diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,8 +1,8 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 91.0,
-  "global_step": 18367713,
+  "epoch": 100.0,
+  "global_step": 20184300,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -221417,11 +221417,21908 @@
       "eval_samples_per_second": 1737.64,
       "eval_steps_per_second": 72.407,
       "step": 18367713
+    },
+    {
+      "epoch": 91.0,
+      "learning_rate": 4.5155665542030194e-06,
+      "loss": 1.8238,
+      "step": 18368000
+    },
+    {
+      "epoch": 91.0,
+      "learning_rate": 4.5143279677769355e-06,
+      "loss": 1.7877,
+      "step": 18368500
+    },
+    {
+      "epoch": 91.01,
+      "learning_rate": 4.513091858523704e-06,
+      "loss": 1.8172,
+      "step": 18369000
+    },
+    {
+      "epoch": 91.01,
+      "learning_rate": 4.51185327209762e-06,
+      "loss": 1.7793,
+      "step": 18369500
+    },
+    {
+      "epoch": 91.01,
+      "learning_rate": 4.5106146856715365e-06,
+      "loss": 1.7981,
+      "step": 18370000
+    },
+    {
+      "epoch": 91.01,
+      "learning_rate": 4.509376099245453e-06,
+      "loss": 1.7923,
+      "step": 18370500
+    },
+    {
+      "epoch": 91.02,
+      "learning_rate": 4.5081375128193695e-06,
+      "loss": 1.8031,
+      "step": 18371000
+    },
+    {
+      "epoch": 91.02,
+      "learning_rate": 4.506898926393286e-06,
+      "loss": 1.795,
+      "step": 18371500
+    },
+    {
+      "epoch": 91.02,
+      "learning_rate": 4.5056603399672025e-06,
+      "loss": 1.8148,
+      "step": 18372000
+    },
+    {
+      "epoch": 91.02,
+      "learning_rate": 4.504421753541119e-06,
+      "loss": 1.7991,
+      "step": 18372500
+    },
+    {
+      "epoch": 91.03,
+      "learning_rate": 4.5031831671150355e-06,
+      "loss": 1.8063,
+      "step": 18373000
+    },
+    {
+      "epoch": 91.03,
+      "learning_rate": 4.501944580688952e-06,
+      "loss": 1.8167,
+      "step": 18373500
+    },
+    {
+      "epoch": 91.03,
+      "learning_rate": 4.50070847143572e-06,
+      "loss": 1.7992,
+      "step": 18374000
+    },
+    {
+      "epoch": 91.03,
+      "learning_rate": 4.4994698850096365e-06,
+      "loss": 1.8178,
+      "step": 18374500
+    },
+    {
+      "epoch": 91.04,
+      "learning_rate": 4.498233775756405e-06,
+      "loss": 1.7876,
+      "step": 18375000
+    },
+    {
+      "epoch": 91.04,
+      "learning_rate": 4.496997666503173e-06,
+      "loss": 1.8133,
+      "step": 18375500
+    },
+    {
+      "epoch": 91.04,
+      "learning_rate": 4.49575908007709e-06,
+      "loss": 1.7867,
+      "step": 18376000
+    },
+    {
+      "epoch": 91.04,
+      "learning_rate": 4.494520493651006e-06,
+      "loss": 1.7892,
+      "step": 18376500
+    },
+    {
+      "epoch": 91.05,
+      "learning_rate": 4.493284384397774e-06,
+      "loss": 1.7739,
+      "step": 18377000
+    },
+    {
+      "epoch": 91.05,
+      "learning_rate": 4.492045797971691e-06,
+      "loss": 1.8005,
+      "step": 18377500
+    },
+    {
+      "epoch": 91.05,
+      "learning_rate": 4.490807211545608e-06,
+      "loss": 1.814,
+      "step": 18378000
+    },
+    {
+      "epoch": 91.05,
+      "learning_rate": 4.489568625119524e-06,
+      "loss": 1.8055,
+      "step": 18378500
+    },
+    {
+      "epoch": 91.06,
+      "learning_rate": 4.48833003869344e-06,
+      "loss": 1.8141,
+      "step": 18379000
+    },
+    {
+      "epoch": 91.06,
+      "learning_rate": 4.487091452267356e-06,
+      "loss": 1.7972,
+      "step": 18379500
+    },
+    {
+      "epoch": 91.06,
+      "learning_rate": 4.485852865841273e-06,
+      "loss": 1.8121,
+      "step": 18380000
+    },
+    {
+      "epoch": 91.06,
+      "learning_rate": 4.484614279415189e-06,
+      "loss": 1.8006,
+      "step": 18380500
+    },
+    {
+      "epoch": 91.07,
+      "learning_rate": 4.483375692989105e-06,
+      "loss": 1.8122,
+      "step": 18381000
+    },
+    {
+      "epoch": 91.07,
+      "learning_rate": 4.4821371065630215e-06,
+      "loss": 1.813,
+      "step": 18381500
+    },
+    {
+      "epoch": 91.07,
+      "learning_rate": 4.48090099730979e-06,
+      "loss": 1.8126,
+      "step": 18382000
+    },
+    {
+      "epoch": 91.07,
+      "learning_rate": 4.479662410883706e-06,
+      "loss": 1.7956,
+      "step": 18382500
+    },
+    {
+      "epoch": 91.08,
+      "learning_rate": 4.478423824457623e-06,
+      "loss": 1.802,
+      "step": 18383000
+    },
+    {
+      "epoch": 91.08,
+      "learning_rate": 4.477185238031539e-06,
+      "loss": 1.7802,
+      "step": 18383500
+    },
+    {
+      "epoch": 91.08,
+      "learning_rate": 4.475946651605456e-06,
+      "loss": 1.8093,
+      "step": 18384000
+    },
+    {
+      "epoch": 91.08,
+      "learning_rate": 4.474710542352224e-06,
+      "loss": 1.8001,
+      "step": 18384500
+    },
+    {
+      "epoch": 91.09,
+      "learning_rate": 4.473471955926141e-06,
+      "loss": 1.8319,
+      "step": 18385000
+    },
+    {
+      "epoch": 91.09,
+      "learning_rate": 4.472233369500057e-06,
+      "loss": 1.8107,
+      "step": 18385500
+    },
+    {
+      "epoch": 91.09,
+      "learning_rate": 4.470994783073973e-06,
+      "loss": 1.8162,
+      "step": 18386000
+    },
+    {
+      "epoch": 91.09,
+      "learning_rate": 4.4697561966478895e-06,
+      "loss": 1.8314,
+      "step": 18386500
+    },
+    {
+      "epoch": 91.1,
+      "learning_rate": 4.468517610221806e-06,
+      "loss": 1.8067,
+      "step": 18387000
+    },
+    {
+      "epoch": 91.1,
+      "learning_rate": 4.467279023795723e-06,
+      "loss": 1.8121,
+      "step": 18387500
+    },
+    {
+      "epoch": 91.1,
+      "learning_rate": 4.4660404373696385e-06,
+      "loss": 1.7994,
+      "step": 18388000
+    },
+    {
+      "epoch": 91.1,
+      "learning_rate": 4.4648018509435555e-06,
+      "loss": 1.7916,
+      "step": 18388500
+    },
+    {
+      "epoch": 91.11,
+      "learning_rate": 4.4635632645174716e-06,
+      "loss": 1.8057,
+      "step": 18389000
+    },
+    {
+      "epoch": 91.11,
+      "learning_rate": 4.4623246780913885e-06,
+      "loss": 1.7974,
+      "step": 18389500
+    },
+    {
+      "epoch": 91.11,
+      "learning_rate": 4.4610885688381564e-06,
+      "loss": 1.7834,
+      "step": 18390000
+    },
+    {
+      "epoch": 91.11,
+      "learning_rate": 4.459849982412073e-06,
+      "loss": 1.8185,
+      "step": 18390500
+    },
+    {
+      "epoch": 91.12,
+      "learning_rate": 4.4586113959859895e-06,
+      "loss": 1.8006,
+      "step": 18391000
+    },
+    {
+      "epoch": 91.12,
+      "learning_rate": 4.4573728095599055e-06,
+      "loss": 1.7961,
+      "step": 18391500
+    },
+    {
+      "epoch": 91.12,
+      "learning_rate": 4.456134223133822e-06,
+      "loss": 1.7955,
+      "step": 18392000
+    },
+    {
+      "epoch": 91.12,
+      "learning_rate": 4.4548956367077385e-06,
+      "loss": 1.8343,
+      "step": 18392500
+    },
+    {
+      "epoch": 91.13,
+      "learning_rate": 4.453657050281655e-06,
+      "loss": 1.8246,
+      "step": 18393000
+    },
+    {
+      "epoch": 91.13,
+      "learning_rate": 4.4524209410284234e-06,
+      "loss": 1.7989,
+      "step": 18393500
+    },
+    {
+      "epoch": 91.13,
+      "learning_rate": 4.451184831775191e-06,
+      "loss": 1.8199,
+      "step": 18394000
+    },
+    {
+      "epoch": 91.13,
+      "learning_rate": 4.449946245349108e-06,
+      "loss": 1.8138,
+      "step": 18394500
+    },
+    {
+      "epoch": 91.14,
+      "learning_rate": 4.448707658923024e-06,
+      "loss": 1.822,
+      "step": 18395000
+    },
+    {
+      "epoch": 91.14,
+      "learning_rate": 4.447469072496941e-06,
+      "loss": 1.7962,
+      "step": 18395500
+    },
+    {
+      "epoch": 91.14,
+      "learning_rate": 4.446230486070857e-06,
+      "loss": 1.7837,
+      "step": 18396000
+    },
+    {
+      "epoch": 91.14,
+      "learning_rate": 4.4449918996447735e-06,
+      "loss": 1.7973,
+      "step": 18396500
+    },
+    {
+      "epoch": 91.15,
+      "learning_rate": 4.44375331321869e-06,
+      "loss": 1.8241,
+      "step": 18397000
+    },
+    {
+      "epoch": 91.15,
+      "learning_rate": 4.442517203965458e-06,
+      "loss": 1.8237,
+      "step": 18397500
+    },
+    {
+      "epoch": 91.15,
+      "learning_rate": 4.4412786175393745e-06,
+      "loss": 1.8102,
+      "step": 18398000
+    },
+    {
+      "epoch": 91.15,
+      "learning_rate": 4.440040031113291e-06,
+      "loss": 1.8286,
+      "step": 18398500
+    },
+    {
+      "epoch": 91.16,
+      "learning_rate": 4.4388014446872075e-06,
+      "loss": 1.7764,
+      "step": 18399000
+    },
+    {
+      "epoch": 91.16,
+      "learning_rate": 4.4375628582611236e-06,
+      "loss": 1.8141,
+      "step": 18399500
+    },
+    {
+      "epoch": 91.16,
+      "learning_rate": 4.4363242718350405e-06,
+      "loss": 1.7887,
+      "step": 18400000
+    },
+    {
+      "epoch": 91.16,
+      "learning_rate": 4.4350856854089566e-06,
+      "loss": 1.823,
+      "step": 18400500
+    },
+    {
+      "epoch": 91.16,
+      "learning_rate": 4.4338470989828735e-06,
+      "loss": 1.7915,
+      "step": 18401000
+    },
+    {
+      "epoch": 91.17,
+      "learning_rate": 4.43260851255679e-06,
+      "loss": 1.8139,
+      "step": 18401500
+    },
+    {
+      "epoch": 91.17,
+      "learning_rate": 4.431369926130706e-06,
+      "loss": 1.8145,
+      "step": 18402000
+    },
+    {
+      "epoch": 91.17,
+      "learning_rate": 4.430131339704622e-06,
+      "loss": 1.7938,
+      "step": 18402500
+    },
+    {
+      "epoch": 91.17,
+      "learning_rate": 4.428892753278539e-06,
+      "loss": 1.8088,
+      "step": 18403000
+    },
+    {
+      "epoch": 91.18,
+      "learning_rate": 4.427654166852455e-06,
+      "loss": 1.8114,
+      "step": 18403500
+    },
+    {
+      "epoch": 91.18,
+      "learning_rate": 4.426415580426372e-06,
+      "loss": 1.8043,
+      "step": 18404000
+    },
+    {
+      "epoch": 91.18,
+      "learning_rate": 4.425176994000287e-06,
+      "loss": 1.7982,
+      "step": 18404500
+    },
+    {
+      "epoch": 91.18,
+      "learning_rate": 4.423938407574204e-06,
+      "loss": 1.7848,
+      "step": 18405000
+    },
+    {
+      "epoch": 91.19,
+      "learning_rate": 4.422699821148121e-06,
+      "loss": 1.8017,
+      "step": 18405500
+    },
+    {
+      "epoch": 91.19,
+      "learning_rate": 4.421463711894889e-06,
+      "loss": 1.8133,
+      "step": 18406000
+    },
+    {
+      "epoch": 91.19,
+      "learning_rate": 4.420230079814509e-06,
+      "loss": 1.8233,
+      "step": 18406500
+    },
+    {
+      "epoch": 91.19,
+      "learning_rate": 4.418991493388426e-06,
+      "loss": 1.7882,
+      "step": 18407000
+    },
+    {
+      "epoch": 91.2,
+      "learning_rate": 4.4177529069623424e-06,
+      "loss": 1.7908,
+      "step": 18407500
+    },
+    {
+      "epoch": 91.2,
+      "learning_rate": 4.416516797709111e-06,
+      "loss": 1.8051,
+      "step": 18408000
+    },
+    {
+      "epoch": 91.2,
+      "learning_rate": 4.415278211283027e-06,
+      "loss": 1.8144,
+      "step": 18408500
+    },
+    {
+      "epoch": 91.2,
+      "learning_rate": 4.414039624856943e-06,
+      "loss": 1.8121,
+      "step": 18409000
+    },
+    {
+      "epoch": 91.21,
+      "learning_rate": 4.4128010384308595e-06,
+      "loss": 1.7947,
+      "step": 18409500
+    },
+    {
+      "epoch": 91.21,
+      "learning_rate": 4.411562452004776e-06,
+      "loss": 1.7938,
+      "step": 18410000
+    },
+    {
+      "epoch": 91.21,
+      "learning_rate": 4.4103238655786925e-06,
+      "loss": 1.8087,
+      "step": 18410500
+    },
+    {
+      "epoch": 91.21,
+      "learning_rate": 4.4090852791526086e-06,
+      "loss": 1.8004,
+      "step": 18411000
+    },
+    {
+      "epoch": 91.22,
+      "learning_rate": 4.407849169899377e-06,
+      "loss": 1.8073,
+      "step": 18411500
+    },
+    {
+      "epoch": 91.22,
+      "learning_rate": 4.406610583473294e-06,
+      "loss": 1.7987,
+      "step": 18412000
+    },
+    {
+      "epoch": 91.22,
+      "learning_rate": 4.4053719970472095e-06,
+      "loss": 1.7817,
+      "step": 18412500
+    },
+    {
+      "epoch": 91.22,
+      "learning_rate": 4.4041334106211265e-06,
+      "loss": 1.7861,
+      "step": 18413000
+    },
+    {
+      "epoch": 91.23,
+      "learning_rate": 4.402894824195043e-06,
+      "loss": 1.8135,
+      "step": 18413500
+    },
+    {
+      "epoch": 91.23,
+      "learning_rate": 4.401658714941811e-06,
+      "loss": 1.7576,
+      "step": 18414000
+    },
+    {
+      "epoch": 91.23,
+      "learning_rate": 4.4004201285157274e-06,
+      "loss": 1.8048,
+      "step": 18414500
+    },
+    {
+      "epoch": 91.23,
+      "learning_rate": 4.399181542089644e-06,
+      "loss": 1.7548,
+      "step": 18415000
+    },
+    {
+      "epoch": 91.24,
+      "learning_rate": 4.3979429556635605e-06,
+      "loss": 1.8192,
+      "step": 18415500
+    },
+    {
+      "epoch": 91.24,
+      "learning_rate": 4.396706846410329e-06,
+      "loss": 1.8046,
+      "step": 18416000
+    },
+    {
+      "epoch": 91.24,
+      "learning_rate": 4.395468259984245e-06,
+      "loss": 1.8224,
+      "step": 18416500
+    },
+    {
+      "epoch": 91.24,
+      "learning_rate": 4.3942296735581614e-06,
+      "loss": 1.7869,
+      "step": 18417000
+    },
+    {
+      "epoch": 91.25,
+      "learning_rate": 4.392991087132078e-06,
+      "loss": 1.8152,
+      "step": 18417500
+    },
+    {
+      "epoch": 91.25,
+      "learning_rate": 4.3917525007059944e-06,
+      "loss": 1.8083,
+      "step": 18418000
+    },
+    {
+      "epoch": 91.25,
+      "learning_rate": 4.390513914279911e-06,
+      "loss": 1.8127,
+      "step": 18418500
+    },
+    {
+      "epoch": 91.25,
+      "learning_rate": 4.3892753278538274e-06,
+      "loss": 1.8243,
+      "step": 18419000
+    },
+    {
+      "epoch": 91.26,
+      "learning_rate": 4.3880367414277435e-06,
+      "loss": 1.7837,
+      "step": 18419500
+    },
+    {
+      "epoch": 91.26,
+      "learning_rate": 4.38679815500166e-06,
+      "loss": 1.8011,
+      "step": 18420000
+    },
+    {
+      "epoch": 91.26,
+      "learning_rate": 4.3855595685755765e-06,
+      "loss": 1.7814,
+      "step": 18420500
+    },
+    {
+      "epoch": 91.26,
+      "learning_rate": 4.384320982149493e-06,
+      "loss": 1.8039,
+      "step": 18421000
+    },
+    {
+      "epoch": 91.27,
+      "learning_rate": 4.383082395723409e-06,
+      "loss": 1.7995,
+      "step": 18421500
+    },
+    {
+      "epoch": 91.27,
+      "learning_rate": 4.3818462864701775e-06,
+      "loss": 1.7886,
+      "step": 18422000
+    },
+    {
+      "epoch": 91.27,
+      "learning_rate": 4.380612654389798e-06,
+      "loss": 1.8072,
+      "step": 18422500
+    },
+    {
+      "epoch": 91.27,
+      "learning_rate": 4.379374067963714e-06,
+      "loss": 1.8142,
+      "step": 18423000
+    },
+    {
+      "epoch": 91.28,
+      "learning_rate": 4.378135481537631e-06,
+      "loss": 1.8281,
+      "step": 18423500
+    },
+    {
+      "epoch": 91.28,
+      "learning_rate": 4.376896895111547e-06,
+      "loss": 1.8153,
+      "step": 18424000
+    },
+    {
+      "epoch": 91.28,
+      "learning_rate": 4.375658308685464e-06,
+      "loss": 1.7936,
+      "step": 18424500
+    },
+    {
+      "epoch": 91.28,
+      "learning_rate": 4.3744197222593795e-06,
+      "loss": 1.7945,
+      "step": 18425000
+    },
+    {
+      "epoch": 91.29,
+      "learning_rate": 4.373181135833296e-06,
+      "loss": 1.7936,
+      "step": 18425500
+    },
+    {
+      "epoch": 91.29,
+      "learning_rate": 4.3719425494072125e-06,
+      "loss": 1.8093,
+      "step": 18426000
+    },
+    {
+      "epoch": 91.29,
+      "learning_rate": 4.370703962981129e-06,
+      "loss": 1.8145,
+      "step": 18426500
+    },
+    {
+      "epoch": 91.29,
+      "learning_rate": 4.3694653765550455e-06,
+      "loss": 1.7938,
+      "step": 18427000
+    },
+    {
+      "epoch": 91.3,
+      "learning_rate": 4.3682267901289615e-06,
+      "loss": 1.7991,
+      "step": 18427500
+    },
+    {
+      "epoch": 91.3,
+      "learning_rate": 4.3669882037028785e-06,
+      "loss": 1.8111,
+      "step": 18428000
+    },
+    {
+      "epoch": 91.3,
+      "learning_rate": 4.3657496172767946e-06,
+      "loss": 1.7835,
+      "step": 18428500
+    },
+    {
+      "epoch": 91.3,
+      "learning_rate": 4.3645135080235625e-06,
+      "loss": 1.8006,
+      "step": 18429000
+    },
+    {
+      "epoch": 91.31,
+      "learning_rate": 4.3632749215974794e-06,
+      "loss": 1.8044,
+      "step": 18429500
+    },
+    {
+      "epoch": 91.31,
+      "learning_rate": 4.362036335171396e-06,
+      "loss": 1.8151,
+      "step": 18430000
+    },
+    {
+      "epoch": 91.31,
+      "learning_rate": 4.3607977487453125e-06,
+      "loss": 1.7854,
+      "step": 18430500
+    },
+    {
+      "epoch": 91.31,
+      "learning_rate": 4.3595591623192285e-06,
+      "loss": 1.8164,
+      "step": 18431000
+    },
+    {
+      "epoch": 91.32,
+      "learning_rate": 4.358320575893145e-06,
+      "loss": 1.8253,
+      "step": 18431500
+    },
+    {
+      "epoch": 91.32,
+      "learning_rate": 4.3570819894670615e-06,
+      "loss": 1.8103,
+      "step": 18432000
+    },
+    {
+      "epoch": 91.32,
+      "learning_rate": 4.355843403040978e-06,
+      "loss": 1.7965,
+      "step": 18432500
+    },
+    {
+      "epoch": 91.32,
+      "learning_rate": 4.3546072937877464e-06,
+      "loss": 1.8188,
+      "step": 18433000
+    },
+    {
+      "epoch": 91.33,
+      "learning_rate": 4.3533687073616625e-06,
+      "loss": 1.807,
+      "step": 18433500
+    },
+    {
+      "epoch": 91.33,
+      "learning_rate": 4.352130120935579e-06,
+      "loss": 1.8107,
+      "step": 18434000
+    },
+    {
+      "epoch": 91.33,
+      "learning_rate": 4.350891534509495e-06,
+      "loss": 1.8244,
+      "step": 18434500
+    },
+    {
+      "epoch": 91.33,
+      "learning_rate": 4.349652948083412e-06,
+      "loss": 1.782,
+      "step": 18435000
+    },
+    {
+      "epoch": 91.34,
+      "learning_rate": 4.348414361657328e-06,
+      "loss": 1.8092,
+      "step": 18435500
+    },
+    {
+      "epoch": 91.34,
+      "learning_rate": 4.347175775231245e-06,
+      "loss": 1.8002,
+      "step": 18436000
+    },
+    {
+      "epoch": 91.34,
+      "learning_rate": 4.345937188805161e-06,
+      "loss": 1.8015,
+      "step": 18436500
+    },
+    {
+      "epoch": 91.34,
+      "learning_rate": 4.344703556724781e-06,
+      "loss": 1.7918,
+      "step": 18437000
+    },
+    {
+      "epoch": 91.35,
+      "learning_rate": 4.343467447471549e-06,
+      "loss": 1.8164,
+      "step": 18437500
+    },
+    {
+      "epoch": 91.35,
+      "learning_rate": 4.342228861045466e-06,
+      "loss": 1.8096,
+      "step": 18438000
+    },
+    {
+      "epoch": 91.35,
+      "learning_rate": 4.340990274619382e-06,
+      "loss": 1.8126,
+      "step": 18438500
+    },
+    {
+      "epoch": 91.35,
+      "learning_rate": 4.339751688193299e-06,
+      "loss": 1.8234,
+      "step": 18439000
+    },
+    {
+      "epoch": 91.36,
+      "learning_rate": 4.338513101767215e-06,
+      "loss": 1.7843,
+      "step": 18439500
+    },
+    {
+      "epoch": 91.36,
+      "learning_rate": 4.3372745153411315e-06,
+      "loss": 1.7892,
+      "step": 18440000
+    },
+    {
+      "epoch": 91.36,
+      "learning_rate": 4.3360359289150475e-06,
+      "loss": 1.7983,
+      "step": 18440500
+    },
+    {
+      "epoch": 91.36,
+      "learning_rate": 4.3347973424889645e-06,
+      "loss": 1.7977,
+      "step": 18441000
+    },
+    {
+      "epoch": 91.37,
+      "learning_rate": 4.3335587560628805e-06,
+      "loss": 1.7831,
+      "step": 18441500
+    },
+    {
+      "epoch": 91.37,
+      "learning_rate": 4.332322646809649e-06,
+      "loss": 1.7906,
+      "step": 18442000
+    },
+    {
+      "epoch": 91.37,
+      "learning_rate": 4.3310840603835654e-06,
+      "loss": 1.8172,
+      "step": 18442500
+    },
+    {
+      "epoch": 91.37,
+      "learning_rate": 4.329847951130334e-06,
+      "loss": 1.8025,
+      "step": 18443000
+    },
+    {
+      "epoch": 91.38,
+      "learning_rate": 4.32860936470425e-06,
+      "loss": 1.7938,
+      "step": 18443500
+    },
+    {
+      "epoch": 91.38,
+      "learning_rate": 4.327370778278167e-06,
+      "loss": 1.7916,
+      "step": 18444000
+    },
+    {
+      "epoch": 91.38,
+      "learning_rate": 4.326132191852083e-06,
+      "loss": 1.7927,
+      "step": 18444500
+    },
+    {
+      "epoch": 91.38,
+      "learning_rate": 4.324893605425999e-06,
+      "loss": 1.7992,
+      "step": 18445000
+    },
+    {
+      "epoch": 91.39,
+      "learning_rate": 4.323655018999916e-06,
+      "loss": 1.8043,
+      "step": 18445500
+    },
+    {
+      "epoch": 91.39,
+      "learning_rate": 4.322418909746684e-06,
+      "loss": 1.8067,
+      "step": 18446000
+    },
+    {
+      "epoch": 91.39,
+      "learning_rate": 4.321180323320601e-06,
+      "loss": 1.8118,
+      "step": 18446500
+    },
+    {
+      "epoch": 91.39,
+      "learning_rate": 4.319941736894517e-06,
+      "loss": 1.8203,
+      "step": 18447000
+    },
+    {
+      "epoch": 91.4,
+      "learning_rate": 4.318703150468434e-06,
+      "loss": 1.8081,
+      "step": 18447500
+    },
+    {
+      "epoch": 91.4,
+      "learning_rate": 4.3174645640423495e-06,
+      "loss": 1.7989,
+      "step": 18448000
+    },
+    {
+      "epoch": 91.4,
+      "learning_rate": 4.316225977616266e-06,
+      "loss": 1.7943,
+      "step": 18448500
+    },
+    {
+      "epoch": 91.4,
+      "learning_rate": 4.3149873911901825e-06,
+      "loss": 1.7726,
+      "step": 18449000
+    },
+    {
+      "epoch": 91.41,
+      "learning_rate": 4.313748804764099e-06,
+      "loss": 1.7879,
+      "step": 18449500
+    },
+    {
+      "epoch": 91.41,
+      "learning_rate": 4.3125102183380155e-06,
+      "loss": 1.8255,
+      "step": 18450000
+    },
+    {
+      "epoch": 91.41,
+      "learning_rate": 4.311271631911932e-06,
+      "loss": 1.8035,
+      "step": 18450500
+    },
+    {
+      "epoch": 91.41,
+      "learning_rate": 4.310033045485848e-06,
+      "loss": 1.7864,
+      "step": 18451000
+    },
+    {
+      "epoch": 91.42,
+      "learning_rate": 4.3087969362326165e-06,
+      "loss": 1.8104,
+      "step": 18451500
+    },
+    {
+      "epoch": 91.42,
+      "learning_rate": 4.3075583498065325e-06,
+      "loss": 1.8231,
+      "step": 18452000
+    },
+    {
+      "epoch": 91.42,
+      "learning_rate": 4.3063197633804495e-06,
+      "loss": 1.8403,
+      "step": 18452500
+    },
+    {
+      "epoch": 91.42,
+      "learning_rate": 4.3050811769543656e-06,
+      "loss": 1.8119,
+      "step": 18453000
+    },
+    {
+      "epoch": 91.43,
+      "learning_rate": 4.3038425905282825e-06,
+      "loss": 1.8011,
+      "step": 18453500
+    },
+    {
+      "epoch": 91.43,
+      "learning_rate": 4.302608958447902e-06,
+      "loss": 1.8145,
+      "step": 18454000
+    },
+    {
+      "epoch": 91.43,
+      "learning_rate": 4.301370372021819e-06,
+      "loss": 1.8077,
+      "step": 18454500
+    },
+    {
+      "epoch": 91.43,
+      "learning_rate": 4.300131785595735e-06,
+      "loss": 1.7986,
+      "step": 18455000
+    },
+    {
+      "epoch": 91.43,
+      "learning_rate": 4.298893199169652e-06,
+      "loss": 1.7884,
+      "step": 18455500
+    },
+    {
+      "epoch": 91.44,
+      "learning_rate": 4.297654612743568e-06,
+      "loss": 1.8042,
+      "step": 18456000
+    },
+    {
+      "epoch": 91.44,
+      "learning_rate": 4.2964160263174844e-06,
+      "loss": 1.7959,
+      "step": 18456500
+    },
+    {
+      "epoch": 91.44,
+      "learning_rate": 4.295179917064253e-06,
+      "loss": 1.7996,
+      "step": 18457000
+    },
+    {
+      "epoch": 91.44,
+      "learning_rate": 4.293941330638169e-06,
+      "loss": 1.8079,
+      "step": 18457500
+    },
+    {
+      "epoch": 91.45,
+      "learning_rate": 4.292702744212085e-06,
+      "loss": 1.8014,
+      "step": 18458000
+    },
+    {
+      "epoch": 91.45,
+      "learning_rate": 4.291464157786002e-06,
+      "loss": 1.8148,
+      "step": 18458500
+    },
+    {
+      "epoch": 91.45,
+      "learning_rate": 4.290225571359919e-06,
+      "loss": 1.7937,
+      "step": 18459000
+    },
+    {
+      "epoch": 91.45,
+      "learning_rate": 4.2889869849338345e-06,
+      "loss": 1.8136,
+      "step": 18459500
+    },
+    {
+      "epoch": 91.46,
+      "learning_rate": 4.287750875680603e-06,
+      "loss": 1.7974,
+      "step": 18460000
+    },
+    {
+      "epoch": 91.46,
+      "learning_rate": 4.28651228925452e-06,
+      "loss": 1.7971,
+      "step": 18460500
+    },
+    {
+      "epoch": 91.46,
+      "learning_rate": 4.285273702828436e-06,
+      "loss": 1.7939,
+      "step": 18461000
+    },
+    {
+      "epoch": 91.46,
+      "learning_rate": 4.284035116402352e-06,
+      "loss": 1.8095,
+      "step": 18461500
+    },
+    {
+      "epoch": 91.47,
+      "learning_rate": 4.282796529976269e-06,
+      "loss": 1.788,
+      "step": 18462000
+    },
+    {
+      "epoch": 91.47,
+      "learning_rate": 4.281560420723037e-06,
+      "loss": 1.8043,
+      "step": 18462500
+    },
+    {
+      "epoch": 91.47,
+      "learning_rate": 4.280321834296954e-06,
+      "loss": 1.8013,
+      "step": 18463000
+    },
+    {
+      "epoch": 91.47,
+      "learning_rate": 4.27908324787087e-06,
+      "loss": 1.8226,
+      "step": 18463500
+    },
+    {
+      "epoch": 91.48,
+      "learning_rate": 4.277844661444786e-06,
+      "loss": 1.791,
+      "step": 18464000
+    },
+    {
+      "epoch": 91.48,
+      "learning_rate": 4.2766060750187025e-06,
+      "loss": 1.7857,
+      "step": 18464500
+    },
+    {
+      "epoch": 91.48,
+      "learning_rate": 4.275367488592619e-06,
+      "loss": 1.809,
+      "step": 18465000
+    },
+    {
+      "epoch": 91.48,
+      "learning_rate": 4.2741289021665355e-06,
+      "loss": 1.8259,
+      "step": 18465500
+    },
+    {
+      "epoch": 91.49,
+      "learning_rate": 4.272890315740452e-06,
+      "loss": 1.8109,
+      "step": 18466000
+    },
+    {
+      "epoch": 91.49,
+      "learning_rate": 4.2716517293143685e-06,
+      "loss": 1.7935,
+      "step": 18466500
+    },
+    {
+      "epoch": 91.49,
+      "learning_rate": 4.2704131428882846e-06,
+      "loss": 1.7839,
+      "step": 18467000
+    },
+    {
+      "epoch": 91.49,
+      "learning_rate": 4.269174556462201e-06,
+      "loss": 1.8161,
+      "step": 18467500
+    },
+    {
+      "epoch": 91.5,
+      "learning_rate": 4.2679384472089694e-06,
+      "loss": 1.7774,
+      "step": 18468000
+    },
+    {
+      "epoch": 91.5,
+      "learning_rate": 4.2666998607828855e-06,
+      "loss": 1.799,
+      "step": 18468500
+    },
+    {
+      "epoch": 91.5,
+      "learning_rate": 4.2654612743568025e-06,
+      "loss": 1.7922,
+      "step": 18469000
+    },
+    {
+      "epoch": 91.5,
+      "learning_rate": 4.2642226879307185e-06,
+      "loss": 1.8079,
+      "step": 18469500
+    },
+    {
+      "epoch": 91.51,
+      "learning_rate": 4.262986578677487e-06,
+      "loss": 1.8125,
+      "step": 18470000
+    },
+    {
+      "epoch": 91.51,
+      "learning_rate": 4.2617479922514034e-06,
+      "loss": 1.7899,
+      "step": 18470500
+    },
+    {
+      "epoch": 91.51,
+      "learning_rate": 4.2605094058253195e-06,
+      "loss": 1.8053,
+      "step": 18471000
+    },
+    {
+      "epoch": 91.51,
+      "learning_rate": 4.2592708193992364e-06,
+      "loss": 1.8038,
+      "step": 18471500
+    },
+    {
+      "epoch": 91.52,
+      "learning_rate": 4.2580322329731525e-06,
+      "loss": 1.7897,
+      "step": 18472000
+    },
+    {
+      "epoch": 91.52,
+      "learning_rate": 4.2567961237199205e-06,
+      "loss": 1.7926,
+      "step": 18472500
+    },
+    {
+      "epoch": 91.52,
+      "learning_rate": 4.255557537293837e-06,
+      "loss": 1.8311,
+      "step": 18473000
+    },
+    {
+      "epoch": 91.52,
+      "learning_rate": 4.254318950867754e-06,
+      "loss": 1.8104,
+      "step": 18473500
+    },
+    {
+      "epoch": 91.53,
+      "learning_rate": 4.25308036444167e-06,
+      "loss": 1.8184,
+      "step": 18474000
+    },
+    {
+      "epoch": 91.53,
+      "learning_rate": 4.251844255188438e-06,
+      "loss": 1.8073,
+      "step": 18474500
+    },
+    {
+      "epoch": 91.53,
+      "learning_rate": 4.250605668762355e-06,
+      "loss": 1.8114,
+      "step": 18475000
+    },
+    {
+      "epoch": 91.53,
+      "learning_rate": 4.249369559509123e-06,
+      "loss": 1.7977,
+      "step": 18475500
+    },
+    {
+      "epoch": 91.54,
+      "learning_rate": 4.24813097308304e-06,
+      "loss": 1.7972,
+      "step": 18476000
+    },
+    {
+      "epoch": 91.54,
+      "learning_rate": 4.246892386656956e-06,
+      "loss": 1.8107,
+      "step": 18476500
+    },
+    {
+      "epoch": 91.54,
+      "learning_rate": 4.245653800230872e-06,
+      "loss": 1.7936,
+      "step": 18477000
+    },
+    {
+      "epoch": 91.54,
+      "learning_rate": 4.244417690977641e-06,
+      "loss": 1.812,
+      "step": 18477500
+    },
+    {
+      "epoch": 91.55,
+      "learning_rate": 4.243179104551557e-06,
+      "loss": 1.812,
+      "step": 18478000
+    },
+    {
+      "epoch": 91.55,
+      "learning_rate": 4.241940518125474e-06,
+      "loss": 1.8184,
+      "step": 18478500
+    },
+    {
+      "epoch": 91.55,
+      "learning_rate": 4.240704408872242e-06,
+      "loss": 1.7866,
+      "step": 18479000
+    },
+    {
+      "epoch": 91.55,
+      "learning_rate": 4.239465822446159e-06,
+      "loss": 1.7878,
+      "step": 18479500
+    },
+    {
+      "epoch": 91.56,
+      "learning_rate": 4.238227236020075e-06,
+      "loss": 1.8121,
+      "step": 18480000
+    },
+    {
+      "epoch": 91.56,
+      "learning_rate": 4.236988649593992e-06,
+      "loss": 1.7962,
+      "step": 18480500
+    },
+    {
+      "epoch": 91.56,
+      "learning_rate": 4.235750063167908e-06,
+      "loss": 1.7987,
+      "step": 18481000
+    },
+    {
+      "epoch": 91.56,
+      "learning_rate": 4.234511476741824e-06,
+      "loss": 1.8132,
+      "step": 18481500
+    },
+    {
+      "epoch": 91.57,
+      "learning_rate": 4.23327289031574e-06,
+      "loss": 1.8095,
+      "step": 18482000
+    },
+    {
+      "epoch": 91.57,
+      "learning_rate": 4.232034303889657e-06,
+      "loss": 1.8208,
+      "step": 18482500
+    },
+    {
+      "epoch": 91.57,
+      "learning_rate": 4.230795717463573e-06,
+      "loss": 1.788,
+      "step": 18483000
+    },
+    {
+      "epoch": 91.57,
+      "learning_rate": 4.22955713103749e-06,
+      "loss": 1.8069,
+      "step": 18483500
+    },
+    {
+      "epoch": 91.58,
+      "learning_rate": 4.2283185446114055e-06,
+      "loss": 1.7952,
+      "step": 18484000
+    },
+    {
+      "epoch": 91.58,
+      "learning_rate": 4.227079958185322e-06,
+      "loss": 1.8092,
+      "step": 18484500
+    },
+    {
+      "epoch": 91.58,
+      "learning_rate": 4.2258413717592385e-06,
+      "loss": 1.8073,
+      "step": 18485000
+    },
+    {
+      "epoch": 91.58,
+      "learning_rate": 4.2246027853331554e-06,
+      "loss": 1.7964,
+      "step": 18485500
+    },
+    {
+      "epoch": 91.59,
+      "learning_rate": 4.2233641989070715e-06,
+      "loss": 1.8035,
+      "step": 18486000
+    },
+    {
+      "epoch": 91.59,
+      "learning_rate": 4.222125612480988e-06,
+      "loss": 1.8093,
+      "step": 18486500
+    },
+    {
+      "epoch": 91.59,
+      "learning_rate": 4.2208870260549045e-06,
+      "loss": 1.8137,
+      "step": 18487000
+    },
+    {
+      "epoch": 91.59,
+      "learning_rate": 4.219648439628821e-06,
+      "loss": 1.7957,
+      "step": 18487500
+    },
+    {
+      "epoch": 91.6,
+      "learning_rate": 4.2184098532027375e-06,
+      "loss": 1.8314,
+      "step": 18488000
+    },
+    {
+      "epoch": 91.6,
+      "learning_rate": 4.217171266776653e-06,
+      "loss": 1.805,
+      "step": 18488500
+    },
+    {
+      "epoch": 91.6,
+      "learning_rate": 4.21593268035057e-06,
+      "loss": 1.7765,
+      "step": 18489000
+    },
+    {
+      "epoch": 91.6,
+      "learning_rate": 4.214694093924486e-06,
+      "loss": 1.7965,
+      "step": 18489500
+    },
+    {
+      "epoch": 91.61,
+      "learning_rate": 4.213455507498403e-06,
+      "loss": 1.7834,
+      "step": 18490000
+    },
+    {
+      "epoch": 91.61,
+      "learning_rate": 4.212216921072319e-06,
+      "loss": 1.7969,
+      "step": 18490500
+    },
+    {
+      "epoch": 91.61,
+      "learning_rate": 4.2109832889919395e-06,
+      "loss": 1.7914,
+      "step": 18491000
+    },
+    {
+      "epoch": 91.61,
+      "learning_rate": 4.209747179738708e-06,
+      "loss": 1.8087,
+      "step": 18491500
+    },
+    {
+      "epoch": 91.62,
+      "learning_rate": 4.208508593312624e-06,
+      "loss": 1.788,
+      "step": 18492000
+    },
+    {
+      "epoch": 91.62,
+      "learning_rate": 4.2072700068865404e-06,
+      "loss": 1.7911,
+      "step": 18492500
+    },
+    {
+      "epoch": 91.62,
+      "learning_rate": 4.206031420460457e-06,
+      "loss": 1.8222,
+      "step": 18493000
+    },
+    {
+      "epoch": 91.62,
+      "learning_rate": 4.2047928340343735e-06,
+      "loss": 1.8048,
+      "step": 18493500
+    },
+    {
+      "epoch": 91.63,
+      "learning_rate": 4.2035542476082895e-06,
+      "loss": 1.7843,
+      "step": 18494000
+    },
+    {
+      "epoch": 91.63,
+      "learning_rate": 4.202315661182206e-06,
+      "loss": 1.8122,
+      "step": 18494500
+    },
+    {
+      "epoch": 91.63,
+      "learning_rate": 4.201079551928975e-06,
+      "loss": 1.7916,
+      "step": 18495000
+    },
+    {
+      "epoch": 91.63,
+      "learning_rate": 4.1998409655028905e-06,
+      "loss": 1.824,
+      "step": 18495500
+    },
+    {
+      "epoch": 91.64,
+      "learning_rate": 4.19860485624966e-06,
+      "loss": 1.8045,
+      "step": 18496000
+    },
+    {
+      "epoch": 91.64,
+      "learning_rate": 4.197366269823575e-06,
+      "loss": 1.7892,
+      "step": 18496500
+    },
+    {
+      "epoch": 91.64,
+      "learning_rate": 4.196127683397492e-06,
+      "loss": 1.7754,
+      "step": 18497000
+    },
+    {
+      "epoch": 91.64,
+      "learning_rate": 4.194889096971408e-06,
+      "loss": 1.8226,
+      "step": 18497500
+    },
+    {
+      "epoch": 91.65,
+      "learning_rate": 4.193650510545325e-06,
+      "loss": 1.8181,
+      "step": 18498000
+    },
+    {
+      "epoch": 91.65,
+      "learning_rate": 4.192414401292093e-06,
+      "loss": 1.7891,
+      "step": 18498500
+    },
+    {
+      "epoch": 91.65,
+      "learning_rate": 4.19117581486601e-06,
+      "loss": 1.8071,
+      "step": 18499000
+    },
+    {
+      "epoch": 91.65,
+      "learning_rate": 4.189937228439926e-06,
+      "loss": 1.7856,
+      "step": 18499500
+    },
+    {
+      "epoch": 91.66,
+      "learning_rate": 4.188698642013842e-06,
+      "loss": 1.766,
+      "step": 18500000
+    },
+    {
+      "epoch": 91.66,
+      "learning_rate": 4.1874600555877585e-06,
+      "loss": 1.8005,
+      "step": 18500500
+    },
+    {
+      "epoch": 91.66,
+      "learning_rate": 4.186223946334527e-06,
+      "loss": 1.8345,
+      "step": 18501000
+    },
+    {
+      "epoch": 91.66,
+      "learning_rate": 4.184985359908443e-06,
+      "loss": 1.7774,
+      "step": 18501500
+    },
+    {
+      "epoch": 91.67,
+      "learning_rate": 4.18374677348236e-06,
+      "loss": 1.8103,
+      "step": 18502000
+    },
+    {
+      "epoch": 91.67,
+      "learning_rate": 4.182508187056276e-06,
+      "loss": 1.8228,
+      "step": 18502500
+    },
+    {
+      "epoch": 91.67,
+      "learning_rate": 4.181269600630193e-06,
+      "loss": 1.8207,
+      "step": 18503000
+    },
+    {
+      "epoch": 91.67,
+      "learning_rate": 4.180031014204109e-06,
+      "loss": 1.7878,
+      "step": 18503500
+    },
+    {
+      "epoch": 91.68,
+      "learning_rate": 4.1787924277780255e-06,
+      "loss": 1.819,
+      "step": 18504000
+    },
+    {
+      "epoch": 91.68,
+      "learning_rate": 4.177553841351942e-06,
+      "loss": 1.835,
+      "step": 18504500
+    },
+    {
+      "epoch": 91.68,
+      "learning_rate": 4.1763152549258585e-06,
+      "loss": 1.7822,
+      "step": 18505000
+    },
+    {
+      "epoch": 91.68,
+      "learning_rate": 4.1750766684997745e-06,
+      "loss": 1.7858,
+      "step": 18505500
+    },
+    {
+      "epoch": 91.69,
+      "learning_rate": 4.173838082073691e-06,
+      "loss": 1.7809,
+      "step": 18506000
+    },
+    {
+      "epoch": 91.69,
+      "learning_rate": 4.1725994956476076e-06,
+      "loss": 1.8198,
+      "step": 18506500
