diff --git "a/checkpoint-969194/trainer_state.json" "b/checkpoint-969194/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-969194/trainer_state.json"
@@ -0,0 +1,11649 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 28.999985039122695,
+  "eval_steps": 500,
+  "global_step": 969194,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.5e-05,
+      "loss": 30.4144,
+      "step": 500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00015,
+      "loss": 5.067,
+      "step": 1000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000225,
+      "loss": 4.7223,
+      "step": 1500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003,
+      "loss": 3.2436,
+      "step": 2000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029985008994603234,
+      "loss": 2.6616,
+      "step": 2500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029970017989206475,
+      "loss": 2.3832,
+      "step": 3000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002995502698380971,
+      "loss": 2.206,
+      "step": 3500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002994003597841295,
+      "loss": 2.1107,
+      "step": 4000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002992504497301619,
+      "loss": 2.0405,
+      "step": 4500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029910053967619426,
+      "loss": 2.0127,
+      "step": 5000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002989506296222266,
+      "loss": 1.9404,
+      "step": 5500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029880071956825903,
+      "loss": 1.9067,
+      "step": 6000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002986508095142914,
+      "loss": 1.9101,
+      "step": 6500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002985008994603238,
+      "loss": 1.8644,
+      "step": 7000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002983509894063562,
+      "loss": 1.8294,
+      "step": 7500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029820107935238854,
+      "loss": 1.811,
+      "step": 8000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002980511692984209,
+      "loss": 1.8067,
+      "step": 8500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002979012592444533,
+      "loss": 1.7794,
+      "step": 9000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002977513491904857,
+      "loss": 1.7366,
+      "step": 9500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029760143913651804,
+      "loss": 1.7349,
+      "step": 10000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029745152908255046,
+      "loss": 1.7221,
+      "step": 10500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002973016190285828,
+      "loss": 1.7171,
+      "step": 11000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029715170897461524,
+      "loss": 1.7112,
+      "step": 11500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002970017989206476,
+      "loss": 1.7109,
+      "step": 12000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029685188886667996,
+      "loss": 1.695,
+      "step": 12500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002967019788127124,
+      "loss": 1.6686,
+      "step": 13000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029655206875874474,
+      "loss": 1.6822,
+      "step": 13500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002964021587047771,
+      "loss": 1.686,
+      "step": 14000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029625224865080946,
+      "loss": 1.6344,
+      "step": 14500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002961023385968419,
+      "loss": 1.6272,
+      "step": 15000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029595242854287424,
+      "loss": 1.6296,
+      "step": 15500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002958025184889066,
+      "loss": 1.6365,
+      "step": 16000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000295652608434939,
+      "loss": 1.6434,
+      "step": 16500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002955026983809714,
+      "loss": 1.6387,
+      "step": 17000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002953527883270038,
+      "loss": 1.6126,
+      "step": 17500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00029520287827303616,
+      "loss": 1.5874,
+      "step": 18000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002950529682190685,
+      "loss": 1.6019,
+      "step": 18500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029490305816510094,
+      "loss": 1.5916,
+      "step": 19000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002947531481111333,
+      "loss": 1.607,
+      "step": 19500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029460323805716567,
+      "loss": 1.5834,
+      "step": 20000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029445332800319803,
+      "loss": 1.5662,
+      "step": 20500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029430341794923045,
+      "loss": 1.5921,
+      "step": 21000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002941535078952628,
+      "loss": 1.5616,
+      "step": 21500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00029400359784129517,
+      "loss": 1.5467,
+      "step": 22000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002938536877873276,
+      "loss": 1.5701,
+      "step": 22500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00029370377773335995,
+      "loss": 1.5428,
+      "step": 23000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00029355386767939237,
+      "loss": 1.5373,
+      "step": 23500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00029340395762542473,
+      "loss": 1.5422,
+      "step": 24000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002932540475714571,
+      "loss": 1.5579,
+      "step": 24500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002931041375174895,
+      "loss": 1.5478,
+      "step": 25000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029295422746352187,
+      "loss": 1.5409,
+      "step": 25500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00029280431740955423,
+      "loss": 1.5254,
+      "step": 26000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002926544073555866,
+      "loss": 1.5163,
+      "step": 26500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000292504497301619,
+      "loss": 1.5214,
+      "step": 27000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00029235458724765137,
+      "loss": 1.5224,
+      "step": 27500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00029220467719368373,
+      "loss": 1.52,
+      "step": 28000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00029205476713971615,
+      "loss": 1.5289,
+      "step": 28500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002919048570857485,
+      "loss": 1.4892,
+      "step": 29000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00029175494703178093,
+      "loss": 1.5054,
+      "step": 29500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002916050369778133,
+      "loss": 1.5075,
+      "step": 30000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002914551269238457,
+      "loss": 1.5178,
+      "step": 30500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00029130521686987807,
+      "loss": 1.4731,
+      "step": 31000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00029115530681591043,
+      "loss": 1.4926,
+      "step": 31500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002910053967619428,
+      "loss": 1.5006,
+      "step": 32000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002908554867079752,
+      "loss": 1.4902,
+      "step": 32500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002907055766540076,
+      "loss": 1.479,
+      "step": 33000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00029055566660003994,
+      "loss": 1.4597,
+      "step": 33500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002904057565460723,
+      "loss": 1.4347,
+      "step": 34000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002902558464921047,
+      "loss": 1.4184,
+      "step": 34500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002901059364381371,
+      "loss": 1.427,
+      "step": 35000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002899560263841695,
+      "loss": 1.416,
+      "step": 35500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00028980611633020186,
+      "loss": 1.3993,
+      "step": 36000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00028965620627623427,
+      "loss": 1.4331,
+      "step": 36500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00028950629622226663,
+      "loss": 1.4232,
+      "step": 37000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.000289356386168299,
+      "loss": 1.4256,
+      "step": 37500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00028920647611433136,
+      "loss": 1.4239,
+      "step": 38000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002890565660603638,
+      "loss": 1.4312,
+      "step": 38500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00028890665600639614,
+      "loss": 1.438,
+      "step": 39000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002887567459524285,
+      "loss": 1.4126,
+      "step": 39500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00028860683589846086,
+      "loss": 1.4097,
+      "step": 40000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002884569258444933,
+      "loss": 1.4009,
+      "step": 40500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00028830701579052564,
+      "loss": 1.4133,
+      "step": 41000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00028815710573655806,
+      "loss": 1.3991,
+      "step": 41500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002880071956825904,
+      "loss": 1.4189,
+      "step": 42000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00028785728562862284,
+      "loss": 1.3955,
+      "step": 42500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002877073755746552,
+      "loss": 1.4144,
+      "step": 43000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00028755746552068756,
+      "loss": 1.4202,
+      "step": 43500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002874075554667199,
+      "loss": 1.3879,
+      "step": 44000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00028725764541275234,
+      "loss": 1.3883,
+      "step": 44500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002871077353587847,
+      "loss": 1.3962,
+      "step": 45000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00028695782530481706,
+      "loss": 1.4029,
+      "step": 45500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00028680791525084943,
+      "loss": 1.3723,
+      "step": 46000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00028665800519688184,
+      "loss": 1.3893,
+      "step": 46500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002865080951429142,
+      "loss": 1.375,
+      "step": 47000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002863581850889466,
+      "loss": 1.3956,
+      "step": 47500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.000286208275034979,
+      "loss": 1.3975,
+      "step": 48000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002860583649810114,
+      "loss": 1.4012,
+      "step": 48500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00028590845492704376,
+      "loss": 1.3918,
+      "step": 49000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002857585448730761,
+      "loss": 1.3892,
+      "step": 49500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002856086348191085,
+      "loss": 1.4061,
+      "step": 50000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002854587247651409,
+      "loss": 1.4248,
+      "step": 50500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00028530881471117327,
+      "loss": 1.4064,
+      "step": 51000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00028515890465720563,
+      "loss": 1.4046,
+      "step": 51500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.000285008994603238,
+      "loss": 1.3945,
+      "step": 52000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002848590845492704,
+      "loss": 1.3688,
+      "step": 52500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00028470917449530277,
+      "loss": 1.413,
+      "step": 53000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002845592644413352,
+      "loss": 1.3868,
+      "step": 53500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00028440935438736755,
+      "loss": 1.3916,
+      "step": 54000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00028425944433339997,
+      "loss": 1.3999,
+      "step": 54500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00028410953427943233,
+      "loss": 1.3678,
+      "step": 55000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0002839596242254647,
+      "loss": 1.3921,
+      "step": 55500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0002838097141714971,
+      "loss": 1.3744,
+      "step": 56000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00028365980411752947,
+      "loss": 1.39,
+      "step": 56500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00028350989406356183,
+      "loss": 1.3837,
+      "step": 57000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0002833599840095942,
+      "loss": 1.3628,
+      "step": 57500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002832100739556266,
+      "loss": 1.3798,
+      "step": 58000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00028306016390165897,
+      "loss": 1.3719,
+      "step": 58500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00028291025384769133,
+      "loss": 1.3689,
+      "step": 59000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00028276034379372375,
+      "loss": 1.393,
+      "step": 59500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0002826104337397561,
+      "loss": 1.3628,
+      "step": 60000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00028246052368578853,
+      "loss": 1.3739,
+      "step": 60500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002823106136318209,
+      "loss": 1.3624,
+      "step": 61000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00028216070357785325,
+      "loss": 1.3475,
+      "step": 61500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00028201079352388567,
+      "loss": 1.3492,
+      "step": 62000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00028186088346991803,
+      "loss": 1.3637,
+      "step": 62500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002817109734159504,
+      "loss": 1.3515,
+      "step": 63000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00028156106336198276,
+      "loss": 1.3759,
+      "step": 63500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002814111533080152,
+      "loss": 1.368,
+      "step": 64000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00028126124325404754,
+      "loss": 1.3684,
+      "step": 64500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002811113332000799,
+      "loss": 1.3591,
+      "step": 65000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002809614231461123,
+      "loss": 1.3438,
+      "step": 65500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002808115130921447,
+      "loss": 1.3392,
+      "step": 66000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0002806616030381771,
+      "loss": 1.3517,
+      "step": 66500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00028051169298420946,
+      "loss": 1.3496,
+      "step": 67000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0002803617829302418,
+      "loss": 1.2833,
+      "step": 67500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00028021187287627424,
+      "loss": 1.3091,
+      "step": 68000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0002800619628223066,
+      "loss": 1.2997,
+      "step": 68500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00027991205276833896,
+      "loss": 1.2987,
+      "step": 69000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0002797621427143713,
+      "loss": 1.2982,
+      "step": 69500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00027961223266040374,
+      "loss": 1.2895,
+      "step": 70000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0002794623226064361,
+      "loss": 1.3113,
+      "step": 70500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00027931241255246846,
+      "loss": 1.3079,
+      "step": 71000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0002791625024985009,
+      "loss": 1.3029,
+      "step": 71500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00027901259244453324,
+      "loss": 1.3064,
+      "step": 72000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00027886268239056566,
+      "loss": 1.2945,
+      "step": 72500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.000278712772336598,
+      "loss": 1.3019,
+      "step": 73000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0002785628622826304,
+      "loss": 1.3151,
+      "step": 73500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0002784129522286628,
+      "loss": 1.3004,
+      "step": 74000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00027826304217469516,
+      "loss": 1.2726,
+      "step": 74500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0002781131321207275,
+      "loss": 1.3012,
+      "step": 75000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0002779632220667599,
+      "loss": 1.2726,
+      "step": 75500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0002778133120127923,
+      "loss": 1.2873,
+      "step": 76000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00027766340195882467,
+      "loss": 1.3153,
+      "step": 76500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0002775134919048571,
+      "loss": 1.3091,
+      "step": 77000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00027736358185088944,
+      "loss": 1.3058,
+      "step": 77500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0002772136717969218,
+      "loss": 1.2905,
+      "step": 78000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0002770637617429542,
+      "loss": 1.2976,
+      "step": 78500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0002769138516889866,
+      "loss": 1.3113,
+      "step": 79000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00027676394163501895,
+      "loss": 1.3111,
+      "step": 79500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00027661403158105136,
+      "loss": 1.27,
+      "step": 80000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0002764641215270837,
+      "loss": 1.2939,
+      "step": 80500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0002763142114731161,
+      "loss": 1.3067,
+      "step": 81000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00027616430141914845,
+      "loss": 1.2816,
+      "step": 81500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00027601439136518087,
+      "loss": 1.2998,
+      "step": 82000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00027586448131121323,
+      "loss": 1.2766,
+      "step": 82500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00027571457125724565,
+      "loss": 1.2851,
+      "step": 83000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.000275564661203278,
+      "loss": 1.2945,
+      "step": 83500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00027541475114931037,
+      "loss": 1.2823,
+      "step": 84000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0002752648410953428,
+      "loss": 1.2745,
+      "step": 84500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00027511493104137515,
+      "loss": 1.2625,
+      "step": 85000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00027496502098740757,
+      "loss": 1.2775,
+      "step": 85500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00027481511093343993,
+      "loss": 1.2713,
+      "step": 86000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0002746652008794723,
+      "loss": 1.2873,
+      "step": 86500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00027451529082550465,
+      "loss": 1.2731,
+      "step": 87000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00027436538077153707,
+      "loss": 1.2847,
+      "step": 87500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00027421547071756943,
+      "loss": 1.2885,
+      "step": 88000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0002740655606636018,
+      "loss": 1.2725,
+      "step": 88500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0002739156506096342,
+      "loss": 1.2936,
+      "step": 89000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00027376574055566657,
+      "loss": 1.2674,
+      "step": 89500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.000273615830501699,
+      "loss": 1.2706,
+      "step": 90000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00027346592044773135,
+      "loss": 1.2623,
+      "step": 90500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0002733160103937637,
+      "loss": 1.2824,
+      "step": 91000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00027316610033979613,
+      "loss": 1.2823,
+      "step": 91500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0002730161902858285,
+      "loss": 1.2593,
+      "step": 92000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00027286628023186085,
+      "loss": 1.2816,
+      "step": 92500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0002727163701778932,
+      "loss": 1.2774,
+      "step": 93000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00027256646012392563,
+      "loss": 1.2799,
+      "step": 93500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.000272416550069958,
+      "loss": 1.2415,
+      "step": 94000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00027226664001599036,
+      "loss": 1.265,
+      "step": 94500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0002721167299620228,
+      "loss": 1.2706,
+      "step": 95000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00027196681990805514,
+      "loss": 1.2417,
+      "step": 95500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00027181690985408755,
+      "loss": 1.271,
+      "step": 96000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0002716669998001199,
+      "loss": 1.248,
+      "step": 96500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0002715170897461523,
+      "loss": 1.2573,
+      "step": 97000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0002713671796921847,
+      "loss": 1.2717,
+      "step": 97500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00027121726963821706,
+      "loss": 1.2769,
+      "step": 98000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0002710673595842494,
+      "loss": 1.2758,
+      "step": 98500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0002709174495302818,
+      "loss": 1.2814,
+      "step": 99000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0002707675394763142,
+      "loss": 1.262,
+      "step": 99500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00027061762942234656,
+      "loss": 1.2618,
+      "step": 100000
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0002704677193683789,
+      "loss": 1.2202,
+      "step": 100500
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00027031780931441134,
+      "loss": 1.2024,
+      "step": 101000
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0002701678992604437,
+      "loss": 1.1845,
+      "step": 101500
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0002700179892064761,
+      "loss": 1.219,
+      "step": 102000
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0002698680791525085,
+      "loss": 1.1941,
+      "step": 102500
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00026971816909854084,
+      "loss": 1.2169,
+      "step": 103000
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00026956825904457326,
+      "loss": 1.2369,
+      "step": 103500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0002694183489906056,
+      "loss": 1.2122,
+      "step": 104000
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.000269268438936638,
+      "loss": 1.2156,
+      "step": 104500
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00026911852888267035,
+      "loss": 1.209,
+      "step": 105000
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00026896861882870276,
+      "loss": 1.2089,
+      "step": 105500
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0002688187087747351,
+      "loss": 1.2065,
+      "step": 106000
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0002686687987207675,
+      "loss": 1.2393,
+      "step": 106500
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0002685188886667999,
+      "loss": 1.212,
+      "step": 107000
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00026836897861283227,
+      "loss": 1.2137,
+      "step": 107500
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0002682190685588647,
+      "loss": 1.2214,
+      "step": 108000
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00026806915850489704,
+      "loss": 1.2086,
+      "step": 108500
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0002679192484509294,
+      "loss": 1.2069,
+      "step": 109000
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0002677693383969618,
+      "loss": 1.2314,
+      "step": 109500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0002676194283429942,
+      "loss": 1.211,
+      "step": 110000
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00026746951828902655,
+      "loss": 1.257,
+      "step": 110500
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00026731960823505896,
+      "loss": 1.1988,
+      "step": 111000
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0002671696981810913,
+      "loss": 1.2205,
+      "step": 111500
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0002670197881271237,
+      "loss": 1.2193,
+      "step": 112000
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00026686987807315605,
+      "loss": 1.2047,
+      "step": 112500
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00026671996801918847,
+      "loss": 1.2171,
+      "step": 113000
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00026657005796522083,
+      "loss": 1.2197,
+      "step": 113500
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00026642014791125325,
+      "loss": 1.244,
+      "step": 114000
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0002662702378572856,
+      "loss": 1.2117,
+      "step": 114500
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.000266120327803318,
+      "loss": 1.2069,
+      "step": 115000
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0002659704177493504,
+      "loss": 1.2264,
+      "step": 115500
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00026582050769538275,
+      "loss": 1.2018,
+      "step": 116000
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0002656705976414151,
+      "loss": 1.2016,
+      "step": 116500
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00026552068758744753,
+      "loss": 1.2417,
+      "step": 117000
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0002653707775334799,
+      "loss": 1.2086,
+      "step": 117500
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00026522086747951225,
+      "loss": 1.2082,
+      "step": 118000
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0002650709574255446,
+      "loss": 1.2135,
+      "step": 118500
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00026492104737157703,
+      "loss": 1.2326,
+      "step": 119000
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0002647711373176094,
+      "loss": 1.1959,
+      "step": 119500
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0002646212272636418,
+      "loss": 1.2416,
+      "step": 120000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0002644713172096742,
+      "loss": 1.23,
+      "step": 120500
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0002643214071557066,
+      "loss": 1.2089,
+      "step": 121000
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00026417149710173895,
+      "loss": 1.2097,
+      "step": 121500
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0002640215870477713,
+      "loss": 1.1949,
+      "step": 122000
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0002638716769938037,
+      "loss": 1.2055,
+      "step": 122500
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0002637217669398361,
+      "loss": 1.2159,
+      "step": 123000
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00026357185688586846,
+      "loss": 1.2248,
+      "step": 123500
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0002634219468319008,
+      "loss": 1.2073,
+      "step": 124000
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0002632720367779332,
+      "loss": 1.1995,
+      "step": 124500
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0002631221267239656,
+      "loss": 1.2039,
+      "step": 125000
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00026297221666999796,
+      "loss": 1.1936,
+      "step": 125500
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0002628223066160304,
+      "loss": 1.2106,
+      "step": 126000
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00026267239656206274,
+      "loss": 1.2228,
+      "step": 126500
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00026252248650809515,
