diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,12060 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 29.999551173680825,
+  "eval_steps": 500,
+  "global_step": 1002600,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00015,
+      "loss": 23.4799,
+      "step": 500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003,
+      "loss": 4.9899,
+      "step": 1000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002998502396166134,
+      "loss": 3.793,
+      "step": 1500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002997004792332268,
+      "loss": 2.6917,
+      "step": 2000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029955071884984027,
+      "loss": 2.4188,
+      "step": 2500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002994009584664536,
+      "loss": 2.2087,
+      "step": 3000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002992511980830671,
+      "loss": 2.108,
+      "step": 3500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002991014376996805,
+      "loss": 2.0303,
+      "step": 4000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002989516773162939,
+      "loss": 1.9718,
+      "step": 4500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029880191693290733,
+      "loss": 1.9275,
+      "step": 5000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029865215654952074,
+      "loss": 1.8732,
+      "step": 5500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029850239616613415,
+      "loss": 1.8399,
+      "step": 6000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00029835263578274757,
+      "loss": 1.8453,
+      "step": 6500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000298202875399361,
+      "loss": 1.7905,
+      "step": 7000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002980531150159744,
+      "loss": 1.7855,
+      "step": 7500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002979033546325878,
+      "loss": 1.7553,
+      "step": 8000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029775359424920127,
+      "loss": 1.7553,
+      "step": 8500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029760383386581463,
+      "loss": 1.7429,
+      "step": 9000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002974540734824281,
+      "loss": 1.7157,
+      "step": 9500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002973043130990415,
+      "loss": 1.699,
+      "step": 10000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002971545527156549,
+      "loss": 1.6889,
+      "step": 10500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029700479233226834,
+      "loss": 1.6879,
+      "step": 11000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029685503194888175,
+      "loss": 1.6451,
+      "step": 11500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002967052715654952,
+      "loss": 1.6394,
+      "step": 12000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002965555111821086,
+      "loss": 1.6772,
+      "step": 12500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029640575079872204,
+      "loss": 1.6283,
+      "step": 13000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029625599041533545,
+      "loss": 1.6453,
+      "step": 13500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029610623003194887,
+      "loss": 1.5847,
+      "step": 14000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002959564696485623,
+      "loss": 1.6174,
+      "step": 14500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002958067092651757,
+      "loss": 1.6078,
+      "step": 15000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002956569488817891,
+      "loss": 1.6151,
+      "step": 15500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002955071884984025,
+      "loss": 1.5976,
+      "step": 16000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029535742811501593,
+      "loss": 1.5757,
+      "step": 16500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002952076677316294,
+      "loss": 1.5864,
+      "step": 17000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00029505790734824275,
+      "loss": 1.5726,
+      "step": 17500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002949081469648562,
+      "loss": 1.5873,
+      "step": 18000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029475838658146963,
+      "loss": 1.574,
+      "step": 18500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029460862619808305,
+      "loss": 1.5688,
+      "step": 19000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029445886581469646,
+      "loss": 1.5547,
+      "step": 19500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029430910543130987,
+      "loss": 1.5151,
+      "step": 20000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002941593450479233,
+      "loss": 1.5491,
+      "step": 20500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002940095846645367,
+      "loss": 1.5408,
+      "step": 21000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029385982428115016,
+      "loss": 1.5167,
+      "step": 21500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002937100638977635,
+      "loss": 1.5214,
+      "step": 22000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000293560303514377,
+      "loss": 1.5405,
+      "step": 22500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002934105431309904,
+      "loss": 1.5109,
+      "step": 23000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002932607827476038,
+      "loss": 1.5214,
+      "step": 23500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00029311102236421723,
+      "loss": 1.5263,
+      "step": 24000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00029296126198083064,
+      "loss": 1.519,
+      "step": 24500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00029281150159744405,
+      "loss": 1.5294,
+      "step": 25000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029266174121405747,
+      "loss": 1.5054,
+      "step": 25500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002925119808306709,
+      "loss": 1.5302,
+      "step": 26000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00029236222044728435,
+      "loss": 1.5025,
+      "step": 26500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002922124600638977,
+      "loss": 1.5263,
+      "step": 27000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00029206269968051117,
+      "loss": 1.5177,
+      "step": 27500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002919129392971246,
+      "loss": 1.4945,
+      "step": 28000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.000291763178913738,
+      "loss": 1.5229,
+      "step": 28500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002916134185303514,
+      "loss": 1.493,
+      "step": 29000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002914636581469648,
+      "loss": 1.498,
+      "step": 29500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002913138977635783,
+      "loss": 1.4723,
+      "step": 30000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00029116413738019165,
+      "loss": 1.4665,
+      "step": 30500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002910143769968051,
+      "loss": 1.5105,
+      "step": 31000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00029086461661341847,
+      "loss": 1.4768,
+      "step": 31500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00029071485623003194,
+      "loss": 1.4581,
+      "step": 32000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00029056509584664535,
+      "loss": 1.5065,
+      "step": 32500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00029041533546325876,
+      "loss": 1.4763,
+      "step": 33000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002902655750798722,
+      "loss": 1.4729,
+      "step": 33500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002901158146964856,
+      "loss": 1.4453,
+      "step": 34000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.000289966054313099,
+      "loss": 1.4323,
+      "step": 34500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002898162939297124,
+      "loss": 1.4344,
+      "step": 35000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00028966653354632583,
+      "loss": 1.4443,
+      "step": 35500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002895167731629393,
+      "loss": 1.4011,
+      "step": 36000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00028936701277955265,
+      "loss": 1.4288,
+      "step": 36500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002892172523961661,
+      "loss": 1.4035,
+      "step": 37000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00028906749201277953,
+      "loss": 1.4278,
+      "step": 37500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00028891773162939295,
+      "loss": 1.4242,
+      "step": 38000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00028876797124600636,
+      "loss": 1.3915,
+      "step": 38500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00028861821086261977,
+      "loss": 1.4098,
+      "step": 39000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00028846845047923324,
+      "loss": 1.4413,
+      "step": 39500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002883186900958466,
+      "loss": 1.4319,
+      "step": 40000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00028816892971246006,
+      "loss": 1.4128,
+      "step": 40500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002880191693290735,
+      "loss": 1.4043,
+      "step": 41000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002878694089456869,
+      "loss": 1.4143,
+      "step": 41500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002877196485623003,
+      "loss": 1.3753,
+      "step": 42000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002875698881789137,
+      "loss": 1.4136,
+      "step": 42500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002874201277955271,
+      "loss": 1.4277,
+      "step": 43000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00028727036741214054,
+      "loss": 1.4248,
+      "step": 43500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00028712060702875395,
+      "loss": 1.4246,
+      "step": 44000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00028697084664536736,
+      "loss": 1.3831,
+      "step": 44500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002868210862619808,
+      "loss": 1.4034,
+      "step": 45000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00028667132587859424,
+      "loss": 1.4165,
+      "step": 45500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002865215654952076,
+      "loss": 1.3971,
+      "step": 46000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00028637180511182107,
+      "loss": 1.413,
+      "step": 46500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002862220447284345,
+      "loss": 1.4016,
+      "step": 47000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002860722843450479,
+      "loss": 1.3903,
+      "step": 47500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002859225239616613,
+      "loss": 1.3947,
+      "step": 48000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002857727635782747,
+      "loss": 1.379,
+      "step": 48500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002856230031948882,
+      "loss": 1.3777,
+      "step": 49000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00028547324281150155,
+      "loss": 1.387,
+      "step": 49500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.000285323482428115,
+      "loss": 1.3775,
+      "step": 50000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002851737220447284,
+      "loss": 1.3784,
+      "step": 50500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00028502396166134184,
+      "loss": 1.3837,
+      "step": 51000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00028487420127795525,
+      "loss": 1.3763,
+      "step": 51500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00028472444089456866,
+      "loss": 1.3744,
+      "step": 52000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002845746805111821,
+      "loss": 1.3657,
+      "step": 52500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0002844249201277955,
+      "loss": 1.348,
+      "step": 53000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002842751597444089,
+      "loss": 1.3662,
+      "step": 53500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00028412539936102237,
+      "loss": 1.3676,
+      "step": 54000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002839756389776357,
+      "loss": 1.3601,
+      "step": 54500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0002838258785942492,
+      "loss": 1.361,
+      "step": 55000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0002836761182108626,
+      "loss": 1.3691,
+      "step": 55500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.000283526357827476,
+      "loss": 1.3578,
+      "step": 56000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00028337659744408943,
+      "loss": 1.3864,
+      "step": 56500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00028322683706070284,
+      "loss": 1.363,
+      "step": 57000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00028307707667731626,
+      "loss": 1.3488,
+      "step": 57500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00028292731629392967,
+      "loss": 1.3706,
+      "step": 58000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00028277755591054314,
+      "loss": 1.39,
+      "step": 58500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002826277955271565,
+      "loss": 1.3524,
+      "step": 59000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00028247803514376996,
+      "loss": 1.3477,
+      "step": 59500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0002823282747603834,
+      "loss": 1.3434,
+      "step": 60000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002821785143769968,
+      "loss": 1.3478,
+      "step": 60500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002820287539936102,
+      "loss": 1.3387,
+      "step": 61000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002818789936102236,
+      "loss": 1.3574,
+      "step": 61500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.000281729233226837,
+      "loss": 1.3426,
+      "step": 62000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00028157947284345044,
+      "loss": 1.3526,
+      "step": 62500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00028142971246006385,
+      "loss": 1.3405,
+      "step": 63000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002812799520766773,
+      "loss": 1.3432,
+      "step": 63500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002811301916932907,
+      "loss": 1.3549,
+      "step": 64000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00028098043130990414,
+      "loss": 1.3422,
+      "step": 64500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00028083067092651756,
+      "loss": 1.3373,
+      "step": 65000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00028068091054313097,
+      "loss": 1.3379,
+      "step": 65500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002805311501597444,
+      "loss": 1.3265,
+      "step": 66000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0002803813897763578,
+      "loss": 1.3608,
+      "step": 66500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00028023162939297126,
+      "loss": 1.3178,
+      "step": 67000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0002800818690095846,
+      "loss": 1.267,
+      "step": 67500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0002799321086261981,
+      "loss": 1.2854,
+      "step": 68000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00027978234824281144,
+      "loss": 1.2774,
+      "step": 68500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0002796325878594249,
+      "loss": 1.2755,
+      "step": 69000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0002794828274760383,
+      "loss": 1.3021,
+      "step": 69500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00027933306709265174,
+      "loss": 1.2467,
+      "step": 70000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00027918330670926515,
+      "loss": 1.2993,
+      "step": 70500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00027903354632587856,
+      "loss": 1.2699,
+      "step": 71000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.000278883785942492,
+      "loss": 1.2788,
+      "step": 71500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0002787340255591054,
+      "loss": 1.2875,
+      "step": 72000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0002785842651757188,
+      "loss": 1.2893,
+      "step": 72500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00027843450479233227,
+      "loss": 1.295,
+      "step": 73000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0002782847444089456,
+      "loss": 1.2726,
+      "step": 73500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0002781349840255591,
+      "loss": 1.2756,
+      "step": 74000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0002779852236421725,
+      "loss": 1.2939,
+      "step": 74500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0002778354632587859,
+      "loss": 1.2891,
+      "step": 75000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00027768570287539933,
+      "loss": 1.3099,
+      "step": 75500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00027753594249201274,
+      "loss": 1.2733,
+      "step": 76000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0002773861821086262,
+      "loss": 1.3034,
+      "step": 76500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00027723642172523957,
+      "loss": 1.2756,
+      "step": 77000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00027708666134185304,
+      "loss": 1.3009,
+      "step": 77500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00027693690095846645,
+      "loss": 1.2973,
+      "step": 78000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00027678714057507986,
+      "loss": 1.2698,
+      "step": 78500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0002766373801916933,
+      "loss": 1.3091,
+      "step": 79000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0002764876198083067,
+      "loss": 1.3076,
+      "step": 79500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0002763378594249201,
+      "loss": 1.2882,
+      "step": 80000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0002761880990415335,
+      "loss": 1.2655,
+      "step": 80500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0002760383386581469,
+      "loss": 1.2583,
+      "step": 81000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00027588857827476034,
+      "loss": 1.2835,
+      "step": 81500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00027573881789137375,
+      "loss": 1.2842,
+      "step": 82000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0002755890575079872,
+      "loss": 1.3011,
+      "step": 82500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0002754392971246006,
+      "loss": 1.2745,
+      "step": 83000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00027528953674121404,
+      "loss": 1.2762,
+      "step": 83500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00027513977635782745,
+      "loss": 1.2641,
+      "step": 84000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00027499001597444087,
+      "loss": 1.2573,
+      "step": 84500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0002748402555910543,
+      "loss": 1.2622,
+      "step": 85000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0002746904952076677,
+      "loss": 1.2864,
+      "step": 85500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00027454073482428116,
+      "loss": 1.2456,
+      "step": 86000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0002743909744408945,
+      "loss": 1.2529,
+      "step": 86500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.000274241214057508,
+      "loss": 1.2761,
+      "step": 87000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0002740914536741214,
+      "loss": 1.2726,
+      "step": 87500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0002739416932907348,
+      "loss": 1.2614,
+      "step": 88000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0002737919329073482,
+      "loss": 1.2803,
+      "step": 88500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00027364217252396164,
+      "loss": 1.2331,
+      "step": 89000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00027349241214057505,
+      "loss": 1.274,
+      "step": 89500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00027334265175718846,
+      "loss": 1.2533,
+      "step": 90000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0002731928913738019,
+      "loss": 1.2712,
+      "step": 90500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00027304313099041534,
+      "loss": 1.2864,
+      "step": 91000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0002728933706070287,
+      "loss": 1.258,
+      "step": 91500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00027274361022364217,
+      "loss": 1.2637,
+      "step": 92000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0002725938498402555,
+      "loss": 1.2523,
+      "step": 92500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.000272444089456869,
+      "loss": 1.2683,
+      "step": 93000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0002722943290734824,
+      "loss": 1.2721,
+      "step": 93500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0002721445686900958,
+      "loss": 1.2675,
+      "step": 94000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00027199480830670923,
+      "loss": 1.2773,
+      "step": 94500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00027184504792332264,
+      "loss": 1.2691,
+      "step": 95000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0002716952875399361,
+      "loss": 1.2614,
+      "step": 95500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00027154552715654947,
+      "loss": 1.2524,
+      "step": 96000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00027139576677316293,
+      "loss": 1.2524,
+      "step": 96500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00027124600638977635,
+      "loss": 1.2764,
+      "step": 97000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00027109624600638976,
+      "loss": 1.2356,
+      "step": 97500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00027094648562300317,
+      "loss": 1.2531,
+      "step": 98000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0002707967252396166,
+      "loss": 1.2488,
+      "step": 98500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00027064696485623,
+      "loss": 1.2497,
+      "step": 99000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0002704972044728434,
+      "loss": 1.2353,
+      "step": 99500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0002703474440894568,
+      "loss": 1.2522,
+      "step": 100000
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0002701976837060703,
+      "loss": 1.2157,
+      "step": 100500
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00027004792332268365,
+      "loss": 1.2059,
+      "step": 101000
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0002698981629392971,
+      "loss": 1.1675,
+      "step": 101500
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00026974840255591053,
+      "loss": 1.1866,
+      "step": 102000
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00026959864217252394,
+      "loss": 1.1894,
+      "step": 102500
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00026944888178913735,
+      "loss": 1.1943,
+      "step": 103000
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00026929912140575077,
+      "loss": 1.214,
+      "step": 103500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00026914936102236423,
+      "loss": 1.197,
+      "step": 104000
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0002689996006389776,
+      "loss": 1.1804,
+      "step": 104500
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00026884984025559106,
+      "loss": 1.2149,
+      "step": 105000
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0002687000798722044,
+      "loss": 1.2021,
+      "step": 105500
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0002685503194888179,
+      "loss": 1.1997,
+      "step": 106000
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0002684005591054313,
+      "loss": 1.2202,
+      "step": 106500
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0002682507987220447,
+      "loss": 1.2099,
+      "step": 107000
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0002681010383386581,
+      "loss": 1.1896,
+      "step": 107500
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00026795127795527153,
+      "loss": 1.1964,
+      "step": 108000
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00026780151757188495,
+      "loss": 1.1964,
+      "step": 108500
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00026765175718849836,
+      "loss": 1.227,
+      "step": 109000
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00026750199680511177,
+      "loss": 1.2226,
+      "step": 109500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00026735223642172524,
+      "loss": 1.2058,
+      "step": 110000
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0002672024760383386,
+      "loss": 1.211,
+      "step": 110500
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00026705271565495206,
+      "loss": 1.2202,
+      "step": 111000
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0002669029552715655,
+      "loss": 1.1997,
+      "step": 111500
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0002667531948881789,
+      "loss": 1.2182,
+      "step": 112000
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0002666034345047923,
+      "loss": 1.1984,
+      "step": 112500
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0002664536741214057,
+      "loss": 1.225,
+      "step": 113000
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0002663039137380192,
+      "loss": 1.1833,
+      "step": 113500
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00026615415335463254,
+      "loss": 1.2283,
+      "step": 114000
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.000266004392971246,
+      "loss": 1.2443,
+      "step": 114500
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0002658546325878594,
+      "loss": 1.2072,
+      "step": 115000
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00026570487220447283,
+      "loss": 1.1984,
+      "step": 115500
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00026555511182108625,
+      "loss": 1.1957,
+      "step": 116000
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00026540535143769966,
+      "loss": 1.1892,
+      "step": 116500
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00026525559105431307,
+      "loss": 1.2016,
+      "step": 117000
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0002651058306709265,
+      "loss": 1.218,
+      "step": 117500
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0002649560702875399,
+      "loss": 1.2045,
+      "step": 118000
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0002648063099041533,
+      "loss": 1.2168,
+      "step": 118500
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0002646565495207667,
+      "loss": 1.2093,
+      "step": 119000
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0002645067891373802,
+      "loss": 1.19,
+      "step": 119500
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00026435702875399355,
+      "loss": 1.2092,
+      "step": 120000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.000264207268370607,
+      "loss": 1.1932,
+      "step": 120500
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0002640575079872204,
+      "loss": 1.2064,
+      "step": 121000
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00026390774760383384,
+      "loss": 1.2128,
+      "step": 121500
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00026375798722044725,
+      "loss": 1.2292,
+      "step": 122000
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00026360822683706066,
+      "loss": 1.2224,
+      "step": 122500
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00026345846645367413,
+      "loss": 1.2055,
+      "step": 123000
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0002633087060702875,
+      "loss": 1.2179,
+      "step": 123500
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00026315894568690096,
+      "loss": 1.2333,
+      "step": 124000
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00026300918530351437,
+      "loss": 1.2168,
+      "step": 124500
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0002628594249201278,
+      "loss": 1.2051,
+      "step": 125000
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0002627096645367412,
+      "loss": 1.2078,
+      "step": 125500
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0002625599041533546,
+      "loss": 1.2075,
+      "step": 126000
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.000262410143769968,
+      "loss": 1.2343,
+      "step": 126500
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00026226038338658143,
+      "loss": 1.2069,
+      "step": 127000
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00026211062300319485,
+      "loss": 1.1925,
+      "step": 127500
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0002619608626198083,
+      "loss": 1.1936,
+      "step": 128000
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00026181110223642167,
+      "loss": 1.2048,
+      "step": 128500
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00026166134185303514,
+      "loss": 1.1921,
+      "step": 129000
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0002615115814696485,
+      "loss": 1.1923,
+      "step": 129500
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00026136182108626196,
+      "loss": 1.2246,
+      "step": 130000
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0002612120607028754,
+      "loss": 1.1842,
+      "step": 130500
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0002610623003194888,
+      "loss": 1.1877,
+      "step": 131000
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0002609125399361022,
+      "loss": 1.2153,
+      "step": 131500
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0002607627795527156,