+    },
+    {
+      "epoch": 91.69,
+      "learning_rate": 4.1713633863943755e-06,
+      "loss": 1.8045,
+      "step": 18507000
+    },
+    {
+      "epoch": 91.69,
+      "learning_rate": 4.170129754313996e-06,
+      "loss": 1.7881,
+      "step": 18507500
+    },
+    {
+      "epoch": 91.7,
+      "learning_rate": 4.168891167887913e-06,
+      "loss": 1.8333,
+      "step": 18508000
+    },
+    {
+      "epoch": 91.7,
+      "learning_rate": 4.16765258146183e-06,
+      "loss": 1.8031,
+      "step": 18508500
+    },
+    {
+      "epoch": 91.7,
+      "learning_rate": 4.166413995035745e-06,
+      "loss": 1.8146,
+      "step": 18509000
+    },
+    {
+      "epoch": 91.7,
+      "learning_rate": 4.165175408609662e-06,
+      "loss": 1.8147,
+      "step": 18509500
+    },
+    {
+      "epoch": 91.7,
+      "learning_rate": 4.163936822183578e-06,
+      "loss": 1.8014,
+      "step": 18510000
+    },
+    {
+      "epoch": 91.71,
+      "learning_rate": 4.162698235757495e-06,
+      "loss": 1.8102,
+      "step": 18510500
+    },
+    {
+      "epoch": 91.71,
+      "learning_rate": 4.161459649331411e-06,
+      "loss": 1.8136,
+      "step": 18511000
+    },
+    {
+      "epoch": 91.71,
+      "learning_rate": 4.160221062905327e-06,
+      "loss": 1.8024,
+      "step": 18511500
+    },
+    {
+      "epoch": 91.71,
+      "learning_rate": 4.1589824764792435e-06,
+      "loss": 1.811,
+      "step": 18512000
+    },
+    {
+      "epoch": 91.72,
+      "learning_rate": 4.15774389005316e-06,
+      "loss": 1.8004,
+      "step": 18512500
+    },
+    {
+      "epoch": 91.72,
+      "learning_rate": 4.1565053036270765e-06,
+      "loss": 1.8176,
+      "step": 18513000
+    },
+    {
+      "epoch": 91.72,
+      "learning_rate": 4.155266717200993e-06,
+      "loss": 1.8076,
+      "step": 18513500
+    },
+    {
+      "epoch": 91.72,
+      "learning_rate": 4.1540281307749095e-06,
+      "loss": 1.8059,
+      "step": 18514000
+    },
+    {
+      "epoch": 91.73,
+      "learning_rate": 4.152792021521678e-06,
+      "loss": 1.7871,
+      "step": 18514500
+    },
+    {
+      "epoch": 91.73,
+      "learning_rate": 4.1515534350955935e-06,
+      "loss": 1.8129,
+      "step": 18515000
+    },
+    {
+      "epoch": 91.73,
+      "learning_rate": 4.1503148486695105e-06,
+      "loss": 1.811,
+      "step": 18515500
+    },
+    {
+      "epoch": 91.73,
+      "learning_rate": 4.149076262243427e-06,
+      "loss": 1.8213,
+      "step": 18516000
+    },
+    {
+      "epoch": 91.74,
+      "learning_rate": 4.1478376758173435e-06,
+      "loss": 1.8046,
+      "step": 18516500
+    },
+    {
+      "epoch": 91.74,
+      "learning_rate": 4.14659908939126e-06,
+      "loss": 1.8373,
+      "step": 18517000
+    },
+    {
+      "epoch": 91.74,
+      "learning_rate": 4.145362980138028e-06,
+      "loss": 1.8008,
+      "step": 18517500
+    },
+    {
+      "epoch": 91.74,
+      "learning_rate": 4.144126870884796e-06,
+      "loss": 1.7921,
+      "step": 18518000
+    },
+    {
+      "epoch": 91.75,
+      "learning_rate": 4.142888284458713e-06,
+      "loss": 1.799,
+      "step": 18518500
+    },
+    {
+      "epoch": 91.75,
+      "learning_rate": 4.14164969803263e-06,
+      "loss": 1.804,
+      "step": 18519000
+    },
+    {
+      "epoch": 91.75,
+      "learning_rate": 4.1404111116065454e-06,
+      "loss": 1.8077,
+      "step": 18519500
+    },
+    {
+      "epoch": 91.75,
+      "learning_rate": 4.139172525180462e-06,
+      "loss": 1.7993,
+      "step": 18520000
+    },
+    {
+      "epoch": 91.76,
+      "learning_rate": 4.1379339387543784e-06,
+      "loss": 1.7859,
+      "step": 18520500
+    },
+    {
+      "epoch": 91.76,
+      "learning_rate": 4.136695352328295e-06,
+      "loss": 1.8017,
+      "step": 18521000
+    },
+    {
+      "epoch": 91.76,
+      "learning_rate": 4.1354567659022114e-06,
+      "loss": 1.8077,
+      "step": 18521500
+    },
+    {
+      "epoch": 91.76,
+      "learning_rate": 4.1342181794761275e-06,
+      "loss": 1.814,
+      "step": 18522000
+    },
+    {
+      "epoch": 91.77,
+      "learning_rate": 4.132979593050044e-06,
+      "loss": 1.7972,
+      "step": 18522500
+    },
+    {
+      "epoch": 91.77,
+      "learning_rate": 4.1317410066239605e-06,
+      "loss": 1.7909,
+      "step": 18523000
+    },
+    {
+      "epoch": 91.77,
+      "learning_rate": 4.130502420197877e-06,
+      "loss": 1.8196,
+      "step": 18523500
+    },
+    {
+      "epoch": 91.77,
+      "learning_rate": 4.1292663109446454e-06,
+      "loss": 1.7972,
+      "step": 18524000
+    },
+    {
+      "epoch": 91.78,
+      "learning_rate": 4.1280277245185615e-06,
+      "loss": 1.8111,
+      "step": 18524500
+    },
+    {
+      "epoch": 91.78,
+      "learning_rate": 4.1267891380924784e-06,
+      "loss": 1.8073,
+      "step": 18525000
+    },
+    {
+      "epoch": 91.78,
+      "learning_rate": 4.125550551666394e-06,
+      "loss": 1.8265,
+      "step": 18525500
+    },
+    {
+      "epoch": 91.78,
+      "learning_rate": 4.124314442413163e-06,
+      "loss": 1.8081,
+      "step": 18526000
+    },
+    {
+      "epoch": 91.79,
+      "learning_rate": 4.123075855987079e-06,
+      "loss": 1.8251,
+      "step": 18526500
+    },
+    {
+      "epoch": 91.79,
+      "learning_rate": 4.121839746733848e-06,
+      "loss": 1.799,
+      "step": 18527000
+    },
+    {
+      "epoch": 91.79,
+      "learning_rate": 4.120601160307764e-06,
+      "loss": 1.8239,
+      "step": 18527500
+    },
+    {
+      "epoch": 91.79,
+      "learning_rate": 4.11936257388168e-06,
+      "loss": 1.8177,
+      "step": 18528000
+    },
+    {
+      "epoch": 91.8,
+      "learning_rate": 4.1181239874555965e-06,
+      "loss": 1.7974,
+      "step": 18528500
+    },
+    {
+      "epoch": 91.8,
+      "learning_rate": 4.116885401029513e-06,
+      "loss": 1.796,
+      "step": 18529000
+    },
+    {
+      "epoch": 91.8,
+      "learning_rate": 4.11564681460343e-06,
+      "loss": 1.8176,
+      "step": 18529500
+    },
+    {
+      "epoch": 91.8,
+      "learning_rate": 4.1144082281773456e-06,
+      "loss": 1.7928,
+      "step": 18530000
+    },
+    {
+      "epoch": 91.81,
+      "learning_rate": 4.1131696417512625e-06,
+      "loss": 1.781,
+      "step": 18530500
+    },
+    {
+      "epoch": 91.81,
+      "learning_rate": 4.1119310553251786e-06,
+      "loss": 1.8005,
+      "step": 18531000
+    },
+    {
+      "epoch": 91.81,
+      "learning_rate": 4.110694946071947e-06,
+      "loss": 1.8261,
+      "step": 18531500
+    },
+    {
+      "epoch": 91.81,
+      "learning_rate": 4.1094563596458634e-06,
+      "loss": 1.8171,
+      "step": 18532000
+    },
+    {
+      "epoch": 91.82,
+      "learning_rate": 4.10821777321978e-06,
+      "loss": 1.7985,
+      "step": 18532500
+    },
+    {
+      "epoch": 91.82,
+      "learning_rate": 4.106981663966548e-06,
+      "loss": 1.7756,
+      "step": 18533000
+    },
+    {
+      "epoch": 91.82,
+      "learning_rate": 4.105743077540465e-06,
+      "loss": 1.7974,
+      "step": 18533500
+    },
+    {
+      "epoch": 91.82,
+      "learning_rate": 4.104504491114381e-06,
+      "loss": 1.8083,
+      "step": 18534000
+    },
+    {
+      "epoch": 91.83,
+      "learning_rate": 4.1032659046882974e-06,
+      "loss": 1.794,
+      "step": 18534500
+    },
+    {
+      "epoch": 91.83,
+      "learning_rate": 4.1020273182622135e-06,
+      "loss": 1.7819,
+      "step": 18535000
+    },
+    {
+      "epoch": 91.83,
+      "learning_rate": 4.1007887318361304e-06,
+      "loss": 1.8181,
+      "step": 18535500
+    },
+    {
+      "epoch": 91.83,
+      "learning_rate": 4.0995501454100465e-06,
+      "loss": 1.8284,
+      "step": 18536000
+    },
+    {
+      "epoch": 91.84,
+      "learning_rate": 4.0983115589839634e-06,
+      "loss": 1.7931,
+      "step": 18536500
+    },
+    {
+      "epoch": 91.84,
+      "learning_rate": 4.097072972557879e-06,
+      "loss": 1.8073,
+      "step": 18537000
+    },
+    {
+      "epoch": 91.84,
+      "learning_rate": 4.095834386131796e-06,
+      "loss": 1.8103,
+      "step": 18537500
+    },
+    {
+      "epoch": 91.84,
+      "learning_rate": 4.094595799705712e-06,
+      "loss": 1.8161,
+      "step": 18538000
+    },
+    {
+      "epoch": 91.85,
+      "learning_rate": 4.093357213279629e-06,
+      "loss": 1.8134,
+      "step": 18538500
+    },
+    {
+      "epoch": 91.85,
+      "learning_rate": 4.092118626853545e-06,
+      "loss": 1.8162,
+      "step": 18539000
+    },
+    {
+      "epoch": 91.85,
+      "learning_rate": 4.090880040427461e-06,
+      "loss": 1.8092,
+      "step": 18539500
+    },
+    {
+      "epoch": 91.85,
+      "learning_rate": 4.089641454001378e-06,
+      "loss": 1.8184,
+      "step": 18540000
+    },
+    {
+      "epoch": 91.86,
+      "learning_rate": 4.088402867575294e-06,
+      "loss": 1.7973,
+      "step": 18540500
+    },
+    {
+      "epoch": 91.86,
+      "learning_rate": 4.087166758322063e-06,
+      "loss": 1.8087,
+      "step": 18541000
+    },
+    {
+      "epoch": 91.86,
+      "learning_rate": 4.085928171895979e-06,
+      "loss": 1.7835,
+      "step": 18541500
+    },
+    {
+      "epoch": 91.86,
+      "learning_rate": 4.084689585469896e-06,
+      "loss": 1.8248,
+      "step": 18542000
+    },
+    {
+      "epoch": 91.87,
+      "learning_rate": 4.0834534762166636e-06,
+      "loss": 1.8075,
+      "step": 18542500
+    },
+    {
+      "epoch": 91.87,
+      "learning_rate": 4.0822148897905805e-06,
+      "loss": 1.8304,
+      "step": 18543000
+    },
+    {
+      "epoch": 91.87,
+      "learning_rate": 4.080976303364497e-06,
+      "loss": 1.7839,
+      "step": 18543500
+    },
+    {
+      "epoch": 91.87,
+      "learning_rate": 4.079737716938413e-06,
+      "loss": 1.7811,
+      "step": 18544000
+    },
+    {
+      "epoch": 91.88,
+      "learning_rate": 4.078499130512329e-06,
+      "loss": 1.7882,
+      "step": 18544500
+    },
+    {
+      "epoch": 91.88,
+      "learning_rate": 4.077260544086246e-06,
+      "loss": 1.8047,
+      "step": 18545000
+    },
+    {
+      "epoch": 91.88,
+      "learning_rate": 4.076021957660162e-06,
+      "loss": 1.7723,
+      "step": 18545500
+    },
+    {
+      "epoch": 91.88,
+      "learning_rate": 4.0747858484069306e-06,
+      "loss": 1.8135,
+      "step": 18546000
+    },
+    {
+      "epoch": 91.89,
+      "learning_rate": 4.073547261980847e-06,
+      "loss": 1.8053,
+      "step": 18546500
+    },
+    {
+      "epoch": 91.89,
+      "learning_rate": 4.0723086755547636e-06,
+      "loss": 1.8002,
+      "step": 18547000
+    },
+    {
+      "epoch": 91.89,
+      "learning_rate": 4.0710725663015315e-06,
+      "loss": 1.8031,
+      "step": 18547500
+    },
+    {
+      "epoch": 91.89,
+      "learning_rate": 4.0698339798754485e-06,
+      "loss": 1.7937,
+      "step": 18548000
+    },
+    {
+      "epoch": 91.9,
+      "learning_rate": 4.068595393449364e-06,
+      "loss": 1.7992,
+      "step": 18548500
+    },
+    {
+      "epoch": 91.9,
+      "learning_rate": 4.067356807023281e-06,
+      "loss": 1.8354,
+      "step": 18549000
+    },
+    {
+      "epoch": 91.9,
+      "learning_rate": 4.066118220597197e-06,
+      "loss": 1.7761,
+      "step": 18549500
+    },
+    {
+      "epoch": 91.9,
+      "learning_rate": 4.064879634171114e-06,
+      "loss": 1.7857,
+      "step": 18550000
+    },
+    {
+      "epoch": 91.91,
+      "learning_rate": 4.06364104774503e-06,
+      "loss": 1.8107,
+      "step": 18550500
+    },
+    {
+      "epoch": 91.91,
+      "learning_rate": 4.062402461318946e-06,
+      "loss": 1.8407,
+      "step": 18551000
+    },
+    {
+      "epoch": 91.91,
+      "learning_rate": 4.061166352065715e-06,
+      "loss": 1.8099,
+      "step": 18551500
+    },
+    {
+      "epoch": 91.91,
+      "learning_rate": 4.059930242812483e-06,
+      "loss": 1.8294,
+      "step": 18552000
+    },
+    {
+      "epoch": 91.92,
+      "learning_rate": 4.0586916563863995e-06,
+      "loss": 1.7949,
+      "step": 18552500
+    },
+    {
+      "epoch": 91.92,
+      "learning_rate": 4.057455547133168e-06,
+      "loss": 1.793,
+      "step": 18553000
+    },
+    {
+      "epoch": 91.92,
+      "learning_rate": 4.056216960707084e-06,
+      "loss": 1.8161,
+      "step": 18553500
+    },
+    {
+      "epoch": 91.92,
+      "learning_rate": 4.0549783742810005e-06,
+      "loss": 1.8104,
+      "step": 18554000
+    },
+    {
+      "epoch": 91.93,
+      "learning_rate": 4.0537397878549166e-06,
+      "loss": 1.7961,
+      "step": 18554500
+    },
+    {
+      "epoch": 91.93,
+      "learning_rate": 4.0525012014288335e-06,
+      "loss": 1.8115,
+      "step": 18555000
+    },
+    {
+      "epoch": 91.93,
+      "learning_rate": 4.0512626150027496e-06,
+      "loss": 1.809,
+      "step": 18555500
+    },
+    {
+      "epoch": 91.93,
+      "learning_rate": 4.0500240285766665e-06,
+      "loss": 1.8076,
+      "step": 18556000
+    },
+    {
+      "epoch": 91.94,
+      "learning_rate": 4.0487879193234345e-06,
+      "loss": 1.8158,
+      "step": 18556500
+    },
+    {
+      "epoch": 91.94,
+      "learning_rate": 4.047549332897351e-06,
+      "loss": 1.7911,
+      "step": 18557000
+    },
+    {
+      "epoch": 91.94,
+      "learning_rate": 4.0463107464712675e-06,
+      "loss": 1.8222,
+      "step": 18557500
+    },
+    {
+      "epoch": 91.94,
+      "learning_rate": 4.0450721600451835e-06,
+      "loss": 1.7996,
+      "step": 18558000
+    },
+    {
+      "epoch": 91.95,
+      "learning_rate": 4.0438335736191005e-06,
+      "loss": 1.7807,
+      "step": 18558500
+    },
+    {
+      "epoch": 91.95,
+      "learning_rate": 4.0425949871930165e-06,
+      "loss": 1.8013,
+      "step": 18559000
+    },
+    {
+      "epoch": 91.95,
+      "learning_rate": 4.0413564007669335e-06,
+      "loss": 1.8112,
+      "step": 18559500
+    },
+    {
+      "epoch": 91.95,
+      "learning_rate": 4.040117814340849e-06,
+      "loss": 1.8027,
+      "step": 18560000
+    },
+    {
+      "epoch": 91.96,
+      "learning_rate": 4.038879227914766e-06,
+      "loss": 1.8141,
+      "step": 18560500
+    },
+    {
+      "epoch": 91.96,
+      "learning_rate": 4.037640641488682e-06,
+      "loss": 1.8385,
+      "step": 18561000
+    },
+    {
+      "epoch": 91.96,
+      "learning_rate": 4.036402055062599e-06,
+      "loss": 1.8135,
+      "step": 18561500
+    },
+    {
+      "epoch": 91.96,
+      "learning_rate": 4.035163468636515e-06,
+      "loss": 1.7873,
+      "step": 18562000
+    },
+    {
+      "epoch": 91.97,
+      "learning_rate": 4.033924882210431e-06,
+      "loss": 1.8105,
+      "step": 18562500
+    },
+    {
+      "epoch": 91.97,
+      "learning_rate": 4.032686295784347e-06,
+      "loss": 1.8076,
+      "step": 18563000
+    },
+    {
+      "epoch": 91.97,
+      "learning_rate": 4.031447709358264e-06,
+      "loss": 1.7805,
+      "step": 18563500
+    },
+    {
+      "epoch": 91.97,
+      "learning_rate": 4.030209122932181e-06,
+      "loss": 1.7807,
+      "step": 18564000
+    },
+    {
+      "epoch": 91.97,
+      "learning_rate": 4.028970536506097e-06,
+      "loss": 1.8205,
+      "step": 18564500
+    },
+    {
+      "epoch": 91.98,
+      "learning_rate": 4.027734427252865e-06,
+      "loss": 1.8214,
+      "step": 18565000
+    },
+    {
+      "epoch": 91.98,
+      "learning_rate": 4.026498317999634e-06,
+      "loss": 1.7879,
+      "step": 18565500
+    },
+    {
+      "epoch": 91.98,
+      "learning_rate": 4.02525973157355e-06,
+      "loss": 1.7966,
+      "step": 18566000
+    },
+    {
+      "epoch": 91.98,
+      "learning_rate": 4.0240236223203185e-06,
+      "loss": 1.7945,
+      "step": 18566500
+    },
+    {
+      "epoch": 91.99,
+      "learning_rate": 4.0227850358942346e-06,
+      "loss": 1.8153,
+      "step": 18567000
+    },
+    {
+      "epoch": 91.99,
+      "learning_rate": 4.0215464494681515e-06,
+      "loss": 1.8056,
+      "step": 18567500
+    },
+    {
+      "epoch": 91.99,
+      "learning_rate": 4.020307863042068e-06,
+      "loss": 1.7954,
+      "step": 18568000
+    },
+    {
+      "epoch": 91.99,
+      "learning_rate": 4.019069276615984e-06,
+      "loss": 1.8094,
+      "step": 18568500
+    },
+    {
+      "epoch": 92.0,
+      "learning_rate": 4.017830690189901e-06,
+      "loss": 1.803,
+      "step": 18569000
+    },
+    {
+      "epoch": 92.0,
+      "learning_rate": 4.016592103763817e-06,
+      "loss": 1.7962,
+      "step": 18569500
+    },
+    {
+      "epoch": 92.0,
+      "eval_accuracy": 0.6854704703021435,
+      "eval_accuracy_mlm": 0.646739844825033,
+      "eval_accuracy_nsp": 0.8680650614412513,
+      "eval_loss": 2.325012445449829,
+      "eval_runtime": 147.0403,
+      "eval_samples_per_second": 1733.939,
+      "eval_steps_per_second": 72.252,
+      "step": 18569556
+    },
+    {
+      "epoch": 92.0,
+      "learning_rate": 4.015353517337734e-06,
+      "loss": 1.7942,
+      "step": 18570000
+    },
+    {
+      "epoch": 92.0,
+      "learning_rate": 4.014114930911649e-06,
+      "loss": 1.791,
+      "step": 18570500
+    },
+    {
+      "epoch": 92.01,
+      "learning_rate": 4.012876344485566e-06,
+      "loss": 1.7914,
+      "step": 18571000
+    },
+    {
+      "epoch": 92.01,
+      "learning_rate": 4.011637758059482e-06,
+      "loss": 1.7903,
+      "step": 18571500
+    },
+    {
+      "epoch": 92.01,
+      "learning_rate": 4.010399171633399e-06,
+      "loss": 1.7944,
+      "step": 18572000
+    },
+    {
+      "epoch": 92.01,
+      "learning_rate": 4.009160585207315e-06,
+      "loss": 1.7605,
+      "step": 18572500
+    },
+    {
+      "epoch": 92.02,
+      "learning_rate": 4.007921998781231e-06,
+      "loss": 1.7905,
+      "step": 18573000
+    },
+    {
+      "epoch": 92.02,
+      "learning_rate": 4.006685889528e-06,
+      "loss": 1.8214,
+      "step": 18573500
+    },
+    {
+      "epoch": 92.02,
+      "learning_rate": 4.005447303101916e-06,
+      "loss": 1.8067,
+      "step": 18574000
+    },
+    {
+      "epoch": 92.02,
+      "learning_rate": 4.004208716675832e-06,
+      "loss": 1.8162,
+      "step": 18574500
+    },
+    {
+      "epoch": 92.03,
+      "learning_rate": 4.002970130249749e-06,
+      "loss": 1.7994,
+      "step": 18575000
+    },
+    {
+      "epoch": 92.03,
+      "learning_rate": 4.001731543823665e-06,
+      "loss": 1.7994,
+      "step": 18575500
+    },
+    {
+      "epoch": 92.03,
+      "learning_rate": 4.000492957397582e-06,
+      "loss": 1.7949,
+      "step": 18576000
+    },
+    {
+      "epoch": 92.03,
+      "learning_rate": 3.99925684814435e-06,
+      "loss": 1.7809,
+      "step": 18576500
+    },
+    {
+      "epoch": 92.04,
+      "learning_rate": 3.998018261718267e-06,
+      "loss": 1.8145,
+      "step": 18577000
+    },
+    {
+      "epoch": 92.04,
+      "learning_rate": 3.996782152465035e-06,
+      "loss": 1.7966,
+      "step": 18577500
+    },
+    {
+      "epoch": 92.04,
+      "learning_rate": 3.9955460432118035e-06,
+      "loss": 1.8132,
+      "step": 18578000
+    },
+    {
+      "epoch": 92.04,
+      "learning_rate": 3.99430745678572e-06,
+      "loss": 1.7801,
+      "step": 18578500
+    },
+    {
+      "epoch": 92.05,
+      "learning_rate": 3.9930688703596365e-06,
+      "loss": 1.7611,
+      "step": 18579000
+    },
+    {
+      "epoch": 92.05,
+      "learning_rate": 3.991830283933553e-06,
+      "loss": 1.7756,
+      "step": 18579500
+    },
+    {
+      "epoch": 92.05,
+      "learning_rate": 3.990591697507469e-06,
+      "loss": 1.8085,
+      "step": 18580000
+    },
+    {
+      "epoch": 92.05,
+      "learning_rate": 3.989353111081385e-06,
+      "loss": 1.7942,
+      "step": 18580500
+    },
+    {
+      "epoch": 92.06,
+      "learning_rate": 3.988114524655302e-06,
+      "loss": 1.7956,
+      "step": 18581000
+    },
+    {
+      "epoch": 92.06,
+      "learning_rate": 3.986875938229219e-06,
+      "loss": 1.8205,
+      "step": 18581500
+    },
+    {
+      "epoch": 92.06,
+      "learning_rate": 3.985637351803134e-06,
+      "loss": 1.7842,
+      "step": 18582000
+    },
+    {
+      "epoch": 92.06,
+      "learning_rate": 3.984398765377051e-06,
+      "loss": 1.7868,
+      "step": 18582500
+    },
+    {
+      "epoch": 92.07,
+      "learning_rate": 3.983160178950967e-06,
+      "loss": 1.7991,
+      "step": 18583000
+    },
+    {
+      "epoch": 92.07,
+      "learning_rate": 3.981921592524884e-06,
+      "loss": 1.7914,
+      "step": 18583500
+    },
+    {
+      "epoch": 92.07,
+      "learning_rate": 3.9806830060988e-06,
+      "loss": 1.7998,
+      "step": 18584000
+    },
+    {
+      "epoch": 92.07,
+      "learning_rate": 3.979446896845569e-06,
+      "loss": 1.7963,
+      "step": 18584500
+    },
+    {
+      "epoch": 92.08,
+      "learning_rate": 3.978210787592337e-06,
+      "loss": 1.8,
+      "step": 18585000
+    },
+    {
+      "epoch": 92.08,
+      "learning_rate": 3.9769722011662536e-06,
+      "loss": 1.7837,
+      "step": 18585500
+    },
+    {
+      "epoch": 92.08,
+      "learning_rate": 3.97573361474017e-06,
+      "loss": 1.8154,
+      "step": 18586000
+    },
+    {
+      "epoch": 92.08,
+      "learning_rate": 3.974495028314086e-06,
+      "loss": 1.8086,
+      "step": 18586500
+    },
+    {
+      "epoch": 92.09,
+      "learning_rate": 3.973256441888002e-06,
+      "loss": 1.789,
+      "step": 18587000
+    },
+    {
+      "epoch": 92.09,
+      "learning_rate": 3.972022809807623e-06,
+      "loss": 1.7797,
+      "step": 18587500
+    },
+    {
+      "epoch": 92.09,
+      "learning_rate": 3.9707842233815394e-06,
+      "loss": 1.8133,
+      "step": 18588000
+    },
+    {
+      "epoch": 92.09,
+      "learning_rate": 3.969545636955456e-06,
+      "loss": 1.7945,
+      "step": 18588500
+    },
+    {
+      "epoch": 92.1,
+      "learning_rate": 3.968307050529372e-06,
+      "loss": 1.7803,
+      "step": 18589000
+    },
+    {
+      "epoch": 92.1,
+      "learning_rate": 3.9670684641032885e-06,
+      "loss": 1.7893,
+      "step": 18589500
+    },
+    {
+      "epoch": 92.1,
+      "learning_rate": 3.965829877677205e-06,
+      "loss": 1.8154,
+      "step": 18590000
+    },
+    {
+      "epoch": 92.1,
+      "learning_rate": 3.9645912912511215e-06,
+      "loss": 1.7905,
+      "step": 18590500
+    },
+    {
+      "epoch": 92.11,
+      "learning_rate": 3.963352704825038e-06,
+      "loss": 1.7984,
+      "step": 18591000
+    },
+    {
+      "epoch": 92.11,
+      "learning_rate": 3.962114118398954e-06,
+      "loss": 1.7857,
+      "step": 18591500
+    },
+    {
+      "epoch": 92.11,
+      "learning_rate": 3.96087553197287e-06,
+      "loss": 1.8122,
+      "step": 18592000
+    },
+    {
+      "epoch": 92.11,
+      "learning_rate": 3.959636945546787e-06,
+      "loss": 1.7881,
+      "step": 18592500
+    },
+    {
+      "epoch": 92.12,
+      "learning_rate": 3.958398359120703e-06,
+      "loss": 1.8289,
+      "step": 18593000
+    },
+    {
+      "epoch": 92.12,
+      "learning_rate": 3.957162249867472e-06,
+      "loss": 1.7944,
+      "step": 18593500
+    },
+    {
+      "epoch": 92.12,
+      "learning_rate": 3.9559261406142396e-06,
+      "loss": 1.7887,
+      "step": 18594000
+    },
+    {
+      "epoch": 92.12,
+      "learning_rate": 3.9546875541881565e-06,
+      "loss": 1.8001,
+      "step": 18594500
+    },
+    {
+      "epoch": 92.13,
+      "learning_rate": 3.9534489677620726e-06,
+      "loss": 1.7967,
+      "step": 18595000
+    },
+    {
+      "epoch": 92.13,
+      "learning_rate": 3.9522103813359895e-06,
+      "loss": 1.7522,
+      "step": 18595500
+    },
+    {
+      "epoch": 92.13,
+      "learning_rate": 3.950971794909905e-06,
+      "loss": 1.778,
+      "step": 18596000
+    },
+    {
+      "epoch": 92.13,
+      "learning_rate": 3.949733208483822e-06,
+      "loss": 1.7917,
+      "step": 18596500
+    },
+    {
+      "epoch": 92.14,
+      "learning_rate": 3.948494622057739e-06,
+      "loss": 1.7987,
+      "step": 18597000
+    },
+    {
+      "epoch": 92.14,
+      "learning_rate": 3.947256035631655e-06,
+      "loss": 1.8182,
+      "step": 18597500
+    },
+    {
+      "epoch": 92.14,
+      "learning_rate": 3.946019926378423e-06,
+      "loss": 1.7826,
+      "step": 18598000
+    },
+    {
+      "epoch": 92.14,
+      "learning_rate": 3.9447813399523396e-06,
+      "loss": 1.7938,
+      "step": 18598500
+    },
+    {
+      "epoch": 92.15,
+      "learning_rate": 3.943542753526256e-06,
+      "loss": 1.8187,
+      "step": 18599000
+    },
+    {
+      "epoch": 92.15,
+      "learning_rate": 3.942304167100172e-06,
+      "loss": 1.7993,
+      "step": 18599500
+    },
+    {
+      "epoch": 92.15,
+      "learning_rate": 3.941065580674089e-06,
+      "loss": 1.8146,
+      "step": 18600000
+    },
+    {
+      "epoch": 92.15,
+      "learning_rate": 3.939829471420857e-06,
+      "loss": 1.7917,
+      "step": 18600500
+    },
+    {
+      "epoch": 92.16,
+      "learning_rate": 3.9385908849947735e-06,
+      "loss": 1.8213,
+      "step": 18601000
+    },
+    {
+      "epoch": 92.16,
+      "learning_rate": 3.93735229856869e-06,
+      "loss": 1.8038,
+      "step": 18601500
+    },
+    {
+      "epoch": 92.16,
+      "learning_rate": 3.9361137121426065e-06,
+      "loss": 1.7937,
+      "step": 18602000
+    },
+    {
+      "epoch": 92.16,
+      "learning_rate": 3.934875125716523e-06,
+      "loss": 1.7866,
+      "step": 18602500
+    },
+    {
+      "epoch": 92.17,
+      "learning_rate": 3.933636539290439e-06,
+      "loss": 1.8031,
+      "step": 18603000
+    },
+    {
+      "epoch": 92.17,
+      "learning_rate": 3.932397952864355e-06,
+      "loss": 1.7807,
+      "step": 18603500
+    },
+    {
+      "epoch": 92.17,
+      "learning_rate": 3.931161843611124e-06,
+      "loss": 1.8292,
+      "step": 18604000
+    },
+    {
+      "epoch": 92.17,
+      "learning_rate": 3.92992325718504e-06,
+      "loss": 1.8037,
+      "step": 18604500
+    },
+    {
+      "epoch": 92.18,
+      "learning_rate": 3.928684670758957e-06,
+      "loss": 1.8093,
+      "step": 18605000
+    },
+    {
+      "epoch": 92.18,
+      "learning_rate": 3.927446084332873e-06,
+      "loss": 1.8048,
+      "step": 18605500
+    },
+    {
+      "epoch": 92.18,
+      "learning_rate": 3.92620749790679e-06,
+      "loss": 1.7917,
+      "step": 18606000
+    },
+    {
+      "epoch": 92.18,
+      "learning_rate": 3.924968911480705e-06,
+      "loss": 1.7729,
+      "step": 18606500
+    },
+    {
+      "epoch": 92.19,
+      "learning_rate": 3.923730325054622e-06,
+      "loss": 1.8028,
+      "step": 18607000
+    },
+    {
+      "epoch": 92.19,
+      "learning_rate": 3.922491738628539e-06,
+      "loss": 1.8052,
+      "step": 18607500
+    },
+    {
+      "epoch": 92.19,
+      "learning_rate": 3.921255629375307e-06,
+      "loss": 1.8231,
+      "step": 18608000
+    },
+    {
+      "epoch": 92.19,
+      "learning_rate": 3.920017042949223e-06,
+      "loss": 1.7973,
+      "step": 18608500
+    },
+    {
+      "epoch": 92.2,
+      "learning_rate": 3.91877845652314e-06,
+      "loss": 1.8125,
+      "step": 18609000
+    },
+    {
+      "epoch": 92.2,
+      "learning_rate": 3.917542347269908e-06,
+      "loss": 1.775,
+      "step": 18609500
+    },
+    {
+      "epoch": 92.2,
+      "learning_rate": 3.9163037608438246e-06,
+      "loss": 1.7731,
+      "step": 18610000
+    },
+    {
+      "epoch": 92.2,
+      "learning_rate": 3.915065174417741e-06,
+      "loss": 1.8171,
+      "step": 18610500
+    },
+    {
+      "epoch": 92.21,
+      "learning_rate": 3.913826587991657e-06,
+      "loss": 1.8017,
+      "step": 18611000
+    },
+    {
+      "epoch": 92.21,
+      "learning_rate": 3.912588001565574e-06,
+      "loss": 1.795,
+      "step": 18611500
+    },
+    {
+      "epoch": 92.21,
+      "learning_rate": 3.91134941513949e-06,
+      "loss": 1.7984,
+      "step": 18612000
+    },
+    {
+      "epoch": 92.21,
+      "learning_rate": 3.910110828713407e-06,
+      "loss": 1.7969,
+      "step": 18612500
+    },
+    {
+      "epoch": 92.22,
+      "learning_rate": 3.908872242287323e-06,
+      "loss": 1.7622,
+      "step": 18613000
+    },
+    {
+      "epoch": 92.22,
+      "learning_rate": 3.907633655861239e-06,
+      "loss": 1.7718,
+      "step": 18613500
+    },
+    {
+      "epoch": 92.22,
+      "learning_rate": 3.906397546608008e-06,
+      "loss": 1.7906,
+      "step": 18614000
+    },
+    {
+      "epoch": 92.22,
+      "learning_rate": 3.905158960181924e-06,
+      "loss": 1.8051,
+      "step": 18614500
+    },
+    {
+      "epoch": 92.23,
+      "learning_rate": 3.903925328101544e-06,
+      "loss": 1.8252,
+      "step": 18615000
+    },
+    {
+      "epoch": 92.23,
+      "learning_rate": 3.9026867416754605e-06,
+      "loss": 1.7962,
+      "step": 18615500
+    },
+    {
+      "epoch": 92.23,
+      "learning_rate": 3.901448155249377e-06,
+      "loss": 1.8158,
+      "step": 18616000
+    },
+    {
+      "epoch": 92.23,
+      "learning_rate": 3.9002095688232935e-06,
+      "loss": 1.7909,
+      "step": 18616500
+    },
+    {
+      "epoch": 92.24,
+      "learning_rate": 3.89897098239721e-06,
+      "loss": 1.807,
+      "step": 18617000
+    },
+    {
+      "epoch": 92.24,
+      "learning_rate": 3.8977323959711265e-06,
+      "loss": 1.7949,
+      "step": 18617500
+    },
+    {
+      "epoch": 92.24,
+      "learning_rate": 3.896493809545043e-06,
+      "loss": 1.7948,
+      "step": 18618000
+    },
+    {
+      "epoch": 92.24,
+      "learning_rate": 3.8952552231189595e-06,
+      "loss": 1.7675,
+      "step": 18618500
+    },
+    {
+      "epoch": 92.24,
+      "learning_rate": 3.894016636692875e-06,
+      "loss": 1.786,
+      "step": 18619000
+    },
+    {
+      "epoch": 92.25,
+      "learning_rate": 3.892778050266792e-06,
+      "loss": 1.8001,
+      "step": 18619500
+    },
+    {
+      "epoch": 92.25,
+      "learning_rate": 3.89154194101356e-06,
+      "loss": 1.8181,
+      "step": 18620000
+    },
+    {
+      "epoch": 92.25,
+      "learning_rate": 3.890305831760329e-06,
+      "loss": 1.7722,
+      "step": 18620500
+    },
+    {
+      "epoch": 92.25,
+      "learning_rate": 3.889067245334245e-06,
+      "loss": 1.7821,
+      "step": 18621000
+    },
+    {
+      "epoch": 92.26,
+      "learning_rate": 3.8878286589081615e-06,
+      "loss": 1.7842,
+      "step": 18621500
+    },
+    {
+      "epoch": 92.26,
+      "learning_rate": 3.8865900724820775e-06,
+      "loss": 1.7998,
+      "step": 18622000
+    },
+    {
+      "epoch": 92.26,
+      "learning_rate": 3.8853514860559945e-06,
+      "loss": 1.7808,
+      "step": 18622500
+    },
+    {
+      "epoch": 92.26,
+      "learning_rate": 3.8841128996299106e-06,
+      "loss": 1.7926,
+      "step": 18623000
+    },
+    {
+      "epoch": 92.27,
+      "learning_rate": 3.882874313203827e-06,
+      "loss": 1.8011,
+      "step": 18623500
+    },
+    {
+      "epoch": 92.27,
+      "learning_rate": 3.881635726777743e-06,
+      "loss": 1.8053,
+      "step": 18624000
+    },
+    {
+      "epoch": 92.27,
+      "learning_rate": 3.88039714035166e-06,
+      "loss": 1.7997,
+      "step": 18624500
+    },
+    {
+      "epoch": 92.27,
+      "learning_rate": 3.8791585539255766e-06,
+      "loss": 1.8283,
+      "step": 18625000
+    },
+    {
+      "epoch": 92.28,
+      "learning_rate": 3.877919967499493e-06,
+      "loss": 1.8005,
+      "step": 18625500
+    },
+    {
+      "epoch": 92.28,
+      "learning_rate": 3.876681381073409e-06,
+      "loss": 1.8112,
+      "step": 18626000
+    },
+    {
+      "epoch": 92.28,
+      "learning_rate": 3.875442794647325e-06,
+      "loss": 1.7964,
+      "step": 18626500
+    },
+    {
+      "epoch": 92.28,
+      "learning_rate": 3.874204208221242e-06,
+      "loss": 1.7872,
+      "step": 18627000
+    },
+    {
+      "epoch": 92.29,
+      "learning_rate": 3.87296809896801e-06,
+      "loss": 1.7815,
+      "step": 18627500
+    },
+    {
+      "epoch": 92.29,
+      "learning_rate": 3.8717319897147785e-06,
+      "loss": 1.8043,
+      "step": 18628000
+    },
+    {
+      "epoch": 92.29,
+      "learning_rate": 3.870493403288695e-06,
+      "loss": 1.8238,
+      "step": 18628500
+    },
+    {
+      "epoch": 92.29,
+      "learning_rate": 3.8692548168626115e-06,
+      "loss": 1.791,
+      "step": 18629000
+    },
+    {
+      "epoch": 92.3,
+      "learning_rate": 3.868016230436528e-06,
+      "loss": 1.8019,
+      "step": 18629500
+    },
+    {
+      "epoch": 92.3,
+      "learning_rate": 3.866780121183296e-06,
+      "loss": 1.7865,
+      "step": 18630000
+    },
+    {
+      "epoch": 92.3,
+      "learning_rate": 3.8655415347572125e-06,
+      "loss": 1.7919,
+      "step": 18630500
+    },
+    {
+      "epoch": 92.3,
+      "learning_rate": 3.8643029483311294e-06,
+      "loss": 1.7932,
+      "step": 18631000
+    },
+    {
+      "epoch": 92.31,
+      "learning_rate": 3.8630643619050455e-06,
+      "loss": 1.8104,
+      "step": 18631500
+    },
+    {
+      "epoch": 92.31,
+      "learning_rate": 3.861828252651814e-06,
+      "loss": 1.7993,
+      "step": 18632000
+    },
+    {
+      "epoch": 92.31,
+      "learning_rate": 3.86058966622573e-06,
+      "loss": 1.7779,
+      "step": 18632500
+    },
+    {
+      "epoch": 92.31,
+      "learning_rate": 3.8593510797996465e-06,
+      "loss": 1.8061,
+      "step": 18633000
+    },
+    {
+      "epoch": 92.32,
+      "learning_rate": 3.8581124933735626e-06,
+      "loss": 1.8067,
+      "step": 18633500
+    },
+    {
+      "epoch": 92.32,
+      "learning_rate": 3.8568739069474795e-06,
+      "loss": 1.8043,
+      "step": 18634000
+    },
+    {
+      "epoch": 92.32,
+      "learning_rate": 3.8556353205213956e-06,
+      "loss": 1.7822,
+      "step": 18634500
+    },
+    {
+      "epoch": 92.32,
+      "learning_rate": 3.854399211268164e-06,
+      "loss": 1.8302,
+      "step": 18635000
+    },
+    {
+      "epoch": 92.33,
+      "learning_rate": 3.8531606248420805e-06,
+      "loss": 1.7923,
+      "step": 18635500
+    },
+    {
+      "epoch": 92.33,
+      "learning_rate": 3.8519220384159965e-06,
+      "loss": 1.7932,
+      "step": 18636000
+    },
+    {
+      "epoch": 92.33,
+      "learning_rate": 3.850683451989913e-06,
+      "loss": 1.8017,
+      "step": 18636500
+    },
+    {
+      "epoch": 92.33,
+      "learning_rate": 3.8494448655638295e-06,
+      "loss": 1.8092,
+      "step": 18637000
+    },
+    {
+      "epoch": 92.34,
+      "learning_rate": 3.848206279137746e-06,
+      "loss": 1.8035,
+      "step": 18637500
+    },
+    {
+      "epoch": 92.34,
+      "learning_rate": 3.8469676927116626e-06,
+      "loss": 1.8179,
+      "step": 18638000
+    },
+    {
+      "epoch": 92.34,
+      "learning_rate": 3.845729106285579e-06,
+      "loss": 1.7945,
+      "step": 18638500
+    },
+    {
+      "epoch": 92.34,
+      "learning_rate": 3.844490519859495e-06,
+      "loss": 1.7748,
+      "step": 18639000
+    },
+    {
+      "epoch": 92.35,
+      "learning_rate": 3.843251933433412e-06,
+      "loss": 1.7926,
+      "step": 18639500
+    },
+    {
+      "epoch": 92.35,
+      "learning_rate": 3.842013347007328e-06,
+      "loss": 1.7776,
+      "step": 18640000
+    },
+    {
+      "epoch": 92.35,
+      "learning_rate": 3.8407772377540965e-06,
+      "loss": 1.816,
+      "step": 18640500
+    },
+    {
+      "epoch": 92.35,
+      "learning_rate": 3.839538651328013e-06,
+      "loss": 1.8118,
+      "step": 18641000
+    },
+    {
+      "epoch": 92.36,
+      "learning_rate": 3.8383000649019295e-06,
+      "loss": 1.7924,
+      "step": 18641500
+    },
+    {
+      "epoch": 92.36,
+      "learning_rate": 3.837061478475845e-06,
+      "loss": 1.8098,
+      "step": 18642000
+    },
+    {
+      "epoch": 92.36,
+      "learning_rate": 3.835822892049762e-06,
+      "loss": 1.8093,
+      "step": 18642500
+    },
+    {
+      "epoch": 92.36,
+      "learning_rate": 3.8345867827965305e-06,
+      "loss": 1.8058,
+      "step": 18643000
+    },
+    {
+      "epoch": 92.37,
+      "learning_rate": 3.833348196370447e-06,
+      "loss": 1.8155,
+      "step": 18643500
+    },
+    {
+      "epoch": 92.37,
+      "learning_rate": 3.832109609944363e-06,
+      "loss": 1.7955,
+      "step": 18644000
+    },
+    {
+      "epoch": 92.37,
+      "learning_rate": 3.83087102351828e-06,
+      "loss": 1.7809,
+      "step": 18644500
+    },
+    {
+      "epoch": 92.37,
+      "learning_rate": 3.829632437092196e-06,
+      "loss": 1.7978,
+      "step": 18645000
+    },
+    {
+      "epoch": 92.38,
+      "learning_rate": 3.828393850666112e-06,
+      "loss": 1.8403,
+      "step": 18645500
+    },
+    {
+      "epoch": 92.38,