+      "loss": 1.219,
+      "step": 127000
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0002623725764541275,
+      "loss": 1.2075,
+      "step": 127500
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0002622226664001599,
+      "loss": 1.1894,
+      "step": 128000
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00026207275634619224,
+      "loss": 1.2125,
+      "step": 128500
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00026192284629222466,
+      "loss": 1.1979,
+      "step": 129000
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.000261772936238257,
+      "loss": 1.1828,
+      "step": 129500
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0002616230261842894,
+      "loss": 1.1993,
+      "step": 130000
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00026147311613032174,
+      "loss": 1.2015,
+      "step": 130500
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00026132320607635416,
+      "loss": 1.1925,
+      "step": 131000
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0002611732960223865,
+      "loss": 1.2037,
+      "step": 131500
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00026102338596841894,
+      "loss": 1.2205,
+      "step": 132000
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0002608734759144513,
+      "loss": 1.1913,
+      "step": 132500
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0002607235658604837,
+      "loss": 1.2173,
+      "step": 133000
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0002605736558065161,
+      "loss": 1.1957,
+      "step": 133500
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00026042374575254844,
+      "loss": 1.1741,
+      "step": 134000
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0002602738356985808,
+      "loss": 1.1411,
+      "step": 134500
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0002601239256446132,
+      "loss": 1.1493,
+      "step": 135000
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0002599740155906456,
+      "loss": 1.133,
+      "step": 135500
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00025982410553667795,
+      "loss": 1.1409,
+      "step": 136000
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00025967419548271036,
+      "loss": 1.139,
+      "step": 136500
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0002595242854287427,
+      "loss": 1.1655,
+      "step": 137000
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0002593743753747751,
+      "loss": 1.1608,
+      "step": 137500
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0002592244653208075,
+      "loss": 1.1603,
+      "step": 138000
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00025907455526683987,
+      "loss": 1.1399,
+      "step": 138500
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0002589246452128723,
+      "loss": 1.1543,
+      "step": 139000
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00025877473515890464,
+      "loss": 1.144,
+      "step": 139500
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.000258624825104937,
+      "loss": 1.1363,
+      "step": 140000
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0002584749150509694,
+      "loss": 1.1544,
+      "step": 140500
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0002583250049970018,
+      "loss": 1.169,
+      "step": 141000
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00025817509494303415,
+      "loss": 1.1481,
+      "step": 141500
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.0002580251848890665,
+      "loss": 1.1531,
+      "step": 142000
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0002578752748350989,
+      "loss": 1.1625,
+      "step": 142500
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0002577253647811313,
+      "loss": 1.1779,
+      "step": 143000
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00025757545472716365,
+      "loss": 1.1435,
+      "step": 143500
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00025742554467319607,
+      "loss": 1.151,
+      "step": 144000
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00025727563461922843,
+      "loss": 1.1468,
+      "step": 144500
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00025712572456526085,
+      "loss": 1.1577,
+      "step": 145000
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.0002569758145112932,
+      "loss": 1.1787,
+      "step": 145500
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00025682590445732557,
+      "loss": 1.1418,
+      "step": 146000
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.000256675994403358,
+      "loss": 1.1805,
+      "step": 146500
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00025652608434939035,
+      "loss": 1.1471,
+      "step": 147000
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.0002563761742954227,
+      "loss": 1.1518,
+      "step": 147500
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.0002562262642414551,
+      "loss": 1.1567,
+      "step": 148000
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.0002560763541874875,
+      "loss": 1.1673,
+      "step": 148500
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00025592644413351985,
+      "loss": 1.1741,
+      "step": 149000
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.0002557765340795522,
+      "loss": 1.1622,
+      "step": 149500
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00025562662402558463,
+      "loss": 1.1753,
+      "step": 150000
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.000255476713971617,
+      "loss": 1.1571,
+      "step": 150500
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.0002553268039176494,
+      "loss": 1.1766,
+      "step": 151000
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.0002551768938636818,
+      "loss": 1.1542,
+      "step": 151500
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00025502698380971414,
+      "loss": 1.1719,
+      "step": 152000
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00025487707375574655,
+      "loss": 1.1891,
+      "step": 152500
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.0002547271637017789,
+      "loss": 1.1458,
+      "step": 153000
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.0002545772536478113,
+      "loss": 1.1779,
+      "step": 153500
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00025442734359384364,
+      "loss": 1.1506,
+      "step": 154000
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00025427743353987606,
+      "loss": 1.1356,
+      "step": 154500
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.0002541275234859084,
+      "loss": 1.1545,
+      "step": 155000
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00025397761343194083,
+      "loss": 1.1592,
+      "step": 155500
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.0002538277033779732,
+      "loss": 1.1687,
+      "step": 156000
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00025367779332400556,
+      "loss": 1.165,
+      "step": 156500
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.000253527883270038,
+      "loss": 1.1673,
+      "step": 157000
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00025337797321607034,
+      "loss": 1.1652,
+      "step": 157500
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0002532280631621027,
+      "loss": 1.1822,
+      "step": 158000
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0002530781531081351,
+      "loss": 1.1549,
+      "step": 158500
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.0002529282430541675,
+      "loss": 1.1391,
+      "step": 159000
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00025277833300019984,
+      "loss": 1.1788,
+      "step": 159500
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.0002526284229462322,
+      "loss": 1.1718,
+      "step": 160000
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.0002524785128922646,
+      "loss": 1.1428,
+      "step": 160500
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.000252328602838297,
+      "loss": 1.1427,
+      "step": 161000
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.0002521786927843294,
+      "loss": 1.1297,
+      "step": 161500
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00025202878273036176,
+      "loss": 1.1884,
+      "step": 162000
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0002518788726763941,
+      "loss": 1.1633,
+      "step": 162500
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00025172896262242654,
+      "loss": 1.167,
+      "step": 163000
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.0002515790525684589,
+      "loss": 1.1662,
+      "step": 163500
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.0002514291425144913,
+      "loss": 1.167,
+      "step": 164000
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.0002512792324605237,
+      "loss": 1.1626,
+      "step": 164500
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00025112932240655604,
+      "loss": 1.1482,
+      "step": 165000
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0002509794123525884,
+      "loss": 1.1431,
+      "step": 165500
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0002508295022986208,
+      "loss": 1.1669,
+      "step": 166000
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.0002506795922446532,
+      "loss": 1.1443,
+      "step": 166500
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00025052968219068555,
+      "loss": 1.1483,
+      "step": 167000
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00025037977213671796,
+      "loss": 1.1247,
+      "step": 167500
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0002502298620827503,
+      "loss": 1.1053,
+      "step": 168000
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00025007995202878274,
+      "loss": 1.0971,
+      "step": 168500
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0002499300419748151,
+      "loss": 1.0936,
+      "step": 169000
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00024978013192084747,
+      "loss": 1.1065,
+      "step": 169500
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.0002496302218668799,
+      "loss": 1.1023,
+      "step": 170000
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00024948031181291225,
+      "loss": 1.0882,
+      "step": 170500
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.0002493304017589446,
+      "loss": 1.105,
+      "step": 171000
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00024918049170497697,
+      "loss": 1.0959,
+      "step": 171500
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.0002490305816510094,
+      "loss": 1.1116,
+      "step": 172000
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00024888067159704175,
+      "loss": 1.1049,
+      "step": 172500
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0002487307615430741,
+      "loss": 1.085,
+      "step": 173000
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00024858085148910653,
+      "loss": 1.093,
+      "step": 173500
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.0002484309414351389,
+      "loss": 1.1143,
+      "step": 174000
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.0002482810313811713,
+      "loss": 1.101,
+      "step": 174500
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00024813112132720367,
+      "loss": 1.115,
+      "step": 175000
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00024798121127323603,
+      "loss": 1.096,
+      "step": 175500
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00024783130121926845,
+      "loss": 1.0904,
+      "step": 176000
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.0002476813911653008,
+      "loss": 1.1095,
+      "step": 176500
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00024753148111133317,
+      "loss": 1.137,
+      "step": 177000
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00024738157105736553,
+      "loss": 1.1363,
+      "step": 177500
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00024723166100339795,
+      "loss": 1.1347,
+      "step": 178000
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.0002470817509494303,
+      "loss": 1.0815,
+      "step": 178500
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.0002469318408954627,
+      "loss": 1.1037,
+      "step": 179000
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.0002467819308414951,
+      "loss": 1.1074,
+      "step": 179500
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00024663202078752745,
+      "loss": 1.1297,
+      "step": 180000
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00024648211073355987,
+      "loss": 1.1063,
+      "step": 180500
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00024633220067959223,
+      "loss": 1.1097,
+      "step": 181000
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.0002461822906256246,
+      "loss": 1.1157,
+      "step": 181500
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.000246032380571657,
+      "loss": 1.1281,
+      "step": 182000
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.0002458824705176894,
+      "loss": 1.098,
+      "step": 182500
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00024573256046372174,
+      "loss": 1.1329,
+      "step": 183000
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.0002455826504097541,
+      "loss": 1.1089,
+      "step": 183500
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.0002454327403557865,
+      "loss": 1.105,
+      "step": 184000
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.0002452828303018189,
+      "loss": 1.1242,
+      "step": 184500
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00024513292024785124,
+      "loss": 1.0894,
+      "step": 185000
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00024498301019388366,
+      "loss": 1.1279,
+      "step": 185500
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.000244833100139916,
+      "loss": 1.1087,
+      "step": 186000
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00024468319008594843,
+      "loss": 1.1328,
+      "step": 186500
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.0002445332800319808,
+      "loss": 1.1211,
+      "step": 187000
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00024438336997801316,
+      "loss": 1.1238,
+      "step": 187500
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.0002442334599240456,
+      "loss": 1.1188,
+      "step": 188000
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00024408354987007794,
+      "loss": 1.1226,
+      "step": 188500
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.0002439336398161103,
+      "loss": 1.0867,
+      "step": 189000
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.0002437837297621427,
+      "loss": 1.1203,
+      "step": 189500
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00024363381970817508,
+      "loss": 1.1332,
+      "step": 190000
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00024348390965420744,
+      "loss": 1.1316,
+      "step": 190500
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00024333399960023983,
+      "loss": 1.1005,
+      "step": 191000
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00024318408954627222,
+      "loss": 1.1159,
+      "step": 191500
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.0002430341794923046,
+      "loss": 1.1314,
+      "step": 192000
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00024288426943833697,
+      "loss": 1.1039,
+      "step": 192500
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00024273435938436936,
+      "loss": 1.1146,
+      "step": 193000
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00024258444933040175,
+      "loss": 1.124,
+      "step": 193500
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.0002424345392764341,
+      "loss": 1.0976,
+      "step": 194000
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.0002422846292224665,
+      "loss": 1.1254,
+      "step": 194500
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00024213471916849886,
+      "loss": 1.1296,
+      "step": 195000
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00024198480911453128,
+      "loss": 1.1354,
+      "step": 195500
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00024183489906056364,
+      "loss": 1.1201,
+      "step": 196000
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00024168498900659603,
+      "loss": 1.1199,
+      "step": 196500
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.0002415350789526284,
+      "loss": 1.1305,
+      "step": 197000
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00024138516889866078,
+      "loss": 1.1289,
+      "step": 197500
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00024123525884469317,
+      "loss": 1.1193,
+      "step": 198000
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00024108534879072554,
+      "loss": 1.127,
+      "step": 198500
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00024093543873675793,
+      "loss": 1.1204,
+      "step": 199000
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00024078552868279031,
+      "loss": 1.1143,
+      "step": 199500
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00024063561862882268,
+      "loss": 1.1198,
+      "step": 200000
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00024048570857485507,
+      "loss": 1.1097,
+      "step": 200500
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00024033579852088743,
+      "loss": 1.0679,
+      "step": 201000
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00024018588846691985,
+      "loss": 1.0696,
+      "step": 201500
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.0002400359784129522,
+      "loss": 1.0768,
+      "step": 202000
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.0002398860683589846,
+      "loss": 1.0765,
+      "step": 202500
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00023973615830501696,
+      "loss": 1.0638,
+      "step": 203000
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00023958624825104935,
+      "loss": 1.0645,
+      "step": 203500
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00023943633819708174,
+      "loss": 1.077,
+      "step": 204000
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.0002392864281431141,
+      "loss": 1.0588,
+      "step": 204500
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.0002391365180891465,
+      "loss": 1.0927,
+      "step": 205000
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00023898660803517888,
+      "loss": 1.0748,
+      "step": 205500
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00023883669798121127,
+      "loss": 1.0775,
+      "step": 206000
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00023868678792724363,
+      "loss": 1.0952,
+      "step": 206500
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.000238536877873276,
+      "loss": 1.0825,
+      "step": 207000
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.0002383869678193084,
+      "loss": 1.0919,
+      "step": 207500
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00023823705776534077,
+      "loss": 1.08,
+      "step": 208000
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00023808714771137316,
+      "loss": 1.0646,
+      "step": 208500
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00023793723765740552,
+      "loss": 1.0872,
+      "step": 209000
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00023778732760343794,
+      "loss": 1.071,
+      "step": 209500
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.0002376374175494703,
+      "loss": 1.0879,
+      "step": 210000
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00023748750749550266,
+      "loss": 1.1079,
+      "step": 210500
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00023733759744153505,
+      "loss": 1.069,
+      "step": 211000
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00023718768738756744,
+      "loss": 1.0817,
+      "step": 211500
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00023703777733359983,
+      "loss": 1.1054,
+      "step": 212000
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.0002368878672796322,
+      "loss": 1.0796,
+      "step": 212500
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00023673795722566456,
+      "loss": 1.0909,
+      "step": 213000
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00023658804717169697,
+      "loss": 1.0552,
+      "step": 213500
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00023643813711772934,
+      "loss": 1.0737,
+      "step": 214000
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00023628822706376173,
+      "loss": 1.0872,
+      "step": 214500
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.0002361383170097941,
+      "loss": 1.0796,
+      "step": 215000
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.0002359884069558265,
+      "loss": 1.0838,
+      "step": 215500
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00023583849690185887,
+      "loss": 1.0849,
+      "step": 216000
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00023568858684789123,
+      "loss": 1.0451,
+      "step": 216500
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00023553867679392362,
+      "loss": 1.1126,
+      "step": 217000
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.000235388766739956,
+      "loss": 1.0708,
+      "step": 217500
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.0002352388566859884,
+      "loss": 1.0912,
+      "step": 218000
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00023508894663202076,
+      "loss": 1.0902,
+      "step": 218500
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00023493903657805318,
+      "loss": 1.0682,
+      "step": 219000
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00023478912652408554,
+      "loss": 1.0921,
+      "step": 219500
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.0002346392164701179,
+      "loss": 1.0653,
+      "step": 220000
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.0002344893064161503,
+      "loss": 1.1109,
+      "step": 220500
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00023433939636218268,
+      "loss": 1.0818,
+      "step": 221000
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00023418948630821507,
+      "loss": 1.0729,
+      "step": 221500
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00023403957625424743,
+      "loss": 1.08,
+      "step": 222000
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.0002338896662002798,
+      "loss": 1.0888,
+      "step": 222500
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.0002337397561463122,
+      "loss": 1.1045,
+      "step": 223000
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00023358984609234457,
+      "loss": 1.0762,
+      "step": 223500
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00023343993603837696,
+      "loss": 1.0806,
+      "step": 224000
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00023329002598440932,
+      "loss": 1.1107,
+      "step": 224500
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00023314011593044174,
+      "loss": 1.0842,
+      "step": 225000
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.0002329902058764741,
+      "loss": 1.0825,
+      "step": 225500
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00023284029582250646,
+      "loss": 1.1027,
+      "step": 226000
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00023269038576853885,
+      "loss": 1.0718,
+      "step": 226500
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00023254047571457124,
+      "loss": 1.1151,
+      "step": 227000
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00023239056566060363,
+      "loss": 1.0906,
+      "step": 227500
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.000232240655606636,
+      "loss": 1.1104,
+      "step": 228000
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00023209074555266836,
+      "loss": 1.0796,
+      "step": 228500
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00023194083549870077,
+      "loss": 1.0795,
+      "step": 229000
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00023179092544473314,
+      "loss": 1.0719,
+      "step": 229500
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00023164101539076553,
+      "loss": 1.073,
+      "step": 230000
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.0002314911053367979,
+      "loss": 1.065,
+      "step": 230500
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.0002313411952828303,
+      "loss": 1.1047,
+      "step": 231000
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00023119128522886267,
+      "loss": 1.0833,
+      "step": 231500
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00023104137517489503,
+      "loss": 1.0887,
+      "step": 232000
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00023089146512092742,
+      "loss": 1.0704,
+      "step": 232500
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.0002307415550669598,
+      "loss": 1.0901,
+      "step": 233000
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.0002305916450129922,
+      "loss": 1.074,
+      "step": 233500
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00023044173495902456,
+      "loss": 1.068,
+      "step": 234000
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 0.00023029182490505692,
+      "loss": 1.0309,
+      "step": 234500
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 0.00023014191485108934,
+      "loss": 1.0351,
+      "step": 235000
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 0.0002299920047971217,
+      "loss": 1.0033,
+      "step": 235500
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 0.0002298420947431541,
+      "loss": 1.0328,
+      "step": 236000
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 0.00022969218468918645,
+      "loss": 1.0369,
+      "step": 236500
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 0.00022954227463521887,
+      "loss": 1.0205,
+      "step": 237000
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 0.00022939236458125123,
+      "loss": 1.0258,
+      "step": 237500
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 0.0002292424545272836,
+      "loss": 1.042,
+      "step": 238000
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 0.00022909254447331598,
+      "loss": 1.0626,
+      "step": 238500
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 0.00022894263441934837,
+      "loss": 1.0357,
+      "step": 239000
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 0.00022879272436538076,
+      "loss": 1.0716,
+      "step": 239500
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 0.00022864281431141312,
+      "loss": 1.0499,
+      "step": 240000
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 0.00022849290425744549,
+      "loss": 1.0401,
+      "step": 240500
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 0.0002283429942034779,
+      "loss": 1.0554,
+      "step": 241000
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 0.00022819308414951027,
+      "loss": 1.0459,
+      "step": 241500
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 0.00022804317409554265,
+      "loss": 1.0593,
+      "step": 242000
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 0.00022789326404157502,
+      "loss": 1.0419,
+      "step": 242500
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 0.00022774335398760743,
+      "loss": 1.0745,
+      "step": 243000
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 0.0002275934439336398,
+      "loss": 1.0466,
+      "step": 243500
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 0.00022744353387967216,
+      "loss": 1.0444,
+      "step": 244000
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 0.00022729362382570457,
+      "loss": 1.0562,
+      "step": 244500
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 0.00022714371377173694,
+      "loss": 1.0787,
+      "step": 245000
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 0.00022699380371776933,
+      "loss": 1.047,
+      "step": 245500
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 0.0002268438936638017,
+      "loss": 1.0636,
+      "step": 246000
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 0.0002266939836098341,
+      "loss": 1.0331,
+      "step": 246500
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 0.00022654407355586647,
+      "loss": 1.055,
+      "step": 247000
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 0.00022639416350189883,
+      "loss": 1.0523,
+      "step": 247500
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 0.00022624425344793122,
+      "loss": 1.0567,
+      "step": 248000
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 0.0002260943433939636,
+      "loss": 1.0808,
+      "step": 248500
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 0.000225944433339996,
+      "loss": 1.0701,
+      "step": 249000
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 0.00022579452328602836,
+      "loss": 1.0627,
+      "step": 249500
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 0.00022564461323206072,
+      "loss": 1.0374,
+      "step": 250000
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 0.00022549470317809314,
+      "loss": 1.052,
+      "step": 250500
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 0.0002253447931241255,
+      "loss": 1.0602,
+      "step": 251000
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 0.0002251948830701579,
+      "loss": 1.0607,