+      "loss": 1.2133,
+      "step": 132000
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0002606130191693291,
+      "loss": 1.1676,
+      "step": 132500
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00026046325878594244,
+      "loss": 1.2033,
+      "step": 133000
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0002603134984025559,
+      "loss": 1.2008,
+      "step": 133500
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0002601637380191693,
+      "loss": 1.1657,
+      "step": 134000
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00026001397763578273,
+      "loss": 1.1436,
+      "step": 134500
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00025986421725239614,
+      "loss": 1.1119,
+      "step": 135000
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00025971445686900956,
+      "loss": 1.1286,
+      "step": 135500
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00025956469648562297,
+      "loss": 1.1546,
+      "step": 136000
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0002594149361022364,
+      "loss": 1.1581,
+      "step": 136500
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0002592651757188498,
+      "loss": 1.1109,
+      "step": 137000
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00025911541533546326,
+      "loss": 1.1457,
+      "step": 137500
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0002589656549520766,
+      "loss": 1.1404,
+      "step": 138000
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0002588158945686901,
+      "loss": 1.1153,
+      "step": 138500
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0002586661341853035,
+      "loss": 1.1447,
+      "step": 139000
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0002585163738019169,
+      "loss": 1.1412,
+      "step": 139500
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0002583666134185303,
+      "loss": 1.137,
+      "step": 140000
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00025821685303514374,
+      "loss": 1.1568,
+      "step": 140500
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0002580670926517572,
+      "loss": 1.1311,
+      "step": 141000
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00025791733226837056,
+      "loss": 1.1469,
+      "step": 141500
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00025776757188498403,
+      "loss": 1.1517,
+      "step": 142000
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0002576178115015974,
+      "loss": 1.1675,
+      "step": 142500
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00025746805111821086,
+      "loss": 1.1359,
+      "step": 143000
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00025731829073482427,
+      "loss": 1.1219,
+      "step": 143500
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.0002571685303514377,
+      "loss": 1.1604,
+      "step": 144000
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.0002570187699680511,
+      "loss": 1.1551,
+      "step": 144500
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.0002568690095846645,
+      "loss": 1.1412,
+      "step": 145000
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.0002567192492012779,
+      "loss": 1.1381,
+      "step": 145500
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00025656948881789133,
+      "loss": 1.1564,
+      "step": 146000
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00025641972843450474,
+      "loss": 1.1327,
+      "step": 146500
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.0002562699680511182,
+      "loss": 1.159,
+      "step": 147000
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00025612020766773157,
+      "loss": 1.1792,
+      "step": 147500
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00025597044728434504,
+      "loss": 1.1506,
+      "step": 148000
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00025582068690095845,
+      "loss": 1.1768,
+      "step": 148500
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00025567092651757186,
+      "loss": 1.1393,
+      "step": 149000
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.0002555211661341853,
+      "loss": 1.1733,
+      "step": 149500
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0002553714057507987,
+      "loss": 1.143,
+      "step": 150000
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00025522164536741215,
+      "loss": 1.1405,
+      "step": 150500
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.0002550718849840255,
+      "loss": 1.1642,
+      "step": 151000
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.000254922124600639,
+      "loss": 1.1664,
+      "step": 151500
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.0002547723642172524,
+      "loss": 1.1479,
+      "step": 152000
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.0002546226038338658,
+      "loss": 1.1437,
+      "step": 152500
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.0002544728434504792,
+      "loss": 1.1486,
+      "step": 153000
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00025432308306709263,
+      "loss": 1.1809,
+      "step": 153500
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00025417332268370604,
+      "loss": 1.1441,
+      "step": 154000
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00025402356230031946,
+      "loss": 1.1379,
+      "step": 154500
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00025387380191693287,
+      "loss": 1.1727,
+      "step": 155000
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.0002537240415335463,
+      "loss": 1.1383,
+      "step": 155500
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.0002535742811501597,
+      "loss": 1.1725,
+      "step": 156000
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00025342452076677316,
+      "loss": 1.1384,
+      "step": 156500
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.0002532747603833865,
+      "loss": 1.1828,
+      "step": 157000
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.000253125,
+      "loss": 1.1684,
+      "step": 157500
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0002529752396166134,
+      "loss": 1.1488,
+      "step": 158000
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0002528254792332268,
+      "loss": 1.1315,
+      "step": 158500
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.0002526757188498402,
+      "loss": 1.15,
+      "step": 159000
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00025252595846645364,
+      "loss": 1.1609,
+      "step": 159500
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.0002523761980830671,
+      "loss": 1.1686,
+      "step": 160000
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00025222643769968046,
+      "loss": 1.1349,
+      "step": 160500
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00025207667731629393,
+      "loss": 1.1485,
+      "step": 161000
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00025192691693290734,
+      "loss": 1.1666,
+      "step": 161500
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00025177715654952075,
+      "loss": 1.1426,
+      "step": 162000
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00025162739616613417,
+      "loss": 1.1571,
+      "step": 162500
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0002514776357827476,
+      "loss": 1.1655,
+      "step": 163000
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.000251327875399361,
+      "loss": 1.148,
+      "step": 163500
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.0002511781150159744,
+      "loss": 1.168,
+      "step": 164000
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.0002510283546325878,
+      "loss": 1.147,
+      "step": 164500
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.0002508785942492013,
+      "loss": 1.1212,
+      "step": 165000
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00025072883386581464,
+      "loss": 1.1505,
+      "step": 165500
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0002505790734824281,
+      "loss": 1.1373,
+      "step": 166000
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00025042931309904147,
+      "loss": 1.167,
+      "step": 166500
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00025027955271565494,
+      "loss": 1.1546,
+      "step": 167000
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00025012979233226835,
+      "loss": 1.1085,
+      "step": 167500
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00024998003194888176,
+      "loss": 1.0952,
+      "step": 168000
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.0002498302715654952,
+      "loss": 1.0728,
+      "step": 168500
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0002496805111821086,
+      "loss": 1.0914,
+      "step": 169000
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00024953075079872205,
+      "loss": 1.0724,
+      "step": 169500
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.0002493809904153354,
+      "loss": 1.0914,
+      "step": 170000
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.0002492312300319489,
+      "loss": 1.097,
+      "step": 170500
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.0002490814696485623,
+      "loss": 1.0977,
+      "step": 171000
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.0002489317092651757,
+      "loss": 1.0884,
+      "step": 171500
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.0002487819488817891,
+      "loss": 1.0892,
+      "step": 172000
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00024863218849840253,
+      "loss": 1.0994,
+      "step": 172500
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00024848242811501594,
+      "loss": 1.1073,
+      "step": 173000
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00024833266773162935,
+      "loss": 1.0973,
+      "step": 173500
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00024818290734824277,
+      "loss": 1.086,
+      "step": 174000
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00024803314696485623,
+      "loss": 1.0895,
+      "step": 174500
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0002478833865814696,
+      "loss": 1.0855,
+      "step": 175000
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00024773362619808306,
+      "loss": 1.1118,
+      "step": 175500
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00024758386581469647,
+      "loss": 1.0935,
+      "step": 176000
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.0002474341054313099,
+      "loss": 1.1035,
+      "step": 176500
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.0002472843450479233,
+      "loss": 1.1094,
+      "step": 177000
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.0002471345846645367,
+      "loss": 1.1199,
+      "step": 177500
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.0002469848242811502,
+      "loss": 1.1008,
+      "step": 178000
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00024683506389776354,
+      "loss": 1.0842,
+      "step": 178500
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.000246685303514377,
+      "loss": 1.1091,
+      "step": 179000
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00024653554313099036,
+      "loss": 1.1025,
+      "step": 179500
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00024638578274760383,
+      "loss": 1.0951,
+      "step": 180000
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00024623602236421724,
+      "loss": 1.0936,
+      "step": 180500
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00024608626198083065,
+      "loss": 1.1002,
+      "step": 181000
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00024593650159744407,
+      "loss": 1.0808,
+      "step": 181500
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.0002457867412140575,
+      "loss": 1.1175,
+      "step": 182000
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.0002456369808306709,
+      "loss": 1.1129,
+      "step": 182500
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0002454872204472843,
+      "loss": 1.1161,
+      "step": 183000
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.0002453374600638977,
+      "loss": 1.0924,
+      "step": 183500
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.0002451876996805112,
+      "loss": 1.0979,
+      "step": 184000
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00024503793929712454,
+      "loss": 1.1188,
+      "step": 184500
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.000244888178913738,
+      "loss": 1.0842,
+      "step": 185000
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.0002447384185303514,
+      "loss": 1.1094,
+      "step": 185500
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00024458865814696483,
+      "loss": 1.1087,
+      "step": 186000
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00024443889776357825,
+      "loss": 1.1102,
+      "step": 186500
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00024428913738019166,
+      "loss": 1.1239,
+      "step": 187000
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.0002441393769968051,
+      "loss": 1.1031,
+      "step": 187500
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.0002439896166134185,
+      "loss": 1.1239,
+      "step": 188000
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00024383985623003192,
+      "loss": 1.1183,
+      "step": 188500
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00024369009584664534,
+      "loss": 1.1168,
+      "step": 189000
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00024354033546325878,
+      "loss": 1.1102,
+      "step": 189500
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00024339057507987216,
+      "loss": 1.0952,
+      "step": 190000
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.0002432408146964856,
+      "loss": 1.1021,
+      "step": 190500
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00024309105431309902,
+      "loss": 1.1145,
+      "step": 191000
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00024294129392971246,
+      "loss": 1.1153,
+      "step": 191500
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00024279153354632584,
+      "loss": 1.0896,
+      "step": 192000
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00024264177316293928,
+      "loss": 1.0811,
+      "step": 192500
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.0002424920127795527,
+      "loss": 1.0697,
+      "step": 193000
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.0002423422523961661,
+      "loss": 1.0948,
+      "step": 193500
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00024219249201277952,
+      "loss": 1.1027,
+      "step": 194000
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00024204273162939296,
+      "loss": 1.0906,
+      "step": 194500
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00024189297124600637,
+      "loss": 1.1038,
+      "step": 195000
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00024174321086261978,
+      "loss": 1.1365,
+      "step": 195500
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00024159345047923322,
+      "loss": 1.0879,
+      "step": 196000
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.0002414436900958466,
+      "loss": 1.1044,
+      "step": 196500
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00024129392971246005,
+      "loss": 1.0886,
+      "step": 197000
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00024114416932907346,
+      "loss": 1.1174,
+      "step": 197500
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.0002409944089456869,
+      "loss": 1.0779,
+      "step": 198000
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.0002408446485623003,
+      "loss": 1.1075,
+      "step": 198500
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00024069488817891373,
+      "loss": 1.1154,
+      "step": 199000
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00024054512779552714,
+      "loss": 1.1139,
+      "step": 199500
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00024039536741214055,
+      "loss": 1.0851,
+      "step": 200000
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00024024560702875396,
+      "loss": 1.0981,
+      "step": 200500
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.0002400958466453674,
+      "loss": 1.0528,
+      "step": 201000
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.0002399460862619808,
+      "loss": 1.0366,
+      "step": 201500
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00023979632587859423,
+      "loss": 1.0546,
+      "step": 202000
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00023964656549520764,
+      "loss": 1.0456,
+      "step": 202500
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00023949680511182106,
+      "loss": 1.0361,
+      "step": 203000
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00023934704472843447,
+      "loss": 1.0735,
+      "step": 203500
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.0002391972843450479,
+      "loss": 1.0479,
+      "step": 204000
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00023904752396166135,
+      "loss": 1.0632,
+      "step": 204500
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00023889776357827473,
+      "loss": 1.071,
+      "step": 205000
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00023874800319488817,
+      "loss": 1.0402,
+      "step": 205500
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00023859824281150156,
+      "loss": 1.0265,
+      "step": 206000
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.000238448482428115,
+      "loss": 1.0542,
+      "step": 206500
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.0002382987220447284,
+      "loss": 1.0457,
+      "step": 207000
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00023814896166134185,
+      "loss": 1.0574,
+      "step": 207500
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00023799920127795524,
+      "loss": 1.0657,
+      "step": 208000
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00023784944089456868,
+      "loss": 1.0522,
+      "step": 208500
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.0002376996805111821,
+      "loss": 1.043,
+      "step": 209000
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.0002375499201277955,
+      "loss": 1.0466,
+      "step": 209500
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00023740015974440891,
+      "loss": 1.0544,
+      "step": 210000
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00023725039936102235,
+      "loss": 1.0619,
+      "step": 210500
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00023710063897763574,
+      "loss": 1.0543,
+      "step": 211000
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00023695087859424918,
+      "loss": 1.0456,
+      "step": 211500
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.0002368011182108626,
+      "loss": 1.0485,
+      "step": 212000
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.000236651357827476,
+      "loss": 1.0476,
+      "step": 212500
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00023650159744408944,
+      "loss": 1.0526,
+      "step": 213000
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00023635183706070286,
+      "loss": 1.0432,
+      "step": 213500
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.0002362020766773163,
+      "loss": 1.0607,
+      "step": 214000
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00023605231629392968,
+      "loss": 1.0486,
+      "step": 214500
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00023590255591054312,
+      "loss": 1.0624,
+      "step": 215000
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00023575279552715654,
+      "loss": 1.0566,
+      "step": 215500
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00023560303514376995,
+      "loss": 1.0847,
+      "step": 216000
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00023545327476038336,
+      "loss": 1.0442,
+      "step": 216500
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.0002353035143769968,
+      "loss": 1.0623,
+      "step": 217000
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00023515375399361019,
+      "loss": 1.0567,
+      "step": 217500
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00023500399361022363,
+      "loss": 1.0609,
+      "step": 218000
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00023485423322683704,
+      "loss": 1.0747,
+      "step": 218500
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00023470447284345045,
+      "loss": 1.0556,
+      "step": 219000
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00023455471246006386,
+      "loss": 1.0605,
+      "step": 219500
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.0002344049520766773,
+      "loss": 1.0856,
+      "step": 220000
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.0002342551916932907,
+      "loss": 1.0964,
+      "step": 220500
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00023410543130990413,
+      "loss": 1.0804,
+      "step": 221000
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00023395567092651754,
+      "loss": 1.0558,
+      "step": 221500
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00023380591054313098,
+      "loss": 1.0578,
+      "step": 222000
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.0002336561501597444,
+      "loss": 1.0621,
+      "step": 222500
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.0002335063897763578,
+      "loss": 1.0664,
+      "step": 223000
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00023335662939297125,
+      "loss": 1.0699,
+      "step": 223500
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00023320686900958463,
+      "loss": 1.066,
+      "step": 224000
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00023305710862619807,
+      "loss": 1.0877,
+      "step": 224500
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00023290734824281148,
+      "loss": 1.059,
+      "step": 225000
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.0002327575878594249,
+      "loss": 1.0639,
+      "step": 225500
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.0002326078274760383,
+      "loss": 1.0546,
+      "step": 226000
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00023245806709265175,
+      "loss": 1.0271,
+      "step": 226500
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00023230830670926514,
+      "loss": 1.0721,
+      "step": 227000
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00023215854632587857,
+      "loss": 1.0476,
+      "step": 227500
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.000232008785942492,
+      "loss": 1.0683,
+      "step": 228000
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00023185902555910543,
+      "loss": 1.0411,
+      "step": 228500
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.0002317092651757188,
+      "loss": 1.0563,
+      "step": 229000
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00023155950479233225,
+      "loss": 1.0823,
+      "step": 229500
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00023140974440894567,
+      "loss": 1.0532,
+      "step": 230000
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00023125998402555908,
+      "loss": 1.0354,
+      "step": 230500
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.0002311102236421725,
+      "loss": 1.0451,
+      "step": 231000
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00023096046325878593,
+      "loss": 1.0479,
+      "step": 231500
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00023081070287539934,
+      "loss": 1.0807,
+      "step": 232000
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00023066094249201276,
+      "loss": 1.0438,
+      "step": 232500
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.0002305111821086262,
+      "loss": 1.0733,
+      "step": 233000
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00023036142172523958,
+      "loss": 1.0792,
+      "step": 233500
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00023021166134185302,
+      "loss": 1.0469,
+      "step": 234000
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 0.00023006190095846643,
+      "loss": 0.9902,
+      "step": 234500
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 0.00022991214057507987,
+      "loss": 0.9927,
+      "step": 235000
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 0.00022976238019169326,
+      "loss": 1.0105,
+      "step": 235500
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 0.0002296126198083067,
+      "loss": 1.018,
+      "step": 236000
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 0.00022946285942492008,
+      "loss": 1.0335,
+      "step": 236500
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 0.00022931309904153352,
+      "loss": 1.0095,
+      "step": 237000
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 0.00022916333865814694,
+      "loss": 0.9966,
+      "step": 237500
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 0.00022901357827476038,
+      "loss": 1.0082,
+      "step": 238000
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 0.00022886381789137376,
+      "loss": 1.0327,
+      "step": 238500
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 0.0002287140575079872,
+      "loss": 1.0122,
+      "step": 239000
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 0.00022856429712460061,
+      "loss": 1.0224,
+      "step": 239500
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 0.00022841453674121403,
+      "loss": 1.0231,
+      "step": 240000
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 0.00022826477635782747,
+      "loss": 1.0052,
+      "step": 240500
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 0.00022811501597444088,
+      "loss": 1.0087,
+      "step": 241000
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 0.00022796525559105432,
+      "loss": 1.0279,
+      "step": 241500
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 0.0002278154952076677,
+      "loss": 1.0263,
+      "step": 242000
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 0.00022766573482428115,
+      "loss": 1.0041,
+      "step": 242500
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 0.00022751597444089453,
+      "loss": 1.0149,
+      "step": 243000
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 0.00022736621405750797,
+      "loss": 1.0164,
+      "step": 243500
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 0.00022721645367412138,
+      "loss": 1.0067,
+      "step": 244000
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 0.00022706669329073482,
+      "loss": 1.0188,
+      "step": 244500
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 0.0002269169329073482,
+      "loss": 1.0233,
+      "step": 245000
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 0.00022676717252396165,
+      "loss": 1.0186,
+      "step": 245500
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 0.00022661741214057506,
+      "loss": 1.0152,
+      "step": 246000
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 0.00022646765175718847,
+      "loss": 0.9957,
+      "step": 246500
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 0.00022631789137380189,
+      "loss": 1.0159,
+      "step": 247000
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 0.00022616813099041533,
+      "loss": 1.0359,
+      "step": 247500
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 0.0002260183706070287,
+      "loss": 1.0079,
+      "step": 248000
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 0.00022586861022364215,
+      "loss": 1.0205,
+      "step": 248500
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 0.00022571884984025556,
+      "loss": 1.0244,
+      "step": 249000
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 0.00022556908945686898,
+      "loss": 1.0302,
+      "step": 249500
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 0.00022541932907348242,
+      "loss": 1.0152,
+      "step": 250000
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 0.00022526956869009583,
+      "loss": 1.0128,
+      "step": 250500
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 0.00022511980830670927,
+      "loss": 1.0201,
+      "step": 251000
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 0.00022497004792332265,
+      "loss": 1.0254,
+      "step": 251500
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 0.0002248202875399361,
+      "loss": 1.0426,
+      "step": 252000
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 0.0002246705271565495,
+      "loss": 1.0105,
+      "step": 252500
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 0.00022452076677316292,
+      "loss": 1.0082,
+      "step": 253000
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 0.00022437100638977633,
+      "loss": 1.0276,
+      "step": 253500
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 0.00022422124600638977,
+      "loss": 1.0261,
+      "step": 254000
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 0.00022407148562300316,
+      "loss": 0.994,
+      "step": 254500
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 0.0002239217252396166,
+      "loss": 1.0095,
+      "step": 255000
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 0.00022377196485623,
+      "loss": 1.0315,
+      "step": 255500
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 0.00022362220447284342,
+      "loss": 1.0029,
+      "step": 256000
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 0.00022347244408945684,
+      "loss": 1.0124,
+      "step": 256500
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 0.00022332268370607028,
+      "loss": 1.0311,
+      "step": 257000
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 0.00022317292332268366,
+      "loss": 1.0021,
+      "step": 257500
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 0.0002230231629392971,
+      "loss": 1.0164,
+      "step": 258000
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 0.00022287340255591051,
+      "loss": 1.0196,
+      "step": 258500
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 0.00022272364217252395,
+      "loss": 1.0362,
+      "step": 259000
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 0.00022257388178913737,
+      "loss": 1.0137,
+      "step": 259500
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 0.00022242412140575078,
+      "loss": 1.0408,
+      "step": 260000
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 0.00022227436102236422,
+      "loss": 1.0095,
+      "step": 260500
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 0.0002221246006389776,