+      "learning_rate": 3.827155264240028e-06,
+      "loss": 1.7823,
+      "step": 18646000
+    },
+    {
+      "epoch": 92.38,
+      "learning_rate": 3.825916677813945e-06,
+      "loss": 1.7931,
+      "step": 18646500
+    },
+    {
+      "epoch": 92.38,
+      "learning_rate": 3.824678091387861e-06,
+      "loss": 1.8006,
+      "step": 18647000
+    },
+    {
+      "epoch": 92.39,
+      "learning_rate": 3.82344198213463e-06,
+      "loss": 1.8078,
+      "step": 18647500
+    },
+    {
+      "epoch": 92.39,
+      "learning_rate": 3.822203395708546e-06,
+      "loss": 1.7972,
+      "step": 18648000
+    },
+    {
+      "epoch": 92.39,
+      "learning_rate": 3.820964809282463e-06,
+      "loss": 1.7782,
+      "step": 18648500
+    },
+    {
+      "epoch": 92.39,
+      "learning_rate": 3.819726222856379e-06,
+      "loss": 1.7916,
+      "step": 18649000
+    },
+    {
+      "epoch": 92.4,
+      "learning_rate": 3.818487636430295e-06,
+      "loss": 1.8131,
+      "step": 18649500
+    },
+    {
+      "epoch": 92.4,
+      "learning_rate": 3.817249050004212e-06,
+      "loss": 1.8025,
+      "step": 18650000
+    },
+    {
+      "epoch": 92.4,
+      "learning_rate": 3.816010463578128e-06,
+      "loss": 1.8131,
+      "step": 18650500
+    },
+    {
+      "epoch": 92.4,
+      "learning_rate": 3.814771877152045e-06,
+      "loss": 1.8002,
+      "step": 18651000
+    },
+    {
+      "epoch": 92.41,
+      "learning_rate": 3.8135357678988127e-06,
+      "loss": 1.7952,
+      "step": 18651500
+    },
+    {
+      "epoch": 92.41,
+      "learning_rate": 3.8122971814727292e-06,
+      "loss": 1.7876,
+      "step": 18652000
+    },
+    {
+      "epoch": 92.41,
+      "learning_rate": 3.8110585950466453e-06,
+      "loss": 1.7738,
+      "step": 18652500
+    },
+    {
+      "epoch": 92.41,
+      "learning_rate": 3.809820008620562e-06,
+      "loss": 1.7869,
+      "step": 18653000
+    },
+    {
+      "epoch": 92.42,
+      "learning_rate": 3.8085838993673302e-06,
+      "loss": 1.801,
+      "step": 18653500
+    },
+    {
+      "epoch": 92.42,
+      "learning_rate": 3.8073453129412467e-06,
+      "loss": 1.7885,
+      "step": 18654000
+    },
+    {
+      "epoch": 92.42,
+      "learning_rate": 3.806106726515163e-06,
+      "loss": 1.7944,
+      "step": 18654500
+    },
+    {
+      "epoch": 92.42,
+      "learning_rate": 3.8048681400890797e-06,
+      "loss": 1.7974,
+      "step": 18655000
+    },
+    {
+      "epoch": 92.43,
+      "learning_rate": 3.8036295536629954e-06,
+      "loss": 1.7943,
+      "step": 18655500
+    },
+    {
+      "epoch": 92.43,
+      "learning_rate": 3.8023909672369123e-06,
+      "loss": 1.7931,
+      "step": 18656000
+    },
+    {
+      "epoch": 92.43,
+      "learning_rate": 3.8011523808108284e-06,
+      "loss": 1.8046,
+      "step": 18656500
+    },
+    {
+      "epoch": 92.43,
+      "learning_rate": 3.799913794384745e-06,
+      "loss": 1.7738,
+      "step": 18657000
+    },
+    {
+      "epoch": 92.44,
+      "learning_rate": 3.7986776851315133e-06,
+      "loss": 1.8027,
+      "step": 18657500
+    },
+    {
+      "epoch": 92.44,
+      "learning_rate": 3.79743909870543e-06,
+      "loss": 1.8111,
+      "step": 18658000
+    },
+    {
+      "epoch": 92.44,
+      "learning_rate": 3.796202989452198e-06,
+      "loss": 1.7983,
+      "step": 18658500
+    },
+    {
+      "epoch": 92.44,
+      "learning_rate": 3.7949644030261147e-06,
+      "loss": 1.8013,
+      "step": 18659000
+    },
+    {
+      "epoch": 92.45,
+      "learning_rate": 3.793728293772883e-06,
+      "loss": 1.7954,
+      "step": 18659500
+    },
+    {
+      "epoch": 92.45,
+      "learning_rate": 3.7924897073467996e-06,
+      "loss": 1.8054,
+      "step": 18660000
+    },
+    {
+      "epoch": 92.45,
+      "learning_rate": 3.7912511209207157e-06,
+      "loss": 1.8,
+      "step": 18660500
+    },
+    {
+      "epoch": 92.45,
+      "learning_rate": 3.790012534494632e-06,
+      "loss": 1.7803,
+      "step": 18661000
+    },
+    {
+      "epoch": 92.46,
+      "learning_rate": 3.7887764252414005e-06,
+      "loss": 1.7881,
+      "step": 18661500
+    },
+    {
+      "epoch": 92.46,
+      "learning_rate": 3.787537838815317e-06,
+      "loss": 1.846,
+      "step": 18662000
+    },
+    {
+      "epoch": 92.46,
+      "learning_rate": 3.786299252389233e-06,
+      "loss": 1.8087,
+      "step": 18662500
+    },
+    {
+      "epoch": 92.46,
+      "learning_rate": 3.78506066596315e-06,
+      "loss": 1.7996,
+      "step": 18663000
+    },
+    {
+      "epoch": 92.47,
+      "learning_rate": 3.7838220795370657e-06,
+      "loss": 1.8094,
+      "step": 18663500
+    },
+    {
+      "epoch": 92.47,
+      "learning_rate": 3.7825834931109826e-06,
+      "loss": 1.8313,
+      "step": 18664000
+    },
+    {
+      "epoch": 92.47,
+      "learning_rate": 3.7813449066848987e-06,
+      "loss": 1.8266,
+      "step": 18664500
+    },
+    {
+      "epoch": 92.47,
+      "learning_rate": 3.7801087974316675e-06,
+      "loss": 1.8024,
+      "step": 18665000
+    },
+    {
+      "epoch": 92.48,
+      "learning_rate": 3.778872688178436e-06,
+      "loss": 1.7978,
+      "step": 18665500
+    },
+    {
+      "epoch": 92.48,
+      "learning_rate": 3.7776341017523524e-06,
+      "loss": 1.8009,
+      "step": 18666000
+    },
+    {
+      "epoch": 92.48,
+      "learning_rate": 3.7763955153262685e-06,
+      "loss": 1.8048,
+      "step": 18666500
+    },
+    {
+      "epoch": 92.48,
+      "learning_rate": 3.775156928900185e-06,
+      "loss": 1.7995,
+      "step": 18667000
+    },
+    {
+      "epoch": 92.49,
+      "learning_rate": 3.773918342474101e-06,
+      "loss": 1.7928,
+      "step": 18667500
+    },
+    {
+      "epoch": 92.49,
+      "learning_rate": 3.7726797560480176e-06,
+      "loss": 1.8094,
+      "step": 18668000
+    },
+    {
+      "epoch": 92.49,
+      "learning_rate": 3.7714411696219337e-06,
+      "loss": 1.8093,
+      "step": 18668500
+    },
+    {
+      "epoch": 92.49,
+      "learning_rate": 3.7702025831958506e-06,
+      "loss": 1.7915,
+      "step": 18669000
+    },
+    {
+      "epoch": 92.5,
+      "learning_rate": 3.768963996769767e-06,
+      "loss": 1.8067,
+      "step": 18669500
+    },
+    {
+      "epoch": 92.5,
+      "learning_rate": 3.767725410343683e-06,
+      "loss": 1.8169,
+      "step": 18670000
+    },
+    {
+      "epoch": 92.5,
+      "learning_rate": 3.766489301090452e-06,
+      "loss": 1.7854,
+      "step": 18670500
+    },
+    {
+      "epoch": 92.5,
+      "learning_rate": 3.765250714664368e-06,
+      "loss": 1.8003,
+      "step": 18671000
+    },
+    {
+      "epoch": 92.51,
+      "learning_rate": 3.7640121282382846e-06,
+      "loss": 1.807,
+      "step": 18671500
+    },
+    {
+      "epoch": 92.51,
+      "learning_rate": 3.7627735418122007e-06,
+      "loss": 1.7963,
+      "step": 18672000
+    },
+    {
+      "epoch": 92.51,
+      "learning_rate": 3.7615374325589695e-06,
+      "loss": 1.8021,
+      "step": 18672500
+    },
+    {
+      "epoch": 92.51,
+      "learning_rate": 3.7602988461328856e-06,
+      "loss": 1.8256,
+      "step": 18673000
+    },
+    {
+      "epoch": 92.51,
+      "learning_rate": 3.759060259706802e-06,
+      "loss": 1.7738,
+      "step": 18673500
+    },
+    {
+      "epoch": 92.52,
+      "learning_rate": 3.757821673280718e-06,
+      "loss": 1.8055,
+      "step": 18674000
+    },
+    {
+      "epoch": 92.52,
+      "learning_rate": 3.756583086854635e-06,
+      "loss": 1.7977,
+      "step": 18674500
+    },
+    {
+      "epoch": 92.52,
+      "learning_rate": 3.7553445004285507e-06,
+      "loss": 1.7798,
+      "step": 18675000
+    },
+    {
+      "epoch": 92.52,
+      "learning_rate": 3.7541059140024677e-06,
+      "loss": 1.7864,
+      "step": 18675500
+    },
+    {
+      "epoch": 92.53,
+      "learning_rate": 3.7528673275763837e-06,
+      "loss": 1.8009,
+      "step": 18676000
+    },
+    {
+      "epoch": 92.53,
+      "learning_rate": 3.7516287411503002e-06,
+      "loss": 1.7966,
+      "step": 18676500
+    },
+    {
+      "epoch": 92.53,
+      "learning_rate": 3.7503926318970686e-06,
+      "loss": 1.8138,
+      "step": 18677000
+    },
+    {
+      "epoch": 92.53,
+      "learning_rate": 3.749154045470985e-06,
+      "loss": 1.7875,
+      "step": 18677500
+    },
+    {
+      "epoch": 92.54,
+      "learning_rate": 3.7479154590449012e-06,
+      "loss": 1.7925,
+      "step": 18678000
+    },
+    {
+      "epoch": 92.54,
+      "learning_rate": 3.7466768726188177e-06,
+      "loss": 1.7753,
+      "step": 18678500
+    },
+    {
+      "epoch": 92.54,
+      "learning_rate": 3.745438286192734e-06,
+      "loss": 1.8137,
+      "step": 18679000
+    },
+    {
+      "epoch": 92.54,
+      "learning_rate": 3.7441996997666503e-06,
+      "loss": 1.7997,
+      "step": 18679500
+    },
+    {
+      "epoch": 92.55,
+      "learning_rate": 3.7429611133405672e-06,
+      "loss": 1.7901,
+      "step": 18680000
+    },
+    {
+      "epoch": 92.55,
+      "learning_rate": 3.7417225269144833e-06,
+      "loss": 1.8008,
+      "step": 18680500
+    },
+    {
+      "epoch": 92.55,
+      "learning_rate": 3.7404839404884e-06,
+      "loss": 1.8076,
+      "step": 18681000
+    },
+    {
+      "epoch": 92.55,
+      "learning_rate": 3.739245354062316e-06,
+      "loss": 1.8097,
+      "step": 18681500
+    },
+    {
+      "epoch": 92.56,
+      "learning_rate": 3.7380067676362324e-06,
+      "loss": 1.7805,
+      "step": 18682000
+    },
+    {
+      "epoch": 92.56,
+      "learning_rate": 3.736773135555853e-06,
+      "loss": 1.7976,
+      "step": 18682500
+    },
+    {
+      "epoch": 92.56,
+      "learning_rate": 3.7355345491297696e-06,
+      "loss": 1.8022,
+      "step": 18683000
+    },
+    {
+      "epoch": 92.56,
+      "learning_rate": 3.7342959627036857e-06,
+      "loss": 1.8189,
+      "step": 18683500
+    },
+    {
+      "epoch": 92.57,
+      "learning_rate": 3.733057376277602e-06,
+      "loss": 1.8047,
+      "step": 18684000
+    },
+    {
+      "epoch": 92.57,
+      "learning_rate": 3.7318187898515183e-06,
+      "loss": 1.7727,
+      "step": 18684500
+    },
+    {
+      "epoch": 92.57,
+      "learning_rate": 3.730580203425435e-06,
+      "loss": 1.8078,
+      "step": 18685000
+    },
+    {
+      "epoch": 92.57,
+      "learning_rate": 3.729341616999351e-06,
+      "loss": 1.7802,
+      "step": 18685500
+    },
+    {
+      "epoch": 92.58,
+      "learning_rate": 3.7281030305732678e-06,
+      "loss": 1.7937,
+      "step": 18686000
+    },
+    {
+      "epoch": 92.58,
+      "learning_rate": 3.726864444147184e-06,
+      "loss": 1.8274,
+      "step": 18686500
+    },
+    {
+      "epoch": 92.58,
+      "learning_rate": 3.7256283348939527e-06,
+      "loss": 1.7949,
+      "step": 18687000
+    },
+    {
+      "epoch": 92.58,
+      "learning_rate": 3.7243897484678688e-06,
+      "loss": 1.7974,
+      "step": 18687500
+    },
+    {
+      "epoch": 92.59,
+      "learning_rate": 3.7231511620417853e-06,
+      "loss": 1.8059,
+      "step": 18688000
+    },
+    {
+      "epoch": 92.59,
+      "learning_rate": 3.7219125756157013e-06,
+      "loss": 1.8018,
+      "step": 18688500
+    },
+    {
+      "epoch": 92.59,
+      "learning_rate": 3.720673989189618e-06,
+      "loss": 1.7764,
+      "step": 18689000
+    },
+    {
+      "epoch": 92.59,
+      "learning_rate": 3.7194378799363862e-06,
+      "loss": 1.7763,
+      "step": 18689500
+    },
+    {
+      "epoch": 92.6,
+      "learning_rate": 3.7181992935103027e-06,
+      "loss": 1.7828,
+      "step": 18690000
+    },
+    {
+      "epoch": 92.6,
+      "learning_rate": 3.716960707084219e-06,
+      "loss": 1.8032,
+      "step": 18690500
+    },
+    {
+      "epoch": 92.6,
+      "learning_rate": 3.7157221206581353e-06,
+      "loss": 1.7766,
+      "step": 18691000
+    },
+    {
+      "epoch": 92.6,
+      "learning_rate": 3.7144835342320514e-06,
+      "loss": 1.8173,
+      "step": 18691500
+    },
+    {
+      "epoch": 92.61,
+      "learning_rate": 3.7132449478059683e-06,
+      "loss": 1.8019,
+      "step": 18692000
+    },
+    {
+      "epoch": 92.61,
+      "learning_rate": 3.712006361379885e-06,
+      "loss": 1.8074,
+      "step": 18692500
+    },
+    {
+      "epoch": 92.61,
+      "learning_rate": 3.710767774953801e-06,
+      "loss": 1.7749,
+      "step": 18693000
+    },
+    {
+      "epoch": 92.61,
+      "learning_rate": 3.7095291885277174e-06,
+      "loss": 1.788,
+      "step": 18693500
+    },
+    {
+      "epoch": 92.62,
+      "learning_rate": 3.7082906021016335e-06,
+      "loss": 1.8198,
+      "step": 18694000
+    },
+    {
+      "epoch": 92.62,
+      "learning_rate": 3.7070544928484023e-06,
+      "loss": 1.7973,
+      "step": 18694500
+    },
+    {
+      "epoch": 92.62,
+      "learning_rate": 3.7058159064223184e-06,
+      "loss": 1.8137,
+      "step": 18695000
+    },
+    {
+      "epoch": 92.62,
+      "learning_rate": 3.704577319996235e-06,
+      "loss": 1.8035,
+      "step": 18695500
+    },
+    {
+      "epoch": 92.63,
+      "learning_rate": 3.703338733570151e-06,
+      "loss": 1.8089,
+      "step": 18696000
+    },
+    {
+      "epoch": 92.63,
+      "learning_rate": 3.702100147144068e-06,
+      "loss": 1.7845,
+      "step": 18696500
+    },
+    {
+      "epoch": 92.63,
+      "learning_rate": 3.7008615607179836e-06,
+      "loss": 1.8107,
+      "step": 18697000
+    },
+    {
+      "epoch": 92.63,
+      "learning_rate": 3.6996229742919005e-06,
+      "loss": 1.7973,
+      "step": 18697500
+    },
+    {
+      "epoch": 92.64,
+      "learning_rate": 3.6983893422115208e-06,
+      "loss": 1.7956,
+      "step": 18698000
+    },
+    {
+      "epoch": 92.64,
+      "learning_rate": 3.6971507557854377e-06,
+      "loss": 1.7992,
+      "step": 18698500
+    },
+    {
+      "epoch": 92.64,
+      "learning_rate": 3.6959121693593538e-06,
+      "loss": 1.7948,
+      "step": 18699000
+    },
+    {
+      "epoch": 92.64,
+      "learning_rate": 3.6946760601061226e-06,
+      "loss": 1.7959,
+      "step": 18699500
+    },
+    {
+      "epoch": 92.65,
+      "learning_rate": 3.6934374736800387e-06,
+      "loss": 1.7861,
+      "step": 18700000
+    },
+    {
+      "epoch": 92.65,
+      "learning_rate": 3.692198887253955e-06,
+      "loss": 1.804,
+      "step": 18700500
+    },
+    {
+      "epoch": 92.65,
+      "learning_rate": 3.6909603008278712e-06,
+      "loss": 1.8015,
+      "step": 18701000
+    },
+    {
+      "epoch": 92.65,
+      "learning_rate": 3.6897217144017878e-06,
+      "loss": 1.7854,
+      "step": 18701500
+    },
+    {
+      "epoch": 92.66,
+      "learning_rate": 3.688485605148556e-06,
+      "loss": 1.7849,
+      "step": 18702000
+    },
+    {
+      "epoch": 92.66,
+      "learning_rate": 3.6872470187224726e-06,
+      "loss": 1.8001,
+      "step": 18702500
+    },
+    {
+      "epoch": 92.66,
+      "learning_rate": 3.6860084322963887e-06,
+      "loss": 1.7843,
+      "step": 18703000
+    },
+    {
+      "epoch": 92.66,
+      "learning_rate": 3.6847698458703057e-06,
+      "loss": 1.7801,
+      "step": 18703500
+    },
+    {
+      "epoch": 92.67,
+      "learning_rate": 3.6835312594442213e-06,
+      "loss": 1.7778,
+      "step": 18704000
+    },
+    {
+      "epoch": 92.67,
+      "learning_rate": 3.6822926730181382e-06,
+      "loss": 1.8161,
+      "step": 18704500
+    },
+    {
+      "epoch": 92.67,
+      "learning_rate": 3.681056563764906e-06,
+      "loss": 1.8009,
+      "step": 18705000
+    },
+    {
+      "epoch": 92.67,
+      "learning_rate": 3.679817977338823e-06,
+      "loss": 1.7838,
+      "step": 18705500
+    },
+    {
+      "epoch": 92.68,
+      "learning_rate": 3.678579390912739e-06,
+      "loss": 1.7842,
+      "step": 18706000
+    },
+    {
+      "epoch": 92.68,
+      "learning_rate": 3.6773408044866557e-06,
+      "loss": 1.801,
+      "step": 18706500
+    },
+    {
+      "epoch": 92.68,
+      "learning_rate": 3.676102218060572e-06,
+      "loss": 1.7955,
+      "step": 18707000
+    },
+    {
+      "epoch": 92.68,
+      "learning_rate": 3.6748636316344883e-06,
+      "loss": 1.795,
+      "step": 18707500
+    },
+    {
+      "epoch": 92.69,
+      "learning_rate": 3.6736250452084052e-06,
+      "loss": 1.7963,
+      "step": 18708000
+    },
+    {
+      "epoch": 92.69,
+      "learning_rate": 3.672386458782321e-06,
+      "loss": 1.8035,
+      "step": 18708500
+    },
+    {
+      "epoch": 92.69,
+      "learning_rate": 3.671147872356238e-06,
+      "loss": 1.7925,
+      "step": 18709000
+    },
+    {
+      "epoch": 92.69,
+      "learning_rate": 3.669909285930154e-06,
+      "loss": 1.8016,
+      "step": 18709500
+    },
+    {
+      "epoch": 92.7,
+      "learning_rate": 3.6686706995040704e-06,
+      "loss": 1.7748,
+      "step": 18710000
+    },
+    {
+      "epoch": 92.7,
+      "learning_rate": 3.6674321130779865e-06,
+      "loss": 1.7911,
+      "step": 18710500
+    },
+    {
+      "epoch": 92.7,
+      "learning_rate": 3.666193526651903e-06,
+      "loss": 1.803,
+      "step": 18711000
+    },
+    {
+      "epoch": 92.7,
+      "learning_rate": 3.6649574173986714e-06,
+      "loss": 1.795,
+      "step": 18711500
+    },
+    {
+      "epoch": 92.71,
+      "learning_rate": 3.663718830972588e-06,
+      "loss": 1.8069,
+      "step": 18712000
+    },
+    {
+      "epoch": 92.71,
+      "learning_rate": 3.6624827217193563e-06,
+      "loss": 1.7909,
+      "step": 18712500
+    },
+    {
+      "epoch": 92.71,
+      "learning_rate": 3.6612441352932728e-06,
+      "loss": 1.8073,
+      "step": 18713000
+    },
+    {
+      "epoch": 92.71,
+      "learning_rate": 3.660005548867189e-06,
+      "loss": 1.803,
+      "step": 18713500
+    },
+    {
+      "epoch": 92.72,
+      "learning_rate": 3.6587669624411054e-06,
+      "loss": 1.8134,
+      "step": 18714000
+    },
+    {
+      "epoch": 92.72,
+      "learning_rate": 3.6575283760150214e-06,
+      "loss": 1.7902,
+      "step": 18714500
+    },
+    {
+      "epoch": 92.72,
+      "learning_rate": 3.6562922667617907e-06,
+      "loss": 1.8082,
+      "step": 18715000
+    },
+    {
+      "epoch": 92.72,
+      "learning_rate": 3.6550561575085586e-06,
+      "loss": 1.8025,
+      "step": 18715500
+    },
+    {
+      "epoch": 92.73,
+      "learning_rate": 3.6538175710824756e-06,
+      "loss": 1.7943,
+      "step": 18716000
+    },
+    {
+      "epoch": 92.73,
+      "learning_rate": 3.6525814618292435e-06,
+      "loss": 1.7941,
+      "step": 18716500
+    },
+    {
+      "epoch": 92.73,
+      "learning_rate": 3.6513428754031604e-06,
+      "loss": 1.7861,
+      "step": 18717000
+    },
+    {
+      "epoch": 92.73,
+      "learning_rate": 3.6501042889770765e-06,
+      "loss": 1.7814,
+      "step": 18717500
+    },
+    {
+      "epoch": 92.74,
+      "learning_rate": 3.648865702550993e-06,
+      "loss": 1.7979,
+      "step": 18718000
+    },
+    {
+      "epoch": 92.74,
+      "learning_rate": 3.647627116124909e-06,
+      "loss": 1.7945,
+      "step": 18718500
+    },
+    {
+      "epoch": 92.74,
+      "learning_rate": 3.6463885296988256e-06,
+      "loss": 1.7903,
+      "step": 18719000
+    },
+    {
+      "epoch": 92.74,
+      "learning_rate": 3.6451499432727417e-06,
+      "loss": 1.794,
+      "step": 18719500
+    },
+    {
+      "epoch": 92.75,
+      "learning_rate": 3.643911356846658e-06,
+      "loss": 1.7847,
+      "step": 18720000
+    },
+    {
+      "epoch": 92.75,
+      "learning_rate": 3.6426727704205743e-06,
+      "loss": 1.7952,
+      "step": 18720500
+    },
+    {
+      "epoch": 92.75,
+      "learning_rate": 3.641434183994491e-06,
+      "loss": 1.7908,
+      "step": 18721000
+    },
+    {
+      "epoch": 92.75,
+      "learning_rate": 3.640195597568407e-06,
+      "loss": 1.7752,
+      "step": 18721500
+    },
+    {
+      "epoch": 92.76,
+      "learning_rate": 3.638957011142324e-06,
+      "loss": 1.7919,
+      "step": 18722000
+    },
+    {
+      "epoch": 92.76,
+      "learning_rate": 3.6377184247162403e-06,
+      "loss": 1.8039,
+      "step": 18722500
+    },
+    {
+      "epoch": 92.76,
+      "learning_rate": 3.6364798382901564e-06,
+      "loss": 1.8153,
+      "step": 18723000
+    },
+    {
+      "epoch": 92.76,
+      "learning_rate": 3.635241251864073e-06,
+      "loss": 1.802,
+      "step": 18723500
+    },
+    {
+      "epoch": 92.77,
+      "learning_rate": 3.634002665437989e-06,
+      "loss": 1.7867,
+      "step": 18724000
+    },
+    {
+      "epoch": 92.77,
+      "learning_rate": 3.6327640790119055e-06,
+      "loss": 1.8005,
+      "step": 18724500
+    },
+    {
+      "epoch": 92.77,
+      "learning_rate": 3.6315254925858216e-06,
+      "loss": 1.7985,
+      "step": 18725000
+    },
+    {
+      "epoch": 92.77,
+      "learning_rate": 3.6302869061597385e-06,
+      "loss": 1.7919,
+      "step": 18725500
+    },
+    {
+      "epoch": 92.78,
+      "learning_rate": 3.629048319733654e-06,
+      "loss": 1.7924,
+      "step": 18726000
+    },
+    {
+      "epoch": 92.78,
+      "learning_rate": 3.627809733307571e-06,
+      "loss": 1.7747,
+      "step": 18726500
+    },
+    {
+      "epoch": 92.78,
+      "learning_rate": 3.626573624054339e-06,
+      "loss": 1.8057,
+      "step": 18727000
+    },
+    {
+      "epoch": 92.78,
+      "learning_rate": 3.625335037628256e-06,
+      "loss": 1.8206,
+      "step": 18727500
+    },
+    {
+      "epoch": 92.78,
+      "learning_rate": 3.6241014055478762e-06,
+      "loss": 1.7886,
+      "step": 18728000
+    },
+    {
+      "epoch": 92.79,
+      "learning_rate": 3.622862819121793e-06,
+      "loss": 1.8036,
+      "step": 18728500
+    },
+    {
+      "epoch": 92.79,
+      "learning_rate": 3.6216242326957092e-06,
+      "loss": 1.7855,
+      "step": 18729000
+    },
+    {
+      "epoch": 92.79,
+      "learning_rate": 3.6203856462696257e-06,
+      "loss": 1.8131,
+      "step": 18729500
+    },
+    {
+      "epoch": 92.79,
+      "learning_rate": 3.619147059843542e-06,
+      "loss": 1.7968,
+      "step": 18730000
+    },
+    {
+      "epoch": 92.8,
+      "learning_rate": 3.6179084734174583e-06,
+      "loss": 1.8049,
+      "step": 18730500
+    },
+    {
+      "epoch": 92.8,
+      "learning_rate": 3.6166723641642267e-06,
+      "loss": 1.8029,
+      "step": 18731000
+    },
+    {
+      "epoch": 92.8,
+      "learning_rate": 3.6154337777381432e-06,
+      "loss": 1.8157,
+      "step": 18731500
+    },
+    {
+      "epoch": 92.8,
+      "learning_rate": 3.6141976684849116e-06,
+      "loss": 1.8092,
+      "step": 18732000
+    },
+    {
+      "epoch": 92.81,
+      "learning_rate": 3.612959082058828e-06,
+      "loss": 1.7927,
+      "step": 18732500
+    },
+    {
+      "epoch": 92.81,
+      "learning_rate": 3.611720495632744e-06,
+      "loss": 1.8063,
+      "step": 18733000
+    },
+    {
+      "epoch": 92.81,
+      "learning_rate": 3.610484386379513e-06,
+      "loss": 1.8148,
+      "step": 18733500
+    },
+    {
+      "epoch": 92.81,
+      "learning_rate": 3.609245799953429e-06,
+      "loss": 1.8215,
+      "step": 18734000
+    },
+    {
+      "epoch": 92.82,
+      "learning_rate": 3.6080096907001983e-06,
+      "loss": 1.8029,
+      "step": 18734500
+    },
+    {
+      "epoch": 92.82,
+      "learning_rate": 3.606771104274114e-06,
+      "loss": 1.7792,
+      "step": 18735000
+    },
+    {
+      "epoch": 92.82,
+      "learning_rate": 3.605532517848031e-06,
+      "loss": 1.794,
+      "step": 18735500
+    },
+    {
+      "epoch": 92.82,
+      "learning_rate": 3.6042939314219466e-06,
+      "loss": 1.8045,
+      "step": 18736000
+    },
+    {
+      "epoch": 92.83,
+      "learning_rate": 3.6030553449958635e-06,
+      "loss": 1.7848,
+      "step": 18736500
+    },
+    {
+      "epoch": 92.83,
+      "learning_rate": 3.6018167585697796e-06,
+      "loss": 1.769,
+      "step": 18737000
+    },
+    {
+      "epoch": 92.83,
+      "learning_rate": 3.600578172143696e-06,
+      "loss": 1.7874,
+      "step": 18737500
+    },
+    {
+      "epoch": 92.83,
+      "learning_rate": 3.599339585717612e-06,
+      "loss": 1.7835,
+      "step": 18738000
+    },
+    {
+      "epoch": 92.84,
+      "learning_rate": 3.5981009992915287e-06,
+      "loss": 1.7826,
+      "step": 18738500
+    },
+    {
+      "epoch": 92.84,
+      "learning_rate": 3.5968624128654447e-06,
+      "loss": 1.7916,
+      "step": 18739000
+    },
+    {
+      "epoch": 92.84,
+      "learning_rate": 3.5956238264393612e-06,
+      "loss": 1.8233,
+      "step": 18739500
+    },
+    {
+      "epoch": 92.84,
+      "learning_rate": 3.594385240013278e-06,
+      "loss": 1.8259,
+      "step": 18740000
+    },
+    {
+      "epoch": 92.85,
+      "learning_rate": 3.5931466535871943e-06,
+      "loss": 1.7696,
+      "step": 18740500
+    },
+    {
+      "epoch": 92.85,
+      "learning_rate": 3.5919080671611108e-06,
+      "loss": 1.7935,
+      "step": 18741000
+    },
+    {
+      "epoch": 92.85,
+      "learning_rate": 3.590669480735027e-06,
+      "loss": 1.8018,
+      "step": 18741500
+    },
+    {
+      "epoch": 92.85,
+      "learning_rate": 3.5894308943089433e-06,
+      "loss": 1.7933,
+      "step": 18742000
+    },
+    {
+      "epoch": 92.86,
+      "learning_rate": 3.5881923078828594e-06,
+      "loss": 1.7779,
+      "step": 18742500
+    },
+    {
+      "epoch": 92.86,
+      "learning_rate": 3.586953721456776e-06,
+      "loss": 1.7873,
+      "step": 18743000
+    },
+    {
+      "epoch": 92.86,
+      "learning_rate": 3.585715135030692e-06,
+      "loss": 1.7958,
+      "step": 18743500
+    },
+    {
+      "epoch": 92.86,
+      "learning_rate": 3.584476548604609e-06,
+      "loss": 1.8019,
+      "step": 18744000
+    },
+    {
+      "epoch": 92.87,
+      "learning_rate": 3.583240439351377e-06,
+      "loss": 1.8079,
+      "step": 18744500
+    },
+    {
+      "epoch": 92.87,
+      "learning_rate": 3.582004330098146e-06,
+      "loss": 1.8157,
+      "step": 18745000
+    },
+    {
+      "epoch": 92.87,
+      "learning_rate": 3.580765743672062e-06,
+      "loss": 1.8049,
+      "step": 18745500
+    },
+    {
+      "epoch": 92.87,
+      "learning_rate": 3.5795271572459787e-06,
+      "loss": 1.8048,
+      "step": 18746000
+    },
+    {
+      "epoch": 92.88,
+      "learning_rate": 3.578288570819895e-06,
+      "loss": 1.8246,
+      "step": 18746500
+    },
+    {
+      "epoch": 92.88,
+      "learning_rate": 3.5770499843938113e-06,
+      "loss": 1.7996,
+      "step": 18747000
+    },
+    {
+      "epoch": 92.88,
+      "learning_rate": 3.5758113979677274e-06,
+      "loss": 1.8273,
+      "step": 18747500
+    },
+    {
+      "epoch": 92.88,
+      "learning_rate": 3.574572811541644e-06,
+      "loss": 1.799,
+      "step": 18748000
+    },
+    {
+      "epoch": 92.89,
+      "learning_rate": 3.573334225115561e-06,
+      "loss": 1.7968,
+      "step": 18748500
+    },
+    {
+      "epoch": 92.89,
+      "learning_rate": 3.5720981158623288e-06,
+      "loss": 1.8109,
+      "step": 18749000
+    },
+    {
+      "epoch": 92.89,
+      "learning_rate": 3.570859529436245e-06,
+      "loss": 1.7984,
+      "step": 18749500
+    },
+    {
+      "epoch": 92.89,
+      "learning_rate": 3.5696234201830137e-06,
+      "loss": 1.7914,
+      "step": 18750000
+    },
+    {
+      "epoch": 92.9,
+      "learning_rate": 3.5683848337569298e-06,
+      "loss": 1.7895,
+      "step": 18750500
+    },
+    {
+      "epoch": 92.9,
+      "learning_rate": 3.5671462473308463e-06,
+      "loss": 1.7943,
+      "step": 18751000
+    },
+    {
+      "epoch": 92.9,
+      "learning_rate": 3.5659076609047623e-06,
+      "loss": 1.7716,
+      "step": 18751500
+    },
+    {
+      "epoch": 92.9,
+      "learning_rate": 3.5646690744786793e-06,
+      "loss": 1.7891,
+      "step": 18752000
+    },
+    {
+      "epoch": 92.91,
+      "learning_rate": 3.5634304880525958e-06,
+      "loss": 1.8066,
+      "step": 18752500
+    },
+    {
+      "epoch": 92.91,
+      "learning_rate": 3.562191901626512e-06,
+      "loss": 1.8156,
+      "step": 18753000
+    },
+    {
+      "epoch": 92.91,
+      "learning_rate": 3.5609533152004284e-06,
+      "loss": 1.8296,
+      "step": 18753500
+    },
+    {
+      "epoch": 92.91,
+      "learning_rate": 3.5597147287743444e-06,
+      "loss": 1.7795,
+      "step": 18754000
+    },
+    {
+      "epoch": 92.92,
+      "learning_rate": 3.5584786195211132e-06,
+      "loss": 1.7846,
+      "step": 18754500
+    },
+    {
+      "epoch": 92.92,
+      "learning_rate": 3.5572425102678816e-06,
+      "loss": 1.8085,
+      "step": 18755000
+    },
+    {
+      "epoch": 92.92,
+      "learning_rate": 3.556003923841798e-06,
+      "loss": 1.8018,
+      "step": 18755500
+    },
+    {
+      "epoch": 92.92,
+      "learning_rate": 3.5547678145885665e-06,
+      "loss": 1.7998,
+      "step": 18756000
+    },
+    {
+      "epoch": 92.93,
+      "learning_rate": 3.553529228162483e-06,
+      "loss": 1.7875,
+      "step": 18756500
+    },
+    {
+      "epoch": 92.93,
+      "learning_rate": 3.552290641736399e-06,
+      "loss": 1.795,
+      "step": 18757000
+    },
+    {
+      "epoch": 92.93,
+      "learning_rate": 3.551052055310316e-06,
+      "loss": 1.8066,
+      "step": 18757500
+    },
+    {
+      "epoch": 92.93,
+      "learning_rate": 3.5498134688842317e-06,
+      "loss": 1.8114,
+      "step": 18758000
+    },
+    {
+      "epoch": 92.94,
+      "learning_rate": 3.5485748824581486e-06,
+      "loss": 1.7943,
+      "step": 18758500
+    },
+    {
+      "epoch": 92.94,
+      "learning_rate": 3.5473362960320647e-06,
+      "loss": 1.7935,
+      "step": 18759000
+    },
+    {
+      "epoch": 92.94,
+      "learning_rate": 3.546097709605981e-06,
+      "loss": 1.8119,
+      "step": 18759500
+    },
+    {
+      "epoch": 92.94,
+      "learning_rate": 3.5448591231798973e-06,
+      "loss": 1.7947,
+      "step": 18760000
+    },
+    {
+      "epoch": 92.95,
+      "learning_rate": 3.543623013926666e-06,
+      "loss": 1.7872,
+      "step": 18760500
+    },
+    {
+      "epoch": 92.95,
+      "learning_rate": 3.5423869046734345e-06,
+      "loss": 1.8065,
+      "step": 18761000
+    },
+    {
+      "epoch": 92.95,
+      "learning_rate": 3.541148318247351e-06,
+      "loss": 1.796,
+      "step": 18761500
+    },
+    {
+      "epoch": 92.95,
+      "learning_rate": 3.539909731821267e-06,
+      "loss": 1.7993,
+      "step": 18762000
+    },
+    {
+      "epoch": 92.96,
+      "learning_rate": 3.5386711453951836e-06,
+      "loss": 1.801,
+      "step": 18762500
+    },
+    {
+      "epoch": 92.96,
+      "learning_rate": 3.537435036141952e-06,
+      "loss": 1.8112,
+      "step": 18763000
+    },
+    {
+      "epoch": 92.96,
+      "learning_rate": 3.5361964497158685e-06,
+      "loss": 1.7721,
+      "step": 18763500
+    },
+    {
+      "epoch": 92.96,
+      "learning_rate": 3.5349578632897846e-06,
+      "loss": 1.7738,
+      "step": 18764000
+    },
+    {
+      "epoch": 92.97,
+      "learning_rate": 3.5337192768637015e-06,
+      "loss": 1.8087,
+      "step": 18764500
+    },
+    {
+      "epoch": 92.97,
+      "learning_rate": 3.532480690437617e-06,
+      "loss": 1.8099,
+      "step": 18765000
+    },
+    {
+      "epoch": 92.97,
+      "learning_rate": 3.531242104011534e-06,
+      "loss": 1.7839,
+      "step": 18765500
+    },
+    {
+      "epoch": 92.97,
+      "learning_rate": 3.53000351758545e-06,
+      "loss": 1.792,
+      "step": 18766000
+    },
+    {
+      "epoch": 92.98,
+      "learning_rate": 3.5287649311593666e-06,
+      "loss": 1.765,
+      "step": 18766500
+    },
+    {
+      "epoch": 92.98,
+      "learning_rate": 3.5275263447332827e-06,
+      "loss": 1.806,
+      "step": 18767000
+    },
+    {
+      "epoch": 92.98,
+      "learning_rate": 3.5262877583071992e-06,
+      "loss": 1.806,
+      "step": 18767500
+    },
+    {
+      "epoch": 92.98,
+      "learning_rate": 3.525049171881116e-06,
+      "loss": 1.7692,
+      "step": 18768000
+    },
+    {
+      "epoch": 92.99,
+      "learning_rate": 3.523810585455032e-06,
+      "loss": 1.8051,
+      "step": 18768500
+    },
+    {
+      "epoch": 92.99,
+      "learning_rate": 3.5225719990289487e-06,
+      "loss": 1.7909,
+      "step": 18769000
+    },
+    {
+      "epoch": 92.99,
+      "learning_rate": 3.521333412602865e-06,
+      "loss": 1.7806,
+      "step": 18769500
+    },
+    {
+      "epoch": 92.99,
+      "learning_rate": 3.5200948261767813e-06,
+      "loss": 1.7932,
+      "step": 18770000
+    },
+    {
+      "epoch": 93.0,
+      "learning_rate": 3.5188562397506974e-06,
+      "loss": 1.7938,
+      "step": 18770500
+    },
+    {
+      "epoch": 93.0,
+      "learning_rate": 3.5176201304974662e-06,
+      "loss": 1.7922,
+      "step": 18771000
+    },
+    {
+      "epoch": 93.0,
+      "eval_accuracy": 0.6856451325370596,
+      "eval_accuracy_mlm": 0.6470140033120039,
+      "eval_accuracy_nsp": 0.8679983840539067,
+      "eval_loss": 2.300736904144287,
+      "eval_runtime": 146.8864,
+      "eval_samples_per_second": 1735.756,
+      "eval_steps_per_second": 72.328,
+      "step": 18771399
+    },
+    {
+      "epoch": 93.0,
+      "learning_rate": 3.5163815440713823e-06,
+      "loss": 1.7909,
+      "step": 18771500
+    },
+    {
+      "epoch": 93.0,
+      "learning_rate": 3.515142957645299e-06,
+      "loss": 1.7971,
+      "step": 18772000
+    },
+    {
+      "epoch": 93.01,
+      "learning_rate": 3.513906848392067e-06,
+      "loss": 1.8084,
+      "step": 18772500
+    },
+    {
+      "epoch": 93.01,
+      "learning_rate": 3.5126682619659837e-06,
+      "loss": 1.7777,
+      "step": 18773000
+    },
+    {
+      "epoch": 93.01,
+      "learning_rate": 3.5114296755398998e-06,
+      "loss": 1.808,
+      "step": 18773500
+    },
+    {
+      "epoch": 93.01,
+      "learning_rate": 3.5101910891138167e-06,
+      "loss": 1.7864,
+      "step": 18774000
+    },
+    {
+      "epoch": 93.02,
+      "learning_rate": 3.5089525026877324e-06,
+      "loss": 1.8119,
+      "step": 18774500
+    },
+    {
+      "epoch": 93.02,
+      "learning_rate": 3.5077139162616493e-06,
+      "loss": 1.7687,
+      "step": 18775000
+    },
+    {
+      "epoch": 93.02,
+      "learning_rate": 3.506475329835565e-06,
+      "loss": 1.8017,
+      "step": 18775500
+    },
+    {
+      "epoch": 93.02,
+      "learning_rate": 3.505236743409482e-06,
+      "loss": 1.805,
+      "step": 18776000
+    },
+    {
+      "epoch": 93.03,
+      "learning_rate": 3.503998156983398e-06,
+      "loss": 1.8142,
+      "step": 18776500
+    },
+    {
+      "epoch": 93.03,
+      "learning_rate": 3.5027595705573145e-06,
+      "loss": 1.7926,
+      "step": 18777000
+    },
+    {
+      "epoch": 93.03,
+      "learning_rate": 3.501525938476935e-06,
+      "loss": 1.787,
+      "step": 18777500
+    },
+    {
+      "epoch": 93.03,
+      "learning_rate": 3.5002873520508517e-06,
+      "loss": 1.8028,
+      "step": 18778000
+    },
+    {
+      "epoch": 93.04,
+      "learning_rate": 3.4990487656247677e-06,
+      "loss": 1.7826,
+      "step": 18778500
+    },
+    {
+      "epoch": 93.04,
+      "learning_rate": 3.4978101791986842e-06,
+      "loss": 1.7904,
+      "step": 18779000
+    },
+    {
+      "epoch": 93.04,
+      "learning_rate": 3.4965715927726003e-06,
+      "loss": 1.8014,
+      "step": 18779500
+    },
+    {
+      "epoch": 93.04,
+      "learning_rate": 3.495333006346517e-06,
+      "loss": 1.7931,
+      "step": 18780000
+    },
+    {
+      "epoch": 93.05,
+      "learning_rate": 3.4940944199204338e-06,
+      "loss": 1.8065,
+      "step": 18780500
+    },
+    {
+      "epoch": 93.05,
+      "learning_rate": 3.4928583106672017e-06,
+      "loss": 1.7975,
+      "step": 18781000
+    },
+    {
+      "epoch": 93.05,
+      "learning_rate": 3.4916197242411187e-06,
+      "loss": 1.7969,
+      "step": 18781500
+    },
+    {
+      "epoch": 93.05,
+      "learning_rate": 3.4903811378150347e-06,
+      "loss": 1.8054,
+      "step": 18782000
+    },
+    {
+      "epoch": 93.05,
+      "learning_rate": 3.4891425513889512e-06,
+      "loss": 1.8042,
+      "step": 18782500
+    },
+    {
+      "epoch": 93.06,
+      "learning_rate": 3.4879064421357196e-06,
+      "loss": 1.8058,
+      "step": 18783000
+    },
+    {
+      "epoch": 93.06,
+      "learning_rate": 3.486667855709636e-06,
+      "loss": 1.8059,
+      "step": 18783500
+    },
+    {
+      "epoch": 93.06,
+      "learning_rate": 3.485429269283552e-06,
+      "loss": 1.7913,
+      "step": 18784000
+    },
+    {
+      "epoch": 93.06,
+      "learning_rate": 3.4841906828574687e-06,
+      "loss": 1.8087,
+      "step": 18784500