+      "step": 251500
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 0.00022504497301619025,
+      "loss": 1.04,
+      "step": 252000
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 0.00022489506296222267,
+      "loss": 1.0593,
+      "step": 252500
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 0.00022474515290825503,
+      "loss": 1.0523,
+      "step": 253000
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 0.0002245952428542874,
+      "loss": 1.039,
+      "step": 253500
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 0.00022444533280031978,
+      "loss": 1.0631,
+      "step": 254000
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 0.00022429542274635217,
+      "loss": 1.0646,
+      "step": 254500
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 0.00022414551269238456,
+      "loss": 1.0412,
+      "step": 255000
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 0.00022399560263841692,
+      "loss": 1.0458,
+      "step": 255500
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 0.0002238456925844493,
+      "loss": 1.0551,
+      "step": 256000
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 0.0002236957825304817,
+      "loss": 1.0807,
+      "step": 256500
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 0.00022354587247651407,
+      "loss": 1.0642,
+      "step": 257000
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 0.00022339596242254645,
+      "loss": 1.0652,
+      "step": 257500
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 0.00022324605236857882,
+      "loss": 1.0507,
+      "step": 258000
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 0.00022309614231461123,
+      "loss": 1.059,
+      "step": 258500
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 0.0002229462322606436,
+      "loss": 1.0689,
+      "step": 259000
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 0.00022279632220667596,
+      "loss": 1.0624,
+      "step": 259500
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 0.00022264641215270835,
+      "loss": 1.0385,
+      "step": 260000
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 0.00022249650209874074,
+      "loss": 1.0617,
+      "step": 260500
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 0.00022234659204477313,
+      "loss": 1.0405,
+      "step": 261000
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 0.0002221966819908055,
+      "loss": 1.0557,
+      "step": 261500
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 0.00022204677193683785,
+      "loss": 1.0661,
+      "step": 262000
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 0.00022189686188287027,
+      "loss": 1.0588,
+      "step": 262500
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 0.00022174695182890263,
+      "loss": 1.0534,
+      "step": 263000
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 0.00022159704177493502,
+      "loss": 1.0431,
+      "step": 263500
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 0.00022144713172096738,
+      "loss": 1.0428,
+      "step": 264000
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 0.0002212972216669998,
+      "loss": 1.0493,
+      "step": 264500
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 0.00022114731161303216,
+      "loss": 1.0231,
+      "step": 265000
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 0.00022099740155906452,
+      "loss": 1.0692,
+      "step": 265500
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 0.0002208474915050969,
+      "loss": 1.0527,
+      "step": 266000
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 0.0002206975814511293,
+      "loss": 1.0426,
+      "step": 266500
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 0.0002205476713971617,
+      "loss": 1.0317,
+      "step": 267000
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 0.00022039776134319405,
+      "loss": 1.0274,
+      "step": 267500
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 0.00022024785128922644,
+      "loss": 1.0115,
+      "step": 268000
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 0.00022009794123525883,
+      "loss": 1.0309,
+      "step": 268500
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 0.0002199480311812912,
+      "loss": 0.9979,
+      "step": 269000
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 0.00021979812112732358,
+      "loss": 1.0112,
+      "step": 269500
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 0.00021964821107335595,
+      "loss": 0.9851,
+      "step": 270000
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 0.00021949830101938836,
+      "loss": 1.0091,
+      "step": 270500
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 0.00021934839096542072,
+      "loss": 1.0126,
+      "step": 271000
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 0.00021919848091145311,
+      "loss": 0.9994,
+      "step": 271500
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 0.0002190485708574855,
+      "loss": 1.0088,
+      "step": 272000
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 0.00021889866080351787,
+      "loss": 1.0244,
+      "step": 272500
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 0.00021874875074955025,
+      "loss": 1.0268,
+      "step": 273000
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 0.00021859884069558262,
+      "loss": 1.0217,
+      "step": 273500
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 0.00021844893064161503,
+      "loss": 1.0068,
+      "step": 274000
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 0.0002182990205876474,
+      "loss": 1.0047,
+      "step": 274500
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 0.00021814911053367979,
+      "loss": 0.9962,
+      "step": 275000
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 0.00021799920047971215,
+      "loss": 1.0165,
+      "step": 275500
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 0.00021784929042574454,
+      "loss": 1.028,
+      "step": 276000
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 0.00021769938037177693,
+      "loss": 1.0038,
+      "step": 276500
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 0.0002175494703178093,
+      "loss": 1.0408,
+      "step": 277000
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 0.00021739956026384168,
+      "loss": 1.0173,
+      "step": 277500
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 0.00021724965020987407,
+      "loss": 1.0205,
+      "step": 278000
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 0.00021709974015590643,
+      "loss": 1.0113,
+      "step": 278500
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 0.00021694983010193882,
+      "loss": 0.9955,
+      "step": 279000
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 0.00021679992004797118,
+      "loss": 1.0178,
+      "step": 279500
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 0.0002166500099940036,
+      "loss": 1.0103,
+      "step": 280000
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 0.00021650009994003596,
+      "loss": 1.0114,
+      "step": 280500
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 0.00021635018988606835,
+      "loss": 1.0205,
+      "step": 281000
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 0.0002162002798321007,
+      "loss": 1.0345,
+      "step": 281500
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 0.0002160503697781331,
+      "loss": 1.0241,
+      "step": 282000
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 0.0002159004597241655,
+      "loss": 1.0025,
+      "step": 282500
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 0.00021575054967019785,
+      "loss": 1.0134,
+      "step": 283000
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 0.00021560063961623024,
+      "loss": 1.0285,
+      "step": 283500
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 0.00021545072956226263,
+      "loss": 1.0104,
+      "step": 284000
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 0.00021530081950829502,
+      "loss": 1.0321,
+      "step": 284500
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 0.00021515090945432738,
+      "loss": 1.0247,
+      "step": 285000
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 0.00021500099940035975,
+      "loss": 1.0604,
+      "step": 285500
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 0.00021485108934639216,
+      "loss": 1.0477,
+      "step": 286000
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 0.00021470117929242452,
+      "loss": 1.0212,
+      "step": 286500
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 0.00021455126923845691,
+      "loss": 0.9966,
+      "step": 287000
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 0.00021440135918448928,
+      "loss": 1.0401,
+      "step": 287500
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 0.0002142514491305217,
+      "loss": 1.0141,
+      "step": 288000
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 0.00021410153907655405,
+      "loss": 1.0218,
+      "step": 288500
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 0.00021395162902258642,
+      "loss": 1.0181,
+      "step": 289000
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 0.0002138017189686188,
+      "loss": 1.0322,
+      "step": 289500
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 0.0002136518089146512,
+      "loss": 1.0203,
+      "step": 290000
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 0.00021350189886068359,
+      "loss": 1.0237,
+      "step": 290500
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 0.00021335198880671595,
+      "loss": 1.0193,
+      "step": 291000
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 0.0002132020787527483,
+      "loss": 1.0192,
+      "step": 291500
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 0.00021305216869878073,
+      "loss": 1.0402,
+      "step": 292000
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 0.0002129022586448131,
+      "loss": 1.0278,
+      "step": 292500
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 0.00021275234859084548,
+      "loss": 1.0245,
+      "step": 293000
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 0.00021260243853687784,
+      "loss": 1.0114,
+      "step": 293500
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 0.00021245252848291026,
+      "loss": 1.0531,
+      "step": 294000
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 0.00021230261842894262,
+      "loss": 1.0417,
+      "step": 294500
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 0.00021215270837497498,
+      "loss": 1.044,
+      "step": 295000
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 0.00021200279832100737,
+      "loss": 1.042,
+      "step": 295500
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 0.00021185288826703976,
+      "loss": 1.023,
+      "step": 296000
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 0.00021170297821307215,
+      "loss": 1.0228,
+      "step": 296500
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 0.0002115530681591045,
+      "loss": 1.0452,
+      "step": 297000
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 0.00021140315810513687,
+      "loss": 1.0476,
+      "step": 297500
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 0.0002112532480511693,
+      "loss": 1.0348,
+      "step": 298000
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 0.00021110333799720165,
+      "loss": 1.0083,
+      "step": 298500
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 0.00021095342794323404,
+      "loss": 1.0312,
+      "step": 299000
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 0.00021080351788926643,
+      "loss": 1.0221,
+      "step": 299500
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 0.00021065360783529882,
+      "loss": 1.0182,
+      "step": 300000
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 0.00021050369778133118,
+      "loss": 1.0241,
+      "step": 300500
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 0.00021035378772736355,
+      "loss": 0.9948,
+      "step": 301000
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 0.00021020387767339596,
+      "loss": 0.9737,
+      "step": 301500
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 0.00021005396761942832,
+      "loss": 0.974,
+      "step": 302000
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 0.00020990405756546071,
+      "loss": 0.9561,
+      "step": 302500
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 0.00020975414751149308,
+      "loss": 0.9664,
+      "step": 303000
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 0.0002096042374575255,
+      "loss": 0.9937,
+      "step": 303500
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 0.00020945432740355786,
+      "loss": 0.9755,
+      "step": 304000
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 0.00020930441734959022,
+      "loss": 0.9804,
+      "step": 304500
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 0.0002091545072956226,
+      "loss": 0.9867,
+      "step": 305000
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 0.000209004597241655,
+      "loss": 0.9938,
+      "step": 305500
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 0.00020885468718768739,
+      "loss": 0.9943,
+      "step": 306000
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 0.00020870477713371975,
+      "loss": 0.9824,
+      "step": 306500
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 0.0002085548670797521,
+      "loss": 0.9998,
+      "step": 307000
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 0.00020840495702578453,
+      "loss": 0.9841,
+      "step": 307500
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 0.0002082550469718169,
+      "loss": 0.9928,
+      "step": 308000
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 0.00020810513691784928,
+      "loss": 0.9668,
+      "step": 308500
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 0.00020795522686388164,
+      "loss": 0.9988,
+      "step": 309000
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 0.00020780531680991406,
+      "loss": 0.9962,
+      "step": 309500
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 0.00020765540675594642,
+      "loss": 0.9967,
+      "step": 310000
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 0.00020750549670197878,
+      "loss": 0.9969,
+      "step": 310500
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 0.00020735558664801117,
+      "loss": 0.9816,
+      "step": 311000
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 0.00020720567659404356,
+      "loss": 0.9854,
+      "step": 311500
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 0.00020705576654007595,
+      "loss": 0.9915,
+      "step": 312000
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 0.0002069058564861083,
+      "loss": 1.0018,
+      "step": 312500
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 0.00020675594643214067,
+      "loss": 0.9815,
+      "step": 313000
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 0.0002066060363781731,
+      "loss": 0.9979,
+      "step": 313500
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 0.00020645612632420545,
+      "loss": 1.0003,
+      "step": 314000
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 0.00020630621627023784,
+      "loss": 1.012,
+      "step": 314500
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 0.0002061563062162702,
+      "loss": 0.9768,
+      "step": 315000
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 0.00020600639616230262,
+      "loss": 1.025,
+      "step": 315500
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 0.00020585648610833498,
+      "loss": 0.9873,
+      "step": 316000
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 0.00020570657605436735,
+      "loss": 1.0034,
+      "step": 316500
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 0.00020555666600039974,
+      "loss": 1.0018,
+      "step": 317000
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 0.00020540675594643212,
+      "loss": 0.9662,
+      "step": 317500
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 0.00020525684589246451,
+      "loss": 0.9955,
+      "step": 318000
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 0.00020510693583849688,
+      "loss": 0.9955,
+      "step": 318500
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 0.00020495702578452924,
+      "loss": 0.9984,
+      "step": 319000
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 0.00020480711573056166,
+      "loss": 0.9734,
+      "step": 319500
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 0.00020465720567659402,
+      "loss": 0.9852,
+      "step": 320000
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 0.0002045072956226264,
+      "loss": 1.0189,
+      "step": 320500
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 0.00020435738556865877,
+      "loss": 1.0255,
+      "step": 321000
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 0.00020420747551469119,
+      "loss": 0.9947,
+      "step": 321500
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 0.00020405756546072355,
+      "loss": 0.9939,
+      "step": 322000
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 0.0002039076554067559,
+      "loss": 0.9875,
+      "step": 322500
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 0.0002037577453527883,
+      "loss": 1.0007,
+      "step": 323000
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 0.0002036078352988207,
+      "loss": 1.0034,
+      "step": 323500
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 0.00020345792524485308,
+      "loss": 1.0097,
+      "step": 324000
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 0.00020330801519088544,
+      "loss": 0.9848,
+      "step": 324500
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 0.00020315810513691786,
+      "loss": 1.0256,
+      "step": 325000
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 0.00020300819508295022,
+      "loss": 0.9808,
+      "step": 325500
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 0.00020285828502898258,
+      "loss": 1.0022,
+      "step": 326000
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 0.00020270837497501497,
+      "loss": 1.0201,
+      "step": 326500
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 0.00020255846492104736,
+      "loss": 0.959,
+      "step": 327000
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 0.00020240855486707975,
+      "loss": 0.9806,
+      "step": 327500
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 0.0002022586448131121,
+      "loss": 1.0115,
+      "step": 328000
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 0.00020210873475914447,
+      "loss": 0.9884,
+      "step": 328500
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 0.0002019588247051769,
+      "loss": 1.0163,
+      "step": 329000
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 0.00020180891465120925,
+      "loss": 1.011,
+      "step": 329500
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 0.00020165900459724164,
+      "loss": 1.0079,
+      "step": 330000
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 0.000201509094543274,
+      "loss": 1.0157,
+      "step": 330500
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 0.00020135918448930642,
+      "loss": 1.0011,
+      "step": 331000
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 0.00020120927443533878,
+      "loss": 0.9829,
+      "step": 331500
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 0.00020105936438137115,
+      "loss": 1.0082,
+      "step": 332000
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 0.00020090945432740354,
+      "loss": 0.9941,
+      "step": 332500
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 0.00020075954427343593,
+      "loss": 0.9964,
+      "step": 333000
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 0.00020060963421946831,
+      "loss": 1.0093,
+      "step": 333500
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 0.00020045972416550068,
+      "loss": 1.0038,
+      "step": 334000
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 0.00020030981411153304,
+      "loss": 0.959,
+      "step": 334500
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 0.00020015990405756546,
+      "loss": 0.9419,
+      "step": 335000
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 0.00020000999400359782,
+      "loss": 0.9299,
+      "step": 335500
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 0.0001998600839496302,
+      "loss": 0.9677,
+      "step": 336000
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 0.00019971017389566257,
+      "loss": 0.9707,
+      "step": 336500
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 0.00019956026384169499,
+      "loss": 0.9857,
+      "step": 337000
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 0.00019941035378772735,
+      "loss": 0.9555,
+      "step": 337500
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 0.0001992604437337597,
+      "loss": 0.9549,
+      "step": 338000
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 0.0001991105336797921,
+      "loss": 0.9652,
+      "step": 338500
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 0.0001989606236258245,
+      "loss": 0.9826,
+      "step": 339000
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 0.00019881071357185688,
+      "loss": 0.969,
+      "step": 339500
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 0.00019866080351788924,
+      "loss": 0.9575,
+      "step": 340000
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 0.0001985108934639216,
+      "loss": 0.9538,
+      "step": 340500
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 0.00019836098340995402,
+      "loss": 0.93,
+      "step": 341000
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 0.00019821107335598638,
+      "loss": 0.959,
+      "step": 341500
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 0.00019806116330201877,
+      "loss": 0.9662,
+      "step": 342000
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 0.00019791125324805113,
+      "loss": 0.9498,
+      "step": 342500
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 0.00019776134319408355,
+      "loss": 0.9753,
+      "step": 343000
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 0.0001976114331401159,
+      "loss": 0.9582,
+      "step": 343500
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 0.00019746152308614827,
+      "loss": 0.9584,
+      "step": 344000
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 0.00019731161303218066,
+      "loss": 0.9476,
+      "step": 344500
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 0.00019716170297821305,
+      "loss": 0.9644,
+      "step": 345000
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 0.00019701179292424544,
+      "loss": 0.9705,
+      "step": 345500
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 0.0001968618828702778,
+      "loss": 0.9617,
+      "step": 346000
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 0.0001967119728163102,
+      "loss": 0.9613,
+      "step": 346500
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 0.00019656206276234258,
+      "loss": 0.9387,
+      "step": 347000
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 0.00019641215270837495,
+      "loss": 0.9691,
+      "step": 347500
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 0.00019626224265440734,
+      "loss": 0.9984,
+      "step": 348000
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 0.0001961123326004397,
+      "loss": 0.9879,
+      "step": 348500
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 0.00019596242254647211,
+      "loss": 0.9735,
+      "step": 349000
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 0.00019581251249250448,
+      "loss": 0.9573,
+      "step": 349500
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 0.00019566260243853687,
+      "loss": 0.9824,
+      "step": 350000
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 0.00019551269238456923,
+      "loss": 0.9736,
+      "step": 350500
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 0.00019536278233060162,
+      "loss": 0.9701,
+      "step": 351000
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 0.000195212872276634,
+      "loss": 0.9647,
+      "step": 351500
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 0.00019506296222266637,
+      "loss": 0.9814,
+      "step": 352000
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 0.00019491305216869879,
+      "loss": 0.9543,
+      "step": 352500
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 0.00019476314211473115,
+      "loss": 0.9871,
+      "step": 353000
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 0.00019461323206076354,
+      "loss": 0.9823,
+      "step": 353500
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 0.0001944633220067959,
+      "loss": 0.9663,
+      "step": 354000
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 0.0001943134119528283,
+      "loss": 1.0079,
+      "step": 354500
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 0.00019416350189886068,
+      "loss": 0.9627,
+      "step": 355000
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 0.00019401359184489304,
+      "loss": 0.9991,
+      "step": 355500
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 0.00019386368179092543,
+      "loss": 0.9705,
+      "step": 356000
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 0.00019371377173695782,
+      "loss": 0.9432,
+      "step": 356500
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 0.00019356386168299018,
+      "loss": 0.9685,
+      "step": 357000
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 0.00019341395162902257,
+      "loss": 0.9643,
+      "step": 357500
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 0.00019326404157505493,
+      "loss": 0.9625,
+      "step": 358000
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 0.00019311413152108735,
+      "loss": 0.9732,
+      "step": 358500
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 0.0001929642214671197,
+      "loss": 0.9731,
+      "step": 359000
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 0.0001928143114131521,
+      "loss": 0.9769,
+      "step": 359500
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 0.00019266440135918446,
+      "loss": 0.9679,
+      "step": 360000
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 0.00019251449130521685,
+      "loss": 0.9545,
+      "step": 360500
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 0.00019236458125124924,
+      "loss": 0.9519,
+      "step": 361000
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 0.0001922146711972816,
+      "loss": 0.9708,
+      "step": 361500
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 0.000192064761143314,
+      "loss": 0.9907,
+      "step": 362000
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 0.00019191485108934638,
+      "loss": 0.9822,
+      "step": 362500
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 0.00019176494103537877,
+      "loss": 0.9835,
+      "step": 363000
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 0.00019161503098141114,
+      "loss": 0.9853,
+      "step": 363500
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 0.0001914651209274435,
+      "loss": 0.9657,
+      "step": 364000
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 0.00019131521087347591,
+      "loss": 0.9775,
+      "step": 364500
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 0.00019116530081950828,
+      "loss": 0.9758,
+      "step": 365000
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 0.00019101539076554067,
+      "loss": 0.9763,
+      "step": 365500
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 0.00019086548071157303,
+      "loss": 0.977,
+      "step": 366000
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 0.00019071557065760545,
+      "loss": 0.9699,
+      "step": 366500
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 0.0001905656606036378,
+      "loss": 0.9689,
+      "step": 367000
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 0.00019041575054967017,
+      "loss": 0.9726,
+      "step": 367500
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 0.00019026584049570256,
+      "loss": 0.9607,
+      "step": 368000
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 0.00019011593044173495,
+      "loss": 0.9214,
+      "step": 368500
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 0.00018996602038776734,
+      "loss": 0.9327,
+      "step": 369000
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 0.0001898161103337997,
+      "loss": 0.9466,
+      "step": 369500
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 0.00018966620027983206,
+      "loss": 0.9361,
+      "step": 370000
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 0.00018951629022586448,
+      "loss": 0.9178,
+      "step": 370500
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 0.00018936638017189684,
+      "loss": 0.9375,
+      "step": 371000
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 0.00018921647011792923,
+      "loss": 0.9297,
+      "step": 371500
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 0.0001890665600639616,
+      "loss": 0.9306,
+      "step": 372000
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 0.000188916650009994,
+      "loss": 0.9331,
+      "step": 372500
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 0.00018876673995602637,
+      "loss": 0.921,
+      "step": 373000
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 0.00018861682990205873,
+      "loss": 0.9353,
+      "step": 373500
