447e-06,
+      "loss": 0.0732,
+      "step": 81030
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 2.9079138296684896e-06,
+      "loss": 0.0432,
+      "step": 81040
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 2.904565241656435e-06,
+      "loss": 0.0642,
+      "step": 81050
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 2.9012166536443803e-06,
+      "loss": 0.0199,
+      "step": 81060
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 2.8978680656323253e-06,
+      "loss": 0.0756,
+      "step": 81070
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 2.89451947762027e-06,
+      "loss": 0.021,
+      "step": 81080
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 2.891170889608215e-06,
+      "loss": 0.061,
+      "step": 81090
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 2.8878223015961604e-06,
+      "loss": 0.0372,
+      "step": 81100
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 2.8844737135841054e-06,
+      "loss": 0.0166,
+      "step": 81110
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 2.8811251255720507e-06,
+      "loss": 0.0762,
+      "step": 81120
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 2.8777765375599956e-06,
+      "loss": 0.0185,
+      "step": 81130
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 2.8744279495479405e-06,
+      "loss": 0.0027,
+      "step": 81140
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 2.871079361535886e-06,
+      "loss": 0.0637,
+      "step": 81150
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 2.867730773523831e-06,
+      "loss": 0.0361,
+      "step": 81160
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 2.8643821855117757e-06,
+      "loss": 0.0206,
+      "step": 81170
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 2.861033597499721e-06,
+      "loss": 0.0429,
+      "step": 81180
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 2.8576850094876664e-06,
+      "loss": 0.0413,
+      "step": 81190
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 2.8543364214756113e-06,
+      "loss": 0.0192,
+      "step": 81200
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 2.8513226922647616e-06,
+      "loss": 0.0095,
+      "step": 81210
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 2.847974104252707e-06,
+      "loss": 0.0568,
+      "step": 81220
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 2.844625516240652e-06,
+      "loss": 0.0402,
+      "step": 81230
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 2.8412769282285972e-06,
+      "loss": 0.0488,
+      "step": 81240
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 2.837928340216542e-06,
+      "loss": 0.0612,
+      "step": 81250
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 2.834579752204487e-06,
+      "loss": 0.0303,
+      "step": 81260
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 2.8312311641924324e-06,
+      "loss": 0.0705,
+      "step": 81270
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 2.8278825761803773e-06,
+      "loss": 0.0166,
+      "step": 81280
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 2.8245339881683222e-06,
+      "loss": 0.3543,
+      "step": 81290
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 2.8211854001562676e-06,
+      "loss": 0.0239,
+      "step": 81300
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 2.817836812144213e-06,
+      "loss": 0.0718,
+      "step": 81310
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 2.814488224132158e-06,
+      "loss": 0.0419,
+      "step": 81320
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 2.8111396361201028e-06,
+      "loss": 0.0756,
+      "step": 81330
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 2.8077910481080477e-06,
+      "loss": 0.054,
+      "step": 81340
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 2.8044424600959926e-06,
+      "loss": 0.033,
+      "step": 81350
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 2.801093872083938e-06,
+      "loss": 0.0228,
+      "step": 81360
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 2.7977452840718833e-06,
+      "loss": 0.0362,
+      "step": 81370
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 2.7943966960598282e-06,
+      "loss": 0.0148,
+      "step": 81380
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 2.791048108047773e-06,
+      "loss": 0.0254,
+      "step": 81390
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 2.7876995200357185e-06,
+      "loss": 0.0535,
+      "step": 81400
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 2.7843509320236634e-06,
+      "loss": 0.0141,
+      "step": 81410
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 2.7810023440116083e-06,
+      "loss": 0.078,
+      "step": 81420
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 2.7776537559995537e-06,
+      "loss": 0.0243,
+      "step": 81430
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 2.7743051679874986e-06,
+      "loss": 0.0593,
+      "step": 81440
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 2.770956579975444e-06,
+      "loss": 0.0301,
+      "step": 81450
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 2.767607991963389e-06,
+      "loss": 0.0374,
+      "step": 81460
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 2.7642594039513338e-06,
+      "loss": 0.028,
+      "step": 81470
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 2.7609108159392787e-06,
+      "loss": 0.0532,
+      "step": 81480
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.757562227927224e-06,
+      "loss": 0.0291,
+      "step": 81490
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.7542136399151694e-06,
+      "loss": 0.0231,
+      "step": 81500
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.7508650519031143e-06,
+      "loss": 0.0063,
+      "step": 81510
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.7475164638910592e-06,
+      "loss": 0.0531,
+      "step": 81520
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.7441678758790046e-06,
+      "loss": 0.0557,
+      "step": 81530
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.7408192878669495e-06,
+      "loss": 0.1028,
+      "step": 81540
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.7374706998548944e-06,
+      "loss": 0.0197,
+      "step": 81550
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.7341221118428398e-06,
+      "loss": 0.0206,
+      "step": 81560
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.7307735238307847e-06,
+      "loss": 0.1,
+      "step": 81570
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.72742493581873e-06,
+      "loss": 0.086,
+      "step": 81580
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.724076347806675e-06,
+      "loss": 0.0131,
+      "step": 81590
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.72072775979462e-06,
+      "loss": 0.0167,
+      "step": 81600
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.717379171782565e-06,
+      "loss": 0.0938,
+      "step": 81610
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.71403058377051e-06,
+      "loss": 0.0562,
+      "step": 81620
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.7106819957584555e-06,
+      "loss": 0.0224,
+      "step": 81630
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.7073334077464004e-06,
+      "loss": 0.0334,
+      "step": 81640
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.7039848197343453e-06,
+      "loss": 0.0493,
+      "step": 81650
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.7006362317222903e-06,
+      "loss": 0.0561,
+      "step": 81660
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.6972876437102356e-06,
+      "loss": 0.0237,
+      "step": 81670
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.6939390556981805e-06,
+      "loss": 0.0132,
+      "step": 81680
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.690590467686126e-06,
+      "loss": 0.0562,
+      "step": 81690
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.6872418796740708e-06,
+      "loss": 0.0185,
+      "step": 81700
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.683893291662016e-06,
+      "loss": 0.0152,
+      "step": 81710
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.680544703649961e-06,
+      "loss": 0.0448,
+      "step": 81720
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.677196115637906e-06,
+      "loss": 0.011,
+      "step": 81730
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.673847527625851e-06,
+      "loss": 0.0269,
+      "step": 81740
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.6704989396137962e-06,
+      "loss": 0.0683,
+      "step": 81750
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.6671503516017416e-06,
+      "loss": 0.0213,
+      "step": 81760
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.6638017635896865e-06,
+      "loss": 0.0383,
+      "step": 81770
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.6604531755776314e-06,
+      "loss": 0.0346,
+      "step": 81780
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.6571045875655763e-06,
+      "loss": 0.0142,
+      "step": 81790
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.6537559995535217e-06,
+      "loss": 0.0646,
+      "step": 81800
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.6504074115414666e-06,
+      "loss": 0.0132,
+      "step": 81810
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.647058823529412e-06,
+      "loss": 0.0187,
+      "step": 81820
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.643710235517357e-06,
+      "loss": 0.0455,
+      "step": 81830
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.6403616475053022e-06,
+      "loss": 0.0312,
+      "step": 81840
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.637013059493247e-06,
+      "loss": 0.0487,
+      "step": 81850
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.633664471481192e-06,
+      "loss": 0.02,
+      "step": 81860
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.630315883469137e-06,
+      "loss": 0.0346,
+      "step": 81870
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.6269672954570823e-06,
+      "loss": 0.033,
+      "step": 81880
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.6236187074450277e-06,
+      "loss": 0.0766,
+      "step": 81890
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.6202701194329726e-06,
+      "loss": 0.0486,
+      "step": 81900
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.6169215314209175e-06,
+      "loss": 0.0202,
+      "step": 81910
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.6135729434088624e-06,
+      "loss": 0.053,
+      "step": 81920
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.6102243553968078e-06,
+      "loss": 0.0339,
+      "step": 81930
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.6068757673847527e-06,
+      "loss": 0.1294,
+      "step": 81940
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.603527179372698e-06,
+      "loss": 0.0325,
+      "step": 81950
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.600178591360643e-06,
+      "loss": 0.023,
+      "step": 81960
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.5968300033485883e-06,
+      "loss": 0.0551,
+      "step": 81970
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.5934814153365332e-06,
+      "loss": 0.0226,
+      "step": 81980
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.590132827324478e-06,
+      "loss": 0.0191,
+      "step": 81990
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.586784239312423e-06,
+      "loss": 0.0291,
+      "step": 82000
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.5834356513003684e-06,
+      "loss": 0.0107,
+      "step": 82010
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.5800870632883138e-06,
+      "loss": 0.0783,
+      "step": 82020
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.5767384752762587e-06,
+      "loss": 0.0347,
+      "step": 82030
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.5733898872642036e-06,
+      "loss": 0.0298,
+      "step": 82040
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.5700412992521485e-06,
+      "loss": 0.0482,
+      "step": 82050
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.566692711240094e-06,
+      "loss": 0.0257,
+      "step": 82060
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.563344123228039e-06,
+      "loss": 0.0219,
+      "step": 82070
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.559995535215984e-06,
+      "loss": 0.0153,
+      "step": 82080
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.556646947203929e-06,
+      "loss": 0.0412,
+      "step": 82090
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.553298359191874e-06,
+      "loss": 0.0094,
+      "step": 82100
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.5499497711798193e-06,
+      "loss": 0.0559,
+      "step": 82110
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.5466011831677642e-06,