+    },
+    {
+      "epoch": 93.07,
+      "learning_rate": 3.482952096431385e-06,
+      "loss": 1.7985,
+      "step": 18785000
+    },
+    {
+      "epoch": 93.07,
+      "learning_rate": 3.4817135100053017e-06,
+      "loss": 1.785,
+      "step": 18785500
+    },
+    {
+      "epoch": 93.07,
+      "learning_rate": 3.4804749235792174e-06,
+      "loss": 1.7971,
+      "step": 18786000
+    },
+    {
+      "epoch": 93.07,
+      "learning_rate": 3.4792363371531343e-06,
+      "loss": 1.799,
+      "step": 18786500
+    },
+    {
+      "epoch": 93.08,
+      "learning_rate": 3.47799775072705e-06,
+      "loss": 1.7865,
+      "step": 18787000
+    },
+    {
+      "epoch": 93.08,
+      "learning_rate": 3.476759164300967e-06,
+      "loss": 1.7974,
+      "step": 18787500
+    },
+    {
+      "epoch": 93.08,
+      "learning_rate": 3.4755230550477353e-06,
+      "loss": 1.7878,
+      "step": 18788000
+    },
+    {
+      "epoch": 93.08,
+      "learning_rate": 3.4742844686216518e-06,
+      "loss": 1.8141,
+      "step": 18788500
+    },
+    {
+      "epoch": 93.09,
+      "learning_rate": 3.473045882195568e-06,
+      "loss": 1.7915,
+      "step": 18789000
+    },
+    {
+      "epoch": 93.09,
+      "learning_rate": 3.4718072957694844e-06,
+      "loss": 1.7877,
+      "step": 18789500
+    },
+    {
+      "epoch": 93.09,
+      "learning_rate": 3.4705711865162528e-06,
+      "loss": 1.8053,
+      "step": 18790000
+    },
+    {
+      "epoch": 93.09,
+      "learning_rate": 3.4693326000901693e-06,
+      "loss": 1.7782,
+      "step": 18790500
+    },
+    {
+      "epoch": 93.1,
+      "learning_rate": 3.4680940136640853e-06,
+      "loss": 1.8005,
+      "step": 18791000
+    },
+    {
+      "epoch": 93.1,
+      "learning_rate": 3.466855427238002e-06,
+      "loss": 1.8045,
+      "step": 18791500
+    },
+    {
+      "epoch": 93.1,
+      "learning_rate": 3.4656193179847702e-06,
+      "loss": 1.7819,
+      "step": 18792000
+    },
+    {
+      "epoch": 93.1,
+      "learning_rate": 3.4643807315586867e-06,
+      "loss": 1.7948,
+      "step": 18792500
+    },
+    {
+      "epoch": 93.11,
+      "learning_rate": 3.463142145132603e-06,
+      "loss": 1.7962,
+      "step": 18793000
+    },
+    {
+      "epoch": 93.11,
+      "learning_rate": 3.4619035587065197e-06,
+      "loss": 1.8065,
+      "step": 18793500
+    },
+    {
+      "epoch": 93.11,
+      "learning_rate": 3.4606649722804363e-06,
+      "loss": 1.7931,
+      "step": 18794000
+    },
+    {
+      "epoch": 93.11,
+      "learning_rate": 3.4594288630272046e-06,
+      "loss": 1.7973,
+      "step": 18794500
+    },
+    {
+      "epoch": 93.12,
+      "learning_rate": 3.4581902766011207e-06,
+      "loss": 1.8238,
+      "step": 18795000
+    },
+    {
+      "epoch": 93.12,
+      "learning_rate": 3.4569516901750372e-06,
+      "loss": 1.7787,
+      "step": 18795500
+    },
+    {
+      "epoch": 93.12,
+      "learning_rate": 3.4557155809218056e-06,
+      "loss": 1.7894,
+      "step": 18796000
+    },
+    {
+      "epoch": 93.12,
+      "learning_rate": 3.454476994495722e-06,
+      "loss": 1.7675,
+      "step": 18796500
+    },
+    {
+      "epoch": 93.13,
+      "learning_rate": 3.453238408069638e-06,
+      "loss": 1.7817,
+      "step": 18797000
+    },
+    {
+      "epoch": 93.13,
+      "learning_rate": 3.4519998216435547e-06,
+      "loss": 1.7877,
+      "step": 18797500
+    },
+    {
+      "epoch": 93.13,
+      "learning_rate": 3.4507612352174716e-06,
+      "loss": 1.8045,
+      "step": 18798000
+    },
+    {
+      "epoch": 93.13,
+      "learning_rate": 3.4495226487913873e-06,
+      "loss": 1.8174,
+      "step": 18798500
+    },
+    {
+      "epoch": 93.14,
+      "learning_rate": 3.4482840623653042e-06,
+      "loss": 1.788,
+      "step": 18799000
+    },
+    {
+      "epoch": 93.14,
+      "learning_rate": 3.4470454759392203e-06,
+      "loss": 1.781,
+      "step": 18799500
+    },
+    {
+      "epoch": 93.14,
+      "learning_rate": 3.445806889513137e-06,
+      "loss": 1.7755,
+      "step": 18800000
+    },
+    {
+      "epoch": 93.14,
+      "learning_rate": 3.444568303087053e-06,
+      "loss": 1.7809,
+      "step": 18800500
+    },
+    {
+      "epoch": 93.15,
+      "learning_rate": 3.4433297166609694e-06,
+      "loss": 1.7728,
+      "step": 18801000
+    },
+    {
+      "epoch": 93.15,
+      "learning_rate": 3.4420911302348855e-06,
+      "loss": 1.7942,
+      "step": 18801500
+    },
+    {
+      "epoch": 93.15,
+      "learning_rate": 3.440852543808802e-06,
+      "loss": 1.808,
+      "step": 18802000
+    },
+    {
+      "epoch": 93.15,
+      "learning_rate": 3.439613957382718e-06,
+      "loss": 1.8147,
+      "step": 18802500
+    },
+    {
+      "epoch": 93.16,
+      "learning_rate": 3.438377848129487e-06,
+      "loss": 1.79,
+      "step": 18803000
+    },
+    {
+      "epoch": 93.16,
+      "learning_rate": 3.437139261703403e-06,
+      "loss": 1.7756,
+      "step": 18803500
+    },
+    {
+      "epoch": 93.16,
+      "learning_rate": 3.43590067527732e-06,
+      "loss": 1.7875,
+      "step": 18804000
+    },
+    {
+      "epoch": 93.16,
+      "learning_rate": 3.4346620888512355e-06,
+      "loss": 1.7897,
+      "step": 18804500
+    },
+    {
+      "epoch": 93.17,
+      "learning_rate": 3.4334235024251525e-06,
+      "loss": 1.7788,
+      "step": 18805000
+    },
+    {
+      "epoch": 93.17,
+      "learning_rate": 3.432184915999069e-06,
+      "loss": 1.7844,
+      "step": 18805500
+    },
+    {
+      "epoch": 93.17,
+      "learning_rate": 3.430946329572985e-06,
+      "loss": 1.806,
+      "step": 18806000
+    },
+    {
+      "epoch": 93.17,
+      "learning_rate": 3.4297077431469015e-06,
+      "loss": 1.8158,
+      "step": 18806500
+    },
+    {
+      "epoch": 93.18,
+      "learning_rate": 3.42847163389367e-06,
+      "loss": 1.776,
+      "step": 18807000
+    },
+    {
+      "epoch": 93.18,
+      "learning_rate": 3.4272330474675864e-06,
+      "loss": 1.8004,
+      "step": 18807500
+    },
+    {
+      "epoch": 93.18,
+      "learning_rate": 3.4259944610415025e-06,
+      "loss": 1.8022,
+      "step": 18808000
+    },
+    {
+      "epoch": 93.18,
+      "learning_rate": 3.4247558746154194e-06,
+      "loss": 1.769,
+      "step": 18808500
+    },
+    {
+      "epoch": 93.19,
+      "learning_rate": 3.4235197653621874e-06,
+      "loss": 1.7843,
+      "step": 18809000
+    },
+    {
+      "epoch": 93.19,
+      "learning_rate": 3.422283656108956e-06,
+      "loss": 1.7783,
+      "step": 18809500
+    },
+    {
+      "epoch": 93.19,
+      "learning_rate": 3.4210450696828723e-06,
+      "loss": 1.7873,
+      "step": 18810000
+    },
+    {
+      "epoch": 93.19,
+      "learning_rate": 3.4198064832567892e-06,
+      "loss": 1.8051,
+      "step": 18810500
+    },
+    {
+      "epoch": 93.2,
+      "learning_rate": 3.4185678968307053e-06,
+      "loss": 1.7925,
+      "step": 18811000
+    },
+    {
+      "epoch": 93.2,
+      "learning_rate": 3.417329310404622e-06,
+      "loss": 1.7948,
+      "step": 18811500
+    },
+    {
+      "epoch": 93.2,
+      "learning_rate": 3.416090723978538e-06,
+      "loss": 1.7905,
+      "step": 18812000
+    },
+    {
+      "epoch": 93.2,
+      "learning_rate": 3.4148521375524544e-06,
+      "loss": 1.7954,
+      "step": 18812500
+    },
+    {
+      "epoch": 93.21,
+      "learning_rate": 3.4136135511263705e-06,
+      "loss": 1.7696,
+      "step": 18813000
+    },
+    {
+      "epoch": 93.21,
+      "learning_rate": 3.412374964700287e-06,
+      "loss": 1.8064,
+      "step": 18813500
+    },
+    {
+      "epoch": 93.21,
+      "learning_rate": 3.4111388554470554e-06,
+      "loss": 1.7944,
+      "step": 18814000
+    },
+    {
+      "epoch": 93.21,
+      "learning_rate": 3.409900269020972e-06,
+      "loss": 1.7974,
+      "step": 18814500
+    },
+    {
+      "epoch": 93.22,
+      "learning_rate": 3.408661682594888e-06,
+      "loss": 1.811,
+      "step": 18815000
+    },
+    {
+      "epoch": 93.22,
+      "learning_rate": 3.407423096168805e-06,
+      "loss": 1.7815,
+      "step": 18815500
+    },
+    {
+      "epoch": 93.22,
+      "learning_rate": 3.4061845097427205e-06,
+      "loss": 1.7933,
+      "step": 18816000
+    },
+    {
+      "epoch": 93.22,
+      "learning_rate": 3.4049459233166375e-06,
+      "loss": 1.7976,
+      "step": 18816500
+    },
+    {
+      "epoch": 93.23,
+      "learning_rate": 3.4037073368905536e-06,
+      "loss": 1.7908,
+      "step": 18817000
+    },
+    {
+      "epoch": 93.23,
+      "learning_rate": 3.40246875046447e-06,
+      "loss": 1.7914,
+      "step": 18817500
+    },
+    {
+      "epoch": 93.23,
+      "learning_rate": 3.4012301640383866e-06,
+      "loss": 1.7888,
+      "step": 18818000
+    },
+    {
+      "epoch": 93.23,
+      "learning_rate": 3.3999915776123026e-06,
+      "loss": 1.7645,
+      "step": 18818500
+    },
+    {
+      "epoch": 93.24,
+      "learning_rate": 3.3987554683590715e-06,
+      "loss": 1.7788,
+      "step": 18819000
+    },
+    {
+      "epoch": 93.24,
+      "learning_rate": 3.3975168819329875e-06,
+      "loss": 1.7963,
+      "step": 18819500
+    },
+    {
+      "epoch": 93.24,
+      "learning_rate": 3.396280772679756e-06,
+      "loss": 1.7745,
+      "step": 18820000
+    },
+    {
+      "epoch": 93.24,
+      "learning_rate": 3.3950446634265247e-06,
+      "loss": 1.7872,
+      "step": 18820500
+    },
+    {
+      "epoch": 93.25,
+      "learning_rate": 3.393806077000441e-06,
+      "loss": 1.7889,
+      "step": 18821000
+    },
+    {
+      "epoch": 93.25,
+      "learning_rate": 3.3925674905743573e-06,
+      "loss": 1.7835,
+      "step": 18821500
+    },
+    {
+      "epoch": 93.25,
+      "learning_rate": 3.3913289041482734e-06,
+      "loss": 1.7679,
+      "step": 18822000
+    },
+    {
+      "epoch": 93.25,
+      "learning_rate": 3.3900903177221903e-06,
+      "loss": 1.8077,
+      "step": 18822500
+    },
+    {
+      "epoch": 93.26,
+      "learning_rate": 3.3888542084689583e-06,
+      "loss": 1.7918,
+      "step": 18823000
+    },
+    {
+      "epoch": 93.26,
+      "learning_rate": 3.3876156220428752e-06,
+      "loss": 1.8009,
+      "step": 18823500
+    },
+    {
+      "epoch": 93.26,
+      "learning_rate": 3.386379512789643e-06,
+      "loss": 1.7837,
+      "step": 18824000
+    },
+    {
+      "epoch": 93.26,
+      "learning_rate": 3.38514092636356e-06,
+      "loss": 1.8044,
+      "step": 18824500
+    },
+    {
+      "epoch": 93.27,
+      "learning_rate": 3.383902339937476e-06,
+      "loss": 1.7845,
+      "step": 18825000
+    },
+    {
+      "epoch": 93.27,
+      "learning_rate": 3.3826637535113927e-06,
+      "loss": 1.7732,
+      "step": 18825500
+    },
+    {
+      "epoch": 93.27,
+      "learning_rate": 3.381425167085309e-06,
+      "loss": 1.7758,
+      "step": 18826000
+    },
+    {
+      "epoch": 93.27,
+      "learning_rate": 3.3801865806592253e-06,
+      "loss": 1.7686,
+      "step": 18826500
+    },
+    {
+      "epoch": 93.28,
+      "learning_rate": 3.378947994233142e-06,
+      "loss": 1.7982,
+      "step": 18827000
+    },
+    {
+      "epoch": 93.28,
+      "learning_rate": 3.377709407807058e-06,
+      "loss": 1.8107,
+      "step": 18827500
+    },
+    {
+      "epoch": 93.28,
+      "learning_rate": 3.376470821380975e-06,
+      "loss": 1.7992,
+      "step": 18828000
+    },
+    {
+      "epoch": 93.28,
+      "learning_rate": 3.3752322349548904e-06,
+      "loss": 1.8088,
+      "step": 18828500
+    },
+    {
+      "epoch": 93.29,
+      "learning_rate": 3.3739936485288074e-06,
+      "loss": 1.795,
+      "step": 18829000
+    },
+    {
+      "epoch": 93.29,
+      "learning_rate": 3.3727550621027235e-06,
+      "loss": 1.7879,
+      "step": 18829500
+    },
+    {
+      "epoch": 93.29,
+      "learning_rate": 3.37151647567664e-06,
+      "loss": 1.7848,
+      "step": 18830000
+    },
+    {
+      "epoch": 93.29,
+      "learning_rate": 3.370277889250556e-06,
+      "loss": 1.7911,
+      "step": 18830500
+    },
+    {
+      "epoch": 93.3,
+      "learning_rate": 3.369041779997325e-06,
+      "loss": 1.7904,
+      "step": 18831000
+    },
+    {
+      "epoch": 93.3,
+      "learning_rate": 3.367803193571241e-06,
+      "loss": 1.7744,
+      "step": 18831500
+    },
+    {
+      "epoch": 93.3,
+      "learning_rate": 3.3665670843180097e-06,
+      "loss": 1.7686,
+      "step": 18832000
+    },
+    {
+      "epoch": 93.3,
+      "learning_rate": 3.365328497891926e-06,
+      "loss": 1.7784,
+      "step": 18832500
+    },
+    {
+      "epoch": 93.31,
+      "learning_rate": 3.3640899114658423e-06,
+      "loss": 1.7707,
+      "step": 18833000
+    },
+    {
+      "epoch": 93.31,
+      "learning_rate": 3.3628513250397584e-06,
+      "loss": 1.7967,
+      "step": 18833500
+    },
+    {
+      "epoch": 93.31,
+      "learning_rate": 3.3616127386136753e-06,
+      "loss": 1.8034,
+      "step": 18834000
+    },
+    {
+      "epoch": 93.31,
+      "learning_rate": 3.360374152187591e-06,
+      "loss": 1.8126,
+      "step": 18834500
+    },
+    {
+      "epoch": 93.32,
+      "learning_rate": 3.359135565761508e-06,
+      "loss": 1.783,
+      "step": 18835000
+    },
+    {
+      "epoch": 93.32,
+      "learning_rate": 3.3578969793354244e-06,
+      "loss": 1.8016,
+      "step": 18835500
+    },
+    {
+      "epoch": 93.32,
+      "learning_rate": 3.3566583929093405e-06,
+      "loss": 1.7832,
+      "step": 18836000
+    },
+    {
+      "epoch": 93.32,
+      "learning_rate": 3.3554222836561093e-06,
+      "loss": 1.7859,
+      "step": 18836500
+    },
+    {
+      "epoch": 93.33,
+      "learning_rate": 3.3541836972300254e-06,
+      "loss": 1.7931,
+      "step": 18837000
+    },
+    {
+      "epoch": 93.33,
+      "learning_rate": 3.352945110803942e-06,
+      "loss": 1.7688,
+      "step": 18837500
+    },
+    {
+      "epoch": 93.33,
+      "learning_rate": 3.351706524377858e-06,
+      "loss": 1.7864,
+      "step": 18838000
+    },
+    {
+      "epoch": 93.33,
+      "learning_rate": 3.350467937951775e-06,
+      "loss": 1.8019,
+      "step": 18838500
+    },
+    {
+      "epoch": 93.33,
+      "learning_rate": 3.3492293515256906e-06,
+      "loss": 1.8103,
+      "step": 18839000
+    },
+    {
+      "epoch": 93.34,
+      "learning_rate": 3.34799324227246e-06,
+      "loss": 1.7715,
+      "step": 18839500
+    },
+    {
+      "epoch": 93.34,
+      "learning_rate": 3.346754655846376e-06,
+      "loss": 1.7754,
+      "step": 18840000
+    },
+    {
+      "epoch": 93.34,
+      "learning_rate": 3.3455160694202924e-06,
+      "loss": 1.7888,
+      "step": 18840500
+    },
+    {
+      "epoch": 93.34,
+      "learning_rate": 3.3442774829942085e-06,
+      "loss": 1.8043,
+      "step": 18841000
+    },
+    {
+      "epoch": 93.35,
+      "learning_rate": 3.343038896568125e-06,
+      "loss": 1.8024,
+      "step": 18841500
+    },
+    {
+      "epoch": 93.35,
+      "learning_rate": 3.341800310142041e-06,
+      "loss": 1.7954,
+      "step": 18842000
+    },
+    {
+      "epoch": 93.35,
+      "learning_rate": 3.3405617237159576e-06,
+      "loss": 1.812,
+      "step": 18842500
+    },
+    {
+      "epoch": 93.35,
+      "learning_rate": 3.3393231372898736e-06,
+      "loss": 1.7865,
+      "step": 18843000
+    },
+    {
+      "epoch": 93.36,
+      "learning_rate": 3.3380895052094948e-06,
+      "loss": 1.7886,
+      "step": 18843500
+    },
+    {
+      "epoch": 93.36,
+      "learning_rate": 3.336853395956263e-06,
+      "loss": 1.7683,
+      "step": 18844000
+    },
+    {
+      "epoch": 93.36,
+      "learning_rate": 3.3356148095301796e-06,
+      "loss": 1.8,
+      "step": 18844500
+    },
+    {
+      "epoch": 93.36,
+      "learning_rate": 3.3343762231040957e-06,
+      "loss": 1.7887,
+      "step": 18845000
+    },
+    {
+      "epoch": 93.37,
+      "learning_rate": 3.3331376366780127e-06,
+      "loss": 1.7836,
+      "step": 18845500
+    },
+    {
+      "epoch": 93.37,
+      "learning_rate": 3.3318990502519283e-06,
+      "loss": 1.7939,
+      "step": 18846000
+    },
+    {
+      "epoch": 93.37,
+      "learning_rate": 3.3306604638258452e-06,
+      "loss": 1.8068,
+      "step": 18846500
+    },
+    {
+      "epoch": 93.37,
+      "learning_rate": 3.329421877399761e-06,
+      "loss": 1.7737,
+      "step": 18847000
+    },
+    {
+      "epoch": 93.38,
+      "learning_rate": 3.32818576814653e-06,
+      "loss": 1.7752,
+      "step": 18847500
+    },
+    {
+      "epoch": 93.38,
+      "learning_rate": 3.3269471817204462e-06,
+      "loss": 1.7801,
+      "step": 18848000
+    },
+    {
+      "epoch": 93.38,
+      "learning_rate": 3.3257085952943627e-06,
+      "loss": 1.7871,
+      "step": 18848500
+    },
+    {
+      "epoch": 93.38,
+      "learning_rate": 3.324470008868279e-06,
+      "loss": 1.7793,
+      "step": 18849000
+    },
+    {
+      "epoch": 93.39,
+      "learning_rate": 3.3232314224421953e-06,
+      "loss": 1.7739,
+      "step": 18849500
+    },
+    {
+      "epoch": 93.39,
+      "learning_rate": 3.3219953131889637e-06,
+      "loss": 1.844,
+      "step": 18850000
+    },
+    {
+      "epoch": 93.39,
+      "learning_rate": 3.32075672676288e-06,
+      "loss": 1.7834,
+      "step": 18850500
+    },
+    {
+      "epoch": 93.39,
+      "learning_rate": 3.3195181403367963e-06,
+      "loss": 1.782,
+      "step": 18851000
+    },
+    {
+      "epoch": 93.4,
+      "learning_rate": 3.3182795539107128e-06,
+      "loss": 1.7855,
+      "step": 18851500
+    },
+    {
+      "epoch": 93.4,
+      "learning_rate": 3.3170409674846297e-06,
+      "loss": 1.7836,
+      "step": 18852000
+    },
+    {
+      "epoch": 93.4,
+      "learning_rate": 3.315802381058546e-06,
+      "loss": 1.8044,
+      "step": 18852500
+    },
+    {
+      "epoch": 93.4,
+      "learning_rate": 3.3145637946324623e-06,
+      "loss": 1.8091,
+      "step": 18853000
+    },
+    {
+      "epoch": 93.41,
+      "learning_rate": 3.3133252082063784e-06,
+      "loss": 1.7835,
+      "step": 18853500
+    },
+    {
+      "epoch": 93.41,
+      "learning_rate": 3.312086621780295e-06,
+      "loss": 1.7842,
+      "step": 18854000
+    },
+    {
+      "epoch": 93.41,
+      "learning_rate": 3.310848035354211e-06,
+      "loss": 1.8017,
+      "step": 18854500
+    },
+    {
+      "epoch": 93.41,
+      "learning_rate": 3.3096119261009798e-06,
+      "loss": 1.8002,
+      "step": 18855000
+    },
+    {
+      "epoch": 93.42,
+      "learning_rate": 3.308373339674896e-06,
+      "loss": 1.7813,
+      "step": 18855500
+    },
+    {
+      "epoch": 93.42,
+      "learning_rate": 3.3071347532488124e-06,
+      "loss": 1.8115,
+      "step": 18856000
+    },
+    {
+      "epoch": 93.42,
+      "learning_rate": 3.3058961668227284e-06,
+      "loss": 1.8006,
+      "step": 18856500
+    },
+    {
+      "epoch": 93.42,
+      "learning_rate": 3.3046575803966454e-06,
+      "loss": 1.8026,
+      "step": 18857000
+    },
+    {
+      "epoch": 93.43,
+      "learning_rate": 3.3034214711434133e-06,
+      "loss": 1.8005,
+      "step": 18857500
+    },
+    {
+      "epoch": 93.43,
+      "learning_rate": 3.3021828847173303e-06,
+      "loss": 1.8104,
+      "step": 18858000
+    },
+    {
+      "epoch": 93.43,
+      "learning_rate": 3.3009467754640982e-06,
+      "loss": 1.7934,
+      "step": 18858500
+    },
+    {
+      "epoch": 93.43,
+      "learning_rate": 3.299708189038015e-06,
+      "loss": 1.8208,
+      "step": 18859000
+    },
+    {
+      "epoch": 93.44,
+      "learning_rate": 3.2984696026119312e-06,
+      "loss": 1.8072,
+      "step": 18859500
+    },
+    {
+      "epoch": 93.44,
+      "learning_rate": 3.2972310161858477e-06,
+      "loss": 1.7895,
+      "step": 18860000
+    },
+    {
+      "epoch": 93.44,
+      "learning_rate": 3.295992429759764e-06,
+      "loss": 1.7906,
+      "step": 18860500
+    },
+    {
+      "epoch": 93.44,
+      "learning_rate": 3.2947538433336803e-06,
+      "loss": 1.7928,
+      "step": 18861000
+    },
+    {
+      "epoch": 93.45,
+      "learning_rate": 3.2935152569075964e-06,
+      "loss": 1.7974,
+      "step": 18861500
+    },
+    {
+      "epoch": 93.45,
+      "learning_rate": 3.292276670481513e-06,
+      "loss": 1.7892,
+      "step": 18862000
+    },
+    {
+      "epoch": 93.45,
+      "learning_rate": 3.2910405612282813e-06,
+      "loss": 1.7789,
+      "step": 18862500
+    },
+    {
+      "epoch": 93.45,
+      "learning_rate": 3.289801974802198e-06,
+      "loss": 1.8062,
+      "step": 18863000
+    },
+    {
+      "epoch": 93.46,
+      "learning_rate": 3.288563388376114e-06,
+      "loss": 1.7808,
+      "step": 18863500
+    },
+    {
+      "epoch": 93.46,
+      "learning_rate": 3.287324801950031e-06,
+      "loss": 1.7965,
+      "step": 18864000
+    },
+    {
+      "epoch": 93.46,
+      "learning_rate": 3.2860862155239473e-06,
+      "loss": 1.8078,
+      "step": 18864500
+    },
+    {
+      "epoch": 93.46,
+      "learning_rate": 3.2848501062707157e-06,
+      "loss": 1.7781,
+      "step": 18865000
+    },
+    {
+      "epoch": 93.47,
+      "learning_rate": 3.2836139970174837e-06,
+      "loss": 1.791,
+      "step": 18865500
+    },
+    {
+      "epoch": 93.47,
+      "learning_rate": 3.2823754105914006e-06,
+      "loss": 1.7911,
+      "step": 18866000
+    },
+    {
+      "epoch": 93.47,
+      "learning_rate": 3.2811368241653167e-06,
+      "loss": 1.8188,
+      "step": 18866500
+    },
+    {
+      "epoch": 93.47,
+      "learning_rate": 3.2799007149120855e-06,
+      "loss": 1.7859,
+      "step": 18867000
+    },
+    {
+      "epoch": 93.48,
+      "learning_rate": 3.2786621284860016e-06,
+      "loss": 1.791,
+      "step": 18867500
+    },
+    {
+      "epoch": 93.48,
+      "learning_rate": 3.277423542059918e-06,
+      "loss": 1.7928,
+      "step": 18868000
+    },
+    {
+      "epoch": 93.48,
+      "learning_rate": 3.276184955633834e-06,
+      "loss": 1.7614,
+      "step": 18868500
+    },
+    {
+      "epoch": 93.48,
+      "learning_rate": 3.2749463692077507e-06,
+      "loss": 1.7915,
+      "step": 18869000
+    },
+    {
+      "epoch": 93.49,
+      "learning_rate": 3.2737077827816676e-06,
+      "loss": 1.7935,
+      "step": 18869500
+    },
+    {
+      "epoch": 93.49,
+      "learning_rate": 3.2724691963555832e-06,
+      "loss": 1.7815,
+      "step": 18870000
+    },
+    {
+      "epoch": 93.49,
+      "learning_rate": 3.2712306099295e-06,
+      "loss": 1.7901,
+      "step": 18870500
+    },
+    {
+      "epoch": 93.49,
+      "learning_rate": 3.2699920235034162e-06,
+      "loss": 1.7793,
+      "step": 18871000
+    },
+    {
+      "epoch": 93.5,
+      "learning_rate": 3.2687534370773327e-06,
+      "loss": 1.8129,
+      "step": 18871500
+    },
+    {
+      "epoch": 93.5,
+      "learning_rate": 3.267514850651249e-06,
+      "loss": 1.7768,
+      "step": 18872000
+    },
+    {
+      "epoch": 93.5,
+      "learning_rate": 3.2662762642251653e-06,
+      "loss": 1.8013,
+      "step": 18872500
+    },
+    {
+      "epoch": 93.5,
+      "learning_rate": 3.2650376777990814e-06,
+      "loss": 1.7743,
+      "step": 18873000
+    },
+    {
+      "epoch": 93.51,
+      "learning_rate": 3.263799091372998e-06,
+      "loss": 1.7868,
+      "step": 18873500
+    },
+    {
+      "epoch": 93.51,
+      "learning_rate": 3.262560504946914e-06,
+      "loss": 1.8108,
+      "step": 18874000
+    },
+    {
+      "epoch": 93.51,
+      "learning_rate": 3.261324395693683e-06,
+      "loss": 1.7759,
+      "step": 18874500
+    },
+    {
+      "epoch": 93.51,
+      "learning_rate": 3.260085809267599e-06,
+      "loss": 1.8062,
+      "step": 18875000
+    },
+    {
+      "epoch": 93.52,
+      "learning_rate": 3.258847222841516e-06,
+      "loss": 1.7874,
+      "step": 18875500
+    },
+    {
+      "epoch": 93.52,
+      "learning_rate": 3.2576086364154315e-06,
+      "loss": 1.7901,
+      "step": 18876000
+    },
+    {
+      "epoch": 93.52,
+      "learning_rate": 3.2563700499893484e-06,
+      "loss": 1.7821,
+      "step": 18876500
+    },
+    {
+      "epoch": 93.52,
+      "learning_rate": 3.2551339407361164e-06,
+      "loss": 1.8003,
+      "step": 18877000
+    },
+    {
+      "epoch": 93.53,
+      "learning_rate": 3.2538953543100333e-06,
+      "loss": 1.7838,
+      "step": 18877500
+    },
+    {
+      "epoch": 93.53,
+      "learning_rate": 3.2526567678839494e-06,
+      "loss": 1.7834,
+      "step": 18878000
+    },
+    {
+      "epoch": 93.53,
+      "learning_rate": 3.251418181457866e-06,
+      "loss": 1.7996,
+      "step": 18878500
+    },
+    {
+      "epoch": 93.53,
+      "learning_rate": 3.2501820722046343e-06,
+      "loss": 1.7932,
+      "step": 18879000
+    },
+    {
+      "epoch": 93.54,
+      "learning_rate": 3.2489434857785508e-06,
+      "loss": 1.7715,
+      "step": 18879500
+    },
+    {
+      "epoch": 93.54,
+      "learning_rate": 3.247707376525319e-06,
+      "loss": 1.7839,
+      "step": 18880000
+    },
+    {
+      "epoch": 93.54,
+      "learning_rate": 3.246471267272088e-06,
+      "loss": 1.8048,
+      "step": 18880500
+    },
+    {
+      "epoch": 93.54,
+      "learning_rate": 3.245232680846004e-06,
+      "loss": 1.7741,
+      "step": 18881000
+    },
+    {
+      "epoch": 93.55,
+      "learning_rate": 3.243996571592773e-06,
+      "loss": 1.8109,
+      "step": 18881500
+    },
+    {
+      "epoch": 93.55,
+      "learning_rate": 3.242757985166689e-06,
+      "loss": 1.7854,
+      "step": 18882000
+    },
+    {
+      "epoch": 93.55,
+      "learning_rate": 3.2415193987406054e-06,
+      "loss": 1.7956,
+      "step": 18882500
+    },
+    {
+      "epoch": 93.55,
+      "learning_rate": 3.2402808123145215e-06,
+      "loss": 1.7822,
+      "step": 18883000
+    },
+    {
+      "epoch": 93.56,
+      "learning_rate": 3.2390422258884385e-06,
+      "loss": 1.8077,
+      "step": 18883500
+    },
+    {
+      "epoch": 93.56,
+      "learning_rate": 3.237803639462354e-06,
+      "loss": 1.7845,
+      "step": 18884000
+    },
+    {
+      "epoch": 93.56,
+      "learning_rate": 3.236565053036271e-06,
+      "loss": 1.7939,
+      "step": 18884500
+    },
+    {
+      "epoch": 93.56,
+      "learning_rate": 3.2353264666101875e-06,
+      "loss": 1.7957,
+      "step": 18885000
+    },
+    {
+      "epoch": 93.57,
+      "learning_rate": 3.2340878801841036e-06,
+      "loss": 1.7861,
+      "step": 18885500
+    },
+    {
+      "epoch": 93.57,
+      "learning_rate": 3.23284929375802e-06,
+      "loss": 1.7794,
+      "step": 18886000
+    },
+    {
+      "epoch": 93.57,
+      "learning_rate": 3.2316107073319362e-06,
+      "loss": 1.8044,
+      "step": 18886500
+    },
+    {
+      "epoch": 93.57,
+      "learning_rate": 3.230372120905853e-06,
+      "loss": 1.786,
+      "step": 18887000
+    },
+    {
+      "epoch": 93.58,
+      "learning_rate": 3.229133534479769e-06,
+      "loss": 1.7841,
+      "step": 18887500
+    },
+    {
+      "epoch": 93.58,
+      "learning_rate": 3.227897425226538e-06,
+      "loss": 1.7995,
+      "step": 18888000
+    },
+    {
+      "epoch": 93.58,
+      "learning_rate": 3.2266588388004537e-06,
+      "loss": 1.7828,
+      "step": 18888500
+    },
+    {
+      "epoch": 93.58,
+      "learning_rate": 3.2254202523743706e-06,
+      "loss": 1.8069,
+      "step": 18889000
+    },
+    {
+      "epoch": 93.59,
+      "learning_rate": 3.2241816659482867e-06,
+      "loss": 1.7938,
+      "step": 18889500
+    },
+    {
+      "epoch": 93.59,
+      "learning_rate": 3.222948033867908e-06,
+      "loss": 1.7846,
+      "step": 18890000
+    },
+    {
+      "epoch": 93.59,
+      "learning_rate": 3.221709447441824e-06,
+      "loss": 1.816,
+      "step": 18890500
+    },
+    {
+      "epoch": 93.59,
+      "learning_rate": 3.2204708610157404e-06,
+      "loss": 1.7686,
+      "step": 18891000
+    },
+    {
+      "epoch": 93.6,
+      "learning_rate": 3.2192322745896565e-06,
+      "loss": 1.8122,
+      "step": 18891500
+    },
+    {
+      "epoch": 93.6,
+      "learning_rate": 3.217993688163573e-06,
+      "loss": 1.7839,
+      "step": 18892000
+    },
+    {
+      "epoch": 93.6,
+      "learning_rate": 3.2167600560831937e-06,
+      "loss": 1.8037,
+      "step": 18892500
+    },
+    {
+      "epoch": 93.6,
+      "learning_rate": 3.2155214696571093e-06,
+      "loss": 1.8005,
+      "step": 18893000
+    },
+    {
+      "epoch": 93.6,
+      "learning_rate": 3.2142828832310263e-06,
+      "loss": 1.792,
+      "step": 18893500
+    },
+    {
+      "epoch": 93.61,
+      "learning_rate": 3.2130442968049428e-06,
+      "loss": 1.7934,
+      "step": 18894000
+    },
+    {
+      "epoch": 93.61,
+      "learning_rate": 3.211805710378859e-06,
+      "loss": 1.8092,
+      "step": 18894500
+    },
+    {
+      "epoch": 93.61,
+      "learning_rate": 3.2105671239527758e-06,
+      "loss": 1.7724,
+      "step": 18895000
+    },
+    {
+      "epoch": 93.61,
+      "learning_rate": 3.2093285375266914e-06,
+      "loss": 1.7932,
+      "step": 18895500
+    },
+    {
+      "epoch": 93.62,
+      "learning_rate": 3.2080899511006084e-06,
+      "loss": 1.7813,
+      "step": 18896000
+    },
+    {
+      "epoch": 93.62,
+      "learning_rate": 3.206851364674524e-06,
+      "loss": 1.7776,
+      "step": 18896500
+    },
+    {
+      "epoch": 93.62,
+      "learning_rate": 3.205612778248441e-06,
+      "loss": 1.78,
+      "step": 18897000
+    },
+    {
+      "epoch": 93.62,
+      "learning_rate": 3.204374191822357e-06,
+      "loss": 1.7979,
+      "step": 18897500
+    },
+    {
+      "epoch": 93.63,
+      "learning_rate": 3.2031356053962735e-06,
+      "loss": 1.811,
+      "step": 18898000
+    },
+    {
+      "epoch": 93.63,
+      "learning_rate": 3.2018970189701896e-06,
+      "loss": 1.7966,
+      "step": 18898500
+    },
+    {
+      "epoch": 93.63,
+      "learning_rate": 3.200658432544106e-06,
+      "loss": 1.7694,
+      "step": 18899000
+    },
+    {
+      "epoch": 93.63,
+      "learning_rate": 3.199419846118023e-06,
+      "loss": 1.7716,
+      "step": 18899500
+    },
+    {
+      "epoch": 93.64,
+      "learning_rate": 3.198183736864791e-06,
+      "loss": 1.7826,
+      "step": 18900000
+    },
+    {
+      "epoch": 93.64,
+      "learning_rate": 3.196945150438707e-06,
+      "loss": 1.8124,
+      "step": 18900500
+    },
+    {
+      "epoch": 93.64,
+      "learning_rate": 3.1957065640126236e-06,
+      "loss": 1.7932,
+      "step": 18901000
+    },
+    {
+      "epoch": 93.64,
+      "learning_rate": 3.1944679775865405e-06,
+      "loss": 1.7874,
+      "step": 18901500
+    },
+    {
+      "epoch": 93.65,
+      "learning_rate": 3.193231868333309e-06,
+      "loss": 1.8039,
+      "step": 18902000
+    },
+    {
+      "epoch": 93.65,
+      "learning_rate": 3.1919932819072254e-06,
+      "loss": 1.8027,
+      "step": 18902500
+    },
+    {
+      "epoch": 93.65,
+      "learning_rate": 3.1907546954811415e-06,
+      "loss": 1.8178,
+      "step": 18903000
+    },
+    {
+      "epoch": 93.65,
+      "learning_rate": 3.189516109055058e-06,
+      "loss": 1.7851,
+      "step": 18903500
+    },
+    {
+      "epoch": 93.66,
+      "learning_rate": 3.188277522628974e-06,
+      "loss": 1.7899,
+      "step": 18904000
+    },
+    {
+      "epoch": 93.66,
+      "learning_rate": 3.1870389362028906e-06,
+      "loss": 1.806,
+      "step": 18904500
+    },
+    {
+      "epoch": 93.66,
+      "learning_rate": 3.1858003497768067e-06,
+      "loss": 1.8083,
+      "step": 18905000
+    },
+    {
+      "epoch": 93.66,
+      "learning_rate": 3.1845617633507236e-06,
+      "loss": 1.7889,
+      "step": 18905500
+    },
+    {
+      "epoch": 93.67,
+      "learning_rate": 3.1833231769246393e-06,
+      "loss": 1.7814,
+      "step": 18906000
+    },
+    {
+      "epoch": 93.67,
+      "learning_rate": 3.182084590498556e-06,
+      "loss": 1.7792,
+      "step": 18906500
+    },
+    {
+      "epoch": 93.67,
+      "learning_rate": 3.180848481245324e-06,
+      "loss": 1.7903,
+      "step": 18907000
+    },
+    {
+      "epoch": 93.67,
+      "learning_rate": 3.1796123719920934e-06,
+      "loss": 1.7905,
+      "step": 18907500
+    },
+    {
+      "epoch": 93.68,
+      "learning_rate": 3.178373785566009e-06,
+      "loss": 1.7716,
+      "step": 18908000
+    },
+    {
+      "epoch": 93.68,
+      "learning_rate": 3.177135199139926e-06,
+      "loss": 1.7834,
+      "step": 18908500
+    },
+    {
+      "epoch": 93.68,
+      "learning_rate": 3.175896612713842e-06,
+      "loss": 1.7942,
+      "step": 18909000
+    },
+    {
+      "epoch": 93.68,
+      "learning_rate": 3.1746580262877585e-06,
+      "loss": 1.777,
+      "step": 18909500
+    },
+    {
+      "epoch": 93.69,
+      "learning_rate": 3.1734194398616746e-06,
+      "loss": 1.7944,
+      "step": 18910000
+    },
+    {
+      "epoch": 93.69,
+      "learning_rate": 3.1721833306084434e-06,
+      "loss": 1.8017,
+      "step": 18910500
+    },
+    {
+      "epoch": 93.69,
+      "learning_rate": 3.1709447441823595e-06,
+      "loss": 1.779,
+      "step": 18911000
+    },
+    {
+      "epoch": 93.69,
+      "learning_rate": 3.169706157756276e-06,
+      "loss": 1.7946,
+      "step": 18911500
+    },
+    {
+      "epoch": 93.7,
+      "learning_rate": 3.1684700485030444e-06,
+      "loss": 1.8024,
+      "step": 18912000
+    },
+    {
+      "epoch": 93.7,
+      "learning_rate": 3.167231462076961e-06,
+      "loss": 1.8022,
+      "step": 18912500
+    },
+    {
+      "epoch": 93.7,
+      "learning_rate": 3.165992875650877e-06,
+      "loss": 1.8069,
+      "step": 18913000
+    },
+    {
+      "epoch": 93.7,
+      "learning_rate": 3.164754289224794e-06,
+      "loss": 1.7762,
+      "step": 18913500
+    },
+    {
+      "epoch": 93.71,
+      "learning_rate": 3.1635157027987096e-06,
+      "loss": 1.759,
+      "step": 18914000
+    },
+    {
+      "epoch": 93.71,
+      "learning_rate": 3.1622771163726265e-06,
+      "loss": 1.7864,
+      "step": 18914500
+    },
+    {
+      "epoch": 93.71,
+      "learning_rate": 3.161038529946543e-06,
+      "loss": 1.7889,
+      "step": 18915000
+    },
+    {
+      "epoch": 93.71,
+      "learning_rate": 3.159799943520459e-06,
+      "loss": 1.7922,
+      "step": 18915500
+    },
+    {
+      "epoch": 93.72,
+      "learning_rate": 3.1585613570943756e-06,
+      "loss": 1.7913,
+      "step": 18916000
+    },
+    {
+      "epoch": 93.72,
+      "learning_rate": 3.1573227706682917e-06,
+      "loss": 1.7879,
+      "step": 18916500
+    },
+    {
+      "epoch": 93.72,
+      "learning_rate": 3.1560841842422086e-06,
+      "loss": 1.7706,
+      "step": 18917000
+    },
+    {
+      "epoch": 93.72,
+      "learning_rate": 3.1548455978161243e-06,
+      "loss": 1.7973,
+      "step": 18917500
+    },
+    {
+      "epoch": 93.73,
+      "learning_rate": 3.153607011390041e-06,
+      "loss": 1.8206,
+      "step": 18918000
+    },
+    {
+      "epoch": 93.73,
+      "learning_rate": 3.1523684249639573e-06,
+      "loss": 1.7904,
+      "step": 18918500
+    },
+    {
+      "epoch": 93.73,
+      "learning_rate": 3.151132315710726e-06,
+      "loss": 1.7843,
+      "step": 18919000
+    },
+    {
+      "epoch": 93.73,
+      "learning_rate": 3.149893729284642e-06,
+      "loss": 1.794,
+      "step": 18919500
+    },
+    {
+      "epoch": 93.74,
+      "learning_rate": 3.1486551428585587e-06,
+      "loss": 1.7932,
+      "step": 18920000
+    },
+    {
+      "epoch": 93.74,
+      "learning_rate": 3.1474165564324748e-06,
+      "loss": 1.7917,
+      "step": 18920500
+    },
+    {
+      "epoch": 93.74,
+      "learning_rate": 3.1461779700063913e-06,
+      "loss": 1.7982,
+      "step": 18921000
+    },
+    {
+      "epoch": 93.74,
+      "learning_rate": 3.1449393835803073e-06,
+      "loss": 1.7867,
+      "step": 18921500
+    },
+    {
+      "epoch": 93.75,
+      "learning_rate": 3.143703274327076e-06,
+      "loss": 1.8069,
+      "step": 18922000
+    },
+    {
+      "epoch": 93.75,
+      "learning_rate": 3.1424646879009922e-06,
+      "loss": 1.7897,
+      "step": 18922500
+    },
+    {
+      "epoch": 93.75,
+      "learning_rate": 3.1412261014749087e-06,
+      "loss": 1.77,
+      "step": 18923000
+    },
+    {
+      "epoch": 93.75,
+      "learning_rate": 3.139987515048825e-06,
+      "loss": 1.8031,
+      "step": 18923500
+    },
+    {
+      "epoch": 93.76,
+      "learning_rate": 3.1387489286227417e-06,
+      "loss": 1.7908,
+      "step": 18924000
+    },
+    {
+      "epoch": 93.76,
+      "learning_rate": 3.1375103421966582e-06,
+      "loss": 1.7824,
+      "step": 18924500
+    },
+    {
+      "epoch": 93.76,