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 0.00018846691984809112,
+      "loss": 0.9464,
+      "step": 374000
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 0.0001883170097941235,
+      "loss": 0.931,
+      "step": 374500
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 0.0001881670997401559,
+      "loss": 0.9276,
+      "step": 375000
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 0.00018801718968618826,
+      "loss": 0.9448,
+      "step": 375500
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 0.00018786727963222063,
+      "loss": 0.9229,
+      "step": 376000
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 0.00018771736957825304,
+      "loss": 0.9417,
+      "step": 376500
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 0.0001875674595242854,
+      "loss": 0.9312,
+      "step": 377000
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 0.0001874175494703178,
+      "loss": 0.945,
+      "step": 377500
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 0.00018726763941635016,
+      "loss": 0.9455,
+      "step": 378000
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 0.00018711772936238257,
+      "loss": 0.9265,
+      "step": 378500
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 0.00018696781930841494,
+      "loss": 0.9421,
+      "step": 379000
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 0.0001868179092544473,
+      "loss": 0.9346,
+      "step": 379500
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 0.00018666799920047971,
+      "loss": 0.9236,
+      "step": 380000
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 0.00018651808914651208,
+      "loss": 0.9323,
+      "step": 380500
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 0.00018636817909254447,
+      "loss": 0.9259,
+      "step": 381000
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 0.00018621826903857683,
+      "loss": 0.9277,
+      "step": 381500
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 0.00018606835898460925,
+      "loss": 0.9224,
+      "step": 382000
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 0.0001859184489306416,
+      "loss": 0.9492,
+      "step": 382500
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 0.00018576853887667397,
+      "loss": 0.9672,
+      "step": 383000
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 0.00018561862882270636,
+      "loss": 0.9576,
+      "step": 383500
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 0.00018546871876873875,
+      "loss": 0.9401,
+      "step": 384000
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 0.00018531880871477114,
+      "loss": 0.9426,
+      "step": 384500
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 0.0001851688986608035,
+      "loss": 0.9591,
+      "step": 385000
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 0.00018501898860683586,
+      "loss": 0.9455,
+      "step": 385500
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 0.00018486907855286828,
+      "loss": 0.9456,
+      "step": 386000
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 0.00018471916849890064,
+      "loss": 0.9433,
+      "step": 386500
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 0.00018456925844493303,
+      "loss": 0.9589,
+      "step": 387000
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 0.0001844193483909654,
+      "loss": 0.9247,
+      "step": 387500
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 0.0001842694383369978,
+      "loss": 0.9239,
+      "step": 388000
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 0.00018411952828303017,
+      "loss": 0.9506,
+      "step": 388500
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 0.00018396961822906253,
+      "loss": 0.9336,
+      "step": 389000
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 0.00018381970817509492,
+      "loss": 0.9592,
+      "step": 389500
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 0.0001836697981211273,
+      "loss": 0.9645,
+      "step": 390000
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 0.0001835198880671597,
+      "loss": 0.9602,
+      "step": 390500
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 0.00018336997801319206,
+      "loss": 0.9246,
+      "step": 391000
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 0.00018322006795922443,
+      "loss": 0.9641,
+      "step": 391500
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 0.00018307015790525684,
+      "loss": 0.9349,
+      "step": 392000
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 0.0001829202478512892,
+      "loss": 0.9634,
+      "step": 392500
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 0.0001827703377973216,
+      "loss": 0.9641,
+      "step": 393000
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 0.00018262042774335396,
+      "loss": 0.9404,
+      "step": 393500
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 0.00018247051768938637,
+      "loss": 0.9647,
+      "step": 394000
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 0.00018232060763541874,
+      "loss": 0.9416,
+      "step": 394500
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 0.0001821706975814511,
+      "loss": 0.9555,
+      "step": 395000
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 0.0001820207875274835,
+      "loss": 0.9487,
+      "step": 395500
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 0.00018187087747351588,
+      "loss": 0.9609,
+      "step": 396000
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 0.00018172096741954827,
+      "loss": 0.9481,
+      "step": 396500
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 0.00018157105736558063,
+      "loss": 0.9449,
+      "step": 397000
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 0.000181421147311613,
+      "loss": 0.9381,
+      "step": 397500
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 0.0001812712372576454,
+      "loss": 0.9726,
+      "step": 398000
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 0.00018112132720367777,
+      "loss": 0.9615,
+      "step": 398500
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 0.00018097141714971016,
+      "loss": 0.9523,
+      "step": 399000
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 0.00018082150709574252,
+      "loss": 0.9286,
+      "step": 399500
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 0.00018067159704177494,
+      "loss": 0.961,
+      "step": 400000
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 0.0001805216869878073,
+      "loss": 0.947,
+      "step": 400500
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 0.00018037177693383966,
+      "loss": 0.9594,
+      "step": 401000
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 0.00018022186687987205,
+      "loss": 0.9237,
+      "step": 401500
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 0.00018007195682590444,
+      "loss": 0.9109,
+      "step": 402000
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 0.00017992204677193683,
+      "loss": 0.9255,
+      "step": 402500
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 0.0001797721367179692,
+      "loss": 0.9117,
+      "step": 403000
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 0.00017962222666400156,
+      "loss": 0.9218,
+      "step": 403500
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 0.00017947231661003397,
+      "loss": 0.9003,
+      "step": 404000
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 0.00017932240655606633,
+      "loss": 0.9156,
+      "step": 404500
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 0.00017917249650209872,
+      "loss": 0.9,
+      "step": 405000
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 0.0001790225864481311,
+      "loss": 0.9066,
+      "step": 405500
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 0.0001788726763941635,
+      "loss": 0.895,
+      "step": 406000
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 0.00017872276634019586,
+      "loss": 0.9223,
+      "step": 406500
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 0.00017857285628622823,
+      "loss": 0.9268,
+      "step": 407000
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 0.00017842294623226064,
+      "loss": 0.9282,
+      "step": 407500
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 0.000178273036178293,
+      "loss": 0.9169,
+      "step": 408000
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 0.0001781231261243254,
+      "loss": 0.9232,
+      "step": 408500
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 0.00017797321607035776,
+      "loss": 0.9058,
+      "step": 409000
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 0.00017782330601639017,
+      "loss": 0.905,
+      "step": 409500
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 0.00017767339596242254,
+      "loss": 0.9204,
+      "step": 410000
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 0.0001775234859084549,
+      "loss": 0.9418,
+      "step": 410500
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 0.0001773735758544873,
+      "loss": 0.8982,
+      "step": 411000
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 0.00017722366580051968,
+      "loss": 0.9224,
+      "step": 411500
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 0.00017707375574655207,
+      "loss": 0.9563,
+      "step": 412000
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 0.00017692384569258443,
+      "loss": 0.8992,
+      "step": 412500
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 0.0001767739356386168,
+      "loss": 0.9165,
+      "step": 413000
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 0.0001766240255846492,
+      "loss": 0.9241,
+      "step": 413500
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 0.00017647411553068157,
+      "loss": 0.9174,
+      "step": 414000
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 0.00017632420547671396,
+      "loss": 0.9213,
+      "step": 414500
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 0.00017617429542274632,
+      "loss": 0.9066,
+      "step": 415000
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 0.00017602438536877874,
+      "loss": 0.9126,
+      "step": 415500
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 0.0001758744753148111,
+      "loss": 0.891,
+      "step": 416000
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 0.00017572456526084346,
+      "loss": 0.9023,
+      "step": 416500
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 0.00017557465520687585,
+      "loss": 0.9307,
+      "step": 417000
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 0.00017542474515290824,
+      "loss": 0.9265,
+      "step": 417500
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 0.00017527483509894063,
+      "loss": 0.8943,
+      "step": 418000
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 0.000175124925044973,
+      "loss": 0.9179,
+      "step": 418500
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 0.00017497501499100536,
+      "loss": 0.9252,
+      "step": 419000
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 0.00017482510493703777,
+      "loss": 0.9232,
+      "step": 419500
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 0.00017467519488307013,
+      "loss": 0.9389,
+      "step": 420000
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 0.00017452528482910252,
+      "loss": 0.898,
+      "step": 420500
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 0.00017437537477513489,
+      "loss": 0.9148,
+      "step": 421000
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 0.0001742254647211673,
+      "loss": 0.9147,
+      "step": 421500
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 0.00017407555466719967,
+      "loss": 0.8992,
+      "step": 422000
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 0.00017392564461323203,
+      "loss": 0.9361,
+      "step": 422500
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 0.00017377573455926442,
+      "loss": 0.9256,
+      "step": 423000
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 0.0001736258245052968,
+      "loss": 0.9341,
+      "step": 423500
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 0.0001734759144513292,
+      "loss": 0.8993,
+      "step": 424000
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 0.00017332600439736156,
+      "loss": 0.9187,
+      "step": 424500
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 0.00017317609434339395,
+      "loss": 0.9244,
+      "step": 425000
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 0.00017302618428942634,
+      "loss": 0.9375,
+      "step": 425500
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 0.0001728762742354587,
+      "loss": 0.9166,
+      "step": 426000
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 0.0001727263641814911,
+      "loss": 0.9181,
+      "step": 426500
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 0.00017257645412752345,
+      "loss": 0.9124,
+      "step": 427000
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 0.00017242654407355587,
+      "loss": 0.9271,
+      "step": 427500
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 0.00017227663401958823,
+      "loss": 0.9301,
+      "step": 428000
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 0.00017212672396562062,
+      "loss": 0.9357,
+      "step": 428500
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 0.00017197681391165298,
+      "loss": 0.9518,
+      "step": 429000
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 0.00017182690385768537,
+      "loss": 0.9481,
+      "step": 429500
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 0.00017167699380371776,
+      "loss": 0.9176,
+      "step": 430000
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 0.00017152708374975012,
+      "loss": 0.9442,
+      "step": 430500
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 0.0001713771736957825,
+      "loss": 0.901,
+      "step": 431000
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 0.0001712272636418149,
+      "loss": 0.9295,
+      "step": 431500
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 0.00017107735358784726,
+      "loss": 0.9319,
+      "step": 432000
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 0.00017092744353387965,
+      "loss": 0.9313,
+      "step": 432500
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 0.00017077753347991204,
+      "loss": 0.9186,
+      "step": 433000
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 0.00017062762342594443,
+      "loss": 0.9097,
+      "step": 433500
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 0.0001704777133719768,
+      "loss": 0.9129,
+      "step": 434000
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 0.00017032780331800918,
+      "loss": 0.9398,
+      "step": 434500
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 0.00017017789326404157,
+      "loss": 0.8845,
+      "step": 435000
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 0.00017002798321007393,
+      "loss": 0.8695,
+      "step": 435500
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 0.00016987807315610632,
+      "loss": 0.8749,
+      "step": 436000
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 0.00016972816310213869,
+      "loss": 0.8873,
+      "step": 436500
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 0.0001695782530481711,
+      "loss": 0.8687,
+      "step": 437000
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 0.00016942834299420347,
+      "loss": 0.8894,
+      "step": 437500
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 0.00016927843294023585,
+      "loss": 0.8918,
+      "step": 438000
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 0.00016912852288626822,
+      "loss": 0.8928,
+      "step": 438500
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 0.0001689786128323006,
+      "loss": 0.8914,
+      "step": 439000
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 0.000168828702778333,
+      "loss": 0.8893,
+      "step": 439500
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 0.00016867879272436536,
+      "loss": 0.8915,
+      "step": 440000
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 0.00016852888267039775,
+      "loss": 0.9007,
+      "step": 440500
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 0.00016837897261643014,
+      "loss": 0.9053,
+      "step": 441000
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 0.00016822906256246253,
+      "loss": 0.8975,
+      "step": 441500
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 0.0001680791525084949,
+      "loss": 0.9001,
+      "step": 442000
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 0.00016792924245452725,
+      "loss": 0.897,
+      "step": 442500
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 0.00016777933240055967,
+      "loss": 0.8888,
+      "step": 443000
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 0.00016762942234659203,
+      "loss": 0.8957,
+      "step": 443500
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 0.00016747951229262442,
+      "loss": 0.9098,
+      "step": 444000
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 0.00016732960223865678,
+      "loss": 0.9032,
+      "step": 444500
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 0.0001671796921846892,
+      "loss": 0.9009,
+      "step": 445000
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 0.00016702978213072156,
+      "loss": 0.8848,
+      "step": 445500
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 0.00016687987207675392,
+      "loss": 0.8927,
+      "step": 446000
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 0.0001667299620227863,
+      "loss": 0.9166,
+      "step": 446500
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 0.0001665800519688187,
+      "loss": 0.8934,
+      "step": 447000
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 0.0001664301419148511,
+      "loss": 0.8561,
+      "step": 447500
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 0.00016628023186088345,
+      "loss": 0.8928,
+      "step": 448000
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 0.00016613032180691582,
+      "loss": 0.8949,
+      "step": 448500
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 0.00016598041175294823,
+      "loss": 0.9076,
+      "step": 449000
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 0.0001658305016989806,
+      "loss": 0.8815,
+      "step": 449500
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 0.00016568059164501298,
+      "loss": 0.9033,
+      "step": 450000
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 0.00016553068159104535,
+      "loss": 0.9035,
+      "step": 450500
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 0.00016538077153707776,
+      "loss": 0.9178,
+      "step": 451000
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 0.00016523086148311012,
+      "loss": 0.8993,
+      "step": 451500
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 0.0001650809514291425,
+      "loss": 0.9015,
+      "step": 452000
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 0.00016493104137517488,
+      "loss": 0.9005,
+      "step": 452500
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 0.00016478113132120727,
+      "loss": 0.8866,
+      "step": 453000
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 0.00016463122126723965,
+      "loss": 0.8973,
+      "step": 453500
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 0.00016448131121327202,
+      "loss": 0.8787,
+      "step": 454000
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 0.00016433140115930438,
+      "loss": 0.8802,
+      "step": 454500
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 0.0001641814911053368,
+      "loss": 0.9065,
+      "step": 455000
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 0.00016403158105136916,
+      "loss": 0.9177,
+      "step": 455500
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 0.00016388167099740155,
+      "loss": 0.9184,
+      "step": 456000
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 0.0001637317609434339,
+      "loss": 0.9112,
+      "step": 456500
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 0.00016358185088946633,
+      "loss": 0.9059,
+      "step": 457000
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 0.0001634319408354987,
+      "loss": 0.9168,
+      "step": 457500
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 0.00016328203078153105,
+      "loss": 0.89,
+      "step": 458000
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 0.00016313212072756344,
+      "loss": 0.9066,
+      "step": 458500
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 0.00016298221067359583,
+      "loss": 0.9041,
+      "step": 459000
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 0.00016283230061962822,
+      "loss": 0.8919,
+      "step": 459500
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 0.00016268239056566058,
+      "loss": 0.8856,
+      "step": 460000
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 0.000162532480511693,
+      "loss": 0.8989,
+      "step": 460500
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 0.00016238257045772536,
+      "loss": 0.8832,
+      "step": 461000
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 0.00016223266040375772,
+      "loss": 0.8902,
+      "step": 461500
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 0.0001620827503497901,
+      "loss": 0.9124,
+      "step": 462000
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 0.0001619328402958225,
+      "loss": 0.9204,
+      "step": 462500
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 0.0001617829302418549,
+      "loss": 0.8828,
+      "step": 463000
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 0.00016163302018788725,
+      "loss": 0.909,
+      "step": 463500
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 0.00016148311013391962,
+      "loss": 0.8917,
+      "step": 464000
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 0.00016133320007995203,
+      "loss": 0.9148,
+      "step": 464500
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 0.0001611832900259844,
+      "loss": 0.8905,
+      "step": 465000
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 0.00016103337997201678,
+      "loss": 0.9057,
+      "step": 465500
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 0.00016088346991804915,
+      "loss": 0.9198,
+      "step": 466000
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 0.00016073355986408156,
+      "loss": 0.9132,
+      "step": 466500
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 0.00016058364981011392,
+      "loss": 0.9184,
+      "step": 467000
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 0.0001604337397561463,
+      "loss": 0.8966,
+      "step": 467500
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 0.00016028382970217868,
+      "loss": 0.8882,
+      "step": 468000
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 0.00016013391964821107,
+      "loss": 0.8713,
+      "step": 468500
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 0.00015998400959424345,
+      "loss": 0.8713,
+      "step": 469000
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 0.00015983409954027582,
+      "loss": 0.8868,
+      "step": 469500
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 0.00015968418948630818,
+      "loss": 0.8559,
+      "step": 470000
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 0.0001595342794323406,
+      "loss": 0.8652,
+      "step": 470500
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 0.00015938436937837296,
+      "loss": 0.8722,
+      "step": 471000
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 0.00015923445932440535,
+      "loss": 0.8768,
+      "step": 471500
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 0.0001590845492704377,
+      "loss": 0.8812,
+      "step": 472000
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 0.00015893463921647013,
+      "loss": 0.8712,
+      "step": 472500
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 0.0001587847291625025,
+      "loss": 0.8546,
+      "step": 473000
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 0.00015863481910853485,
+      "loss": 0.8618,
+      "step": 473500
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 0.00015848490905456724,
+      "loss": 0.8755,
+      "step": 474000
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 0.00015833499900059963,
+      "loss": 0.8632,
+      "step": 474500
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 0.00015818508894663202,
+      "loss": 0.8687,
+      "step": 475000
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 0.00015803517889266438,
+      "loss": 0.8771,
+      "step": 475500
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 0.00015788526883869674,
+      "loss": 0.8504,
+      "step": 476000
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 0.00015773535878472916,
+      "loss": 0.8639,
+      "step": 476500
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 0.00015758544873076152,
+      "loss": 0.8786,
+      "step": 477000
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 0.0001574355386767939,
+      "loss": 0.8752,
+      "step": 477500
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 0.00015728562862282627,
+      "loss": 0.8673,
+      "step": 478000
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 0.0001571357185688587,
+      "loss": 0.867,
+      "step": 478500
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 0.00015698580851489105,
+      "loss": 0.8565,
+      "step": 479000
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 0.00015683589846092342,
+      "loss": 0.8564,
+      "step": 479500
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 0.0001566859884069558,
+      "loss": 0.8573,
+      "step": 480000
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 0.0001565360783529882,
+      "loss": 0.8612,
+      "step": 480500
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 0.00015638616829902058,
+      "loss": 0.8686,
+      "step": 481000
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 0.00015623625824505295,
+      "loss": 0.8795,
+      "step": 481500
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 0.0001560863481910853,
+      "loss": 0.8617,
+      "step": 482000
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 0.00015593643813711772,
+      "loss": 0.8723,
+      "step": 482500
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 0.0001557865280831501,
+      "loss": 0.8706,
+      "step": 483000
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 0.00015563661802918248,
+      "loss": 0.8566,
+      "step": 483500
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 0.00015548670797521484,
+      "loss": 0.8658,
+      "step": 484000
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 0.00015533679792124726,
+      "loss": 0.8949,
+      "step": 484500
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 0.00015518688786727962,
+      "loss": 0.8854,
+      "step": 485000
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 0.00015503697781331198,
+      "loss": 0.8628,
+      "step": 485500
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 0.00015488706775934437,
+      "loss": 0.869,
+      "step": 486000
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 0.00015473715770537676,
+      "loss": 0.8877,
+      "step": 486500
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 0.00015458724765140915,
+      "loss": 0.886,
+      "step": 487000
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 0.0001544373375974415,
+      "loss": 0.8834,
+      "step": 487500
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 0.00015428742754347393,
+      "loss": 0.8778,
+      "step": 488000
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 0.0001541375174895063,
+      "loss": 0.8602,
+      "step": 488500
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 0.00015398760743553865,
+      "loss": 0.8778,
+      "step": 489000
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 0.00015383769738157104,
+      "loss": 0.8854,
+      "step": 489500
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 0.00015368778732760343,
+      "loss": 0.8619,
+      "step": 490000
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 0.00015353787727363582,
+      "loss": 0.8669,
+      "step": 490500
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 0.00015338796721966818,
+      "loss": 0.8851,
+      "step": 491000
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 0.00015323805716570054,
+      "loss": 0.8724,
+      "step": 491500
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 0.00015308814711173296,
+      "loss": 0.8658,
+      "step": 492000
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 0.00015293823705776532,
+      "loss": 0.8839,
+      "step": 492500
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 0.0001527883270037977,
+      "loss": 0.8859,
+      "step": 493000
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 0.00015263841694983007,
+      "loss": 0.8813,
+      "step": 493500
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 0.0001524885068958625,
+      "loss": 0.8656,
+      "step": 494000
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 0.00015233859684189485,
+      "loss": 0.8635,
+      "step": 494500
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 0.00015218868678792722,
+      "loss": 0.8685,
+      "step": 495000
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 0.0001520387767339596,
+      "loss": 0.8987,
+      "step": 495500
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 0.000151888866679992,