+      "loss": 1.0165,
+      "step": 261000
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 0.00022197484025559104,
+      "loss": 1.0419,
+      "step": 261500
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 0.00022182507987220446,
+      "loss": 1.0107,
+      "step": 262000
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 0.00022167531948881787,
+      "loss": 1.0269,
+      "step": 262500
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 0.00022152555910543128,
+      "loss": 1.0432,
+      "step": 263000
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 0.00022137579872204472,
+      "loss": 1.0556,
+      "step": 263500
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 0.0002212260383386581,
+      "loss": 1.0343,
+      "step": 264000
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 0.00022107627795527155,
+      "loss": 1.0128,
+      "step": 264500
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 0.00022092651757188496,
+      "loss": 1.0328,
+      "step": 265000
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 0.0002207767571884984,
+      "loss": 1.0329,
+      "step": 265500
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 0.00022062699680511179,
+      "loss": 1.0161,
+      "step": 266000
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 0.00022047723642172522,
+      "loss": 1.0081,
+      "step": 266500
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 0.0002203274760383386,
+      "loss": 1.0233,
+      "step": 267000
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 0.00022017771565495205,
+      "loss": 1.03,
+      "step": 267500
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 0.0002200279552715655,
+      "loss": 0.975,
+      "step": 268000
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 0.0002198781948881789,
+      "loss": 0.9583,
+      "step": 268500
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 0.00021972843450479232,
+      "loss": 0.9618,
+      "step": 269000
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 0.00021957867412140573,
+      "loss": 0.9676,
+      "step": 269500
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 0.00021942891373801917,
+      "loss": 0.9655,
+      "step": 270000
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 0.00021927915335463255,
+      "loss": 0.9536,
+      "step": 270500
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 0.000219129392971246,
+      "loss": 0.9672,
+      "step": 271000
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 0.0002189796325878594,
+      "loss": 0.9976,
+      "step": 271500
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 0.00021882987220447285,
+      "loss": 0.9612,
+      "step": 272000
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 0.00021868011182108623,
+      "loss": 0.9824,
+      "step": 272500
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 0.00021853035143769967,
+      "loss": 0.9593,
+      "step": 273000
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 0.00021838059105431306,
+      "loss": 0.9749,
+      "step": 273500
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 0.0002182308306709265,
+      "loss": 0.9759,
+      "step": 274000
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 0.0002180810702875399,
+      "loss": 1.0068,
+      "step": 274500
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 0.00021793130990415335,
+      "loss": 0.9743,
+      "step": 275000
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 0.00021778154952076673,
+      "loss": 0.9575,
+      "step": 275500
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 0.00021763178913738017,
+      "loss": 0.9643,
+      "step": 276000
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 0.0002174820287539936,
+      "loss": 0.9848,
+      "step": 276500
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 0.000217332268370607,
+      "loss": 0.9677,
+      "step": 277000
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 0.00021718250798722044,
+      "loss": 0.9734,
+      "step": 277500
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 0.00021703274760383385,
+      "loss": 0.9661,
+      "step": 278000
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 0.0002168829872204473,
+      "loss": 0.9774,
+      "step": 278500
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 0.00021673322683706068,
+      "loss": 0.9593,
+      "step": 279000
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 0.00021658346645367412,
+      "loss": 0.978,
+      "step": 279500
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 0.0002164337060702875,
+      "loss": 0.9788,
+      "step": 280000
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 0.00021628394568690094,
+      "loss": 0.9596,
+      "step": 280500
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 0.00021613418530351436,
+      "loss": 0.963,
+      "step": 281000
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 0.0002159844249201278,
+      "loss": 0.9841,
+      "step": 281500
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 0.00021583466453674118,
+      "loss": 0.9746,
+      "step": 282000
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 0.00021568490415335462,
+      "loss": 0.9659,
+      "step": 282500
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 0.00021553514376996803,
+      "loss": 0.997,
+      "step": 283000
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 0.00021538538338658145,
+      "loss": 0.9981,
+      "step": 283500
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 0.00021523562300319486,
+      "loss": 0.9688,
+      "step": 284000
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 0.0002150858626198083,
+      "loss": 0.9983,
+      "step": 284500
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 0.00021493610223642168,
+      "loss": 0.9915,
+      "step": 285000
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 0.00021478634185303512,
+      "loss": 1.0039,
+      "step": 285500
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 0.00021463658146964854,
+      "loss": 0.9768,
+      "step": 286000
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 0.00021448682108626195,
+      "loss": 0.9796,
+      "step": 286500
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 0.0002143370607028754,
+      "loss": 0.9887,
+      "step": 287000
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 0.0002141873003194888,
+      "loss": 0.9728,
+      "step": 287500
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 0.00021403753993610224,
+      "loss": 0.9841,
+      "step": 288000
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 0.00021388777955271563,
+      "loss": 0.9786,
+      "step": 288500
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 0.00021373801916932907,
+      "loss": 1.0024,
+      "step": 289000
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 0.00021358825878594248,
+      "loss": 0.9853,
+      "step": 289500
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 0.0002134384984025559,
+      "loss": 0.9969,
+      "step": 290000
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 0.0002132887380191693,
+      "loss": 0.9956,
+      "step": 290500
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 0.00021313897763578274,
+      "loss": 0.9579,
+      "step": 291000
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 0.00021298921725239613,
+      "loss": 0.9637,
+      "step": 291500
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 0.00021283945686900957,
+      "loss": 0.9834,
+      "step": 292000
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 0.00021268969648562298,
+      "loss": 0.9773,
+      "step": 292500
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 0.0002125399361022364,
+      "loss": 0.99,
+      "step": 293000
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 0.0002123901757188498,
+      "loss": 0.967,
+      "step": 293500
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 0.00021224041533546325,
+      "loss": 1.006,
+      "step": 294000
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 0.00021209065495207663,
+      "loss": 0.9965,
+      "step": 294500
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 0.00021194089456869007,
+      "loss": 0.9844,
+      "step": 295000
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 0.00021179113418530349,
+      "loss": 0.9895,
+      "step": 295500
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 0.00021164137380191693,
+      "loss": 0.9691,
+      "step": 296000
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 0.00021149161341853034,
+      "loss": 0.9526,
+      "step": 296500
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 0.00021134185303514375,
+      "loss": 0.9893,
+      "step": 297000
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 0.0002111920926517572,
+      "loss": 0.9982,
+      "step": 297500
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 0.00021104233226837058,
+      "loss": 0.9752,
+      "step": 298000
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 0.00021089257188498402,
+      "loss": 0.9992,
+      "step": 298500
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 0.00021074281150159743,
+      "loss": 0.9908,
+      "step": 299000
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 0.00021059305111821084,
+      "loss": 0.9746,
+      "step": 299500
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 0.00021044329073482425,
+      "loss": 0.98,
+      "step": 300000
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 0.0002102935303514377,
+      "loss": 0.9883,
+      "step": 300500
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 0.00021014376996805108,
+      "loss": 0.9545,
+      "step": 301000
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 0.00020999400958466452,
+      "loss": 0.9327,
+      "step": 301500
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 0.00020984424920127793,
+      "loss": 0.9137,
+      "step": 302000
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 0.00020969448881789137,
+      "loss": 0.9335,
+      "step": 302500
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 0.00020954472843450476,
+      "loss": 0.9391,
+      "step": 303000
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 0.0002093949680511182,
+      "loss": 0.9427,
+      "step": 303500
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 0.00020924520766773158,
+      "loss": 0.9229,
+      "step": 304000
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 0.00020909544728434502,
+      "loss": 0.9228,
+      "step": 304500
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 0.00020894568690095846,
+      "loss": 0.9331,
+      "step": 305000
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 0.00020879592651757187,
+      "loss": 0.9496,
+      "step": 305500
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 0.0002086461661341853,
+      "loss": 0.9249,
+      "step": 306000
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 0.0002084964057507987,
+      "loss": 0.9254,
+      "step": 306500
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 0.00020834664536741214,
+      "loss": 0.9379,
+      "step": 307000
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 0.00020819688498402553,
+      "loss": 0.922,
+      "step": 307500
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 0.00020804712460063897,
+      "loss": 0.9219,
+      "step": 308000
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 0.00020789736421725238,
+      "loss": 0.9407,
+      "step": 308500
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 0.00020774760383386582,
+      "loss": 0.9381,
+      "step": 309000
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 0.0002075978434504792,
+      "loss": 0.9248,
+      "step": 309500
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 0.00020744808306709264,
+      "loss": 0.9409,
+      "step": 310000
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 0.00020729832268370603,
+      "loss": 0.9301,
+      "step": 310500
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 0.00020714856230031947,
+      "loss": 0.9537,
+      "step": 311000
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 0.00020699880191693288,
+      "loss": 0.9375,
+      "step": 311500
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 0.00020684904153354632,
+      "loss": 0.9418,
+      "step": 312000
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 0.0002066992811501597,
+      "loss": 0.9384,
+      "step": 312500
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 0.00020654952076677315,
+      "loss": 0.9366,
+      "step": 313000
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 0.00020639976038338656,
+      "loss": 0.9455,
+      "step": 313500
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 0.00020624999999999997,
+      "loss": 0.9523,
+      "step": 314000
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 0.0002061002396166134,
+      "loss": 0.9383,
+      "step": 314500
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 0.00020595047923322682,
+      "loss": 0.9259,
+      "step": 315000
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 0.00020580071884984026,
+      "loss": 0.9421,
+      "step": 315500
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 0.00020565095846645365,
+      "loss": 0.9396,
+      "step": 316000
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 0.0002055011980830671,
+      "loss": 0.9658,
+      "step": 316500
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 0.00020535143769968047,
+      "loss": 0.9442,
+      "step": 317000
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 0.00020520167731629391,
+      "loss": 0.9236,
+      "step": 317500
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 0.00020505191693290733,
+      "loss": 0.9394,
+      "step": 318000
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 0.00020490215654952077,
+      "loss": 0.9503,
+      "step": 318500
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 0.00020475239616613415,
+      "loss": 0.9422,
+      "step": 319000
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 0.0002046026357827476,
+      "loss": 0.9356,
+      "step": 319500
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 0.000204452875399361,
+      "loss": 0.9452,
+      "step": 320000
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 0.00020430311501597442,
+      "loss": 0.9599,
+      "step": 320500
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 0.00020415335463258783,
+      "loss": 0.9557,
+      "step": 321000
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 0.00020400359424920127,
+      "loss": 0.9537,
+      "step": 321500
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 0.00020385383386581466,
+      "loss": 0.9409,
+      "step": 322000
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 0.0002037040734824281,
+      "loss": 0.9482,
+      "step": 322500
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 0.0002035543130990415,
+      "loss": 0.9729,
+      "step": 323000
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 0.00020340455271565492,
+      "loss": 0.944,
+      "step": 323500
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 0.00020325479233226836,
+      "loss": 0.9456,
+      "step": 324000
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 0.00020310503194888177,
+      "loss": 0.9653,
+      "step": 324500
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 0.0002029552715654952,
+      "loss": 0.9549,
+      "step": 325000
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 0.0002028055111821086,
+      "loss": 0.9371,
+      "step": 325500
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 0.00020265575079872204,
+      "loss": 0.9567,
+      "step": 326000
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 0.00020250599041533545,
+      "loss": 0.9383,
+      "step": 326500
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 0.00020235623003194886,
+      "loss": 0.9597,
+      "step": 327000
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 0.00020220646964856228,
+      "loss": 0.947,
+      "step": 327500
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 0.00020205670926517572,
+      "loss": 0.9646,
+      "step": 328000
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 0.0002019069488817891,
+      "loss": 0.9422,
+      "step": 328500
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 0.00020175718849840254,
+      "loss": 0.9359,
+      "step": 329000
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 0.00020160742811501595,
+      "loss": 0.9531,
+      "step": 329500
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 0.00020145766773162937,
+      "loss": 0.9517,
+      "step": 330000
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 0.00020130790734824278,
+      "loss": 0.9477,
+      "step": 330500
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 0.00020115814696485622,
+      "loss": 0.9444,
+      "step": 331000
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 0.0002010083865814696,
+      "loss": 0.9414,
+      "step": 331500
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 0.00020085862619808305,
+      "loss": 0.9675,
+      "step": 332000
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 0.00020070886581469648,
+      "loss": 0.9142,
+      "step": 332500
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 0.0002005591054313099,
+      "loss": 0.9577,
+      "step": 333000
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 0.0002004093450479233,
+      "loss": 0.929,
+      "step": 333500
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 0.00020025958466453672,
+      "loss": 0.9588,
+      "step": 334000
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 0.00020010982428115016,
+      "loss": 0.9139,
+      "step": 334500
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 0.00019996006389776355,
+      "loss": 0.8883,
+      "step": 335000
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 0.000199810303514377,
+      "loss": 0.8883,
+      "step": 335500
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 0.0001996605431309904,
+      "loss": 0.8905,
+      "step": 336000
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 0.0001995107827476038,
+      "loss": 0.8876,
+      "step": 336500
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 0.00019936102236421723,
+      "loss": 0.8912,
+      "step": 337000
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 0.00019921126198083067,
+      "loss": 0.8805,
+      "step": 337500
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 0.00019906150159744405,
+      "loss": 0.883,
+      "step": 338000
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 0.0001989117412140575,
+      "loss": 0.8893,
+      "step": 338500
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 0.0001987619808306709,
+      "loss": 0.9196,
+      "step": 339000
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 0.00019861222044728434,
+      "loss": 0.8943,
+      "step": 339500
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 0.00019846246006389773,
+      "loss": 0.894,
+      "step": 340000
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 0.00019831269968051117,
+      "loss": 0.8997,
+      "step": 340500
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 0.00019816293929712455,
+      "loss": 0.9136,
+      "step": 341000
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 0.000198013178913738,
+      "loss": 0.8754,
+      "step": 341500
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 0.00019786341853035143,
+      "loss": 0.9284,
+      "step": 342000
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 0.00019771365814696485,
+      "loss": 0.9203,
+      "step": 342500
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 0.00019756389776357826,
+      "loss": 0.9126,
+      "step": 343000
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 0.00019741413738019167,
+      "loss": 0.9211,
+      "step": 343500
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 0.0001972643769968051,
+      "loss": 0.9092,
+      "step": 344000
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 0.0001971146166134185,
+      "loss": 0.8982,
+      "step": 344500
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 0.00019696485623003194,
+      "loss": 0.9141,
+      "step": 345000
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 0.00019681509584664535,
+      "loss": 0.9151,
+      "step": 345500
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 0.0001966653354632588,
+      "loss": 0.9094,
+      "step": 346000
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 0.00019651557507987218,
+      "loss": 0.9028,
+      "step": 346500
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 0.00019636581469648562,
+      "loss": 0.8975,
+      "step": 347000
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 0.000196216054313099,
+      "loss": 0.9045,
+      "step": 347500
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 0.00019606629392971244,
+      "loss": 0.9189,
+      "step": 348000
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 0.00019591653354632585,
+      "loss": 0.8872,
+      "step": 348500
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 0.0001957667731629393,
+      "loss": 0.9209,
+      "step": 349000
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 0.00019561701277955268,
+      "loss": 0.9011,
+      "step": 349500
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 0.00019546725239616612,
+      "loss": 0.9138,
+      "step": 350000
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 0.00019531749201277953,
+      "loss": 0.9452,
+      "step": 350500
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 0.00019516773162939294,
+      "loss": 0.8997,
+      "step": 351000
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 0.00019501797124600638,
+      "loss": 0.9215,
+      "step": 351500
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 0.0001948682108626198,
+      "loss": 0.9152,
+      "step": 352000
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 0.00019471845047923324,
+      "loss": 0.9013,
+      "step": 352500
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 0.00019456869009584662,
+      "loss": 0.9208,
+      "step": 353000
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 0.00019441892971246006,
+      "loss": 0.9118,
+      "step": 353500
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 0.00019426916932907345,
+      "loss": 0.9171,
+      "step": 354000
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 0.0001941194089456869,
+      "loss": 0.9118,
+      "step": 354500
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 0.0001939696485623003,
+      "loss": 0.9145,
+      "step": 355000
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 0.00019381988817891374,
+      "loss": 0.904,
+      "step": 355500
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 0.00019367012779552712,
+      "loss": 0.9131,
+      "step": 356000
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 0.00019352036741214056,
+      "loss": 0.9092,
+      "step": 356500
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 0.00019337060702875398,
+      "loss": 0.9211,
+      "step": 357000
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 0.0001932208466453674,
+      "loss": 0.9214,
+      "step": 357500
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 0.0001930710862619808,
+      "loss": 0.9003,
+      "step": 358000
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 0.00019292132587859424,
+      "loss": 0.9108,
+      "step": 358500
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 0.00019277156549520763,
+      "loss": 0.9372,
+      "step": 359000
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 0.00019262180511182107,
+      "loss": 0.924,
+      "step": 359500
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 0.00019247204472843448,
+      "loss": 0.896,
+      "step": 360000
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 0.0001923222843450479,
+      "loss": 0.903,
+      "step": 360500
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 0.00019217252396166133,
+      "loss": 0.9346,
+      "step": 361000
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 0.00019202276357827475,
+      "loss": 0.9126,
+      "step": 361500
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 0.00019187300319488819,
+      "loss": 0.9324,
+      "step": 362000
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 0.00019172324281150157,
+      "loss": 0.9199,
+      "step": 362500
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 0.000191573482428115,
+      "loss": 0.9182,
+      "step": 363000
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 0.00019142372204472842,
+      "loss": 0.9318,
+      "step": 363500
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 0.00019127396166134184,
+      "loss": 0.9305,
+      "step": 364000
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 0.00019112420127795525,
+      "loss": 0.9116,
+      "step": 364500
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 0.0001909744408945687,
+      "loss": 0.8984,
+      "step": 365000
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 0.00019082468051118207,
+      "loss": 0.9241,
+      "step": 365500
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 0.00019067492012779551,
+      "loss": 0.9081,
+      "step": 366000
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 0.00019052515974440893,
+      "loss": 0.9122,
+      "step": 366500
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 0.00019037539936102234,
+      "loss": 0.9316,
+      "step": 367000
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 0.00019022563897763575,
+      "loss": 0.9095,
+      "step": 367500
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 0.0001900758785942492,
+      "loss": 0.8827,
+      "step": 368000
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 0.00018992611821086258,
+      "loss": 0.8385,
+      "step": 368500
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 0.00018977635782747602,
+      "loss": 0.867,
+      "step": 369000
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 0.00018962659744408946,
+      "loss": 0.8622,
+      "step": 369500
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 0.00018947683706070287,
+      "loss": 0.8609,
+      "step": 370000
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 0.00018932707667731628,
+      "loss": 0.8795,
+      "step": 370500
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 0.0001891773162939297,
+      "loss": 0.8614,
+      "step": 371000
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 0.00018902755591054313,
+      "loss": 0.8555,
+      "step": 371500
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 0.00018887779552715652,
+      "loss": 0.8726,
+      "step": 372000
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 0.00018872803514376996,
+      "loss": 0.8529,
+      "step": 372500
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 0.00018857827476038337,
+      "loss": 0.8832,
+      "step": 373000
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 0.00018842851437699679,
+      "loss": 0.8645,
+      "step": 373500
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 0.0001882787539936102,
+      "loss": 0.8643,
+      "step": 374000
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 0.00018812899361022364,
+      "loss": 0.879,
+      "step": 374500
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 0.00018797923322683702,
+      "loss": 0.8742,
+      "step": 375000
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 0.00018782947284345046,
+      "loss": 0.8701,
+      "step": 375500
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 0.00018767971246006388,
+      "loss": 0.8786,
+      "step": 376000
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 0.00018752995207667732,
+      "loss": 0.8797,
+      "step": 376500
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 0.0001873801916932907,
+      "loss": 0.8542,
+      "step": 377000
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 0.00018723043130990414,
+      "loss": 0.8729,
+      "step": 377500
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 0.00018708067092651753,
+      "loss": 0.8445,
+      "step": 378000
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 0.00018693091054313097,
+      "loss": 0.871,
+      "step": 378500
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 0.0001867811501597444,
+      "loss": 0.8748,
+      "step": 379000
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 0.00018663138977635782,
+      "loss": 0.8865,
+      "step": 379500
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 0.00018648162939297123,
+      "loss": 0.874,
+      "step": 380000
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 0.00018633186900958464,
+      "loss": 0.8693,
+      "step": 380500
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 0.00018618210862619808,
+      "loss": 0.8794,
+      "step": 381000
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 0.00018603234824281147,
+      "loss": 0.8791,
+      "step": 381500
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 0.0001858825878594249,
+      "loss": 0.8671,
+      "step": 382000
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 0.00018573282747603832,
+      "loss": 0.8491,
+      "step": 382500
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 0.00018558306709265176,
+      "loss": 0.8819,
+      "step": 383000
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 0.00018543330670926515,
+      "loss": 0.8748,
+      "step": 383500
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 0.0001852835463258786,
+      "loss": 0.8733,
+      "step": 384000
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 0.00018513378594249197,
+      "loss": 0.8792,
+      "step": 384500
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 0.0001849840255591054,
+      "loss": 0.8864,
+      "step": 385000
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 0.00018483426517571883,
+      "loss": 0.8507,
+      "step": 385500
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 0.00018468450479233227,
+      "loss": 0.8696,
+      "step": 386000
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 0.00018453474440894565,
+      "loss": 0.8492,
+      "step": 386500
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 0.0001843849840255591,
+      "loss": 0.8739,
+      "step": 387000
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 0.0001842352236421725,
+      "loss": 0.8491,
+      "step": 387500