+      "loss": 0.0865,
+      "step": 82120
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.543252595155709e-06,
+      "loss": 0.0133,
+      "step": 82130
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.5399040071436545e-06,
+      "loss": 0.0373,
+      "step": 82140
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.5365554191316e-06,
+      "loss": 0.0288,
+      "step": 82150
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.5332068311195448e-06,
+      "loss": 0.0235,
+      "step": 82160
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.5298582431074897e-06,
+      "loss": 0.1161,
+      "step": 82170
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.5265096550954346e-06,
+      "loss": 0.0285,
+      "step": 82180
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.52316106708338e-06,
+      "loss": 0.0359,
+      "step": 82190
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.519812479071325e-06,
+      "loss": 0.0336,
+      "step": 82200
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.5164638910592702e-06,
+      "loss": 0.0082,
+      "step": 82210
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.513115303047215e-06,
+      "loss": 0.0319,
+      "step": 82220
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.50976671503516e-06,
+      "loss": 0.0328,
+      "step": 82230
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.5064181270231054e-06,
+      "loss": 0.1067,
+      "step": 82240
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.5030695390110503e-06,
+      "loss": 0.0284,
+      "step": 82250
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.4997209509989953e-06,
+      "loss": 0.0176,
+      "step": 82260
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.4963723629869406e-06,
+      "loss": 0.0493,
+      "step": 82270
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.493023774974886e-06,
+      "loss": 0.0626,
+      "step": 82280
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.489675186962831e-06,
+      "loss": 0.0396,
+      "step": 82290
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.486326598950776e-06,
+      "loss": 0.0462,
+      "step": 82300
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.4829780109387207e-06,
+      "loss": 0.0323,
+      "step": 82310
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.4796294229266656e-06,
+      "loss": 0.107,
+      "step": 82320
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.4762808349146114e-06,
+      "loss": 0.0506,
+      "step": 82330
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.4729322469025563e-06,
+      "loss": 0.0569,
+      "step": 82340
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.4695836588905012e-06,
+      "loss": 0.0425,
+      "step": 82350
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.466235070878446e-06,
+      "loss": 0.0088,
+      "step": 82360
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.4628864828663915e-06,
+      "loss": 0.0113,
+      "step": 82370
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.4595378948543364e-06,
+      "loss": 0.0543,
+      "step": 82380
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.4561893068422814e-06,
+      "loss": 0.0172,
+      "step": 82390
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.4528407188302267e-06,
+      "loss": 0.0355,
+      "step": 82400
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.449492130818172e-06,
+      "loss": 0.0284,
+      "step": 82410
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.446143542806117e-06,
+      "loss": 0.0299,
+      "step": 82420
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.442794954794062e-06,
+      "loss": 0.1558,
+      "step": 82430
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.439446366782007e-06,
+      "loss": 0.0531,
+      "step": 82440
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.4360977787699517e-06,
+      "loss": 0.0131,
+      "step": 82450
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.4327491907578975e-06,
+      "loss": 0.0158,
+      "step": 82460
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.4294006027458424e-06,
+      "loss": 0.1034,
+      "step": 82470
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.4260520147337873e-06,
+      "loss": 0.0368,
+      "step": 82480
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.4227034267217323e-06,
+      "loss": 0.0302,
+      "step": 82490
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.4193548387096776e-06,
+      "loss": 0.0337,
+      "step": 82500
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.4160062506976225e-06,
+      "loss": 0.0343,
+      "step": 82510
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.4126576626855674e-06,
+      "loss": 0.0869,
+      "step": 82520
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.409309074673513e-06,
+      "loss": 0.0278,
+      "step": 82530
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.4059604866614577e-06,
+      "loss": 0.0893,
+      "step": 82540
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.402611898649403e-06,
+      "loss": 0.0185,
+      "step": 82550
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.399263310637348e-06,
+      "loss": 0.0086,
+      "step": 82560
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.395914722625293e-06,
+      "loss": 0.0547,
+      "step": 82570
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.392566134613238e-06,
+      "loss": 0.0169,
+      "step": 82580
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.3892175466011836e-06,
+      "loss": 0.0848,
+      "step": 82590
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.3858689585891285e-06,
+      "loss": 0.0665,
+      "step": 82600
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.3825203705770734e-06,
+      "loss": 0.0435,
+      "step": 82610
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.3791717825650184e-06,
+      "loss": 0.0283,
+      "step": 82620
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.3758231945529637e-06,
+      "loss": 0.0495,
+      "step": 82630
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.3724746065409086e-06,
+      "loss": 0.0447,
+      "step": 82640
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.3691260185288535e-06,
+      "loss": 0.0157,
+      "step": 82650
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.365777430516799e-06,
+      "loss": 0.0211,
+      "step": 82660
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.362428842504744e-06,
+      "loss": 0.0493,
+      "step": 82670
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.359080254492689e-06,
+      "loss": 0.0489,
+      "step": 82680
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.355731666480634e-06,
+      "loss": 0.0223,
+      "step": 82690
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.352383078468579e-06,
+      "loss": 0.0592,
+      "step": 82700
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.349034490456524e-06,
+      "loss": 0.043,
+      "step": 82710
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.3456859024444697e-06,
+      "loss": 0.0436,
+      "step": 82720
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.3423373144324146e-06,
+      "loss": 0.0597,
+      "step": 82730
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.3389887264203595e-06,
+      "loss": 0.032,
+      "step": 82740
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.3356401384083044e-06,
+      "loss": 0.0502,
+      "step": 82750
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.3322915503962494e-06,
+      "loss": 0.0556,
+      "step": 82760
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.3289429623841947e-06,
+      "loss": 0.0164,
+      "step": 82770
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.32559437437214e-06,
+      "loss": 0.0312,
+      "step": 82780
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.322245786360085e-06,
+      "loss": 0.0225,
+      "step": 82790
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.31889719834803e-06,
+      "loss": 0.0293,
+      "step": 82800
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.3155486103359752e-06,
+      "loss": 0.0154,
+      "step": 82810
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.31220002232392e-06,
+      "loss": 0.0426,
+      "step": 82820
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.308851434311865e-06,
+      "loss": 0.0277,
+      "step": 82830
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.30550284629981e-06,
+      "loss": 0.0885,
+      "step": 82840
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.3021542582877553e-06,
+      "loss": 0.0233,
+      "step": 82850
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.2988056702757007e-06,
+      "loss": 0.0234,
+      "step": 82860
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.2954570822636456e-06,
+      "loss": 0.0374,
+      "step": 82870
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.2921084942515905e-06,
+      "loss": 0.0426,
+      "step": 82880
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.2887599062395355e-06,
+      "loss": 0.033,
+      "step": 82890
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.285411318227481e-06,
+      "loss": 0.1115,
+      "step": 82900
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.282062730215426e-06,
+      "loss": 0.0164,
+      "step": 82910
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.278714142203371e-06,
+      "loss": 0.0595,
+      "step": 82920
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.275365554191316e-06,
+      "loss": 0.0517,
+      "step": 82930
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.2720169661792613e-06,
+      "loss": 0.0522,
+      "step": 82940
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.2686683781672063e-06,
+      "loss": 0.0165,
+      "step": 82950
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.265319790155151e-06,
+      "loss": 0.0238,
+      "step": 82960
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.261971202143096e-06,
+      "loss": 0.0747,
+      "step": 82970
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.2586226141310414e-06,
+      "loss": 0.0297,
+      "step": 82980
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.2552740261189868e-06,
+      "loss": 0.0232,
+      "step": 82990
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.2519254381069317e-06,
+      "loss": 0.049,
+      "step": 83000
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.2485768500948766e-06,
+      "loss": 0.0303,
+      "step": 83010
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.2452282620828215e-06,
+      "loss": 0.0671,
+      "step": 83020
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.241879674070767e-06,
+      "loss": 0.0355,
+      "step": 83030
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.2385310860587122e-06,
+      "loss": 0.035,
+      "step": 83040
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.235182498046657e-06,
+      "loss": 0.0867,
+      "step": 83050
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.231833910034602e-06,
+      "loss": 0.0233,
+      "step": 83060
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.228485322022547e-06,
+      "loss": 0.0977,
+      "step": 83070
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.2251367340104923e-06,
+      "loss": 0.0402,
+      "step": 83080
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.2217881459984373e-06,
+      "loss": 0.0239,
+      "step": 83090
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.218439557986382e-06,
+      "loss": 0.0633,
+      "step": 83100
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.2150909699743275e-06,
+      "loss": 0.0145,
+      "step": 83110
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.211742381962273e-06,
+      "loss": 0.0438,
+      "step": 83120
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.208393793950218e-06,
+      "loss": 0.0444,
+      "step": 83130
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.2050452059381627e-06,
+      "loss": 0.0448,
+      "step": 83140
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.2016966179261076e-06,
+      "loss": 0.0301,
+      "step": 83150
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.198348029914053e-06,
+      "loss": 0.0516,
+      "step": 83160
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.1949994419019983e-06,
+      "loss": 0.0386,
+      "step": 83170
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.1916508538899433e-06,
+      "loss": 0.0543,
+      "step": 83180
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.188302265877888e-06,
+      "loss": 0.0722,
+      "step": 83190
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.184953677865833e-06,
+      "loss": 0.0343,
+      "step": 83200
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.1816050898537784e-06,
+      "loss": 0.2629,
+      "step": 83210