+      "loss": 0.9011,
+      "step": 496000
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 0.00015173895662602438,
+      "loss": 0.8785,
+      "step": 496500
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 0.00015158904657205675,
+      "loss": 0.8568,
+      "step": 497000
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 0.0001514391365180891,
+      "loss": 0.8688,
+      "step": 497500
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 0.00015128922646412152,
+      "loss": 0.8804,
+      "step": 498000
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 0.0001511393164101539,
+      "loss": 0.8795,
+      "step": 498500
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 0.00015098940635618628,
+      "loss": 0.8786,
+      "step": 499000
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 0.00015083949630221864,
+      "loss": 0.8608,
+      "step": 499500
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 0.00015068958624825106,
+      "loss": 0.9066,
+      "step": 500000
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 0.00015053967619428342,
+      "loss": 0.8672,
+      "step": 500500
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 0.00015038976614031578,
+      "loss": 0.8646,
+      "step": 501000
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 0.00015023985608634817,
+      "loss": 0.8523,
+      "step": 501500
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 0.00015008994603238056,
+      "loss": 0.8298,
+      "step": 502000
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 0.00014994003597841295,
+      "loss": 0.8402,
+      "step": 502500
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 0.0001497901259244453,
+      "loss": 0.8216,
+      "step": 503000
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 0.0001496402158704777,
+      "loss": 0.8306,
+      "step": 503500
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 0.00014949030581651006,
+      "loss": 0.8117,
+      "step": 504000
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 0.00014934039576254245,
+      "loss": 0.8351,
+      "step": 504500
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 0.00014919048570857484,
+      "loss": 0.8559,
+      "step": 505000
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 0.00014904057565460723,
+      "loss": 0.8573,
+      "step": 505500
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 0.0001488906656006396,
+      "loss": 0.8333,
+      "step": 506000
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 0.00014874075554667198,
+      "loss": 0.8369,
+      "step": 506500
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 0.00014859084549270437,
+      "loss": 0.8515,
+      "step": 507000
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 0.00014844093543873673,
+      "loss": 0.8362,
+      "step": 507500
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 0.00014829102538476912,
+      "loss": 0.854,
+      "step": 508000
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 0.0001481411153308015,
+      "loss": 0.8529,
+      "step": 508500
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 0.00014799120527683387,
+      "loss": 0.8332,
+      "step": 509000
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 0.00014784129522286626,
+      "loss": 0.8574,
+      "step": 509500
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 0.00014769138516889865,
+      "loss": 0.8549,
+      "step": 510000
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 0.00014754147511493102,
+      "loss": 0.8634,
+      "step": 510500
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 0.0001473915650609634,
+      "loss": 0.8262,
+      "step": 511000
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 0.0001472416550069958,
+      "loss": 0.8455,
+      "step": 511500
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 0.00014709174495302816,
+      "loss": 0.8336,
+      "step": 512000
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 0.00014694183489906055,
+      "loss": 0.8493,
+      "step": 512500
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 0.00014679192484509294,
+      "loss": 0.8368,
+      "step": 513000
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 0.00014664201479112532,
+      "loss": 0.8441,
+      "step": 513500
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 0.0001464921047371577,
+      "loss": 0.8549,
+      "step": 514000
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 0.00014634219468319008,
+      "loss": 0.8619,
+      "step": 514500
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 0.00014619228462922247,
+      "loss": 0.8299,
+      "step": 515000
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 0.00014604237457525483,
+      "loss": 0.8297,
+      "step": 515500
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 0.00014589246452128722,
+      "loss": 0.8415,
+      "step": 516000
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 0.0001457425544673196,
+      "loss": 0.8484,
+      "step": 516500
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 0.000145592644413352,
+      "loss": 0.8505,
+      "step": 517000
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 0.00014544273435938436,
+      "loss": 0.858,
+      "step": 517500
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 0.00014529282430541675,
+      "loss": 0.8429,
+      "step": 518000
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 0.0001451429142514491,
+      "loss": 0.8437,
+      "step": 518500
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 0.0001449930041974815,
+      "loss": 0.8411,
+      "step": 519000
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 0.0001448430941435139,
+      "loss": 0.8565,
+      "step": 519500
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 0.00014469318408954628,
+      "loss": 0.8379,
+      "step": 520000
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 0.00014454327403557864,
+      "loss": 0.8672,
+      "step": 520500
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 0.00014439336398161103,
+      "loss": 0.8565,
+      "step": 521000
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 0.0001442434539276434,
+      "loss": 0.8351,
+      "step": 521500
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 0.00014409354387367578,
+      "loss": 0.8663,
+      "step": 522000
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 0.00014394363381970817,
+      "loss": 0.8497,
+      "step": 522500
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 0.00014379372376574056,
+      "loss": 0.8472,
+      "step": 523000
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 0.00014364381371177292,
+      "loss": 0.8612,
+      "step": 523500
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 0.0001434939036578053,
+      "loss": 0.8223,
+      "step": 524000
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 0.00014334399360383767,
+      "loss": 0.8514,
+      "step": 524500
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 0.00014319408354987006,
+      "loss": 0.8906,
+      "step": 525000
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 0.00014304417349590245,
+      "loss": 0.8522,
+      "step": 525500
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 0.00014289426344193484,
+      "loss": 0.8439,
+      "step": 526000
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 0.0001427443533879672,
+      "loss": 0.8482,
+      "step": 526500
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 0.0001425944433339996,
+      "loss": 0.8656,
+      "step": 527000
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 0.00014244453328003196,
+      "loss": 0.8668,
+      "step": 527500
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 0.00014229462322606435,
+      "loss": 0.8606,
+      "step": 528000
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 0.00014214471317209674,
+      "loss": 0.8567,
+      "step": 528500
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 0.00014199480311812913,
+      "loss": 0.8632,
+      "step": 529000
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 0.0001418448930641615,
+      "loss": 0.8368,
+      "step": 529500
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 0.00014169498301019388,
+      "loss": 0.868,
+      "step": 530000
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 0.00014154507295622624,
+      "loss": 0.8833,
+      "step": 530500
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 0.00014139516290225863,
+      "loss": 0.8784,
+      "step": 531000
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 0.00014124525284829102,
+      "loss": 0.8749,
+      "step": 531500
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 0.0001410953427943234,
+      "loss": 0.8728,
+      "step": 532000
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 0.00014094543274035577,
+      "loss": 0.8622,
+      "step": 532500
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 0.00014079552268638816,
+      "loss": 0.8663,
+      "step": 533000
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 0.00014064561263242052,
+      "loss": 0.8635,
+      "step": 533500
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 0.0001404957025784529,
+      "loss": 0.8524,
+      "step": 534000
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 0.0001403457925244853,
+      "loss": 0.8512,
+      "step": 534500
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 0.0001401958824705177,
+      "loss": 0.8336,
+      "step": 535000
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 0.00014004597241655005,
+      "loss": 0.8164,
+      "step": 535500
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 0.00013989606236258244,
+      "loss": 0.8124,
+      "step": 536000
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 0.0001397461523086148,
+      "loss": 0.8113,
+      "step": 536500
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 0.0001395962422546472,
+      "loss": 0.8323,
+      "step": 537000
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 0.00013944633220067958,
+      "loss": 0.8541,
+      "step": 537500
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 0.00013929642214671197,
+      "loss": 0.8327,
+      "step": 538000
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 0.00013914651209274433,
+      "loss": 0.8329,
+      "step": 538500
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 0.00013899660203877672,
+      "loss": 0.8303,
+      "step": 539000
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 0.00013884669198480909,
+      "loss": 0.8462,
+      "step": 539500
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 0.00013869678193084147,
+      "loss": 0.8202,
+      "step": 540000
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 0.00013854687187687386,
+      "loss": 0.8453,
+      "step": 540500
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 0.00013839696182290625,
+      "loss": 0.8205,
+      "step": 541000
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 0.00013824705176893864,
+      "loss": 0.8206,
+      "step": 541500
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 0.000138097141714971,
+      "loss": 0.8158,
+      "step": 542000
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 0.0001379472316610034,
+      "loss": 0.847,
+      "step": 542500
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 0.00013779732160703576,
+      "loss": 0.8279,
+      "step": 543000
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 0.00013764741155306815,
+      "loss": 0.8234,
+      "step": 543500
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 0.00013749750149910054,
+      "loss": 0.8109,
+      "step": 544000
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 0.00013734759144513293,
+      "loss": 0.8215,
+      "step": 544500
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 0.0001371976813911653,
+      "loss": 0.8124,
+      "step": 545000
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 0.00013704777133719768,
+      "loss": 0.8266,
+      "step": 545500
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 0.00013689786128323004,
+      "loss": 0.8288,
+      "step": 546000
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 0.00013674795122926243,
+      "loss": 0.8218,
+      "step": 546500
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 0.00013659804117529482,
+      "loss": 0.8223,
+      "step": 547000
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 0.0001364481311213272,
+      "loss": 0.8328,
+      "step": 547500
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 0.00013629822106735957,
+      "loss": 0.8132,
+      "step": 548000
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 0.00013614831101339196,
+      "loss": 0.8351,
+      "step": 548500
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 0.00013599840095942432,
+      "loss": 0.8316,
+      "step": 549000
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 0.0001358484909054567,
+      "loss": 0.8163,
+      "step": 549500
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 0.0001356985808514891,
+      "loss": 0.8258,
+      "step": 550000
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 0.0001355486707975215,
+      "loss": 0.8321,
+      "step": 550500
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 0.00013539876074355385,
+      "loss": 0.8229,
+      "step": 551000
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 0.00013524885068958624,
+      "loss": 0.8357,
+      "step": 551500
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 0.0001350989406356186,
+      "loss": 0.8641,
+      "step": 552000
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 0.000134949030581651,
+      "loss": 0.8429,
+      "step": 552500
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 0.00013479912052768338,
+      "loss": 0.807,
+      "step": 553000
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 0.00013464921047371577,
+      "loss": 0.8377,
+      "step": 553500
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 0.00013449930041974813,
+      "loss": 0.8206,
+      "step": 554000
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 0.00013434939036578052,
+      "loss": 0.8149,
+      "step": 554500
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 0.00013419948031181289,
+      "loss": 0.8204,
+      "step": 555000
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 0.00013404957025784528,
+      "loss": 0.8339,
+      "step": 555500
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 0.00013389966020387766,
+      "loss": 0.8129,
+      "step": 556000
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 0.00013374975014991005,
+      "loss": 0.8141,
+      "step": 556500
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 0.00013359984009594242,
+      "loss": 0.8182,
+      "step": 557000
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 0.0001334499300419748,
+      "loss": 0.8326,
+      "step": 557500
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 0.00013330001998800717,
+      "loss": 0.815,
+      "step": 558000
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 0.00013315010993403956,
+      "loss": 0.8301,
+      "step": 558500
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 0.00013300019988007195,
+      "loss": 0.8261,
+      "step": 559000
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 0.00013285028982610434,
+      "loss": 0.8431,
+      "step": 559500
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 0.0001327003797721367,
+      "loss": 0.8281,
+      "step": 560000
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 0.0001325504697181691,
+      "loss": 0.8453,
+      "step": 560500
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 0.00013240055966420145,
+      "loss": 0.8377,
+      "step": 561000
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 0.00013225064961023384,
+      "loss": 0.8218,
+      "step": 561500
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 0.00013210073955626623,
+      "loss": 0.83,
+      "step": 562000
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 0.00013195082950229862,
+      "loss": 0.8409,
+      "step": 562500
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 0.00013180091944833098,
+      "loss": 0.8288,
+      "step": 563000
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 0.00013165100939436337,
+      "loss": 0.8411,
+      "step": 563500
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 0.00013150109934039573,
+      "loss": 0.8228,
+      "step": 564000
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 0.00013135118928642812,
+      "loss": 0.8482,
+      "step": 564500
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 0.0001312012792324605,
+      "loss": 0.8383,
+      "step": 565000
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 0.0001310513691784929,
+      "loss": 0.8329,
+      "step": 565500
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 0.00013090145912452526,
+      "loss": 0.8434,
+      "step": 566000
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 0.00013075154907055765,
+      "loss": 0.8338,
+      "step": 566500
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 0.00013060163901659004,
+      "loss": 0.8355,
+      "step": 567000
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 0.0001304517289626224,
+      "loss": 0.8423,
+      "step": 567500
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 0.0001303018189086548,
+      "loss": 0.8477,
+      "step": 568000
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 0.00013015190885468718,
+      "loss": 0.791,
+      "step": 568500
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 0.00013000199880071957,
+      "loss": 0.7844,
+      "step": 569000
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 0.00012985208874675193,
+      "loss": 0.837,
+      "step": 569500
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 0.00012970217869278432,
+      "loss": 0.7895,
+      "step": 570000
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 0.00012955226863881669,
+      "loss": 0.8245,
+      "step": 570500
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 0.00012940235858484908,
+      "loss": 0.8173,
+      "step": 571000
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 0.00012925244853088146,
+      "loss": 0.8003,
+      "step": 571500
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 0.00012910253847691385,
+      "loss": 0.8038,
+      "step": 572000
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 0.00012895262842294622,
+      "loss": 0.8083,
+      "step": 572500
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 0.0001288027183689786,
+      "loss": 0.8155,
+      "step": 573000
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 0.00012865280831501097,
+      "loss": 0.807,
+      "step": 573500
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 0.00012850289826104336,
+      "loss": 0.795,
+      "step": 574000
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 0.00012835298820707575,
+      "loss": 0.7878,
+      "step": 574500
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 0.00012820307815310814,
+      "loss": 0.8038,
+      "step": 575000
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 0.0001280531680991405,
+      "loss": 0.8051,
+      "step": 575500
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 0.0001279032580451729,
+      "loss": 0.8151,
+      "step": 576000
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 0.00012775334799120525,
+      "loss": 0.8015,
+      "step": 576500
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 0.00012760343793723764,
+      "loss": 0.8142,
+      "step": 577000
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 0.00012745352788327003,
+      "loss": 0.813,
+      "step": 577500
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 0.00012730361782930242,
+      "loss": 0.815,
+      "step": 578000
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 0.00012715370777533478,
+      "loss": 0.8185,
+      "step": 578500
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 0.00012700379772136717,
+      "loss": 0.8199,
+      "step": 579000
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 0.00012685388766739953,
+      "loss": 0.7992,
+      "step": 579500
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 0.00012670397761343192,
+      "loss": 0.819,
+      "step": 580000
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 0.0001265540675594643,
+      "loss": 0.8076,
+      "step": 580500
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 0.0001264041575054967,
+      "loss": 0.8015,
+      "step": 581000
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 0.00012625424745152906,
+      "loss": 0.7965,
+      "step": 581500
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 0.00012610433739756145,
+      "loss": 0.8002,
+      "step": 582000
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 0.00012595442734359381,
+      "loss": 0.8406,
+      "step": 582500
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 0.0001258045172896262,
+      "loss": 0.7868,
+      "step": 583000
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 0.0001256546072356586,
+      "loss": 0.8061,
+      "step": 583500
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 0.00012550469718169098,
+      "loss": 0.7948,
+      "step": 584000
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 0.00012535478712772334,
+      "loss": 0.8087,
+      "step": 584500
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 0.00012520487707375573,
+      "loss": 0.7958,
+      "step": 585000
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 0.00012505496701978812,
+      "loss": 0.8001,
+      "step": 585500
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 0.00012490505696582049,
+      "loss": 0.8074,
+      "step": 586000
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 0.00012475514691185288,
+      "loss": 0.7942,
+      "step": 586500
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 0.00012460523685788526,
+      "loss": 0.8019,
+      "step": 587000
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 0.00012445532680391763,
+      "loss": 0.8053,
+      "step": 587500
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 0.00012430541674995002,
+      "loss": 0.7859,
+      "step": 588000
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 0.0001241555066959824,
+      "loss": 0.8108,
+      "step": 588500
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 0.00012400559664201477,
+      "loss": 0.7906,
+      "step": 589000
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 0.00012385568658804716,
+      "loss": 0.796,
+      "step": 589500
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 0.00012370577653407955,
+      "loss": 0.804,
+      "step": 590000
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 0.0001235558664801119,
+      "loss": 0.8052,
+      "step": 590500
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 0.0001234059564261443,
+      "loss": 0.8186,
+      "step": 591000
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 0.0001232560463721767,
+      "loss": 0.7831,
+      "step": 591500
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 0.00012310613631820908,
+      "loss": 0.8121,
+      "step": 592000
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 0.00012295622626424144,
+      "loss": 0.818,
+      "step": 592500
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 0.00012280631621027383,
+      "loss": 0.837,
+      "step": 593000
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 0.0001226564061563062,
+      "loss": 0.801,
+      "step": 593500
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 0.00012250649610233858,
+      "loss": 0.823,
+      "step": 594000
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 0.00012235658604837097,
+      "loss": 0.8078,
+      "step": 594500
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 0.00012220667599440336,
+      "loss": 0.8081,
+      "step": 595000
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 0.00012205676594043574,
+      "loss": 0.821,
+      "step": 595500
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 0.00012190685588646811,
+      "loss": 0.8113,
+      "step": 596000
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 0.0001217569458325005,
+      "loss": 0.8105,
+      "step": 596500
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 0.00012160703577853286,
+      "loss": 0.7872,
+      "step": 597000
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 0.00012145712572456525,
+      "loss": 0.823,
+      "step": 597500
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 0.00012130721567059763,
+      "loss": 0.8174,
+      "step": 598000
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 0.00012115730561663002,
+      "loss": 0.8317,
+      "step": 598500
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 0.0001210073955626624,
+      "loss": 0.812,
+      "step": 599000
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 0.00012085748550869478,
+      "loss": 0.8176,
+      "step": 599500
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 0.00012070757545472715,
+      "loss": 0.8172,
+      "step": 600000
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 0.00012055766540075953,
+      "loss": 0.8251,
+      "step": 600500
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 0.00012040775534679191,
+      "loss": 0.8163,
+      "step": 601000
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 0.0001202578452928243,
+      "loss": 0.8245,
+      "step": 601500
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 0.00012010793523885668,
+      "loss": 0.7855,
+      "step": 602000
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 0.00011995802518488907,
+      "loss": 0.7866,
+      "step": 602500
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 0.00011980811513092143,
+      "loss": 0.7867,
+      "step": 603000
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 0.00011965820507695382,
+      "loss": 0.774,
+      "step": 603500
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 0.00011950829502298619,
+      "loss": 0.7726,
+      "step": 604000
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 0.00011935838496901858,
+      "loss": 0.7776,
+      "step": 604500
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 0.00011920847491505096,
+      "loss": 0.7951,
+      "step": 605000
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 0.00011905856486108335,
+      "loss": 0.7891,
+      "step": 605500
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 0.00011890865480711572,
+      "loss": 0.7854,
+      "step": 606000
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 0.0001187587447531481,
+      "loss": 0.7797,
+      "step": 606500
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 0.00011860883469918047,
+      "loss": 0.7791,
+      "step": 607000
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 0.00011845892464521286,
+      "loss": 0.7851,
+      "step": 607500
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 0.00011830901459124524,
+      "loss": 0.7862,
+      "step": 608000
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 0.00011815910453727763,
+      "loss": 0.7797,
+      "step": 608500
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 0.00011800919448331,
+      "loss": 0.7657,
+      "step": 609000
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 0.00011785928442934238,
+      "loss": 0.7805,
+      "step": 609500
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 0.00011770937437537476,
+      "loss": 0.769,
+      "step": 610000
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 0.00011755946432140715,
+      "loss": 0.7749,
+      "step": 610500
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 0.00011740955426743952,
+      "loss": 0.7916,
+      "step": 611000
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 0.00011725964421347191,
+      "loss": 0.7865,
+      "step": 611500
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 0.00011710973415950429,
+      "loss": 0.7756,
+      "step": 612000
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 0.00011695982410553668,
+      "loss": 0.7845,
+      "step": 612500
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 0.00011680991405156904,
+      "loss": 0.7873,
+      "step": 613000
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 0.00011666000399760143,
+      "loss": 0.7715,
+      "step": 613500
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 0.0001165100939436338,
+      "loss": 0.8001,
+      "step": 614000
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 0.0001163601838896662,
+      "loss": 0.7767,
+      "step": 614500
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 0.00011621027383569857,
+      "loss": 0.77,
+      "step": 615000
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 0.00011606036378173096,
+      "loss": 0.7755,
+      "step": 615500
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 0.00011591045372776332,
+      "loss": 0.7775,
+      "step": 616000
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 0.00011576054367379571,
+      "loss": 0.782,
+      "step": 616500
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 0.00011561063361982809,
+      "loss": 0.7852,
+      "step": 617000
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 0.00011546072356586048,
+      "loss": 0.7937,
+      "step": 617500
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 0.00011531081351189285,
+      "loss": 0.7965,