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 0.00018408546325878592,
+      "loss": 0.8955,
+      "step": 388000
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 0.00018393570287539936,
+      "loss": 0.8871,
+      "step": 388500
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 0.00018378594249201277,
+      "loss": 0.8769,
+      "step": 389000
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 0.0001836361821086262,
+      "loss": 0.8873,
+      "step": 389500
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 0.0001834864217252396,
+      "loss": 0.8805,
+      "step": 390000
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 0.00018333666134185303,
+      "loss": 0.8884,
+      "step": 390500
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 0.00018318690095846642,
+      "loss": 0.9073,
+      "step": 391000
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 0.00018303714057507986,
+      "loss": 0.8826,
+      "step": 391500
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 0.00018288738019169327,
+      "loss": 0.8643,
+      "step": 392000
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 0.0001827376198083067,
+      "loss": 0.8662,
+      "step": 392500
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 0.0001825878594249201,
+      "loss": 0.8806,
+      "step": 393000
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 0.00018243809904153354,
+      "loss": 0.8792,
+      "step": 393500
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 0.00018228833865814695,
+      "loss": 0.8839,
+      "step": 394000
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 0.00018213857827476036,
+      "loss": 0.8894,
+      "step": 394500
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 0.00018198881789137377,
+      "loss": 0.8785,
+      "step": 395000
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 0.00018183905750798721,
+      "loss": 0.8999,
+      "step": 395500
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 0.0001816892971246006,
+      "loss": 0.9009,
+      "step": 396000
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 0.00018153953674121404,
+      "loss": 0.9065,
+      "step": 396500
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 0.00018138977635782748,
+      "loss": 0.8804,
+      "step": 397000
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 0.00018124001597444087,
+      "loss": 0.8743,
+      "step": 397500
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 0.0001810902555910543,
+      "loss": 0.8761,
+      "step": 398000
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 0.00018094049520766772,
+      "loss": 0.8911,
+      "step": 398500
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 0.00018079073482428116,
+      "loss": 0.8875,
+      "step": 399000
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 0.00018064097444089454,
+      "loss": 0.8898,
+      "step": 399500
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 0.00018049121405750798,
+      "loss": 0.8921,
+      "step": 400000
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 0.0001803414536741214,
+      "loss": 0.8861,
+      "step": 400500
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 0.0001801916932907348,
+      "loss": 0.8986,
+      "step": 401000
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 0.00018004193290734822,
+      "loss": 0.8277,
+      "step": 401500
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 0.00017989217252396166,
+      "loss": 0.8461,
+      "step": 402000
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 0.00017974241214057505,
+      "loss": 0.8266,
+      "step": 402500
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 0.00017959265175718849,
+      "loss": 0.8216,
+      "step": 403000
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 0.0001794428913738019,
+      "loss": 0.8162,
+      "step": 403500
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 0.0001792931309904153,
+      "loss": 0.826,
+      "step": 404000
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 0.00017914337060702872,
+      "loss": 0.8092,
+      "step": 404500
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 0.00017899361022364216,
+      "loss": 0.8225,
+      "step": 405000
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 0.00017884384984025555,
+      "loss": 0.831,
+      "step": 405500
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 0.000178694089456869,
+      "loss": 0.8258,
+      "step": 406000
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 0.00017854432907348243,
+      "loss": 0.8143,
+      "step": 406500
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 0.00017839456869009584,
+      "loss": 0.8502,
+      "step": 407000
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 0.00017824480830670925,
+      "loss": 0.8489,
+      "step": 407500
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 0.00017809504792332267,
+      "loss": 0.8188,
+      "step": 408000
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 0.0001779452875399361,
+      "loss": 0.8419,
+      "step": 408500
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 0.0001777955271565495,
+      "loss": 0.8405,
+      "step": 409000
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 0.00017764576677316293,
+      "loss": 0.832,
+      "step": 409500
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 0.00017749600638977635,
+      "loss": 0.8396,
+      "step": 410000
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 0.00017734624600638976,
+      "loss": 0.85,
+      "step": 410500
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 0.00017719648562300317,
+      "loss": 0.863,
+      "step": 411000
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 0.0001770467252396166,
+      "loss": 0.8287,
+      "step": 411500
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 0.00017689696485623,
+      "loss": 0.8462,
+      "step": 412000
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 0.00017674720447284344,
+      "loss": 0.8257,
+      "step": 412500
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 0.00017659744408945685,
+      "loss": 0.8568,
+      "step": 413000
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 0.0001764476837060703,
+      "loss": 0.8281,
+      "step": 413500
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 0.00017629792332268367,
+      "loss": 0.8465,
+      "step": 414000
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 0.0001761481629392971,
+      "loss": 0.8418,
+      "step": 414500
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 0.0001759984025559105,
+      "loss": 0.8365,
+      "step": 415000
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 0.00017584864217252394,
+      "loss": 0.8302,
+      "step": 415500
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 0.00017569888178913738,
+      "loss": 0.8441,
+      "step": 416000
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 0.0001755491214057508,
+      "loss": 0.8491,
+      "step": 416500
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 0.0001753993610223642,
+      "loss": 0.8389,
+      "step": 417000
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 0.00017524960063897762,
+      "loss": 0.8377,
+      "step": 417500
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 0.00017509984025559106,
+      "loss": 0.8543,
+      "step": 418000
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 0.00017495007987220444,
+      "loss": 0.85,
+      "step": 418500
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 0.00017480031948881788,
+      "loss": 0.8455,
+      "step": 419000
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 0.0001746505591054313,
+      "loss": 0.8586,
+      "step": 419500
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 0.00017450079872204473,
+      "loss": 0.8455,
+      "step": 420000
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 0.00017435103833865812,
+      "loss": 0.8282,
+      "step": 420500
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 0.00017420127795527156,
+      "loss": 0.8499,
+      "step": 421000
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 0.00017405151757188495,
+      "loss": 0.8598,
+      "step": 421500
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 0.00017390175718849838,
+      "loss": 0.8478,
+      "step": 422000
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 0.0001737519968051118,
+      "loss": 0.8584,
+      "step": 422500
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 0.00017360223642172524,
+      "loss": 0.8485,
+      "step": 423000
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 0.00017345247603833862,
+      "loss": 0.8584,
+      "step": 423500
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 0.00017330271565495206,
+      "loss": 0.8533,
+      "step": 424000
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 0.0001731529552715655,
+      "loss": 0.8437,
+      "step": 424500
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 0.0001730031948881789,
+      "loss": 0.8562,
+      "step": 425000
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 0.00017285343450479233,
+      "loss": 0.8379,
+      "step": 425500
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 0.00017270367412140574,
+      "loss": 0.856,
+      "step": 426000
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 0.00017255391373801918,
+      "loss": 0.8626,
+      "step": 426500
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 0.00017240415335463257,
+      "loss": 0.8447,
+      "step": 427000
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 0.000172254392971246,
+      "loss": 0.8677,
+      "step": 427500
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 0.0001721046325878594,
+      "loss": 0.8726,
+      "step": 428000
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 0.00017195487220447283,
+      "loss": 0.8402,
+      "step": 428500
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 0.00017180511182108624,
+      "loss": 0.8469,
+      "step": 429000
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 0.00017165535143769968,
+      "loss": 0.8464,
+      "step": 429500
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 0.00017150559105431307,
+      "loss": 0.856,
+      "step": 430000
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 0.0001713558306709265,
+      "loss": 0.8455,
+      "step": 430500
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 0.00017120607028753992,
+      "loss": 0.8425,
+      "step": 431000
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 0.00017105630990415333,
+      "loss": 0.8373,
+      "step": 431500
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 0.00017090654952076675,
+      "loss": 0.8199,
+      "step": 432000
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 0.0001707567891373802,
+      "loss": 0.8551,
+      "step": 432500
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 0.00017060702875399357,
+      "loss": 0.8328,
+      "step": 433000
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 0.000170457268370607,
+      "loss": 0.8844,
+      "step": 433500
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 0.00017030750798722045,
+      "loss": 0.8498,
+      "step": 434000
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 0.00017015774760383384,
+      "loss": 0.8396,
+      "step": 434500
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 0.00017000798722044728,
+      "loss": 0.786,
+      "step": 435000
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 0.0001698582268370607,
+      "loss": 0.7915,
+      "step": 435500
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 0.00016970846645367413,
+      "loss": 0.8049,
+      "step": 436000
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 0.00016955870607028752,
+      "loss": 0.7845,
+      "step": 436500
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 0.00016940894568690096,
+      "loss": 0.7951,
+      "step": 437000
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 0.00016925918530351437,
+      "loss": 0.8176,
+      "step": 437500
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 0.00016910942492012778,
+      "loss": 0.7952,
+      "step": 438000
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 0.0001689596645367412,
+      "loss": 0.7845,
+      "step": 438500
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 0.00016880990415335463,
+      "loss": 0.7961,
+      "step": 439000
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 0.00016866014376996802,
+      "loss": 0.7973,
+      "step": 439500
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 0.00016851038338658146,
+      "loss": 0.8213,
+      "step": 440000
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 0.00016836062300319487,
+      "loss": 0.8302,
+      "step": 440500
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 0.00016821086261980828,
+      "loss": 0.8054,
+      "step": 441000
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 0.0001680611022364217,
+      "loss": 0.8056,
+      "step": 441500
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 0.00016791134185303514,
+      "loss": 0.8041,
+      "step": 442000
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 0.00016776158146964852,
+      "loss": 0.8022,
+      "step": 442500
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 0.00016761182108626196,
+      "loss": 0.802,
+      "step": 443000
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 0.0001674620607028754,
+      "loss": 0.8195,
+      "step": 443500
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 0.00016731230031948881,
+      "loss": 0.8102,
+      "step": 444000
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 0.00016716253993610223,
+      "loss": 0.7935,
+      "step": 444500
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 0.00016701277955271564,
+      "loss": 0.8194,
+      "step": 445000
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 0.00016686301916932908,
+      "loss": 0.8154,
+      "step": 445500
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 0.00016671325878594246,
+      "loss": 0.8031,
+      "step": 446000
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 0.0001665634984025559,
+      "loss": 0.8072,
+      "step": 446500
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 0.00016641373801916932,
+      "loss": 0.7967,
+      "step": 447000
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 0.00016626397763578273,
+      "loss": 0.8133,
+      "step": 447500
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 0.00016611421725239614,
+      "loss": 0.8108,
+      "step": 448000
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 0.00016596445686900958,
+      "loss": 0.8147,
+      "step": 448500
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 0.00016581469648562297,
+      "loss": 0.8213,
+      "step": 449000
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 0.0001656649361022364,
+      "loss": 0.8125,
+      "step": 449500
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 0.00016551517571884982,
+      "loss": 0.8216,
+      "step": 450000
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 0.00016536541533546326,
+      "loss": 0.8135,
+      "step": 450500
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 0.00016521565495207665,
+      "loss": 0.8352,
+      "step": 451000
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 0.00016506589456869009,
+      "loss": 0.8212,
+      "step": 451500
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 0.00016491613418530347,
+      "loss": 0.8092,
+      "step": 452000
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 0.0001647663738019169,
+      "loss": 0.8217,
+      "step": 452500
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 0.00016461661341853035,
+      "loss": 0.7998,
+      "step": 453000
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 0.00016446685303514376,
+      "loss": 0.8134,
+      "step": 453500
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 0.00016431709265175718,
+      "loss": 0.8082,
+      "step": 454000
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 0.0001641673322683706,
+      "loss": 0.8266,
+      "step": 454500
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 0.00016401757188498403,
+      "loss": 0.8233,
+      "step": 455000
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 0.00016386781150159741,
+      "loss": 0.8049,
+      "step": 455500
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 0.00016371805111821085,
+      "loss": 0.8258,
+      "step": 456000
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 0.00016356829073482427,
+      "loss": 0.8327,
+      "step": 456500
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 0.0001634185303514377,
+      "loss": 0.8187,
+      "step": 457000
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 0.0001632687699680511,
+      "loss": 0.8268,
+      "step": 457500
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 0.00016311900958466453,
+      "loss": 0.823,
+      "step": 458000
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 0.00016296924920127792,
+      "loss": 0.823,
+      "step": 458500
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 0.00016281948881789136,
+      "loss": 0.8051,
+      "step": 459000
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 0.00016266972843450477,
+      "loss": 0.8203,
+      "step": 459500
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 0.0001625199680511182,
+      "loss": 0.8123,
+      "step": 460000
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 0.0001623702076677316,
+      "loss": 0.8327,
+      "step": 460500
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 0.00016222044728434503,
+      "loss": 0.8125,
+      "step": 461000
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 0.00016207068690095847,
+      "loss": 0.8234,
+      "step": 461500
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 0.00016192092651757186,
+      "loss": 0.8232,
+      "step": 462000
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 0.0001617711661341853,
+      "loss": 0.8297,
+      "step": 462500
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 0.0001616214057507987,
+      "loss": 0.7991,
+      "step": 463000
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 0.00016147164536741215,
+      "loss": 0.8201,
+      "step": 463500
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 0.00016132188498402554,
+      "loss": 0.816,
+      "step": 464000
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 0.00016117212460063898,
+      "loss": 0.8229,
+      "step": 464500
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 0.00016102236421725236,
+      "loss": 0.8152,
+      "step": 465000
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 0.0001608726038338658,
+      "loss": 0.8228,
+      "step": 465500
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 0.00016072284345047922,
+      "loss": 0.8307,
+      "step": 466000
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 0.00016057308306709266,
+      "loss": 0.8177,
+      "step": 466500
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 0.00016042332268370604,
+      "loss": 0.8448,
+      "step": 467000
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 0.00016027356230031948,
+      "loss": 0.8249,
+      "step": 467500
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 0.0001601238019169329,
+      "loss": 0.8075,
+      "step": 468000
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 0.0001599740415335463,
+      "loss": 0.7742,
+      "step": 468500
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 0.00015982428115015972,
+      "loss": 0.7701,
+      "step": 469000
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 0.00015967452076677316,
+      "loss": 0.7619,
+      "step": 469500
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 0.00015952476038338654,
+      "loss": 0.7634,
+      "step": 470000
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 0.00015937499999999998,
+      "loss": 0.7544,
+      "step": 470500
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 0.00015922523961661342,
+      "loss": 0.736,
+      "step": 471000
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 0.0001590754792332268,
+      "loss": 0.769,
+      "step": 471500
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 0.00015892571884984025,
+      "loss": 0.7586,
+      "step": 472000
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 0.00015877595846645366,
+      "loss": 0.7972,
+      "step": 472500
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 0.0001586261980830671,
+      "loss": 0.7779,
+      "step": 473000
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 0.0001584764376996805,
+      "loss": 0.7586,
+      "step": 473500
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 0.00015832667731629393,
+      "loss": 0.793,
+      "step": 474000
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 0.00015817691693290734,
+      "loss": 0.7762,
+      "step": 474500
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 0.00015802715654952075,
+      "loss": 0.7788,
+      "step": 475000
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 0.00015787739616613417,
+      "loss": 0.7793,
+      "step": 475500
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 0.0001577276357827476,
+      "loss": 0.7799,
+      "step": 476000
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 0.000157577875399361,
+      "loss": 0.7891,
+      "step": 476500
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 0.00015742811501597443,
+      "loss": 0.7616,
+      "step": 477000
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 0.00015727835463258784,
+      "loss": 0.7831,
+      "step": 477500
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 0.00015712859424920126,
+      "loss": 0.7857,
+      "step": 478000
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 0.00015697883386581467,
+      "loss": 0.7895,
+      "step": 478500
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 0.0001568290734824281,
+      "loss": 0.7896,
+      "step": 479000
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 0.0001566793130990415,
+      "loss": 0.7566,
+      "step": 479500
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 0.00015652955271565493,
+      "loss": 0.776,
+      "step": 480000
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 0.00015637979233226837,
+      "loss": 0.786,
+      "step": 480500
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 0.00015623003194888179,
+      "loss": 0.7902,
+      "step": 481000
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 0.0001560802715654952,
+      "loss": 0.76,
+      "step": 481500
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 0.0001559305111821086,
+      "loss": 0.777,
+      "step": 482000
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 0.00015578075079872205,
+      "loss": 0.7818,
+      "step": 482500
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 0.00015563099041533544,
+      "loss": 0.7836,
+      "step": 483000
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 0.00015548123003194888,
+      "loss": 0.7863,
+      "step": 483500
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 0.0001553314696485623,
+      "loss": 0.8053,
+      "step": 484000
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 0.0001551817092651757,
+      "loss": 0.7706,
+      "step": 484500
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 0.00015503194888178911,
+      "loss": 0.7723,
+      "step": 485000
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 0.00015488218849840255,
+      "loss": 0.7985,
+      "step": 485500
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 0.00015473242811501594,
+      "loss": 0.7988,
+      "step": 486000
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 0.00015458266773162938,
+      "loss": 0.792,
+      "step": 486500
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 0.0001544329073482428,
+      "loss": 0.7991,
+      "step": 487000
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 0.00015428314696485623,
+      "loss": 0.7773,
+      "step": 487500
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 0.00015413338658146962,
+      "loss": 0.8028,
+      "step": 488000
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 0.00015398362619808306,
+      "loss": 0.7839,
+      "step": 488500
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 0.0001538338658146965,
+      "loss": 0.7911,
+      "step": 489000
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 0.00015368410543130988,
+      "loss": 0.7754,
+      "step": 489500
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 0.00015353434504792332,
+      "loss": 0.8065,
+      "step": 490000
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 0.00015338458466453674,
+      "loss": 0.7753,
+      "step": 490500
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 0.00015323482428115015,
+      "loss": 0.7965,
+      "step": 491000
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 0.00015308506389776356,
+      "loss": 0.7935,
+      "step": 491500
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 0.000152935303514377,
+      "loss": 0.7756,
+      "step": 492000
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 0.00015278554313099039,
+      "loss": 0.7933,
+      "step": 492500
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 0.00015263578274760383,
+      "loss": 0.7836,
+      "step": 493000
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 0.00015248602236421724,
+      "loss": 0.7926,
+      "step": 493500
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 0.00015233626198083068,
+      "loss": 0.7751,
+      "step": 494000
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 0.00015218650159744406,
+      "loss": 0.8028,
+      "step": 494500
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 0.0001520367412140575,
+      "loss": 0.796,
+      "step": 495000
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 0.0001518869808306709,
+      "loss": 0.786,
+      "step": 495500
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 0.00015173722044728433,
+      "loss": 0.8018,
+      "step": 496000
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 0.00015158746006389774,
+      "loss": 0.7879,
+      "step": 496500
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 0.00015143769968051118,
+      "loss": 0.7795,
+      "step": 497000
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 0.00015128793929712457,
+      "loss": 0.7955,
+      "step": 497500
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 0.000151138178913738,
+      "loss": 0.7927,
+      "step": 498000
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 0.00015098841853035145,
+      "loss": 0.7785,
+      "step": 498500
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 0.00015083865814696483,
+      "loss": 0.7751,
+      "step": 499000
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 0.00015068889776357827,
+      "loss": 0.7819,
+      "step": 499500
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 0.00015053913738019168,
+      "loss": 0.7877,
+      "step": 500000
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 0.0001503893769968051,
+      "loss": 0.7913,
+      "step": 500500
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 0.0001502396166134185,
+      "loss": 0.8015,
+      "step": 501000
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 0.00015008985623003195,
+      "loss": 0.7618,
+      "step": 501500
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 0.00014994009584664534,
+      "loss": 0.7476,
+      "step": 502000
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 0.00014979033546325878,
+      "loss": 0.7429,
+      "step": 502500
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 0.0001496405750798722,
+      "loss": 0.741,
+      "step": 503000
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 0.00014949081469648563,
+      "loss": 0.7469,
+      "step": 503500
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 0.00014934105431309904,
+      "loss": 0.7342,
+      "step": 504000
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 0.00014919129392971245,
+      "loss": 0.741,
+      "step": 504500
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 0.00014904153354632587,
+      "loss": 0.7526,
+      "step": 505000
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 0.00014889177316293928,
+      "loss": 0.7404,
+      "step": 505500
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 0.00014874201277955272,
+      "loss": 0.7454,
+      "step": 506000
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 0.00014859225239616613,
+      "loss": 0.7607,
+      "step": 506500
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 0.00014844249201277954,
+      "loss": 0.7347,
+      "step": 507000
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 0.00014829273162939296,
+      "loss": 0.7669,
+      "step": 507500
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 0.00014814297124600637,
+      "loss": 0.7517,
+      "step": 508000
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 0.00014799321086261978,
+      "loss": 0.7614,
+      "step": 508500
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 0.00014784345047923322,
+      "loss": 0.7447,
+      "step": 509000
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 0.00014769369009584663,
+      "loss": 0.7519,
+      "step": 509500
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 0.00014754392971246005,
+      "loss": 0.7502,
+      "step": 510000
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 0.00014739416932907346,
+      "loss": 0.7417,
+      "step": 510500
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 0.00014724440894568687,
+      "loss": 0.7281,
+      "step": 511000
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 0.0001470946485623003,
+      "loss": 0.7512,
+      "step": 511500
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 0.00014694488817891372,
+      "loss": 0.7534,
+      "step": 512000
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 0.00014679512779552714,
+      "loss": 0.7424,
+      "step": 512500
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 0.00014664536741214058,
+      "loss": 0.7669,
+      "step": 513000
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 0.000146495607028754,
+      "loss": 0.7618,
+      "step": 513500
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 0.0001463458466453674,
+      "loss": 0.7452,
+      "step": 514000
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 0.00014619608626198082,