+      "learning_rate": 3.1362717557705743e-06,
+      "loss": 1.8045,
+      "step": 18925000
+    },
+    {
+      "epoch": 93.76,
+      "learning_rate": 3.135033169344491e-06,
+      "loss": 1.8008,
+      "step": 18925500
+    },
+    {
+      "epoch": 93.77,
+      "learning_rate": 3.133794582918407e-06,
+      "loss": 1.7845,
+      "step": 18926000
+    },
+    {
+      "epoch": 93.77,
+      "learning_rate": 3.1325584736651757e-06,
+      "loss": 1.7937,
+      "step": 18926500
+    },
+    {
+      "epoch": 93.77,
+      "learning_rate": 3.131319887239092e-06,
+      "loss": 1.77,
+      "step": 18927000
+    },
+    {
+      "epoch": 93.77,
+      "learning_rate": 3.1300813008130083e-06,
+      "loss": 1.7937,
+      "step": 18927500
+    },
+    {
+      "epoch": 93.78,
+      "learning_rate": 3.1288427143869244e-06,
+      "loss": 1.8022,
+      "step": 18928000
+    },
+    {
+      "epoch": 93.78,
+      "learning_rate": 3.1276041279608413e-06,
+      "loss": 1.7752,
+      "step": 18928500
+    },
+    {
+      "epoch": 93.78,
+      "learning_rate": 3.1263680187076093e-06,
+      "loss": 1.7846,
+      "step": 18929000
+    },
+    {
+      "epoch": 93.78,
+      "learning_rate": 3.125129432281526e-06,
+      "loss": 1.7893,
+      "step": 18929500
+    },
+    {
+      "epoch": 93.79,
+      "learning_rate": 3.1238908458554423e-06,
+      "loss": 1.804,
+      "step": 18930000
+    },
+    {
+      "epoch": 93.79,
+      "learning_rate": 3.1226547366022107e-06,
+      "loss": 1.7819,
+      "step": 18930500
+    },
+    {
+      "epoch": 93.79,
+      "learning_rate": 3.121416150176127e-06,
+      "loss": 1.8084,
+      "step": 18931000
+    },
+    {
+      "epoch": 93.79,
+      "learning_rate": 3.1201775637500433e-06,
+      "loss": 1.7843,
+      "step": 18931500
+    },
+    {
+      "epoch": 93.8,
+      "learning_rate": 3.11893897732396e-06,
+      "loss": 1.7786,
+      "step": 18932000
+    },
+    {
+      "epoch": 93.8,
+      "learning_rate": 3.1177003908978763e-06,
+      "loss": 1.7851,
+      "step": 18932500
+    },
+    {
+      "epoch": 93.8,
+      "learning_rate": 3.1164618044717928e-06,
+      "loss": 1.7898,
+      "step": 18933000
+    },
+    {
+      "epoch": 93.8,
+      "learning_rate": 3.115225695218561e-06,
+      "loss": 1.7748,
+      "step": 18933500
+    },
+    {
+      "epoch": 93.81,
+      "learning_rate": 3.1139871087924777e-06,
+      "loss": 1.807,
+      "step": 18934000
+    },
+    {
+      "epoch": 93.81,
+      "learning_rate": 3.112750999539246e-06,
+      "loss": 1.7932,
+      "step": 18934500
+    },
+    {
+      "epoch": 93.81,
+      "learning_rate": 3.111512413113162e-06,
+      "loss": 1.8051,
+      "step": 18935000
+    },
+    {
+      "epoch": 93.81,
+      "learning_rate": 3.110273826687079e-06,
+      "loss": 1.81,
+      "step": 18935500
+    },
+    {
+      "epoch": 93.82,
+      "learning_rate": 3.109035240260995e-06,
+      "loss": 1.7973,
+      "step": 18936000
+    },
+    {
+      "epoch": 93.82,
+      "learning_rate": 3.1077966538349116e-06,
+      "loss": 1.7701,
+      "step": 18936500
+    },
+    {
+      "epoch": 93.82,
+      "learning_rate": 3.10656054458168e-06,
+      "loss": 1.8103,
+      "step": 18937000
+    },
+    {
+      "epoch": 93.82,
+      "learning_rate": 3.1053219581555965e-06,
+      "loss": 1.8216,
+      "step": 18937500
+    },
+    {
+      "epoch": 93.83,
+      "learning_rate": 3.1040858489023645e-06,
+      "loss": 1.7571,
+      "step": 18938000
+    },
+    {
+      "epoch": 93.83,
+      "learning_rate": 3.1028472624762814e-06,
+      "loss": 1.7926,
+      "step": 18938500
+    },
+    {
+      "epoch": 93.83,
+      "learning_rate": 3.1016086760501975e-06,
+      "loss": 1.8123,
+      "step": 18939000
+    },
+    {
+      "epoch": 93.83,
+      "learning_rate": 3.100370089624114e-06,
+      "loss": 1.785,
+      "step": 18939500
+    },
+    {
+      "epoch": 93.84,
+      "learning_rate": 3.0991315031980305e-06,
+      "loss": 1.7812,
+      "step": 18940000
+    },
+    {
+      "epoch": 93.84,
+      "learning_rate": 3.0978929167719466e-06,
+      "loss": 1.7968,
+      "step": 18940500
+    },
+    {
+      "epoch": 93.84,
+      "learning_rate": 3.096654330345863e-06,
+      "loss": 1.7751,
+      "step": 18941000
+    },
+    {
+      "epoch": 93.84,
+      "learning_rate": 3.095415743919779e-06,
+      "loss": 1.7904,
+      "step": 18941500
+    },
+    {
+      "epoch": 93.85,
+      "learning_rate": 3.094179634666548e-06,
+      "loss": 1.7737,
+      "step": 18942000
+    },
+    {
+      "epoch": 93.85,
+      "learning_rate": 3.0929410482404645e-06,
+      "loss": 1.7832,
+      "step": 18942500
+    },
+    {
+      "epoch": 93.85,
+      "learning_rate": 3.091704938987233e-06,
+      "loss": 1.7803,
+      "step": 18943000
+    },
+    {
+      "epoch": 93.85,
+      "learning_rate": 3.0904663525611494e-06,
+      "loss": 1.8158,
+      "step": 18943500
+    },
+    {
+      "epoch": 93.86,
+      "learning_rate": 3.0892277661350655e-06,
+      "loss": 1.7918,
+      "step": 18944000
+    },
+    {
+      "epoch": 93.86,
+      "learning_rate": 3.087989179708982e-06,
+      "loss": 1.7865,
+      "step": 18944500
+    },
+    {
+      "epoch": 93.86,
+      "learning_rate": 3.086750593282898e-06,
+      "loss": 1.7823,
+      "step": 18945000
+    },
+    {
+      "epoch": 93.86,
+      "learning_rate": 3.0855120068568146e-06,
+      "loss": 1.7789,
+      "step": 18945500
+    },
+    {
+      "epoch": 93.87,
+      "learning_rate": 3.084273420430731e-06,
+      "loss": 1.771,
+      "step": 18946000
+    },
+    {
+      "epoch": 93.87,
+      "learning_rate": 3.083034834004647e-06,
+      "loss": 1.7853,
+      "step": 18946500
+    },
+    {
+      "epoch": 93.87,
+      "learning_rate": 3.0817962475785637e-06,
+      "loss": 1.7984,
+      "step": 18947000
+    },
+    {
+      "epoch": 93.87,
+      "learning_rate": 3.080560138325332e-06,
+      "loss": 1.8054,
+      "step": 18947500
+    },
+    {
+      "epoch": 93.87,
+      "learning_rate": 3.0793215518992485e-06,
+      "loss": 1.7869,
+      "step": 18948000
+    },
+    {
+      "epoch": 93.88,
+      "learning_rate": 3.0780829654731646e-06,
+      "loss": 1.7762,
+      "step": 18948500
+    },
+    {
+      "epoch": 93.88,
+      "learning_rate": 3.076844379047081e-06,
+      "loss": 1.8133,
+      "step": 18949000
+    },
+    {
+      "epoch": 93.88,
+      "learning_rate": 3.0756057926209976e-06,
+      "loss": 1.7843,
+      "step": 18949500
+    },
+    {
+      "epoch": 93.88,
+      "learning_rate": 3.074367206194914e-06,
+      "loss": 1.8145,
+      "step": 18950000
+    },
+    {
+      "epoch": 93.89,
+      "learning_rate": 3.0731310969416825e-06,
+      "loss": 1.8061,
+      "step": 18950500
+    },
+    {
+      "epoch": 93.89,
+      "learning_rate": 3.071894987688451e-06,
+      "loss": 1.7989,
+      "step": 18951000
+    },
+    {
+      "epoch": 93.89,
+      "learning_rate": 3.0706564012623674e-06,
+      "loss": 1.8015,
+      "step": 18951500
+    },
+    {
+      "epoch": 93.89,
+      "learning_rate": 3.0694178148362835e-06,
+      "loss": 1.7971,
+      "step": 18952000
+    },
+    {
+      "epoch": 93.9,
+      "learning_rate": 3.0681792284102004e-06,
+      "loss": 1.7661,
+      "step": 18952500
+    },
+    {
+      "epoch": 93.9,
+      "learning_rate": 3.0669406419841165e-06,
+      "loss": 1.7656,
+      "step": 18953000
+    },
+    {
+      "epoch": 93.9,
+      "learning_rate": 3.065702055558033e-06,
+      "loss": 1.7871,
+      "step": 18953500
+    },
+    {
+      "epoch": 93.9,
+      "learning_rate": 3.0644634691319495e-06,
+      "loss": 1.7849,
+      "step": 18954000
+    },
+    {
+      "epoch": 93.91,
+      "learning_rate": 3.0632248827058656e-06,
+      "loss": 1.7845,
+      "step": 18954500
+    },
+    {
+      "epoch": 93.91,
+      "learning_rate": 3.061986296279782e-06,
+      "loss": 1.776,
+      "step": 18955000
+    },
+    {
+      "epoch": 93.91,
+      "learning_rate": 3.060747709853698e-06,
+      "loss": 1.7764,
+      "step": 18955500
+    },
+    {
+      "epoch": 93.91,
+      "learning_rate": 3.0595091234276147e-06,
+      "loss": 1.783,
+      "step": 18956000
+    },
+    {
+      "epoch": 93.92,
+      "learning_rate": 3.0582705370015308e-06,
+      "loss": 1.7904,
+      "step": 18956500
+    },
+    {
+      "epoch": 93.92,
+      "learning_rate": 3.0570319505754473e-06,
+      "loss": 1.7831,
+      "step": 18957000
+    },
+    {
+      "epoch": 93.92,
+      "learning_rate": 3.0557933641493638e-06,
+      "loss": 1.7751,
+      "step": 18957500
+    },
+    {
+      "epoch": 93.92,
+      "learning_rate": 3.05455477772328e-06,
+      "loss": 1.8009,
+      "step": 18958000
+    },
+    {
+      "epoch": 93.93,
+      "learning_rate": 3.0533161912971968e-06,
+      "loss": 1.7967,
+      "step": 18958500
+    },
+    {
+      "epoch": 93.93,
+      "learning_rate": 3.052077604871113e-06,
+      "loss": 1.7878,
+      "step": 18959000
+    },
+    {
+      "epoch": 93.93,
+      "learning_rate": 3.0508390184450294e-06,
+      "loss": 1.7758,
+      "step": 18959500
+    },
+    {
+      "epoch": 93.93,
+      "learning_rate": 3.0496004320189454e-06,
+      "loss": 1.7797,
+      "step": 18960000
+    },
+    {
+      "epoch": 93.94,
+      "learning_rate": 3.048361845592862e-06,
+      "loss": 1.8009,
+      "step": 18960500
+    },
+    {
+      "epoch": 93.94,
+      "learning_rate": 3.0471232591667785e-06,
+      "loss": 1.7813,
+      "step": 18961000
+    },
+    {
+      "epoch": 93.94,
+      "learning_rate": 3.0458846727406945e-06,
+      "loss": 1.7787,
+      "step": 18961500
+    },
+    {
+      "epoch": 93.94,
+      "learning_rate": 3.0446485634874633e-06,
+      "loss": 1.7881,
+      "step": 18962000
+    },
+    {
+      "epoch": 93.95,
+      "learning_rate": 3.0434149314070836e-06,
+      "loss": 1.8035,
+      "step": 18962500
+    },
+    {
+      "epoch": 93.95,
+      "learning_rate": 3.042176344981e-06,
+      "loss": 1.8033,
+      "step": 18963000
+    },
+    {
+      "epoch": 93.95,
+      "learning_rate": 3.0409377585549166e-06,
+      "loss": 1.8143,
+      "step": 18963500
+    },
+    {
+      "epoch": 93.95,
+      "learning_rate": 3.039699172128833e-06,
+      "loss": 1.7874,
+      "step": 18964000
+    },
+    {
+      "epoch": 93.96,
+      "learning_rate": 3.0384605857027492e-06,
+      "loss": 1.7938,
+      "step": 18964500
+    },
+    {
+      "epoch": 93.96,
+      "learning_rate": 3.0372219992766657e-06,
+      "loss": 1.7641,
+      "step": 18965000
+    },
+    {
+      "epoch": 93.96,
+      "learning_rate": 3.0359834128505822e-06,
+      "loss": 1.7898,
+      "step": 18965500
+    },
+    {
+      "epoch": 93.96,
+      "learning_rate": 3.0347448264244983e-06,
+      "loss": 1.798,
+      "step": 18966000
+    },
+    {
+      "epoch": 93.97,
+      "learning_rate": 3.033506239998415e-06,
+      "loss": 1.7702,
+      "step": 18966500
+    },
+    {
+      "epoch": 93.97,
+      "learning_rate": 3.0322726079180355e-06,
+      "loss": 1.802,
+      "step": 18967000
+    },
+    {
+      "epoch": 93.97,
+      "learning_rate": 3.031034021491952e-06,
+      "loss": 1.7983,
+      "step": 18967500
+    },
+    {
+      "epoch": 93.97,
+      "learning_rate": 3.029795435065868e-06,
+      "loss": 1.8094,
+      "step": 18968000
+    },
+    {
+      "epoch": 93.98,
+      "learning_rate": 3.0285568486397846e-06,
+      "loss": 1.7761,
+      "step": 18968500
+    },
+    {
+      "epoch": 93.98,
+      "learning_rate": 3.027318262213701e-06,
+      "loss": 1.7913,
+      "step": 18969000
+    },
+    {
+      "epoch": 93.98,
+      "learning_rate": 3.026079675787617e-06,
+      "loss": 1.7928,
+      "step": 18969500
+    },
+    {
+      "epoch": 93.98,
+      "learning_rate": 3.0248410893615337e-06,
+      "loss": 1.7787,
+      "step": 18970000
+    },
+    {
+      "epoch": 93.99,
+      "learning_rate": 3.0236025029354498e-06,
+      "loss": 1.7798,
+      "step": 18970500
+    },
+    {
+      "epoch": 93.99,
+      "learning_rate": 3.0223639165093663e-06,
+      "loss": 1.7707,
+      "step": 18971000
+    },
+    {
+      "epoch": 93.99,
+      "learning_rate": 3.0211278072561347e-06,
+      "loss": 1.7885,
+      "step": 18971500
+    },
+    {
+      "epoch": 93.99,
+      "learning_rate": 3.019889220830051e-06,
+      "loss": 1.7921,
+      "step": 18972000
+    },
+    {
+      "epoch": 94.0,
+      "learning_rate": 3.01865311157682e-06,
+      "loss": 1.8087,
+      "step": 18972500
+    },
+    {
+      "epoch": 94.0,
+      "learning_rate": 3.017414525150736e-06,
+      "loss": 1.8017,
+      "step": 18973000
+    },
+    {
+      "epoch": 94.0,
+      "eval_accuracy": 0.6856816384535349,
+      "eval_accuracy_mlm": 0.6469539919371597,
+      "eval_accuracy_nsp": 0.8684219815735079,
+      "eval_loss": 2.311725616455078,
+      "eval_runtime": 147.0378,
+      "eval_samples_per_second": 1733.969,
+      "eval_steps_per_second": 72.254,
+      "step": 18973242
+    },
+    {
+      "epoch": 94.0,
+      "learning_rate": 3.0161759387246526e-06,
+      "loss": 1.7706,
+      "step": 18973500
+    },
+    {
+      "epoch": 94.0,
+      "learning_rate": 3.014939829471421e-06,
+      "loss": 1.7808,
+      "step": 18974000
+    },
+    {
+      "epoch": 94.01,
+      "learning_rate": 3.0137012430453374e-06,
+      "loss": 1.7786,
+      "step": 18974500
+    },
+    {
+      "epoch": 94.01,
+      "learning_rate": 3.0124626566192535e-06,
+      "loss": 1.7906,
+      "step": 18975000
+    },
+    {
+      "epoch": 94.01,
+      "learning_rate": 3.01122407019317e-06,
+      "loss": 1.7993,
+      "step": 18975500
+    },
+    {
+      "epoch": 94.01,
+      "learning_rate": 3.0099854837670865e-06,
+      "loss": 1.8006,
+      "step": 18976000
+    },
+    {
+      "epoch": 94.02,
+      "learning_rate": 3.0087468973410026e-06,
+      "loss": 1.7736,
+      "step": 18976500
+    },
+    {
+      "epoch": 94.02,
+      "learning_rate": 3.007508310914919e-06,
+      "loss": 1.7823,
+      "step": 18977000
+    },
+    {
+      "epoch": 94.02,
+      "learning_rate": 3.0062697244888356e-06,
+      "loss": 1.7935,
+      "step": 18977500
+    },
+    {
+      "epoch": 94.02,
+      "learning_rate": 3.005031138062752e-06,
+      "loss": 1.7881,
+      "step": 18978000
+    },
+    {
+      "epoch": 94.03,
+      "learning_rate": 3.003792551636668e-06,
+      "loss": 1.7739,
+      "step": 18978500
+    },
+    {
+      "epoch": 94.03,
+      "learning_rate": 3.0025539652105847e-06,
+      "loss": 1.8044,
+      "step": 18979000
+    },
+    {
+      "epoch": 94.03,
+      "learning_rate": 3.0013153787845012e-06,
+      "loss": 1.8031,
+      "step": 18979500
+    },
+    {
+      "epoch": 94.03,
+      "learning_rate": 3.0000792695312696e-06,
+      "loss": 1.7797,
+      "step": 18980000
+    },
+    {
+      "epoch": 94.04,
+      "learning_rate": 2.998840683105186e-06,
+      "loss": 1.7759,
+      "step": 18980500
+    },
+    {
+      "epoch": 94.04,
+      "learning_rate": 2.9976045738519545e-06,
+      "loss": 1.7828,
+      "step": 18981000
+    },
+    {
+      "epoch": 94.04,
+      "learning_rate": 2.996365987425871e-06,
+      "loss": 1.7745,
+      "step": 18981500
+    },
+    {
+      "epoch": 94.04,
+      "learning_rate": 2.995127400999787e-06,
+      "loss": 1.7915,
+      "step": 18982000
+    },
+    {
+      "epoch": 94.05,
+      "learning_rate": 2.9938888145737036e-06,
+      "loss": 1.7697,
+      "step": 18982500
+    },
+    {
+      "epoch": 94.05,
+      "learning_rate": 2.9926502281476197e-06,
+      "loss": 1.779,
+      "step": 18983000
+    },
+    {
+      "epoch": 94.05,
+      "learning_rate": 2.991411641721536e-06,
+      "loss": 1.7818,
+      "step": 18983500
+    },
+    {
+      "epoch": 94.05,
+      "learning_rate": 2.9901730552954527e-06,
+      "loss": 1.7884,
+      "step": 18984000
+    },
+    {
+      "epoch": 94.06,
+      "learning_rate": 2.9889344688693688e-06,
+      "loss": 1.7988,
+      "step": 18984500
+    },
+    {
+      "epoch": 94.06,
+      "learning_rate": 2.9876958824432853e-06,
+      "loss": 1.766,
+      "step": 18985000
+    },
+    {
+      "epoch": 94.06,
+      "learning_rate": 2.9864572960172013e-06,
+      "loss": 1.7899,
+      "step": 18985500
+    },
+    {
+      "epoch": 94.06,
+      "learning_rate": 2.985218709591118e-06,
+      "loss": 1.786,
+      "step": 18986000
+    },
+    {
+      "epoch": 94.07,
+      "learning_rate": 2.9839801231650343e-06,
+      "loss": 1.7699,
+      "step": 18986500
+    },
+    {
+      "epoch": 94.07,
+      "learning_rate": 2.982741536738951e-06,
+      "loss": 1.7587,
+      "step": 18987000
+    },
+    {
+      "epoch": 94.07,
+      "learning_rate": 2.9815029503128674e-06,
+      "loss": 1.793,
+      "step": 18987500
+    },
+    {
+      "epoch": 94.07,
+      "learning_rate": 2.9802668410596353e-06,
+      "loss": 1.7885,
+      "step": 18988000
+    },
+    {
+      "epoch": 94.08,
+      "learning_rate": 2.9790282546335522e-06,
+      "loss": 1.7987,
+      "step": 18988500
+    },
+    {
+      "epoch": 94.08,
+      "learning_rate": 2.9777896682074683e-06,
+      "loss": 1.7905,
+      "step": 18989000
+    },
+    {
+      "epoch": 94.08,
+      "learning_rate": 2.976551081781385e-06,
+      "loss": 1.7878,
+      "step": 18989500
+    },
+    {
+      "epoch": 94.08,
+      "learning_rate": 2.975312495355301e-06,
+      "loss": 1.8126,
+      "step": 18990000
+    },
+    {
+      "epoch": 94.09,
+      "learning_rate": 2.9740739089292174e-06,
+      "loss": 1.7934,
+      "step": 18990500
+    },
+    {
+      "epoch": 94.09,
+      "learning_rate": 2.972835322503134e-06,
+      "loss": 1.8086,
+      "step": 18991000
+    },
+    {
+      "epoch": 94.09,
+      "learning_rate": 2.9715992132499023e-06,
+      "loss": 1.7799,
+      "step": 18991500
+    },
+    {
+      "epoch": 94.09,
+      "learning_rate": 2.970363103996671e-06,
+      "loss": 1.7925,
+      "step": 18992000
+    },
+    {
+      "epoch": 94.1,
+      "learning_rate": 2.969124517570587e-06,
+      "loss": 1.7913,
+      "step": 18992500
+    },
+    {
+      "epoch": 94.1,
+      "learning_rate": 2.9678859311445037e-06,
+      "loss": 1.798,
+      "step": 18993000
+    },
+    {
+      "epoch": 94.1,
+      "learning_rate": 2.9666473447184198e-06,
+      "loss": 1.789,
+      "step": 18993500
+    },
+    {
+      "epoch": 94.1,
+      "learning_rate": 2.9654087582923363e-06,
+      "loss": 1.7856,
+      "step": 18994000
+    },
+    {
+      "epoch": 94.11,
+      "learning_rate": 2.964170171866253e-06,
+      "loss": 1.7789,
+      "step": 18994500
+    },
+    {
+      "epoch": 94.11,
+      "learning_rate": 2.962934062613021e-06,
+      "loss": 1.7783,
+      "step": 18995000
+    },
+    {
+      "epoch": 94.11,
+      "learning_rate": 2.9616954761869377e-06,
+      "loss": 1.7762,
+      "step": 18995500
+    },
+    {
+      "epoch": 94.11,
+      "learning_rate": 2.9604568897608538e-06,
+      "loss": 1.7801,
+      "step": 18996000
+    },
+    {
+      "epoch": 94.12,
+      "learning_rate": 2.9592183033347703e-06,
+      "loss": 1.7886,
+      "step": 18996500
+    },
+    {
+      "epoch": 94.12,
+      "learning_rate": 2.9579797169086864e-06,
+      "loss": 1.802,
+      "step": 18997000
+    },
+    {
+      "epoch": 94.12,
+      "learning_rate": 2.956741130482603e-06,
+      "loss": 1.8109,
+      "step": 18997500
+    },
+    {
+      "epoch": 94.12,
+      "learning_rate": 2.9555025440565194e-06,
+      "loss": 1.7921,
+      "step": 18998000
+    },
+    {
+      "epoch": 94.13,
+      "learning_rate": 2.9542639576304354e-06,
+      "loss": 1.7824,
+      "step": 18998500
+    },
+    {
+      "epoch": 94.13,
+      "learning_rate": 2.953025371204352e-06,
+      "loss": 1.785,
+      "step": 18999000
+    },
+    {
+      "epoch": 94.13,
+      "learning_rate": 2.9517892619511203e-06,
+      "loss": 1.7788,
+      "step": 18999500
+    },
+    {
+      "epoch": 94.13,
+      "learning_rate": 2.950550675525037e-06,
+      "loss": 1.7781,
+      "step": 19000000
+    },
+    {
+      "epoch": 94.14,
+      "learning_rate": 2.949312089098953e-06,
+      "loss": 1.8023,
+      "step": 19000500
+    },
+    {
+      "epoch": 94.14,
+      "learning_rate": 2.94807350267287e-06,
+      "loss": 1.7759,
+      "step": 19001000
+    },
+    {
+      "epoch": 94.14,
+      "learning_rate": 2.946834916246786e-06,
+      "loss": 1.8143,
+      "step": 19001500
+    },
+    {
+      "epoch": 94.14,
+      "learning_rate": 2.9455963298207024e-06,
+      "loss": 1.7864,
+      "step": 19002000
+    },
+    {
+      "epoch": 94.14,
+      "learning_rate": 2.944357743394619e-06,
+      "loss": 1.7917,
+      "step": 19002500
+    },
+    {
+      "epoch": 94.15,
+      "learning_rate": 2.943119156968535e-06,
+      "loss": 1.7953,
+      "step": 19003000
+    },
+    {
+      "epoch": 94.15,
+      "learning_rate": 2.9418805705424515e-06,
+      "loss": 1.7768,
+      "step": 19003500
+    },
+    {
+      "epoch": 94.15,
+      "learning_rate": 2.9406419841163676e-06,
+      "loss": 1.7558,
+      "step": 19004000
+    },
+    {
+      "epoch": 94.15,
+      "learning_rate": 2.939403397690284e-06,
+      "loss": 1.7807,
+      "step": 19004500
+    },
+    {
+      "epoch": 94.16,
+      "learning_rate": 2.9381648112642006e-06,
+      "loss": 1.7821,
+      "step": 19005000
+    },
+    {
+      "epoch": 94.16,
+      "learning_rate": 2.9369262248381167e-06,
+      "loss": 1.7927,
+      "step": 19005500
+    },
+    {
+      "epoch": 94.16,
+      "learning_rate": 2.935687638412033e-06,
+      "loss": 1.7877,
+      "step": 19006000
+    },
+    {
+      "epoch": 94.16,
+      "learning_rate": 2.934454006331654e-06,
+      "loss": 1.7728,
+      "step": 19006500
+    },
+    {
+      "epoch": 94.17,
+      "learning_rate": 2.9332178970784227e-06,
+      "loss": 1.7957,
+      "step": 19007000
+    },
+    {
+      "epoch": 94.17,
+      "learning_rate": 2.9319793106523388e-06,
+      "loss": 1.7903,
+      "step": 19007500
+    },
+    {
+      "epoch": 94.17,
+      "learning_rate": 2.9307407242262553e-06,
+      "loss": 1.7897,
+      "step": 19008000
+    },
+    {
+      "epoch": 94.17,
+      "learning_rate": 2.9295021378001714e-06,
+      "loss": 1.7797,
+      "step": 19008500
+    },
+    {
+      "epoch": 94.18,
+      "learning_rate": 2.928263551374088e-06,
+      "loss": 1.7694,
+      "step": 19009000
+    },
+    {
+      "epoch": 94.18,
+      "learning_rate": 2.9270249649480044e-06,
+      "loss": 1.7736,
+      "step": 19009500
+    },
+    {
+      "epoch": 94.18,
+      "learning_rate": 2.9257863785219205e-06,
+      "loss": 1.7912,
+      "step": 19010000
+    },
+    {
+      "epoch": 94.18,
+      "learning_rate": 2.924547792095837e-06,
+      "loss": 1.7731,
+      "step": 19010500
+    },
+    {
+      "epoch": 94.19,
+      "learning_rate": 2.9233116828426054e-06,
+      "loss": 1.8031,
+      "step": 19011000
+    },
+    {
+      "epoch": 94.19,
+      "learning_rate": 2.922073096416522e-06,
+      "loss": 1.78,
+      "step": 19011500
+    },
+    {
+      "epoch": 94.19,
+      "learning_rate": 2.920834509990438e-06,
+      "loss": 1.8002,
+      "step": 19012000
+    },
+    {
+      "epoch": 94.19,
+      "learning_rate": 2.9195959235643544e-06,
+      "loss": 1.7856,
+      "step": 19012500
+    },
+    {
+      "epoch": 94.2,
+      "learning_rate": 2.9183598143111233e-06,
+      "loss": 1.7818,
+      "step": 19013000
+    },
+    {
+      "epoch": 94.2,
+      "learning_rate": 2.9171237050578916e-06,
+      "loss": 1.7625,
+      "step": 19013500
+    },
+    {
+      "epoch": 94.2,
+      "learning_rate": 2.915885118631808e-06,
+      "loss": 1.7899,
+      "step": 19014000
+    },
+    {
+      "epoch": 94.2,
+      "learning_rate": 2.9146465322057242e-06,
+      "loss": 1.783,
+      "step": 19014500
+    },
+    {
+      "epoch": 94.21,
+      "learning_rate": 2.9134079457796407e-06,
+      "loss": 1.7868,
+      "step": 19015000
+    },
+    {
+      "epoch": 94.21,
+      "learning_rate": 2.912171836526409e-06,
+      "loss": 1.7605,
+      "step": 19015500
+    },
+    {
+      "epoch": 94.21,
+      "learning_rate": 2.9109332501003256e-06,
+      "loss": 1.7741,
+      "step": 19016000
+    },
+    {
+      "epoch": 94.21,
+      "learning_rate": 2.909697140847094e-06,
+      "loss": 1.7783,
+      "step": 19016500
+    },
+    {
+      "epoch": 94.22,
+      "learning_rate": 2.9084585544210105e-06,
+      "loss": 1.8065,
+      "step": 19017000
+    },
+    {
+      "epoch": 94.22,
+      "learning_rate": 2.907219967994927e-06,
+      "loss": 1.7691,
+      "step": 19017500
+    },
+    {
+      "epoch": 94.22,
+      "learning_rate": 2.905981381568843e-06,
+      "loss": 1.798,
+      "step": 19018000
+    },
+    {
+      "epoch": 94.22,
+      "learning_rate": 2.9047427951427596e-06,
+      "loss": 1.7897,
+      "step": 19018500
+    },
+    {
+      "epoch": 94.23,
+      "learning_rate": 2.9035042087166757e-06,
+      "loss": 1.7859,
+      "step": 19019000
+    },
+    {
+      "epoch": 94.23,
+      "learning_rate": 2.902265622290592e-06,
+      "loss": 1.7506,
+      "step": 19019500
+    },
+    {
+      "epoch": 94.23,
+      "learning_rate": 2.9010270358645087e-06,
+      "loss": 1.7807,
+      "step": 19020000
+    },
+    {
+      "epoch": 94.23,
+      "learning_rate": 2.899790926611277e-06,
+      "loss": 1.8035,
+      "step": 19020500
+    },
+    {
+      "epoch": 94.24,
+      "learning_rate": 2.8985523401851936e-06,
+      "loss": 1.7819,
+      "step": 19021000
+    },
+    {
+      "epoch": 94.24,
+      "learning_rate": 2.89731375375911e-06,
+      "loss": 1.7718,
+      "step": 19021500
+    },
+    {
+      "epoch": 94.24,
+      "learning_rate": 2.8960751673330266e-06,
+      "loss": 1.7901,
+      "step": 19022000
+    },
+    {
+      "epoch": 94.24,
+      "learning_rate": 2.8948365809069427e-06,
+      "loss": 1.7961,
+      "step": 19022500
+    },
+    {
+      "epoch": 94.25,
+      "learning_rate": 2.893597994480859e-06,
+      "loss": 1.7958,
+      "step": 19023000
+    },
+    {
+      "epoch": 94.25,
+      "learning_rate": 2.8923594080547753e-06,
+      "loss": 1.7719,
+      "step": 19023500
+    },
+    {
+      "epoch": 94.25,
+      "learning_rate": 2.8911208216286918e-06,
+      "loss": 1.7711,
+      "step": 19024000
+    },
+    {
+      "epoch": 94.25,
+      "learning_rate": 2.8898822352026083e-06,
+      "loss": 1.8014,
+      "step": 19024500
+    },
+    {
+      "epoch": 94.26,
+      "learning_rate": 2.8886436487765243e-06,
+      "loss": 1.787,
+      "step": 19025000
+    },
+    {
+      "epoch": 94.26,
+      "learning_rate": 2.887405062350441e-06,
+      "loss": 1.7888,
+      "step": 19025500
+    },
+    {
+      "epoch": 94.26,
+      "learning_rate": 2.8861689530972092e-06,
+      "loss": 1.7918,
+      "step": 19026000
+    },
+    {
+      "epoch": 94.26,
+      "learning_rate": 2.8849303666711257e-06,
+      "loss": 1.7551,
+      "step": 19026500
+    },
+    {
+      "epoch": 94.27,
+      "learning_rate": 2.883691780245042e-06,
+      "loss": 1.7809,
+      "step": 19027000
+    },
+    {
+      "epoch": 94.27,
+      "learning_rate": 2.8824531938189583e-06,
+      "loss": 1.7595,
+      "step": 19027500
+    },
+    {
+      "epoch": 94.27,
+      "learning_rate": 2.881214607392875e-06,
+      "loss": 1.7917,
+      "step": 19028000
+    },
+    {
+      "epoch": 94.27,
+      "learning_rate": 2.879976020966791e-06,
+      "loss": 1.7628,
+      "step": 19028500
+    },
+    {
+      "epoch": 94.28,
+      "learning_rate": 2.878737434540708e-06,
+      "loss": 1.7992,
+      "step": 19029000
+    },
+    {
+      "epoch": 94.28,
+      "learning_rate": 2.877498848114624e-06,
+      "loss": 1.7884,
+      "step": 19029500
+    },
+    {
+      "epoch": 94.28,
+      "learning_rate": 2.8762602616885404e-06,
+      "loss": 1.782,
+      "step": 19030000
+    },
+    {
+      "epoch": 94.28,
+      "learning_rate": 2.8750216752624565e-06,
+      "loss": 1.8174,
+      "step": 19030500
+    },
+    {
+      "epoch": 94.29,
+      "learning_rate": 2.873783088836373e-06,
+      "loss": 1.7862,
+      "step": 19031000
+    },
+    {
+      "epoch": 94.29,
+      "learning_rate": 2.8725445024102895e-06,
+      "loss": 1.7875,
+      "step": 19031500
+    },
+    {
+      "epoch": 94.29,
+      "learning_rate": 2.8713059159842056e-06,
+      "loss": 1.7931,
+      "step": 19032000
+    },
+    {
+      "epoch": 94.29,
+      "learning_rate": 2.8700698067309744e-06,
+      "loss": 1.7708,
+      "step": 19032500
+    },
+    {
+      "epoch": 94.3,
+      "learning_rate": 2.8688312203048905e-06,
+      "loss": 1.8122,
+      "step": 19033000
+    },
+    {
+      "epoch": 94.3,
+      "learning_rate": 2.867592633878807e-06,
+      "loss": 1.8106,
+      "step": 19033500
+    },
+    {
+      "epoch": 94.3,
+      "learning_rate": 2.8663565246255754e-06,
+      "loss": 1.7678,
+      "step": 19034000
+    },
+    {
+      "epoch": 94.3,
+      "learning_rate": 2.865117938199492e-06,
+      "loss": 1.7791,
+      "step": 19034500
+    },
+    {
+      "epoch": 94.31,
+      "learning_rate": 2.8638793517734084e-06,
+      "loss": 1.7877,
+      "step": 19035000
+    },
+    {
+      "epoch": 94.31,
+      "learning_rate": 2.8626407653473245e-06,
+      "loss": 1.7883,
+      "step": 19035500
+    },
+    {
+      "epoch": 94.31,
+      "learning_rate": 2.861402178921241e-06,
+      "loss": 1.7926,
+      "step": 19036000
+    },
+    {
+      "epoch": 94.31,
+      "learning_rate": 2.8601660696680094e-06,
+      "loss": 1.7699,
+      "step": 19036500
+    },
+    {
+      "epoch": 94.32,
+      "learning_rate": 2.858929960414778e-06,
+      "loss": 1.7886,
+      "step": 19037000
+    },
+    {
+      "epoch": 94.32,
+      "learning_rate": 2.8576913739886943e-06,
+      "loss": 1.8086,
+      "step": 19037500
+    },
+    {
+      "epoch": 94.32,
+      "learning_rate": 2.8564527875626108e-06,
+      "loss": 1.7875,
+      "step": 19038000
+    },
+    {
+      "epoch": 94.32,
+      "learning_rate": 2.855214201136527e-06,
+      "loss": 1.7865,
+      "step": 19038500
+    },
+    {
+      "epoch": 94.33,
+      "learning_rate": 2.8539756147104433e-06,
+      "loss": 1.7723,
+      "step": 19039000
+    },
+    {
+      "epoch": 94.33,
+      "learning_rate": 2.85273702828436e-06,
+      "loss": 1.7876,
+      "step": 19039500
+    },
+    {
+      "epoch": 94.33,
+      "learning_rate": 2.8515009190311282e-06,
+      "loss": 1.789,
+      "step": 19040000
+    },
+    {
+      "epoch": 94.33,
+      "learning_rate": 2.8502623326050447e-06,
+      "loss": 1.7817,
+      "step": 19040500
+    },
+    {
+      "epoch": 94.34,
+      "learning_rate": 2.849023746178961e-06,
+      "loss": 1.776,
+      "step": 19041000
+    },
+    {
+      "epoch": 94.34,
+      "learning_rate": 2.8477851597528773e-06,
+      "loss": 1.7838,
+      "step": 19041500
+    },
+    {
+      "epoch": 94.34,
+      "learning_rate": 2.8465465733267934e-06,
+      "loss": 1.7771,
+      "step": 19042000
+    },
+    {
+      "epoch": 94.34,
+      "learning_rate": 2.84530798690071e-06,
+      "loss": 1.7854,
+      "step": 19042500
+    },
+    {
+      "epoch": 94.35,
+      "learning_rate": 2.8440718776474787e-06,
+      "loss": 1.7869,
+      "step": 19043000
+    },
+    {
+      "epoch": 94.35,
+      "learning_rate": 2.842833291221395e-06,
+      "loss": 1.7829,
+      "step": 19043500
+    },
+    {
+      "epoch": 94.35,
+      "learning_rate": 2.8415947047953113e-06,
+      "loss": 1.7785,
+      "step": 19044000
+    },
+    {
+      "epoch": 94.35,
+      "learning_rate": 2.8403561183692274e-06,
+      "loss": 1.7703,
+      "step": 19044500
+    },
+    {
+      "epoch": 94.36,
+      "learning_rate": 2.8391175319431443e-06,
+      "loss": 1.7743,
+      "step": 19045000
+    },
+    {
+      "epoch": 94.36,
+      "learning_rate": 2.8378789455170604e-06,
+      "loss": 1.7859,
+      "step": 19045500
+    },
+    {
+      "epoch": 94.36,
+      "learning_rate": 2.836640359090977e-06,
+      "loss": 1.7851,
+      "step": 19046000
+    },
+    {
+      "epoch": 94.36,
+      "learning_rate": 2.8354017726648934e-06,
+      "loss": 1.7743,
+      "step": 19046500
+    },
+    {
+      "epoch": 94.37,
+      "learning_rate": 2.8341631862388095e-06,
+      "loss": 1.7792,
+      "step": 19047000
+    },
+    {
+      "epoch": 94.37,
+      "learning_rate": 2.832924599812726e-06,
+      "loss": 1.785,
+      "step": 19047500
+    },
+    {
+      "epoch": 94.37,
+      "learning_rate": 2.831686013386642e-06,
+      "loss": 1.7986,
+      "step": 19048000
+    },
+    {
+      "epoch": 94.37,
+      "learning_rate": 2.830449904133411e-06,
+      "loss": 1.7905,
+      "step": 19048500
+    },
+    {
+      "epoch": 94.38,
+      "learning_rate": 2.829211317707327e-06,
+      "loss": 1.7828,
+      "step": 19049000
+    },
+    {
+      "epoch": 94.38,
+      "learning_rate": 2.8279727312812435e-06,
+      "loss": 1.7617,
+      "step": 19049500
+    },
+    {
+      "epoch": 94.38,
+      "learning_rate": 2.82673414485516e-06,
+      "loss": 1.784,
+      "step": 19050000
+    },
+    {
+      "epoch": 94.38,
+      "learning_rate": 2.825495558429076e-06,
+      "loss": 1.7956,
+      "step": 19050500
+    },
+    {
+      "epoch": 94.39,
+      "learning_rate": 2.8242569720029926e-06,
+      "loss": 1.7666,
+      "step": 19051000
+    },
+    {
+      "epoch": 94.39,
+      "learning_rate": 2.8230183855769086e-06,
+      "loss": 1.8154,
+      "step": 19051500
+    },
+    {
+      "epoch": 94.39,
+      "learning_rate": 2.8217822763236774e-06,
+      "loss": 1.8179,
+      "step": 19052000
+    },
+    {
+      "epoch": 94.39,
+      "learning_rate": 2.820546167070446e-06,
+      "loss": 1.7849,
+      "step": 19052500
+    },
+    {
+      "epoch": 94.4,
+      "learning_rate": 2.8193075806443623e-06,
+      "loss": 1.7849,
+      "step": 19053000
+    },
+    {
+      "epoch": 94.4,
+      "learning_rate": 2.818068994218279e-06,
+      "loss": 1.7889,
+      "step": 19053500
+    },
+    {
+      "epoch": 94.4,
+      "learning_rate": 2.8168328849650472e-06,
+      "loss": 1.7754,
+      "step": 19054000
+    },
+    {
+      "epoch": 94.4,
+      "learning_rate": 2.8155942985389637e-06,
+      "loss": 1.8015,
+      "step": 19054500
+    },
+    {
+      "epoch": 94.41,
+      "learning_rate": 2.81435571211288e-06,
+      "loss": 1.7805,
+      "step": 19055000
+    },
+    {
+      "epoch": 94.41,
+      "learning_rate": 2.8131171256867963e-06,
+      "loss": 1.8082,
+      "step": 19055500
+    },
+    {
+      "epoch": 94.41,
+      "learning_rate": 2.8118785392607124e-06,
+      "loss": 1.7813,
+      "step": 19056000
+    },
+    {
+      "epoch": 94.41,
+      "learning_rate": 2.810639952834629e-06,
+      "loss": 1.8022,
+      "step": 19056500
+    },
+    {
+      "epoch": 94.41,
+      "learning_rate": 2.8094013664085454e-06,
+      "loss": 1.7762,
+      "step": 19057000
+    },
+    {
+      "epoch": 94.42,
+      "learning_rate": 2.808162779982462e-06,
+      "loss": 1.7979,
+      "step": 19057500
+    },
+    {
+      "epoch": 94.42,
+      "learning_rate": 2.8069266707292303e-06,
+      "loss": 1.764,
+      "step": 19058000
+    },
+    {
+      "epoch": 94.42,
+      "learning_rate": 2.8056880843031464e-06,
+      "loss": 1.7663,
+      "step": 19058500
+    },
+    {
+      "epoch": 94.42,
+      "learning_rate": 2.8044494978770633e-06,
+      "loss": 1.7839,
+      "step": 19059000
+    },
+    {
+      "epoch": 94.43,
+      "learning_rate": 2.8032109114509794e-06,
+      "loss": 1.7816,
+      "step": 19059500
+    },
+    {
+      "epoch": 94.43,
+      "learning_rate": 2.801972325024896e-06,
+      "loss": 1.791,
+      "step": 19060000
+    },
+    {
+      "epoch": 94.43,
+      "learning_rate": 2.8007362157716643e-06,
+      "loss": 1.8139,
+      "step": 19060500
+    },
+    {
+      "epoch": 94.43,
+      "learning_rate": 2.7994976293455808e-06,
+      "loss": 1.7884,
+      "step": 19061000
+    },
+    {
+      "epoch": 94.44,
+      "learning_rate": 2.798259042919497e-06,
+      "loss": 1.7868,
+      "step": 19061500
+    },
+    {
+      "epoch": 94.44,
+      "learning_rate": 2.7970204564934134e-06,
+      "loss": 1.7891,
+      "step": 19062000
+    },
+    {
+      "epoch": 94.44,
+      "learning_rate": 2.79578187006733e-06,
+      "loss": 1.7903,
+      "step": 19062500
+    },
+    {
+      "epoch": 94.44,
+      "learning_rate": 2.794543283641246e-06,
+      "loss": 1.7806,
+      "step": 19063000
+    },
+    {
+      "epoch": 94.45,
+      "learning_rate": 2.7933046972151625e-06,
+      "loss": 1.7768,
+      "step": 19063500
+    },
+    {