+      "step": 618000
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 0.00011516090345792524,
+      "loss": 0.7754,
+      "step": 618500
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 0.0001150109934039576,
+      "loss": 0.7859,
+      "step": 619000
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 0.00011486108334998999,
+      "loss": 0.7706,
+      "step": 619500
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 0.00011471117329602237,
+      "loss": 0.7759,
+      "step": 620000
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 0.00011456126324205476,
+      "loss": 0.7903,
+      "step": 620500
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 0.00011441135318808715,
+      "loss": 0.7997,
+      "step": 621000
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 0.00011426144313411952,
+      "loss": 0.7801,
+      "step": 621500
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 0.00011411153308015191,
+      "loss": 0.7991,
+      "step": 622000
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 0.00011396162302618427,
+      "loss": 0.7887,
+      "step": 622500
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 0.00011381171297221666,
+      "loss": 0.7778,
+      "step": 623000
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 0.00011366180291824904,
+      "loss": 0.7711,
+      "step": 623500
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 0.00011351189286428143,
+      "loss": 0.7817,
+      "step": 624000
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 0.0001133619828103138,
+      "loss": 0.7875,
+      "step": 624500
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 0.0001132120727563462,
+      "loss": 0.7986,
+      "step": 625000
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 0.00011306216270237856,
+      "loss": 0.7796,
+      "step": 625500
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 0.00011291225264841095,
+      "loss": 0.7672,
+      "step": 626000
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 0.00011276234259444332,
+      "loss": 0.7805,
+      "step": 626500
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 0.00011261243254047571,
+      "loss": 0.793,
+      "step": 627000
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 0.00011246252248650809,
+      "loss": 0.7845,
+      "step": 627500
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 0.00011231261243254048,
+      "loss": 0.7806,
+      "step": 628000
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 0.00011216270237857284,
+      "loss": 0.7976,
+      "step": 628500
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 0.00011201279232460523,
+      "loss": 0.7803,
+      "step": 629000
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 0.0001118628822706376,
+      "loss": 0.7834,
+      "step": 629500
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 0.00011171297221667,
+      "loss": 0.813,
+      "step": 630000
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 0.00011156306216270237,
+      "loss": 0.7723,
+      "step": 630500
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 0.00011141315210873476,
+      "loss": 0.802,
+      "step": 631000
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 0.00011126324205476712,
+      "loss": 0.7794,
+      "step": 631500
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 0.00011111333200079951,
+      "loss": 0.7882,
+      "step": 632000
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 0.00011096342194683189,
+      "loss": 0.786,
+      "step": 632500
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 0.00011081351189286428,
+      "loss": 0.7925,
+      "step": 633000
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 0.00011066360183889665,
+      "loss": 0.7908,
+      "step": 633500
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 0.00011051369178492904,
+      "loss": 0.8137,
+      "step": 634000
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 0.0001103637817309614,
+      "loss": 0.7613,
+      "step": 634500
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 0.00011021387167699379,
+      "loss": 0.7876,
+      "step": 635000
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 0.00011006396162302617,
+      "loss": 0.7526,
+      "step": 635500
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 0.00010991405156905856,
+      "loss": 0.7465,
+      "step": 636000
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 0.00010976414151509093,
+      "loss": 0.7527,
+      "step": 636500
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 0.00010961423146112332,
+      "loss": 0.7546,
+      "step": 637000
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 0.00010946432140715569,
+      "loss": 0.7488,
+      "step": 637500
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 0.00010931441135318808,
+      "loss": 0.7402,
+      "step": 638000
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 0.00010916450129922045,
+      "loss": 0.7522,
+      "step": 638500
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 0.00010901459124525284,
+      "loss": 0.7469,
+      "step": 639000
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 0.00010886468119128522,
+      "loss": 0.7611,
+      "step": 639500
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 0.0001087147711373176,
+      "loss": 0.7638,
+      "step": 640000
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 0.00010856486108334997,
+      "loss": 0.7631,
+      "step": 640500
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 0.00010841495102938236,
+      "loss": 0.7469,
+      "step": 641000
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 0.00010826504097541473,
+      "loss": 0.7798,
+      "step": 641500
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 0.00010811513092144712,
+      "loss": 0.7436,
+      "step": 642000
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 0.0001079652208674795,
+      "loss": 0.7482,
+      "step": 642500
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 0.00010781531081351189,
+      "loss": 0.7438,
+      "step": 643000
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 0.00010766540075954426,
+      "loss": 0.7723,
+      "step": 643500
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 0.00010751549070557664,
+      "loss": 0.7569,
+      "step": 644000
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 0.00010736558065160902,
+      "loss": 0.7418,
+      "step": 644500
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 0.0001072156705976414,
+      "loss": 0.777,
+      "step": 645000
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 0.00010706576054367378,
+      "loss": 0.777,
+      "step": 645500
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 0.00010691585048970617,
+      "loss": 0.7607,
+      "step": 646000
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 0.00010676594043573855,
+      "loss": 0.7646,
+      "step": 646500
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 0.00010661603038177092,
+      "loss": 0.7722,
+      "step": 647000
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 0.0001064661203278033,
+      "loss": 0.7728,
+      "step": 647500
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 0.00010631621027383569,
+      "loss": 0.7713,
+      "step": 648000
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 0.00010616630021986808,
+      "loss": 0.769,
+      "step": 648500
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 0.00010601639016590045,
+      "loss": 0.7701,
+      "step": 649000
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 0.00010586648011193284,
+      "loss": 0.7586,
+      "step": 649500
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 0.00010571657005796522,
+      "loss": 0.7718,
+      "step": 650000
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 0.00010556666000399759,
+      "loss": 0.7559,
+      "step": 650500
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 0.00010541674995002997,
+      "loss": 0.7584,
+      "step": 651000
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 0.00010526683989606236,
+      "loss": 0.7553,
+      "step": 651500
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 0.00010511692984209473,
+      "loss": 0.7665,
+      "step": 652000
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 0.00010496701978812712,
+      "loss": 0.7818,
+      "step": 652500
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 0.0001048171097341595,
+      "loss": 0.769,
+      "step": 653000
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 0.00010466719968019189,
+      "loss": 0.78,
+      "step": 653500
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 0.00010451728962622425,
+      "loss": 0.7654,
+      "step": 654000
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 0.00010436737957225664,
+      "loss": 0.7674,
+      "step": 654500
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 0.00010421746951828902,
+      "loss": 0.7649,
+      "step": 655000
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 0.0001040675594643214,
+      "loss": 0.7472,
+      "step": 655500
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 0.00010391764941035378,
+      "loss": 0.7706,
+      "step": 656000
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 0.00010376773935638617,
+      "loss": 0.7545,
+      "step": 656500
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 0.00010361782930241853,
+      "loss": 0.7544,
+      "step": 657000
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 0.00010346791924845092,
+      "loss": 0.7495,
+      "step": 657500
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 0.0001033180091944833,
+      "loss": 0.7769,
+      "step": 658000
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 0.00010316809914051569,
+      "loss": 0.7843,
+      "step": 658500
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 0.00010301818908654806,
+      "loss": 0.7679,
+      "step": 659000
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 0.00010286827903258045,
+      "loss": 0.7687,
+      "step": 659500
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 0.00010271836897861282,
+      "loss": 0.7832,
+      "step": 660000
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 0.0001025684589246452,
+      "loss": 0.778,
+      "step": 660500
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 0.00010241854887067758,
+      "loss": 0.77,
+      "step": 661000
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 0.00010226863881670997,
+      "loss": 0.7664,
+      "step": 661500
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 0.00010211872876274235,
+      "loss": 0.7876,
+      "step": 662000
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 0.00010196881870877474,
+      "loss": 0.7612,
+      "step": 662500
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 0.0001018189086548071,
+      "loss": 0.7588,
+      "step": 663000
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 0.00010166899860083949,
+      "loss": 0.7595,
+      "step": 663500
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 0.00010151908854687186,
+      "loss": 0.7599,
+      "step": 664000
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 0.00010136917849290425,
+      "loss": 0.7689,
+      "step": 664500
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 0.00010121926843893663,
+      "loss": 0.764,
+      "step": 665000
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 0.00010106935838496902,
+      "loss": 0.7808,
+      "step": 665500
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 0.00010091944833100138,
+      "loss": 0.7665,
+      "step": 666000
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 0.00010076953827703377,
+      "loss": 0.7719,
+      "step": 666500
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 0.00010061962822306614,
+      "loss": 0.7652,
+      "step": 667000
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 0.00010046971816909853,
+      "loss": 0.7793,
+      "step": 667500
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 0.00010031980811513091,
+      "loss": 0.7464,
+      "step": 668000
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 0.0001001698980611633,
+      "loss": 0.7796,
+      "step": 668500
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 0.00010001998800719566,
+      "loss": 0.724,
+      "step": 669000
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 9.987007795322805e-05,
+      "loss": 0.7415,
+      "step": 669500
+    },
+    {
+      "epoch": 20.05,
+      "learning_rate": 9.972016789926043e-05,
+      "loss": 0.7282,
+      "step": 670000
+    },
+    {
+      "epoch": 20.06,
+      "learning_rate": 9.957025784529282e-05,
+      "loss": 0.7349,
+      "step": 670500
+    },
+    {
+      "epoch": 20.08,
+      "learning_rate": 9.942034779132519e-05,
+      "loss": 0.7546,
+      "step": 671000
+    },
+    {
+      "epoch": 20.09,
+      "learning_rate": 9.927043773735758e-05,
+      "loss": 0.7421,
+      "step": 671500
+    },
+    {
+      "epoch": 20.11,
+      "learning_rate": 9.912052768338994e-05,
+      "loss": 0.726,
+      "step": 672000
+    },
+    {
+      "epoch": 20.12,
+      "learning_rate": 9.897061762942233e-05,
+      "loss": 0.725,
+      "step": 672500
+    },
+    {
+      "epoch": 20.14,
+      "learning_rate": 9.882070757545471e-05,
+      "loss": 0.7467,
+      "step": 673000
+    },
+    {
+      "epoch": 20.15,
+      "learning_rate": 9.86707975214871e-05,
+      "loss": 0.7317,
+      "step": 673500
+    },
+    {
+      "epoch": 20.17,
+      "learning_rate": 9.852088746751947e-05,
+      "loss": 0.7282,
+      "step": 674000
+    },
+    {
+      "epoch": 20.18,
+      "learning_rate": 9.837097741355186e-05,
+      "loss": 0.7455,
+      "step": 674500
+    },
+    {
+      "epoch": 20.2,
+      "learning_rate": 9.822106735958425e-05,
+      "loss": 0.7315,
+      "step": 675000
+    },
+    {
+      "epoch": 20.21,
+      "learning_rate": 9.807115730561662e-05,
+      "loss": 0.7273,
+      "step": 675500
+    },
+    {
+      "epoch": 20.23,
+      "learning_rate": 9.7921247251649e-05,
+      "loss": 0.7523,
+      "step": 676000
+    },
+    {
+      "epoch": 20.24,
+      "learning_rate": 9.777133719768138e-05,
+      "loss": 0.7213,
+      "step": 676500
+    },
+    {
+      "epoch": 20.26,
+      "learning_rate": 9.762142714371377e-05,
+      "loss": 0.7507,
+      "step": 677000
+    },
+    {
+      "epoch": 20.27,
+      "learning_rate": 9.747151708974615e-05,
+      "loss": 0.733,
+      "step": 677500
+    },
+    {
+      "epoch": 20.29,
+      "learning_rate": 9.732160703577854e-05,
+      "loss": 0.7152,
+      "step": 678000
+    },
+    {
+      "epoch": 20.3,
+      "learning_rate": 9.71716969818109e-05,
+      "loss": 0.7536,
+      "step": 678500
+    },
+    {
+      "epoch": 20.32,
+      "learning_rate": 9.702178692784329e-05,
+      "loss": 0.7405,
+      "step": 679000
+    },
+    {
+      "epoch": 20.33,
+      "learning_rate": 9.687187687387566e-05,
+      "loss": 0.7512,
+      "step": 679500
+    },
+    {
+      "epoch": 20.35,
+      "learning_rate": 9.672196681990805e-05,
+      "loss": 0.7611,
+      "step": 680000
+    },
+    {
+      "epoch": 20.36,
+      "learning_rate": 9.657205676594043e-05,
+      "loss": 0.7535,
+      "step": 680500
+    },
+    {
+      "epoch": 20.38,
+      "learning_rate": 9.642214671197282e-05,
+      "loss": 0.7368,
+      "step": 681000
+    },
+    {
+      "epoch": 20.39,
+      "learning_rate": 9.627223665800518e-05,
+      "loss": 0.7505,
+      "step": 681500
+    },
+    {
+      "epoch": 20.41,
+      "learning_rate": 9.612232660403757e-05,
+      "loss": 0.7415,
+      "step": 682000
+    },
+    {
+      "epoch": 20.42,
+      "learning_rate": 9.597241655006995e-05,
+      "loss": 0.7752,
+      "step": 682500
+    },
+    {
+      "epoch": 20.44,
+      "learning_rate": 9.582250649610233e-05,
+      "loss": 0.7458,
+      "step": 683000
+    },
+    {
+      "epoch": 20.45,
+      "learning_rate": 9.567259644213471e-05,
+      "loss": 0.7421,
+      "step": 683500
+    },
+    {
+      "epoch": 20.47,
+      "learning_rate": 9.55226863881671e-05,
+      "loss": 0.7389,
+      "step": 684000
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 9.537277633419948e-05,
+      "loss": 0.7445,
+      "step": 684500
+    },
+    {
+      "epoch": 20.5,
+      "learning_rate": 9.522286628023185e-05,
+      "loss": 0.741,
+      "step": 685000
+    },
+    {
+      "epoch": 20.51,
+      "learning_rate": 9.507295622626423e-05,
+      "loss": 0.7429,
+      "step": 685500
+    },
+    {
+      "epoch": 20.53,
+      "learning_rate": 9.492304617229662e-05,
+      "loss": 0.7493,
+      "step": 686000
+    },
+    {
+      "epoch": 20.54,
+      "learning_rate": 9.477313611832899e-05,
+      "loss": 0.7443,
+      "step": 686500
+    },
+    {
+      "epoch": 20.56,
+      "learning_rate": 9.462322606436138e-05,
+      "loss": 0.7411,
+      "step": 687000
+    },
+    {
+      "epoch": 20.57,
+      "learning_rate": 9.447331601039376e-05,
+      "loss": 0.7465,
+      "step": 687500
+    },
+    {
+      "epoch": 20.59,
+      "learning_rate": 9.432340595642613e-05,
+      "loss": 0.751,
+      "step": 688000
+    },
+    {
+      "epoch": 20.6,
+      "learning_rate": 9.417349590245851e-05,
+      "loss": 0.7332,
+      "step": 688500
+    },
+    {
+      "epoch": 20.62,
+      "learning_rate": 9.40235858484909e-05,
+      "loss": 0.7249,
+      "step": 689000
+    },
+    {
+      "epoch": 20.63,
+      "learning_rate": 9.387367579452327e-05,
+      "loss": 0.7333,
+      "step": 689500
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 9.372376574055566e-05,
+      "loss": 0.75,
+      "step": 690000
+    },
+    {
+      "epoch": 20.66,
+      "learning_rate": 9.357385568658804e-05,
+      "loss": 0.7328,
+      "step": 690500
+    },
+    {
+      "epoch": 20.68,
+      "learning_rate": 9.342394563262043e-05,
+      "loss": 0.7474,
+      "step": 691000
+    },
+    {
+      "epoch": 20.69,
+      "learning_rate": 9.327403557865279e-05,
+      "loss": 0.7118,
+      "step": 691500
+    },
+    {
+      "epoch": 20.71,
+      "learning_rate": 9.312412552468518e-05,
+      "loss": 0.761,
+      "step": 692000
+    },
+    {
+      "epoch": 20.72,
+      "learning_rate": 9.297421547071756e-05,
+      "loss": 0.7404,
+      "step": 692500
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 9.282430541674995e-05,
+      "loss": 0.7349,
+      "step": 693000
+    },
+    {
+      "epoch": 20.75,
+      "learning_rate": 9.267439536278232e-05,
+      "loss": 0.7429,
+      "step": 693500
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 9.252448530881471e-05,
+      "loss": 0.7511,
+      "step": 694000
+    },
+    {
+      "epoch": 20.78,
+      "learning_rate": 9.237457525484707e-05,
+      "loss": 0.7384,
+      "step": 694500
+    },
+    {
+      "epoch": 20.8,
+      "learning_rate": 9.222466520087946e-05,
+      "loss": 0.7365,
+      "step": 695000
+    },
+    {
+      "epoch": 20.81,
+      "learning_rate": 9.207475514691184e-05,
+      "loss": 0.7331,
+      "step": 695500
+    },
+    {
+      "epoch": 20.83,
+      "learning_rate": 9.192484509294423e-05,
+      "loss": 0.7343,
+      "step": 696000
+    },
+    {
+      "epoch": 20.84,
+      "learning_rate": 9.17749350389766e-05,
+      "loss": 0.7515,
+      "step": 696500
+    },
+    {
+      "epoch": 20.86,
+      "learning_rate": 9.1625024985009e-05,
+      "loss": 0.7536,
+      "step": 697000
+    },
+    {
+      "epoch": 20.87,
+      "learning_rate": 9.147511493104136e-05,
+      "loss": 0.7434,
+      "step": 697500
+    },
+    {
+      "epoch": 20.89,
+      "learning_rate": 9.132520487707375e-05,
+      "loss": 0.742,
+      "step": 698000
+    },
+    {
+      "epoch": 20.9,
+      "learning_rate": 9.117529482310612e-05,
+      "loss": 0.7389,
+      "step": 698500
+    },
+    {
+      "epoch": 20.92,
+      "learning_rate": 9.102538476913851e-05,
+      "loss": 0.7434,
+      "step": 699000
+    },
+    {
+      "epoch": 20.93,
+      "learning_rate": 9.087547471517089e-05,
+      "loss": 0.74,
+      "step": 699500
+    },
+    {
+      "epoch": 20.95,
+      "learning_rate": 9.072556466120328e-05,
+      "loss": 0.7416,
+      "step": 700000
+    },
+    {
+      "epoch": 20.96,
+      "learning_rate": 9.057565460723564e-05,
+      "loss": 0.7508,
+      "step": 700500
+    },
+    {
+      "epoch": 20.98,
+      "learning_rate": 9.042574455326803e-05,
+      "loss": 0.7466,
+      "step": 701000
+    },
+    {
+      "epoch": 20.99,
+      "learning_rate": 9.027583449930042e-05,
+      "loss": 0.7605,
+      "step": 701500
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 9.012592444533279e-05,
+      "loss": 0.7463,
+      "step": 702000
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 8.997601439136518e-05,
+      "loss": 0.6942,
+      "step": 702500
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 8.982610433739756e-05,
+      "loss": 0.7358,
+      "step": 703000
+    },
+    {
+      "epoch": 21.05,
+      "learning_rate": 8.967619428342995e-05,
+      "loss": 0.7123,
+      "step": 703500
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 8.952628422946231e-05,
+      "loss": 0.7118,
+      "step": 704000
+    },
+    {
+      "epoch": 21.08,
+      "learning_rate": 8.93763741754947e-05,
+      "loss": 0.7199,
+      "step": 704500
+    },
+    {
+      "epoch": 21.09,
+      "learning_rate": 8.922646412152707e-05,
+      "loss": 0.7228,
+      "step": 705000
+    },
+    {
+      "epoch": 21.11,
+      "learning_rate": 8.907655406755946e-05,
+      "loss": 0.722,
+      "step": 705500
+    },
+    {
+      "epoch": 21.12,
+      "learning_rate": 8.892664401359184e-05,
+      "loss": 0.7148,
+      "step": 706000
+    },
+    {
+      "epoch": 21.14,
+      "learning_rate": 8.877673395962423e-05,
+      "loss": 0.7344,
+      "step": 706500
+    },
+    {
+      "epoch": 21.15,
+      "learning_rate": 8.862682390565659e-05,
+      "loss": 0.7272,
+      "step": 707000
+    },
+    {
+      "epoch": 21.17,
+      "learning_rate": 8.847691385168898e-05,
+      "loss": 0.7139,
+      "step": 707500
+    },
+    {
+      "epoch": 21.18,
+      "learning_rate": 8.832700379772136e-05,
+      "loss": 0.7126,
+      "step": 708000
+    },
+    {
+      "epoch": 21.2,
+      "learning_rate": 8.817709374375375e-05,
+      "loss": 0.7043,
+      "step": 708500
+    },
+    {
+      "epoch": 21.21,
+      "learning_rate": 8.802718368978612e-05,
+      "loss": 0.7254,
+      "step": 709000
+    },
+    {
+      "epoch": 21.23,
+      "learning_rate": 8.787727363581851e-05,
+      "loss": 0.7072,
+      "step": 709500
+    },
+    {
+      "epoch": 21.24,
+      "learning_rate": 8.772736358185087e-05,
+      "loss": 0.7152,
+      "step": 710000
+    },
+    {
+      "epoch": 21.26,
+      "learning_rate": 8.757745352788326e-05,
+      "loss": 0.7134,
+      "step": 710500
+    },
+    {
+      "epoch": 21.27,
+      "learning_rate": 8.742754347391564e-05,
+      "loss": 0.7019,
+      "step": 711000
+    },
+    {
+      "epoch": 21.29,
+      "learning_rate": 8.727763341994803e-05,
+      "loss": 0.7126,
+      "step": 711500
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 8.71277233659804e-05,
+      "loss": 0.7193,
+      "step": 712000
+    },
+    {
+      "epoch": 21.32,
+      "learning_rate": 8.69778133120128e-05,
+      "loss": 0.7195,
+      "step": 712500
+    },
+    {
+      "epoch": 21.33,
+      "learning_rate": 8.682790325804516e-05,
+      "loss": 0.7284,
+      "step": 713000
+    },
+    {
+      "epoch": 21.35,
+      "learning_rate": 8.667799320407755e-05,
+      "loss": 0.7042,
+      "step": 713500
+    },
+    {
+      "epoch": 21.36,
+      "learning_rate": 8.652808315010992e-05,
+      "loss": 0.7415,
+      "step": 714000
+    },
+    {
+      "epoch": 21.38,
+      "learning_rate": 8.637817309614231e-05,
+      "loss": 0.7127,
+      "step": 714500
+    },
+    {
+      "epoch": 21.39,
+      "learning_rate": 8.622826304217469e-05,
+      "loss": 0.7264,
+      "step": 715000
+    },
+    {
+      "epoch": 21.41,
+      "learning_rate": 8.607835298820708e-05,
+      "loss": 0.7006,
+      "step": 715500
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 8.592844293423944e-05,
+      "loss": 0.7225,
+      "step": 716000
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 8.577853288027183e-05,
+      "loss": 0.7225,
+      "step": 716500
+    },
+    {
+      "epoch": 21.45,
+      "learning_rate": 8.56286228263042e-05,
+      "loss": 0.7278,
+      "step": 717000
+    },
+    {
+      "epoch": 21.47,
+      "learning_rate": 8.547871277233659e-05,
+      "loss": 0.7053,
+      "step": 717500
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 8.532880271836897e-05,
+      "loss": 0.718,
+      "step": 718000
+    },
+    {
+      "epoch": 21.5,
+      "learning_rate": 8.517889266440136e-05,
+      "loss": 0.7299,
+      "step": 718500
+    },
+    {
+      "epoch": 21.51,
+      "learning_rate": 8.502898261043372e-05,
+      "loss": 0.7361,
+      "step": 719000
+    },
+    {
+      "epoch": 21.53,
+      "learning_rate": 8.487907255646611e-05,
+      "loss": 0.7269,
+      "step": 719500
+    },
+    {
+      "epoch": 21.54,
+      "learning_rate": 8.472916250249849e-05,
+      "loss": 0.7084,
+      "step": 720000
+    },
+    {
+      "epoch": 21.56,
+      "learning_rate": 8.457925244853087e-05,
+      "loss": 0.703,
+      "step": 720500
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 8.442934239456325e-05,
+      "loss": 0.7262,
+      "step": 721000
+    },
+    {
+      "epoch": 21.59,
+      "learning_rate": 8.427943234059564e-05,
+      "loss": 0.7207,
+      "step": 721500
+    },
+    {
+      "epoch": 21.6,
+      "learning_rate": 8.412952228662802e-05,
+      "loss": 0.7139,
+      "step": 722000
+    },
+    {
+      "epoch": 21.62,
+      "learning_rate": 8.397961223266039e-05,
+      "loss": 0.7218,
+      "step": 722500
+    },
+    {
+      "epoch": 21.63,
+      "learning_rate": 8.382970217869277e-05,
+      "loss": 0.7413,
+      "step": 723000
+    },
+    {
+      "epoch": 21.65,
+      "learning_rate": 8.367979212472516e-05,
+      "loss": 0.7296,
+      "step": 723500
+    },
+    {
+      "epoch": 21.66,
+      "learning_rate": 8.352988207075753e-05,
+      "loss": 0.7139,
+      "step": 724000
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 8.337997201678992e-05,
+      "loss": 0.7162,
+      "step": 724500
+    },
+    {
+      "epoch": 21.69,
+      "learning_rate": 8.32300619628223e-05,
+      "loss": 0.7115,
+      "step": 725000
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 8.308015190885467e-05,
+      "loss": 0.7271,
+      "step": 725500
+    },
+    {
+      "epoch": 21.72,
+      "learning_rate": 8.293024185488705e-05,
+      "loss": 0.7199,
+      "step": 726000
+    },
+    {
+      "epoch": 21.74,
+      "learning_rate": 8.278033180091944e-05,
+      "loss": 0.7108,
+      "step": 726500
+    },
+    {
+      "epoch": 21.75,
+      "learning_rate": 8.263042174695182e-05,
+      "loss": 0.7193,
+      "step": 727000
+    },
+    {
+      "epoch": 21.77,
+      "learning_rate": 8.24805116929842e-05,
+      "loss": 0.737,
+      "step": 727500
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 8.233060163901658e-05,
+      "loss": 0.7248,
+      "step": 728000
+    },
+    {
+      "epoch": 21.8,
+      "learning_rate": 8.218069158504897e-05,
+      "loss": 0.723,
+      "step": 728500
+    },
+    {
+      "epoch": 21.81,
+      "learning_rate": 8.203078153108135e-05,
+      "loss": 0.7185,
+      "step": 729000
+    },
+    {
+      "epoch": 21.83,
+      "learning_rate": 8.188087147711372e-05,
+      "loss": 0.7156,
+      "step": 729500
+    },
+    {
+      "epoch": 21.84,
+      "learning_rate": 8.173096142314611e-05,
+      "loss": 0.7256,
+      "step": 730000
+    },
+    {
+      "epoch": 21.86,
+      "learning_rate": 8.158105136917849e-05,
+      "loss": 0.7303,
+      "step": 730500
+    },
+    {
+      "epoch": 21.87,
+      "learning_rate": 8.143114131521088e-05,
+      "loss": 0.7176,
+      "step": 731000
+    },
+    {
+      "epoch": 21.89,
+      "learning_rate": 8.128123126124325e-05,
+      "loss": 0.7116,
+      "step": 731500
+    },
+    {
+      "epoch": 21.9,
+      "learning_rate": 8.113132120727564e-05,
+      "loss": 0.7327,
+      "step": 732000
+    },
+    {
+      "epoch": 21.92,
+      "learning_rate": 8.0981411153308e-05,
+      "loss": 0.7312,
+      "step": 732500
+    },
+    {
+      "epoch": 21.93,
+      "learning_rate": 8.083150109934039e-05,
+      "loss": 0.7383,
+      "step": 733000
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 8.068159104537277e-05,
+      "loss": 0.7232,
+      "step": 733500
+    },
+    {
+      "epoch": 21.96,
+      "learning_rate": 8.053168099140516e-05,
+      "loss": 0.7309,
+      "step": 734000
+    },
+    {
+      "epoch": 21.98,
+      "learning_rate": 8.038177093743753e-05,
+      "loss": 0.7291,
+      "step": 734500
+    },
+    {
+      "epoch": 21.99,
+      "learning_rate": 8.023186088346992e-05,
+      "loss": 0.7378,
+      "step": 735000
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 8.008195082950229e-05,
+      "loss": 0.7132,
+      "step": 735500
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 7.993204077553468e-05,
+      "loss": 0.6955,
+      "step": 736000
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 7.978213072156705e-05,
+      "loss": 0.6798,
+      "step": 736500
+    },
+    {
+      "epoch": 22.05,
+      "learning_rate": 7.963222066759944e-05,