+      "loss": 0.7585,
+      "step": 514500
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 0.00014604632587859423,
+      "loss": 0.7386,
+      "step": 515000
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 0.00014589656549520767,
+      "loss": 0.7296,
+      "step": 515500
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 0.00014574680511182108,
+      "loss": 0.7617,
+      "step": 516000
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 0.0001455970447284345,
+      "loss": 0.737,
+      "step": 516500
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 0.0001454472843450479,
+      "loss": 0.7617,
+      "step": 517000
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 0.00014529752396166132,
+      "loss": 0.7592,
+      "step": 517500
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 0.00014514776357827476,
+      "loss": 0.7461,
+      "step": 518000
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 0.00014499800319488817,
+      "loss": 0.7703,
+      "step": 518500
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 0.00014484824281150158,
+      "loss": 0.7502,
+      "step": 519000
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 0.000144698482428115,
+      "loss": 0.7794,
+      "step": 519500
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 0.0001445487220447284,
+      "loss": 0.7628,
+      "step": 520000
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 0.00014439896166134182,
+      "loss": 0.7406,
+      "step": 520500
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 0.00014424920127795526,
+      "loss": 0.7603,
+      "step": 521000
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 0.00014409944089456867,
+      "loss": 0.7726,
+      "step": 521500
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 0.00014394968051118211,
+      "loss": 0.7609,
+      "step": 522000
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 0.00014379992012779553,
+      "loss": 0.7569,
+      "step": 522500
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 0.00014365015974440894,
+      "loss": 0.7617,
+      "step": 523000
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 0.00014350039936102235,
+      "loss": 0.7659,
+      "step": 523500
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 0.00014335063897763576,
+      "loss": 0.7501,
+      "step": 524000
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 0.0001432008785942492,
+      "loss": 0.7756,
+      "step": 524500
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 0.00014305111821086262,
+      "loss": 0.7639,
+      "step": 525000
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 0.00014290135782747603,
+      "loss": 0.7649,
+      "step": 525500
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 0.00014275159744408944,
+      "loss": 0.7533,
+      "step": 526000
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 0.00014260183706070286,
+      "loss": 0.7705,
+      "step": 526500
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 0.00014245207667731627,
+      "loss": 0.7488,
+      "step": 527000
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 0.0001423023162939297,
+      "loss": 0.76,
+      "step": 527500
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 0.00014215255591054312,
+      "loss": 0.7538,
+      "step": 528000
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 0.00014200279552715653,
+      "loss": 0.7447,
+      "step": 528500
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 0.00014185303514376995,
+      "loss": 0.7618,
+      "step": 529000
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 0.00014170327476038336,
+      "loss": 0.746,
+      "step": 529500
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 0.0001415535143769968,
+      "loss": 0.743,
+      "step": 530000
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 0.0001414037539936102,
+      "loss": 0.7609,
+      "step": 530500
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 0.00014125399361022362,
+      "loss": 0.7626,
+      "step": 531000
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 0.00014110423322683706,
+      "loss": 0.7642,
+      "step": 531500
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 0.00014095447284345048,
+      "loss": 0.7704,
+      "step": 532000
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 0.0001408047124600639,
+      "loss": 0.7812,
+      "step": 532500
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 0.0001406549520766773,
+      "loss": 0.7636,
+      "step": 533000
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 0.00014050519169329071,
+      "loss": 0.7507,
+      "step": 533500
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 0.00014035543130990415,
+      "loss": 0.7543,
+      "step": 534000
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 0.00014020567092651757,
+      "loss": 0.7434,
+      "step": 534500
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 0.00014005591054313098,
+      "loss": 0.7545,
+      "step": 535000
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 0.0001399061501597444,
+      "loss": 0.7298,
+      "step": 535500
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 0.0001397563897763578,
+      "loss": 0.7424,
+      "step": 536000
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 0.00013960662939297124,
+      "loss": 0.7235,
+      "step": 536500
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 0.00013945686900958466,
+      "loss": 0.7383,
+      "step": 537000
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 0.00013930710862619807,
+      "loss": 0.7146,
+      "step": 537500
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 0.00013915734824281148,
+      "loss": 0.7321,
+      "step": 538000
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 0.0001390075878594249,
+      "loss": 0.7193,
+      "step": 538500
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 0.0001388578274760383,
+      "loss": 0.7208,
+      "step": 539000
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 0.00013870806709265175,
+      "loss": 0.723,
+      "step": 539500
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 0.00013855830670926516,
+      "loss": 0.7167,
+      "step": 540000
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 0.0001384085463258786,
+      "loss": 0.7197,
+      "step": 540500
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 0.000138258785942492,
+      "loss": 0.7433,
+      "step": 541000
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 0.00013810902555910543,
+      "loss": 0.7213,
+      "step": 541500
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 0.00013795926517571884,
+      "loss": 0.7064,
+      "step": 542000
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 0.00013780950479233225,
+      "loss": 0.7291,
+      "step": 542500
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 0.00013765974440894566,
+      "loss": 0.7418,
+      "step": 543000
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 0.0001375099840255591,
+      "loss": 0.7328,
+      "step": 543500
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 0.00013736022364217252,
+      "loss": 0.7027,
+      "step": 544000
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 0.00013721046325878593,
+      "loss": 0.7325,
+      "step": 544500
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 0.00013706070287539934,
+      "loss": 0.7201,
+      "step": 545000
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 0.00013691094249201275,
+      "loss": 0.6999,
+      "step": 545500
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 0.0001367611821086262,
+      "loss": 0.7339,
+      "step": 546000
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 0.0001366114217252396,
+      "loss": 0.7281,
+      "step": 546500
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 0.00013646166134185302,
+      "loss": 0.7337,
+      "step": 547000
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 0.00013631190095846643,
+      "loss": 0.727,
+      "step": 547500
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 0.00013616214057507984,
+      "loss": 0.721,
+      "step": 548000
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 0.00013601238019169328,
+      "loss": 0.7241,
+      "step": 548500
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 0.0001358626198083067,
+      "loss": 0.7328,
+      "step": 549000
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 0.0001357128594249201,
+      "loss": 0.7358,
+      "step": 549500
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 0.00013556309904153355,
+      "loss": 0.7349,
+      "step": 550000
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 0.00013541333865814696,
+      "loss": 0.7249,
+      "step": 550500
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 0.00013526357827476037,
+      "loss": 0.7335,
+      "step": 551000
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 0.0001351138178913738,
+      "loss": 0.7257,
+      "step": 551500
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 0.0001349640575079872,
+      "loss": 0.7244,
+      "step": 552000
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 0.00013481429712460064,
+      "loss": 0.7023,
+      "step": 552500
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 0.00013466453674121405,
+      "loss": 0.7252,
+      "step": 553000
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 0.00013451477635782747,
+      "loss": 0.7182,
+      "step": 553500
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 0.00013436501597444088,
+      "loss": 0.7225,
+      "step": 554000
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 0.0001342152555910543,
+      "loss": 0.7167,
+      "step": 554500
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 0.0001340654952076677,
+      "loss": 0.7173,
+      "step": 555000
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 0.00013391573482428114,
+      "loss": 0.7273,
+      "step": 555500
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 0.00013376597444089456,
+      "loss": 0.7334,
+      "step": 556000
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 0.00013361621405750797,
+      "loss": 0.743,
+      "step": 556500
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 0.00013346645367412138,
+      "loss": 0.7181,
+      "step": 557000
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 0.0001333166932907348,
+      "loss": 0.7413,
+      "step": 557500
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 0.00013316693290734823,
+      "loss": 0.7387,
+      "step": 558000
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 0.00013301717252396165,
+      "loss": 0.7424,
+      "step": 558500
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 0.00013286741214057509,
+      "loss": 0.7257,
+      "step": 559000
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 0.0001327176517571885,
+      "loss": 0.7216,
+      "step": 559500
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 0.0001325678913738019,
+      "loss": 0.7192,
+      "step": 560000
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 0.00013241813099041532,
+      "loss": 0.7404,
+      "step": 560500
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 0.00013226837060702874,
+      "loss": 0.7358,
+      "step": 561000
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 0.00013211861022364215,
+      "loss": 0.721,
+      "step": 561500
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 0.0001319688498402556,
+      "loss": 0.7317,
+      "step": 562000
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 0.000131819089456869,
+      "loss": 0.7365,
+      "step": 562500
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 0.00013166932907348241,
+      "loss": 0.7322,
+      "step": 563000
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 0.00013151956869009583,
+      "loss": 0.7315,
+      "step": 563500
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 0.00013136980830670924,
+      "loss": 0.7203,
+      "step": 564000
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 0.00013122004792332268,
+      "loss": 0.7319,
+      "step": 564500
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 0.0001310702875399361,
+      "loss": 0.7267,
+      "step": 565000
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 0.0001309205271565495,
+      "loss": 0.7434,
+      "step": 565500
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 0.00013077076677316292,
+      "loss": 0.7392,
+      "step": 566000
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 0.00013062100638977633,
+      "loss": 0.7398,
+      "step": 566500
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 0.00013047124600638977,
+      "loss": 0.7369,
+      "step": 567000
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 0.00013032148562300318,
+      "loss": 0.713,
+      "step": 567500
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 0.0001301717252396166,
+      "loss": 0.7305,
+      "step": 568000
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 0.00013002196485623004,
+      "loss": 0.7188,
+      "step": 568500
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 0.00012987220447284345,
+      "loss": 0.6891,
+      "step": 569000
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 0.00012972244408945686,
+      "loss": 0.6791,
+      "step": 569500
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 0.00012957268370607027,
+      "loss": 0.6889,
+      "step": 570000
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 0.00012942292332268369,
+      "loss": 0.7055,
+      "step": 570500
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 0.00012927316293929713,
+      "loss": 0.7004,
+      "step": 571000
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 0.00012912340255591054,
+      "loss": 0.7069,
+      "step": 571500
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 0.00012897364217252395,
+      "loss": 0.6956,
+      "step": 572000
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 0.00012882388178913736,
+      "loss": 0.707,
+      "step": 572500
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 0.00012867412140575078,
+      "loss": 0.7038,
+      "step": 573000
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 0.0001285243610223642,
+      "loss": 0.6913,
+      "step": 573500
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 0.00012837460063897763,
+      "loss": 0.7031,
+      "step": 574000
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 0.00012822484025559104,
+      "loss": 0.6875,
+      "step": 574500
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 0.00012807507987220445,
+      "loss": 0.6997,
+      "step": 575000
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 0.00012792531948881787,
+      "loss": 0.6969,
+      "step": 575500
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 0.00012777555910543128,
+      "loss": 0.6893,
+      "step": 576000
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 0.00012762579872204472,
+      "loss": 0.7044,
+      "step": 576500
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 0.00012747603833865813,
+      "loss": 0.6771,
+      "step": 577000
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 0.00012732627795527157,
+      "loss": 0.6982,
+      "step": 577500
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 0.00012717651757188498,
+      "loss": 0.6927,
+      "step": 578000
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 0.0001270267571884984,
+      "loss": 0.6778,
+      "step": 578500
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 0.0001268769968051118,
+      "loss": 0.7042,
+      "step": 579000
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 0.00012672723642172522,
+      "loss": 0.6952,
+      "step": 579500
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 0.00012657747603833864,
+      "loss": 0.7018,
+      "step": 580000
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 0.00012642771565495208,
+      "loss": 0.6961,
+      "step": 580500
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 0.0001262779552715655,
+      "loss": 0.6976,
+      "step": 581000
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 0.0001261281948881789,
+      "loss": 0.7028,
+      "step": 581500
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 0.0001259784345047923,
+      "loss": 0.6939,
+      "step": 582000
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 0.00012582867412140573,
+      "loss": 0.7031,
+      "step": 582500
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 0.00012567891373801917,
+      "loss": 0.7029,
+      "step": 583000
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 0.00012552915335463258,
+      "loss": 0.6896,
+      "step": 583500
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 0.000125379392971246,
+      "loss": 0.6986,
+      "step": 584000
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 0.0001252296325878594,
+      "loss": 0.6988,
+      "step": 584500
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 0.00012507987220447282,
+      "loss": 0.6927,
+      "step": 585000
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 0.00012493011182108623,
+      "loss": 0.6938,
+      "step": 585500
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 0.00012478035143769967,
+      "loss": 0.6858,
+      "step": 586000
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 0.00012463059105431308,
+      "loss": 0.708,
+      "step": 586500
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 0.00012448083067092652,
+      "loss": 0.6989,
+      "step": 587000
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 0.00012433107028753993,
+      "loss": 0.7048,
+      "step": 587500
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 0.00012418130990415335,
+      "loss": 0.6934,
+      "step": 588000
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 0.00012403154952076676,
+      "loss": 0.6995,
+      "step": 588500
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 0.00012388178913738017,
+      "loss": 0.7221,
+      "step": 589000
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 0.0001237320287539936,
+      "loss": 0.685,
+      "step": 589500
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 0.00012358226837060702,
+      "loss": 0.7036,
+      "step": 590000
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 0.00012343250798722044,
+      "loss": 0.703,
+      "step": 590500
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 0.00012328274760383385,
+      "loss": 0.7053,
+      "step": 591000
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 0.00012313298722044726,
+      "loss": 0.7038,
+      "step": 591500
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 0.00012298322683706068,
+      "loss": 0.709,
+      "step": 592000
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 0.00012283346645367412,
+      "loss": 0.7197,
+      "step": 592500
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 0.00012268370607028753,
+      "loss": 0.6999,
+      "step": 593000
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 0.00012253394568690094,
+      "loss": 0.6781,
+      "step": 593500
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 0.00012238418530351435,
+      "loss": 0.7092,
+      "step": 594000
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 0.00012223442492012777,
+      "loss": 0.7128,
+      "step": 594500
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 0.0001220846645367412,
+      "loss": 0.7028,
+      "step": 595000
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 0.00012193490415335463,
+      "loss": 0.6816,
+      "step": 595500
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 0.00012178514376996804,
+      "loss": 0.6977,
+      "step": 596000
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 0.00012163538338658147,
+      "loss": 0.686,
+      "step": 596500
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 0.00012148562300319488,
+      "loss": 0.7018,
+      "step": 597000
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 0.0001213358626198083,
+      "loss": 0.706,
+      "step": 597500
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 0.00012118610223642172,
+      "loss": 0.7007,
+      "step": 598000
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 0.00012103634185303514,
+      "loss": 0.7209,
+      "step": 598500
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 0.00012088658146964855,
+      "loss": 0.6842,
+      "step": 599000
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 0.00012073682108626197,
+      "loss": 0.7254,
+      "step": 599500
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 0.00012058706070287539,
+      "loss": 0.7002,
+      "step": 600000
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 0.0001204373003194888,
+      "loss": 0.7104,
+      "step": 600500
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 0.00012028753993610223,
+      "loss": 0.6904,
+      "step": 601000
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 0.00012013777955271564,
+      "loss": 0.701,
+      "step": 601500
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 0.00011998801916932906,
+      "loss": 0.6627,
+      "step": 602000
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 0.00011983825878594248,
+      "loss": 0.6556,
+      "step": 602500
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 0.00011968849840255589,
+      "loss": 0.6574,
+      "step": 603000
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 0.00011953873801916932,
+      "loss": 0.6614,
+      "step": 603500
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 0.00011938897763578274,
+      "loss": 0.6708,
+      "step": 604000
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 0.00011923921725239617,
+      "loss": 0.6599,
+      "step": 604500
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 0.00011908945686900958,
+      "loss": 0.6544,
+      "step": 605000
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 0.000118939696485623,
+      "loss": 0.6669,
+      "step": 605500
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 0.00011878993610223642,
+      "loss": 0.6554,
+      "step": 606000
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 0.00011864017571884983,
+      "loss": 0.6605,
+      "step": 606500
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 0.00011849041533546325,
+      "loss": 0.6642,
+      "step": 607000
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 0.00011834065495207667,
+      "loss": 0.6842,
+      "step": 607500
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 0.00011819089456869008,
+      "loss": 0.642,
+      "step": 608000
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 0.00011804113418530351,
+      "loss": 0.6834,
+      "step": 608500
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 0.00011789137380191692,
+      "loss": 0.6699,
+      "step": 609000
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 0.00011774161341853034,
+      "loss": 0.6668,
+      "step": 609500
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 0.00011759185303514376,
+      "loss": 0.6582,
+      "step": 610000
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 0.00011744209265175718,
+      "loss": 0.6706,
+      "step": 610500
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 0.00011729233226837059,
+      "loss": 0.6657,
+      "step": 611000
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 0.00011714257188498401,
+      "loss": 0.6659,
+      "step": 611500
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 0.00011699281150159743,
+      "loss": 0.6756,
+      "step": 612000
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 0.00011684305111821084,
+      "loss": 0.6782,
+      "step": 612500
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 0.00011669329073482427,
+      "loss": 0.6717,
+      "step": 613000
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 0.00011654353035143769,
+      "loss": 0.6667,
+      "step": 613500
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 0.00011639376996805112,
+      "loss": 0.676,
+      "step": 614000
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 0.00011624400958466453,
+      "loss": 0.6645,
+      "step": 614500
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 0.00011609424920127794,
+      "loss": 0.6581,
+      "step": 615000
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 0.00011594448881789137,
+      "loss": 0.6725,
+      "step": 615500
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 0.00011579472843450478,
+      "loss": 0.6819,
+      "step": 616000
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 0.00011564496805111821,
+      "loss": 0.6817,
+      "step": 616500
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 0.00011549520766773162,
+      "loss": 0.675,
+      "step": 617000
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 0.00011534544728434503,
+      "loss": 0.6622,
+      "step": 617500
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 0.00011519568690095846,
+      "loss": 0.6671,
+      "step": 618000
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 0.00011504592651757187,
+      "loss": 0.6644,
+      "step": 618500
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 0.00011489616613418529,
+      "loss": 0.6672,
+      "step": 619000
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 0.00011474640575079871,
+      "loss": 0.6704,
+      "step": 619500
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 0.00011459664536741212,
+      "loss": 0.6843,
+      "step": 620000
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 0.00011444688498402555,
+      "loss": 0.6625,
+      "step": 620500
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 0.00011429712460063896,
+      "loss": 0.6738,
+      "step": 621000
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 0.00011414736421725238,
+      "loss": 0.6707,
+      "step": 621500
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 0.0001139976038338658,
+      "loss": 0.6902,
+      "step": 622000
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 0.00011384784345047923,
+      "loss": 0.6878,
+      "step": 622500
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 0.00011369808306709265,
+      "loss": 0.6807,
+      "step": 623000
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 0.00011354832268370607,
+      "loss": 0.6625,
+      "step": 623500
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 0.00011339856230031948,
+      "loss": 0.6482,
+      "step": 624000
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 0.0001132488019169329,
+      "loss": 0.6695,
+      "step": 624500
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 0.00011309904153354632,
+      "loss": 0.6884,
+      "step": 625000
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 0.00011294928115015973,
+      "loss": 0.6723,
+      "step": 625500
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 0.00011279952076677316,
+      "loss": 0.6747,
+      "step": 626000
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 0.00011264976038338657,
+      "loss": 0.6717,
+      "step": 626500
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 0.0001125,
+      "loss": 0.6875,
+      "step": 627000
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 0.00011235023961661341,
+      "loss": 0.678,
+      "step": 627500
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 0.00011220047923322682,
+      "loss": 0.6767,
+      "step": 628000
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 0.00011205071884984025,
+      "loss": 0.6751,
+      "step": 628500
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 0.00011190095846645366,
+      "loss": 0.6791,
+      "step": 629000
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 0.00011175119808306707,
+      "loss": 0.6752,
+      "step": 629500
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 0.0001116014376996805,
+      "loss": 0.672,
+      "step": 630000
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 0.00011145167731629391,
+      "loss": 0.689,
+      "step": 630500
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 0.00011130191693290733,
+      "loss": 0.6855,
+      "step": 631000
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 0.00011115215654952075,
+      "loss": 0.6821,
+      "step": 631500
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 0.00011100239616613418,
+      "loss": 0.6854,
+      "step": 632000
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 0.0001108526357827476,
+      "loss": 0.7044,
+      "step": 632500
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 0.00011070287539936102,
+      "loss": 0.67,
+      "step": 633000
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 0.00011055311501597443,
+      "loss": 0.6718,
+      "step": 633500
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 0.00011040335463258786,
+      "loss": 0.6851,
+      "step": 634000
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 0.00011025359424920127,
+      "loss": 0.6841,
+      "step": 634500
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 0.0001101038338658147,
+      "loss": 0.6732,
+      "step": 635000
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 0.00010995407348242811,
+      "loss": 0.6391,
+      "step": 635500
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 0.00010980431309904152,
+      "loss": 0.6357,
+      "step": 636000
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 0.00010965455271565495,
+      "loss": 0.6554,
+      "step": 636500
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 0.00010950479233226836,
+      "loss": 0.6347,
+      "step": 637000
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 0.00010935503194888177,
+      "loss": 0.6331,
+      "step": 637500
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 0.0001092052715654952,
+      "loss": 0.641,
+      "step": 638000
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 0.00010905551118210861,
+      "loss": 0.6286,
+      "step": 638500
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 0.00010890575079872204,
+      "loss": 0.6428,
+      "step": 639000
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 0.00010875599041533545,
+      "loss": 0.6394,
+      "step": 639500
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 0.00010860623003194886,
+      "loss": 0.6444,
+      "step": 640000
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 0.00010845646964856229,
+      "loss": 0.6471,
+      "step": 640500
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 0.00010830670926517571,
+      "loss": 0.6544,