+      "epoch": 94.45,
+      "learning_rate": 2.7920661107890785e-06,
+      "loss": 1.8142,
+      "step": 19064000
+    },
+    {
+      "epoch": 94.45,
+      "learning_rate": 2.790827524362995e-06,
+      "loss": 1.7767,
+      "step": 19064500
+    },
+    {
+      "epoch": 94.45,
+      "learning_rate": 2.7895889379369115e-06,
+      "loss": 1.7811,
+      "step": 19065000
+    },
+    {
+      "epoch": 94.46,
+      "learning_rate": 2.7883503515108276e-06,
+      "loss": 1.7939,
+      "step": 19065500
+    },
+    {
+      "epoch": 94.46,
+      "learning_rate": 2.787111765084744e-06,
+      "loss": 1.7768,
+      "step": 19066000
+    },
+    {
+      "epoch": 94.46,
+      "learning_rate": 2.7858731786586606e-06,
+      "loss": 1.7838,
+      "step": 19066500
+    },
+    {
+      "epoch": 94.46,
+      "learning_rate": 2.784634592232577e-06,
+      "loss": 1.8244,
+      "step": 19067000
+    },
+    {
+      "epoch": 94.47,
+      "learning_rate": 2.7833960058064932e-06,
+      "loss": 1.8021,
+      "step": 19067500
+    },
+    {
+      "epoch": 94.47,
+      "learning_rate": 2.7821574193804097e-06,
+      "loss": 1.7898,
+      "step": 19068000
+    },
+    {
+      "epoch": 94.47,
+      "learning_rate": 2.7809188329543262e-06,
+      "loss": 1.7873,
+      "step": 19068500
+    },
+    {
+      "epoch": 94.47,
+      "learning_rate": 2.7796802465282423e-06,
+      "loss": 1.7725,
+      "step": 19069000
+    },
+    {
+      "epoch": 94.48,
+      "learning_rate": 2.778444137275011e-06,
+      "loss": 1.753,
+      "step": 19069500
+    },
+    {
+      "epoch": 94.48,
+      "learning_rate": 2.777205550848927e-06,
+      "loss": 1.788,
+      "step": 19070000
+    },
+    {
+      "epoch": 94.48,
+      "learning_rate": 2.7759669644228437e-06,
+      "loss": 1.7931,
+      "step": 19070500
+    },
+    {
+      "epoch": 94.48,
+      "learning_rate": 2.77472837799676e-06,
+      "loss": 1.7979,
+      "step": 19071000
+    },
+    {
+      "epoch": 94.49,
+      "learning_rate": 2.7734922687435286e-06,
+      "loss": 1.7824,
+      "step": 19071500
+    },
+    {
+      "epoch": 94.49,
+      "learning_rate": 2.772253682317445e-06,
+      "loss": 1.7889,
+      "step": 19072000
+    },
+    {
+      "epoch": 94.49,
+      "learning_rate": 2.771015095891361e-06,
+      "loss": 1.7819,
+      "step": 19072500
+    },
+    {
+      "epoch": 94.49,
+      "learning_rate": 2.7697765094652777e-06,
+      "loss": 1.7773,
+      "step": 19073000
+    },
+    {
+      "epoch": 94.5,
+      "learning_rate": 2.7685379230391938e-06,
+      "loss": 1.7764,
+      "step": 19073500
+    },
+    {
+      "epoch": 94.5,
+      "learning_rate": 2.7673018137859626e-06,
+      "loss": 1.7796,
+      "step": 19074000
+    },
+    {
+      "epoch": 94.5,
+      "learning_rate": 2.7660632273598787e-06,
+      "loss": 1.789,
+      "step": 19074500
+    },
+    {
+      "epoch": 94.5,
+      "learning_rate": 2.764824640933795e-06,
+      "loss": 1.7888,
+      "step": 19075000
+    },
+    {
+      "epoch": 94.51,
+      "learning_rate": 2.7635860545077117e-06,
+      "loss": 1.7988,
+      "step": 19075500
+    },
+    {
+      "epoch": 94.51,
+      "learning_rate": 2.7623474680816278e-06,
+      "loss": 1.7856,
+      "step": 19076000
+    },
+    {
+      "epoch": 94.51,
+      "learning_rate": 2.7611088816555443e-06,
+      "loss": 1.7791,
+      "step": 19076500
+    },
+    {
+      "epoch": 94.51,
+      "learning_rate": 2.7598702952294603e-06,
+      "loss": 1.8048,
+      "step": 19077000
+    },
+    {
+      "epoch": 94.52,
+      "learning_rate": 2.7586317088033773e-06,
+      "loss": 1.7938,
+      "step": 19077500
+    },
+    {
+      "epoch": 94.52,
+      "learning_rate": 2.7573931223772933e-06,
+      "loss": 1.7729,
+      "step": 19078000
+    },
+    {
+      "epoch": 94.52,
+      "learning_rate": 2.75615453595121e-06,
+      "loss": 1.7658,
+      "step": 19078500
+    },
+    {
+      "epoch": 94.52,
+      "learning_rate": 2.7549184266979782e-06,
+      "loss": 1.7645,
+      "step": 19079000
+    },
+    {
+      "epoch": 94.53,
+      "learning_rate": 2.7536798402718947e-06,
+      "loss": 1.7661,
+      "step": 19079500
+    },
+    {
+      "epoch": 94.53,
+      "learning_rate": 2.752443731018663e-06,
+      "loss": 1.8053,
+      "step": 19080000
+    },
+    {
+      "epoch": 94.53,
+      "learning_rate": 2.7512051445925796e-06,
+      "loss": 1.7842,
+      "step": 19080500
+    },
+    {
+      "epoch": 94.53,
+      "learning_rate": 2.749966558166496e-06,
+      "loss": 1.769,
+      "step": 19081000
+    },
+    {
+      "epoch": 94.54,
+      "learning_rate": 2.7487279717404122e-06,
+      "loss": 1.7903,
+      "step": 19081500
+    },
+    {
+      "epoch": 94.54,
+      "learning_rate": 2.7474893853143287e-06,
+      "loss": 1.7793,
+      "step": 19082000
+    },
+    {
+      "epoch": 94.54,
+      "learning_rate": 2.746250798888245e-06,
+      "loss": 1.7721,
+      "step": 19082500
+    },
+    {
+      "epoch": 94.54,
+      "learning_rate": 2.7450122124621613e-06,
+      "loss": 1.7806,
+      "step": 19083000
+    },
+    {
+      "epoch": 94.55,
+      "learning_rate": 2.743773626036078e-06,
+      "loss": 1.7845,
+      "step": 19083500
+    },
+    {
+      "epoch": 94.55,
+      "learning_rate": 2.742537516782846e-06,
+      "loss": 1.8153,
+      "step": 19084000
+    },
+    {
+      "epoch": 94.55,
+      "learning_rate": 2.7412989303567627e-06,
+      "loss": 1.7922,
+      "step": 19084500
+    },
+    {
+      "epoch": 94.55,
+      "learning_rate": 2.740062821103531e-06,
+      "loss": 1.7773,
+      "step": 19085000
+    },
+    {
+      "epoch": 94.56,
+      "learning_rate": 2.7388242346774476e-06,
+      "loss": 1.7878,
+      "step": 19085500
+    },
+    {
+      "epoch": 94.56,
+      "learning_rate": 2.7375856482513637e-06,
+      "loss": 1.7595,
+      "step": 19086000
+    },
+    {
+      "epoch": 94.56,
+      "learning_rate": 2.73634706182528e-06,
+      "loss": 1.7657,
+      "step": 19086500
+    },
+    {
+      "epoch": 94.56,
+      "learning_rate": 2.7351084753991967e-06,
+      "loss": 1.8135,
+      "step": 19087000
+    },
+    {
+      "epoch": 94.57,
+      "learning_rate": 2.7338698889731128e-06,
+      "loss": 1.7815,
+      "step": 19087500
+    },
+    {
+      "epoch": 94.57,
+      "learning_rate": 2.7326313025470293e-06,
+      "loss": 1.7969,
+      "step": 19088000
+    },
+    {
+      "epoch": 94.57,
+      "learning_rate": 2.7313927161209454e-06,
+      "loss": 1.7668,
+      "step": 19088500
+    },
+    {
+      "epoch": 94.57,
+      "learning_rate": 2.730154129694862e-06,
+      "loss": 1.7799,
+      "step": 19089000
+    },
+    {
+      "epoch": 94.58,
+      "learning_rate": 2.7289155432687784e-06,
+      "loss": 1.7694,
+      "step": 19089500
+    },
+    {
+      "epoch": 94.58,
+      "learning_rate": 2.727676956842695e-06,
+      "loss": 1.7835,
+      "step": 19090000
+    },
+    {
+      "epoch": 94.58,
+      "learning_rate": 2.7264383704166114e-06,
+      "loss": 1.7843,
+      "step": 19090500
+    },
+    {
+      "epoch": 94.58,
+      "learning_rate": 2.7252022611633793e-06,
+      "loss": 1.8114,
+      "step": 19091000
+    },
+    {
+      "epoch": 94.59,
+      "learning_rate": 2.723966151910148e-06,
+      "loss": 1.7716,
+      "step": 19091500
+    },
+    {
+      "epoch": 94.59,
+      "learning_rate": 2.7227275654840642e-06,
+      "loss": 1.8063,
+      "step": 19092000
+    },
+    {
+      "epoch": 94.59,
+      "learning_rate": 2.7214889790579807e-06,
+      "loss": 1.7947,
+      "step": 19092500
+    },
+    {
+      "epoch": 94.59,
+      "learning_rate": 2.7202503926318972e-06,
+      "loss": 1.7934,
+      "step": 19093000
+    },
+    {
+      "epoch": 94.6,
+      "learning_rate": 2.7190118062058137e-06,
+      "loss": 1.7819,
+      "step": 19093500
+    },
+    {
+      "epoch": 94.6,
+      "learning_rate": 2.71777321977973e-06,
+      "loss": 1.7793,
+      "step": 19094000
+    },
+    {
+      "epoch": 94.6,
+      "learning_rate": 2.7165371105264986e-06,
+      "loss": 1.7779,
+      "step": 19094500
+    },
+    {
+      "epoch": 94.6,
+      "learning_rate": 2.715301001273267e-06,
+      "loss": 1.7786,
+      "step": 19095000
+    },
+    {
+      "epoch": 94.61,
+      "learning_rate": 2.714062414847183e-06,
+      "loss": 1.7944,
+      "step": 19095500
+    },
+    {
+      "epoch": 94.61,
+      "learning_rate": 2.7128238284210996e-06,
+      "loss": 1.8044,
+      "step": 19096000
+    },
+    {
+      "epoch": 94.61,
+      "learning_rate": 2.711585241995016e-06,
+      "loss": 1.789,
+      "step": 19096500
+    },
+    {
+      "epoch": 94.61,
+      "learning_rate": 2.7103466555689326e-06,
+      "loss": 1.7962,
+      "step": 19097000
+    },
+    {
+      "epoch": 94.62,
+      "learning_rate": 2.7091080691428487e-06,
+      "loss": 1.7924,
+      "step": 19097500
+    },
+    {
+      "epoch": 94.62,
+      "learning_rate": 2.707869482716765e-06,
+      "loss": 1.7919,
+      "step": 19098000
+    },
+    {
+      "epoch": 94.62,
+      "learning_rate": 2.7066308962906817e-06,
+      "loss": 1.7673,
+      "step": 19098500
+    },
+    {
+      "epoch": 94.62,
+      "learning_rate": 2.70539478703745e-06,
+      "loss": 1.7832,
+      "step": 19099000
+    },
+    {
+      "epoch": 94.63,
+      "learning_rate": 2.7041562006113666e-06,
+      "loss": 1.7558,
+      "step": 19099500
+    },
+    {
+      "epoch": 94.63,
+      "learning_rate": 2.7029176141852827e-06,
+      "loss": 1.7939,
+      "step": 19100000
+    },
+    {
+      "epoch": 94.63,
+      "learning_rate": 2.701679027759199e-06,
+      "loss": 1.7735,
+      "step": 19100500
+    },
+    {
+      "epoch": 94.63,
+      "learning_rate": 2.7004404413331153e-06,
+      "loss": 1.7668,
+      "step": 19101000
+    },
+    {
+      "epoch": 94.64,
+      "learning_rate": 2.699204332079884e-06,
+      "loss": 1.7858,
+      "step": 19101500
+    },
+    {
+      "epoch": 94.64,
+      "learning_rate": 2.6979657456538006e-06,
+      "loss": 1.7556,
+      "step": 19102000
+    },
+    {
+      "epoch": 94.64,
+      "learning_rate": 2.6967271592277167e-06,
+      "loss": 1.7833,
+      "step": 19102500
+    },
+    {
+      "epoch": 94.64,
+      "learning_rate": 2.695488572801633e-06,
+      "loss": 1.7727,
+      "step": 19103000
+    },
+    {
+      "epoch": 94.65,
+      "learning_rate": 2.6942499863755492e-06,
+      "loss": 1.8078,
+      "step": 19103500
+    },
+    {
+      "epoch": 94.65,
+      "learning_rate": 2.6930113999494657e-06,
+      "loss": 1.777,
+      "step": 19104000
+    },
+    {
+      "epoch": 94.65,
+      "learning_rate": 2.691775290696234e-06,
+      "loss": 1.7803,
+      "step": 19104500
+    },
+    {
+      "epoch": 94.65,
+      "learning_rate": 2.6905367042701506e-06,
+      "loss": 1.768,
+      "step": 19105000
+    },
+    {
+      "epoch": 94.66,
+      "learning_rate": 2.689298117844067e-06,
+      "loss": 1.7802,
+      "step": 19105500
+    },
+    {
+      "epoch": 94.66,
+      "learning_rate": 2.6880620085908355e-06,
+      "loss": 1.7846,
+      "step": 19106000
+    },
+    {
+      "epoch": 94.66,
+      "learning_rate": 2.686823422164752e-06,
+      "loss": 1.8075,
+      "step": 19106500
+    },
+    {
+      "epoch": 94.66,
+      "learning_rate": 2.685584835738668e-06,
+      "loss": 1.7818,
+      "step": 19107000
+    },
+    {
+      "epoch": 94.67,
+      "learning_rate": 2.6843462493125846e-06,
+      "loss": 1.7729,
+      "step": 19107500
+    },
+    {
+      "epoch": 94.67,
+      "learning_rate": 2.6831076628865007e-06,
+      "loss": 1.787,
+      "step": 19108000
+    },
+    {
+      "epoch": 94.67,
+      "learning_rate": 2.681869076460417e-06,
+      "loss": 1.7819,
+      "step": 19108500
+    },
+    {
+      "epoch": 94.67,
+      "learning_rate": 2.6806304900343337e-06,
+      "loss": 1.7744,
+      "step": 19109000
+    },
+    {
+      "epoch": 94.68,
+      "learning_rate": 2.67939190360825e-06,
+      "loss": 1.7872,
+      "step": 19109500
+    },
+    {
+      "epoch": 94.68,
+      "learning_rate": 2.6781533171821667e-06,
+      "loss": 1.7856,
+      "step": 19110000
+    },
+    {
+      "epoch": 94.68,
+      "learning_rate": 2.676917207928935e-06,
+      "loss": 1.7854,
+      "step": 19110500
+    },
+    {
+      "epoch": 94.68,
+      "learning_rate": 2.6756786215028516e-06,
+      "loss": 1.7898,
+      "step": 19111000
+    },
+    {
+      "epoch": 94.68,
+      "learning_rate": 2.6744400350767677e-06,
+      "loss": 1.7831,
+      "step": 19111500
+    },
+    {
+      "epoch": 94.69,
+      "learning_rate": 2.673201448650684e-06,
+      "loss": 1.7926,
+      "step": 19112000
+    },
+    {
+      "epoch": 94.69,
+      "learning_rate": 2.6719628622246003e-06,
+      "loss": 1.7796,
+      "step": 19112500
+    },
+    {
+      "epoch": 94.69,
+      "learning_rate": 2.670726752971369e-06,
+      "loss": 1.7571,
+      "step": 19113000
+    },
+    {
+      "epoch": 94.69,
+      "learning_rate": 2.6694881665452856e-06,
+      "loss": 1.7926,
+      "step": 19113500
+    },
+    {
+      "epoch": 94.7,
+      "learning_rate": 2.6682495801192017e-06,
+      "loss": 1.7767,
+      "step": 19114000
+    },
+    {
+      "epoch": 94.7,
+      "learning_rate": 2.667010993693118e-06,
+      "loss": 1.786,
+      "step": 19114500
+    },
+    {
+      "epoch": 94.7,
+      "learning_rate": 2.6657724072670343e-06,
+      "loss": 1.7915,
+      "step": 19115000
+    },
+    {
+      "epoch": 94.7,
+      "learning_rate": 2.664536298013803e-06,
+      "loss": 1.7664,
+      "step": 19115500
+    },
+    {
+      "epoch": 94.71,
+      "learning_rate": 2.6633001887605714e-06,
+      "loss": 1.7858,
+      "step": 19116000
+    },
+    {
+      "epoch": 94.71,
+      "learning_rate": 2.662061602334488e-06,
+      "loss": 1.7976,
+      "step": 19116500
+    },
+    {
+      "epoch": 94.71,
+      "learning_rate": 2.660823015908404e-06,
+      "loss": 1.7741,
+      "step": 19117000
+    },
+    {
+      "epoch": 94.71,
+      "learning_rate": 2.659586906655173e-06,
+      "loss": 1.795,
+      "step": 19117500
+    },
+    {
+      "epoch": 94.72,
+      "learning_rate": 2.6583483202290893e-06,
+      "loss": 1.7809,
+      "step": 19118000
+    },
+    {
+      "epoch": 94.72,
+      "learning_rate": 2.6571097338030054e-06,
+      "loss": 1.7669,
+      "step": 19118500
+    },
+    {
+      "epoch": 94.72,
+      "learning_rate": 2.655871147376922e-06,
+      "loss": 1.7951,
+      "step": 19119000
+    },
+    {
+      "epoch": 94.72,
+      "learning_rate": 2.654632560950838e-06,
+      "loss": 1.7775,
+      "step": 19119500
+    },
+    {
+      "epoch": 94.73,
+      "learning_rate": 2.6533939745247545e-06,
+      "loss": 1.7972,
+      "step": 19120000
+    },
+    {
+      "epoch": 94.73,
+      "learning_rate": 2.652155388098671e-06,
+      "loss": 1.7852,
+      "step": 19120500
+    },
+    {
+      "epoch": 94.73,
+      "learning_rate": 2.650916801672587e-06,
+      "loss": 1.777,
+      "step": 19121000
+    },
+    {
+      "epoch": 94.73,
+      "learning_rate": 2.6496782152465036e-06,
+      "loss": 1.8212,
+      "step": 19121500
+    },
+    {
+      "epoch": 94.74,
+      "learning_rate": 2.6484396288204197e-06,
+      "loss": 1.7855,
+      "step": 19122000
+    },
+    {
+      "epoch": 94.74,
+      "learning_rate": 2.647201042394336e-06,
+      "loss": 1.7689,
+      "step": 19122500
+    },
+    {
+      "epoch": 94.74,
+      "learning_rate": 2.6459624559682527e-06,
+      "loss": 1.8029,
+      "step": 19123000
+    },
+    {
+      "epoch": 94.74,
+      "learning_rate": 2.644723869542169e-06,
+      "loss": 1.7764,
+      "step": 19123500
+    },
+    {
+      "epoch": 94.75,
+      "learning_rate": 2.6434852831160853e-06,
+      "loss": 1.7802,
+      "step": 19124000
+    },
+    {
+      "epoch": 94.75,
+      "learning_rate": 2.642246696690002e-06,
+      "loss": 1.778,
+      "step": 19124500
+    },
+    {
+      "epoch": 94.75,
+      "learning_rate": 2.6410081102639183e-06,
+      "loss": 1.7692,
+      "step": 19125000
+    },
+    {
+      "epoch": 94.75,
+      "learning_rate": 2.6397695238378344e-06,
+      "loss": 1.779,
+      "step": 19125500
+    },
+    {
+      "epoch": 94.76,
+      "learning_rate": 2.638530937411751e-06,
+      "loss": 1.776,
+      "step": 19126000
+    },
+    {
+      "epoch": 94.76,
+      "learning_rate": 2.637292350985667e-06,
+      "loss": 1.7913,
+      "step": 19126500
+    },
+    {
+      "epoch": 94.76,
+      "learning_rate": 2.6360562417324358e-06,
+      "loss": 1.8007,
+      "step": 19127000
+    },
+    {
+      "epoch": 94.76,
+      "learning_rate": 2.6348176553063523e-06,
+      "loss": 1.794,
+      "step": 19127500
+    },
+    {
+      "epoch": 94.77,
+      "learning_rate": 2.6335790688802684e-06,
+      "loss": 1.7763,
+      "step": 19128000
+    },
+    {
+      "epoch": 94.77,
+      "learning_rate": 2.632340482454185e-06,
+      "loss": 1.7839,
+      "step": 19128500
+    },
+    {
+      "epoch": 94.77,
+      "learning_rate": 2.6311043732009532e-06,
+      "loss": 1.7907,
+      "step": 19129000
+    },
+    {
+      "epoch": 94.77,
+      "learning_rate": 2.6298657867748698e-06,
+      "loss": 1.7654,
+      "step": 19129500
+    },
+    {
+      "epoch": 94.78,
+      "learning_rate": 2.628627200348786e-06,
+      "loss": 1.7717,
+      "step": 19130000
+    },
+    {
+      "epoch": 94.78,
+      "learning_rate": 2.6273886139227023e-06,
+      "loss": 1.7832,
+      "step": 19130500
+    },
+    {
+      "epoch": 94.78,
+      "learning_rate": 2.626150027496619e-06,
+      "loss": 1.8032,
+      "step": 19131000
+    },
+    {
+      "epoch": 94.78,
+      "learning_rate": 2.624911441070535e-06,
+      "loss": 1.8002,
+      "step": 19131500
+    },
+    {
+      "epoch": 94.79,
+      "learning_rate": 2.623672854644452e-06,
+      "loss": 1.7984,
+      "step": 19132000
+    },
+    {
+      "epoch": 94.79,
+      "learning_rate": 2.622434268218368e-06,
+      "loss": 1.7995,
+      "step": 19132500
+    },
+    {
+      "epoch": 94.79,
+      "learning_rate": 2.6211956817922844e-06,
+      "loss": 1.7806,
+      "step": 19133000
+    },
+    {
+      "epoch": 94.79,
+      "learning_rate": 2.6199570953662005e-06,
+      "loss": 1.766,
+      "step": 19133500
+    },
+    {
+      "epoch": 94.8,
+      "learning_rate": 2.618718508940117e-06,
+      "loss": 1.7993,
+      "step": 19134000
+    },
+    {
+      "epoch": 94.8,
+      "learning_rate": 2.6174823996868854e-06,
+      "loss": 1.7861,
+      "step": 19134500
+    },
+    {
+      "epoch": 94.8,
+      "learning_rate": 2.616243813260802e-06,
+      "loss": 1.7629,
+      "step": 19135000
+    },
+    {
+      "epoch": 94.8,
+      "learning_rate": 2.6150052268347184e-06,
+      "loss": 1.7818,
+      "step": 19135500
+    },
+    {
+      "epoch": 94.81,
+      "learning_rate": 2.613769117581487e-06,
+      "loss": 1.7672,
+      "step": 19136000
+    },
+    {
+      "epoch": 94.81,
+      "learning_rate": 2.6125305311554033e-06,
+      "loss": 1.7833,
+      "step": 19136500
+    },
+    {
+      "epoch": 94.81,
+      "learning_rate": 2.6112919447293194e-06,
+      "loss": 1.8134,
+      "step": 19137000
+    },
+    {
+      "epoch": 94.81,
+      "learning_rate": 2.610055835476088e-06,
+      "loss": 1.7798,
+      "step": 19137500
+    },
+    {
+      "epoch": 94.82,
+      "learning_rate": 2.6088172490500043e-06,
+      "loss": 1.7839,
+      "step": 19138000
+    },
+    {
+      "epoch": 94.82,
+      "learning_rate": 2.6075786626239208e-06,
+      "loss": 1.78,
+      "step": 19138500
+    },
+    {
+      "epoch": 94.82,
+      "learning_rate": 2.6063400761978373e-06,
+      "loss": 1.7945,
+      "step": 19139000
+    },
+    {
+      "epoch": 94.82,
+      "learning_rate": 2.6051014897717534e-06,
+      "loss": 1.7941,
+      "step": 19139500
+    },
+    {
+      "epoch": 94.83,
+      "learning_rate": 2.60386290334567e-06,
+      "loss": 1.7754,
+      "step": 19140000
+    },
+    {
+      "epoch": 94.83,
+      "learning_rate": 2.602624316919586e-06,
+      "loss": 1.7676,
+      "step": 19140500
+    },
+    {
+      "epoch": 94.83,
+      "learning_rate": 2.6013857304935025e-06,
+      "loss": 1.7822,
+      "step": 19141000
+    },
+    {
+      "epoch": 94.83,
+      "learning_rate": 2.600147144067419e-06,
+      "loss": 1.8139,
+      "step": 19141500
+    },
+    {
+      "epoch": 94.84,
+      "learning_rate": 2.598908557641335e-06,
+      "loss": 1.7825,
+      "step": 19142000
+    },
+    {
+      "epoch": 94.84,
+      "learning_rate": 2.597672448388104e-06,
+      "loss": 1.7897,
+      "step": 19142500
+    },
+    {
+      "epoch": 94.84,
+      "learning_rate": 2.59643386196202e-06,
+      "loss": 1.7896,
+      "step": 19143000
+    },
+    {
+      "epoch": 94.84,
+      "learning_rate": 2.5951977527087887e-06,
+      "loss": 1.7815,
+      "step": 19143500
+    },
+    {
+      "epoch": 94.85,
+      "learning_rate": 2.593959166282705e-06,
+      "loss": 1.8005,
+      "step": 19144000
+    },
+    {
+      "epoch": 94.85,
+      "learning_rate": 2.5927205798566213e-06,
+      "loss": 1.7781,
+      "step": 19144500
+    },
+    {
+      "epoch": 94.85,
+      "learning_rate": 2.5914819934305374e-06,
+      "loss": 1.7633,
+      "step": 19145000
+    },
+    {
+      "epoch": 94.85,
+      "learning_rate": 2.590243407004454e-06,
+      "loss": 1.7878,
+      "step": 19145500
+    },
+    {
+      "epoch": 94.86,
+      "learning_rate": 2.5890072977512227e-06,
+      "loss": 1.7861,
+      "step": 19146000
+    },
+    {
+      "epoch": 94.86,
+      "learning_rate": 2.587768711325139e-06,
+      "loss": 1.8038,
+      "step": 19146500
+    },
+    {
+      "epoch": 94.86,
+      "learning_rate": 2.5865301248990553e-06,
+      "loss": 1.7828,
+      "step": 19147000
+    },
+    {
+      "epoch": 94.86,
+      "learning_rate": 2.5852915384729714e-06,
+      "loss": 1.7884,
+      "step": 19147500
+    },
+    {
+      "epoch": 94.87,
+      "learning_rate": 2.5840529520468883e-06,
+      "loss": 1.7782,
+      "step": 19148000
+    },
+    {
+      "epoch": 94.87,
+      "learning_rate": 2.5828143656208044e-06,
+      "loss": 1.7777,
+      "step": 19148500
+    },
+    {
+      "epoch": 94.87,
+      "learning_rate": 2.581575779194721e-06,
+      "loss": 1.7758,
+      "step": 19149000
+    },
+    {
+      "epoch": 94.87,
+      "learning_rate": 2.580337192768637e-06,
+      "loss": 1.7844,
+      "step": 19149500
+    },
+    {
+      "epoch": 94.88,
+      "learning_rate": 2.579101083515406e-06,
+      "loss": 1.8014,
+      "step": 19150000
+    },
+    {
+      "epoch": 94.88,
+      "learning_rate": 2.5778624970893223e-06,
+      "loss": 1.8092,
+      "step": 19150500
+    },
+    {
+      "epoch": 94.88,
+      "learning_rate": 2.5766239106632384e-06,
+      "loss": 1.7891,
+      "step": 19151000
+    },
+    {
+      "epoch": 94.88,
+      "learning_rate": 2.575385324237155e-06,
+      "loss": 1.7632,
+      "step": 19151500
+    },
+    {
+      "epoch": 94.89,
+      "learning_rate": 2.574146737811071e-06,
+      "loss": 1.7766,
+      "step": 19152000
+    },
+    {
+      "epoch": 94.89,
+      "learning_rate": 2.5729081513849875e-06,
+      "loss": 1.782,
+      "step": 19152500
+    },
+    {
+      "epoch": 94.89,
+      "learning_rate": 2.571669564958904e-06,
+      "loss": 1.7702,
+      "step": 19153000
+    },
+    {
+      "epoch": 94.89,
+      "learning_rate": 2.57043097853282e-06,
+      "loss": 1.7791,
+      "step": 19153500
+    },
+    {
+      "epoch": 94.9,
+      "learning_rate": 2.569194869279589e-06,
+      "loss": 1.7834,
+      "step": 19154000
+    },
+    {
+      "epoch": 94.9,
+      "learning_rate": 2.567956282853505e-06,
+      "loss": 1.7946,
+      "step": 19154500
+    },
+    {
+      "epoch": 94.9,
+      "learning_rate": 2.5667201736002738e-06,
+      "loss": 1.7825,
+      "step": 19155000
+    },
+    {
+      "epoch": 94.9,
+      "learning_rate": 2.56548158717419e-06,
+      "loss": 1.7861,
+      "step": 19155500
+    },
+    {
+      "epoch": 94.91,
+      "learning_rate": 2.5642430007481063e-06,
+      "loss": 1.7611,
+      "step": 19156000
+    },
+    {
+      "epoch": 94.91,
+      "learning_rate": 2.5630044143220224e-06,
+      "loss": 1.8016,
+      "step": 19156500
+    },
+    {
+      "epoch": 94.91,
+      "learning_rate": 2.561765827895939e-06,
+      "loss": 1.7969,
+      "step": 19157000
+    },
+    {
+      "epoch": 94.91,
+      "learning_rate": 2.5605272414698554e-06,
+      "loss": 1.7756,
+      "step": 19157500
+    },
+    {
+      "epoch": 94.92,
+      "learning_rate": 2.5592886550437715e-06,
+      "loss": 1.7743,
+      "step": 19158000
+    },
+    {
+      "epoch": 94.92,
+      "learning_rate": 2.5580500686176884e-06,
+      "loss": 1.7728,
+      "step": 19158500
+    },
+    {
+      "epoch": 94.92,
+      "learning_rate": 2.5568114821916045e-06,
+      "loss": 1.7987,
+      "step": 19159000
+    },
+    {
+      "epoch": 94.92,
+      "learning_rate": 2.555572895765521e-06,
+      "loss": 1.755,
+      "step": 19159500
+    },
+    {
+      "epoch": 94.93,
+      "learning_rate": 2.554336786512289e-06,
+      "loss": 1.7985,
+      "step": 19160000
+    },
+    {
+      "epoch": 94.93,
+      "learning_rate": 2.553098200086206e-06,
+      "loss": 1.7689,
+      "step": 19160500
+    },
+    {
+      "epoch": 94.93,
+      "learning_rate": 2.551859613660122e-06,
+      "loss": 1.7855,
+      "step": 19161000
+    },
+    {
+      "epoch": 94.93,
+      "learning_rate": 2.5506210272340385e-06,
+      "loss": 1.7808,
+      "step": 19161500
+    },
+    {
+      "epoch": 94.94,
+      "learning_rate": 2.5493849179808073e-06,
+      "loss": 1.779,
+      "step": 19162000
+    },
+    {
+      "epoch": 94.94,
+      "learning_rate": 2.5481463315547234e-06,
+      "loss": 1.8109,
+      "step": 19162500
+    },
+    {
+      "epoch": 94.94,
+      "learning_rate": 2.5469102223014918e-06,
+      "loss": 1.7911,
+      "step": 19163000
+    },
+    {
+      "epoch": 94.94,
+      "learning_rate": 2.5456716358754083e-06,
+      "loss": 1.8005,
+      "step": 19163500
+    },
+    {
+      "epoch": 94.95,
+      "learning_rate": 2.544433049449325e-06,
+      "loss": 1.7797,
+      "step": 19164000
+    },
+    {
+      "epoch": 94.95,
+      "learning_rate": 2.543194463023241e-06,
+      "loss": 1.7902,
+      "step": 19164500
+    },
+    {
+      "epoch": 94.95,
+      "learning_rate": 2.5419558765971574e-06,
+      "loss": 1.7855,
+      "step": 19165000
+    },
+    {
+      "epoch": 94.95,
+      "learning_rate": 2.540717290171074e-06,
+      "loss": 1.7704,
+      "step": 19165500
+    },
+    {
+      "epoch": 94.95,
+      "learning_rate": 2.53947870374499e-06,
+      "loss": 1.7872,
+      "step": 19166000
+    },
+    {
+      "epoch": 94.96,
+      "learning_rate": 2.5382401173189065e-06,
+      "loss": 1.7741,
+      "step": 19166500
+    },
+    {
+      "epoch": 94.96,
+      "learning_rate": 2.5370015308928226e-06,
+      "loss": 1.7751,
+      "step": 19167000
+    },
+    {
+      "epoch": 94.96,
+      "learning_rate": 2.535762944466739e-06,
+      "loss": 1.8031,
+      "step": 19167500
+    },
+    {
+      "epoch": 94.96,
+      "learning_rate": 2.5345243580406556e-06,
+      "loss": 1.7732,
+      "step": 19168000
+    },
+    {
+      "epoch": 94.97,
+      "learning_rate": 2.533288248787424e-06,
+      "loss": 1.7678,
+      "step": 19168500
+    },
+    {
+      "epoch": 94.97,
+      "learning_rate": 2.5320496623613405e-06,
+      "loss": 1.7752,
+      "step": 19169000
+    },
+    {
+      "epoch": 94.97,
+      "learning_rate": 2.5308110759352565e-06,
+      "loss": 1.7934,
+      "step": 19169500
+    },
+    {
+      "epoch": 94.97,
+      "learning_rate": 2.529572489509173e-06,
+      "loss": 1.7499,
+      "step": 19170000
+    },
+    {
+      "epoch": 94.98,
+      "learning_rate": 2.528333903083089e-06,
+      "loss": 1.7852,
+      "step": 19170500
+    },
+    {
+      "epoch": 94.98,
+      "learning_rate": 2.527095316657006e-06,
+      "loss": 1.7836,
+      "step": 19171000
+    },
+    {
+      "epoch": 94.98,
+      "learning_rate": 2.5258592074037744e-06,
+      "loss": 1.7839,
+      "step": 19171500
+    },
+    {
+      "epoch": 94.98,
+      "learning_rate": 2.5246206209776905e-06,
+      "loss": 1.7973,
+      "step": 19172000
+    },
+    {
+      "epoch": 94.99,
+      "learning_rate": 2.523382034551607e-06,
+      "loss": 1.783,
+      "step": 19172500
+    },
+    {
+      "epoch": 94.99,
+      "learning_rate": 2.5221434481255235e-06,
+      "loss": 1.7964,
+      "step": 19173000
+    },
+    {
+      "epoch": 94.99,
+      "learning_rate": 2.52090486169944e-06,
+      "loss": 1.781,
+      "step": 19173500
+    },
+    {
+      "epoch": 94.99,
+      "learning_rate": 2.519666275273356e-06,
+      "loss": 1.7983,
+      "step": 19174000
+    },
+    {
+      "epoch": 95.0,
+      "learning_rate": 2.518430166020125e-06,
+      "loss": 1.7738,
+      "step": 19174500
+    },
+    {
+      "epoch": 95.0,
+      "learning_rate": 2.517191579594041e-06,
+      "loss": 1.7774,
+      "step": 19175000
+    },
+    {
+      "epoch": 95.0,
+      "eval_accuracy": 0.6867611295760832,
+      "eval_accuracy_mlm": 0.6480042502386186,
+      "eval_accuracy_nsp": 0.8695947191509223,
+      "eval_loss": 2.3160958290100098,
+      "eval_runtime": 147.1337,
+      "eval_samples_per_second": 1732.839,
+      "eval_steps_per_second": 72.206,
+      "step": 19175085
+    },
+    {
+      "epoch": 95.0,
+      "learning_rate": 2.5159554703408094e-06,
+      "loss": 1.7699,
+      "step": 19175500
+    },
+    {
+      "epoch": 95.0,
+      "learning_rate": 2.514716883914726e-06,
+      "loss": 1.7926,
+      "step": 19176000
+    },
+    {
+      "epoch": 95.01,
+      "learning_rate": 2.5134782974886424e-06,
+      "loss": 1.766,
+      "step": 19176500
+    },
+    {
+      "epoch": 95.01,
+      "learning_rate": 2.512239711062559e-06,
+      "loss": 1.7971,
+      "step": 19177000
+    },
+    {
+      "epoch": 95.01,
+      "learning_rate": 2.511001124636475e-06,
+      "loss": 1.7513,
+      "step": 19177500
+    },
+    {
+      "epoch": 95.01,
+      "learning_rate": 2.5097625382103915e-06,
+      "loss": 1.7864,
+      "step": 19178000
+    },
+    {
+      "epoch": 95.02,
+      "learning_rate": 2.50852642895716e-06,
+      "loss": 1.7785,
+      "step": 19178500
+    },
+    {
+      "epoch": 95.02,
+      "learning_rate": 2.5072878425310764e-06,
+      "loss": 1.7682,
+      "step": 19179000
+    },
+    {
+      "epoch": 95.02,
+      "learning_rate": 2.5060492561049925e-06,
+      "loss": 1.79,
+      "step": 19179500
+    },
+    {
+      "epoch": 95.02,
+      "learning_rate": 2.504810669678909e-06,
+      "loss": 1.7668,
+      "step": 19180000
+    },
+    {
+      "epoch": 95.03,
+      "learning_rate": 2.5035720832528255e-06,
+      "loss": 1.7786,
+      "step": 19180500
+    },
+    {
+      "epoch": 95.03,
+      "learning_rate": 2.5023334968267415e-06,
+      "loss": 1.8015,
+      "step": 19181000
+    },
+    {
+      "epoch": 95.03,
+      "learning_rate": 2.5010973875735104e-06,
+      "loss": 1.7668,
+      "step": 19181500
+    },
+    {
+      "epoch": 95.03,
+      "learning_rate": 2.4998588011474264e-06,
+      "loss": 1.7944,
+      "step": 19182000
+    },
+    {
+      "epoch": 95.04,
+      "learning_rate": 2.498620214721343e-06,
+      "loss": 1.7903,
+      "step": 19182500
+    },
+    {
+      "epoch": 95.04,
+      "learning_rate": 2.4973816282952594e-06,
+      "loss": 1.7741,
+      "step": 19183000
+    },
+    {
+      "epoch": 95.04,
+      "learning_rate": 2.4961430418691755e-06,
+      "loss": 1.7717,
+      "step": 19183500
+    },
+    {
+      "epoch": 95.04,
+      "learning_rate": 2.494904455443092e-06,
+      "loss": 1.7897,
+      "step": 19184000
+    },
+    {
+      "epoch": 95.05,
+      "learning_rate": 2.493665869017008e-06,
+      "loss": 1.7654,
+      "step": 19184500
+    },
+    {
+      "epoch": 95.05,
+      "learning_rate": 2.4924272825909246e-06,
+      "loss": 1.7832,
+      "step": 19185000
+    },
+    {
+      "epoch": 95.05,
+      "learning_rate": 2.491188696164841e-06,
+      "loss": 1.7861,
+      "step": 19185500
+    },
+    {
+      "epoch": 95.05,
+      "learning_rate": 2.4899501097387576e-06,
+      "loss": 1.8033,
+      "step": 19186000
+    },
+    {
+      "epoch": 95.06,
+      "learning_rate": 2.4887115233126737e-06,
+      "loss": 1.7701,
+      "step": 19186500
+    },
+    {
+      "epoch": 95.06,
+      "learning_rate": 2.4874754140594425e-06,
+      "loss": 1.7972,
+      "step": 19187000
+    },
+    {
+      "epoch": 95.06,
+      "learning_rate": 2.486236827633359e-06,
+      "loss": 1.7841,
+      "step": 19187500
+    },
+    {
+      "epoch": 95.06,
+      "learning_rate": 2.484998241207275e-06,
+      "loss": 1.7831,
+      "step": 19188000
+    },
+    {
+      "epoch": 95.07,
+      "learning_rate": 2.4837596547811916e-06,
+      "loss": 1.7622,
+      "step": 19188500
+    },
+    {
+      "epoch": 95.07,
+      "learning_rate": 2.4825210683551077e-06,
+      "loss": 1.7803,
+      "step": 19189000
+    },
+    {
+      "epoch": 95.07,
+      "learning_rate": 2.481282481929024e-06,
+      "loss": 1.7915,
+      "step": 19189500
+    },
+    {
+      "epoch": 95.07,
+      "learning_rate": 2.4800438955029407e-06,
+      "loss": 1.7947,
+      "step": 19190000
+    },
+    {
+      "epoch": 95.08,
+      "learning_rate": 2.4788053090768568e-06,
+      "loss": 1.7716,
+      "step": 19190500
+    },
+    {
+      "epoch": 95.08,
+      "learning_rate": 2.4775691998236256e-06,
+      "loss": 1.7854,
+      "step": 19191000
+    },
+    {
+      "epoch": 95.08,
+      "learning_rate": 2.4763306133975417e-06,
+      "loss": 1.7624,
+      "step": 19191500
+    },
+    {
+      "epoch": 95.08,
+      "learning_rate": 2.475092026971458e-06,
+      "loss": 1.7697,
+      "step": 19192000
+    },
+    {
+      "epoch": 95.09,
+      "learning_rate": 2.4738534405453743e-06,
+      "loss": 1.774,
+      "step": 19192500
+    },
+    {
+      "epoch": 95.09,
+      "learning_rate": 2.4726148541192908e-06,
+      "loss": 1.7863,
+      "step": 19193000
+    },
+    {
+      "epoch": 95.09,
+      "learning_rate": 2.471378744866059e-06,
+      "loss": 1.7803,
+      "step": 19193500
+    },
+    {
+      "epoch": 95.09,
+      "learning_rate": 2.4701401584399757e-06,
+      "loss": 1.799,
+      "step": 19194000
+    },
+    {
+      "epoch": 95.1,
+      "learning_rate": 2.468901572013892e-06,
+      "loss": 1.767,
+      "step": 19194500
+    },
+    {
+      "epoch": 95.1,
+      "learning_rate": 2.4676629855878082e-06,
+      "loss": 1.781,
+      "step": 19195000
+    },
+    {
+      "epoch": 95.1,
+      "learning_rate": 2.4664243991617247e-06,
+      "loss": 1.7648,
+      "step": 19195500
+    },
+    {
+      "epoch": 95.1,
+      "learning_rate": 2.4651858127356412e-06,
+      "loss": 1.7906,
+      "step": 19196000
+    },
+    {
+      "epoch": 95.11,
+      "learning_rate": 2.4639472263095578e-06,
+      "loss": 1.7788,
+      "step": 19196500
+    },
+    {
+      "epoch": 95.11,
+      "learning_rate": 2.462708639883474e-06,
+      "loss": 1.7864,
+      "step": 19197000
+    },
+    {
+      "epoch": 95.11,
+      "learning_rate": 2.4614700534573903e-06,
+      "loss": 1.7809,
+      "step": 19197500
+    },
+    {
+      "epoch": 95.11,
+      "learning_rate": 2.460231467031307e-06,
+      "loss": 1.7882,
+      "step": 19198000
+    },
+    {
+      "epoch": 95.12,
+      "learning_rate": 2.4589953577780752e-06,
+      "loss": 1.7968,
+      "step": 19198500
+    },
+    {
+      "epoch": 95.12,
+      "learning_rate": 2.4577567713519917e-06,
+      "loss": 1.7892,
+      "step": 19199000
+    },
+    {
+      "epoch": 95.12,
+      "learning_rate": 2.456518184925908e-06,
+      "loss": 1.8076,
+      "step": 19199500
+    },
+    {
+      "epoch": 95.12,
+      "learning_rate": 2.4552795984998243e-06,
+      "loss": 1.7728,
+      "step": 19200000
+    },
+    {
+      "epoch": 95.13,
+      "learning_rate": 2.4540410120737404e-06,
+      "loss": 1.7617,
+      "step": 19200500
+    },
+    {
+      "epoch": 95.13,
+      "learning_rate": 2.452804902820509e-06,
+      "loss": 1.786,
+      "step": 19201000
+    },
+    {
+      "epoch": 95.13,
+      "learning_rate": 2.4515663163944257e-06,
+      "loss": 1.7917,
+      "step": 19201500
+    },
+    {
+      "epoch": 95.13,
+      "learning_rate": 2.450330207141194e-06,
+      "loss": 1.7594,
+      "step": 19202000
+    },
+    {
+      "epoch": 95.14,
+      "learning_rate": 2.449094097887963e-06,
+      "loss": 1.7743,