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.1782565018417234e-06,
+      "loss": 0.0349,
+      "step": 83220
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.1749079138296687e-06,
+      "loss": 0.0217,
+      "step": 83230
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.1715593258176136e-06,
+      "loss": 0.0204,
+      "step": 83240
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.168210737805559e-06,
+      "loss": 0.0255,
+      "step": 83250
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.164862149793504e-06,
+      "loss": 0.0609,
+      "step": 83260
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.161513561781449e-06,
+      "loss": 0.0707,
+      "step": 83270
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.1584998325705995e-06,
+      "loss": 0.0694,
+      "step": 83280
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.155151244558545e-06,
+      "loss": 0.0315,
+      "step": 83290
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.1518026565464898e-06,
+      "loss": 0.0425,
+      "step": 83300
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.1484540685344347e-06,
+      "loss": 0.0387,
+      "step": 83310
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.1451054805223796e-06,
+      "loss": 0.0252,
+      "step": 83320
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.141756892510325e-06,
+      "loss": 0.0211,
+      "step": 83330
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.13840830449827e-06,
+      "loss": 0.0986,
+      "step": 83340
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.135059716486215e-06,
+      "loss": 0.0297,
+      "step": 83350
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.13171112847416e-06,
+      "loss": 0.0557,
+      "step": 83360
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.1283625404621055e-06,
+      "loss": 0.0353,
+      "step": 83370
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.1250139524500504e-06,
+      "loss": 0.0522,
+      "step": 83380
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.1216653644379953e-06,
+      "loss": 0.0075,
+      "step": 83390
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.1183167764259402e-06,
+      "loss": 0.072,
+      "step": 83400
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.114968188413885e-06,
+      "loss": 0.0136,
+      "step": 83410
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.111619600401831e-06,
+      "loss": 0.0312,
+      "step": 83420
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.108271012389776e-06,
+      "loss": 0.0367,
+      "step": 83430
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.1049224243777208e-06,
+      "loss": 0.0489,
+      "step": 83440
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.1015738363656657e-06,
+      "loss": 0.1087,
+      "step": 83450
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.098225248353611e-06,
+      "loss": 0.0366,
+      "step": 83460
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.094876660341556e-06,
+      "loss": 0.0429,
+      "step": 83470
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.091528072329501e-06,
+      "loss": 0.0335,
+      "step": 83480
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.0881794843174462e-06,
+      "loss": 0.0383,
+      "step": 83490
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.0848308963053916e-06,
+      "loss": 0.0534,
+      "step": 83500
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.0814823082933365e-06,
+      "loss": 0.0586,
+      "step": 83510
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.0781337202812814e-06,
+      "loss": 0.0266,
+      "step": 83520
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.0747851322692263e-06,
+      "loss": 0.0874,
+      "step": 83530
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.0714365442571713e-06,
+      "loss": 0.0238,
+      "step": 83540
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.068087956245117e-06,
+      "loss": 0.0496,
+      "step": 83550
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.064739368233062e-06,
+      "loss": 0.0384,
+      "step": 83560
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.061390780221007e-06,
+      "loss": 0.0239,
+      "step": 83570
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.0580421922089518e-06,
+      "loss": 0.0347,
+      "step": 83580
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.054693604196897e-06,
+      "loss": 0.024,
+      "step": 83590
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.051345016184842e-06,
+      "loss": 0.0476,
+      "step": 83600
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.0479964281727874e-06,
+      "loss": 0.0441,
+      "step": 83610
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.0446478401607323e-06,
+      "loss": 0.0129,
+      "step": 83620
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.0412992521486772e-06,
+      "loss": 0.0759,
+      "step": 83630
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.0379506641366226e-06,
+      "loss": 0.0154,
+      "step": 83640
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.0346020761245675e-06,
+      "loss": 0.0409,
+      "step": 83650
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.0312534881125124e-06,
+      "loss": 0.0192,
+      "step": 83660
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.0279049001004573e-06,
+      "loss": 0.0157,
+      "step": 83670
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.024556312088403e-06,
+      "loss": 0.0682,
+      "step": 83680
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.021207724076348e-06,
+      "loss": 0.0636,
+      "step": 83690
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.017859136064293e-06,
+      "loss": 0.0302,
+      "step": 83700
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.014510548052238e-06,
+      "loss": 0.0224,
+      "step": 83710
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.0111619600401832e-06,
+      "loss": 0.0689,
+      "step": 83720
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.007813372028128e-06,
+      "loss": 0.0597,
+      "step": 83730
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.0044647840160735e-06,
+      "loss": 0.0454,
+      "step": 83740
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.0011161960040184e-06,
+      "loss": 0.0311,
+      "step": 83750
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.9977676079919633e-06,
+      "loss": 0.0199,
+      "step": 83760
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.9944190199799087e-06,
+      "loss": 0.1108,
+      "step": 83770
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.9910704319678536e-06,
+      "loss": 0.0738,
+      "step": 83780
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.9877218439557985e-06,
+      "loss": 0.0066,
+      "step": 83790
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.9843732559437434e-06,
+      "loss": 0.0477,
+      "step": 83800
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.981024667931689e-06,
+      "loss": 0.0355,
+      "step": 83810
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.977676079919634e-06,
+      "loss": 0.0336,
+      "step": 83820
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.974327491907579e-06,
+      "loss": 0.0714,
+      "step": 83830
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.970978903895524e-06,
+      "loss": 0.0348,
+      "step": 83840
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.967630315883469e-06,
+      "loss": 0.0071,
+      "step": 83850
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.9642817278714142e-06,
+      "loss": 0.0368,
+      "step": 83860
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.9609331398593596e-06,
+      "loss": 0.0395,
+      "step": 83870
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.9575845518473045e-06,
+      "loss": 0.0255,
+      "step": 83880
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.9542359638352494e-06,
+      "loss": 0.0306,
+      "step": 83890
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.9508873758231948e-06,
+      "loss": 0.0253,
+      "step": 83900
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.9475387878111397e-06,
+      "loss": 0.0171,
+      "step": 83910
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.9441901997990846e-06,
+      "loss": 0.0257,
+      "step": 83920
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.9408416117870295e-06,
+      "loss": 0.04,
+      "step": 83930
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.937493023774975e-06,
+      "loss": 0.0115,
+      "step": 83940
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.9341444357629202e-06,
+      "loss": 0.0193,
+      "step": 83950
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.930795847750865e-06,
+      "loss": 0.1324,
+      "step": 83960
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.92744725973881e-06,
+      "loss": 0.0534,
+      "step": 83970
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.924098671726755e-06,
+      "loss": 0.1446,
+      "step": 83980
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.9207500837147003e-06,
+      "loss": 0.0363,
+      "step": 83990
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.9174014957026457e-06,
+      "loss": 0.0439,
+      "step": 84000
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.9140529076905906e-06,
+      "loss": 0.0126,
+      "step": 84010
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.9107043196785355e-06,
+      "loss": 0.038,
+      "step": 84020
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.907355731666481e-06,
+      "loss": 0.092,
+      "step": 84030
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.9040071436544258e-06,
+      "loss": 0.0273,
+      "step": 84040
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.9006585556423707e-06,
+      "loss": 0.0151,
+      "step": 84050
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.897309967630316e-06,
+      "loss": 0.0634,
+      "step": 84060
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.8939613796182612e-06,
+      "loss": 0.0586,
+      "step": 84070
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.890612791606206e-06,
+      "loss": 0.0216,
+      "step": 84080
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.8872642035941512e-06,
+      "loss": 0.0145,
+      "step": 84090
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.8839156155820962e-06,
+      "loss": 0.0144,
+      "step": 84100
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.8805670275700413e-06,
+      "loss": 0.0592,
+      "step": 84110
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.8772184395579862e-06,
+      "loss": 0.0597,
+      "step": 84120
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.8738698515459316e-06,
+      "loss": 0.0071,
+      "step": 84130
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.8705212635338765e-06,
+      "loss": 0.1319,
+      "step": 84140
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.8671726755218218e-06,
+      "loss": 0.0247,
+      "step": 84150
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.8638240875097667e-06,
+      "loss": 0.0195,
+      "step": 84160
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.8604754994977117e-06,
+      "loss": 0.0513,
+      "step": 84170
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.857126911485657e-06,
+      "loss": 0.0132,
+      "step": 84180
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.853778323473602e-06,
+      "loss": 0.0111,
+      "step": 84190
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.850429735461547e-06,
+      "loss": 0.1572,
+      "step": 84200
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.8470811474494922e-06,
+      "loss": 0.046,
+      "step": 84210
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.8437325594374373e-06,
+      "loss": 0.1151,
+      "step": 84220
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.8403839714253822e-06,
+      "loss": 0.0288,
+      "step": 84230
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.8370353834133276e-06,
+      "loss": 0.0286,
+      "step": 84240
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.8336867954012725e-06,
+      "loss": 0.0414,
+      "step": 84250
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.8303382073892176e-06,
+      "loss": 0.0703,
+      "step": 84260
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.8269896193771628e-06,
+      "loss": 0.0849,
+      "step": 84270
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.8236410313651077e-06,
+      "loss": 0.0479,
+      "step": 84280
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.8202924433530528e-06,
+      "loss": 0.016,
+      "step": 84290
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.8169438553409978e-06,
+      "loss": 0.0274,
+      "step": 84300
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.813595267328943e-06,