+      "step": 641000
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 0.00010815694888178914,
+      "loss": 0.6425,
+      "step": 641500
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 0.00010800718849840255,
+      "loss": 0.6436,
+      "step": 642000
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 0.00010785742811501597,
+      "loss": 0.6475,
+      "step": 642500
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 0.00010770766773162939,
+      "loss": 0.644,
+      "step": 643000
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 0.0001075579073482428,
+      "loss": 0.6394,
+      "step": 643500
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 0.00010740814696485622,
+      "loss": 0.6365,
+      "step": 644000
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 0.00010725838658146964,
+      "loss": 0.6381,
+      "step": 644500
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 0.00010710862619808306,
+      "loss": 0.6429,
+      "step": 645000
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 0.00010695886581469647,
+      "loss": 0.6517,
+      "step": 645500
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 0.0001068091054313099,
+      "loss": 0.6415,
+      "step": 646000
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 0.00010665934504792331,
+      "loss": 0.6512,
+      "step": 646500
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 0.00010650958466453673,
+      "loss": 0.6405,
+      "step": 647000
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 0.00010635982428115015,
+      "loss": 0.653,
+      "step": 647500
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 0.00010621006389776356,
+      "loss": 0.6442,
+      "step": 648000
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 0.00010606030351437699,
+      "loss": 0.6403,
+      "step": 648500
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 0.0001059105431309904,
+      "loss": 0.658,
+      "step": 649000
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 0.00010576078274760381,
+      "loss": 0.6369,
+      "step": 649500
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 0.00010561102236421724,
+      "loss": 0.6631,
+      "step": 650000
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 0.00010546126198083066,
+      "loss": 0.6481,
+      "step": 650500
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 0.00010531150159744409,
+      "loss": 0.6662,
+      "step": 651000
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 0.0001051617412140575,
+      "loss": 0.6534,
+      "step": 651500
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 0.00010501198083067092,
+      "loss": 0.6626,
+      "step": 652000
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 0.00010486222044728434,
+      "loss": 0.6583,
+      "step": 652500
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 0.00010471246006389775,
+      "loss": 0.6543,
+      "step": 653000
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 0.00010456269968051118,
+      "loss": 0.6501,
+      "step": 653500
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 0.0001044129392971246,
+      "loss": 0.6396,
+      "step": 654000
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 0.000104263178913738,
+      "loss": 0.6523,
+      "step": 654500
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 0.00010411341853035143,
+      "loss": 0.6461,
+      "step": 655000
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 0.00010396365814696485,
+      "loss": 0.6575,
+      "step": 655500
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 0.00010381389776357826,
+      "loss": 0.652,
+      "step": 656000
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 0.00010366413738019168,
+      "loss": 0.6514,
+      "step": 656500
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 0.0001035143769968051,
+      "loss": 0.6585,
+      "step": 657000
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 0.00010336461661341851,
+      "loss": 0.6628,
+      "step": 657500
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 0.00010321485623003194,
+      "loss": 0.6346,
+      "step": 658000
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 0.00010306509584664535,
+      "loss": 0.6607,
+      "step": 658500
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 0.00010291533546325877,
+      "loss": 0.6544,
+      "step": 659000
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 0.0001027655750798722,
+      "loss": 0.6641,
+      "step": 659500
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 0.00010261581469648563,
+      "loss": 0.6381,
+      "step": 660000
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 0.00010246605431309904,
+      "loss": 0.6571,
+      "step": 660500
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 0.00010231629392971245,
+      "loss": 0.6525,
+      "step": 661000
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 0.00010216653354632588,
+      "loss": 0.6363,
+      "step": 661500
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 0.00010201677316293929,
+      "loss": 0.6473,
+      "step": 662000
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 0.0001018670127795527,
+      "loss": 0.6518,
+      "step": 662500
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 0.00010171725239616613,
+      "loss": 0.6555,
+      "step": 663000
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 0.00010156749201277954,
+      "loss": 0.6586,
+      "step": 663500
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 0.00010141773162939296,
+      "loss": 0.6328,
+      "step": 664000
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 0.00010126797124600638,
+      "loss": 0.6404,
+      "step": 664500
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 0.0001011182108626198,
+      "loss": 0.6557,
+      "step": 665000
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 0.00010096845047923322,
+      "loss": 0.6517,
+      "step": 665500
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 0.00010081869009584663,
+      "loss": 0.665,
+      "step": 666000
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 0.00010066892971246005,
+      "loss": 0.6451,
+      "step": 666500
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 0.00010051916932907347,
+      "loss": 0.6385,
+      "step": 667000
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 0.00010036940894568688,
+      "loss": 0.6343,
+      "step": 667500
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 0.0001002196485623003,
+      "loss": 0.6499,
+      "step": 668000
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 0.00010006988817891374,
+      "loss": 0.6356,
+      "step": 668500
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 9.992012779552715e-05,
+      "loss": 0.6135,
+      "step": 669000
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 9.977036741214058e-05,
+      "loss": 0.623,
+      "step": 669500
+    },
+    {
+      "epoch": 20.05,
+      "learning_rate": 9.962060702875399e-05,
+      "loss": 0.6182,
+      "step": 670000
+    },
+    {
+      "epoch": 20.06,
+      "learning_rate": 9.94708466453674e-05,
+      "loss": 0.6088,
+      "step": 670500
+    },
+    {
+      "epoch": 20.08,
+      "learning_rate": 9.932108626198083e-05,
+      "loss": 0.6177,
+      "step": 671000
+    },
+    {
+      "epoch": 20.09,
+      "learning_rate": 9.917132587859424e-05,
+      "loss": 0.6235,
+      "step": 671500
+    },
+    {
+      "epoch": 20.11,
+      "learning_rate": 9.902156549520767e-05,
+      "loss": 0.6101,
+      "step": 672000
+    },
+    {
+      "epoch": 20.12,
+      "learning_rate": 9.887180511182108e-05,
+      "loss": 0.6066,
+      "step": 672500
+    },
+    {
+      "epoch": 20.14,
+      "learning_rate": 9.872204472843449e-05,
+      "loss": 0.6111,
+      "step": 673000
+    },
+    {
+      "epoch": 20.15,
+      "learning_rate": 9.857228434504792e-05,
+      "loss": 0.6285,
+      "step": 673500
+    },
+    {
+      "epoch": 20.17,
+      "learning_rate": 9.842252396166133e-05,
+      "loss": 0.6286,
+      "step": 674000
+    },
+    {
+      "epoch": 20.18,
+      "learning_rate": 9.827276357827474e-05,
+      "loss": 0.6114,
+      "step": 674500
+    },
+    {
+      "epoch": 20.2,
+      "learning_rate": 9.812300319488817e-05,
+      "loss": 0.6379,
+      "step": 675000
+    },
+    {
+      "epoch": 20.21,
+      "learning_rate": 9.797324281150158e-05,
+      "loss": 0.6178,
+      "step": 675500
+    },
+    {
+      "epoch": 20.23,
+      "learning_rate": 9.7823482428115e-05,
+      "loss": 0.6289,
+      "step": 676000
+    },
+    {
+      "epoch": 20.24,
+      "learning_rate": 9.767372204472842e-05,
+      "loss": 0.6276,
+      "step": 676500
+    },
+    {
+      "epoch": 20.26,
+      "learning_rate": 9.752396166134183e-05,
+      "loss": 0.619,
+      "step": 677000
+    },
+    {
+      "epoch": 20.27,
+      "learning_rate": 9.737420127795526e-05,
+      "loss": 0.6073,
+      "step": 677500
+    },
+    {
+      "epoch": 20.29,
+      "learning_rate": 9.722444089456869e-05,
+      "loss": 0.6175,
+      "step": 678000
+    },
+    {
+      "epoch": 20.3,
+      "learning_rate": 9.707468051118211e-05,
+      "loss": 0.6318,
+      "step": 678500
+    },
+    {
+      "epoch": 20.32,
+      "learning_rate": 9.692492012779553e-05,
+      "loss": 0.6148,
+      "step": 679000
+    },
+    {
+      "epoch": 20.33,
+      "learning_rate": 9.677515974440894e-05,
+      "loss": 0.6151,
+      "step": 679500
+    },
+    {
+      "epoch": 20.35,
+      "learning_rate": 9.662539936102236e-05,
+      "loss": 0.6328,
+      "step": 680000
+    },
+    {
+      "epoch": 20.36,
+      "learning_rate": 9.647563897763578e-05,
+      "loss": 0.6228,
+      "step": 680500
+    },
+    {
+      "epoch": 20.38,
+      "learning_rate": 9.632587859424919e-05,
+      "loss": 0.6218,
+      "step": 681000
+    },
+    {
+      "epoch": 20.39,
+      "learning_rate": 9.617611821086262e-05,
+      "loss": 0.633,
+      "step": 681500
+    },
+    {
+      "epoch": 20.41,
+      "learning_rate": 9.602635782747603e-05,
+      "loss": 0.6038,
+      "step": 682000
+    },
+    {
+      "epoch": 20.42,
+      "learning_rate": 9.587659744408944e-05,
+      "loss": 0.5946,
+      "step": 682500
+    },
+    {
+      "epoch": 20.44,
+      "learning_rate": 9.572683706070287e-05,
+      "loss": 0.6091,
+      "step": 683000
+    },
+    {
+      "epoch": 20.45,
+      "learning_rate": 9.557707667731628e-05,
+      "loss": 0.6293,
+      "step": 683500
+    },
+    {
+      "epoch": 20.47,
+      "learning_rate": 9.54273162939297e-05,
+      "loss": 0.6308,
+      "step": 684000
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 9.527755591054312e-05,
+      "loss": 0.6329,
+      "step": 684500
+    },
+    {
+      "epoch": 20.5,
+      "learning_rate": 9.512779552715653e-05,
+      "loss": 0.6182,
+      "step": 685000
+    },
+    {
+      "epoch": 20.51,
+      "learning_rate": 9.497803514376996e-05,
+      "loss": 0.62,
+      "step": 685500
+    },
+    {
+      "epoch": 20.53,
+      "learning_rate": 9.482827476038337e-05,
+      "loss": 0.6212,
+      "step": 686000
+    },
+    {
+      "epoch": 20.54,
+      "learning_rate": 9.467851437699678e-05,
+      "loss": 0.6291,
+      "step": 686500
+    },
+    {
+      "epoch": 20.56,
+      "learning_rate": 9.452875399361022e-05,
+      "loss": 0.6317,
+      "step": 687000
+    },
+    {
+      "epoch": 20.57,
+      "learning_rate": 9.437899361022364e-05,
+      "loss": 0.6198,
+      "step": 687500
+    },
+    {
+      "epoch": 20.59,
+      "learning_rate": 9.422923322683706e-05,
+      "loss": 0.6327,
+      "step": 688000
+    },
+    {
+      "epoch": 20.6,
+      "learning_rate": 9.407947284345048e-05,
+      "loss": 0.6228,
+      "step": 688500
+    },
+    {
+      "epoch": 20.62,
+      "learning_rate": 9.392971246006389e-05,
+      "loss": 0.6258,
+      "step": 689000
+    },
+    {
+      "epoch": 20.63,
+      "learning_rate": 9.377995207667731e-05,
+      "loss": 0.6172,
+      "step": 689500
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 9.363019169329073e-05,
+      "loss": 0.6056,
+      "step": 690000
+    },
+    {
+      "epoch": 20.66,
+      "learning_rate": 9.348043130990415e-05,
+      "loss": 0.6214,
+      "step": 690500
+    },
+    {
+      "epoch": 20.68,
+      "learning_rate": 9.333067092651757e-05,
+      "loss": 0.6188,
+      "step": 691000
+    },
+    {
+      "epoch": 20.69,
+      "learning_rate": 9.318091054313098e-05,
+      "loss": 0.6257,
+      "step": 691500
+    },
+    {
+      "epoch": 20.71,
+      "learning_rate": 9.30311501597444e-05,
+      "loss": 0.6379,
+      "step": 692000
+    },
+    {
+      "epoch": 20.72,
+      "learning_rate": 9.288138977635782e-05,
+      "loss": 0.6226,
+      "step": 692500
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 9.273162939297123e-05,
+      "loss": 0.6286,
+      "step": 693000
+    },
+    {
+      "epoch": 20.75,
+      "learning_rate": 9.258186900958466e-05,
+      "loss": 0.638,
+      "step": 693500
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 9.243210862619807e-05,
+      "loss": 0.619,
+      "step": 694000
+    },
+    {
+      "epoch": 20.78,
+      "learning_rate": 9.228234824281148e-05,
+      "loss": 0.6058,
+      "step": 694500
+    },
+    {
+      "epoch": 20.8,
+      "learning_rate": 9.213258785942491e-05,
+      "loss": 0.6197,
+      "step": 695000
+    },
+    {
+      "epoch": 20.81,
+      "learning_rate": 9.198282747603832e-05,
+      "loss": 0.6173,
+      "step": 695500
+    },
+    {
+      "epoch": 20.83,
+      "learning_rate": 9.183306709265175e-05,
+      "loss": 0.6072,
+      "step": 696000
+    },
+    {
+      "epoch": 20.84,
+      "learning_rate": 9.168330670926517e-05,
+      "loss": 0.6319,
+      "step": 696500
+    },
+    {
+      "epoch": 20.86,
+      "learning_rate": 9.15335463258786e-05,
+      "loss": 0.6163,
+      "step": 697000
+    },
+    {
+      "epoch": 20.87,
+      "learning_rate": 9.138378594249201e-05,
+      "loss": 0.6363,
+      "step": 697500
+    },
+    {
+      "epoch": 20.89,
+      "learning_rate": 9.123402555910542e-05,
+      "loss": 0.6149,
+      "step": 698000
+    },
+    {
+      "epoch": 20.9,
+      "learning_rate": 9.108426517571885e-05,
+      "loss": 0.6409,
+      "step": 698500
+    },
+    {
+      "epoch": 20.92,
+      "learning_rate": 9.093450479233226e-05,
+      "loss": 0.639,
+      "step": 699000
+    },
+    {
+      "epoch": 20.93,
+      "learning_rate": 9.078474440894568e-05,
+      "loss": 0.6178,
+      "step": 699500
+    },
+    {
+      "epoch": 20.95,
+      "learning_rate": 9.06349840255591e-05,
+      "loss": 0.6145,
+      "step": 700000
+    },
+    {
+      "epoch": 20.96,
+      "learning_rate": 9.048522364217251e-05,
+      "loss": 0.6243,
+      "step": 700500
+    },
+    {
+      "epoch": 20.98,
+      "learning_rate": 9.033546325878593e-05,
+      "loss": 0.6367,
+      "step": 701000
+    },
+    {
+      "epoch": 20.99,
+      "learning_rate": 9.018570287539935e-05,
+      "loss": 0.6263,
+      "step": 701500
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 9.003594249201277e-05,
+      "loss": 0.634,
+      "step": 702000
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 8.988618210862619e-05,
+      "loss": 0.5902,
+      "step": 702500
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 8.97364217252396e-05,
+      "loss": 0.5895,
+      "step": 703000
+    },
+    {
+      "epoch": 21.05,
+      "learning_rate": 8.958666134185302e-05,
+      "loss": 0.5957,
+      "step": 703500
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 8.943690095846644e-05,
+      "loss": 0.6017,
+      "step": 704000
+    },
+    {
+      "epoch": 21.08,
+      "learning_rate": 8.928714057507986e-05,
+      "loss": 0.5856,
+      "step": 704500
+    },
+    {
+      "epoch": 21.09,
+      "learning_rate": 8.913738019169327e-05,
+      "loss": 0.5786,
+      "step": 705000
+    },
+    {
+      "epoch": 21.11,
+      "learning_rate": 8.898761980830671e-05,
+      "loss": 0.5863,
+      "step": 705500
+    },
+    {
+      "epoch": 21.12,
+      "learning_rate": 8.883785942492012e-05,
+      "loss": 0.5995,
+      "step": 706000
+    },
+    {
+      "epoch": 21.14,
+      "learning_rate": 8.868809904153355e-05,
+      "loss": 0.5961,
+      "step": 706500
+    },
+    {
+      "epoch": 21.15,
+      "learning_rate": 8.853833865814696e-05,
+      "loss": 0.6043,
+      "step": 707000
+    },
+    {
+      "epoch": 21.17,
+      "learning_rate": 8.838857827476037e-05,
+      "loss": 0.5834,
+      "step": 707500
+    },
+    {
+      "epoch": 21.18,
+      "learning_rate": 8.82388178913738e-05,
+      "loss": 0.5949,
+      "step": 708000
+    },
+    {
+      "epoch": 21.2,
+      "learning_rate": 8.808905750798721e-05,
+      "loss": 0.601,
+      "step": 708500
+    },
+    {
+      "epoch": 21.21,
+      "learning_rate": 8.793929712460064e-05,
+      "loss": 0.5851,
+      "step": 709000
+    },
+    {
+      "epoch": 21.23,
+      "learning_rate": 8.778953674121405e-05,
+      "loss": 0.586,
+      "step": 709500
+    },
+    {
+      "epoch": 21.24,
+      "learning_rate": 8.763977635782746e-05,
+      "loss": 0.6063,
+      "step": 710000
+    },
+    {
+      "epoch": 21.26,
+      "learning_rate": 8.749001597444089e-05,
+      "loss": 0.6036,
+      "step": 710500
+    },
+    {
+      "epoch": 21.27,
+      "learning_rate": 8.73402555910543e-05,
+      "loss": 0.6005,
+      "step": 711000
+    },
+    {
+      "epoch": 21.29,
+      "learning_rate": 8.719049520766772e-05,
+      "loss": 0.5963,
+      "step": 711500
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 8.704073482428114e-05,
+      "loss": 0.5883,
+      "step": 712000
+    },
+    {
+      "epoch": 21.32,
+      "learning_rate": 8.689097444089455e-05,
+      "loss": 0.6098,
+      "step": 712500
+    },
+    {
+      "epoch": 21.33,
+      "learning_rate": 8.674121405750797e-05,
+      "loss": 0.5819,
+      "step": 713000
+    },
+    {
+      "epoch": 21.35,
+      "learning_rate": 8.65914536741214e-05,
+      "loss": 0.6014,
+      "step": 713500
+    },
+    {
+      "epoch": 21.36,
+      "learning_rate": 8.64416932907348e-05,
+      "loss": 0.5902,
+      "step": 714000
+    },
+    {
+      "epoch": 21.38,
+      "learning_rate": 8.629193290734825e-05,
+      "loss": 0.5958,
+      "step": 714500
+    },
+    {
+      "epoch": 21.39,
+      "learning_rate": 8.614217252396166e-05,
+      "loss": 0.5962,
+      "step": 715000
+    },
+    {
+      "epoch": 21.41,
+      "learning_rate": 8.599241214057509e-05,
+      "loss": 0.6028,
+      "step": 715500
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 8.58426517571885e-05,
+      "loss": 0.5855,
+      "step": 716000
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 8.569289137380191e-05,
+      "loss": 0.6091,
+      "step": 716500
+    },
+    {
+      "epoch": 21.45,
+      "learning_rate": 8.554313099041534e-05,
+      "loss": 0.5944,
+      "step": 717000
+    },
+    {
+      "epoch": 21.47,
+      "learning_rate": 8.539337060702875e-05,
+      "loss": 0.5984,
+      "step": 717500
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 8.524361022364216e-05,
+      "loss": 0.5983,
+      "step": 718000
+    },
+    {
+      "epoch": 21.5,
+      "learning_rate": 8.509384984025559e-05,
+      "loss": 0.5925,
+      "step": 718500
+    },
+    {
+      "epoch": 21.51,
+      "learning_rate": 8.4944089456869e-05,
+      "loss": 0.5872,
+      "step": 719000
+    },
+    {
+      "epoch": 21.53,
+      "learning_rate": 8.479432907348241e-05,
+      "loss": 0.5903,
+      "step": 719500
+    },
+    {
+      "epoch": 21.54,
+      "learning_rate": 8.464456869009584e-05,
+      "loss": 0.5967,
+      "step": 720000
+    },
+    {
+      "epoch": 21.56,
+      "learning_rate": 8.449480830670925e-05,
+      "loss": 0.6012,
+      "step": 720500
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 8.434504792332268e-05,
+      "loss": 0.5969,
+      "step": 721000
+    },
+    {
+      "epoch": 21.59,
+      "learning_rate": 8.419528753993609e-05,
+      "loss": 0.583,
+      "step": 721500
+    },
+    {
+      "epoch": 21.6,
+      "learning_rate": 8.40455271565495e-05,
+      "loss": 0.6063,
+      "step": 722000
+    },
+    {
+      "epoch": 21.62,
+      "learning_rate": 8.389576677316293e-05,
+      "loss": 0.5981,
+      "step": 722500
+    },
+    {
+      "epoch": 21.63,
+      "learning_rate": 8.374600638977634e-05,
+      "loss": 0.6052,
+      "step": 723000
+    },
+    {
+      "epoch": 21.65,
+      "learning_rate": 8.359624600638976e-05,
+      "loss": 0.6212,
+      "step": 723500
+    },
+    {
+      "epoch": 21.66,
+      "learning_rate": 8.34464856230032e-05,
+      "loss": 0.5931,
+      "step": 724000
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 8.329672523961661e-05,
+      "loss": 0.6077,
+      "step": 724500
+    },
+    {
+      "epoch": 21.69,
+      "learning_rate": 8.314696485623003e-05,
+      "loss": 0.5955,
+      "step": 725000
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 8.299720447284345e-05,
+      "loss": 0.5914,
+      "step": 725500
+    },
+    {
+      "epoch": 21.72,
+      "learning_rate": 8.284744408945686e-05,
+      "loss": 0.5971,
+      "step": 726000
+    },
+    {
+      "epoch": 21.74,
+      "learning_rate": 8.269768370607029e-05,
+      "loss": 0.5951,
+      "step": 726500
+    },
+    {
+      "epoch": 21.75,
+      "learning_rate": 8.25479233226837e-05,
+      "loss": 0.6002,
+      "step": 727000
+    },
+    {
+      "epoch": 21.77,
+      "learning_rate": 8.239816293929713e-05,
+      "loss": 0.6096,
+      "step": 727500
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 8.224840255591054e-05,
+      "loss": 0.5933,
+      "step": 728000
+    },
+    {
+      "epoch": 21.8,
+      "learning_rate": 8.209864217252395e-05,
+      "loss": 0.5874,
+      "step": 728500
+    },
+    {
+      "epoch": 21.81,
+      "learning_rate": 8.194888178913738e-05,
+      "loss": 0.6129,
+      "step": 729000
+    },
+    {
+      "epoch": 21.83,
+      "learning_rate": 8.179912140575079e-05,
+      "loss": 0.5955,
+      "step": 729500
+    },
+    {
+      "epoch": 21.84,
+      "learning_rate": 8.16493610223642e-05,
+      "loss": 0.6011,
+      "step": 730000
+    },
+    {
+      "epoch": 21.86,
+      "learning_rate": 8.149960063897763e-05,
+      "loss": 0.6015,
+      "step": 730500
+    },
+    {
+      "epoch": 21.87,
+      "learning_rate": 8.134984025559104e-05,
+      "loss": 0.6211,
+      "step": 731000
+    },
+    {
+      "epoch": 21.89,
+      "learning_rate": 8.120007987220445e-05,
+      "loss": 0.5971,
+      "step": 731500
+    },
+    {
+      "epoch": 21.9,
+      "learning_rate": 8.105031948881788e-05,
+      "loss": 0.579,
+      "step": 732000
+    },
+    {
+      "epoch": 21.92,
+      "learning_rate": 8.090055910543129e-05,
+      "loss": 0.6153,
+      "step": 732500
+    },
+    {
+      "epoch": 21.93,
+      "learning_rate": 8.075079872204473e-05,
+      "loss": 0.6175,
+      "step": 733000
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 8.060103833865814e-05,
+      "loss": 0.6132,
+      "step": 733500
+    },
+    {
+      "epoch": 21.96,
+      "learning_rate": 8.045127795527157e-05,
+      "loss": 0.6018,
+      "step": 734000
+    },
+    {
+      "epoch": 21.98,
+      "learning_rate": 8.030151757188498e-05,
+      "loss": 0.5964,
+      "step": 734500
+    },
+    {
+      "epoch": 21.99,
+      "learning_rate": 8.01517571884984e-05,
+      "loss": 0.6181,
+      "step": 735000
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 8.000199680511182e-05,
+      "loss": 0.5697,
+      "step": 735500
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 7.985223642172524e-05,
+      "loss": 0.5621,
+      "step": 736000
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 7.970247603833865e-05,
+      "loss": 0.5808,
+      "step": 736500
+    },
+    {
+      "epoch": 22.05,
+      "learning_rate": 7.955271565495207e-05,
+      "loss": 0.5742,
+      "step": 737000
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 7.940295527156549e-05,
+      "loss": 0.5721,
+      "step": 737500
+    },
+    {
+      "epoch": 22.08,
+      "learning_rate": 7.92531948881789e-05,
+      "loss": 0.5634,
+      "step": 738000
+    },
+    {
+      "epoch": 22.1,
+      "learning_rate": 7.910343450479233e-05,
+      "loss": 0.572,
+      "step": 738500
+    },
+    {
+      "epoch": 22.11,
+      "learning_rate": 7.895367412140574e-05,
+      "loss": 0.5839,
+      "step": 739000
+    },
+    {
+      "epoch": 22.13,
+      "learning_rate": 7.880391373801916e-05,
+      "loss": 0.5761,
+      "step": 739500
+    },
+    {
+      "epoch": 22.14,
+      "learning_rate": 7.865415335463258e-05,
+      "loss": 0.5771,
+      "step": 740000
+    },
+    {
+      "epoch": 22.16,
+      "learning_rate": 7.850439297124599e-05,
+      "loss": 0.5763,
+      "step": 740500
+    },
+    {
+      "epoch": 22.17,
+      "learning_rate": 7.835463258785942e-05,
+      "loss": 0.5688,
+      "step": 741000
+    },
+    {
+      "epoch": 22.19,
+      "learning_rate": 7.820487220447283e-05,
+      "loss": 0.569,
+      "step": 741500
+    },
+    {
+      "epoch": 22.2,
+      "learning_rate": 7.805511182108624e-05,
+      "loss": 0.5737,
+      "step": 742000
+    },
+    {
+      "epoch": 22.22,
+      "learning_rate": 7.790535143769968e-05,
+      "loss": 0.57,
+      "step": 742500
+    },
+    {
+      "epoch": 22.23,
+      "learning_rate": 7.77555910543131e-05,
+      "loss": 0.5575,
+      "step": 743000
+    },
+    {
+      "epoch": 22.25,
+      "learning_rate": 7.760583067092652e-05,
+      "loss": 0.5636,
+      "step": 743500
+    },
+    {
+      "epoch": 22.26,
+      "learning_rate": 7.745607028753993e-05,
+      "loss": 0.5683,
+      "step": 744000
+    },
+    {
+      "epoch": 22.28,
+      "learning_rate": 7.730630990415335e-05,
+      "loss": 0.5666,
+      "step": 744500
+    },
+    {
+      "epoch": 22.29,
+      "learning_rate": 7.715654952076677e-05,
+      "loss": 0.5681,
+      "step": 745000
+    },
+    {
+      "epoch": 22.31,
+      "learning_rate": 7.700678913738018e-05,
+      "loss": 0.5856,
+      "step": 745500
+    },
+    {
+      "epoch": 22.32,
+      "learning_rate": 7.685702875399361e-05,
+      "loss": 0.5856,
+      "step": 746000
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 7.670726837060702e-05,
+      "loss": 0.5908,
+      "step": 746500
+    },
+    {
+      "epoch": 22.35,
+      "learning_rate": 7.655750798722044e-05,
+      "loss": 0.5792,
+      "step": 747000
+    },
+    {
+      "epoch": 22.37,
+      "learning_rate": 7.640774760383386e-05,
+      "loss": 0.5803,
+      "step": 747500
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 7.625798722044728e-05,
+      "loss": 0.5732,
+      "step": 748000
+    },
+    {
+      "epoch": 22.4,
+      "learning_rate": 7.610822683706069e-05,
+      "loss": 0.5846,
+      "step": 748500
+    },
+    {
+      "epoch": 22.41,
+      "learning_rate": 7.595846645367411e-05,
+      "loss": 0.5723,
+      "step": 749000
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 7.580870607028753e-05,
+      "loss": 0.5622,
+      "step": 749500
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 7.565894568690094e-05,
+      "loss": 0.5673,
+      "step": 750000
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 7.550918530351437e-05,
+      "loss": 0.5806,
+      "step": 750500
+    },
+    {
+      "epoch": 22.47,
+      "learning_rate": 7.535942492012778e-05,
+      "loss": 0.5653,
+      "step": 751000
+    },
+    {
+      "epoch": 22.49,
+      "learning_rate": 7.520966453674122e-05,
+      "loss": 0.568,
+      "step": 751500
+    },
+    {
+      "epoch": 22.5,
+      "learning_rate": 7.505990415335463e-05,
+      "loss": 0.5679,
+      "step": 752000
+    },
+    {
+      "epoch": 22.52,
+      "learning_rate": 7.491014376996804e-05,
+      "loss": 0.5796,
+      "step": 752500
+    },
+    {
+      "epoch": 22.53,
+      "learning_rate": 7.476038338658146e-05,
+      "loss": 0.5845,
+      "step": 753000
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 7.461062300319488e-05,
+      "loss": 0.5758,
+      "step": 753500
+    },
+    {
+      "epoch": 22.56,
+      "learning_rate": 7.446086261980831e-05,
+      "loss": 0.5836,
+      "step": 754000
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 7.431110223642172e-05,
+      "loss": 0.5636,
+      "step": 754500
+    },
+    {
+      "epoch": 22.59,
+      "learning_rate": 7.416134185303513e-05,
+      "loss": 0.5778,
+      "step": 755000
+    },
+    {
+      "epoch": 22.61,
+      "learning_rate": 7.401158146964856e-05,
+      "loss": 0.5698,
+      "step": 755500
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 7.386182108626197e-05,
+      "loss": 0.5984,
+      "step": 756000
+    },
+    {
+      "epoch": 22.64,
+      "learning_rate": 7.371206070287539e-05,
+      "loss": 0.5675,
+      "step": 756500
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 7.356230031948881e-05,
+      "loss": 0.5734,
+      "step": 757000
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 7.341253993610222e-05,
+      "loss": 0.5831,
+      "step": 757500
+    },
+    {
+      "epoch": 22.68,
+      "learning_rate": 7.326277955271565e-05,
+      "loss": 0.5838,
+      "step": 758000
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 7.311301916932908e-05,
+      "loss": 0.5799,
+      "step": 758500
+    },
+    {
+      "epoch": 22.71,
+      "learning_rate": 7.296325878594249e-05,
+      "loss": 0.5832,
+      "step": 759000
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 7.28134984025559e-05,
+      "loss": 0.5781,
+      "step": 759500
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 7.266373801916933e-05,
+      "loss": 0.5695,
+      "step": 760000
+    },
+    {
+      "epoch": 22.76,
+      "learning_rate": 7.251397763578274e-05,
+      "loss": 0.5707,
+      "step": 760500
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 7.236421725239615e-05,
+      "loss": 0.5765,
+      "step": 761000
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 7.221445686900958e-05,
+      "loss": 0.5814,
+      "step": 761500
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 7.206469648562299e-05,
+      "loss": 0.572,
+      "step": 762000
+    },
+    {
+      "epoch": 22.82,
+      "learning_rate": 7.19149361022364e-05,
+      "loss": 0.5733,
+      "step": 762500
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 7.176517571884983e-05,
+      "loss": 0.5663,
+      "step": 763000
+    },
+    {
+      "epoch": 22.85,
+      "learning_rate": 7.161541533546326e-05,
+      "loss": 0.5773,