+      "step": 19202500
+    },
+    {
+      "epoch": 95.14,
+      "learning_rate": 2.447855511461879e-06,
+      "loss": 1.769,
+      "step": 19203000
+    },
+    {
+      "epoch": 95.14,
+      "learning_rate": 2.4466169250357955e-06,
+      "loss": 1.779,
+      "step": 19203500
+    },
+    {
+      "epoch": 95.14,
+      "learning_rate": 2.4453783386097116e-06,
+      "loss": 1.7958,
+      "step": 19204000
+    },
+    {
+      "epoch": 95.15,
+      "learning_rate": 2.444139752183628e-06,
+      "loss": 1.7733,
+      "step": 19204500
+    },
+    {
+      "epoch": 95.15,
+      "learning_rate": 2.442901165757544e-06,
+      "loss": 1.7757,
+      "step": 19205000
+    },
+    {
+      "epoch": 95.15,
+      "learning_rate": 2.4416625793314607e-06,
+      "loss": 1.7666,
+      "step": 19205500
+    },
+    {
+      "epoch": 95.15,
+      "learning_rate": 2.440423992905377e-06,
+      "loss": 1.8024,
+      "step": 19206000
+    },
+    {
+      "epoch": 95.16,
+      "learning_rate": 2.4391854064792933e-06,
+      "loss": 1.7743,
+      "step": 19206500
+    },
+    {
+      "epoch": 95.16,
+      "learning_rate": 2.4379468200532098e-06,
+      "loss": 1.7901,
+      "step": 19207000
+    },
+    {
+      "epoch": 95.16,
+      "learning_rate": 2.436710710799978e-06,
+      "loss": 1.7956,
+      "step": 19207500
+    },
+    {
+      "epoch": 95.16,
+      "learning_rate": 2.4354721243738946e-06,
+      "loss": 1.7703,
+      "step": 19208000
+    },
+    {
+      "epoch": 95.17,
+      "learning_rate": 2.434233537947811e-06,
+      "loss": 1.7647,
+      "step": 19208500
+    },
+    {
+      "epoch": 95.17,
+      "learning_rate": 2.4329974286945795e-06,
+      "loss": 1.7695,
+      "step": 19209000
+    },
+    {
+      "epoch": 95.17,
+      "learning_rate": 2.431761319441348e-06,
+      "loss": 1.7887,
+      "step": 19209500
+    },
+    {
+      "epoch": 95.17,
+      "learning_rate": 2.4305227330152644e-06,
+      "loss": 1.77,
+      "step": 19210000
+    },
+    {
+      "epoch": 95.18,
+      "learning_rate": 2.429284146589181e-06,
+      "loss": 1.7812,
+      "step": 19210500
+    },
+    {
+      "epoch": 95.18,
+      "learning_rate": 2.428045560163097e-06,
+      "loss": 1.8016,
+      "step": 19211000
+    },
+    {
+      "epoch": 95.18,
+      "learning_rate": 2.426809450909866e-06,
+      "loss": 1.7719,
+      "step": 19211500
+    },
+    {
+      "epoch": 95.18,
+      "learning_rate": 2.425570864483782e-06,
+      "loss": 1.7753,
+      "step": 19212000
+    },
+    {
+      "epoch": 95.19,
+      "learning_rate": 2.4243347552305507e-06,
+      "loss": 1.8011,
+      "step": 19212500
+    },
+    {
+      "epoch": 95.19,
+      "learning_rate": 2.423096168804467e-06,
+      "loss": 1.7846,
+      "step": 19213000
+    },
+    {
+      "epoch": 95.19,
+      "learning_rate": 2.4218575823783833e-06,
+      "loss": 1.776,
+      "step": 19213500
+    },
+    {
+      "epoch": 95.19,
+      "learning_rate": 2.4206189959523e-06,
+      "loss": 1.7702,
+      "step": 19214000
+    },
+    {
+      "epoch": 95.2,
+      "learning_rate": 2.419380409526216e-06,
+      "loss": 1.7859,
+      "step": 19214500
+    },
+    {
+      "epoch": 95.2,
+      "learning_rate": 2.4181418231001324e-06,
+      "loss": 1.7739,
+      "step": 19215000
+    },
+    {
+      "epoch": 95.2,
+      "learning_rate": 2.4169032366740485e-06,
+      "loss": 1.7756,
+      "step": 19215500
+    },
+    {
+      "epoch": 95.2,
+      "learning_rate": 2.415664650247965e-06,
+      "loss": 1.789,
+      "step": 19216000
+    },
+    {
+      "epoch": 95.21,
+      "learning_rate": 2.4144260638218815e-06,
+      "loss": 1.7904,
+      "step": 19216500
+    },
+    {
+      "epoch": 95.21,
+      "learning_rate": 2.413187477395798e-06,
+      "loss": 1.7745,
+      "step": 19217000
+    },
+    {
+      "epoch": 95.21,
+      "learning_rate": 2.4119488909697145e-06,
+      "loss": 1.7847,
+      "step": 19217500
+    },
+    {
+      "epoch": 95.21,
+      "learning_rate": 2.4107103045436306e-06,
+      "loss": 1.7718,
+      "step": 19218000
+    },
+    {
+      "epoch": 95.22,
+      "learning_rate": 2.409471718117547e-06,
+      "loss": 1.7868,
+      "step": 19218500
+    },
+    {
+      "epoch": 95.22,
+      "learning_rate": 2.408233131691463e-06,
+      "loss": 1.7799,
+      "step": 19219000
+    },
+    {
+      "epoch": 95.22,
+      "learning_rate": 2.4069945452653797e-06,
+      "loss": 1.7666,
+      "step": 19219500
+    },
+    {
+      "epoch": 95.22,
+      "learning_rate": 2.405755958839296e-06,
+      "loss": 1.7843,
+      "step": 19220000
+    },
+    {
+      "epoch": 95.23,
+      "learning_rate": 2.4045198495860646e-06,
+      "loss": 1.7739,
+      "step": 19220500
+    },
+    {
+      "epoch": 95.23,
+      "learning_rate": 2.403281263159981e-06,
+      "loss": 1.7625,
+      "step": 19221000
+    },
+    {
+      "epoch": 95.23,
+      "learning_rate": 2.402042676733897e-06,
+      "loss": 1.7715,
+      "step": 19221500
+    },
+    {
+      "epoch": 95.23,
+      "learning_rate": 2.4008040903078136e-06,
+      "loss": 1.7747,
+      "step": 19222000
+    },
+    {
+      "epoch": 95.23,
+      "learning_rate": 2.3995655038817297e-06,
+      "loss": 1.7736,
+      "step": 19222500
+    },
+    {
+      "epoch": 95.24,
+      "learning_rate": 2.3983269174556462e-06,
+      "loss": 1.7791,
+      "step": 19223000
+    },
+    {
+      "epoch": 95.24,
+      "learning_rate": 2.3970883310295627e-06,
+      "loss": 1.7653,
+      "step": 19223500
+    },
+    {
+      "epoch": 95.24,
+      "learning_rate": 2.395849744603479e-06,
+      "loss": 1.7886,
+      "step": 19224000
+    },
+    {
+      "epoch": 95.24,
+      "learning_rate": 2.3946111581773957e-06,
+      "loss": 1.7865,
+      "step": 19224500
+    },
+    {
+      "epoch": 95.25,
+      "learning_rate": 2.393372571751312e-06,
+      "loss": 1.7772,
+      "step": 19225000
+    },
+    {
+      "epoch": 95.25,
+      "learning_rate": 2.3921339853252283e-06,
+      "loss": 1.7801,
+      "step": 19225500
+    },
+    {
+      "epoch": 95.25,
+      "learning_rate": 2.3908953988991444e-06,
+      "loss": 1.7852,
+      "step": 19226000
+    },
+    {
+      "epoch": 95.25,
+      "learning_rate": 2.3896592896459132e-06,
+      "loss": 1.7986,
+      "step": 19226500
+    },
+    {
+      "epoch": 95.26,
+      "learning_rate": 2.3884207032198293e-06,
+      "loss": 1.7721,
+      "step": 19227000
+    },
+    {
+      "epoch": 95.26,
+      "learning_rate": 2.387182116793746e-06,
+      "loss": 1.7649,
+      "step": 19227500
+    },
+    {
+      "epoch": 95.26,
+      "learning_rate": 2.3859435303676623e-06,
+      "loss": 1.793,
+      "step": 19228000
+    },
+    {
+      "epoch": 95.26,
+      "learning_rate": 2.3847049439415784e-06,
+      "loss": 1.7781,
+      "step": 19228500
+    },
+    {
+      "epoch": 95.27,
+      "learning_rate": 2.383466357515495e-06,
+      "loss": 1.7878,
+      "step": 19229000
+    },
+    {
+      "epoch": 95.27,
+      "learning_rate": 2.3822302482622633e-06,
+      "loss": 1.7951,
+      "step": 19229500
+    },
+    {
+      "epoch": 95.27,
+      "learning_rate": 2.3809916618361798e-06,
+      "loss": 1.7914,
+      "step": 19230000
+    },
+    {
+      "epoch": 95.27,
+      "learning_rate": 2.379753075410096e-06,
+      "loss": 1.7843,
+      "step": 19230500
+    },
+    {
+      "epoch": 95.28,
+      "learning_rate": 2.3785144889840124e-06,
+      "loss": 1.7815,
+      "step": 19231000
+    },
+    {
+      "epoch": 95.28,
+      "learning_rate": 2.377275902557929e-06,
+      "loss": 1.7633,
+      "step": 19231500
+    },
+    {
+      "epoch": 95.28,
+      "learning_rate": 2.3760397933046973e-06,
+      "loss": 1.7666,
+      "step": 19232000
+    },
+    {
+      "epoch": 95.28,
+      "learning_rate": 2.3748012068786138e-06,
+      "loss": 1.7821,
+      "step": 19232500
+    },
+    {
+      "epoch": 95.29,
+      "learning_rate": 2.37356262045253e-06,
+      "loss": 1.7747,
+      "step": 19233000
+    },
+    {
+      "epoch": 95.29,
+      "learning_rate": 2.3723265111992987e-06,
+      "loss": 1.7888,
+      "step": 19233500
+    },
+    {
+      "epoch": 95.29,
+      "learning_rate": 2.3710879247732147e-06,
+      "loss": 1.7766,
+      "step": 19234000
+    },
+    {
+      "epoch": 95.29,
+      "learning_rate": 2.3698493383471312e-06,
+      "loss": 1.784,
+      "step": 19234500
+    },
+    {
+      "epoch": 95.3,
+      "learning_rate": 2.3686107519210477e-06,
+      "loss": 1.7766,
+      "step": 19235000
+    },
+    {
+      "epoch": 95.3,
+      "learning_rate": 2.367372165494964e-06,
+      "loss": 1.7869,
+      "step": 19235500
+    },
+    {
+      "epoch": 95.3,
+      "learning_rate": 2.3661360562417326e-06,
+      "loss": 1.7785,
+      "step": 19236000
+    },
+    {
+      "epoch": 95.3,
+      "learning_rate": 2.3648974698156487e-06,
+      "loss": 1.8124,
+      "step": 19236500
+    },
+    {
+      "epoch": 95.31,
+      "learning_rate": 2.3636588833895652e-06,
+      "loss": 1.7888,
+      "step": 19237000
+    },
+    {
+      "epoch": 95.31,
+      "learning_rate": 2.3624202969634813e-06,
+      "loss": 1.7629,
+      "step": 19237500
+    },
+    {
+      "epoch": 95.31,
+      "learning_rate": 2.36118418771025e-06,
+      "loss": 1.7613,
+      "step": 19238000
+    },
+    {
+      "epoch": 95.31,
+      "learning_rate": 2.3599456012841666e-06,
+      "loss": 1.7812,
+      "step": 19238500
+    },
+    {
+      "epoch": 95.32,
+      "learning_rate": 2.3587070148580827e-06,
+      "loss": 1.7722,
+      "step": 19239000
+    },
+    {
+      "epoch": 95.32,
+      "learning_rate": 2.357468428431999e-06,
+      "loss": 1.7904,
+      "step": 19239500
+    },
+    {
+      "epoch": 95.32,
+      "learning_rate": 2.3562298420059157e-06,
+      "loss": 1.7838,
+      "step": 19240000
+    },
+    {
+      "epoch": 95.32,
+      "learning_rate": 2.3549912555798322e-06,
+      "loss": 1.7761,
+      "step": 19240500
+    },
+    {
+      "epoch": 95.33,
+      "learning_rate": 2.3537526691537483e-06,
+      "loss": 1.7916,
+      "step": 19241000
+    },
+    {
+      "epoch": 95.33,
+      "learning_rate": 2.352514082727665e-06,
+      "loss": 1.7503,
+      "step": 19241500
+    },
+    {
+      "epoch": 95.33,
+      "learning_rate": 2.351275496301581e-06,
+      "loss": 1.7839,
+      "step": 19242000
+    },
+    {
+      "epoch": 95.33,
+      "learning_rate": 2.3500369098754974e-06,
+      "loss": 1.7655,
+      "step": 19242500
+    },
+    {
+      "epoch": 95.34,
+      "learning_rate": 2.348798323449414e-06,
+      "loss": 1.7803,
+      "step": 19243000
+    },
+    {
+      "epoch": 95.34,
+      "learning_rate": 2.3475622141961823e-06,
+      "loss": 1.7911,
+      "step": 19243500
+    },
+    {
+      "epoch": 95.34,
+      "learning_rate": 2.346326104942951e-06,
+      "loss": 1.8022,
+      "step": 19244000
+    },
+    {
+      "epoch": 95.34,
+      "learning_rate": 2.345087518516867e-06,
+      "loss": 1.7884,
+      "step": 19244500
+    },
+    {
+      "epoch": 95.35,
+      "learning_rate": 2.3438489320907837e-06,
+      "loss": 1.7738,
+      "step": 19245000
+    },
+    {
+      "epoch": 95.35,
+      "learning_rate": 2.3426103456646998e-06,
+      "loss": 1.7733,
+      "step": 19245500
+    },
+    {
+      "epoch": 95.35,
+      "learning_rate": 2.3413717592386163e-06,
+      "loss": 1.7866,
+      "step": 19246000
+    },
+    {
+      "epoch": 95.35,
+      "learning_rate": 2.3401331728125328e-06,
+      "loss": 1.7518,
+      "step": 19246500
+    },
+    {
+      "epoch": 95.36,
+      "learning_rate": 2.338894586386449e-06,
+      "loss": 1.783,
+      "step": 19247000
+    },
+    {
+      "epoch": 95.36,
+      "learning_rate": 2.3376559999603653e-06,
+      "loss": 1.7806,
+      "step": 19247500
+    },
+    {
+      "epoch": 95.36,
+      "learning_rate": 2.3364198907071337e-06,
+      "loss": 1.7813,
+      "step": 19248000
+    },
+    {
+      "epoch": 95.36,
+      "learning_rate": 2.3351813042810502e-06,
+      "loss": 1.7883,
+      "step": 19248500
+    },
+    {
+      "epoch": 95.37,
+      "learning_rate": 2.3339427178549663e-06,
+      "loss": 1.786,
+      "step": 19249000
+    },
+    {
+      "epoch": 95.37,
+      "learning_rate": 2.332704131428883e-06,
+      "loss": 1.7645,
+      "step": 19249500
+    },
+    {
+      "epoch": 95.37,
+      "learning_rate": 2.3314655450027993e-06,
+      "loss": 1.7768,
+      "step": 19250000
+    },
+    {
+      "epoch": 95.37,
+      "learning_rate": 2.3302269585767154e-06,
+      "loss": 1.8067,
+      "step": 19250500
+    },
+    {
+      "epoch": 95.38,
+      "learning_rate": 2.3289908493234842e-06,
+      "loss": 1.7615,
+      "step": 19251000
+    },
+    {
+      "epoch": 95.38,
+      "learning_rate": 2.3277522628974003e-06,
+      "loss": 1.7923,
+      "step": 19251500
+    },
+    {
+      "epoch": 95.38,
+      "learning_rate": 2.326513676471317e-06,
+      "loss": 1.7584,
+      "step": 19252000
+    },
+    {
+      "epoch": 95.38,
+      "learning_rate": 2.3252750900452333e-06,
+      "loss": 1.7694,
+      "step": 19252500
+    },
+    {
+      "epoch": 95.39,
+      "learning_rate": 2.32403650361915e-06,
+      "loss": 1.7763,
+      "step": 19253000
+    },
+    {
+      "epoch": 95.39,
+      "learning_rate": 2.3227979171930663e-06,
+      "loss": 1.7745,
+      "step": 19253500
+    },
+    {
+      "epoch": 95.39,
+      "learning_rate": 2.3215593307669824e-06,
+      "loss": 1.7969,
+      "step": 19254000
+    },
+    {
+      "epoch": 95.39,
+      "learning_rate": 2.320323221513751e-06,
+      "loss": 1.7941,
+      "step": 19254500
+    },
+    {
+      "epoch": 95.4,
+      "learning_rate": 2.319087112260519e-06,
+      "loss": 1.8008,
+      "step": 19255000
+    },
+    {
+      "epoch": 95.4,
+      "learning_rate": 2.317848525834436e-06,
+      "loss": 1.769,
+      "step": 19255500
+    },
+    {
+      "epoch": 95.4,
+      "learning_rate": 2.316609939408352e-06,
+      "loss": 1.7494,
+      "step": 19256000
+    },
+    {
+      "epoch": 95.4,
+      "learning_rate": 2.3153713529822687e-06,
+      "loss": 1.7927,
+      "step": 19256500
+    },
+    {
+      "epoch": 95.41,
+      "learning_rate": 2.3141327665561848e-06,
+      "loss": 1.7827,
+      "step": 19257000
+    },
+    {
+      "epoch": 95.41,
+      "learning_rate": 2.3128941801301013e-06,
+      "loss": 1.7973,
+      "step": 19257500
+    },
+    {
+      "epoch": 95.41,
+      "learning_rate": 2.31165807087687e-06,
+      "loss": 1.7943,
+      "step": 19258000
+    },
+    {
+      "epoch": 95.41,
+      "learning_rate": 2.310419484450786e-06,
+      "loss": 1.7655,
+      "step": 19258500
+    },
+    {
+      "epoch": 95.42,
+      "learning_rate": 2.3091808980247027e-06,
+      "loss": 1.7814,
+      "step": 19259000
+    },
+    {
+      "epoch": 95.42,
+      "learning_rate": 2.3079423115986187e-06,
+      "loss": 1.7706,
+      "step": 19259500
+    },
+    {
+      "epoch": 95.42,
+      "learning_rate": 2.3067037251725353e-06,
+      "loss": 1.7797,
+      "step": 19260000
+    },
+    {
+      "epoch": 95.42,
+      "learning_rate": 2.3054651387464513e-06,
+      "loss": 1.7858,
+      "step": 19260500
+    },
+    {
+      "epoch": 95.43,
+      "learning_rate": 2.304226552320368e-06,
+      "loss": 1.7738,
+      "step": 19261000
+    },
+    {
+      "epoch": 95.43,
+      "learning_rate": 2.3029879658942843e-06,
+      "loss": 1.7789,
+      "step": 19261500
+    },
+    {
+      "epoch": 95.43,
+      "learning_rate": 2.3017493794682004e-06,
+      "loss": 1.8064,
+      "step": 19262000
+    },
+    {
+      "epoch": 95.43,
+      "learning_rate": 2.3005132702149692e-06,
+      "loss": 1.7772,
+      "step": 19262500
+    },
+    {
+      "epoch": 95.44,
+      "learning_rate": 2.2992746837888853e-06,
+      "loss": 1.7657,
+      "step": 19263000
+    },
+    {
+      "epoch": 95.44,
+      "learning_rate": 2.298036097362802e-06,
+      "loss": 1.7974,
+      "step": 19263500
+    },
+    {
+      "epoch": 95.44,
+      "learning_rate": 2.2967975109367183e-06,
+      "loss": 1.7792,
+      "step": 19264000
+    },
+    {
+      "epoch": 95.44,
+      "learning_rate": 2.2955614016834867e-06,
+      "loss": 1.7666,
+      "step": 19264500
+    },
+    {
+      "epoch": 95.45,
+      "learning_rate": 2.2943228152574032e-06,
+      "loss": 1.7571,
+      "step": 19265000
+    },
+    {
+      "epoch": 95.45,
+      "learning_rate": 2.2930842288313193e-06,
+      "loss": 1.7567,
+      "step": 19265500
+    },
+    {
+      "epoch": 95.45,
+      "learning_rate": 2.291845642405236e-06,
+      "loss": 1.7521,
+      "step": 19266000
+    },
+    {
+      "epoch": 95.45,
+      "learning_rate": 2.2906070559791523e-06,
+      "loss": 1.7603,
+      "step": 19266500
+    },
+    {
+      "epoch": 95.46,
+      "learning_rate": 2.289368469553069e-06,
+      "loss": 1.7682,
+      "step": 19267000
+    },
+    {
+      "epoch": 95.46,
+      "learning_rate": 2.288129883126985e-06,
+      "loss": 1.7923,
+      "step": 19267500
+    },
+    {
+      "epoch": 95.46,
+      "learning_rate": 2.2868912967009014e-06,
+      "loss": 1.7772,
+      "step": 19268000
+    },
+    {
+      "epoch": 95.46,
+      "learning_rate": 2.285652710274818e-06,
+      "loss": 1.77,
+      "step": 19268500
+    },
+    {
+      "epoch": 95.47,
+      "learning_rate": 2.284414123848734e-06,
+      "loss": 1.806,
+      "step": 19269000
+    },
+    {
+      "epoch": 95.47,
+      "learning_rate": 2.2831755374226505e-06,
+      "loss": 1.7583,
+      "step": 19269500
+    },
+    {
+      "epoch": 95.47,
+      "learning_rate": 2.2819369509965666e-06,
+      "loss": 1.7915,
+      "step": 19270000
+    },
+    {
+      "epoch": 95.47,
+      "learning_rate": 2.2807008417433354e-06,
+      "loss": 1.8001,
+      "step": 19270500
+    },
+    {
+      "epoch": 95.48,
+      "learning_rate": 2.2794622553172515e-06,
+      "loss": 1.7579,
+      "step": 19271000
+    },
+    {
+      "epoch": 95.48,
+      "learning_rate": 2.278223668891168e-06,
+      "loss": 1.7743,
+      "step": 19271500
+    },
+    {
+      "epoch": 95.48,
+      "learning_rate": 2.2769850824650845e-06,
+      "loss": 1.7612,
+      "step": 19272000
+    },
+    {
+      "epoch": 95.48,
+      "learning_rate": 2.275748973211853e-06,
+      "loss": 1.7548,
+      "step": 19272500
+    },
+    {
+      "epoch": 95.49,
+      "learning_rate": 2.2745103867857694e-06,
+      "loss": 1.7796,
+      "step": 19273000
+    },
+    {
+      "epoch": 95.49,
+      "learning_rate": 2.2732718003596854e-06,
+      "loss": 1.7831,
+      "step": 19273500
+    },
+    {
+      "epoch": 95.49,
+      "learning_rate": 2.272033213933602e-06,
+      "loss": 1.7765,
+      "step": 19274000
+    },
+    {
+      "epoch": 95.49,
+      "learning_rate": 2.270794627507518e-06,
+      "loss": 1.772,
+      "step": 19274500
+    },
+    {
+      "epoch": 95.5,
+      "learning_rate": 2.2695560410814345e-06,
+      "loss": 1.7667,
+      "step": 19275000
+    },
+    {
+      "epoch": 95.5,
+      "learning_rate": 2.2683199318282033e-06,
+      "loss": 1.7932,
+      "step": 19275500
+    },
+    {
+      "epoch": 95.5,
+      "learning_rate": 2.2670813454021194e-06,
+      "loss": 1.7702,
+      "step": 19276000
+    },
+    {
+      "epoch": 95.5,
+      "learning_rate": 2.265842758976036e-06,
+      "loss": 1.7743,
+      "step": 19276500
+    },
+    {
+      "epoch": 95.5,
+      "learning_rate": 2.264604172549952e-06,
+      "loss": 1.776,
+      "step": 19277000
+    },
+    {
+      "epoch": 95.51,
+      "learning_rate": 2.263365586123869e-06,
+      "loss": 1.7782,
+      "step": 19277500
+    },
+    {
+      "epoch": 95.51,
+      "learning_rate": 2.262126999697785e-06,
+      "loss": 1.8033,
+      "step": 19278000
+    },
+    {
+      "epoch": 95.51,
+      "learning_rate": 2.2608884132717015e-06,
+      "loss": 1.7861,
+      "step": 19278500
+    },
+    {
+      "epoch": 95.51,
+      "learning_rate": 2.25965230401847e-06,
+      "loss": 1.7779,
+      "step": 19279000
+    },
+    {
+      "epoch": 95.52,
+      "learning_rate": 2.2584137175923864e-06,
+      "loss": 1.7673,
+      "step": 19279500
+    },
+    {
+      "epoch": 95.52,
+      "learning_rate": 2.257177608339155e-06,
+      "loss": 1.7834,
+      "step": 19280000
+    },
+    {
+      "epoch": 95.52,
+      "learning_rate": 2.2559390219130713e-06,
+      "loss": 1.7692,
+      "step": 19280500
+    },
+    {
+      "epoch": 95.52,
+      "learning_rate": 2.254700435486988e-06,
+      "loss": 1.7755,
+      "step": 19281000
+    },
+    {
+      "epoch": 95.53,
+      "learning_rate": 2.253461849060904e-06,
+      "loss": 1.7784,
+      "step": 19281500
+    },
+    {
+      "epoch": 95.53,
+      "learning_rate": 2.2522232626348204e-06,
+      "loss": 1.7838,
+      "step": 19282000
+    },
+    {
+      "epoch": 95.53,
+      "learning_rate": 2.2509846762087365e-06,
+      "loss": 1.7835,
+      "step": 19282500
+    },
+    {
+      "epoch": 95.53,
+      "learning_rate": 2.249746089782653e-06,
+      "loss": 1.78,
+      "step": 19283000
+    },
+    {
+      "epoch": 95.54,
+      "learning_rate": 2.2485075033565695e-06,
+      "loss": 1.7876,
+      "step": 19283500
+    },
+    {
+      "epoch": 95.54,
+      "learning_rate": 2.247271394103338e-06,
+      "loss": 1.781,
+      "step": 19284000
+    },
+    {
+      "epoch": 95.54,
+      "learning_rate": 2.2460352848501067e-06,
+      "loss": 1.7781,
+      "step": 19284500
+    },
+    {
+      "epoch": 95.54,
+      "learning_rate": 2.2447966984240228e-06,
+      "loss": 1.7746,
+      "step": 19285000
+    },
+    {
+      "epoch": 95.55,
+      "learning_rate": 2.2435581119979393e-06,
+      "loss": 1.7811,
+      "step": 19285500
+    },
+    {
+      "epoch": 95.55,
+      "learning_rate": 2.2423195255718553e-06,
+      "loss": 1.7791,
+      "step": 19286000
+    },
+    {
+      "epoch": 95.55,
+      "learning_rate": 2.241083416318624e-06,
+      "loss": 1.7726,
+      "step": 19286500
+    },
+    {
+      "epoch": 95.55,
+      "learning_rate": 2.2398448298925402e-06,
+      "loss": 1.7936,
+      "step": 19287000
+    },
+    {
+      "epoch": 95.56,
+      "learning_rate": 2.2386062434664567e-06,
+      "loss": 1.8031,
+      "step": 19287500
+    },
+    {
+      "epoch": 95.56,
+      "learning_rate": 2.2373676570403732e-06,
+      "loss": 1.7963,
+      "step": 19288000
+    },
+    {
+      "epoch": 95.56,
+      "learning_rate": 2.2361290706142893e-06,
+      "loss": 1.7779,
+      "step": 19288500
+    },
+    {
+      "epoch": 95.56,
+      "learning_rate": 2.234890484188206e-06,
+      "loss": 1.7586,
+      "step": 19289000
+    },
+    {
+      "epoch": 95.57,
+      "learning_rate": 2.233651897762122e-06,
+      "loss": 1.8082,
+      "step": 19289500
+    },
+    {
+      "epoch": 95.57,
+      "learning_rate": 2.2324133113360384e-06,
+      "loss": 1.7824,
+      "step": 19290000
+    },
+    {
+      "epoch": 95.57,
+      "learning_rate": 2.231174724909955e-06,
+      "loss": 1.7797,
+      "step": 19290500
+    },
+    {
+      "epoch": 95.57,
+      "learning_rate": 2.2299386156567233e-06,
+      "loss": 1.7757,
+      "step": 19291000
+    },
+    {
+      "epoch": 95.58,
+      "learning_rate": 2.22870002923064e-06,
+      "loss": 1.7681,
+      "step": 19291500
+    },
+    {
+      "epoch": 95.58,
+      "learning_rate": 2.227463919977408e-06,
+      "loss": 1.7798,
+      "step": 19292000
+    },
+    {
+      "epoch": 95.58,
+      "learning_rate": 2.2262253335513247e-06,
+      "loss": 1.777,
+      "step": 19292500
+    },
+    {
+      "epoch": 95.58,
+      "learning_rate": 2.2249867471252408e-06,
+      "loss": 1.7618,
+      "step": 19293000
+    },
+    {
+      "epoch": 95.59,
+      "learning_rate": 2.2237481606991573e-06,
+      "loss": 1.7747,
+      "step": 19293500
+    },
+    {
+      "epoch": 95.59,
+      "learning_rate": 2.222509574273074e-06,
+      "loss": 1.795,
+      "step": 19294000
+    },
+    {
+      "epoch": 95.59,
+      "learning_rate": 2.221273465019842e-06,
+      "loss": 1.7661,
+      "step": 19294500
+    },
+    {
+      "epoch": 95.59,
+      "learning_rate": 2.2200348785937587e-06,
+      "loss": 1.7753,
+      "step": 19295000
+    },
+    {
+      "epoch": 95.6,
+      "learning_rate": 2.2187962921676748e-06,
+      "loss": 1.7859,
+      "step": 19295500
+    },
+    {
+      "epoch": 95.6,
+      "learning_rate": 2.2175577057415913e-06,
+      "loss": 1.7889,
+      "step": 19296000
+    },
+    {
+      "epoch": 95.6,
+      "learning_rate": 2.2163191193155078e-06,
+      "loss": 1.777,
+      "step": 19296500
+    },
+    {
+      "epoch": 95.6,
+      "learning_rate": 2.2150805328894243e-06,
+      "loss": 1.7994,
+      "step": 19297000
+    },
+    {
+      "epoch": 95.61,
+      "learning_rate": 2.2138419464633404e-06,
+      "loss": 1.784,
+      "step": 19297500
+    },
+    {
+      "epoch": 95.61,
+      "learning_rate": 2.212603360037257e-06,
+      "loss": 1.7798,
+      "step": 19298000
+    },
+    {
+      "epoch": 95.61,
+      "learning_rate": 2.2113647736111734e-06,
+      "loss": 1.7655,
+      "step": 19298500
+    },
+    {
+      "epoch": 95.61,
+      "learning_rate": 2.2101261871850894e-06,
+      "loss": 1.7662,
+      "step": 19299000
+    },
+    {
+      "epoch": 95.62,
+      "learning_rate": 2.2088900779318583e-06,
+      "loss": 1.788,
+      "step": 19299500
+    },
+    {
+      "epoch": 95.62,
+      "learning_rate": 2.2076514915057743e-06,
+      "loss": 1.7931,
+      "step": 19300000
+    },
+    {
+      "epoch": 95.62,
+      "learning_rate": 2.206412905079691e-06,
+      "loss": 1.7769,
+      "step": 19300500
+    },
+    {
+      "epoch": 95.62,
+      "learning_rate": 2.205174318653607e-06,
+      "loss": 1.7652,
+      "step": 19301000
+    },
+    {
+      "epoch": 95.63,
+      "learning_rate": 2.2039357322275234e-06,
+      "loss": 1.7962,
+      "step": 19301500
+    },
+    {
+      "epoch": 95.63,
+      "learning_rate": 2.20269714580144e-06,
+      "loss": 1.772,
+      "step": 19302000
+    },
+    {
+      "epoch": 95.63,
+      "learning_rate": 2.201458559375356e-06,
+      "loss": 1.7937,
+      "step": 19302500
+    },
+    {
+      "epoch": 95.63,
+      "learning_rate": 2.2002199729492725e-06,
+      "loss": 1.7844,
+      "step": 19303000
+    },
+    {
+      "epoch": 95.64,
+      "learning_rate": 2.1989813865231886e-06,
+      "loss": 1.7538,
+      "step": 19303500
+    },
+    {
+      "epoch": 95.64,
+      "learning_rate": 2.1977452772699574e-06,
+      "loss": 1.7793,
+      "step": 19304000
+    },
+    {
+      "epoch": 95.64,
+      "learning_rate": 2.1965066908438735e-06,
+      "loss": 1.7916,
+      "step": 19304500
+    },
+    {
+      "epoch": 95.64,
+      "learning_rate": 2.19526810441779e-06,
+      "loss": 1.7497,
+      "step": 19305000
+    },
+    {
+      "epoch": 95.65,
+      "learning_rate": 2.194031995164559e-06,
+      "loss": 1.7705,
+      "step": 19305500
+    },
+    {
+      "epoch": 95.65,
+      "learning_rate": 2.192793408738475e-06,
+      "loss": 1.7758,
+      "step": 19306000
+    },
+    {
+      "epoch": 95.65,
+      "learning_rate": 2.1915548223123914e-06,
+      "loss": 1.7949,
+      "step": 19306500
+    },
+    {
+      "epoch": 95.65,
+      "learning_rate": 2.1903187130591598e-06,
+      "loss": 1.7774,
+      "step": 19307000
+    },
+    {
+      "epoch": 95.66,
+      "learning_rate": 2.1890801266330763e-06,
+      "loss": 1.795,
+      "step": 19307500
+    },
+    {
+      "epoch": 95.66,
+      "learning_rate": 2.1878415402069924e-06,
+      "loss": 1.7584,
+      "step": 19308000
+    },
+    {
+      "epoch": 95.66,
+      "learning_rate": 2.186602953780909e-06,
+      "loss": 1.7712,
+      "step": 19308500
+    },
+    {
+      "epoch": 95.66,
+      "learning_rate": 2.1853643673548254e-06,
+      "loss": 1.7751,
+      "step": 19309000
+    },
+    {
+      "epoch": 95.67,
+      "learning_rate": 2.184125780928742e-06,
+      "loss": 1.7937,
+      "step": 19309500
+    },
+    {
+      "epoch": 95.67,
+      "learning_rate": 2.1828871945026584e-06,
+      "loss": 1.7669,
+      "step": 19310000
+    },
+    {
+      "epoch": 95.67,
+      "learning_rate": 2.1816486080765745e-06,
+      "loss": 1.7961,
+      "step": 19310500
+    },
+    {
+      "epoch": 95.67,
+      "learning_rate": 2.180410021650491e-06,
+      "loss": 1.79,
+      "step": 19311000
+    },
+    {
+      "epoch": 95.68,
+      "learning_rate": 2.1791739123972594e-06,
+      "loss": 1.7637,
+      "step": 19311500
+    },
+    {
+      "epoch": 95.68,
+      "learning_rate": 2.177935325971176e-06,
+      "loss": 1.7751,
+      "step": 19312000
+    },
+    {
+      "epoch": 95.68,
+      "learning_rate": 2.1766992167179442e-06,
+      "loss": 1.7949,
+      "step": 19312500
+    },
+    {
+      "epoch": 95.68,
+      "learning_rate": 2.1754606302918607e-06,
+      "loss": 1.7857,
+      "step": 19313000
+    },
+    {
+      "epoch": 95.69,
+      "learning_rate": 2.1742220438657773e-06,
+      "loss": 1.7743,
+      "step": 19313500
+    },
+    {
+      "epoch": 95.69,
+      "learning_rate": 2.1729834574396933e-06,
+      "loss": 1.7623,
+      "step": 19314000
+    },
+    {
+      "epoch": 95.69,
+      "learning_rate": 2.17174487101361e-06,
+      "loss": 1.7695,
+      "step": 19314500
+    },
+    {
+      "epoch": 95.69,
+      "learning_rate": 2.170506284587526e-06,
+      "loss": 1.7634,
+      "step": 19315000
+    },
+    {
+      "epoch": 95.7,
+      "learning_rate": 2.1692676981614424e-06,
+      "loss": 1.7728,
+      "step": 19315500
+    },
+    {
+      "epoch": 95.7,
+      "learning_rate": 2.1680291117353585e-06,
+      "loss": 1.7798,
+      "step": 19316000
+    },
+    {
+      "epoch": 95.7,
+      "learning_rate": 2.166790525309275e-06,
+      "loss": 1.786,
+      "step": 19316500
+    },
+    {
+      "epoch": 95.7,
+      "learning_rate": 2.1655519388831915e-06,
+      "loss": 1.7546,
+      "step": 19317000
+    },
+    {
+      "epoch": 95.71,
+      "learning_rate": 2.1643133524571076e-06,
+      "loss": 1.7667,
+      "step": 19317500
+    },
+    {
+      "epoch": 95.71,
+      "learning_rate": 2.1630747660310245e-06,
+      "loss": 1.7648,
+      "step": 19318000
+    },
+    {
+      "epoch": 95.71,
+      "learning_rate": 2.1618386567777925e-06,
+      "loss": 1.7661,
+      "step": 19318500
+    },
+    {
+      "epoch": 95.71,
+      "learning_rate": 2.1606025475245613e-06,
+      "loss": 1.7738,
+      "step": 19319000
+    },
+    {
+      "epoch": 95.72,
+      "learning_rate": 2.1593639610984774e-06,
+      "loss": 1.7729,
+      "step": 19319500
+    },
+    {
+      "epoch": 95.72,
+      "learning_rate": 2.158125374672394e-06,
+      "loss": 1.7653,
+      "step": 19320000
+    },
+    {
+      "epoch": 95.72,
+      "learning_rate": 2.1568867882463104e-06,
+      "loss": 1.7913,
+      "step": 19320500
+    },
+    {
+      "epoch": 95.72,
+      "learning_rate": 2.1556482018202265e-06,
+      "loss": 1.7664,
+      "step": 19321000
+    },
+    {
+      "epoch": 95.73,
+      "learning_rate": 2.1544096153941434e-06,
+      "loss": 1.7951,
+      "step": 19321500
+    },
+    {
+      "epoch": 95.73,
+      "learning_rate": 2.1531710289680595e-06,
+      "loss": 1.7883,
+      "step": 19322000
+    },
+    {
+      "epoch": 95.73,
+      "learning_rate": 2.151932442541976e-06,
+      "loss": 1.7627,
+      "step": 19322500
+    },
+    {
+      "epoch": 95.73,
+      "learning_rate": 2.1506963332887444e-06,
+      "loss": 1.7895,
+      "step": 19323000
+    },
+    {
+      "epoch": 95.74,
+      "learning_rate": 2.149457746862661e-06,
+      "loss": 1.7951,
+      "step": 19323500
+    },
+    {
+      "epoch": 95.74,
+      "learning_rate": 2.148219160436577e-06,
+      "loss": 1.7846,
+      "step": 19324000
+    },
+    {
+      "epoch": 95.74,
+      "learning_rate": 2.1469805740104935e-06,
+      "loss": 1.7595,
+      "step": 19324500
+    },
+    {
+      "epoch": 95.74,
+      "learning_rate": 2.1457444647572623e-06,
+      "loss": 1.7894,
+      "step": 19325000
+    },
+    {
+      "epoch": 95.75,
+      "learning_rate": 2.1445058783311783e-06,
+      "loss": 1.7823,
+      "step": 19325500
+    },
+    {
+      "epoch": 95.75,
+      "learning_rate": 2.143272246250799e-06,
+      "loss": 1.7673,
+      "step": 19326000
+    },
+    {
+      "epoch": 95.75,
+      "learning_rate": 2.142033659824715e-06,
+      "loss": 1.7956,
+      "step": 19326500
+    },
+    {
+      "epoch": 95.75,
+      "learning_rate": 2.1407950733986316e-06,
+      "loss": 1.7881,
+      "step": 19327000
+    },
+    {
+      "epoch": 95.76,
+      "learning_rate": 2.1395564869725477e-06,
+      "loss": 1.786,
+      "step": 19327500
+    },
+    {
+      "epoch": 95.76,
+      "learning_rate": 2.1383179005464646e-06,
+      "loss": 1.7762,
+      "step": 19328000
+    },
+    {
+      "epoch": 95.76,
+      "learning_rate": 2.1370793141203807e-06,
+      "loss": 1.7826,
+      "step": 19328500
+    },
+    {
+      "epoch": 95.76,
+      "learning_rate": 2.1358407276942972e-06,
+      "loss": 1.7769,
+      "step": 19329000
+    },
+    {
+      "epoch": 95.77,
+      "learning_rate": 2.1346021412682137e-06,
+      "loss": 1.7692,
+      "step": 19329500
+    },
+    {
+      "epoch": 95.77,
+      "learning_rate": 2.13336355484213e-06,
+      "loss": 1.757,
+      "step": 19330000
+    },
+    {
+      "epoch": 95.77,
+      "learning_rate": 2.1321249684160463e-06,
+      "loss": 1.7852,
+      "step": 19330500
+    },
+    {
+      "epoch": 95.77,
+      "learning_rate": 2.1308863819899624e-06,
+      "loss": 1.7794,
+      "step": 19331000
+    },
+    {
+      "epoch": 95.77,
+      "learning_rate": 2.129647795563879e-06,
+      "loss": 1.7686,
+      "step": 19331500
+    },
+    {
+      "epoch": 95.78,
+      "learning_rate": 2.1284092091377954e-06,
+      "loss": 1.7793,
+      "step": 19332000
+    },
+    {
+      "epoch": 95.78,
+      "learning_rate": 2.1271706227117115e-06,
+      "loss": 1.7668,
+      "step": 19332500
+    },
+    {
+      "epoch": 95.78,
+      "learning_rate": 2.1259345134584803e-06,
+      "loss": 1.7765,
+      "step": 19333000
+    },
+    {
+      "epoch": 95.78,
+      "learning_rate": 2.1246959270323964e-06,
+      "loss": 1.7793,
+      "step": 19333500
+    },
+    {
+      "epoch": 95.79,
+      "learning_rate": 2.123457340606313e-06,
+      "loss": 1.7761,
+      "step": 19334000
+    },
+    {
+      "epoch": 95.79,
+      "learning_rate": 2.122218754180229e-06,
+      "loss": 1.7803,
+      "step": 19334500
+    },
+    {
+      "epoch": 95.79,
+      "learning_rate": 2.1209801677541455e-06,
+      "loss": 1.7993,
+      "step": 19335000
+    },
+    {
+      "epoch": 95.79,
+      "learning_rate": 2.119746535673766e-06,
+      "loss": 1.7927,
+      "step": 19335500
+    },
+    {
+      "epoch": 95.8,
+      "learning_rate": 2.118510426420535e-06,
+      "loss": 1.7933,
+      "step": 19336000
+    },
+    {
+      "epoch": 95.8,
+      "learning_rate": 2.1172718399944515e-06,
+      "loss": 1.7701,
+      "step": 19336500
+    },
+    {
+      "epoch": 95.8,
+      "learning_rate": 2.1160332535683675e-06,
+      "loss": 1.7852,
+      "step": 19337000
+    },
+    {
+      "epoch": 95.8,
+      "learning_rate": 2.114794667142284e-06,
+      "loss": 1.7595,
+      "step": 19337500
+    },
+    {
+      "epoch": 95.81,
+      "learning_rate": 2.1135560807162e-06,
+      "loss": 1.7846,
+      "step": 19338000
+    },
+    {
+      "epoch": 95.81,
+      "learning_rate": 2.1123174942901166e-06,
+      "loss": 1.7756,
+      "step": 19338500
+    },
+    {
+      "epoch": 95.81,
+      "learning_rate": 2.1110789078640327e-06,
+      "loss": 1.7763,
+      "step": 19339000
+    },
+    {
+      "epoch": 95.81,
+      "learning_rate": 2.1098403214379492e-06,
+      "loss": 1.7705,
+      "step": 19339500
+    },
+    {
+      "epoch": 95.82,
+      "learning_rate": 2.1086017350118657e-06,
+      "loss": 1.7811,
+      "step": 19340000
+    },
+    {
+      "epoch": 95.82,
+      "learning_rate": 2.1073631485857822e-06,
+      "loss": 1.7544,
+      "step": 19340500
+    },
+    {
+      "epoch": 95.82,
+      "learning_rate": 2.1061245621596987e-06,
+      "loss": 1.7634,
+      "step": 19341000
+    },
+    {
+      "epoch": 95.82,
+      "learning_rate": 2.104885975733615e-06,
+      "loss": 1.7715,
+      "step": 19341500
+    },
+    {
+      "epoch": 95.83,
+      "learning_rate": 2.1036473893075313e-06,
+      "loss": 1.7645,
+      "step": 19342000
+    },
+    {
+      "epoch": 95.83,
+      "learning_rate": 2.1024088028814474e-06,
+      "loss": 1.7916,
+      "step": 19342500
+    },
+    {
+      "epoch": 95.83,