+      "loss": 0.0317,
+      "step": 84310
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.810246679316888e-06,
+      "loss": 0.0541,
+      "step": 84320
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.8068980913048332e-06,
+      "loss": 0.0255,
+      "step": 84330
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.8035495032927783e-06,
+      "loss": 0.0316,
+      "step": 84340
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.8002009152807234e-06,
+      "loss": 0.0241,
+      "step": 84350
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.7968523272686683e-06,
+      "loss": 0.0175,
+      "step": 84360
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.7935037392566137e-06,
+      "loss": 0.0623,
+      "step": 84370
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.7901551512445586e-06,
+      "loss": 0.0214,
+      "step": 84380
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.7868065632325035e-06,
+      "loss": 0.0354,
+      "step": 84390
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.7834579752204489e-06,
+      "loss": 0.0571,
+      "step": 84400
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.7801093872083938e-06,
+      "loss": 0.0166,
+      "step": 84410
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.776760799196339e-06,
+      "loss": 0.0379,
+      "step": 84420
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.773412211184284e-06,
+      "loss": 0.0477,
+      "step": 84430
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.7700636231722292e-06,
+      "loss": 0.0425,
+      "step": 84440
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.7667150351601741e-06,
+      "loss": 0.0333,
+      "step": 84450
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.7633664471481192e-06,
+      "loss": 0.0207,
+      "step": 84460
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.7600178591360644e-06,
+      "loss": 0.0535,
+      "step": 84470
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.7566692711240095e-06,
+      "loss": 0.061,
+      "step": 84480
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.7533206831119544e-06,
+      "loss": 0.044,
+      "step": 84490
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.7499720950998996e-06,
+      "loss": 0.0408,
+      "step": 84500
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.7466235070878447e-06,
+      "loss": 0.0221,
+      "step": 84510
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.7432749190757896e-06,
+      "loss": 0.0482,
+      "step": 84520
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.739926331063735e-06,
+      "loss": 0.0409,
+      "step": 84530
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.7365777430516799e-06,
+      "loss": 0.0224,
+      "step": 84540
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.733229155039625e-06,
+      "loss": 0.0188,
+      "step": 84550
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.7298805670275701e-06,
+      "loss": 0.0796,
+      "step": 84560
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.7265319790155153e-06,
+      "loss": 0.0382,
+      "step": 84570
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.7231833910034602e-06,
+      "loss": 0.0792,
+      "step": 84580
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.7198348029914053e-06,
+      "loss": 0.0211,
+      "step": 84590
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.7164862149793505e-06,
+      "loss": 0.0359,
+      "step": 84600
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.7131376269672954e-06,
+      "loss": 0.0433,
+      "step": 84610
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.7097890389552405e-06,
+      "loss": 0.0387,
+      "step": 84620
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.7064404509431857e-06,
+      "loss": 0.0991,
+      "step": 84630
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.7030918629311308e-06,
+      "loss": 0.0241,
+      "step": 84640
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.6997432749190757e-06,
+      "loss": 0.0425,
+      "step": 84650
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.696394686907021e-06,
+      "loss": 0.0257,
+      "step": 84660
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.693046098894966e-06,
+      "loss": 0.0355,
+      "step": 84670
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.6896975108829111e-06,
+      "loss": 0.0211,
+      "step": 84680
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.6863489228708562e-06,
+      "loss": 0.0211,
+      "step": 84690
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.6830003348588012e-06,
+      "loss": 0.0284,
+      "step": 84700
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.6796517468467463e-06,
+      "loss": 0.0381,
+      "step": 84710
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.6763031588346914e-06,
+      "loss": 0.0745,
+      "step": 84720
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.6729545708226366e-06,
+      "loss": 0.0289,
+      "step": 84730
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.6696059828105815e-06,
+      "loss": 0.0652,
+      "step": 84740
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.6662573947985266e-06,
+      "loss": 0.0284,
+      "step": 84750
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.6629088067864717e-06,
+      "loss": 0.0297,
+      "step": 84760
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.6595602187744169e-06,
+      "loss": 0.0596,
+      "step": 84770
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.6562116307623618e-06,
+      "loss": 0.0356,
+      "step": 84780
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.6528630427503071e-06,
+      "loss": 0.0305,
+      "step": 84790
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.649514454738252e-06,
+      "loss": 0.0254,
+      "step": 84800
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.646165866726197e-06,
+      "loss": 0.0321,
+      "step": 84810
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.6428172787141423e-06,
+      "loss": 0.0669,
+      "step": 84820
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.6394686907020873e-06,
+      "loss": 0.0275,
+      "step": 84830
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.6361201026900324e-06,
+      "loss": 0.0623,
+      "step": 84840
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.6327715146779775e-06,
+      "loss": 0.0399,
+      "step": 84850
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.6294229266659227e-06,
+      "loss": 0.0313,
+      "step": 84860
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.6260743386538676e-06,
+      "loss": 0.1166,
+      "step": 84870
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.622725750641813e-06,
+      "loss": 0.0562,
+      "step": 84880
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.6193771626297578e-06,
+      "loss": 0.0222,
+      "step": 84890
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.616028574617703e-06,
+      "loss": 0.0645,
+      "step": 84900
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.612679986605648e-06,
+      "loss": 0.0891,
+      "step": 84910
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.609331398593593e-06,
+      "loss": 0.0363,
+      "step": 84920
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.6059828105815382e-06,
+      "loss": 0.0457,
+      "step": 84930
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.602634222569483e-06,
+      "loss": 0.0237,
+      "step": 84940
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.5992856345574284e-06,
+      "loss": 0.0544,
+      "step": 84950
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.5959370465453733e-06,
+      "loss": 0.0243,
+      "step": 84960
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.5925884585333185e-06,
+      "loss": 0.048,
+      "step": 84970
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.5892398705212636e-06,
+      "loss": 0.0352,
+      "step": 84980
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.5858912825092087e-06,
+      "loss": 0.0183,
+      "step": 84990
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.5825426944971537e-06,
+      "loss": 0.0639,
+      "step": 85000
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.579194106485099e-06,
+      "loss": 0.0869,
+      "step": 85010
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.575845518473044e-06,
+      "loss": 0.0495,
+      "step": 85020
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.5724969304609889e-06,
+      "loss": 0.1112,
+      "step": 85030
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.569148342448934e-06,
+      "loss": 0.0199,
+      "step": 85040
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.5657997544368791e-06,
+      "loss": 0.0394,
+      "step": 85050
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.5624511664248243e-06,
+      "loss": 0.0404,
+      "step": 85060
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.5591025784127692e-06,
+      "loss": 0.0346,
+      "step": 85070
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.5557539904007145e-06,
+      "loss": 0.0699,
+      "step": 85080
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.5524054023886594e-06,
+      "loss": 0.0195,
+      "step": 85090
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.5490568143766046e-06,
+      "loss": 0.0321,
+      "step": 85100
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.5457082263645497e-06,
+      "loss": 0.0545,
+      "step": 85110
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.5423596383524948e-06,
+      "loss": 0.0702,
+      "step": 85120
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.5390110503404398e-06,
+      "loss": 0.0473,
+      "step": 85130
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.5356624623283849e-06,
+      "loss": 0.0753,
+      "step": 85140
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.53231387431633e-06,
+      "loss": 0.0387,
+      "step": 85150
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.528965286304275e-06,
+      "loss": 0.1087,
+      "step": 85160
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.5256166982922203e-06,
+      "loss": 0.0456,
+      "step": 85170
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.5222681102801652e-06,
+      "loss": 0.0056,
+      "step": 85180
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.5189195222681103e-06,
+      "loss": 0.0186,
+      "step": 85190
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.5155709342560553e-06,
+      "loss": 0.0101,
+      "step": 85200
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.5122223462440006e-06,
+      "loss": 0.0625,
+      "step": 85210
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.5088737582319455e-06,
+      "loss": 0.0272,
+      "step": 85220
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.5055251702198907e-06,
+      "loss": 0.0713,
+      "step": 85230
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.5021765822078358e-06,
+      "loss": 0.0337,
+      "step": 85240
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.4988279941957807e-06,
+      "loss": 0.0566,
+      "step": 85250
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.4954794061837259e-06,
+      "loss": 0.036,
+      "step": 85260
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.492130818171671e-06,
+      "loss": 0.0252,
+      "step": 85270
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.4887822301596161e-06,
+      "loss": 0.0294,
+      "step": 85280
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.485433642147561e-06,
+      "loss": 0.0491,
+      "step": 85290
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.4820850541355064e-06,
+      "loss": 0.0448,
+      "step": 85300
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.4787364661234513e-06,
+      "loss": 0.0191,
+      "step": 85310
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.4753878781113964e-06,
+      "loss": 0.0365,
+      "step": 85320
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.4720392900993414e-06,
+      "loss": 0.0619,
+      "step": 85330
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.4686907020872867e-06,
+      "loss": 0.063,
+      "step": 85340
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.4653421140752316e-06,
+      "loss": 0.0554,
+      "step": 85350
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.4619935260631765e-06,
+      "loss": 0.0315,
+      "step": 85360
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.4586449380511219e-06,
+      "loss": 0.0569,
+      "step": 85370
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.4552963500390668e-06,
+      "loss": 0.0198,
+      "step": 85380
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.451947762027012e-06,
+      "loss": 0.3457,
+      "step": 85390
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.448599174014957e-06,
+      "loss": 0.028,