+      "loss": 0.7013,
+      "step": 737000
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 7.948231061363182e-05,
+      "loss": 0.6835,
+      "step": 737500
+    },
+    {
+      "epoch": 22.08,
+      "learning_rate": 7.93324005596642e-05,
+      "loss": 0.7122,
+      "step": 738000
+    },
+    {
+      "epoch": 22.1,
+      "learning_rate": 7.918249050569657e-05,
+      "loss": 0.6835,
+      "step": 738500
+    },
+    {
+      "epoch": 22.11,
+      "learning_rate": 7.903258045172896e-05,
+      "loss": 0.6997,
+      "step": 739000
+    },
+    {
+      "epoch": 22.13,
+      "learning_rate": 7.888267039776133e-05,
+      "loss": 0.7069,
+      "step": 739500
+    },
+    {
+      "epoch": 22.14,
+      "learning_rate": 7.873276034379372e-05,
+      "loss": 0.6875,
+      "step": 740000
+    },
+    {
+      "epoch": 22.16,
+      "learning_rate": 7.85828502898261e-05,
+      "loss": 0.7001,
+      "step": 740500
+    },
+    {
+      "epoch": 22.17,
+      "learning_rate": 7.843294023585849e-05,
+      "loss": 0.706,
+      "step": 741000
+    },
+    {
+      "epoch": 22.19,
+      "learning_rate": 7.828303018189085e-05,
+      "loss": 0.6986,
+      "step": 741500
+    },
+    {
+      "epoch": 22.2,
+      "learning_rate": 7.813312012792324e-05,
+      "loss": 0.6979,
+      "step": 742000
+    },
+    {
+      "epoch": 22.22,
+      "learning_rate": 7.798321007395562e-05,
+      "loss": 0.7109,
+      "step": 742500
+    },
+    {
+      "epoch": 22.23,
+      "learning_rate": 7.7833300019988e-05,
+      "loss": 0.6914,
+      "step": 743000
+    },
+    {
+      "epoch": 22.25,
+      "learning_rate": 7.768338996602038e-05,
+      "loss": 0.7009,
+      "step": 743500
+    },
+    {
+      "epoch": 22.26,
+      "learning_rate": 7.753347991205277e-05,
+      "loss": 0.6839,
+      "step": 744000
+    },
+    {
+      "epoch": 22.28,
+      "learning_rate": 7.738356985808513e-05,
+      "loss": 0.7103,
+      "step": 744500
+    },
+    {
+      "epoch": 22.29,
+      "learning_rate": 7.723365980411752e-05,
+      "loss": 0.6977,
+      "step": 745000
+    },
+    {
+      "epoch": 22.31,
+      "learning_rate": 7.70837497501499e-05,
+      "loss": 0.7084,
+      "step": 745500
+    },
+    {
+      "epoch": 22.32,
+      "learning_rate": 7.693383969618229e-05,
+      "loss": 0.7139,
+      "step": 746000
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 7.678392964221466e-05,
+      "loss": 0.7109,
+      "step": 746500
+    },
+    {
+      "epoch": 22.35,
+      "learning_rate": 7.663401958824705e-05,
+      "loss": 0.6968,
+      "step": 747000
+    },
+    {
+      "epoch": 22.37,
+      "learning_rate": 7.648410953427941e-05,
+      "loss": 0.6996,
+      "step": 747500
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 7.63341994803118e-05,
+      "loss": 0.7147,
+      "step": 748000
+    },
+    {
+      "epoch": 22.4,
+      "learning_rate": 7.618428942634418e-05,
+      "loss": 0.7082,
+      "step": 748500
+    },
+    {
+      "epoch": 22.41,
+      "learning_rate": 7.603437937237657e-05,
+      "loss": 0.7011,
+      "step": 749000
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 7.588446931840894e-05,
+      "loss": 0.7058,
+      "step": 749500
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 7.573455926444133e-05,
+      "loss": 0.7101,
+      "step": 750000
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 7.55846492104737e-05,
+      "loss": 0.7117,
+      "step": 750500
+    },
+    {
+      "epoch": 22.47,
+      "learning_rate": 7.543473915650609e-05,
+      "loss": 0.6966,
+      "step": 751000
+    },
+    {
+      "epoch": 22.49,
+      "learning_rate": 7.528482910253846e-05,
+      "loss": 0.7064,
+      "step": 751500
+    },
+    {
+      "epoch": 22.5,
+      "learning_rate": 7.513491904857085e-05,
+      "loss": 0.701,
+      "step": 752000
+    },
+    {
+      "epoch": 22.52,
+      "learning_rate": 7.498500899460323e-05,
+      "loss": 0.7059,
+      "step": 752500
+    },
+    {
+      "epoch": 22.53,
+      "learning_rate": 7.483509894063562e-05,
+      "loss": 0.6802,
+      "step": 753000
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 7.468518888666799e-05,
+      "loss": 0.699,
+      "step": 753500
+    },
+    {
+      "epoch": 22.56,
+      "learning_rate": 7.453527883270037e-05,
+      "loss": 0.7006,
+      "step": 754000
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 7.438536877873276e-05,
+      "loss": 0.7014,
+      "step": 754500
+    },
+    {
+      "epoch": 22.59,
+      "learning_rate": 7.423545872476513e-05,
+      "loss": 0.6971,
+      "step": 755000
+    },
+    {
+      "epoch": 22.61,
+      "learning_rate": 7.408554867079751e-05,
+      "loss": 0.6937,
+      "step": 755500
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 7.39356386168299e-05,
+      "loss": 0.7018,
+      "step": 756000
+    },
+    {
+      "epoch": 22.64,
+      "learning_rate": 7.378572856286227e-05,
+      "loss": 0.7045,
+      "step": 756500
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 7.363581850889465e-05,
+      "loss": 0.7015,
+      "step": 757000
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 7.348590845492704e-05,
+      "loss": 0.7073,
+      "step": 757500
+    },
+    {
+      "epoch": 22.68,
+      "learning_rate": 7.333599840095942e-05,
+      "loss": 0.7053,
+      "step": 758000
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 7.318608834699179e-05,
+      "loss": 0.7054,
+      "step": 758500
+    },
+    {
+      "epoch": 22.71,
+      "learning_rate": 7.303617829302418e-05,
+      "loss": 0.7045,
+      "step": 759000
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 7.288626823905656e-05,
+      "loss": 0.72,
+      "step": 759500
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 7.273635818508893e-05,
+      "loss": 0.712,
+      "step": 760000
+    },
+    {
+      "epoch": 22.76,
+      "learning_rate": 7.258644813112132e-05,
+      "loss": 0.6888,
+      "step": 760500
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 7.24365380771537e-05,
+      "loss": 0.6955,
+      "step": 761000
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 7.228662802318607e-05,
+      "loss": 0.7007,
+      "step": 761500
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 7.213671796921846e-05,
+      "loss": 0.7146,
+      "step": 762000
+    },
+    {
+      "epoch": 22.82,
+      "learning_rate": 7.198680791525084e-05,
+      "loss": 0.7208,
+      "step": 762500
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 7.183689786128323e-05,
+      "loss": 0.685,
+      "step": 763000
+    },
+    {
+      "epoch": 22.85,
+      "learning_rate": 7.16869878073156e-05,
+      "loss": 0.7014,
+      "step": 763500
+    },
+    {
+      "epoch": 22.86,
+      "learning_rate": 7.153707775334798e-05,
+      "loss": 0.6998,
+      "step": 764000
+    },
+    {
+      "epoch": 22.88,
+      "learning_rate": 7.138716769938037e-05,
+      "loss": 0.7147,
+      "step": 764500
+    },
+    {
+      "epoch": 22.89,
+      "learning_rate": 7.123725764541274e-05,
+      "loss": 0.6832,
+      "step": 765000
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 7.108734759144512e-05,
+      "loss": 0.704,
+      "step": 765500
+    },
+    {
+      "epoch": 22.92,
+      "learning_rate": 7.093743753747751e-05,
+      "loss": 0.6902,
+      "step": 766000
+    },
+    {
+      "epoch": 22.94,
+      "learning_rate": 7.078752748350989e-05,
+      "loss": 0.6992,
+      "step": 766500
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 7.063761742954226e-05,
+      "loss": 0.6888,
+      "step": 767000
+    },
+    {
+      "epoch": 22.96,
+      "learning_rate": 7.048770737557465e-05,
+      "loss": 0.7033,
+      "step": 767500
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 7.033779732160703e-05,
+      "loss": 0.7144,
+      "step": 768000
+    },
+    {
+      "epoch": 22.99,
+      "learning_rate": 7.01878872676394e-05,
+      "loss": 0.7113,
+      "step": 768500
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 7.003797721367179e-05,
+      "loss": 0.6798,
+      "step": 769000
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 6.988806715970418e-05,
+      "loss": 0.6728,
+      "step": 769500
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 6.973815710573656e-05,
+      "loss": 0.6779,
+      "step": 770000
+    },
+    {
+      "epoch": 23.05,
+      "learning_rate": 6.958824705176893e-05,
+      "loss": 0.6711,
+      "step": 770500
+    },
+    {
+      "epoch": 23.07,
+      "learning_rate": 6.943833699780132e-05,
+      "loss": 0.6738,
+      "step": 771000
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 6.92884269438337e-05,
+      "loss": 0.6686,
+      "step": 771500
+    },
+    {
+      "epoch": 23.1,
+      "learning_rate": 6.913851688986607e-05,
+      "loss": 0.6869,
+      "step": 772000
+    },
+    {
+      "epoch": 23.11,
+      "learning_rate": 6.898860683589846e-05,
+      "loss": 0.6738,
+      "step": 772500
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 6.883869678193084e-05,
+      "loss": 0.6937,
+      "step": 773000
+    },
+    {
+      "epoch": 23.14,
+      "learning_rate": 6.868878672796322e-05,
+      "loss": 0.6813,
+      "step": 773500
+    },
+    {
+      "epoch": 23.16,
+      "learning_rate": 6.85388766739956e-05,
+      "loss": 0.6874,
+      "step": 774000
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 6.838896662002798e-05,
+      "loss": 0.6621,
+      "step": 774500
+    },
+    {
+      "epoch": 23.19,
+      "learning_rate": 6.823905656606036e-05,
+      "loss": 0.683,
+      "step": 775000
+    },
+    {
+      "epoch": 23.2,
+      "learning_rate": 6.808914651209275e-05,
+      "loss": 0.6892,
+      "step": 775500
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 6.793923645812512e-05,
+      "loss": 0.672,
+      "step": 776000
+    },
+    {
+      "epoch": 23.23,
+      "learning_rate": 6.77893264041575e-05,
+      "loss": 0.6959,
+      "step": 776500
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 6.763941635018989e-05,
+      "loss": 0.6863,
+      "step": 777000
+    },
+    {
+      "epoch": 23.26,
+      "learning_rate": 6.748950629622226e-05,
+      "loss": 0.6691,
+      "step": 777500
+    },
+    {
+      "epoch": 23.28,
+      "learning_rate": 6.733959624225464e-05,
+      "loss": 0.685,
+      "step": 778000
+    },
+    {
+      "epoch": 23.29,
+      "learning_rate": 6.718968618828703e-05,
+      "loss": 0.6724,
+      "step": 778500
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 6.70397761343194e-05,
+      "loss": 0.6846,
+      "step": 779000
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 6.688986608035178e-05,
+      "loss": 0.6773,
+      "step": 779500
+    },
+    {
+      "epoch": 23.34,
+      "learning_rate": 6.673995602638417e-05,
+      "loss": 0.6831,
+      "step": 780000
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 6.659004597241655e-05,
+      "loss": 0.6989,
+      "step": 780500
+    },
+    {
+      "epoch": 23.37,
+      "learning_rate": 6.644013591844892e-05,
+      "loss": 0.681,
+      "step": 781000
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 6.629022586448131e-05,
+      "loss": 0.6751,
+      "step": 781500
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 6.614031581051369e-05,
+      "loss": 0.6723,
+      "step": 782000
+    },
+    {
+      "epoch": 23.41,
+      "learning_rate": 6.599040575654606e-05,
+      "loss": 0.6995,
+      "step": 782500
+    },
+    {
+      "epoch": 23.43,
+      "learning_rate": 6.584049570257845e-05,
+      "loss": 0.6783,
+      "step": 783000
+    },
+    {
+      "epoch": 23.44,
+      "learning_rate": 6.569058564861083e-05,
+      "loss": 0.6738,
+      "step": 783500
+    },
+    {
+      "epoch": 23.46,
+      "learning_rate": 6.55406755946432e-05,
+      "loss": 0.6933,
+      "step": 784000
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 6.539076554067559e-05,
+      "loss": 0.706,
+      "step": 784500
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 6.524085548670797e-05,
+      "loss": 0.6589,
+      "step": 785000
+    },
+    {
+      "epoch": 23.5,
+      "learning_rate": 6.509094543274034e-05,
+      "loss": 0.68,
+      "step": 785500
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 6.494103537877273e-05,
+      "loss": 0.6614,
+      "step": 786000
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 6.479112532480511e-05,
+      "loss": 0.6856,
+      "step": 786500
+    },
+    {
+      "epoch": 23.55,
+      "learning_rate": 6.464121527083749e-05,
+      "loss": 0.6812,
+      "step": 787000
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 6.449130521686987e-05,
+      "loss": 0.6689,
+      "step": 787500
+    },
+    {
+      "epoch": 23.58,
+      "learning_rate": 6.434139516290225e-05,
+      "loss": 0.6791,
+      "step": 788000
+    },
+    {
+      "epoch": 23.59,
+      "learning_rate": 6.419148510893463e-05,
+      "loss": 0.69,
+      "step": 788500
+    },
+    {
+      "epoch": 23.61,
+      "learning_rate": 6.404157505496702e-05,
+      "loss": 0.6721,
+      "step": 789000
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 6.389166500099939e-05,
+      "loss": 0.6702,
+      "step": 789500
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 6.374175494703177e-05,
+      "loss": 0.6939,
+      "step": 790000
+    },
+    {
+      "epoch": 23.65,
+      "learning_rate": 6.359184489306416e-05,
+      "loss": 0.6791,
+      "step": 790500
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 6.344193483909653e-05,
+      "loss": 0.6826,
+      "step": 791000
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 6.329202478512891e-05,
+      "loss": 0.6859,
+      "step": 791500
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 6.31421147311613e-05,
+      "loss": 0.6778,
+      "step": 792000
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 6.299220467719367e-05,
+      "loss": 0.6745,
+      "step": 792500
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 6.284229462322605e-05,
+      "loss": 0.7018,
+      "step": 793000
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 6.269238456925844e-05,
+      "loss": 0.6882,
+      "step": 793500
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 6.254247451529081e-05,
+      "loss": 0.6737,
+      "step": 794000
+    },
+    {
+      "epoch": 23.77,
+      "learning_rate": 6.239256446132319e-05,
+      "loss": 0.681,
+      "step": 794500
+    },
+    {
+      "epoch": 23.79,
+      "learning_rate": 6.224265440735558e-05,
+      "loss": 0.6998,
+      "step": 795000
+    },
+    {
+      "epoch": 23.8,
+      "learning_rate": 6.209274435338797e-05,
+      "loss": 0.6691,
+      "step": 795500
+    },
+    {
+      "epoch": 23.82,
+      "learning_rate": 6.194283429942035e-05,
+      "loss": 0.668,
+      "step": 796000
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 6.179292424545272e-05,
+      "loss": 0.6894,
+      "step": 796500
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 6.164301419148511e-05,
+      "loss": 0.674,
+      "step": 797000
+    },
+    {
+      "epoch": 23.86,
+      "learning_rate": 6.149310413751749e-05,
+      "loss": 0.675,
+      "step": 797500
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 6.134319408354986e-05,
+      "loss": 0.6843,
+      "step": 798000
+    },
+    {
+      "epoch": 23.89,
+      "learning_rate": 6.119328402958225e-05,
+      "loss": 0.6909,
+      "step": 798500
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 6.104337397561463e-05,
+      "loss": 0.6875,
+      "step": 799000
+    },
+    {
+      "epoch": 23.92,
+      "learning_rate": 6.089346392164701e-05,
+      "loss": 0.6623,
+      "step": 799500
+    },
+    {
+      "epoch": 23.94,
+      "learning_rate": 6.074355386767939e-05,
+      "loss": 0.6976,
+      "step": 800000
+    },
+    {
+      "epoch": 23.95,
+      "learning_rate": 6.059364381371177e-05,
+      "loss": 0.6769,
+      "step": 800500
+    },
+    {
+      "epoch": 23.97,
+      "learning_rate": 6.044373375974415e-05,
+      "loss": 0.6848,
+      "step": 801000
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 6.0293823705776534e-05,
+      "loss": 0.677,
+      "step": 801500
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 6.014391365180891e-05,
+      "loss": 0.6994,
+      "step": 802000
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 5.999400359784129e-05,
+      "loss": 0.6587,
+      "step": 802500
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 5.9844093543873675e-05,
+      "loss": 0.6472,
+      "step": 803000
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 5.969418348990605e-05,
+      "loss": 0.6612,
+      "step": 803500
+    },
+    {
+      "epoch": 24.06,
+      "learning_rate": 5.954427343593843e-05,
+      "loss": 0.666,
+      "step": 804000
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 5.9394363381970816e-05,
+      "loss": 0.6726,
+      "step": 804500
+    },
+    {
+      "epoch": 24.09,
+      "learning_rate": 5.924445332800319e-05,
+      "loss": 0.6788,
+      "step": 805000
+    },
+    {
+      "epoch": 24.1,
+      "learning_rate": 5.9094543274035574e-05,
+      "loss": 0.6556,
+      "step": 805500
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 5.894463322006796e-05,
+      "loss": 0.6668,
+      "step": 806000
+    },
+    {
+      "epoch": 24.13,
+      "learning_rate": 5.879472316610033e-05,
+      "loss": 0.6642,
+      "step": 806500
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 5.8644813112132715e-05,
+      "loss": 0.6391,
+      "step": 807000
+    },
+    {
+      "epoch": 24.16,
+      "learning_rate": 5.84949030581651e-05,
+      "loss": 0.6513,
+      "step": 807500
+    },
+    {
+      "epoch": 24.18,
+      "learning_rate": 5.8344993004197474e-05,
+      "loss": 0.6436,
+      "step": 808000
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 5.8195082950229857e-05,
+      "loss": 0.666,
+      "step": 808500
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 5.804517289626224e-05,
+      "loss": 0.6687,
+      "step": 809000
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 5.7895262842294615e-05,
+      "loss": 0.6581,
+      "step": 809500
+    },
+    {
+      "epoch": 24.24,
+      "learning_rate": 5.7745352788327e-05,
+      "loss": 0.6699,
+      "step": 810000
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 5.759544273435938e-05,
+      "loss": 0.6666,
+      "step": 810500
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 5.7445532680391756e-05,
+      "loss": 0.6646,
+      "step": 811000
+    },
+    {
+      "epoch": 24.28,
+      "learning_rate": 5.729562262642414e-05,
+      "loss": 0.6724,
+      "step": 811500
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 5.714571257245652e-05,
+      "loss": 0.6653,
+      "step": 812000
+    },
+    {
+      "epoch": 24.31,
+      "learning_rate": 5.6995802518488904e-05,
+      "loss": 0.6727,
+      "step": 812500
+    },
+    {
+      "epoch": 24.33,
+      "learning_rate": 5.684589246452128e-05,
+      "loss": 0.667,
+      "step": 813000
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 5.669598241055366e-05,
+      "loss": 0.6709,
+      "step": 813500
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 5.6546072356586045e-05,
+      "loss": 0.6627,
+      "step": 814000
+    },
+    {
+      "epoch": 24.37,
+      "learning_rate": 5.639616230261842e-05,
+      "loss": 0.6675,
+      "step": 814500
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 5.6246252248650803e-05,
+      "loss": 0.672,
+      "step": 815000
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 5.6096342194683186e-05,
+      "loss": 0.6638,
+      "step": 815500
+    },
+    {
+      "epoch": 24.42,
+      "learning_rate": 5.594643214071556e-05,
+      "loss": 0.6593,
+      "step": 816000
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 5.5796522086747945e-05,
+      "loss": 0.6783,
+      "step": 816500
+    },
+    {
+      "epoch": 24.45,
+      "learning_rate": 5.564661203278033e-05,
+      "loss": 0.6722,
+      "step": 817000
+    },
+    {
+      "epoch": 24.46,
+      "learning_rate": 5.54967019788127e-05,
+      "loss": 0.6607,
+      "step": 817500
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 5.5346791924845086e-05,
+      "loss": 0.6755,
+      "step": 818000
+    },
+    {
+      "epoch": 24.49,
+      "learning_rate": 5.519688187087747e-05,
+      "loss": 0.6711,
+      "step": 818500
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 5.5046971816909844e-05,
+      "loss": 0.6612,
+      "step": 819000
+    },
+    {
+      "epoch": 24.52,
+      "learning_rate": 5.489706176294223e-05,
+      "loss": 0.6583,
+      "step": 819500
+    },
+    {
+      "epoch": 24.54,
+      "learning_rate": 5.474715170897461e-05,
+      "loss": 0.65,
+      "step": 820000
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 5.4597241655006985e-05,
+      "loss": 0.6659,
+      "step": 820500
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 5.444733160103937e-05,
+      "loss": 0.6639,
+      "step": 821000
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 5.429742154707175e-05,
+      "loss": 0.6719,
+      "step": 821500
+    },
+    {
+      "epoch": 24.6,
+      "learning_rate": 5.4147511493104126e-05,
+      "loss": 0.6724,
+      "step": 822000
+    },
+    {
+      "epoch": 24.61,
+      "learning_rate": 5.3997601439136516e-05,
+      "loss": 0.6769,
+      "step": 822500
+    },
+    {
+      "epoch": 24.63,
+      "learning_rate": 5.38476913851689e-05,
+      "loss": 0.6643,
+      "step": 823000
+    },
+    {
+      "epoch": 24.64,
+      "learning_rate": 5.369778133120128e-05,
+      "loss": 0.6632,
+      "step": 823500
+    },
+    {
+      "epoch": 24.66,
+      "learning_rate": 5.354787127723366e-05,
+      "loss": 0.6651,
+      "step": 824000
+    },
+    {
+      "epoch": 24.67,
+      "learning_rate": 5.339796122326604e-05,
+      "loss": 0.6684,
+      "step": 824500
+    },
+    {
+      "epoch": 24.69,
+      "learning_rate": 5.324805116929842e-05,
+      "loss": 0.6965,
+      "step": 825000
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 5.30981411153308e-05,
+      "loss": 0.6577,
+      "step": 825500
+    },
+    {
+      "epoch": 24.72,
+      "learning_rate": 5.294823106136318e-05,
+      "loss": 0.6623,
+      "step": 826000
+    },
+    {
+      "epoch": 24.73,
+      "learning_rate": 5.279832100739556e-05,
+      "loss": 0.6553,
+      "step": 826500
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 5.264841095342794e-05,
+      "loss": 0.6655,
+      "step": 827000
+    },
+    {
+      "epoch": 24.76,
+      "learning_rate": 5.249850089946032e-05,
+      "loss": 0.6778,
+      "step": 827500
+    },
+    {
+      "epoch": 24.78,
+      "learning_rate": 5.2348590845492704e-05,
+      "loss": 0.6642,
+      "step": 828000
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 5.219868079152508e-05,
+      "loss": 0.651,
+      "step": 828500
+    },
+    {
+      "epoch": 24.81,
+      "learning_rate": 5.204877073755746e-05,
+      "loss": 0.6645,
+      "step": 829000
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 5.1898860683589845e-05,
+      "loss": 0.6534,
+      "step": 829500
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 5.174895062962222e-05,
+      "loss": 0.6581,
+      "step": 830000
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 5.1599040575654604e-05,
+      "loss": 0.6703,
+      "step": 830500
+    },
+    {
+      "epoch": 24.86,
+      "learning_rate": 5.1449130521686986e-05,
+      "loss": 0.6563,
+      "step": 831000
+    },
+    {
+      "epoch": 24.88,
+      "learning_rate": 5.129922046771936e-05,
+      "loss": 0.6623,
+      "step": 831500
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 5.1149310413751745e-05,
+      "loss": 0.6609,
+      "step": 832000
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 5.099940035978413e-05,
+      "loss": 0.6681,
+      "step": 832500
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 5.08494903058165e-05,
+      "loss": 0.6611,
+      "step": 833000
+    },
+    {
+      "epoch": 24.94,
+      "learning_rate": 5.0699580251848886e-05,
+      "loss": 0.6411,
+      "step": 833500
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 5.054967019788127e-05,
+      "loss": 0.6546,
+      "step": 834000
+    },
+    {
+      "epoch": 24.97,
+      "learning_rate": 5.039976014391365e-05,
+      "loss": 0.6809,
+      "step": 834500
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 5.024985008994603e-05,
+      "loss": 0.6626,
+      "step": 835000
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 5.009994003597841e-05,
+      "loss": 0.6552,
+      "step": 835500
+    },
+    {
+      "epoch": 25.01,
+      "learning_rate": 4.995002998201079e-05,
+      "loss": 0.6281,
+      "step": 836000
+    },
+    {
+      "epoch": 25.03,
+      "learning_rate": 4.980011992804317e-05,
+      "loss": 0.6428,
+      "step": 836500
+    },
+    {
+      "epoch": 25.04,
+      "learning_rate": 4.965020987407555e-05,
+      "loss": 0.6264,
+      "step": 837000
+    },
+    {
+      "epoch": 25.06,
+      "learning_rate": 4.950029982010793e-05,
+      "loss": 0.6402,
+      "step": 837500
+    },
+    {
+      "epoch": 25.07,
+      "learning_rate": 4.935038976614031e-05,
+      "loss": 0.6562,
+      "step": 838000
+    },
+    {
+      "epoch": 25.09,
+      "learning_rate": 4.920047971217269e-05,
+      "loss": 0.6611,
+      "step": 838500
+    },
+    {
+      "epoch": 25.1,
+      "learning_rate": 4.9050569658205074e-05,
+      "loss": 0.6615,
+      "step": 839000
+    },
+    {
+      "epoch": 25.12,
+      "learning_rate": 4.890065960423745e-05,
+      "loss": 0.6239,
+      "step": 839500
+    },
+    {
+      "epoch": 25.13,
+      "learning_rate": 4.875074955026983e-05,
+      "loss": 0.6465,
+      "step": 840000
+    },
+    {
+      "epoch": 25.15,
+      "learning_rate": 4.8600839496302215e-05,
+      "loss": 0.6687,
+      "step": 840500
+    },
+    {
+      "epoch": 25.16,
+      "learning_rate": 4.845092944233459e-05,
+      "loss": 0.6259,
+      "step": 841000
+    },
+    {
+      "epoch": 25.18,
+      "learning_rate": 4.8301019388366974e-05,
+      "loss": 0.6472,
+      "step": 841500
+    },
+    {
+      "epoch": 25.19,
+      "learning_rate": 4.8151109334399356e-05,
+      "loss": 0.6368,
+      "step": 842000
+    },
+    {
+      "epoch": 25.21,
+      "learning_rate": 4.800119928043173e-05,
+      "loss": 0.6379,
+      "step": 842500
+    },
+    {
+      "epoch": 25.22,
+      "learning_rate": 4.7851289226464115e-05,
+      "loss": 0.6472,
+      "step": 843000
+    },
+    {
+      "epoch": 25.24,
+      "learning_rate": 4.77013791724965e-05,
+      "loss": 0.6604,
+      "step": 843500
+    },
+    {
+      "epoch": 25.25,
+      "learning_rate": 4.755146911852887e-05,
+      "loss": 0.6564,
+      "step": 844000
+    },
+    {
+      "epoch": 25.27,
+      "learning_rate": 4.7401559064561256e-05,
+      "loss": 0.6387,
+      "step": 844500
+    },
+    {
+      "epoch": 25.28,
+      "learning_rate": 4.725164901059364e-05,
+      "loss": 0.6429,
+      "step": 845000
+    },
+    {
+      "epoch": 25.3,
+      "learning_rate": 4.7101738956626014e-05,
+      "loss": 0.643,
+      "step": 845500
+    },
+    {
+      "epoch": 25.31,
+      "learning_rate": 4.69518289026584e-05,
+      "loss": 0.6486,
+      "step": 846000
+    },
+    {
+      "epoch": 25.33,
+      "learning_rate": 4.680191884869078e-05,
+      "loss": 0.6445,
+      "step": 846500
+    },
+    {
+      "epoch": 25.34,
+      "learning_rate": 4.6652008794723155e-05,
+      "loss": 0.6459,
+      "step": 847000
+    },
+    {
+      "epoch": 25.36,
+      "learning_rate": 4.650209874075554e-05,
+      "loss": 0.6522,
+      "step": 847500
+    },
+    {
+      "epoch": 25.37,
+      "learning_rate": 4.635218868678792e-05,
+      "loss": 0.6538,
+      "step": 848000
+    },
+    {
+      "epoch": 25.39,
+      "learning_rate": 4.6202278632820297e-05,
+      "loss": 0.6454,
+      "step": 848500
+    },
+    {
+      "epoch": 25.4,
+      "learning_rate": 4.605236857885268e-05,
+      "loss": 0.639,
+      "step": 849000
+    },
+    {
+      "epoch": 25.42,
+      "learning_rate": 4.590245852488507e-05,
+      "loss": 0.6469,
+      "step": 849500
+    },
+    {
+      "epoch": 25.43,
+      "learning_rate": 4.575254847091745e-05,
+      "loss": 0.6471,
+      "step": 850000
+    },
+    {
+      "epoch": 25.45,
+      "learning_rate": 4.560263841694983e-05,
+      "loss": 0.6511,
+      "step": 850500
+    },
+    {
+      "epoch": 25.46,
+      "learning_rate": 4.545272836298221e-05,
+      "loss": 0.6218,
+      "step": 851000
+    },
+    {
+      "epoch": 25.48,
+      "learning_rate": 4.530281830901459e-05,
+      "loss": 0.6448,
+      "step": 851500
+    },
+    {
+      "epoch": 25.49,
+      "learning_rate": 4.515290825504697e-05,
+      "loss": 0.6532,
+      "step": 852000
+    },
+    {
+      "epoch": 25.51,
+      "learning_rate": 4.500299820107935e-05,
+      "loss": 0.6322,
+      "step": 852500
+    },
+    {
+      "epoch": 25.52,
+      "learning_rate": 4.485308814711173e-05,
+      "loss": 0.6316,
+      "step": 853000
+    },
+    {
+      "epoch": 25.54,
+      "learning_rate": 4.470317809314411e-05,
+      "loss": 0.6508,
+      "step": 853500
+    },
+    {
+      "epoch": 25.55,
+      "learning_rate": 4.455326803917649e-05,
+      "loss": 0.6638,
+      "step": 854000
+    },
+    {
+      "epoch": 25.57,
+      "learning_rate": 4.4403357985208874e-05,
+      "loss": 0.6425,
+      "step": 854500
+    },
+    {
+      "epoch": 25.58,
+      "learning_rate": 4.425344793124126e-05,