+      "step": 763500
+    },
+    {
+      "epoch": 22.86,
+      "learning_rate": 7.146565495207667e-05,
+      "loss": 0.5639,
+      "step": 764000
+    },
+    {
+      "epoch": 22.88,
+      "learning_rate": 7.13158945686901e-05,
+      "loss": 0.56,
+      "step": 764500
+    },
+    {
+      "epoch": 22.89,
+      "learning_rate": 7.116613418530351e-05,
+      "loss": 0.5603,
+      "step": 765000
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 7.101637380191692e-05,
+      "loss": 0.5662,
+      "step": 765500
+    },
+    {
+      "epoch": 22.92,
+      "learning_rate": 7.086661341853035e-05,
+      "loss": 0.5735,
+      "step": 766000
+    },
+    {
+      "epoch": 22.94,
+      "learning_rate": 7.071685303514376e-05,
+      "loss": 0.5789,
+      "step": 766500
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 7.056709265175717e-05,
+      "loss": 0.5632,
+      "step": 767000
+    },
+    {
+      "epoch": 22.96,
+      "learning_rate": 7.04173322683706e-05,
+      "loss": 0.5686,
+      "step": 767500
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 7.026757188498403e-05,
+      "loss": 0.5852,
+      "step": 768000
+    },
+    {
+      "epoch": 22.99,
+      "learning_rate": 7.011781150159744e-05,
+      "loss": 0.5775,
+      "step": 768500
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 6.996805111821085e-05,
+      "loss": 0.5587,
+      "step": 769000
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 6.981829073482428e-05,
+      "loss": 0.5653,
+      "step": 769500
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 6.966853035143769e-05,
+      "loss": 0.5455,
+      "step": 770000
+    },
+    {
+      "epoch": 23.05,
+      "learning_rate": 6.951876996805112e-05,
+      "loss": 0.549,
+      "step": 770500
+    },
+    {
+      "epoch": 23.07,
+      "learning_rate": 6.936900958466453e-05,
+      "loss": 0.5207,
+      "step": 771000
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 6.921924920127794e-05,
+      "loss": 0.5476,
+      "step": 771500
+    },
+    {
+      "epoch": 23.1,
+      "learning_rate": 6.906948881789137e-05,
+      "loss": 0.5457,
+      "step": 772000
+    },
+    {
+      "epoch": 23.11,
+      "learning_rate": 6.89197284345048e-05,
+      "loss": 0.5307,
+      "step": 772500
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 6.876996805111821e-05,
+      "loss": 0.5558,
+      "step": 773000
+    },
+    {
+      "epoch": 23.14,
+      "learning_rate": 6.862020766773162e-05,
+      "loss": 0.5327,
+      "step": 773500
+    },
+    {
+      "epoch": 23.16,
+      "learning_rate": 6.847044728434505e-05,
+      "loss": 0.5543,
+      "step": 774000
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 6.832068690095846e-05,
+      "loss": 0.5537,
+      "step": 774500
+    },
+    {
+      "epoch": 23.19,
+      "learning_rate": 6.817092651757187e-05,
+      "loss": 0.5539,
+      "step": 775000
+    },
+    {
+      "epoch": 23.2,
+      "learning_rate": 6.80211661341853e-05,
+      "loss": 0.5566,
+      "step": 775500
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 6.787140575079871e-05,
+      "loss": 0.5548,
+      "step": 776000
+    },
+    {
+      "epoch": 23.23,
+      "learning_rate": 6.772164536741214e-05,
+      "loss": 0.5596,
+      "step": 776500
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 6.757188498402556e-05,
+      "loss": 0.5729,
+      "step": 777000
+    },
+    {
+      "epoch": 23.26,
+      "learning_rate": 6.742212460063898e-05,
+      "loss": 0.5634,
+      "step": 777500
+    },
+    {
+      "epoch": 23.28,
+      "learning_rate": 6.727236421725239e-05,
+      "loss": 0.5553,
+      "step": 778000
+    },
+    {
+      "epoch": 23.29,
+      "learning_rate": 6.712260383386581e-05,
+      "loss": 0.5498,
+      "step": 778500
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 6.697284345047923e-05,
+      "loss": 0.5458,
+      "step": 779000
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 6.682308306709264e-05,
+      "loss": 0.5445,
+      "step": 779500
+    },
+    {
+      "epoch": 23.34,
+      "learning_rate": 6.667332268370607e-05,
+      "loss": 0.5665,
+      "step": 780000
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 6.652356230031948e-05,
+      "loss": 0.5741,
+      "step": 780500
+    },
+    {
+      "epoch": 23.37,
+      "learning_rate": 6.637380191693289e-05,
+      "loss": 0.5481,
+      "step": 781000
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 6.622404153354632e-05,
+      "loss": 0.5394,
+      "step": 781500
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 6.607428115015974e-05,
+      "loss": 0.5531,
+      "step": 782000
+    },
+    {
+      "epoch": 23.41,
+      "learning_rate": 6.592452076677316e-05,
+      "loss": 0.5396,
+      "step": 782500
+    },
+    {
+      "epoch": 23.43,
+      "learning_rate": 6.577476038338658e-05,
+      "loss": 0.5662,
+      "step": 783000
+    },
+    {
+      "epoch": 23.44,
+      "learning_rate": 6.5625e-05,
+      "loss": 0.5522,
+      "step": 783500
+    },
+    {
+      "epoch": 23.46,
+      "learning_rate": 6.547523961661341e-05,
+      "loss": 0.5631,
+      "step": 784000
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 6.532547923322683e-05,
+      "loss": 0.5614,
+      "step": 784500
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 6.517571884984025e-05,
+      "loss": 0.5453,
+      "step": 785000
+    },
+    {
+      "epoch": 23.5,
+      "learning_rate": 6.502595846645366e-05,
+      "loss": 0.5575,
+      "step": 785500
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 6.487619808306709e-05,
+      "loss": 0.5443,
+      "step": 786000
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 6.472643769968051e-05,
+      "loss": 0.5701,
+      "step": 786500
+    },
+    {
+      "epoch": 23.55,
+      "learning_rate": 6.457667731629393e-05,
+      "loss": 0.5633,
+      "step": 787000
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 6.442691693290734e-05,
+      "loss": 0.5536,
+      "step": 787500
+    },
+    {
+      "epoch": 23.58,
+      "learning_rate": 6.427715654952076e-05,
+      "loss": 0.5453,
+      "step": 788000
+    },
+    {
+      "epoch": 23.59,
+      "learning_rate": 6.412739616613418e-05,
+      "loss": 0.5425,
+      "step": 788500
+    },
+    {
+      "epoch": 23.61,
+      "learning_rate": 6.39776357827476e-05,
+      "loss": 0.5394,
+      "step": 789000
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 6.382787539936102e-05,
+      "loss": 0.5696,
+      "step": 789500
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 6.367811501597443e-05,
+      "loss": 0.559,
+      "step": 790000
+    },
+    {
+      "epoch": 23.65,
+      "learning_rate": 6.352835463258785e-05,
+      "loss": 0.5539,
+      "step": 790500
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 6.337859424920128e-05,
+      "loss": 0.5493,
+      "step": 791000
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 6.32288338658147e-05,
+      "loss": 0.5632,
+      "step": 791500
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 6.30790734824281e-05,
+      "loss": 0.5485,
+      "step": 792000
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 6.292931309904153e-05,
+      "loss": 0.55,
+      "step": 792500
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 6.277955271565495e-05,
+      "loss": 0.561,
+      "step": 793000
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 6.262979233226836e-05,
+      "loss": 0.551,
+      "step": 793500
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 6.248003194888178e-05,
+      "loss": 0.5602,
+      "step": 794000
+    },
+    {
+      "epoch": 23.77,
+      "learning_rate": 6.23302715654952e-05,
+      "loss": 0.5444,
+      "step": 794500
+    },
+    {
+      "epoch": 23.79,
+      "learning_rate": 6.218051118210862e-05,
+      "loss": 0.5244,
+      "step": 795000
+    },
+    {
+      "epoch": 23.8,
+      "learning_rate": 6.203075079872204e-05,
+      "loss": 0.5545,
+      "step": 795500
+    },
+    {
+      "epoch": 23.82,
+      "learning_rate": 6.188099041533546e-05,
+      "loss": 0.5522,
+      "step": 796000
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 6.173123003194887e-05,
+      "loss": 0.5553,
+      "step": 796500
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 6.15814696485623e-05,
+      "loss": 0.5523,
+      "step": 797000
+    },
+    {
+      "epoch": 23.86,
+      "learning_rate": 6.143170926517571e-05,
+      "loss": 0.5532,
+      "step": 797500
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 6.128194888178913e-05,
+      "loss": 0.5565,
+      "step": 798000
+    },
+    {
+      "epoch": 23.89,
+      "learning_rate": 6.113218849840255e-05,
+      "loss": 0.5556,
+      "step": 798500
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 6.0982428115015965e-05,
+      "loss": 0.5497,
+      "step": 799000
+    },
+    {
+      "epoch": 23.92,
+      "learning_rate": 6.0832667731629385e-05,
+      "loss": 0.5494,
+      "step": 799500
+    },
+    {
+      "epoch": 23.94,
+      "learning_rate": 6.068290734824281e-05,
+      "loss": 0.5575,
+      "step": 800000
+    },
+    {
+      "epoch": 23.95,
+      "learning_rate": 6.053314696485623e-05,
+      "loss": 0.5659,
+      "step": 800500
+    },
+    {
+      "epoch": 23.97,
+      "learning_rate": 6.038338658146964e-05,
+      "loss": 0.5659,
+      "step": 801000
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 6.023362619808306e-05,
+      "loss": 0.5612,
+      "step": 801500
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 6.008386581469648e-05,
+      "loss": 0.5535,
+      "step": 802000
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 5.9934105431309895e-05,
+      "loss": 0.5224,
+      "step": 802500
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 5.9784345047923314e-05,
+      "loss": 0.5445,
+      "step": 803000
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 5.9634584664536734e-05,
+      "loss": 0.5171,
+      "step": 803500
+    },
+    {
+      "epoch": 24.06,
+      "learning_rate": 5.948482428115015e-05,
+      "loss": 0.5264,
+      "step": 804000
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 5.933506389776358e-05,
+      "loss": 0.5268,
+      "step": 804500
+    },
+    {
+      "epoch": 24.09,
+      "learning_rate": 5.918530351437699e-05,
+      "loss": 0.5303,
+      "step": 805000
+    },
+    {
+      "epoch": 24.1,
+      "learning_rate": 5.903554313099041e-05,
+      "loss": 0.5581,
+      "step": 805500
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 5.888578274760383e-05,
+      "loss": 0.5234,
+      "step": 806000
+    },
+    {
+      "epoch": 24.13,
+      "learning_rate": 5.873602236421725e-05,
+      "loss": 0.5372,
+      "step": 806500
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 5.858626198083066e-05,
+      "loss": 0.5225,
+      "step": 807000
+    },
+    {
+      "epoch": 24.16,
+      "learning_rate": 5.843650159744408e-05,
+      "loss": 0.5254,
+      "step": 807500
+    },
+    {
+      "epoch": 24.18,
+      "learning_rate": 5.82867412140575e-05,
+      "loss": 0.5203,
+      "step": 808000
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 5.8136980830670915e-05,
+      "loss": 0.5261,
+      "step": 808500
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 5.798722044728434e-05,
+      "loss": 0.5365,
+      "step": 809000
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 5.783746006389776e-05,
+      "loss": 0.547,
+      "step": 809500
+    },
+    {
+      "epoch": 24.24,
+      "learning_rate": 5.768769968051118e-05,
+      "loss": 0.5386,
+      "step": 810000
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 5.75379392971246e-05,
+      "loss": 0.5294,
+      "step": 810500
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 5.738817891373801e-05,
+      "loss": 0.5364,
+      "step": 811000
+    },
+    {
+      "epoch": 24.28,
+      "learning_rate": 5.723841853035143e-05,
+      "loss": 0.5454,
+      "step": 811500
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 5.708865814696485e-05,
+      "loss": 0.5302,
+      "step": 812000
+    },
+    {
+      "epoch": 24.31,
+      "learning_rate": 5.693889776357827e-05,
+      "loss": 0.5415,
+      "step": 812500
+    },
+    {
+      "epoch": 24.33,
+      "learning_rate": 5.678913738019168e-05,
+      "loss": 0.5464,
+      "step": 813000
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 5.663937699680511e-05,
+      "loss": 0.5327,
+      "step": 813500
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 5.648961661341853e-05,
+      "loss": 0.5463,
+      "step": 814000
+    },
+    {
+      "epoch": 24.37,
+      "learning_rate": 5.633985623003195e-05,
+      "loss": 0.539,
+      "step": 814500
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 5.619009584664536e-05,
+      "loss": 0.5299,
+      "step": 815000
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 5.604033546325878e-05,
+      "loss": 0.5311,
+      "step": 815500
+    },
+    {
+      "epoch": 24.42,
+      "learning_rate": 5.58905750798722e-05,
+      "loss": 0.5399,
+      "step": 816000
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 5.574081469648562e-05,
+      "loss": 0.5236,
+      "step": 816500
+    },
+    {
+      "epoch": 24.45,
+      "learning_rate": 5.559105431309903e-05,
+      "loss": 0.5401,
+      "step": 817000
+    },
+    {
+      "epoch": 24.46,
+      "learning_rate": 5.544129392971245e-05,
+      "loss": 0.5268,
+      "step": 817500
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 5.529153354632587e-05,
+      "loss": 0.5314,
+      "step": 818000
+    },
+    {
+      "epoch": 24.49,
+      "learning_rate": 5.51417731629393e-05,
+      "loss": 0.5373,
+      "step": 818500
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 5.4992012779552716e-05,
+      "loss": 0.5365,
+      "step": 819000
+    },
+    {
+      "epoch": 24.52,
+      "learning_rate": 5.484225239616613e-05,
+      "loss": 0.542,
+      "step": 819500
+    },
+    {
+      "epoch": 24.54,
+      "learning_rate": 5.469249201277955e-05,
+      "loss": 0.544,
+      "step": 820000
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 5.454273162939297e-05,
+      "loss": 0.5427,
+      "step": 820500
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 5.439297124600638e-05,
+      "loss": 0.518,
+      "step": 821000
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 5.42432108626198e-05,
+      "loss": 0.5258,
+      "step": 821500
+    },
+    {
+      "epoch": 24.6,
+      "learning_rate": 5.409345047923322e-05,
+      "loss": 0.5218,
+      "step": 822000
+    },
+    {
+      "epoch": 24.61,
+      "learning_rate": 5.394369009584664e-05,
+      "loss": 0.5127,
+      "step": 822500
+    },
+    {
+      "epoch": 24.63,
+      "learning_rate": 5.3793929712460065e-05,
+      "loss": 0.5409,
+      "step": 823000
+    },
+    {
+      "epoch": 24.64,
+      "learning_rate": 5.364416932907348e-05,
+      "loss": 0.5301,
+      "step": 823500
+    },
+    {
+      "epoch": 24.66,
+      "learning_rate": 5.34944089456869e-05,
+      "loss": 0.5138,
+      "step": 824000
+    },
+    {
+      "epoch": 24.67,
+      "learning_rate": 5.334464856230032e-05,
+      "loss": 0.5297,
+      "step": 824500
+    },
+    {
+      "epoch": 24.69,
+      "learning_rate": 5.3194888178913736e-05,
+      "loss": 0.5252,
+      "step": 825000
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 5.304512779552715e-05,
+      "loss": 0.5449,
+      "step": 825500
+    },
+    {
+      "epoch": 24.72,
+      "learning_rate": 5.289536741214057e-05,
+      "loss": 0.5354,
+      "step": 826000
+    },
+    {
+      "epoch": 24.73,
+      "learning_rate": 5.274560702875399e-05,
+      "loss": 0.5309,
+      "step": 826500
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 5.25958466453674e-05,
+      "loss": 0.5366,
+      "step": 827000
+    },
+    {
+      "epoch": 24.76,
+      "learning_rate": 5.244608626198083e-05,
+      "loss": 0.5278,
+      "step": 827500
+    },
+    {
+      "epoch": 24.78,
+      "learning_rate": 5.2296325878594246e-05,
+      "loss": 0.5416,
+      "step": 828000
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 5.2146565495207666e-05,
+      "loss": 0.5228,
+      "step": 828500
+    },
+    {
+      "epoch": 24.81,
+      "learning_rate": 5.1996805111821085e-05,
+      "loss": 0.5331,
+      "step": 829000
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 5.18470447284345e-05,
+      "loss": 0.5285,
+      "step": 829500
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 5.169728434504792e-05,
+      "loss": 0.5245,
+      "step": 830000
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 5.154752396166134e-05,
+      "loss": 0.5333,
+      "step": 830500
+    },
+    {
+      "epoch": 24.86,
+      "learning_rate": 5.1397763578274756e-05,
+      "loss": 0.5417,
+      "step": 831000
+    },
+    {
+      "epoch": 24.88,
+      "learning_rate": 5.124800319488817e-05,
+      "loss": 0.5446,
+      "step": 831500
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 5.1098242811501595e-05,
+      "loss": 0.5215,
+      "step": 832000
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 5.0948482428115015e-05,
+      "loss": 0.5384,
+      "step": 832500
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 5.0798722044728434e-05,
+      "loss": 0.5386,
+      "step": 833000
+    },
+    {
+      "epoch": 24.94,
+      "learning_rate": 5.064896166134185e-05,
+      "loss": 0.5332,
+      "step": 833500
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 5.0499201277955266e-05,
+      "loss": 0.5345,
+      "step": 834000
+    },
+    {
+      "epoch": 24.97,
+      "learning_rate": 5.0349440894568686e-05,
+      "loss": 0.5333,
+      "step": 834500
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 5.0199680511182105e-05,
+      "loss": 0.5428,
+      "step": 835000
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 5.004992012779552e-05,
+      "loss": 0.5283,
+      "step": 835500
+    },
+    {
+      "epoch": 25.01,
+      "learning_rate": 4.990015974440894e-05,
+      "loss": 0.5146,
+      "step": 836000
+    },
+    {
+      "epoch": 25.03,
+      "learning_rate": 4.9750399361022364e-05,
+      "loss": 0.5071,
+      "step": 836500
+    },
+    {
+      "epoch": 25.04,
+      "learning_rate": 4.960063897763578e-05,
+      "loss": 0.512,
+      "step": 837000
+    },
+    {
+      "epoch": 25.06,
+      "learning_rate": 4.94508785942492e-05,
+      "loss": 0.5052,
+      "step": 837500
+    },
+    {
+      "epoch": 25.07,
+      "learning_rate": 4.9301118210862615e-05,
+      "loss": 0.5082,
+      "step": 838000
+    },
+    {
+      "epoch": 25.09,
+      "learning_rate": 4.9151357827476035e-05,
+      "loss": 0.5107,
+      "step": 838500
+    },
+    {
+      "epoch": 25.1,
+      "learning_rate": 4.9001597444089454e-05,
+      "loss": 0.5401,
+      "step": 839000
+    },
+    {
+      "epoch": 25.12,
+      "learning_rate": 4.885183706070287e-05,
+      "loss": 0.5105,
+      "step": 839500
+    },
+    {
+      "epoch": 25.13,
+      "learning_rate": 4.8702076677316286e-05,
+      "loss": 0.5168,
+      "step": 840000
+    },
+    {
+      "epoch": 25.15,
+      "learning_rate": 4.8552316293929706e-05,
+      "loss": 0.5038,
+      "step": 840500
+    },
+    {
+      "epoch": 25.16,
+      "learning_rate": 4.8402555910543125e-05,
+      "loss": 0.5096,
+      "step": 841000
+    },
+    {
+      "epoch": 25.18,
+      "learning_rate": 4.825279552715655e-05,
+      "loss": 0.5241,
+      "step": 841500
+    },
+    {
+      "epoch": 25.19,
+      "learning_rate": 4.8103035143769964e-05,
+      "loss": 0.5136,
+      "step": 842000
+    },
+    {
+      "epoch": 25.21,
+      "learning_rate": 4.7953274760383384e-05,
+      "loss": 0.5136,
+      "step": 842500
+    },
+    {
+      "epoch": 25.22,
+      "learning_rate": 4.78035143769968e-05,
+      "loss": 0.5089,
+      "step": 843000
+    },
+    {
+      "epoch": 25.24,
+      "learning_rate": 4.765375399361022e-05,
+      "loss": 0.5147,
+      "step": 843500
+    },
+    {
+      "epoch": 25.25,
+      "learning_rate": 4.7503993610223635e-05,
+      "loss": 0.5061,
+      "step": 844000
+    },
+    {
+      "epoch": 25.27,
+      "learning_rate": 4.7354233226837055e-05,
+      "loss": 0.5102,
+      "step": 844500
+    },
+    {
+      "epoch": 25.28,
+      "learning_rate": 4.7204472843450474e-05,
+      "loss": 0.5126,
+      "step": 845000
+    },
+    {
+      "epoch": 25.3,
+      "learning_rate": 4.705471246006389e-05,
+      "loss": 0.5259,
+      "step": 845500
+    },
+    {
+      "epoch": 25.31,
+      "learning_rate": 4.690495207667731e-05,
+      "loss": 0.5129,
+      "step": 846000
+    },
+    {
+      "epoch": 25.33,
+      "learning_rate": 4.675519169329073e-05,
+      "loss": 0.5066,
+      "step": 846500
+    },
+    {
+      "epoch": 25.34,
+      "learning_rate": 4.660543130990415e-05,
+      "loss": 0.5127,
+      "step": 847000
+    },
+    {
+      "epoch": 25.36,
+      "learning_rate": 4.645567092651757e-05,
+      "loss": 0.5169,
+      "step": 847500
+    },
+    {
+      "epoch": 25.37,
+      "learning_rate": 4.6305910543130984e-05,
+      "loss": 0.4951,
+      "step": 848000
+    },
+    {
+      "epoch": 25.39,
+      "learning_rate": 4.6156150159744404e-05,
+      "loss": 0.5076,
+      "step": 848500
+    },
+    {
+      "epoch": 25.4,
+      "learning_rate": 4.600638977635782e-05,
+      "loss": 0.5142,
+      "step": 849000
+    },
+    {
+      "epoch": 25.42,
+      "learning_rate": 4.585662939297124e-05,
+      "loss": 0.5138,
+      "step": 849500
+    },
+    {
+      "epoch": 25.43,
+      "learning_rate": 4.5706869009584655e-05,
+      "loss": 0.5185,
+      "step": 850000
+    },
+    {
+      "epoch": 25.45,
+      "learning_rate": 4.555710862619808e-05,
+      "loss": 0.521,
+      "step": 850500
+    },
+    {
+      "epoch": 25.46,
+      "learning_rate": 4.54073482428115e-05,
+      "loss": 0.4925,
+      "step": 851000
+    },
+    {
+      "epoch": 25.48,
+      "learning_rate": 4.525758785942492e-05,
+      "loss": 0.5164,
+      "step": 851500
+    },
+    {
+      "epoch": 25.49,
+      "learning_rate": 4.510782747603833e-05,
+      "loss": 0.5023,
+      "step": 852000
+    },
+    {
+      "epoch": 25.51,
+      "learning_rate": 4.495806709265175e-05,
+      "loss": 0.5219,
+      "step": 852500
+    },
+    {
+      "epoch": 25.52,
+      "learning_rate": 4.480830670926517e-05,
+      "loss": 0.5215,
+      "step": 853000
+    },
+    {
+      "epoch": 25.54,
+      "learning_rate": 4.465854632587859e-05,
+      "loss": 0.5232,
+      "step": 853500
+    },
+    {
+      "epoch": 25.55,
+      "learning_rate": 4.4508785942492004e-05,
+      "loss": 0.5113,
+      "step": 854000
+    },
+    {
+      "epoch": 25.57,
+      "learning_rate": 4.4359025559105423e-05,
+      "loss": 0.5055,
+      "step": 854500
+    },
+    {
+      "epoch": 25.58,
+      "learning_rate": 4.420926517571885e-05,
+      "loss": 0.514,
+      "step": 855000
+    },
+    {
+      "epoch": 25.6,
+      "learning_rate": 4.405950479233227e-05,
+      "loss": 0.5168,
+      "step": 855500
+    },
+    {
+      "epoch": 25.61,
+      "learning_rate": 4.390974440894568e-05,
+      "loss": 0.521,
+      "step": 856000
+    },
+    {
+      "epoch": 25.63,
+      "learning_rate": 4.37599840255591e-05,
+      "loss": 0.5077,
+      "step": 856500
+    },
+    {
+      "epoch": 25.64,
+      "learning_rate": 4.361022364217252e-05,
+      "loss": 0.5123,
+      "step": 857000
+    },
+    {
+      "epoch": 25.66,
+      "learning_rate": 4.346046325878594e-05,
+      "loss": 0.5189,
+      "step": 857500
+    },
+    {
+      "epoch": 25.67,
+      "learning_rate": 4.331070287539935e-05,
+      "loss": 0.5124,
+      "step": 858000
+    },
+    {
+      "epoch": 25.69,
+      "learning_rate": 4.316094249201277e-05,
+      "loss": 0.5086,
+      "step": 858500
+    },
+    {
+      "epoch": 25.7,
+      "learning_rate": 4.301118210862619e-05,
+      "loss": 0.5175,
+      "step": 859000
+    },
+    {
+      "epoch": 25.72,
+      "learning_rate": 4.286142172523962e-05,
+      "loss": 0.5216,
+      "step": 859500
+    },
+    {
+      "epoch": 25.73,
+      "learning_rate": 4.271166134185304e-05,
+      "loss": 0.5226,
+      "step": 860000
+    },
+    {
+      "epoch": 25.75,
+      "learning_rate": 4.256190095846645e-05,
+      "loss": 0.525,
+      "step": 860500
+    },
+    {
+      "epoch": 25.76,
+      "learning_rate": 4.241214057507987e-05,
+      "loss": 0.499,
+      "step": 861000
+    },
+    {
+      "epoch": 25.78,
+      "learning_rate": 4.226238019169329e-05,
+      "loss": 0.512,
+      "step": 861500
+    },
+    {
+      "epoch": 25.79,
+      "learning_rate": 4.21126198083067e-05,
+      "loss": 0.5263,
+      "step": 862000
+    },
+    {
+      "epoch": 25.81,
+      "learning_rate": 4.196285942492012e-05,
+      "loss": 0.5219,
+      "step": 862500
+    },
+    {
+      "epoch": 25.82,
+      "learning_rate": 4.181309904153354e-05,
+      "loss": 0.5195,
+      "step": 863000
+    },
+    {
+      "epoch": 25.84,
+      "learning_rate": 4.166333865814696e-05,
+      "loss": 0.5072,
+      "step": 863500
+    },
+    {
+      "epoch": 25.85,
+      "learning_rate": 4.151357827476037e-05,
+      "loss": 0.5175,
+      "step": 864000
+    },
+    {
+      "epoch": 25.87,
+      "learning_rate": 4.13638178913738e-05,
+      "loss": 0.5059,
+      "step": 864500
+    },
+    {
+      "epoch": 25.88,
+      "learning_rate": 4.121405750798722e-05,
+      "loss": 0.4973,
+      "step": 865000
+    },
+    {
+      "epoch": 25.9,
+      "learning_rate": 4.106429712460064e-05,
+      "loss": 0.5007,
+      "step": 865500
+    },
+    {
+      "epoch": 25.91,
+      "learning_rate": 4.091453674121406e-05,
+      "loss": 0.5164,
+      "step": 866000
+    },
+    {
+      "epoch": 25.93,
+      "learning_rate": 4.076477635782747e-05,
+      "loss": 0.5178,
+      "step": 866500
+    },
+    {
+      "epoch": 25.94,
+      "learning_rate": 4.061501597444089e-05,
+      "loss": 0.5126,
+      "step": 867000
+    },
+    {
+      "epoch": 25.96,
+      "learning_rate": 4.046525559105431e-05,
+      "loss": 0.5091,
+      "step": 867500
+    },
+    {
+      "epoch": 25.97,
+      "learning_rate": 4.031549520766772e-05,
+      "loss": 0.5069,
+      "step": 868000
+    },
+    {
+      "epoch": 25.99,
+      "learning_rate": 4.016573482428114e-05,
+      "loss": 0.512,
+      "step": 868500
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 4.001597444089457e-05,
+      "loss": 0.5214,
+      "step": 869000
+    },
+    {
+      "epoch": 26.02,
+      "learning_rate": 3.986621405750799e-05,
+      "loss": 0.5034,
+      "step": 869500
+    },
+    {
+      "epoch": 26.03,
+      "learning_rate": 3.9716453674121406e-05,
+      "loss": 0.4907,
+      "step": 870000
+    },
+    {
+      "epoch": 26.05,
+      "learning_rate": 3.956669329073482e-05,
+      "loss": 0.4864,
+      "step": 870500
+    },
+    {
+      "epoch": 26.06,
+      "learning_rate": 3.941693290734824e-05,
+      "loss": 0.4909,
+      "step": 871000
+    },
+    {
+      "epoch": 26.08,
+      "learning_rate": 3.926717252396166e-05,
+      "loss": 0.493,
+      "step": 871500
+    },
+    {
+      "epoch": 26.09,
+      "learning_rate": 3.911741214057508e-05,
+      "loss": 0.494,
+      "step": 872000
+    },
+    {
+      "epoch": 26.11,
+      "learning_rate": 3.896765175718849e-05,
+      "loss": 0.5021,
+      "step": 872500
+    },
+    {
+      "epoch": 26.12,
+      "learning_rate": 3.881789137380191e-05,
+      "loss": 0.5094,
+      "step": 873000
+    },
+    {
+      "epoch": 26.14,
+      "learning_rate": 3.8668130990415336e-05,
+      "loss": 0.4912,
+      "step": 873500
+    },
+    {
+      "epoch": 26.15,
+      "learning_rate": 3.8518370607028755e-05,
+      "loss": 0.4909,
+      "step": 874000
+    },
+    {
+      "epoch": 26.17,
+      "learning_rate": 3.836861022364217e-05,
+      "loss": 0.4977,
+      "step": 874500
+    },
+    {
+      "epoch": 26.18,
+      "learning_rate": 3.821884984025559e-05,
+      "loss": 0.4962,
+      "step": 875000
+    },
+    {
+      "epoch": 26.2,
+      "learning_rate": 3.806908945686901e-05,
+      "loss": 0.4924,
+      "step": 875500
+    },
+    {
+      "epoch": 26.21,
+      "learning_rate": 3.7919329073482426e-05,
+      "loss": 0.4924,
+      "step": 876000
+    },
+    {
+      "epoch": 26.23,
+      "learning_rate": 3.776956869009584e-05,
+      "loss": 0.4955,
+      "step": 876500
+    },
+    {
+      "epoch": 26.24,
+      "learning_rate": 3.761980830670926e-05,
+      "loss": 0.5015,
+      "step": 877000
+    },
+    {
+      "epoch": 26.26,
+      "learning_rate": 3.747004792332268e-05,
+      "loss": 0.4967,
+      "step": 877500
+    },
+    {
+      "epoch": 26.27,
+      "learning_rate": 3.73202875399361e-05,
+      "loss": 0.4857,
+      "step": 878000
+    },
+    {
+      "epoch": 26.29,
+      "learning_rate": 3.717052715654952e-05,
+      "loss": 0.4936,
+      "step": 878500
+    },
+    {
+      "epoch": 26.3,
+      "learning_rate": 3.7020766773162936e-05,
+      "loss": 0.5054,
+      "step": 879000
+    },
+    {
+      "epoch": 26.32,
+      "learning_rate": 3.6871006389776356e-05,
+      "loss": 0.4927,
+      "step": 879500
+    },
+    {
+      "epoch": 26.33,
+      "learning_rate": 3.6721246006389775e-05,
+      "loss": 0.4929,
+      "step": 880000
+    },
+    {
+      "epoch": 26.35,
+      "learning_rate": 3.657148562300319e-05,
+      "loss": 0.5001,
+      "step": 880500
+    },
+    {
+      "epoch": 26.36,
+      "learning_rate": 3.642172523961661e-05,
+      "loss": 0.501,
+      "step": 881000
+    },
+    {
+      "epoch": 26.38,
+      "learning_rate": 3.6271964856230034e-05,
+      "loss": 0.4847,
+      "step": 881500
+    },
+    {
+      "epoch": 26.39,
+      "learning_rate": 3.6122204472843446e-05,
+      "loss": 0.4855,
+      "step": 882000
+    },
+    {
+      "epoch": 26.41,
+      "learning_rate": 3.5972444089456866e-05,
+      "loss": 0.4966,
+      "step": 882500
+    },
+    {
+      "epoch": 26.42,
+      "learning_rate": 3.5822683706070285e-05,
+      "loss": 0.4968,
+      "step": 883000
+    },
+    {
+      "epoch": 26.44,
+      "learning_rate": 3.56729233226837e-05,
+      "loss": 0.4969,
+      "step": 883500
+    },
+    {
+      "epoch": 26.45,
+      "learning_rate": 3.5523162939297124e-05,
+      "loss": 0.4977,
+      "step": 884000
+    },
+    {
+      "epoch": 26.47,
+      "learning_rate": 3.5373402555910544e-05,
+      "loss": 0.4885,
+      "step": 884500
+    },
+    {
+      "epoch": 26.48,
+      "learning_rate": 3.5223642172523956e-05,
+      "loss": 0.4988,
+      "step": 885000
+    },
+    {
+      "epoch": 26.5,