+      "step": 85400
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.4452505860029022e-06,
+      "loss": 0.0208,
+      "step": 85410
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.4419019979908471e-06,
+      "loss": 0.02,
+      "step": 85420
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.4385534099787925e-06,
+      "loss": 0.0772,
+      "step": 85430
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.4352048219667374e-06,
+      "loss": 0.0357,
+      "step": 85440
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.4318562339546825e-06,
+      "loss": 0.0361,
+      "step": 85450
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.4285076459426277e-06,
+      "loss": 0.0426,
+      "step": 85460
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.4251590579305726e-06,
+      "loss": 0.0585,
+      "step": 85470
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.4218104699185177e-06,
+      "loss": 0.0193,
+      "step": 85480
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.4184618819064626e-06,
+      "loss": 0.0052,
+      "step": 85490
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.415113293894408e-06,
+      "loss": 0.0429,
+      "step": 85500
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.411764705882353e-06,
+      "loss": 0.0932,
+      "step": 85510
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.408416117870298e-06,
+      "loss": 0.0343,
+      "step": 85520
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.4050675298582432e-06,
+      "loss": 0.0389,
+      "step": 85530
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.4017189418461883e-06,
+      "loss": 0.0145,
+      "step": 85540
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.3983703538341332e-06,
+      "loss": 0.094,
+      "step": 85550
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.3950217658220786e-06,
+      "loss": 0.0581,
+      "step": 85560
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.3916731778100235e-06,
+      "loss": 0.033,
+      "step": 85570
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.3883245897979684e-06,
+      "loss": 0.082,
+      "step": 85580
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.3849760017859138e-06,
+      "loss": 0.0342,
+      "step": 85590
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.3816274137738587e-06,
+      "loss": 0.0409,
+      "step": 85600
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.3782788257618038e-06,
+      "loss": 0.0483,
+      "step": 85610
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.374930237749749e-06,
+      "loss": 0.2147,
+      "step": 85620
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.371581649737694e-06,
+      "loss": 0.0391,
+      "step": 85630
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.368233061725639e-06,
+      "loss": 0.0912,
+      "step": 85640
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.3648844737135841e-06,
+      "loss": 0.0202,
+      "step": 85650
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.3615358857015293e-06,
+      "loss": 0.0079,
+      "step": 85660
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.3581872976894744e-06,
+      "loss": 0.1097,
+      "step": 85670
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.3548387096774193e-06,
+      "loss": 0.0485,
+      "step": 85680
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.3514901216653644e-06,
+      "loss": 0.0427,
+      "step": 85690
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.3481415336533096e-06,
+      "loss": 0.0272,
+      "step": 85700
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.3447929456412545e-06,
+      "loss": 0.0084,
+      "step": 85710
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.3414443576291998e-06,
+      "loss": 0.0675,
+      "step": 85720
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.3380957696171448e-06,
+      "loss": 0.0079,
+      "step": 85730
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.33474718160509e-06,
+      "loss": 0.0213,
+      "step": 85740
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.331398593593035e-06,
+      "loss": 0.0469,
+      "step": 85750
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.3280500055809802e-06,
+      "loss": 0.029,
+      "step": 85760
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.324701417568925e-06,
+      "loss": 0.1677,
+      "step": 85770
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.3213528295568702e-06,
+      "loss": 0.0195,
+      "step": 85780
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.3180042415448154e-06,
+      "loss": 0.0107,
+      "step": 85790
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.3146556535327603e-06,
+      "loss": 0.0466,
+      "step": 85800
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.3113070655207054e-06,
+      "loss": 0.0328,
+      "step": 85810
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.3079584775086505e-06,
+      "loss": 0.0648,
+      "step": 85820
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.3046098894965957e-06,
+      "loss": 0.0526,
+      "step": 85830
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.3012613014845406e-06,
+      "loss": 0.0091,
+      "step": 85840
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.297912713472486e-06,
+      "loss": 0.0077,
+      "step": 85850
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.2945641254604309e-06,
+      "loss": 0.0229,
+      "step": 85860
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.291215537448376e-06,
+      "loss": 0.0476,
+      "step": 85870
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.2878669494363211e-06,
+      "loss": 0.0473,
+      "step": 85880
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.2845183614242663e-06,
+      "loss": 0.0315,
+      "step": 85890
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.2811697734122112e-06,
+      "loss": 0.031,
+      "step": 85900
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.2778211854001563e-06,
+      "loss": 0.0383,
+      "step": 85910
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.2744725973881014e-06,
+      "loss": 0.054,
+      "step": 85920
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.2711240093760464e-06,
+      "loss": 0.0285,
+      "step": 85930
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.2677754213639915e-06,
+      "loss": 0.0642,
+      "step": 85940
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.2644268333519366e-06,
+      "loss": 0.0123,
+      "step": 85950
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.2610782453398818e-06,
+      "loss": 0.014,
+      "step": 85960
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.2577296573278267e-06,
+      "loss": 0.0292,
+      "step": 85970
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.254381069315772e-06,
+      "loss": 0.0536,
+      "step": 85980
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.251032481303717e-06,
+      "loss": 0.0262,
+      "step": 85990
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.247683893291662e-06,
+      "loss": 0.0468,
+      "step": 86000
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.2443353052796072e-06,
+      "loss": 0.0341,
+      "step": 86010
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.2409867172675521e-06,
+      "loss": 0.0461,
+      "step": 86020
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.2376381292554973e-06,
+      "loss": 0.0771,
+      "step": 86030
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.2342895412434424e-06,
+      "loss": 0.0674,
+      "step": 86040
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.2309409532313875e-06,
+      "loss": 0.0289,
+      "step": 86050
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.2275923652193325e-06,
+      "loss": 0.0447,
+      "step": 86060
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.2242437772072778e-06,
+      "loss": 0.1099,
+      "step": 86070
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.2208951891952227e-06,
+      "loss": 0.1883,
+      "step": 86080
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.2175466011831679e-06,
+      "loss": 0.0326,
+      "step": 86090
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.2141980131711128e-06,
+      "loss": 0.0504,
+      "step": 86100
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.210849425159058e-06,
+      "loss": 0.0174,
+      "step": 86110
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.207500837147003e-06,
+      "loss": 0.0259,
+      "step": 86120
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.204152249134948e-06,
+      "loss": 0.0153,
+      "step": 86130
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.2008036611228933e-06,
+      "loss": 0.0563,
+      "step": 86140
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.1974550731108382e-06,
+      "loss": 0.0153,
+      "step": 86150
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.1941064850987834e-06,
+      "loss": 0.0987,
+      "step": 86160
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.1907578970867285e-06,
+      "loss": 0.0567,
+      "step": 86170
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.1874093090746736e-06,
+      "loss": 0.0123,
+      "step": 86180
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.1840607210626185e-06,
+      "loss": 0.0091,
+      "step": 86190
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.1807121330505639e-06,
+      "loss": 0.0184,
+      "step": 86200
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.1773635450385088e-06,
+      "loss": 0.025,
+      "step": 86210
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.1740149570264537e-06,
+      "loss": 0.1012,
+      "step": 86220
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.1706663690143989e-06,
+      "loss": 0.0421,
+      "step": 86230
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.167317781002344e-06,
+      "loss": 0.0806,
+      "step": 86240
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.1639691929902891e-06,
+      "loss": 0.028,
+      "step": 86250
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.160620604978234e-06,
+      "loss": 0.0196,
+      "step": 86260
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.1572720169661794e-06,
+      "loss": 0.0469,
+      "step": 86270
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.1539234289541243e-06,
+      "loss": 0.0456,
+      "step": 86280
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.1505748409420695e-06,
+      "loss": 0.0197,
+      "step": 86290
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.1472262529300146e-06,
+      "loss": 0.0436,
+      "step": 86300
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.1438776649179597e-06,
+      "loss": 0.0389,
+      "step": 86310
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.1405290769059046e-06,
+      "loss": 0.0285,
+      "step": 86320
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.1371804888938498e-06,
+      "loss": 0.0206,
+      "step": 86330
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.133831900881795e-06,
+      "loss": 0.018,
+      "step": 86340
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.1304833128697398e-06,
+      "loss": 0.021,
+      "step": 86350
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.1271347248576852e-06,
+      "loss": 0.0489,
+      "step": 86360
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.12378613684563e-06,
+      "loss": 0.0144,
+      "step": 86370
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.1204375488335752e-06,
+      "loss": 0.0742,
+      "step": 86380
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.1170889608215201e-06,
+      "loss": 0.0324,
+      "step": 86390
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.1137403728094655e-06,
+      "loss": 0.0302,
+      "step": 86400
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.1103917847974104e-06,
+      "loss": 0.0622,
+      "step": 86410
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.1070431967853555e-06,
+      "loss": 0.0466,
+      "step": 86420
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.1036946087733007e-06,
+      "loss": 0.0355,
+      "step": 86430
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.1003460207612456e-06,
+      "loss": 0.0807,
+      "step": 86440
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.0969974327491907e-06,
+      "loss": 0.0138,
+      "step": 86450
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.0936488447371359e-06,
+      "loss": 0.0635,
+      "step": 86460
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.090300256725081e-06,
+      "loss": 0.0758,
+      "step": 86470
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.086951668713026e-06,
+      "loss": 0.0242,
+      "step": 86480
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.0836030807009713e-06,
+      "loss": 0.0546,
+      "step": 86490