+      "loss": 0.6351,
+      "step": 855000
+    },
+    {
+      "epoch": 25.6,
+      "learning_rate": 4.410353787727363e-05,
+      "loss": 0.6314,
+      "step": 855500
+    },
+    {
+      "epoch": 25.61,
+      "learning_rate": 4.3953627823306015e-05,
+      "loss": 0.6572,
+      "step": 856000
+    },
+    {
+      "epoch": 25.63,
+      "learning_rate": 4.38037177693384e-05,
+      "loss": 0.66,
+      "step": 856500
+    },
+    {
+      "epoch": 25.64,
+      "learning_rate": 4.3653807715370774e-05,
+      "loss": 0.663,
+      "step": 857000
+    },
+    {
+      "epoch": 25.66,
+      "learning_rate": 4.3503897661403157e-05,
+      "loss": 0.6635,
+      "step": 857500
+    },
+    {
+      "epoch": 25.67,
+      "learning_rate": 4.335398760743554e-05,
+      "loss": 0.6706,
+      "step": 858000
+    },
+    {
+      "epoch": 25.69,
+      "learning_rate": 4.3204077553467915e-05,
+      "loss": 0.6471,
+      "step": 858500
+    },
+    {
+      "epoch": 25.7,
+      "learning_rate": 4.30541674995003e-05,
+      "loss": 0.6481,
+      "step": 859000
+    },
+    {
+      "epoch": 25.72,
+      "learning_rate": 4.290425744553268e-05,
+      "loss": 0.6528,
+      "step": 859500
+    },
+    {
+      "epoch": 25.73,
+      "learning_rate": 4.2754347391565056e-05,
+      "loss": 0.6387,
+      "step": 860000
+    },
+    {
+      "epoch": 25.75,
+      "learning_rate": 4.260443733759744e-05,
+      "loss": 0.6513,
+      "step": 860500
+    },
+    {
+      "epoch": 25.76,
+      "learning_rate": 4.245452728362982e-05,
+      "loss": 0.6378,
+      "step": 861000
+    },
+    {
+      "epoch": 25.78,
+      "learning_rate": 4.23046172296622e-05,
+      "loss": 0.6323,
+      "step": 861500
+    },
+    {
+      "epoch": 25.79,
+      "learning_rate": 4.215470717569458e-05,
+      "loss": 0.6511,
+      "step": 862000
+    },
+    {
+      "epoch": 25.81,
+      "learning_rate": 4.200479712172696e-05,
+      "loss": 0.6434,
+      "step": 862500
+    },
+    {
+      "epoch": 25.82,
+      "learning_rate": 4.185488706775934e-05,
+      "loss": 0.651,
+      "step": 863000
+    },
+    {
+      "epoch": 25.84,
+      "learning_rate": 4.170497701379172e-05,
+      "loss": 0.6527,
+      "step": 863500
+    },
+    {
+      "epoch": 25.85,
+      "learning_rate": 4.1555066959824103e-05,
+      "loss": 0.647,
+      "step": 864000
+    },
+    {
+      "epoch": 25.87,
+      "learning_rate": 4.140515690585648e-05,
+      "loss": 0.6418,
+      "step": 864500
+    },
+    {
+      "epoch": 25.88,
+      "learning_rate": 4.125524685188886e-05,
+      "loss": 0.6563,
+      "step": 865000
+    },
+    {
+      "epoch": 25.9,
+      "learning_rate": 4.1105336797921245e-05,
+      "loss": 0.6521,
+      "step": 865500
+    },
+    {
+      "epoch": 25.91,
+      "learning_rate": 4.095542674395362e-05,
+      "loss": 0.6432,
+      "step": 866000
+    },
+    {
+      "epoch": 25.93,
+      "learning_rate": 4.0805516689986e-05,
+      "loss": 0.6474,
+      "step": 866500
+    },
+    {
+      "epoch": 25.94,
+      "learning_rate": 4.0655606636018386e-05,
+      "loss": 0.6422,
+      "step": 867000
+    },
+    {
+      "epoch": 25.96,
+      "learning_rate": 4.050569658205076e-05,
+      "loss": 0.6438,
+      "step": 867500
+    },
+    {
+      "epoch": 25.97,
+      "learning_rate": 4.0355786528083144e-05,
+      "loss": 0.6535,
+      "step": 868000
+    },
+    {
+      "epoch": 25.99,
+      "learning_rate": 4.020587647411553e-05,
+      "loss": 0.6365,
+      "step": 868500
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 4.00559664201479e-05,
+      "loss": 0.6319,
+      "step": 869000
+    },
+    {
+      "epoch": 26.02,
+      "learning_rate": 3.9906056366180285e-05,
+      "loss": 0.63,
+      "step": 869500
+    },
+    {
+      "epoch": 26.03,
+      "learning_rate": 3.975614631221267e-05,
+      "loss": 0.6262,
+      "step": 870000
+    },
+    {
+      "epoch": 26.05,
+      "learning_rate": 3.960623625824505e-05,
+      "loss": 0.6163,
+      "step": 870500
+    },
+    {
+      "epoch": 26.06,
+      "learning_rate": 3.9456326204277426e-05,
+      "loss": 0.6219,
+      "step": 871000
+    },
+    {
+      "epoch": 26.08,
+      "learning_rate": 3.930641615030981e-05,
+      "loss": 0.6415,
+      "step": 871500
+    },
+    {
+      "epoch": 26.09,
+      "learning_rate": 3.915650609634219e-05,
+      "loss": 0.6316,
+      "step": 872000
+    },
+    {
+      "epoch": 26.11,
+      "learning_rate": 3.900659604237457e-05,
+      "loss": 0.6261,
+      "step": 872500
+    },
+    {
+      "epoch": 26.12,
+      "learning_rate": 3.885668598840695e-05,
+      "loss": 0.6339,
+      "step": 873000
+    },
+    {
+      "epoch": 26.14,
+      "learning_rate": 3.870677593443933e-05,
+      "loss": 0.6443,
+      "step": 873500
+    },
+    {
+      "epoch": 26.15,
+      "learning_rate": 3.855686588047171e-05,
+      "loss": 0.63,
+      "step": 874000
+    },
+    {
+      "epoch": 26.17,
+      "learning_rate": 3.840695582650409e-05,
+      "loss": 0.6502,
+      "step": 874500
+    },
+    {
+      "epoch": 26.18,
+      "learning_rate": 3.8257045772536474e-05,
+      "loss": 0.6242,
+      "step": 875000
+    },
+    {
+      "epoch": 26.2,
+      "learning_rate": 3.810713571856885e-05,
+      "loss": 0.6333,
+      "step": 875500
+    },
+    {
+      "epoch": 26.21,
+      "learning_rate": 3.795722566460123e-05,
+      "loss": 0.6432,
+      "step": 876000
+    },
+    {
+      "epoch": 26.23,
+      "learning_rate": 3.780731561063362e-05,
+      "loss": 0.6342,
+      "step": 876500
+    },
+    {
+      "epoch": 26.24,
+      "learning_rate": 3.7657405556666004e-05,
+      "loss": 0.6199,
+      "step": 877000
+    },
+    {
+      "epoch": 26.26,
+      "learning_rate": 3.750749550269838e-05,
+      "loss": 0.6216,
+      "step": 877500
+    },
+    {
+      "epoch": 26.27,
+      "learning_rate": 3.7357585448730756e-05,
+      "loss": 0.6354,
+      "step": 878000
+    },
+    {
+      "epoch": 26.29,
+      "learning_rate": 3.720767539476314e-05,
+      "loss": 0.6477,
+      "step": 878500
+    },
+    {
+      "epoch": 26.3,
+      "learning_rate": 3.705776534079552e-05,
+      "loss": 0.629,
+      "step": 879000
+    },
+    {
+      "epoch": 26.32,
+      "learning_rate": 3.69078552868279e-05,
+      "loss": 0.6368,
+      "step": 879500
+    },
+    {
+      "epoch": 26.33,
+      "learning_rate": 3.675794523286028e-05,
+      "loss": 0.6343,
+      "step": 880000
+    },
+    {
+      "epoch": 26.35,
+      "learning_rate": 3.660803517889266e-05,
+      "loss": 0.6169,
+      "step": 880500
+    },
+    {
+      "epoch": 26.36,
+      "learning_rate": 3.645812512492504e-05,
+      "loss": 0.6238,
+      "step": 881000
+    },
+    {
+      "epoch": 26.38,
+      "learning_rate": 3.630821507095742e-05,
+      "loss": 0.6187,
+      "step": 881500
+    },
+    {
+      "epoch": 26.39,
+      "learning_rate": 3.61583050169898e-05,
+      "loss": 0.6278,
+      "step": 882000
+    },
+    {
+      "epoch": 26.41,
+      "learning_rate": 3.600839496302218e-05,
+      "loss": 0.6354,
+      "step": 882500
+    },
+    {
+      "epoch": 26.42,
+      "learning_rate": 3.585848490905457e-05,
+      "loss": 0.6208,
+      "step": 883000
+    },
+    {
+      "epoch": 26.44,
+      "learning_rate": 3.5708574855086944e-05,
+      "loss": 0.619,
+      "step": 883500
+    },
+    {
+      "epoch": 26.45,
+      "learning_rate": 3.555866480111933e-05,
+      "loss": 0.6245,
+      "step": 884000
+    },
+    {
+      "epoch": 26.47,
+      "learning_rate": 3.540875474715171e-05,
+      "loss": 0.6249,
+      "step": 884500
+    },
+    {
+      "epoch": 26.48,
+      "learning_rate": 3.5258844693184085e-05,
+      "loss": 0.6148,
+      "step": 885000
+    },
+    {
+      "epoch": 26.5,
+      "learning_rate": 3.510893463921647e-05,
+      "loss": 0.6341,
+      "step": 885500
+    },
+    {
+      "epoch": 26.51,
+      "learning_rate": 3.495902458524885e-05,
+      "loss": 0.6235,
+      "step": 886000
+    },
+    {
+      "epoch": 26.53,
+      "learning_rate": 3.4809114531281226e-05,
+      "loss": 0.6234,
+      "step": 886500
+    },
+    {
+      "epoch": 26.54,
+      "learning_rate": 3.465920447731361e-05,
+      "loss": 0.6301,
+      "step": 887000
+    },
+    {
+      "epoch": 26.56,
+      "learning_rate": 3.450929442334599e-05,
+      "loss": 0.6289,
+      "step": 887500
+    },
+    {
+      "epoch": 26.57,
+      "learning_rate": 3.435938436937837e-05,
+      "loss": 0.6293,
+      "step": 888000
+    },
+    {
+      "epoch": 26.59,
+      "learning_rate": 3.420947431541075e-05,
+      "loss": 0.638,
+      "step": 888500
+    },
+    {
+      "epoch": 26.6,
+      "learning_rate": 3.405956426144313e-05,
+      "loss": 0.63,
+      "step": 889000
+    },
+    {
+      "epoch": 26.62,
+      "learning_rate": 3.390965420747551e-05,
+      "loss": 0.629,
+      "step": 889500
+    },
+    {
+      "epoch": 26.63,
+      "learning_rate": 3.375974415350789e-05,
+      "loss": 0.6204,
+      "step": 890000
+    },
+    {
+      "epoch": 26.65,
+      "learning_rate": 3.3609834099540274e-05,
+      "loss": 0.6449,
+      "step": 890500
+    },
+    {
+      "epoch": 26.66,
+      "learning_rate": 3.3459924045572656e-05,
+      "loss": 0.6446,
+      "step": 891000
+    },
+    {
+      "epoch": 26.68,
+      "learning_rate": 3.331001399160503e-05,
+      "loss": 0.6259,
+      "step": 891500
+    },
+    {
+      "epoch": 26.69,
+      "learning_rate": 3.3160103937637415e-05,
+      "loss": 0.6269,
+      "step": 892000
+    },
+    {
+      "epoch": 26.71,
+      "learning_rate": 3.30101938836698e-05,
+      "loss": 0.6279,
+      "step": 892500
+    },
+    {
+      "epoch": 26.72,
+      "learning_rate": 3.286028382970217e-05,
+      "loss": 0.6399,
+      "step": 893000
+    },
+    {
+      "epoch": 26.74,
+      "learning_rate": 3.2710373775734556e-05,
+      "loss": 0.6228,
+      "step": 893500
+    },
+    {
+      "epoch": 26.75,
+      "learning_rate": 3.256046372176694e-05,
+      "loss": 0.6184,
+      "step": 894000
+    },
+    {
+      "epoch": 26.77,
+      "learning_rate": 3.2410553667799314e-05,
+      "loss": 0.6319,
+      "step": 894500
+    },
+    {
+      "epoch": 26.78,
+      "learning_rate": 3.22606436138317e-05,
+      "loss": 0.6289,
+      "step": 895000
+    },
+    {
+      "epoch": 26.79,
+      "learning_rate": 3.211073355986408e-05,
+      "loss": 0.6381,
+      "step": 895500
+    },
+    {
+      "epoch": 26.81,
+      "learning_rate": 3.1960823505896455e-05,
+      "loss": 0.6255,
+      "step": 896000
+    },
+    {
+      "epoch": 26.82,
+      "learning_rate": 3.1810913451928845e-05,
+      "loss": 0.6276,
+      "step": 896500
+    },
+    {
+      "epoch": 26.84,
+      "learning_rate": 3.166100339796122e-05,
+      "loss": 0.6345,
+      "step": 897000
+    },
+    {
+      "epoch": 26.85,
+      "learning_rate": 3.15110933439936e-05,
+      "loss": 0.6359,
+      "step": 897500
+    },
+    {
+      "epoch": 26.87,
+      "learning_rate": 3.1361183290025986e-05,
+      "loss": 0.6383,
+      "step": 898000
+    },
+    {
+      "epoch": 26.88,
+      "learning_rate": 3.121127323605836e-05,
+      "loss": 0.6459,
+      "step": 898500
+    },
+    {
+      "epoch": 26.9,
+      "learning_rate": 3.1061363182090744e-05,
+      "loss": 0.6469,
+      "step": 899000
+    },
+    {
+      "epoch": 26.91,
+      "learning_rate": 3.091145312812313e-05,
+      "loss": 0.6389,
+      "step": 899500
+    },
+    {
+      "epoch": 26.93,
+      "learning_rate": 3.07615430741555e-05,
+      "loss": 0.6246,
+      "step": 900000
+    },
+    {
+      "epoch": 26.94,
+      "learning_rate": 3.0611633020187886e-05,
+      "loss": 0.6405,
+      "step": 900500
+    },
+    {
+      "epoch": 26.96,
+      "learning_rate": 3.0461722966220265e-05,
+      "loss": 0.6212,
+      "step": 901000
+    },
+    {
+      "epoch": 26.97,
+      "learning_rate": 3.0311812912252647e-05,
+      "loss": 0.6167,
+      "step": 901500
+    },
+    {
+      "epoch": 26.99,
+      "learning_rate": 3.0161902858285027e-05,
+      "loss": 0.6309,
+      "step": 902000
+    },
+    {
+      "epoch": 27.0,
+      "learning_rate": 3.0011992804317406e-05,
+      "loss": 0.6253,
+      "step": 902500
+    },
+    {
+      "epoch": 27.02,
+      "learning_rate": 2.986208275034979e-05,
+      "loss": 0.611,
+      "step": 903000
+    },
+    {
+      "epoch": 27.03,
+      "learning_rate": 2.9712172696382168e-05,
+      "loss": 0.6216,
+      "step": 903500
+    },
+    {
+      "epoch": 27.05,
+      "learning_rate": 2.9562262642414547e-05,
+      "loss": 0.6245,
+      "step": 904000
+    },
+    {
+      "epoch": 27.06,
+      "learning_rate": 2.941235258844693e-05,
+      "loss": 0.623,
+      "step": 904500
+    },
+    {
+      "epoch": 27.08,
+      "learning_rate": 2.926244253447931e-05,
+      "loss": 0.6232,
+      "step": 905000
+    },
+    {
+      "epoch": 27.09,
+      "learning_rate": 2.911253248051169e-05,
+      "loss": 0.6223,
+      "step": 905500
+    },
+    {
+      "epoch": 27.11,
+      "learning_rate": 2.896262242654407e-05,
+      "loss": 0.6055,
+      "step": 906000
+    },
+    {
+      "epoch": 27.12,
+      "learning_rate": 2.881271237257645e-05,
+      "loss": 0.6126,
+      "step": 906500
+    },
+    {
+      "epoch": 27.14,
+      "learning_rate": 2.8662802318608832e-05,
+      "loss": 0.6177,
+      "step": 907000
+    },
+    {
+      "epoch": 27.15,
+      "learning_rate": 2.851289226464121e-05,
+      "loss": 0.6049,
+      "step": 907500
+    },
+    {
+      "epoch": 27.17,
+      "learning_rate": 2.836298221067359e-05,
+      "loss": 0.6093,
+      "step": 908000
+    },
+    {
+      "epoch": 27.18,
+      "learning_rate": 2.8213072156705974e-05,
+      "loss": 0.6124,
+      "step": 908500
+    },
+    {
+      "epoch": 27.2,
+      "learning_rate": 2.8063162102738353e-05,
+      "loss": 0.6187,
+      "step": 909000
+    },
+    {
+      "epoch": 27.21,
+      "learning_rate": 2.7913252048770732e-05,
+      "loss": 0.6217,
+      "step": 909500
+    },
+    {
+      "epoch": 27.23,
+      "learning_rate": 2.7763341994803118e-05,
+      "loss": 0.6142,
+      "step": 910000
+    },
+    {
+      "epoch": 27.24,
+      "learning_rate": 2.7613431940835497e-05,
+      "loss": 0.6215,
+      "step": 910500
+    },
+    {
+      "epoch": 27.26,
+      "learning_rate": 2.746352188686788e-05,
+      "loss": 0.6221,
+      "step": 911000
+    },
+    {
+      "epoch": 27.27,
+      "learning_rate": 2.731361183290026e-05,
+      "loss": 0.6259,
+      "step": 911500
+    },
+    {
+      "epoch": 27.29,
+      "learning_rate": 2.7163701778932638e-05,
+      "loss": 0.6171,
+      "step": 912000
+    },
+    {
+      "epoch": 27.3,
+      "learning_rate": 2.701379172496502e-05,
+      "loss": 0.6168,
+      "step": 912500
+    },
+    {
+      "epoch": 27.32,
+      "learning_rate": 2.68638816709974e-05,
+      "loss": 0.6197,
+      "step": 913000
+    },
+    {
+      "epoch": 27.33,
+      "learning_rate": 2.671397161702978e-05,
+      "loss": 0.6158,
+      "step": 913500
+    },
+    {
+      "epoch": 27.35,
+      "learning_rate": 2.6564061563062162e-05,
+      "loss": 0.6203,
+      "step": 914000
+    },
+    {
+      "epoch": 27.36,
+      "learning_rate": 2.641415150909454e-05,
+      "loss": 0.5991,
+      "step": 914500
+    },
+    {
+      "epoch": 27.38,
+      "learning_rate": 2.626424145512692e-05,
+      "loss": 0.6053,
+      "step": 915000
+    },
+    {
+      "epoch": 27.39,
+      "learning_rate": 2.6114331401159303e-05,
+      "loss": 0.63,
+      "step": 915500
+    },
+    {
+      "epoch": 27.41,
+      "learning_rate": 2.5964421347191682e-05,
+      "loss": 0.5957,
+      "step": 916000
+    },
+    {
+      "epoch": 27.42,
+      "learning_rate": 2.5814511293224065e-05,
+      "loss": 0.627,
+      "step": 916500
+    },
+    {
+      "epoch": 27.44,
+      "learning_rate": 2.5664601239256444e-05,
+      "loss": 0.6242,
+      "step": 917000
+    },
+    {
+      "epoch": 27.45,
+      "learning_rate": 2.5514691185288823e-05,
+      "loss": 0.6362,
+      "step": 917500
+    },
+    {
+      "epoch": 27.47,
+      "learning_rate": 2.5364781131321206e-05,
+      "loss": 0.6132,
+      "step": 918000
+    },
+    {
+      "epoch": 27.48,
+      "learning_rate": 2.5214871077353585e-05,
+      "loss": 0.6239,
+      "step": 918500
+    },
+    {
+      "epoch": 27.5,
+      "learning_rate": 2.5064961023385964e-05,
+      "loss": 0.6159,
+      "step": 919000
+    },
+    {
+      "epoch": 27.51,
+      "learning_rate": 2.4915050969418347e-05,
+      "loss": 0.6065,
+      "step": 919500
+    },
+    {
+      "epoch": 27.53,
+      "learning_rate": 2.4765140915450726e-05,
+      "loss": 0.6253,
+      "step": 920000
+    },
+    {
+      "epoch": 27.54,
+      "learning_rate": 2.4615230861483106e-05,
+      "loss": 0.6141,
+      "step": 920500
+    },
+    {
+      "epoch": 27.56,
+      "learning_rate": 2.4465320807515488e-05,
+      "loss": 0.6172,
+      "step": 921000
+    },
+    {
+      "epoch": 27.57,
+      "learning_rate": 2.4315410753547867e-05,
+      "loss": 0.6115,
+      "step": 921500
+    },
+    {
+      "epoch": 27.59,
+      "learning_rate": 2.4165500699580247e-05,
+      "loss": 0.6192,
+      "step": 922000
+    },
+    {
+      "epoch": 27.6,
+      "learning_rate": 2.401559064561263e-05,
+      "loss": 0.6169,
+      "step": 922500
+    },
+    {
+      "epoch": 27.62,
+      "learning_rate": 2.386568059164501e-05,
+      "loss": 0.6115,
+      "step": 923000
+    },
+    {
+      "epoch": 27.63,
+      "learning_rate": 2.3715770537677394e-05,
+      "loss": 0.6232,
+      "step": 923500
+    },
+    {
+      "epoch": 27.65,
+      "learning_rate": 2.3565860483709774e-05,
+      "loss": 0.6213,
+      "step": 924000
+    },
+    {
+      "epoch": 27.66,
+      "learning_rate": 2.3415950429742153e-05,
+      "loss": 0.6113,
+      "step": 924500
+    },
+    {
+      "epoch": 27.68,
+      "learning_rate": 2.3266040375774536e-05,
+      "loss": 0.6111,
+      "step": 925000
+    },
+    {
+      "epoch": 27.69,
+      "learning_rate": 2.3116130321806915e-05,
+      "loss": 0.6208,
+      "step": 925500
+    },
+    {
+      "epoch": 27.71,
+      "learning_rate": 2.2966220267839294e-05,
+      "loss": 0.6097,
+      "step": 926000
+    },
+    {
+      "epoch": 27.72,
+      "learning_rate": 2.2816310213871677e-05,
+      "loss": 0.6122,
+      "step": 926500
+    },
+    {
+      "epoch": 27.74,
+      "learning_rate": 2.2666400159904056e-05,
+      "loss": 0.5979,
+      "step": 927000
+    },
+    {
+      "epoch": 27.75,
+      "learning_rate": 2.251649010593644e-05,
+      "loss": 0.6052,
+      "step": 927500
+    },
+    {
+      "epoch": 27.77,
+      "learning_rate": 2.2366580051968818e-05,
+      "loss": 0.5997,
+      "step": 928000
+    },
+    {
+      "epoch": 27.78,
+      "learning_rate": 2.2216669998001197e-05,
+      "loss": 0.6175,
+      "step": 928500
+    },
+    {
+      "epoch": 27.8,
+      "learning_rate": 2.206675994403358e-05,
+      "loss": 0.62,
+      "step": 929000
+    },
+    {
+      "epoch": 27.81,
+      "learning_rate": 2.191684989006596e-05,
+      "loss": 0.6212,
+      "step": 929500
+    },
+    {
+      "epoch": 27.83,
+      "learning_rate": 2.1766939836098338e-05,
+      "loss": 0.6213,
+      "step": 930000
+    },
+    {
+      "epoch": 27.84,
+      "learning_rate": 2.161702978213072e-05,
+      "loss": 0.6364,
+      "step": 930500
+    },
+    {
+      "epoch": 27.86,
+      "learning_rate": 2.14671197281631e-05,
+      "loss": 0.6241,
+      "step": 931000
+    },
+    {
+      "epoch": 27.87,
+      "learning_rate": 2.131720967419548e-05,
+      "loss": 0.6028,
+      "step": 931500
+    },
+    {
+      "epoch": 27.89,
+      "learning_rate": 2.116729962022786e-05,
+      "loss": 0.6253,
+      "step": 932000
+    },
+    {
+      "epoch": 27.9,
+      "learning_rate": 2.101738956626024e-05,
+      "loss": 0.6185,
+      "step": 932500
+    },
+    {
+      "epoch": 27.92,
+      "learning_rate": 2.086747951229262e-05,
+      "loss": 0.6058,
+      "step": 933000
+    },
+    {
+      "epoch": 27.93,
+      "learning_rate": 2.0717569458325003e-05,
+      "loss": 0.6167,
+      "step": 933500
+    },
+    {
+      "epoch": 27.95,
+      "learning_rate": 2.0567659404357382e-05,
+      "loss": 0.624,
+      "step": 934000
+    },
+    {
+      "epoch": 27.96,
+      "learning_rate": 2.0417749350389765e-05,
+      "loss": 0.6288,
+      "step": 934500
+    },
+    {
+      "epoch": 27.98,
+      "learning_rate": 2.0267839296422144e-05,
+      "loss": 0.624,
+      "step": 935000
+    },
+    {
+      "epoch": 27.99,
+      "learning_rate": 2.0117929242454523e-05,
+      "loss": 0.6234,
+      "step": 935500
+    },
+    {
+      "epoch": 28.01,
+      "learning_rate": 1.9968019188486906e-05,
+      "loss": 0.6059,
+      "step": 936000
+    },
+    {
+      "epoch": 28.02,
+      "learning_rate": 1.9818109134519285e-05,
+      "loss": 0.6089,
+      "step": 936500
+    },
+    {
+      "epoch": 28.04,
+      "learning_rate": 1.966819908055167e-05,
+      "loss": 0.6003,
+      "step": 937000
+    },
+    {
+      "epoch": 28.05,
+      "learning_rate": 1.951828902658405e-05,
+      "loss": 0.5908,
+      "step": 937500
+    },
+    {
+      "epoch": 28.07,
+      "learning_rate": 1.936837897261643e-05,
+      "loss": 0.6169,
+      "step": 938000
+    },
+    {
+      "epoch": 28.08,
+      "learning_rate": 1.9218468918648812e-05,
+      "loss": 0.5899,
+      "step": 938500
+    },
+    {
+      "epoch": 28.1,
+      "learning_rate": 1.906855886468119e-05,
+      "loss": 0.606,
+      "step": 939000
+    },
+    {
+      "epoch": 28.11,
+      "learning_rate": 1.891864881071357e-05,
+      "loss": 0.6065,
+      "step": 939500
+    },
+    {
+      "epoch": 28.13,
+      "learning_rate": 1.8768738756745953e-05,
+      "loss": 0.6151,
+      "step": 940000
+    },
+    {
+      "epoch": 28.14,
+      "learning_rate": 1.8618828702778332e-05,
+      "loss": 0.6042,
+      "step": 940500
+    },
+    {
+      "epoch": 28.16,
+      "learning_rate": 1.846891864881071e-05,
+      "loss": 0.5988,
+      "step": 941000
+    },
+    {
+      "epoch": 28.17,
+      "learning_rate": 1.8319008594843094e-05,
+      "loss": 0.6099,
+      "step": 941500
+    },
+    {
+      "epoch": 28.19,
+      "learning_rate": 1.8169098540875473e-05,
+      "loss": 0.6023,
+      "step": 942000
+    },
+    {
+      "epoch": 28.2,
+      "learning_rate": 1.8019188486907853e-05,
+      "loss": 0.6139,
+      "step": 942500
+    },
+    {
+      "epoch": 28.22,
+      "learning_rate": 1.7869278432940235e-05,
+      "loss": 0.6062,
+      "step": 943000
+    },
+    {
+      "epoch": 28.23,
+      "learning_rate": 1.7719368378972614e-05,
+      "loss": 0.6136,
+      "step": 943500
+    },
+    {
+      "epoch": 28.25,
+      "learning_rate": 1.7569458325004994e-05,
+      "loss": 0.6063,
+      "step": 944000
+    },
+    {
+      "epoch": 28.26,
+      "learning_rate": 1.7419548271037376e-05,
+      "loss": 0.6142,
+      "step": 944500
+    },
+    {
+      "epoch": 28.28,
+      "learning_rate": 1.7269638217069756e-05,
+      "loss": 0.6106,
+      "step": 945000
+    },
+    {
+      "epoch": 28.29,
+      "learning_rate": 1.7119728163102138e-05,
+      "loss": 0.5985,
+      "step": 945500
+    },
+    {
+      "epoch": 28.31,
+      "learning_rate": 1.6969818109134517e-05,
+      "loss": 0.6193,
+      "step": 946000
+    },
+    {
+      "epoch": 28.32,
+      "learning_rate": 1.6819908055166897e-05,
+      "loss": 0.6159,
+      "step": 946500
+    },
+    {
+      "epoch": 28.34,
+      "learning_rate": 1.666999800119928e-05,
+      "loss": 0.6019,
+      "step": 947000
+    },
+    {
+      "epoch": 28.35,
+      "learning_rate": 1.6520087947231662e-05,
+      "loss": 0.5911,
+      "step": 947500
+    },
+    {
+      "epoch": 28.37,
+      "learning_rate": 1.637017789326404e-05,
+      "loss": 0.6013,
+      "step": 948000
+    },
+    {
+      "epoch": 28.38,
+      "learning_rate": 1.622026783929642e-05,
+      "loss": 0.6036,
+      "step": 948500
+    },
+    {
+      "epoch": 28.4,
+      "learning_rate": 1.6070357785328803e-05,
+      "loss": 0.6076,
+      "step": 949000
+    },
+    {
+      "epoch": 28.41,
+      "learning_rate": 1.5920447731361182e-05,
+      "loss": 0.6173,
+      "step": 949500
+    },
+    {
+      "epoch": 28.43,
+      "learning_rate": 1.577053767739356e-05,
+      "loss": 0.601,
+      "step": 950000
+    },
+    {
+      "epoch": 28.44,
+      "learning_rate": 1.5620627623425944e-05,
+      "loss": 0.578,
+      "step": 950500
+    },
+    {
+      "epoch": 28.46,
+      "learning_rate": 1.5470717569458323e-05,
+      "loss": 0.6151,
+      "step": 951000
+    },
+    {
+      "epoch": 28.47,
+      "learning_rate": 1.5320807515490702e-05,
+      "loss": 0.609,
+      "step": 951500
+    },
+    {
+      "epoch": 28.49,
+      "learning_rate": 1.5170897461523085e-05,
+      "loss": 0.6098,
+      "step": 952000
+    },
+    {
+      "epoch": 28.5,
+      "learning_rate": 1.5020987407555464e-05,
+      "loss": 0.6045,
+      "step": 952500
+    },
+    {
+      "epoch": 28.52,
+      "learning_rate": 1.4871077353587845e-05,
+      "loss": 0.6059,
+      "step": 953000
+    },
+    {
+      "epoch": 28.53,
+      "learning_rate": 1.4721167299620226e-05,
+      "loss": 0.6018,
+      "step": 953500
+    },
+    {
+      "epoch": 28.55,
+      "learning_rate": 1.4571257245652609e-05,
+      "loss": 0.611,
+      "step": 954000
+    },
+    {
+      "epoch": 28.56,
+      "learning_rate": 1.4421347191684988e-05,
+      "loss": 0.5968,
+      "step": 954500
+    },
+    {
+      "epoch": 28.58,
+      "learning_rate": 1.4271437137717369e-05,
+      "loss": 0.6085,
+      "step": 955000
+    },
+    {
+      "epoch": 28.59,
+      "learning_rate": 1.412152708374975e-05,
+      "loss": 0.5988,
+      "step": 955500
+    },
+    {
+      "epoch": 28.61,
+      "learning_rate": 1.3971617029782129e-05,
+      "loss": 0.5933,
+      "step": 956000
+    },
+    {
+      "epoch": 28.62,
+      "learning_rate": 1.382170697581451e-05,
+      "loss": 0.5988,
+      "step": 956500
+    },
+    {
+      "epoch": 28.64,
+      "learning_rate": 1.3671796921846891e-05,
+      "loss": 0.6041,
+      "step": 957000
+    },
+    {
+      "epoch": 28.65,
+      "learning_rate": 1.3521886867879272e-05,
+      "loss": 0.6055,
+      "step": 957500
+    },
+    {
+      "epoch": 28.67,
+      "learning_rate": 1.3371976813911651e-05,
+      "loss": 0.6016,
+      "step": 958000
+    },
+    {
+      "epoch": 28.68,
+      "learning_rate": 1.3222066759944032e-05,
+      "loss": 0.5995,
+      "step": 958500
+    },
+    {
+      "epoch": 28.69,
+      "learning_rate": 1.3072156705976413e-05,
+      "loss": 0.5996,
+      "step": 959000
+    },
+    {
+      "epoch": 28.71,
+      "learning_rate": 1.2922246652008792e-05,
+      "loss": 0.6126,
+      "step": 959500
+    },
+    {
+      "epoch": 28.72,
+      "learning_rate": 1.2772336598041173e-05,
+      "loss": 0.5835,
+      "step": 960000
+    },
+    {
+      "epoch": 28.74,
+      "learning_rate": 1.2622426544073556e-05,
+      "loss": 0.5804,
+      "step": 960500
+    },
+    {
+      "epoch": 28.75,
+      "learning_rate": 1.2472516490105937e-05,
+      "loss": 0.6171,
+      "step": 961000
+    },
+    {
+      "epoch": 28.77,
+      "learning_rate": 1.2322606436138316e-05,
+      "loss": 0.5997,
+      "step": 961500
+    },
+    {
+      "epoch": 28.78,
+      "learning_rate": 1.2172696382170697e-05,
+      "loss": 0.6153,
+      "step": 962000
+    },
+    {
+      "epoch": 28.8,
+      "learning_rate": 1.2022786328203078e-05,
+      "loss": 0.5941,
+      "step": 962500
+    },
+    {
+      "epoch": 28.81,
+      "learning_rate": 1.1872876274235459e-05,
+      "loss": 0.6022,
+      "step": 963000
+    },
+    {
+      "epoch": 28.83,
+      "learning_rate": 1.1722966220267838e-05,
+      "loss": 0.6248,
+      "step": 963500
+    },
+    {
+      "epoch": 28.84,
+      "learning_rate": 1.1573056166300219e-05,
+      "loss": 0.6033,
+      "step": 964000
+    },
+    {
+      "epoch": 28.86,
+      "learning_rate": 1.14231461123326e-05,
+      "loss": 0.6059,
+      "step": 964500
+    },
+    {
+      "epoch": 28.87,
+      "learning_rate": 1.1273236058364979e-05,
+      "loss": 0.6137,
+      "step": 965000
+    },
+    {
+      "epoch": 28.89,
+      "learning_rate": 1.112332600439736e-05,
+      "loss": 0.6019,
+      "step": 965500
+    },
+    {
+      "epoch": 28.9,
+      "learning_rate": 1.097341595042974e-05,
+      "loss": 0.6003,
+      "step": 966000
+    },
+    {
+      "epoch": 28.92,
+      "learning_rate": 1.0823505896462122e-05,
+      "loss": 0.619,
+      "step": 966500
+    },
+    {
+      "epoch": 28.93,
+      "learning_rate": 1.0673595842494503e-05,
+      "loss": 0.6056,
+      "step": 967000
+    },
+    {
+      "epoch": 28.95,
+      "learning_rate": 1.0523685788526884e-05,
+      "loss": 0.5954,
+      "step": 967500
+    },
+    {
+      "epoch": 28.96,
+      "learning_rate": 1.0373775734559264e-05,
+      "loss": 0.6115,
+      "step": 968000
+    },
+    {
+      "epoch": 28.98,
+      "learning_rate": 1.0223865680591645e-05,
+      "loss": 0.6067,
+      "step": 968500
+    },
+    {
+      "epoch": 28.99,
+      "learning_rate": 1.0073955626624025e-05,
+      "loss": 0.5987,
+      "step": 969000
+    }
+  ],
+  "logging_steps": 500,
+  "max_steps": 1002600,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 30,
+  "save_steps": 500,
+  "total_flos": 2.445394000189791e+21,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}