+      "learning_rate": 3.5073881789137376e-05,
+      "loss": 0.5054,
+      "step": 885500
+    },
+    {
+      "epoch": 26.51,
+      "learning_rate": 3.4924121405750795e-05,
+      "loss": 0.4956,
+      "step": 886000
+    },
+    {
+      "epoch": 26.53,
+      "learning_rate": 3.4774361022364215e-05,
+      "loss": 0.4913,
+      "step": 886500
+    },
+    {
+      "epoch": 26.54,
+      "learning_rate": 3.4624600638977634e-05,
+      "loss": 0.4927,
+      "step": 887000
+    },
+    {
+      "epoch": 26.56,
+      "learning_rate": 3.4474840255591054e-05,
+      "loss": 0.4845,
+      "step": 887500
+    },
+    {
+      "epoch": 26.57,
+      "learning_rate": 3.4325079872204466e-05,
+      "loss": 0.5005,
+      "step": 888000
+    },
+    {
+      "epoch": 26.59,
+      "learning_rate": 3.417531948881789e-05,
+      "loss": 0.5098,
+      "step": 888500
+    },
+    {
+      "epoch": 26.6,
+      "learning_rate": 3.4025559105431305e-05,
+      "loss": 0.4953,
+      "step": 889000
+    },
+    {
+      "epoch": 26.62,
+      "learning_rate": 3.3875798722044725e-05,
+      "loss": 0.4909,
+      "step": 889500
+    },
+    {
+      "epoch": 26.63,
+      "learning_rate": 3.3726038338658144e-05,
+      "loss": 0.4913,
+      "step": 890000
+    },
+    {
+      "epoch": 26.65,
+      "learning_rate": 3.3576277955271563e-05,
+      "loss": 0.5076,
+      "step": 890500
+    },
+    {
+      "epoch": 26.66,
+      "learning_rate": 3.342651757188498e-05,
+      "loss": 0.4814,
+      "step": 891000
+    },
+    {
+      "epoch": 26.68,
+      "learning_rate": 3.32767571884984e-05,
+      "loss": 0.4886,
+      "step": 891500
+    },
+    {
+      "epoch": 26.69,
+      "learning_rate": 3.3126996805111815e-05,
+      "loss": 0.4826,
+      "step": 892000
+    },
+    {
+      "epoch": 26.71,
+      "learning_rate": 3.2977236421725235e-05,
+      "loss": 0.4945,
+      "step": 892500
+    },
+    {
+      "epoch": 26.72,
+      "learning_rate": 3.2827476038338654e-05,
+      "loss": 0.503,
+      "step": 893000
+    },
+    {
+      "epoch": 26.74,
+      "learning_rate": 3.2677715654952073e-05,
+      "loss": 0.4845,
+      "step": 893500
+    },
+    {
+      "epoch": 26.75,
+      "learning_rate": 3.252795527156549e-05,
+      "loss": 0.5016,
+      "step": 894000
+    },
+    {
+      "epoch": 26.77,
+      "learning_rate": 3.237819488817891e-05,
+      "loss": 0.4941,
+      "step": 894500
+    },
+    {
+      "epoch": 26.78,
+      "learning_rate": 3.2228434504792325e-05,
+      "loss": 0.4896,
+      "step": 895000
+    },
+    {
+      "epoch": 26.79,
+      "learning_rate": 3.207867412140575e-05,
+      "loss": 0.5019,
+      "step": 895500
+    },
+    {
+      "epoch": 26.81,
+      "learning_rate": 3.1928913738019164e-05,
+      "loss": 0.492,
+      "step": 896000
+    },
+    {
+      "epoch": 26.82,
+      "learning_rate": 3.1779153354632583e-05,
+      "loss": 0.4949,
+      "step": 896500
+    },
+    {
+      "epoch": 26.84,
+      "learning_rate": 3.1629392971246e-05,
+      "loss": 0.4991,
+      "step": 897000
+    },
+    {
+      "epoch": 26.85,
+      "learning_rate": 3.147963258785942e-05,
+      "loss": 0.4874,
+      "step": 897500
+    },
+    {
+      "epoch": 26.87,
+      "learning_rate": 3.132987220447284e-05,
+      "loss": 0.4964,
+      "step": 898000
+    },
+    {
+      "epoch": 26.88,
+      "learning_rate": 3.118011182108626e-05,
+      "loss": 0.5036,
+      "step": 898500
+    },
+    {
+      "epoch": 26.9,
+      "learning_rate": 3.1030351437699674e-05,
+      "loss": 0.4805,
+      "step": 899000
+    },
+    {
+      "epoch": 26.91,
+      "learning_rate": 3.0880591054313093e-05,
+      "loss": 0.4914,
+      "step": 899500
+    },
+    {
+      "epoch": 26.93,
+      "learning_rate": 3.073083067092652e-05,
+      "loss": 0.5006,
+      "step": 900000
+    },
+    {
+      "epoch": 26.94,
+      "learning_rate": 3.058107028753993e-05,
+      "loss": 0.4889,
+      "step": 900500
+    },
+    {
+      "epoch": 26.96,
+      "learning_rate": 3.0431309904153352e-05,
+      "loss": 0.5026,
+      "step": 901000
+    },
+    {
+      "epoch": 26.97,
+      "learning_rate": 3.0281549520766768e-05,
+      "loss": 0.4827,
+      "step": 901500
+    },
+    {
+      "epoch": 26.99,
+      "learning_rate": 3.0131789137380187e-05,
+      "loss": 0.5045,
+      "step": 902000
+    },
+    {
+      "epoch": 27.0,
+      "learning_rate": 2.998202875399361e-05,
+      "loss": 0.4896,
+      "step": 902500
+    },
+    {
+      "epoch": 27.02,
+      "learning_rate": 2.9832268370607026e-05,
+      "loss": 0.4794,
+      "step": 903000
+    },
+    {
+      "epoch": 27.03,
+      "learning_rate": 2.9682507987220446e-05,
+      "loss": 0.4838,
+      "step": 903500
+    },
+    {
+      "epoch": 27.05,
+      "learning_rate": 2.9532747603833862e-05,
+      "loss": 0.4751,
+      "step": 904000
+    },
+    {
+      "epoch": 27.06,
+      "learning_rate": 2.9382987220447285e-05,
+      "loss": 0.4841,
+      "step": 904500
+    },
+    {
+      "epoch": 27.08,
+      "learning_rate": 2.92332268370607e-05,
+      "loss": 0.4701,
+      "step": 905000
+    },
+    {
+      "epoch": 27.09,
+      "learning_rate": 2.908346645367412e-05,
+      "loss": 0.4883,
+      "step": 905500
+    },
+    {
+      "epoch": 27.11,
+      "learning_rate": 2.8933706070287536e-05,
+      "loss": 0.4777,
+      "step": 906000
+    },
+    {
+      "epoch": 27.12,
+      "learning_rate": 2.8783945686900956e-05,
+      "loss": 0.4667,
+      "step": 906500
+    },
+    {
+      "epoch": 27.14,
+      "learning_rate": 2.8634185303514375e-05,
+      "loss": 0.4768,
+      "step": 907000
+    },
+    {
+      "epoch": 27.15,
+      "learning_rate": 2.8484424920127795e-05,
+      "loss": 0.4763,
+      "step": 907500
+    },
+    {
+      "epoch": 27.17,
+      "learning_rate": 2.833466453674121e-05,
+      "loss": 0.4736,
+      "step": 908000
+    },
+    {
+      "epoch": 27.18,
+      "learning_rate": 2.818490415335463e-05,
+      "loss": 0.4783,
+      "step": 908500
+    },
+    {
+      "epoch": 27.2,
+      "learning_rate": 2.803514376996805e-05,
+      "loss": 0.481,
+      "step": 909000
+    },
+    {
+      "epoch": 27.21,
+      "learning_rate": 2.788538338658147e-05,
+      "loss": 0.4786,
+      "step": 909500
+    },
+    {
+      "epoch": 27.23,
+      "learning_rate": 2.7735623003194885e-05,
+      "loss": 0.4928,
+      "step": 910000
+    },
+    {
+      "epoch": 27.24,
+      "learning_rate": 2.7585862619808305e-05,
+      "loss": 0.4687,
+      "step": 910500
+    },
+    {
+      "epoch": 27.26,
+      "learning_rate": 2.743610223642172e-05,
+      "loss": 0.4784,
+      "step": 911000
+    },
+    {
+      "epoch": 27.27,
+      "learning_rate": 2.7286341853035143e-05,
+      "loss": 0.4705,
+      "step": 911500
+    },
+    {
+      "epoch": 27.29,
+      "learning_rate": 2.713658146964856e-05,
+      "loss": 0.4758,
+      "step": 912000
+    },
+    {
+      "epoch": 27.3,
+      "learning_rate": 2.698682108626198e-05,
+      "loss": 0.4702,
+      "step": 912500
+    },
+    {
+      "epoch": 27.32,
+      "learning_rate": 2.6837060702875395e-05,
+      "loss": 0.4761,
+      "step": 913000
+    },
+    {
+      "epoch": 27.33,
+      "learning_rate": 2.6687300319488815e-05,
+      "loss": 0.4753,
+      "step": 913500
+    },
+    {
+      "epoch": 27.35,
+      "learning_rate": 2.6537539936102234e-05,
+      "loss": 0.4776,
+      "step": 914000
+    },
+    {
+      "epoch": 27.36,
+      "learning_rate": 2.6387779552715653e-05,
+      "loss": 0.4772,
+      "step": 914500
+    },
+    {
+      "epoch": 27.38,
+      "learning_rate": 2.623801916932907e-05,
+      "loss": 0.4872,
+      "step": 915000
+    },
+    {
+      "epoch": 27.39,
+      "learning_rate": 2.608825878594249e-05,
+      "loss": 0.4725,
+      "step": 915500
+    },
+    {
+      "epoch": 27.41,
+      "learning_rate": 2.593849840255591e-05,
+      "loss": 0.4801,
+      "step": 916000
+    },
+    {
+      "epoch": 27.42,
+      "learning_rate": 2.5788738019169328e-05,
+      "loss": 0.4765,
+      "step": 916500
+    },
+    {
+      "epoch": 27.44,
+      "learning_rate": 2.5638977635782744e-05,
+      "loss": 0.475,
+      "step": 917000
+    },
+    {
+      "epoch": 27.45,
+      "learning_rate": 2.5489217252396163e-05,
+      "loss": 0.4799,
+      "step": 917500
+    },
+    {
+      "epoch": 27.47,
+      "learning_rate": 2.533945686900958e-05,
+      "loss": 0.4743,
+      "step": 918000
+    },
+    {
+      "epoch": 27.48,
+      "learning_rate": 2.5189696485623002e-05,
+      "loss": 0.4766,
+      "step": 918500
+    },
+    {
+      "epoch": 27.5,
+      "learning_rate": 2.503993610223642e-05,
+      "loss": 0.4771,
+      "step": 919000
+    },
+    {
+      "epoch": 27.51,
+      "learning_rate": 2.4890175718849838e-05,
+      "loss": 0.4715,
+      "step": 919500
+    },
+    {
+      "epoch": 27.53,
+      "learning_rate": 2.4740415335463254e-05,
+      "loss": 0.4709,
+      "step": 920000
+    },
+    {
+      "epoch": 27.54,
+      "learning_rate": 2.4590654952076677e-05,
+      "loss": 0.4803,
+      "step": 920500
+    },
+    {
+      "epoch": 27.56,
+      "learning_rate": 2.4440894568690096e-05,
+      "loss": 0.48,
+      "step": 921000
+    },
+    {
+      "epoch": 27.57,
+      "learning_rate": 2.4291134185303512e-05,
+      "loss": 0.4808,
+      "step": 921500
+    },
+    {
+      "epoch": 27.59,
+      "learning_rate": 2.414137380191693e-05,
+      "loss": 0.4841,
+      "step": 922000
+    },
+    {
+      "epoch": 27.6,
+      "learning_rate": 2.3991613418530348e-05,
+      "loss": 0.4733,
+      "step": 922500
+    },
+    {
+      "epoch": 27.62,
+      "learning_rate": 2.384185303514377e-05,
+      "loss": 0.477,
+      "step": 923000
+    },
+    {
+      "epoch": 27.63,
+      "learning_rate": 2.3692092651757187e-05,
+      "loss": 0.4886,
+      "step": 923500
+    },
+    {
+      "epoch": 27.65,
+      "learning_rate": 2.3542332268370606e-05,
+      "loss": 0.4722,
+      "step": 924000
+    },
+    {
+      "epoch": 27.66,
+      "learning_rate": 2.3392571884984022e-05,
+      "loss": 0.4783,
+      "step": 924500
+    },
+    {
+      "epoch": 27.68,
+      "learning_rate": 2.3242811501597442e-05,
+      "loss": 0.4586,
+      "step": 925000
+    },
+    {
+      "epoch": 27.69,
+      "learning_rate": 2.309305111821086e-05,
+      "loss": 0.4803,
+      "step": 925500
+    },
+    {
+      "epoch": 27.71,
+      "learning_rate": 2.294329073482428e-05,
+      "loss": 0.4693,
+      "step": 926000
+    },
+    {
+      "epoch": 27.72,
+      "learning_rate": 2.2793530351437697e-05,
+      "loss": 0.468,
+      "step": 926500
+    },
+    {
+      "epoch": 27.74,
+      "learning_rate": 2.2643769968051116e-05,
+      "loss": 0.4757,
+      "step": 927000
+    },
+    {
+      "epoch": 27.75,
+      "learning_rate": 2.2494009584664536e-05,
+      "loss": 0.4878,
+      "step": 927500
+    },
+    {
+      "epoch": 27.77,
+      "learning_rate": 2.2344249201277955e-05,
+      "loss": 0.4909,
+      "step": 928000
+    },
+    {
+      "epoch": 27.78,
+      "learning_rate": 2.219448881789137e-05,
+      "loss": 0.4907,
+      "step": 928500
+    },
+    {
+      "epoch": 27.8,
+      "learning_rate": 2.204472843450479e-05,
+      "loss": 0.4685,
+      "step": 929000
+    },
+    {
+      "epoch": 27.81,
+      "learning_rate": 2.1894968051118207e-05,
+      "loss": 0.4822,
+      "step": 929500
+    },
+    {
+      "epoch": 27.83,
+      "learning_rate": 2.174520766773163e-05,
+      "loss": 0.4942,
+      "step": 930000
+    },
+    {
+      "epoch": 27.84,
+      "learning_rate": 2.1595447284345046e-05,
+      "loss": 0.4958,
+      "step": 930500
+    },
+    {
+      "epoch": 27.86,
+      "learning_rate": 2.1445686900958465e-05,
+      "loss": 0.4744,
+      "step": 931000
+    },
+    {
+      "epoch": 27.87,
+      "learning_rate": 2.129592651757188e-05,
+      "loss": 0.4801,
+      "step": 931500
+    },
+    {
+      "epoch": 27.89,
+      "learning_rate": 2.1146166134185304e-05,
+      "loss": 0.4919,
+      "step": 932000
+    },
+    {
+      "epoch": 27.9,
+      "learning_rate": 2.099640575079872e-05,
+      "loss": 0.4751,
+      "step": 932500
+    },
+    {
+      "epoch": 27.92,
+      "learning_rate": 2.084664536741214e-05,
+      "loss": 0.4875,
+      "step": 933000
+    },
+    {
+      "epoch": 27.93,
+      "learning_rate": 2.0696884984025556e-05,
+      "loss": 0.4882,
+      "step": 933500
+    },
+    {
+      "epoch": 27.95,
+      "learning_rate": 2.0547124600638975e-05,
+      "loss": 0.4844,
+      "step": 934000
+    },
+    {
+      "epoch": 27.96,
+      "learning_rate": 2.0397364217252394e-05,
+      "loss": 0.4925,
+      "step": 934500
+    },
+    {
+      "epoch": 27.98,
+      "learning_rate": 2.0247603833865814e-05,
+      "loss": 0.4708,
+      "step": 935000
+    },
+    {
+      "epoch": 27.99,
+      "learning_rate": 2.009784345047923e-05,
+      "loss": 0.487,
+      "step": 935500
+    },
+    {
+      "epoch": 28.01,
+      "learning_rate": 1.994808306709265e-05,
+      "loss": 0.4779,
+      "step": 936000
+    },
+    {
+      "epoch": 28.02,
+      "learning_rate": 1.9798322683706066e-05,
+      "loss": 0.4721,
+      "step": 936500
+    },
+    {
+      "epoch": 28.04,
+      "learning_rate": 1.964856230031949e-05,
+      "loss": 0.4752,
+      "step": 937000
+    },
+    {
+      "epoch": 28.05,
+      "learning_rate": 1.9498801916932904e-05,
+      "loss": 0.463,
+      "step": 937500
+    },
+    {
+      "epoch": 28.07,
+      "learning_rate": 1.9349041533546324e-05,
+      "loss": 0.4547,
+      "step": 938000
+    },
+    {
+      "epoch": 28.08,
+      "learning_rate": 1.919928115015974e-05,
+      "loss": 0.4566,
+      "step": 938500
+    },
+    {
+      "epoch": 28.1,
+      "learning_rate": 1.9049520766773163e-05,
+      "loss": 0.4707,
+      "step": 939000
+    },
+    {
+      "epoch": 28.11,
+      "learning_rate": 1.8899760383386582e-05,
+      "loss": 0.4728,
+      "step": 939500
+    },
+    {
+      "epoch": 28.13,
+      "learning_rate": 1.875e-05,
+      "loss": 0.462,
+      "step": 940000
+    },
+    {
+      "epoch": 28.14,
+      "learning_rate": 1.8600239616613414e-05,
+      "loss": 0.4562,
+      "step": 940500
+    },
+    {
+      "epoch": 28.16,
+      "learning_rate": 1.8450479233226837e-05,
+      "loss": 0.4651,
+      "step": 941000
+    },
+    {
+      "epoch": 28.17,
+      "learning_rate": 1.8300718849840253e-05,
+      "loss": 0.4679,
+      "step": 941500
+    },
+    {
+      "epoch": 28.19,
+      "learning_rate": 1.8150958466453673e-05,
+      "loss": 0.4634,
+      "step": 942000
+    },
+    {
+      "epoch": 28.2,
+      "learning_rate": 1.8001198083067092e-05,
+      "loss": 0.4724,
+      "step": 942500
+    },
+    {
+      "epoch": 28.22,
+      "learning_rate": 1.7851437699680512e-05,
+      "loss": 0.4594,
+      "step": 943000
+    },
+    {
+      "epoch": 28.23,
+      "learning_rate": 1.7701677316293928e-05,
+      "loss": 0.4658,
+      "step": 943500
+    },
+    {
+      "epoch": 28.25,
+      "learning_rate": 1.7551916932907347e-05,
+      "loss": 0.4655,
+      "step": 944000
+    },
+    {
+      "epoch": 28.26,
+      "learning_rate": 1.7402156549520767e-05,
+      "loss": 0.4656,
+      "step": 944500
+    },
+    {
+      "epoch": 28.28,
+      "learning_rate": 1.7252396166134183e-05,
+      "loss": 0.4611,
+      "step": 945000
+    },
+    {
+      "epoch": 28.29,
+      "learning_rate": 1.7102635782747602e-05,
+      "loss": 0.4735,
+      "step": 945500
+    },
+    {
+      "epoch": 28.31,
+      "learning_rate": 1.695287539936102e-05,
+      "loss": 0.4737,
+      "step": 946000
+    },
+    {
+      "epoch": 28.32,
+      "learning_rate": 1.680311501597444e-05,
+      "loss": 0.4686,
+      "step": 946500
+    },
+    {
+      "epoch": 28.34,
+      "learning_rate": 1.6653354632587857e-05,
+      "loss": 0.4628,
+      "step": 947000
+    },
+    {
+      "epoch": 28.35,
+      "learning_rate": 1.6503594249201277e-05,
+      "loss": 0.4703,
+      "step": 947500
+    },
+    {
+      "epoch": 28.37,
+      "learning_rate": 1.6353833865814696e-05,
+      "loss": 0.4717,
+      "step": 948000
+    },
+    {
+      "epoch": 28.38,
+      "learning_rate": 1.6204073482428112e-05,
+      "loss": 0.4775,
+      "step": 948500
+    },
+    {
+      "epoch": 28.4,
+      "learning_rate": 1.605431309904153e-05,
+      "loss": 0.4674,
+      "step": 949000
+    },
+    {
+      "epoch": 28.41,
+      "learning_rate": 1.590455271565495e-05,
+      "loss": 0.4652,
+      "step": 949500
+    },
+    {
+      "epoch": 28.43,
+      "learning_rate": 1.575479233226837e-05,
+      "loss": 0.4692,
+      "step": 950000
+    },
+    {
+      "epoch": 28.44,
+      "learning_rate": 1.5605031948881787e-05,
+      "loss": 0.4603,
+      "step": 950500
+    },
+    {
+      "epoch": 28.46,
+      "learning_rate": 1.5455271565495206e-05,
+      "loss": 0.476,
+      "step": 951000
+    },
+    {
+      "epoch": 28.47,
+      "learning_rate": 1.5305511182108626e-05,
+      "loss": 0.4585,
+      "step": 951500
+    },
+    {
+      "epoch": 28.49,
+      "learning_rate": 1.5155750798722043e-05,
+      "loss": 0.457,
+      "step": 952000
+    },
+    {
+      "epoch": 28.5,
+      "learning_rate": 1.5005990415335463e-05,
+      "loss": 0.4599,
+      "step": 952500
+    },
+    {
+      "epoch": 28.52,
+      "learning_rate": 1.485623003194888e-05,
+      "loss": 0.4542,
+      "step": 953000
+    },
+    {
+      "epoch": 28.53,
+      "learning_rate": 1.47064696485623e-05,
+      "loss": 0.4702,
+      "step": 953500
+    },
+    {
+      "epoch": 28.55,
+      "learning_rate": 1.4556709265175718e-05,
+      "loss": 0.4616,
+      "step": 954000
+    },
+    {
+      "epoch": 28.56,
+      "learning_rate": 1.4406948881789137e-05,
+      "loss": 0.4708,
+      "step": 954500
+    },
+    {
+      "epoch": 28.58,
+      "learning_rate": 1.4257188498402555e-05,
+      "loss": 0.4719,
+      "step": 955000
+    },
+    {
+      "epoch": 28.59,
+      "learning_rate": 1.4107428115015973e-05,
+      "loss": 0.4726,
+      "step": 955500
+    },
+    {
+      "epoch": 28.61,
+      "learning_rate": 1.3957667731629392e-05,
+      "loss": 0.4578,
+      "step": 956000
+    },
+    {
+      "epoch": 28.62,
+      "learning_rate": 1.380790734824281e-05,
+      "loss": 0.4662,
+      "step": 956500
+    },
+    {
+      "epoch": 28.64,
+      "learning_rate": 1.365814696485623e-05,
+      "loss": 0.4568,
+      "step": 957000
+    },
+    {
+      "epoch": 28.65,
+      "learning_rate": 1.3508386581469647e-05,
+      "loss": 0.4701,
+      "step": 957500
+    },
+    {
+      "epoch": 28.67,
+      "learning_rate": 1.3358626198083067e-05,
+      "loss": 0.4643,
+      "step": 958000
+    },
+    {
+      "epoch": 28.68,
+      "learning_rate": 1.3208865814696484e-05,
+      "loss": 0.4605,
+      "step": 958500
+    },
+    {
+      "epoch": 28.69,
+      "learning_rate": 1.3059105431309904e-05,
+      "loss": 0.4681,
+      "step": 959000
+    },
+    {
+      "epoch": 28.71,
+      "learning_rate": 1.2909345047923322e-05,
+      "loss": 0.4645,
+      "step": 959500
+    },
+    {
+      "epoch": 28.72,
+      "learning_rate": 1.275958466453674e-05,
+      "loss": 0.4617,
+      "step": 960000
+    },
+    {
+      "epoch": 28.74,
+      "learning_rate": 1.2609824281150159e-05,
+      "loss": 0.4633,
+      "step": 960500
+    },
+    {
+      "epoch": 28.75,
+      "learning_rate": 1.2460063897763577e-05,
+      "loss": 0.4739,
+      "step": 961000
+    },
+    {
+      "epoch": 28.77,
+      "learning_rate": 1.2310303514376996e-05,
+      "loss": 0.4639,
+      "step": 961500
+    },
+    {
+      "epoch": 28.78,
+      "learning_rate": 1.2160543130990414e-05,
+      "loss": 0.4652,
+      "step": 962000
+    },
+    {
+      "epoch": 28.8,
+      "learning_rate": 1.2010782747603833e-05,
+      "loss": 0.4606,
+      "step": 962500
+    },
+    {
+      "epoch": 28.81,
+      "learning_rate": 1.1861022364217251e-05,
+      "loss": 0.4603,
+      "step": 963000
+    },
+    {
+      "epoch": 28.83,
+      "learning_rate": 1.1711261980830669e-05,
+      "loss": 0.4788,
+      "step": 963500
+    },
+    {
+      "epoch": 28.84,
+      "learning_rate": 1.1561501597444088e-05,
+      "loss": 0.458,
+      "step": 964000
+    },
+    {
+      "epoch": 28.86,
+      "learning_rate": 1.1411741214057506e-05,
+      "loss": 0.4735,
+      "step": 964500
+    },
+    {
+      "epoch": 28.87,
+      "learning_rate": 1.1261980830670926e-05,
+      "loss": 0.4561,
+      "step": 965000
+    },
+    {
+      "epoch": 28.89,
+      "learning_rate": 1.1112220447284343e-05,
+      "loss": 0.4668,
+      "step": 965500
+    },
+    {
+      "epoch": 28.9,
+      "learning_rate": 1.0962460063897763e-05,
+      "loss": 0.4776,
+      "step": 966000
+    },
+    {
+      "epoch": 28.92,
+      "learning_rate": 1.081269968051118e-05,
+      "loss": 0.4597,
+      "step": 966500
+    },
+    {
+      "epoch": 28.93,
+      "learning_rate": 1.0662939297124598e-05,
+      "loss": 0.4663,
+      "step": 967000
+    },
+    {
+      "epoch": 28.95,
+      "learning_rate": 1.0513178913738018e-05,
+      "loss": 0.4638,
+      "step": 967500
+    },
+    {
+      "epoch": 28.96,
+      "learning_rate": 1.0363418530351435e-05,
+      "loss": 0.4807,
+      "step": 968000
+    },
+    {
+      "epoch": 28.98,
+      "learning_rate": 1.0213658146964857e-05,
+      "loss": 0.4511,
+      "step": 968500
+    },
+    {
+      "epoch": 28.99,
+      "learning_rate": 1.0063897763578273e-05,
+      "loss": 0.4616,
+      "step": 969000
+    },
+    {
+      "epoch": 29.01,
+      "learning_rate": 9.914137380191694e-06,
+      "loss": 0.46,
+      "step": 969500
+    },
+    {
+      "epoch": 29.02,
+      "learning_rate": 9.764376996805112e-06,
+      "loss": 0.4528,
+      "step": 970000
+    },
+    {
+      "epoch": 29.04,
+      "learning_rate": 9.614616613418531e-06,
+      "loss": 0.4607,
+      "step": 970500
+    },
+    {
+      "epoch": 29.05,
+      "learning_rate": 9.464856230031949e-06,
+      "loss": 0.4574,
+      "step": 971000
+    },
+    {
+      "epoch": 29.07,
+      "learning_rate": 9.315095846645367e-06,
+      "loss": 0.4693,
+      "step": 971500
+    },
+    {
+      "epoch": 29.08,
+      "learning_rate": 9.165335463258786e-06,
+      "loss": 0.4492,
+      "step": 972000
+    },
+    {
+      "epoch": 29.1,
+      "learning_rate": 9.015575079872204e-06,
+      "loss": 0.4543,
+      "step": 972500
+    },
+    {
+      "epoch": 29.11,
+      "learning_rate": 8.865814696485622e-06,
+      "loss": 0.4593,
+      "step": 973000
+    },
+    {
+      "epoch": 29.13,
+      "learning_rate": 8.716054313099041e-06,
+      "loss": 0.4558,
+      "step": 973500
+    },
+    {
+      "epoch": 29.14,
+      "learning_rate": 8.566293929712459e-06,
+      "loss": 0.4522,
+      "step": 974000
+    },
+    {
+      "epoch": 29.16,
+      "learning_rate": 8.416533546325878e-06,
+      "loss": 0.4383,
+      "step": 974500
+    },
+    {
+      "epoch": 29.17,
+      "learning_rate": 8.266773162939296e-06,
+      "loss": 0.4482,
+      "step": 975000
+    },
+    {
+      "epoch": 29.19,
+      "learning_rate": 8.117012779552716e-06,
+      "loss": 0.4519,
+      "step": 975500
+    },
+    {
+      "epoch": 29.2,
+      "learning_rate": 7.967252396166133e-06,
+      "loss": 0.4584,
+      "step": 976000
+    },
+    {
+      "epoch": 29.22,
+      "learning_rate": 7.817492012779551e-06,
+      "loss": 0.4649,
+      "step": 976500
+    },
+    {
+      "epoch": 29.23,
+      "learning_rate": 7.66773162939297e-06,
+      "loss": 0.4541,
+      "step": 977000
+    },
+    {
+      "epoch": 29.25,
+      "learning_rate": 7.517971246006389e-06,
+      "loss": 0.4776,
+      "step": 977500
+    },
+    {
+      "epoch": 29.26,
+      "learning_rate": 7.368210862619808e-06,
+      "loss": 0.4529,
+      "step": 978000
+    },
+    {
+      "epoch": 29.28,
+      "learning_rate": 7.218450479233226e-06,
+      "loss": 0.4644,
+      "step": 978500
+    },
+    {
+      "epoch": 29.29,
+      "learning_rate": 7.068690095846645e-06,
+      "loss": 0.4621,
+      "step": 979000
+    },
+    {
+      "epoch": 29.31,
+      "learning_rate": 6.9189297124600636e-06,
+      "loss": 0.4485,
+      "step": 979500
+    },
+    {
+      "epoch": 29.32,
+      "learning_rate": 6.769169329073482e-06,
+      "loss": 0.4631,
+      "step": 980000
+    },
+    {
+      "epoch": 29.34,
+      "learning_rate": 6.6194089456869e-06,
+      "loss": 0.4607,
+      "step": 980500
+    },
+    {
+      "epoch": 29.35,
+      "learning_rate": 6.4696485623003185e-06,
+      "loss": 0.4559,
+      "step": 981000
+    },
+    {
+      "epoch": 29.37,
+      "learning_rate": 6.319888178913737e-06,
+      "loss": 0.4526,
+      "step": 981500
+    },
+    {
+      "epoch": 29.38,
+      "learning_rate": 6.170127795527156e-06,
+      "loss": 0.4651,
+      "step": 982000
+    },
+    {
+      "epoch": 29.4,
+      "learning_rate": 6.020367412140574e-06,
+      "loss": 0.4524,
+      "step": 982500
+    },
+    {
+      "epoch": 29.41,
+      "learning_rate": 5.870607028753993e-06,
+      "loss": 0.4374,
+      "step": 983000
+    },
+    {
+      "epoch": 29.43,
+      "learning_rate": 5.7208466453674124e-06,
+      "loss": 0.4516,
+      "step": 983500
+    },
+    {
+      "epoch": 29.44,
+      "learning_rate": 5.571086261980829e-06,
+      "loss": 0.4546,
+      "step": 984000
+    },
+    {
+      "epoch": 29.46,
+      "learning_rate": 5.421325878594248e-06,
+      "loss": 0.4571,
+      "step": 984500
+    },
+    {
+      "epoch": 29.47,
+      "learning_rate": 5.2715654952076674e-06,
+      "loss": 0.4552,
+      "step": 985000
+    },
+    {
+      "epoch": 29.49,
+      "learning_rate": 5.121805111821086e-06,
+      "loss": 0.4575,
+      "step": 985500
+    },
+    {
+      "epoch": 29.5,
+      "learning_rate": 4.972044728434505e-06,
+      "loss": 0.4429,
+      "step": 986000
+    },
+    {
+      "epoch": 29.52,
+      "learning_rate": 4.822284345047923e-06,
+      "loss": 0.464,
+      "step": 986500
+    },
+    {
+      "epoch": 29.53,
+      "learning_rate": 4.672523961661341e-06,
+      "loss": 0.4534,
+      "step": 987000
+    },
+    {
+      "epoch": 29.55,
+      "learning_rate": 4.52276357827476e-06,
+      "loss": 0.4497,
+      "step": 987500
+    },
+    {
+      "epoch": 29.56,
+      "learning_rate": 4.373003194888178e-06,
+      "loss": 0.4608,
+      "step": 988000
+    },
+    {
+      "epoch": 29.58,
+      "learning_rate": 4.223242811501598e-06,
+      "loss": 0.4706,
+      "step": 988500
+    },
+    {
+      "epoch": 29.59,
+      "learning_rate": 4.0734824281150155e-06,
+      "loss": 0.4396,
+      "step": 989000
+    },
+    {
+      "epoch": 29.61,
+      "learning_rate": 3.923722044728434e-06,
+      "loss": 0.4476,
+      "step": 989500
+    },
+    {
+      "epoch": 29.62,
+      "learning_rate": 3.7739616613418527e-06,
+      "loss": 0.4337,
+      "step": 990000
+    },
+    {
+      "epoch": 29.64,
+      "learning_rate": 3.6242012779552713e-06,
+      "loss": 0.4526,
+      "step": 990500
+    },
+    {
+      "epoch": 29.65,
+      "learning_rate": 3.4744408945686895e-06,
+      "loss": 0.4516,
+      "step": 991000
+    },
+    {
+      "epoch": 29.67,
+      "learning_rate": 3.3246805111821085e-06,
+      "loss": 0.4625,
+      "step": 991500
+    },
+    {
+      "epoch": 29.68,
+      "learning_rate": 3.174920127795527e-06,
+      "loss": 0.4573,
+      "step": 992000
+    },
+    {
+      "epoch": 29.7,
+      "learning_rate": 3.0251597444089453e-06,
+      "loss": 0.4449,
+      "step": 992500
+    },
+    {
+      "epoch": 29.71,
+      "learning_rate": 2.875399361022364e-06,
+      "loss": 0.4566,
+      "step": 993000
+    },
+    {
+      "epoch": 29.73,
+      "learning_rate": 2.7256389776357825e-06,
+      "loss": 0.4613,
+      "step": 993500
+    },
+    {
+      "epoch": 29.74,
+      "learning_rate": 2.575878594249201e-06,
+      "loss": 0.4537,
+      "step": 994000
+    },
+    {
+      "epoch": 29.76,
+      "learning_rate": 2.4261182108626193e-06,
+      "loss": 0.4538,
+      "step": 994500
+    },
+    {
+      "epoch": 29.77,
+      "learning_rate": 2.276357827476038e-06,
+      "loss": 0.4598,
+      "step": 995000
+    },
+    {
+      "epoch": 29.79,
+      "learning_rate": 2.1265974440894566e-06,
+      "loss": 0.4446,
+      "step": 995500
+    },
+    {
+      "epoch": 29.8,
+      "learning_rate": 1.976837060702875e-06,
+      "loss": 0.4605,
+      "step": 996000
+    },
+    {
+      "epoch": 29.82,
+      "learning_rate": 1.8270766773162938e-06,
+      "loss": 0.4578,
+      "step": 996500
+    },
+    {
+      "epoch": 29.83,
+      "learning_rate": 1.6773162939297122e-06,
+      "loss": 0.4654,
+      "step": 997000
+    },
+    {
+      "epoch": 29.85,
+      "learning_rate": 1.527555910543131e-06,
+      "loss": 0.4562,
+      "step": 997500
+    },
+    {
+      "epoch": 29.86,
+      "learning_rate": 1.3777955271565494e-06,
+      "loss": 0.4506,
+      "step": 998000
+    },
+    {
+      "epoch": 29.88,
+      "learning_rate": 1.228035143769968e-06,
+      "loss": 0.4784,
+      "step": 998500
+    },
+    {
+      "epoch": 29.89,
+      "learning_rate": 1.0782747603833864e-06,
+      "loss": 0.4626,
+      "step": 999000
+    },
+    {
+      "epoch": 29.91,
+      "learning_rate": 9.28514376996805e-07,
+      "loss": 0.46,
+      "step": 999500
+    },
+    {
+      "epoch": 29.92,
+      "learning_rate": 7.787539936102235e-07,
+      "loss": 0.4552,
+      "step": 1000000
+    },
+    {
+      "epoch": 29.94,
+      "learning_rate": 6.28993610223642e-07,
+      "loss": 0.4525,
+      "step": 1000500
+    },
+    {
+      "epoch": 29.95,
+      "learning_rate": 4.792332268370607e-07,
+      "loss": 0.4381,
+      "step": 1001000
+    },
+    {
+      "epoch": 29.97,
+      "learning_rate": 3.294728434504792e-07,
+      "loss": 0.4544,
+      "step": 1001500
+    },
+    {
+      "epoch": 29.98,
+      "learning_rate": 1.7971246006389775e-07,
+      "loss": 0.4555,
+      "step": 1002000
+    },
+    {
+      "epoch": 30.0,
+      "learning_rate": 2.995207667731629e-08,
+      "loss": 0.4606,
+      "step": 1002500
+    },
+    {
+      "epoch": 30.0,
+      "step": 1002600,
+      "total_flos": 2.535910859688022e+21,
+      "train_loss": 0.8395041381863332,
+      "train_runtime": 838269.5782,
+      "train_samples_per_second": 19.137,
+      "train_steps_per_second": 1.196
+    }
+  ],
+  "logging_steps": 500,
+  "max_steps": 1002600,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 30,
+  "save_steps": 500,
+  "total_flos": 2.535910859688022e+21,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}