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.0802544926889162e-06,
+      "loss": 0.054,
+      "step": 86500
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.0769059046768613e-06,
+      "loss": 0.0367,
+      "step": 86510
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.0735573166648065e-06,
+      "loss": 0.0439,
+      "step": 86520
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.0702087286527516e-06,
+      "loss": 0.0239,
+      "step": 86530
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.0668601406406965e-06,
+      "loss": 0.0553,
+      "step": 86540
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.0635115526286414e-06,
+      "loss": 0.0283,
+      "step": 86550
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.0601629646165868e-06,
+      "loss": 0.0465,
+      "step": 86560
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.0568143766045317e-06,
+      "loss": 0.0677,
+      "step": 86570
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.0534657885924768e-06,
+      "loss": 0.0384,
+      "step": 86580
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.050117200580422e-06,
+      "loss": 0.0486,
+      "step": 86590
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.046768612568367e-06,
+      "loss": 0.0209,
+      "step": 86600
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.043420024556312e-06,
+      "loss": 0.0451,
+      "step": 86610
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.0400714365442574e-06,
+      "loss": 0.062,
+      "step": 86620
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.0367228485322023e-06,
+      "loss": 0.0214,
+      "step": 86630
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.0333742605201474e-06,
+      "loss": 0.0299,
+      "step": 86640
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.0300256725080925e-06,
+      "loss": 0.032,
+      "step": 86650
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.0266770844960375e-06,
+      "loss": 0.1084,
+      "step": 86660
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.0233284964839826e-06,
+      "loss": 0.0227,
+      "step": 86670
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.0199799084719275e-06,
+      "loss": 0.0364,
+      "step": 86680
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.0166313204598729e-06,
+      "loss": 0.071,
+      "step": 86690
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.0132827324478178e-06,
+      "loss": 0.0111,
+      "step": 86700
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.009934144435763e-06,
+      "loss": 0.0313,
+      "step": 86710
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.006585556423708e-06,
+      "loss": 0.072,
+      "step": 86720
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.0032369684116532e-06,
+      "loss": 0.0201,
+      "step": 86730
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 9.99888380399598e-07,
+      "loss": 0.0385,
+      "step": 86740
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 9.965397923875434e-07,
+      "loss": 0.0325,
+      "step": 86750
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 9.931912043754884e-07,
+      "loss": 0.0295,
+      "step": 86760
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 9.898426163634333e-07,
+      "loss": 0.0297,
+      "step": 86770
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 9.864940283513786e-07,
+      "loss": 0.0772,
+      "step": 86780
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 9.831454403393236e-07,
+      "loss": 0.0481,
+      "step": 86790
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 9.797968523272687e-07,
+      "loss": 0.0204,
+      "step": 86800
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 9.764482643152138e-07,
+      "loss": 0.0242,
+      "step": 86810
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 9.73099676303159e-07,
+      "loss": 0.0224,
+      "step": 86820
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 9.697510882911039e-07,
+      "loss": 0.0195,
+      "step": 86830
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 9.66402500279049e-07,
+      "loss": 0.0221,
+      "step": 86840
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 9.630539122669941e-07,
+      "loss": 0.025,
+      "step": 86850
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 9.597053242549393e-07,
+      "loss": 0.0888,
+      "step": 86860
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 9.563567362428842e-07,
+      "loss": 0.0286,
+      "step": 86870
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 9.530081482308294e-07,
+      "loss": 0.0299,
+      "step": 86880
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 9.496595602187745e-07,
+      "loss": 0.0415,
+      "step": 86890
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 9.463109722067195e-07,
+      "loss": 0.081,
+      "step": 86900
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 9.429623841946647e-07,
+      "loss": 0.017,
+      "step": 86910
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 9.396137961826098e-07,
+      "loss": 0.0382,
+      "step": 86920
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 9.362652081705548e-07,
+      "loss": 0.0963,
+      "step": 86930
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 9.329166201584998e-07,
+      "loss": 0.0364,
+      "step": 86940
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 9.295680321464449e-07,
+      "loss": 0.0403,
+      "step": 86950
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 9.2621944413439e-07,
+      "loss": 0.0247,
+      "step": 86960
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 9.228708561223351e-07,
+      "loss": 0.0618,
+      "step": 86970
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 9.195222681102802e-07,
+      "loss": 0.0347,
+      "step": 86980
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 9.161736800982253e-07,
+      "loss": 0.0228,
+      "step": 86990
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 9.128250920861704e-07,
+      "loss": 0.0504,
+      "step": 87000
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 9.094765040741154e-07,
+      "loss": 0.026,
+      "step": 87010
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 9.061279160620606e-07,
+      "loss": 0.0298,
+      "step": 87020
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 9.027793280500056e-07,
+      "loss": 0.0773,
+      "step": 87030
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 8.994307400379506e-07,
+      "loss": 0.0643,
+      "step": 87040
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 8.960821520258957e-07,
+      "loss": 0.0251,
+      "step": 87050
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 8.927335640138409e-07,
+      "loss": 0.0418,
+      "step": 87060
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 8.893849760017859e-07,
+      "loss": 0.0832,
+      "step": 87070
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 8.86036387989731e-07,
+      "loss": 0.0233,
+      "step": 87080
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 8.826877999776761e-07,
+      "loss": 0.0348,
+      "step": 87090
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 8.793392119656212e-07,
+      "loss": 0.0386,
+      "step": 87100
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 8.759906239535663e-07,
+      "loss": 0.0406,
+      "step": 87110
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 8.726420359415114e-07,
+      "loss": 0.0727,
+      "step": 87120
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 8.692934479294565e-07,
+      "loss": 0.0638,
+      "step": 87130
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 8.659448599174015e-07,
+      "loss": 0.0949,
+      "step": 87140
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 8.625962719053465e-07,
+      "loss": 0.0424,
+      "step": 87150
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 8.592476838932917e-07,
+      "loss": 0.0744,
+      "step": 87160
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 8.558990958812367e-07,
+      "loss": 0.0353,
+      "step": 87170
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 8.525505078691818e-07,
+      "loss": 0.0365,
+      "step": 87180
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 8.49201919857127e-07,
+      "loss": 0.0296,
+      "step": 87190
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 8.45853331845072e-07,
+      "loss": 0.044,
+      "step": 87200
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 8.425047438330171e-07,
+      "loss": 0.0252,
+      "step": 87210
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 8.391561558209623e-07,
+      "loss": 0.1282,
+      "step": 87220
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 8.358075678089073e-07,
+      "loss": 0.0554,
+      "step": 87230
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 8.324589797968524e-07,
+      "loss": 0.0683,
+      "step": 87240
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 8.291103917847973e-07,
+      "loss": 0.045,
+      "step": 87250
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 8.257618037727425e-07,
+      "loss": 0.052,
+      "step": 87260
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 8.224132157606876e-07,
+      "loss": 0.0501,
+      "step": 87270
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 8.190646277486326e-07,
+      "loss": 0.0298,
+      "step": 87280
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 8.157160397365778e-07,
+      "loss": 0.0492,
+      "step": 87290
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 8.123674517245229e-07,
+      "loss": 0.0841,
+      "step": 87300
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 8.090188637124679e-07,
+      "loss": 0.0329,
+      "step": 87310
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.056702757004131e-07,
+      "loss": 0.0659,
+      "step": 87320
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.023216876883581e-07,
+      "loss": 0.018,
+      "step": 87330
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 7.989730996763032e-07,
+      "loss": 0.0133,
+      "step": 87340
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 7.956245116642484e-07,
+      "loss": 0.1206,
+      "step": 87350
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 7.922759236521933e-07,
+      "loss": 0.0533,
+      "step": 87360
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 7.889273356401384e-07,
+      "loss": 0.0789,
+      "step": 87370
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 7.855787476280834e-07,
+      "loss": 0.0277,
+      "step": 87380
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 7.822301596160286e-07,
+      "loss": 0.0129,
+      "step": 87390
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 7.788815716039737e-07,
+      "loss": 0.076,
+      "step": 87400
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 7.755329835919187e-07,
+      "loss": 0.0391,
+      "step": 87410
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 7.721843955798639e-07,
+      "loss": 0.0603,
+      "step": 87420
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 7.68835807567809e-07,
+      "loss": 0.0499,
+      "step": 87430
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 7.65487219555754e-07,
+      "loss": 0.0292,
+      "step": 87440
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 7.621386315436992e-07,
+      "loss": 0.0704,
+      "step": 87450
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 7.587900435316442e-07,
+      "loss": 0.052,
+      "step": 87460
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 7.554414555195892e-07,
+      "loss": 0.0257,
+      "step": 87470
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 7.520928675075343e-07,
+      "loss": 0.0574,
+      "step": 87480
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 7.487442794954794e-07,
+      "loss": 0.0145,
+      "step": 87490
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 7.453956914834245e-07,
+      "loss": 0.0284,
+      "step": 87500
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 7.420471034713696e-07,
+      "loss": 0.0186,
+      "step": 87510
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 7.386985154593147e-07,
+      "loss": 0.0462,
+      "step": 87520
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 7.353499274472598e-07,
+      "loss": 0.0314,
+      "step": 87530
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 7.320013394352048e-07,
+      "loss": 0.0412,
+      "step": 87540
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 7.2865275142315e-07,
+      "loss": 0.0289,
+      "step": 87550
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 7.253041634110951e-07,
+      "loss": 0.0378,
+      "step": 87560
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 7.219555753990401e-07,
+      "loss": 0.069,
+      "step": 87570
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 7.186069873869851e-07,
+      "loss": 0.0256,
+      "step": 87580
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.152583993749303e-07,
+      "loss": 0.0491,
+      "step": 87590
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.119098113628753e-07,
+      "loss": 0.009,
+      "step": 87600
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.085612233508204e-07,
+      "loss": 0.0321,
+      "step": 87610
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.052126353387655e-07,
+      "loss": 0.0285,
+      "step": 87620
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.018640473267106e-07,
+      "loss": 0.1004,
+      "step": 87630
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 6.985154593146557e-07,
+      "loss": 0.023,
+      "step": 87640
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 6.951668713026007e-07,
+      "loss": 0.0515,
+      "step": 87650
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 6.918182832905459e-07,
+      "loss": 0.0481,
+      "step": 87660
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 6.88469695278491e-07,
+      "loss": 0.0409,
+      "step": 87670
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 6.851211072664359e-07,
+      "loss": 0.025,
+      "step": 87680
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 6.817725192543811e-07,
+      "loss": 0.0217,
+      "step": 87690
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 6.784239312423261e-07,
+      "loss": 0.0091,
+      "step": 87700
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 6.754102020314767e-07,
+      "loss": 0.0677,
+      "step": 87710
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 6.720616140194218e-07,
+      "loss": 0.0543,
+      "step": 87720
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 6.687130260073669e-07,
+      "loss": 0.0965,
+      "step": 87730
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 6.65364437995312e-07,
+      "loss": 0.03,
+      "step": 87740
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 6.620158499832571e-07,
+      "loss": 0.0142,
+      "step": 87750
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 6.586672619712021e-07,
+      "loss": 0.0414,
+      "step": 87760
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 6.553186739591473e-07,
+      "loss": 0.0183,
+      "step": 87770
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 6.519700859470924e-07,
+      "loss": 0.0388,
+      "step": 87780
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 6.486214979350374e-07,
+      "loss": 0.0289,
+      "step": 87790
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 6.452729099229824e-07,
+      "loss": 0.0382,
+      "step": 87800
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 6.419243219109276e-07,
+      "loss": 0.0293,
+      "step": 87810
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 6.385757338988726e-07,
+      "loss": 0.0427,
+      "step": 87820
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 6.352271458868177e-07,
+      "loss": 0.0271,
+      "step": 87830
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 6.318785578747628e-07,
+      "loss": 0.0219,
+      "step": 87840
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 6.285299698627079e-07,
+      "loss": 0.0494,
+      "step": 87850
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.25181381850653e-07,
+      "loss": 0.0414,
+      "step": 87860
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.218327938385981e-07,
+      "loss": 0.1333,
+      "step": 87870
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.184842058265432e-07,
+      "loss": 0.0277,
+      "step": 87880
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.151356178144883e-07,
+      "loss": 0.0211,
+      "step": 87890
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.117870298024334e-07,
+      "loss": 0.0767,
+      "step": 87900
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.084384417903784e-07,
+      "loss": 0.0362,
+      "step": 87910
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.050898537783234e-07,
+      "loss": 0.0173,
+      "step": 87920
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.017412657662685e-07,
+      "loss": 0.1024,
+      "step": 87930
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.983926777542137e-07,
+      "loss": 0.0216,
+      "step": 87940
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.950440897421587e-07,
+      "loss": 0.0291,
+      "step": 87950
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.916955017301038e-07,
+      "loss": 0.0356,
+      "step": 87960
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.88346913718049e-07,
+      "loss": 0.0182,
+      "step": 87970
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.84998325705994e-07,
+      "loss": 0.0303,
+      "step": 87980
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.816497376939391e-07,
+      "loss": 0.097,
+      "step": 87990
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.783011496818842e-07,
+      "loss": 0.0578,
+      "step": 88000
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.749525616698293e-07,
+      "loss": 0.0466,
+      "step": 88010
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.716039736577743e-07,
+      "loss": 0.0226,
+      "step": 88020
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.682553856457193e-07,
+      "loss": 0.0252,
+      "step": 88030
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.649067976336645e-07,
+      "loss": 0.0744,
+      "step": 88040
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.615582096216095e-07,
+      "loss": 0.0344,
+      "step": 88050
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.582096216095546e-07,
+      "loss": 0.0198,
+      "step": 88060
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.548610335974998e-07,
+      "loss": 0.0727,
+      "step": 88070
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.515124455854448e-07,
+      "loss": 0.0484,
+      "step": 88080
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.481638575733899e-07,
+      "loss": 0.0168,
+      "step": 88090
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.448152695613351e-07,
+      "loss": 0.0111,
+      "step": 88100
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.414666815492801e-07,
+      "loss": 0.0943,
+      "step": 88110
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.381180935372252e-07,
+      "loss": 0.0513,
+      "step": 88120
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.347695055251701e-07,
+      "loss": 0.0105,
+      "step": 88130
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.314209175131153e-07,
+      "loss": 0.047,
+      "step": 88140
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.280723295010604e-07,
+      "loss": 0.0626,
+      "step": 88150
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.247237414890054e-07,
+      "loss": 0.085,
+      "step": 88160
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.213751534769506e-07,
+      "loss": 0.1431,
+      "step": 88170
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.180265654648957e-07,
+      "loss": 0.0625,
+      "step": 88180
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.146779774528407e-07,
+      "loss": 0.0279,
+      "step": 88190
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.113293894407859e-07,
+      "loss": 0.1092,
+      "step": 88200
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.079808014287309e-07,
+      "loss": 0.0314,
+      "step": 88210
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.04632213416676e-07,
+      "loss": 0.0242,
+      "step": 88220
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.012836254046211e-07,
+      "loss": 0.0122,
+      "step": 88230
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 4.979350373925661e-07,
+      "loss": 0.0402,
+      "step": 88240
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 4.945864493805112e-07,
+      "loss": 0.0504,
+      "step": 88250
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 4.912378613684563e-07,
+      "loss": 0.044,
+      "step": 88260
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 4.878892733564014e-07,
+      "loss": 0.0368,
+      "step": 88270
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 4.845406853443465e-07,
+      "loss": 0.0221,
+      "step": 88280
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 4.811920973322915e-07,
+      "loss": 0.0635,
+      "step": 88290
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.778435093202367e-07,
+      "loss": 0.0153,
+      "step": 88300
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.744949213081818e-07,
+      "loss": 0.027,
+      "step": 88310
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.7114633329612676e-07,
+      "loss": 0.0378,
+      "step": 88320
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.677977452840719e-07,
+      "loss": 0.0164,
+      "step": 88330
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.64449157272017e-07,
+      "loss": 0.0119,
+      "step": 88340
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.6110056925996205e-07,
+      "loss": 0.0374,
+      "step": 88350
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.577519812479072e-07,
+      "loss": 0.0411,
+      "step": 88360
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.544033932358522e-07,
+      "loss": 0.0403,
+      "step": 88370
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.510548052237973e-07,
+      "loss": 0.0627,
+      "step": 88380
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.4770621721174237e-07,
+      "loss": 0.0304,
+      "step": 88390
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.443576291996875e-07,
+      "loss": 0.0205,
+      "step": 88400
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.410090411876326e-07,
+      "loss": 0.0695,
+      "step": 88410
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.3766045317557767e-07,
+      "loss": 0.0382,
+      "step": 88420
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.343118651635227e-07,
+      "loss": 0.0334,
+      "step": 88430
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.309632771514678e-07,
+      "loss": 0.043,
+      "step": 88440
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.276146891394129e-07,
+      "loss": 0.0639,
+      "step": 88450
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.24266101127358e-07,
+      "loss": 0.0282,
+      "step": 88460
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.2091751311530307e-07,
+      "loss": 0.1088,
+      "step": 88470
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.175689251032481e-07,
+      "loss": 0.059,
+      "step": 88480
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.142203370911932e-07,
+      "loss": 0.0247,
+      "step": 88490
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.108717490791383e-07,
+      "loss": 0.0499,
+      "step": 88500
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.075231610670834e-07,
+      "loss": 0.0547,
+      "step": 88510
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.0417457305502847e-07,
+      "loss": 0.0403,
+      "step": 88520
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.0082598504297355e-07,
+      "loss": 0.0138,
+      "step": 88530
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 3.974773970309186e-07,
+      "loss": 0.0635,
+      "step": 88540
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 3.941288090188637e-07,
+      "loss": 0.0145,
+      "step": 88550
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 3.907802210068088e-07,
+      "loss": 0.0335,
+      "step": 88560
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.874316329947539e-07,
+      "loss": 0.0299,
+      "step": 88570
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.84083044982699e-07,
+      "loss": 0.0624,
+      "step": 88580
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.80734456970644e-07,
+      "loss": 0.0254,
+      "step": 88590
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.773858689585891e-07,
+      "loss": 0.0398,
+      "step": 88600
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.7403728094653424e-07,
+      "loss": 0.0219,
+      "step": 88610
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.706886929344793e-07,
+      "loss": 0.0147,
+      "step": 88620
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.673401049224244e-07,
+      "loss": 0.0639,
+      "step": 88630
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.639915169103694e-07,
+      "loss": 0.0253,
+      "step": 88640
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.6064292889831456e-07,
+      "loss": 0.0268,
+      "step": 88650
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.5729434088625964e-07,
+      "loss": 0.0566,
+      "step": 88660
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.539457528742047e-07,
+      "loss": 0.0283,
+      "step": 88670
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.505971648621498e-07,
+      "loss": 0.0296,
+      "step": 88680
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.4724857685009493e-07,
+      "loss": 0.0593,
+      "step": 88690
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.4389998883803996e-07,
+      "loss": 0.0305,
+      "step": 88700
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.4055140082598504e-07,
+      "loss": 0.0335,
+      "step": 88710
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.372028128139301e-07,
+      "loss": 0.085,
+      "step": 88720
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.3385422480187525e-07,
+      "loss": 0.0222,
+      "step": 88730
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.3050563678982033e-07,
+      "loss": 0.0193,
+      "step": 88740
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.2715704877776536e-07,
+      "loss": 0.0549,
+      "step": 88750
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.2380846076571044e-07,
+      "loss": 0.0475,
+      "step": 88760
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.2045987275365557e-07,
+      "loss": 0.0247,
+      "step": 88770
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.1711128474160065e-07,
+      "loss": 0.0229,
+      "step": 88780
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.1376269672954573e-07,
+      "loss": 0.0245,
+      "step": 88790
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.104141087174908e-07,
+      "loss": 0.0218,
+      "step": 88800
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.070655207054359e-07,
+      "loss": 0.0916,
+      "step": 88810
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.0371693269338097e-07,
+      "loss": 0.0282,
+      "step": 88820
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.0036834468132605e-07,
+      "loss": 0.0147,
+      "step": 88830
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 2.9701975666927113e-07,
+      "loss": 0.023,
+      "step": 88840
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 2.9367116865721626e-07,
+      "loss": 0.0404,
+      "step": 88850
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 2.903225806451613e-07,
+      "loss": 0.0226,
+      "step": 88860
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 2.8697399263310637e-07,
+      "loss": 0.0498,
+      "step": 88870
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 2.8362540462105145e-07,
+      "loss": 0.051,
+      "step": 88880
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 2.8027681660899653e-07,
+      "loss": 0.0323,
+      "step": 88890
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 2.7692822859694166e-07,
+      "loss": 0.0444,
+      "step": 88900
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 2.735796405848867e-07,
+      "loss": 0.0139,
+      "step": 88910
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 2.7023105257283177e-07,
+      "loss": 0.0136,
+      "step": 88920
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.6688246456077685e-07,
+      "loss": 0.0458,
+      "step": 88930
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.63533876548722e-07,
+      "loss": 0.0024,
+      "step": 88940
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.6018528853666706e-07,
+      "loss": 0.0528,
+      "step": 88950
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.5683670052461214e-07,
+      "loss": 0.0484,
+      "step": 88960
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.5348811251255717e-07,
+      "loss": 0.1044,
+      "step": 88970
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.501395245005023e-07,
+      "loss": 0.0213,
+      "step": 88980
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.467909364884474e-07,
+      "loss": 0.0735,
+      "step": 88990
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.4344234847639246e-07,
+      "loss": 0.0166,
+      "step": 89000
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.4009376046433754e-07,
+      "loss": 0.0086,
+      "step": 89010
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.3674517245228265e-07,
+      "loss": 0.0088,
+      "step": 89020
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.3339658444022773e-07,
+      "loss": 0.0118,
+      "step": 89030
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.3004799642817278e-07,
+      "loss": 0.0259,
+      "step": 89040
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.266994084161179e-07,
+      "loss": 0.0418,
+      "step": 89050
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.2335082040406294e-07,
+      "loss": 0.0128,
+      "step": 89060
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.2000223239200805e-07,
+      "loss": 0.0291,
+      "step": 89070
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.1665364437995313e-07,
+      "loss": 0.0319,
+      "step": 89080
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.133050563678982e-07,
+      "loss": 0.0443,
+      "step": 89090
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.099564683558433e-07,
+      "loss": 0.0409,
+      "step": 89100
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.066078803437884e-07,
+      "loss": 0.0244,
+      "step": 89110
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.0325929233173345e-07,
+      "loss": 0.0193,
+      "step": 89120
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.9991070431967855e-07,
+      "loss": 0.0321,
+      "step": 89130
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.965621163076236e-07,
+      "loss": 0.0329,
+      "step": 89140
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.932135282955687e-07,
+      "loss": 0.0545,
+      "step": 89150
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.898649402835138e-07,
+      "loss": 0.0584,
+      "step": 89160
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.8651635227145887e-07,
+      "loss": 0.0706,
+      "step": 89170
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.8316776425940395e-07,
+      "loss": 0.0079,
+      "step": 89180
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.7981917624734906e-07,
+      "loss": 0.0102,
+      "step": 89190
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.764705882352941e-07,
+      "loss": 0.0475,
+      "step": 89200
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.7312200022323922e-07,
+      "loss": 0.0344,
+      "step": 89210
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.697734122111843e-07,
+      "loss": 0.0379,
+      "step": 89220
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.6642482419912938e-07,
+      "loss": 0.0571,
+      "step": 89230
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.6307623618707446e-07,
+      "loss": 0.0249,
+      "step": 89240
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.5972764817501954e-07,
+      "loss": 0.0395,
+      "step": 89250
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.5637906016296462e-07,
+      "loss": 0.0552,
+      "step": 89260
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.5303047215090973e-07,
+      "loss": 0.0473,
+      "step": 89270
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.4968188413885478e-07,
+      "loss": 0.0493,
+      "step": 89280
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.4633329612679986e-07,
+      "loss": 0.0254,
+      "step": 89290
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.4298470811474497e-07,
+      "loss": 0.0087,
+      "step": 89300
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.3963612010269002e-07,
+      "loss": 0.0317,
+      "step": 89310
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.3628753209063512e-07,
+      "loss": 0.0215,
+      "step": 89320
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.3293894407858018e-07,
+      "loss": 0.049,
+      "step": 89330
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.2959035606652528e-07,
+      "loss": 0.0481,
+      "step": 89340
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.2624176805447036e-07,
+      "loss": 0.0777,
+      "step": 89350
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.2289318004241544e-07,
+      "loss": 0.0449,
+      "step": 89360
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.1987945083156602e-07,
+      "loss": 0.0519,
+      "step": 89370
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.165308628195111e-07,
+      "loss": 0.084,
+      "step": 89380
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.1318227480745618e-07,
+      "loss": 0.0437,
+      "step": 89390
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.0983368679540127e-07,
+      "loss": 0.173,
+      "step": 89400
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.0648509878334635e-07,
+      "loss": 0.0177,
+      "step": 89410
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.0313651077129143e-07,
+      "loss": 0.0511,
+      "step": 89420
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 9.978792275923651e-08,
+      "loss": 0.1003,
+      "step": 89430
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 9.643933474718161e-08,
+      "loss": 0.0179,
+      "step": 89440
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 9.309074673512669e-08,
+      "loss": 0.0317,
+      "step": 89450
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 8.974215872307177e-08,
+      "loss": 0.0439,
+      "step": 89460
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 8.639357071101685e-08,
+      "loss": 0.0302,
+      "step": 89470
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 8.304498269896194e-08,
+      "loss": 0.0535,
+      "step": 89480
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 7.969639468690702e-08,
+      "loss": 0.0817,
+      "step": 89490
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 7.63478066748521e-08,
+      "loss": 0.0277,
+      "step": 89500
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 7.29992186627972e-08,
+      "loss": 0.0515,
+      "step": 89510
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 6.965063065074227e-08,
+      "loss": 0.0198,
+      "step": 89520
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 6.630204263868735e-08,
+      "loss": 0.0424,
+      "step": 89530
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 6.295345462663243e-08,
+      "loss": 0.0238,
+      "step": 89540
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.960486661457753e-08,
+      "loss": 0.0601,
+      "step": 89550
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.625627860252261e-08,
+      "loss": 0.0966,
+      "step": 89560
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.290769059046769e-08,
+      "loss": 0.0374,
+      "step": 89570
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 4.955910257841277e-08,
+      "loss": 0.0438,
+      "step": 89580
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 4.621051456635785e-08,
+      "loss": 0.0209,
+      "step": 89590
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 4.286192655430294e-08,
+      "loss": 0.048,
+      "step": 89600
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.951333854224802e-08,
+      "loss": 0.019,
+      "step": 89610
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.6164750530193106e-08,
+      "loss": 0.0654,
+      "step": 89620
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.2816162518138186e-08,
+      "loss": 0.0398,
+      "step": 89630
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 2.9467574506083273e-08,
+      "loss": 0.0725,
+      "step": 89640
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 2.6118986494028353e-08,
+      "loss": 0.0233,
+      "step": 89650
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 2.2770398481973436e-08,
+      "loss": 0.0435,
+      "step": 89660
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.942181046991852e-08,
+      "loss": 0.0096,
+      "step": 89670
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.6073222457863602e-08,
+      "loss": 0.0353,
+      "step": 89680
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.2724634445808686e-08,
+      "loss": 0.0274,
+      "step": 89690
+    },
+    {
+      "epoch": 10.0,
+      "eval_loss": 0.3072720468044281,
+      "eval_runtime": 252.6023,
+      "eval_samples_per_second": 7.894,
+      "eval_steps_per_second": 3.947,
+      "eval_wer": 0.08156879424005833,
+      "step": 89690
+    },
+    {
+      "epoch": 10.0,
+      "step": 89690,
+      "total_flos": 5.436213924030689e+19,
+      "train_loss": 0.14124298615650893,
+      "train_runtime": 41277.8445,
+      "train_samples_per_second": 4.345,
+      "train_steps_per_second": 2.173
+    },
+    {
+      "epoch": 10.0,
+      "step": 89690,
+      "total_flos": 5.436213924030689e+19,
+      "train_loss": 0.0,
+      "train_runtime": 1567.9322,
+      "train_samples_per_second": 294.528,
+      "train_steps_per_second": 4.592
+    }
+  ],
+  "max_steps": 7200,
+  "num_train_epochs": 100,
+  "total_flos": 5.436213924030689e+19,
+  "trial_name": null,
+  "trial_params": null
+}