+      "learning_rate": 2.101170216455364e-06,
+      "loss": 1.7634,
+      "step": 19343000
+    },
+    {
+      "epoch": 95.83,
+      "learning_rate": 2.0999316300292804e-06,
+      "loss": 1.7928,
+      "step": 19343500
+    },
+    {
+      "epoch": 95.84,
+      "learning_rate": 2.098695520776049e-06,
+      "loss": 1.7647,
+      "step": 19344000
+    },
+    {
+      "epoch": 95.84,
+      "learning_rate": 2.0974569343499653e-06,
+      "loss": 1.7659,
+      "step": 19344500
+    },
+    {
+      "epoch": 95.84,
+      "learning_rate": 2.0962183479238814e-06,
+      "loss": 1.747,
+      "step": 19345000
+    },
+    {
+      "epoch": 95.84,
+      "learning_rate": 2.094979761497798e-06,
+      "loss": 1.7817,
+      "step": 19345500
+    },
+    {
+      "epoch": 95.85,
+      "learning_rate": 2.0937436522445663e-06,
+      "loss": 1.7599,
+      "step": 19346000
+    },
+    {
+      "epoch": 95.85,
+      "learning_rate": 2.0925050658184828e-06,
+      "loss": 1.7689,
+      "step": 19346500
+    },
+    {
+      "epoch": 95.85,
+      "learning_rate": 2.0912664793923993e-06,
+      "loss": 1.7888,
+      "step": 19347000
+    },
+    {
+      "epoch": 95.85,
+      "learning_rate": 2.0900278929663154e-06,
+      "loss": 1.7823,
+      "step": 19347500
+    },
+    {
+      "epoch": 95.86,
+      "learning_rate": 2.088789306540232e-06,
+      "loss": 1.7693,
+      "step": 19348000
+    },
+    {
+      "epoch": 95.86,
+      "learning_rate": 2.0875531972870003e-06,
+      "loss": 1.7625,
+      "step": 19348500
+    },
+    {
+      "epoch": 95.86,
+      "learning_rate": 2.0863146108609168e-06,
+      "loss": 1.7887,
+      "step": 19349000
+    },
+    {
+      "epoch": 95.86,
+      "learning_rate": 2.085076024434833e-06,
+      "loss": 1.7924,
+      "step": 19349500
+    },
+    {
+      "epoch": 95.87,
+      "learning_rate": 2.0838374380087493e-06,
+      "loss": 1.7861,
+      "step": 19350000
+    },
+    {
+      "epoch": 95.87,
+      "learning_rate": 2.082598851582666e-06,
+      "loss": 1.7927,
+      "step": 19350500
+    },
+    {
+      "epoch": 95.87,
+      "learning_rate": 2.0813627423294342e-06,
+      "loss": 1.7761,
+      "step": 19351000
+    },
+    {
+      "epoch": 95.87,
+      "learning_rate": 2.080126633076203e-06,
+      "loss": 1.7589,
+      "step": 19351500
+    },
+    {
+      "epoch": 95.88,
+      "learning_rate": 2.078888046650119e-06,
+      "loss": 1.7784,
+      "step": 19352000
+    },
+    {
+      "epoch": 95.88,
+      "learning_rate": 2.0776494602240356e-06,
+      "loss": 1.7953,
+      "step": 19352500
+    },
+    {
+      "epoch": 95.88,
+      "learning_rate": 2.076413350970804e-06,
+      "loss": 1.7706,
+      "step": 19353000
+    },
+    {
+      "epoch": 95.88,
+      "learning_rate": 2.0751747645447205e-06,
+      "loss": 1.7818,
+      "step": 19353500
+    },
+    {
+      "epoch": 95.89,
+      "learning_rate": 2.0739361781186366e-06,
+      "loss": 1.7869,
+      "step": 19354000
+    },
+    {
+      "epoch": 95.89,
+      "learning_rate": 2.072697591692553e-06,
+      "loss": 1.7663,
+      "step": 19354500
+    },
+    {
+      "epoch": 95.89,
+      "learning_rate": 2.071461482439322e-06,
+      "loss": 1.7977,
+      "step": 19355000
+    },
+    {
+      "epoch": 95.89,
+      "learning_rate": 2.070222896013238e-06,
+      "loss": 1.791,
+      "step": 19355500
+    },
+    {
+      "epoch": 95.9,
+      "learning_rate": 2.0689843095871545e-06,
+      "loss": 1.7679,
+      "step": 19356000
+    },
+    {
+      "epoch": 95.9,
+      "learning_rate": 2.0677457231610706e-06,
+      "loss": 1.788,
+      "step": 19356500
+    },
+    {
+      "epoch": 95.9,
+      "learning_rate": 2.066507136734987e-06,
+      "loss": 1.7774,
+      "step": 19357000
+    },
+    {
+      "epoch": 95.9,
+      "learning_rate": 2.0652685503089036e-06,
+      "loss": 1.7657,
+      "step": 19357500
+    },
+    {
+      "epoch": 95.91,
+      "learning_rate": 2.06402996388282e-06,
+      "loss": 1.7578,
+      "step": 19358000
+    },
+    {
+      "epoch": 95.91,
+      "learning_rate": 2.062791377456736e-06,
+      "loss": 1.7751,
+      "step": 19358500
+    },
+    {
+      "epoch": 95.91,
+      "learning_rate": 2.061555268203505e-06,
+      "loss": 1.7794,
+      "step": 19359000
+    },
+    {
+      "epoch": 95.91,
+      "learning_rate": 2.0603166817774215e-06,
+      "loss": 1.7497,
+      "step": 19359500
+    },
+    {
+      "epoch": 95.92,
+      "learning_rate": 2.0590780953513376e-06,
+      "loss": 1.7672,
+      "step": 19360000
+    },
+    {
+      "epoch": 95.92,
+      "learning_rate": 2.057839508925254e-06,
+      "loss": 1.7984,
+      "step": 19360500
+    },
+    {
+      "epoch": 95.92,
+      "learning_rate": 2.05660092249917e-06,
+      "loss": 1.7865,
+      "step": 19361000
+    },
+    {
+      "epoch": 95.92,
+      "learning_rate": 2.055364813245939e-06,
+      "loss": 1.7728,
+      "step": 19361500
+    },
+    {
+      "epoch": 95.93,
+      "learning_rate": 2.054126226819855e-06,
+      "loss": 1.7761,
+      "step": 19362000
+    },
+    {
+      "epoch": 95.93,
+      "learning_rate": 2.0528876403937716e-06,
+      "loss": 1.7643,
+      "step": 19362500
+    },
+    {
+      "epoch": 95.93,
+      "learning_rate": 2.051649053967688e-06,
+      "loss": 1.7969,
+      "step": 19363000
+    },
+    {
+      "epoch": 95.93,
+      "learning_rate": 2.050410467541604e-06,
+      "loss": 1.7856,
+      "step": 19363500
+    },
+    {
+      "epoch": 95.94,
+      "learning_rate": 2.0491718811155206e-06,
+      "loss": 1.7894,
+      "step": 19364000
+    },
+    {
+      "epoch": 95.94,
+      "learning_rate": 2.0479332946894367e-06,
+      "loss": 1.7699,
+      "step": 19364500
+    },
+    {
+      "epoch": 95.94,
+      "learning_rate": 2.0466947082633532e-06,
+      "loss": 1.8014,
+      "step": 19365000
+    },
+    {
+      "epoch": 95.94,
+      "learning_rate": 2.0454561218372697e-06,
+      "loss": 1.7585,
+      "step": 19365500
+    },
+    {
+      "epoch": 95.95,
+      "learning_rate": 2.044217535411186e-06,
+      "loss": 1.7759,
+      "step": 19366000
+    },
+    {
+      "epoch": 95.95,
+      "learning_rate": 2.0429789489851023e-06,
+      "loss": 1.798,
+      "step": 19366500
+    },
+    {
+      "epoch": 95.95,
+      "learning_rate": 2.041740362559019e-06,
+      "loss": 1.7799,
+      "step": 19367000
+    },
+    {
+      "epoch": 95.95,
+      "learning_rate": 2.0405017761329353e-06,
+      "loss": 1.7999,
+      "step": 19367500
+    },
+    {
+      "epoch": 95.96,
+      "learning_rate": 2.0392631897068514e-06,
+      "loss": 1.7584,
+      "step": 19368000
+    },
+    {
+      "epoch": 95.96,
+      "learning_rate": 2.038024603280768e-06,
+      "loss": 1.7941,
+      "step": 19368500
+    },
+    {
+      "epoch": 95.96,
+      "learning_rate": 2.0367860168546844e-06,
+      "loss": 1.766,
+      "step": 19369000
+    },
+    {
+      "epoch": 95.96,
+      "learning_rate": 2.035549907601453e-06,
+      "loss": 1.7744,
+      "step": 19369500
+    },
+    {
+      "epoch": 95.97,
+      "learning_rate": 2.0343113211753693e-06,
+      "loss": 1.79,
+      "step": 19370000
+    },
+    {
+      "epoch": 95.97,
+      "learning_rate": 2.0330727347492854e-06,
+      "loss": 1.7835,
+      "step": 19370500
+    },
+    {
+      "epoch": 95.97,
+      "learning_rate": 2.031836625496054e-06,
+      "loss": 1.7816,
+      "step": 19371000
+    },
+    {
+      "epoch": 95.97,
+      "learning_rate": 2.0305980390699703e-06,
+      "loss": 1.764,
+      "step": 19371500
+    },
+    {
+      "epoch": 95.98,
+      "learning_rate": 2.029359452643887e-06,
+      "loss": 1.7758,
+      "step": 19372000
+    },
+    {
+      "epoch": 95.98,
+      "learning_rate": 2.028120866217803e-06,
+      "loss": 1.7707,
+      "step": 19372500
+    },
+    {
+      "epoch": 95.98,
+      "learning_rate": 2.0268822797917194e-06,
+      "loss": 1.7951,
+      "step": 19373000
+    },
+    {
+      "epoch": 95.98,
+      "learning_rate": 2.02564864771134e-06,
+      "loss": 1.7834,
+      "step": 19373500
+    },
+    {
+      "epoch": 95.99,
+      "learning_rate": 2.0244100612852566e-06,
+      "loss": 1.7795,
+      "step": 19374000
+    },
+    {
+      "epoch": 95.99,
+      "learning_rate": 2.023171474859173e-06,
+      "loss": 1.7742,
+      "step": 19374500
+    },
+    {
+      "epoch": 95.99,
+      "learning_rate": 2.021932888433089e-06,
+      "loss": 1.7895,
+      "step": 19375000
+    },
+    {
+      "epoch": 95.99,
+      "learning_rate": 2.0206943020070057e-06,
+      "loss": 1.764,
+      "step": 19375500
+    },
+    {
+      "epoch": 96.0,
+      "learning_rate": 2.0194557155809217e-06,
+      "loss": 1.8001,
+      "step": 19376000
+    },
+    {
+      "epoch": 96.0,
+      "learning_rate": 2.0182171291548382e-06,
+      "loss": 1.791,
+      "step": 19376500
+    },
+    {
+      "epoch": 96.0,
+      "eval_accuracy": 0.6873047121683882,
+      "eval_accuracy_mlm": 0.6488415920257267,
+      "eval_accuracy_nsp": 0.8686180915362862,
+      "eval_loss": 2.316138982772827,
+      "eval_runtime": 147.2368,
+      "eval_samples_per_second": 1731.625,
+      "eval_steps_per_second": 72.156,
+      "step": 19376928
+    },
+    {
+      "epoch": 96.0,
+      "learning_rate": 2.0169785427287548e-06,
+      "loss": 1.7872,
+      "step": 19377000
+    },
+    {
+      "epoch": 96.0,
+      "learning_rate": 2.015739956302671e-06,
+      "loss": 1.7637,
+      "step": 19377500
+    },
+    {
+      "epoch": 96.01,
+      "learning_rate": 2.0145038470494396e-06,
+      "loss": 1.7844,
+      "step": 19378000
+    },
+    {
+      "epoch": 96.01,
+      "learning_rate": 2.0132652606233557e-06,
+      "loss": 1.7857,
+      "step": 19378500
+    },
+    {
+      "epoch": 96.01,
+      "learning_rate": 2.0120266741972722e-06,
+      "loss": 1.7675,
+      "step": 19379000
+    },
+    {
+      "epoch": 96.01,
+      "learning_rate": 2.0107880877711883e-06,
+      "loss": 1.7646,
+      "step": 19379500
+    },
+    {
+      "epoch": 96.02,
+      "learning_rate": 2.009549501345105e-06,
+      "loss": 1.7882,
+      "step": 19380000
+    },
+    {
+      "epoch": 96.02,
+      "learning_rate": 2.0083133920918736e-06,
+      "loss": 1.7747,
+      "step": 19380500
+    },
+    {
+      "epoch": 96.02,
+      "learning_rate": 2.0070748056657897e-06,
+      "loss": 1.7838,
+      "step": 19381000
+    },
+    {
+      "epoch": 96.02,
+      "learning_rate": 2.005836219239706e-06,
+      "loss": 1.7776,
+      "step": 19381500
+    },
+    {
+      "epoch": 96.03,
+      "learning_rate": 2.0046001099864746e-06,
+      "loss": 1.7752,
+      "step": 19382000
+    },
+    {
+      "epoch": 96.03,
+      "learning_rate": 2.003361523560391e-06,
+      "loss": 1.7769,
+      "step": 19382500
+    },
+    {
+      "epoch": 96.03,
+      "learning_rate": 2.002122937134307e-06,
+      "loss": 1.7679,
+      "step": 19383000
+    },
+    {
+      "epoch": 96.03,
+      "learning_rate": 2.0008843507082237e-06,
+      "loss": 1.7943,
+      "step": 19383500
+    },
+    {
+      "epoch": 96.04,
+      "learning_rate": 1.99964576428214e-06,
+      "loss": 1.7702,
+      "step": 19384000
+    },
+    {
+      "epoch": 96.04,
+      "learning_rate": 1.9984071778560567e-06,
+      "loss": 1.7644,
+      "step": 19384500
+    },
+    {
+      "epoch": 96.04,
+      "learning_rate": 1.997168591429973e-06,
+      "loss": 1.7493,
+      "step": 19385000
+    },
+    {
+      "epoch": 96.04,
+      "learning_rate": 1.9959300050038893e-06,
+      "loss": 1.7654,
+      "step": 19385500
+    },
+    {
+      "epoch": 96.04,
+      "learning_rate": 1.9946914185778058e-06,
+      "loss": 1.7654,
+      "step": 19386000
+    },
+    {
+      "epoch": 96.05,
+      "learning_rate": 1.993452832151722e-06,
+      "loss": 1.7713,
+      "step": 19386500
+    },
+    {
+      "epoch": 96.05,
+      "learning_rate": 1.9922142457256384e-06,
+      "loss": 1.7692,
+      "step": 19387000
+    },
+    {
+      "epoch": 96.05,
+      "learning_rate": 1.990975659299555e-06,
+      "loss": 1.7794,
+      "step": 19387500
+    },
+    {
+      "epoch": 96.05,
+      "learning_rate": 1.9897395500463233e-06,
+      "loss": 1.7653,
+      "step": 19388000
+    },
+    {
+      "epoch": 96.06,
+      "learning_rate": 1.9885009636202398e-06,
+      "loss": 1.786,
+      "step": 19388500
+    },
+    {
+      "epoch": 96.06,
+      "learning_rate": 1.987262377194156e-06,
+      "loss": 1.751,
+      "step": 19389000
+    },
+    {
+      "epoch": 96.06,
+      "learning_rate": 1.9860237907680724e-06,
+      "loss": 1.7677,
+      "step": 19389500
+    },
+    {
+      "epoch": 96.06,
+      "learning_rate": 1.9847852043419884e-06,
+      "loss": 1.775,
+      "step": 19390000
+    },
+    {
+      "epoch": 96.07,
+      "learning_rate": 1.9835515722616095e-06,
+      "loss": 1.7797,
+      "step": 19390500
+    },
+    {
+      "epoch": 96.07,
+      "learning_rate": 1.982315463008378e-06,
+      "loss": 1.7773,
+      "step": 19391000
+    },
+    {
+      "epoch": 96.07,
+      "learning_rate": 1.9810768765822944e-06,
+      "loss": 1.7536,
+      "step": 19391500
+    },
+    {
+      "epoch": 96.07,
+      "learning_rate": 1.9798382901562105e-06,
+      "loss": 1.7718,
+      "step": 19392000
+    },
+    {
+      "epoch": 96.08,
+      "learning_rate": 1.978599703730127e-06,
+      "loss": 1.7754,
+      "step": 19392500
+    },
+    {
+      "epoch": 96.08,
+      "learning_rate": 1.9773611173040435e-06,
+      "loss": 1.7857,
+      "step": 19393000
+    },
+    {
+      "epoch": 96.08,
+      "learning_rate": 1.9761225308779596e-06,
+      "loss": 1.7729,
+      "step": 19393500
+    },
+    {
+      "epoch": 96.08,
+      "learning_rate": 1.974883944451876e-06,
+      "loss": 1.7891,
+      "step": 19394000
+    },
+    {
+      "epoch": 96.09,
+      "learning_rate": 1.9736478351986445e-06,
+      "loss": 1.7645,
+      "step": 19394500
+    },
+    {
+      "epoch": 96.09,
+      "learning_rate": 1.972409248772561e-06,
+      "loss": 1.7889,
+      "step": 19395000
+    },
+    {
+      "epoch": 96.09,
+      "learning_rate": 1.971170662346477e-06,
+      "loss": 1.7585,
+      "step": 19395500
+    },
+    {
+      "epoch": 96.09,
+      "learning_rate": 1.9699320759203936e-06,
+      "loss": 1.7713,
+      "step": 19396000
+    },
+    {
+      "epoch": 96.1,
+      "learning_rate": 1.96869348949431e-06,
+      "loss": 1.7819,
+      "step": 19396500
+    },
+    {
+      "epoch": 96.1,
+      "learning_rate": 1.9674573802410785e-06,
+      "loss": 1.7963,
+      "step": 19397000
+    },
+    {
+      "epoch": 96.1,
+      "learning_rate": 1.966218793814995e-06,
+      "loss": 1.783,
+      "step": 19397500
+    },
+    {
+      "epoch": 96.1,
+      "learning_rate": 1.964980207388911e-06,
+      "loss": 1.7834,
+      "step": 19398000
+    },
+    {
+      "epoch": 96.11,
+      "learning_rate": 1.9637416209628276e-06,
+      "loss": 1.7907,
+      "step": 19398500
+    },
+    {
+      "epoch": 96.11,
+      "learning_rate": 1.9625030345367437e-06,
+      "loss": 1.7696,
+      "step": 19399000
+    },
+    {
+      "epoch": 96.11,
+      "learning_rate": 1.96126444811066e-06,
+      "loss": 1.7712,
+      "step": 19399500
+    },
+    {
+      "epoch": 96.11,
+      "learning_rate": 1.9600258616845767e-06,
+      "loss": 1.7661,
+      "step": 19400000
+    },
+    {
+      "epoch": 96.12,
+      "learning_rate": 1.958789752431345e-06,
+      "loss": 1.7976,
+      "step": 19400500
+    },
+    {
+      "epoch": 96.12,
+      "learning_rate": 1.9575511660052616e-06,
+      "loss": 1.7645,
+      "step": 19401000
+    },
+    {
+      "epoch": 96.12,
+      "learning_rate": 1.956312579579178e-06,
+      "loss": 1.7648,
+      "step": 19401500
+    },
+    {
+      "epoch": 96.12,
+      "learning_rate": 1.9550739931530946e-06,
+      "loss": 1.8045,
+      "step": 19402000
+    },
+    {
+      "epoch": 96.13,
+      "learning_rate": 1.9538354067270106e-06,
+      "loss": 1.7874,
+      "step": 19402500
+    },
+    {
+      "epoch": 96.13,
+      "learning_rate": 1.9525992974737795e-06,
+      "loss": 1.7568,
+      "step": 19403000
+    },
+    {
+      "epoch": 96.13,
+      "learning_rate": 1.9513607110476955e-06,
+      "loss": 1.7455,
+      "step": 19403500
+    },
+    {
+      "epoch": 96.13,
+      "learning_rate": 1.950122124621612e-06,
+      "loss": 1.7711,
+      "step": 19404000
+    },
+    {
+      "epoch": 96.14,
+      "learning_rate": 1.9488835381955285e-06,
+      "loss": 1.77,
+      "step": 19404500
+    },
+    {
+      "epoch": 96.14,
+      "learning_rate": 1.9476449517694446e-06,
+      "loss": 1.7728,
+      "step": 19405000
+    },
+    {
+      "epoch": 96.14,
+      "learning_rate": 1.9464088425162134e-06,
+      "loss": 1.756,
+      "step": 19405500
+    },
+    {
+      "epoch": 96.14,
+      "learning_rate": 1.9451702560901295e-06,
+      "loss": 1.7455,
+      "step": 19406000
+    },
+    {
+      "epoch": 96.15,
+      "learning_rate": 1.943931669664046e-06,
+      "loss": 1.7821,
+      "step": 19406500
+    },
+    {
+      "epoch": 96.15,
+      "learning_rate": 1.942693083237962e-06,
+      "loss": 1.7711,
+      "step": 19407000
+    },
+    {
+      "epoch": 96.15,
+      "learning_rate": 1.9414544968118786e-06,
+      "loss": 1.7727,
+      "step": 19407500
+    },
+    {
+      "epoch": 96.15,
+      "learning_rate": 1.940215910385795e-06,
+      "loss": 1.7787,
+      "step": 19408000
+    },
+    {
+      "epoch": 96.16,
+      "learning_rate": 1.938977323959711e-06,
+      "loss": 1.7777,
+      "step": 19408500
+    },
+    {
+      "epoch": 96.16,
+      "learning_rate": 1.9377387375336277e-06,
+      "loss": 1.7313,
+      "step": 19409000
+    },
+    {
+      "epoch": 96.16,
+      "learning_rate": 1.936502628280396e-06,
+      "loss": 1.7619,
+      "step": 19409500
+    },
+    {
+      "epoch": 96.16,
+      "learning_rate": 1.9352640418543126e-06,
+      "loss": 1.7638,
+      "step": 19410000
+    },
+    {
+      "epoch": 96.17,
+      "learning_rate": 1.934025455428229e-06,
+      "loss": 1.7694,
+      "step": 19410500
+    },
+    {
+      "epoch": 96.17,
+      "learning_rate": 1.932786869002145e-06,
+      "loss": 1.7807,
+      "step": 19411000
+    },
+    {
+      "epoch": 96.17,
+      "learning_rate": 1.9315482825760617e-06,
+      "loss": 1.7888,
+      "step": 19411500
+    },
+    {
+      "epoch": 96.17,
+      "learning_rate": 1.9303096961499778e-06,
+      "loss": 1.7831,
+      "step": 19412000
+    },
+    {
+      "epoch": 96.18,
+      "learning_rate": 1.9290711097238947e-06,
+      "loss": 1.7615,
+      "step": 19412500
+    },
+    {
+      "epoch": 96.18,
+      "learning_rate": 1.9278325232978108e-06,
+      "loss": 1.7811,
+      "step": 19413000
+    },
+    {
+      "epoch": 96.18,
+      "learning_rate": 1.9265939368717273e-06,
+      "loss": 1.7717,
+      "step": 19413500
+    },
+    {
+      "epoch": 96.18,
+      "learning_rate": 1.9253578276184957e-06,
+      "loss": 1.7778,
+      "step": 19414000
+    },
+    {
+      "epoch": 96.19,
+      "learning_rate": 1.924121718365264e-06,
+      "loss": 1.7513,
+      "step": 19414500
+    },
+    {
+      "epoch": 96.19,
+      "learning_rate": 1.9228831319391805e-06,
+      "loss": 1.7684,
+      "step": 19415000
+    },
+    {
+      "epoch": 96.19,
+      "learning_rate": 1.921644545513097e-06,
+      "loss": 1.7681,
+      "step": 19415500
+    },
+    {
+      "epoch": 96.19,
+      "learning_rate": 1.9204059590870136e-06,
+      "loss": 1.7838,
+      "step": 19416000
+    },
+    {
+      "epoch": 96.2,
+      "learning_rate": 1.9191673726609296e-06,
+      "loss": 1.7716,
+      "step": 19416500
+    },
+    {
+      "epoch": 96.2,
+      "learning_rate": 1.917928786234846e-06,
+      "loss": 1.7642,
+      "step": 19417000
+    },
+    {
+      "epoch": 96.2,
+      "learning_rate": 1.9166901998087622e-06,
+      "loss": 1.7804,
+      "step": 19417500
+    },
+    {
+      "epoch": 96.2,
+      "learning_rate": 1.915454090555531e-06,
+      "loss": 1.7558,
+      "step": 19418000
+    },
+    {
+      "epoch": 96.21,
+      "learning_rate": 1.914215504129447e-06,
+      "loss": 1.7849,
+      "step": 19418500
+    },
+    {
+      "epoch": 96.21,
+      "learning_rate": 1.912979394876216e-06,
+      "loss": 1.7461,
+      "step": 19419000
+    },
+    {
+      "epoch": 96.21,
+      "learning_rate": 1.9117432856229843e-06,
+      "loss": 1.7468,
+      "step": 19419500
+    },
+    {
+      "epoch": 96.21,
+      "learning_rate": 1.9105046991969004e-06,
+      "loss": 1.7637,
+      "step": 19420000
+    },
+    {
+      "epoch": 96.22,
+      "learning_rate": 1.9092661127708173e-06,
+      "loss": 1.7726,
+      "step": 19420500
+    },
+    {
+      "epoch": 96.22,
+      "learning_rate": 1.9080275263447334e-06,
+      "loss": 1.772,
+      "step": 19421000
+    },
+    {
+      "epoch": 96.22,
+      "learning_rate": 1.90678893991865e-06,
+      "loss": 1.7556,
+      "step": 19421500
+    },
+    {
+      "epoch": 96.22,
+      "learning_rate": 1.9055503534925662e-06,
+      "loss": 1.7624,
+      "step": 19422000
+    },
+    {
+      "epoch": 96.23,
+      "learning_rate": 1.9043117670664825e-06,
+      "loss": 1.769,
+      "step": 19422500
+    },
+    {
+      "epoch": 96.23,
+      "learning_rate": 1.9030731806403988e-06,
+      "loss": 1.7401,
+      "step": 19423000
+    },
+    {
+      "epoch": 96.23,
+      "learning_rate": 1.901834594214315e-06,
+      "loss": 1.7865,
+      "step": 19423500
+    },
+    {
+      "epoch": 96.23,
+      "learning_rate": 1.9005960077882316e-06,
+      "loss": 1.7866,
+      "step": 19424000
+    },
+    {
+      "epoch": 96.24,
+      "learning_rate": 1.8993598985350002e-06,
+      "loss": 1.7742,
+      "step": 19424500
+    },
+    {
+      "epoch": 96.24,
+      "learning_rate": 1.8981213121089165e-06,
+      "loss": 1.758,
+      "step": 19425000
+    },
+    {
+      "epoch": 96.24,
+      "learning_rate": 1.8968827256828328e-06,
+      "loss": 1.7676,
+      "step": 19425500
+    },
+    {
+      "epoch": 96.24,
+      "learning_rate": 1.895644139256749e-06,
+      "loss": 1.7869,
+      "step": 19426000
+    },
+    {
+      "epoch": 96.25,
+      "learning_rate": 1.8944080300035177e-06,
+      "loss": 1.7515,
+      "step": 19426500
+    },
+    {
+      "epoch": 96.25,
+      "learning_rate": 1.893169443577434e-06,
+      "loss": 1.747,
+      "step": 19427000
+    },
+    {
+      "epoch": 96.25,
+      "learning_rate": 1.8919308571513502e-06,
+      "loss": 1.7568,
+      "step": 19427500
+    },
+    {
+      "epoch": 96.25,
+      "learning_rate": 1.8906922707252667e-06,
+      "loss": 1.7698,
+      "step": 19428000
+    },
+    {
+      "epoch": 96.26,
+      "learning_rate": 1.889453684299183e-06,
+      "loss": 1.7609,
+      "step": 19428500
+    },
+    {
+      "epoch": 96.26,
+      "learning_rate": 1.8882175750459516e-06,
+      "loss": 1.7738,
+      "step": 19429000
+    },
+    {
+      "epoch": 96.26,
+      "learning_rate": 1.886978988619868e-06,
+      "loss": 1.7792,
+      "step": 19429500
+    },
+    {
+      "epoch": 96.26,
+      "learning_rate": 1.8857404021937842e-06,
+      "loss": 1.7755,
+      "step": 19430000
+    },
+    {
+      "epoch": 96.27,
+      "learning_rate": 1.8845018157677005e-06,
+      "loss": 1.7561,
+      "step": 19430500
+    },
+    {
+      "epoch": 96.27,
+      "learning_rate": 1.883263229341617e-06,
+      "loss": 1.7788,
+      "step": 19431000
+    },
+    {
+      "epoch": 96.27,
+      "learning_rate": 1.8820246429155335e-06,
+      "loss": 1.7869,
+      "step": 19431500
+    },
+    {
+      "epoch": 96.27,
+      "learning_rate": 1.88078605648945e-06,
+      "loss": 1.7798,
+      "step": 19432000
+    },
+    {
+      "epoch": 96.28,
+      "learning_rate": 1.8795474700633663e-06,
+      "loss": 1.7906,
+      "step": 19432500
+    },
+    {
+      "epoch": 96.28,
+      "learning_rate": 1.8783088836372826e-06,
+      "loss": 1.7366,
+      "step": 19433000
+    },
+    {
+      "epoch": 96.28,
+      "learning_rate": 1.877070297211199e-06,
+      "loss": 1.7729,
+      "step": 19433500
+    },
+    {
+      "epoch": 96.28,
+      "learning_rate": 1.8758317107851152e-06,
+      "loss": 1.7999,
+      "step": 19434000
+    },
+    {
+      "epoch": 96.29,
+      "learning_rate": 1.8745931243590315e-06,
+      "loss": 1.7681,
+      "step": 19434500
+    },
+    {
+      "epoch": 96.29,
+      "learning_rate": 1.873354537932948e-06,
+      "loss": 1.7897,
+      "step": 19435000
+    },
+    {
+      "epoch": 96.29,
+      "learning_rate": 1.8721184286797166e-06,
+      "loss": 1.7846,
+      "step": 19435500
+    },
+    {
+      "epoch": 96.29,
+      "learning_rate": 1.8708798422536329e-06,
+      "loss": 1.7692,
+      "step": 19436000
+    },
+    {
+      "epoch": 96.3,
+      "learning_rate": 1.8696437330004015e-06,
+      "loss": 1.7834,
+      "step": 19436500
+    },
+    {
+      "epoch": 96.3,
+      "learning_rate": 1.8684051465743178e-06,
+      "loss": 1.7857,
+      "step": 19437000
+    },
+    {
+      "epoch": 96.3,
+      "learning_rate": 1.867166560148234e-06,
+      "loss": 1.777,
+      "step": 19437500
+    },
+    {
+      "epoch": 96.3,
+      "learning_rate": 1.8659279737221504e-06,
+      "loss": 1.7707,
+      "step": 19438000
+    },
+    {
+      "epoch": 96.31,
+      "learning_rate": 1.864691864468919e-06,
+      "loss": 1.7588,
+      "step": 19438500
+    },
+    {
+      "epoch": 96.31,
+      "learning_rate": 1.8634532780428353e-06,
+      "loss": 1.7778,
+      "step": 19439000
+    },
+    {
+      "epoch": 96.31,
+      "learning_rate": 1.8622146916167518e-06,
+      "loss": 1.7751,
+      "step": 19439500
+    },
+    {
+      "epoch": 96.31,
+      "learning_rate": 1.860976105190668e-06,
+      "loss": 1.77,
+      "step": 19440000
+    },
+    {
+      "epoch": 96.31,
+      "learning_rate": 1.8597375187645843e-06,
+      "loss": 1.7558,
+      "step": 19440500
+    },
+    {
+      "epoch": 96.32,
+      "learning_rate": 1.8584989323385006e-06,
+      "loss": 1.7792,
+      "step": 19441000
+    },
+    {
+      "epoch": 96.32,
+      "learning_rate": 1.857260345912417e-06,
+      "loss": 1.7606,
+      "step": 19441500
+    },
+    {
+      "epoch": 96.32,
+      "learning_rate": 1.8560217594863336e-06,
+      "loss": 1.785,
+      "step": 19442000
+    },
+    {
+      "epoch": 96.32,
+      "learning_rate": 1.854785650233102e-06,
+      "loss": 1.7502,
+      "step": 19442500
+    },
+    {
+      "epoch": 96.33,
+      "learning_rate": 1.8535495409798706e-06,
+      "loss": 1.7639,
+      "step": 19443000
+    },
+    {
+      "epoch": 96.33,
+      "learning_rate": 1.852310954553787e-06,
+      "loss": 1.7676,
+      "step": 19443500
+    },
+    {
+      "epoch": 96.33,
+      "learning_rate": 1.8510723681277032e-06,
+      "loss": 1.7683,
+      "step": 19444000
+    },
+    {
+      "epoch": 96.33,
+      "learning_rate": 1.8498337817016195e-06,
+      "loss": 1.7692,
+      "step": 19444500
+    },
+    {
+      "epoch": 96.34,
+      "learning_rate": 1.8485976724483881e-06,
+      "loss": 1.7961,
+      "step": 19445000
+    },
+    {
+      "epoch": 96.34,
+      "learning_rate": 1.8473590860223044e-06,
+      "loss": 1.7885,
+      "step": 19445500
+    },
+    {
+      "epoch": 96.34,
+      "learning_rate": 1.8461204995962207e-06,
+      "loss": 1.7633,
+      "step": 19446000
+    },
+    {
+      "epoch": 96.34,
+      "learning_rate": 1.8448819131701372e-06,
+      "loss": 1.8001,
+      "step": 19446500
+    },
+    {
+      "epoch": 96.35,
+      "learning_rate": 1.8436433267440537e-06,
+      "loss": 1.7694,
+      "step": 19447000
+    },
+    {
+      "epoch": 96.35,
+      "learning_rate": 1.8424047403179702e-06,
+      "loss": 1.7665,
+      "step": 19447500
+    },
+    {
+      "epoch": 96.35,
+      "learning_rate": 1.8411661538918865e-06,
+      "loss": 1.7793,
+      "step": 19448000
+    },
+    {
+      "epoch": 96.35,
+      "learning_rate": 1.8399275674658028e-06,
+      "loss": 1.7837,
+      "step": 19448500
+    },
+    {
+      "epoch": 96.36,
+      "learning_rate": 1.838688981039719e-06,
+      "loss": 1.7564,
+      "step": 19449000
+    },
+    {
+      "epoch": 96.36,
+      "learning_rate": 1.8374503946136354e-06,
+      "loss": 1.7553,
+      "step": 19449500
+    },
+    {
+      "epoch": 96.36,
+      "learning_rate": 1.8362118081875519e-06,
+      "loss": 1.7409,
+      "step": 19450000
+    },
+    {
+      "epoch": 96.36,
+      "learning_rate": 1.8349732217614682e-06,
+      "loss": 1.7487,
+      "step": 19450500
+    },
+    {
+      "epoch": 96.37,
+      "learning_rate": 1.8337395896810889e-06,
+      "loss": 1.756,
+      "step": 19451000
+    },
+    {
+      "epoch": 96.37,
+      "learning_rate": 1.8325034804278573e-06,
+      "loss": 1.7658,
+      "step": 19451500
+    },
+    {
+      "epoch": 96.37,
+      "learning_rate": 1.831264894001774e-06,
+      "loss": 1.77,
+      "step": 19452000
+    },
+    {
+      "epoch": 96.37,
+      "learning_rate": 1.8300263075756903e-06,
+      "loss": 1.8026,
+      "step": 19452500
+    },
+    {
+      "epoch": 96.38,
+      "learning_rate": 1.8287877211496066e-06,
+      "loss": 1.7447,
+      "step": 19453000
+    },
+    {
+      "epoch": 96.38,
+      "learning_rate": 1.8275491347235228e-06,
+      "loss": 1.7742,
+      "step": 19453500
+    },
+    {
+      "epoch": 96.38,
+      "learning_rate": 1.8263130254702914e-06,
+      "loss": 1.7909,
+      "step": 19454000
+    },
+    {
+      "epoch": 96.38,
+      "learning_rate": 1.8250744390442077e-06,
+      "loss": 1.7487,
+      "step": 19454500
+    },
+    {
+      "epoch": 96.39,
+      "learning_rate": 1.8238358526181242e-06,
+      "loss": 1.7742,
+      "step": 19455000
+    },
+    {
+      "epoch": 96.39,
+      "learning_rate": 1.8225972661920405e-06,
+      "loss": 1.7735,
+      "step": 19455500
+    },
+    {
+      "epoch": 96.39,
+      "learning_rate": 1.8213586797659568e-06,
+      "loss": 1.7729,
+      "step": 19456000
+    },
+    {
+      "epoch": 96.39,
+      "learning_rate": 1.8201200933398731e-06,
+      "loss": 1.7763,
+      "step": 19456500
+    },
+    {
+      "epoch": 96.4,
+      "learning_rate": 1.8188815069137894e-06,
+      "loss": 1.7633,
+      "step": 19457000
+    },
+    {
+      "epoch": 96.4,
+      "learning_rate": 1.817645397660558e-06,
+      "loss": 1.7945,
+      "step": 19457500
+    },
+    {
+      "epoch": 96.4,
+      "learning_rate": 1.8164068112344743e-06,
+      "loss": 1.7932,
+      "step": 19458000
+    },
+    {
+      "epoch": 96.4,
+      "learning_rate": 1.8151682248083908e-06,
+      "loss": 1.7745,
+      "step": 19458500
+    },
+    {
+      "epoch": 96.41,
+      "learning_rate": 1.813929638382307e-06,
+      "loss": 1.7511,
+      "step": 19459000
+    },
+    {
+      "epoch": 96.41,
+      "learning_rate": 1.8126910519562234e-06,
+      "loss": 1.774,
+      "step": 19459500
+    },
+    {
+      "epoch": 96.41,
+      "learning_rate": 1.8114524655301397e-06,
+      "loss": 1.7727,
+      "step": 19460000
+    },
+    {
+      "epoch": 96.41,
+      "learning_rate": 1.810213879104056e-06,
+      "loss": 1.784,
+      "step": 19460500
+    },
+    {
+      "epoch": 96.42,
+      "learning_rate": 1.8089752926779727e-06,
+      "loss": 1.7868,
+      "step": 19461000
+    },
+    {
+      "epoch": 96.42,
+      "learning_rate": 1.807736706251889e-06,
+      "loss": 1.7691,
+      "step": 19461500
+    },
+    {
+      "epoch": 96.42,
+      "learning_rate": 1.8064981198258055e-06,
+      "loss": 1.7785,
+      "step": 19462000
+    },
+    {
+      "epoch": 96.42,
+      "learning_rate": 1.8052595333997218e-06,
+      "loss": 1.7849,
+      "step": 19462500
+    },
+    {
+      "epoch": 96.43,
+      "learning_rate": 1.804020946973638e-06,
+      "loss": 1.7861,
+      "step": 19463000
+    },
+    {
+      "epoch": 96.43,
+      "learning_rate": 1.8027823605475544e-06,
+      "loss": 1.7854,
+      "step": 19463500
+    },
+    {
+      "epoch": 96.43,
+      "learning_rate": 1.801546251294323e-06,
+      "loss": 1.7604,
+      "step": 19464000
+    },
+    {
+      "epoch": 96.43,
+      "learning_rate": 1.8003076648682393e-06,
+      "loss": 1.7854,
+      "step": 19464500
+    },
+    {
+      "epoch": 96.44,
+      "learning_rate": 1.7990715556150079e-06,
+      "loss": 1.7571,
+      "step": 19465000
+    },
+    {
+      "epoch": 96.44,
+      "learning_rate": 1.7978329691889242e-06,
+      "loss": 1.7415,
+      "step": 19465500
+    },
+    {
+      "epoch": 96.44,
+      "learning_rate": 1.7965993371085448e-06,
+      "loss": 1.7779,
+      "step": 19466000
+    },
+    {
+      "epoch": 96.44,
+      "learning_rate": 1.7953607506824611e-06,
+      "loss": 1.77,
+      "step": 19466500
+    },
+    {
+      "epoch": 96.45,
+      "learning_rate": 1.7941221642563774e-06,
+      "loss": 1.7762,
+      "step": 19467000
+    },
+    {
+      "epoch": 96.45,
+      "learning_rate": 1.7928835778302941e-06,
+      "loss": 1.7579,
+      "step": 19467500
+    },
+    {
+      "epoch": 96.45,
+      "learning_rate": 1.7916449914042104e-06,
+      "loss": 1.7831,
+      "step": 19468000
+    },
+    {
+      "epoch": 96.45,
+      "learning_rate": 1.7904064049781267e-06,
+      "loss": 1.7733,
+      "step": 19468500
+    },
+    {
+      "epoch": 96.46,
+      "learning_rate": 1.789167818552043e-06,
+      "loss": 1.772,
+      "step": 19469000
+    },
+    {
+      "epoch": 96.46,
+      "learning_rate": 1.7879317092988116e-06,
+      "loss": 1.7711,
+      "step": 19469500
+    },
+    {
+      "epoch": 96.46,
+      "learning_rate": 1.786693122872728e-06,
+      "loss": 1.7819,
+      "step": 19470000
+    },
+    {
+      "epoch": 96.46,
+      "learning_rate": 1.7854545364466444e-06,
+      "loss": 1.7565,
+      "step": 19470500
+    },
+    {
+      "epoch": 96.47,
+      "learning_rate": 1.7842159500205607e-06,
+      "loss": 1.7655,
+      "step": 19471000
+    },
+    {
+      "epoch": 96.47,
+      "learning_rate": 1.782977363594477e-06,
+      "loss": 1.7633,
+      "step": 19471500
+    },
+    {
+      "epoch": 96.47,
+      "learning_rate": 1.7817387771683933e-06,
+      "loss": 1.7601,
+      "step": 19472000
+    },
+    {
+      "epoch": 96.47,
+      "learning_rate": 1.7805001907423096e-06,
+      "loss": 1.773,
+      "step": 19472500
+    },
+    {
+      "epoch": 96.48,
+      "learning_rate": 1.779261604316226e-06,
+      "loss": 1.7568,
+      "step": 19473000
+    },
+    {
+      "epoch": 96.48,
+      "learning_rate": 1.7780230178901424e-06,
+      "loss": 1.7783,
+      "step": 19473500
+    },
+    {
+      "epoch": 96.48,
+      "learning_rate": 1.7767844314640587e-06,
+      "loss": 1.754,
+      "step": 19474000
+    },
+    {
+      "epoch": 96.48,
+      "learning_rate": 1.775545845037975e-06,
+      "loss": 1.7819,
+      "step": 19474500
+    },
+    {
+      "epoch": 96.49,
+      "learning_rate": 1.7743072586118917e-06,
+      "loss": 1.7975,
+      "step": 19475000
+    },
+    {
+      "epoch": 96.49,
+      "learning_rate": 1.7730711493586599e-06,
+      "loss": 1.7487,
+      "step": 19475500
+    },
+    {
+      "epoch": 96.49,
+      "learning_rate": 1.7718325629325762e-06,
+      "loss": 1.7486,
+      "step": 19476000
+    },
+    {
+      "epoch": 96.49,
+      "learning_rate": 1.7705939765064927e-06,
+      "loss": 1.7754,
+      "step": 19476500
+    },
+    {
+      "epoch": 96.5,
+      "learning_rate": 1.7693553900804092e-06,
+      "loss": 1.775,
+      "step": 19477000
+    },
+    {
+      "epoch": 96.5,
+      "learning_rate": 1.7681168036543257e-06,
+      "loss": 1.8026,
+      "step": 19477500
+    },
+    {
+      "epoch": 96.5,
+      "learning_rate": 1.766878217228242e-06,
+      "loss": 1.7793,
+      "step": 19478000
+    },
+    {
+      "epoch": 96.5,
+      "learning_rate": 1.7656421079750106e-06,
+      "loss": 1.7813,
+      "step": 19478500
+    },
+    {
+      "epoch": 96.51,
+      "learning_rate": 1.7644035215489269e-06,
+      "loss": 1.7477,
+      "step": 19479000
+    },
+    {
+      "epoch": 96.51,
+      "learning_rate": 1.7631649351228432e-06,
+      "loss": 1.7596,
+      "step": 19479500
+    },
+    {
+      "epoch": 96.51,
+      "learning_rate": 1.7619263486967594e-06,
+      "loss": 1.7585,
+      "step": 19480000
+    },
+    {
+      "epoch": 96.51,
+      "learning_rate": 1.760687762270676e-06,
+      "loss": 1.8015,
+      "step": 19480500
+    },
+    {
+      "epoch": 96.52,
+      "learning_rate": 1.7594491758445922e-06,
+      "loss": 1.7809,
+      "step": 19481000
+    },
+    {
+      "epoch": 96.52,
+      "learning_rate": 1.7582105894185085e-06,
+      "loss": 1.7832,
+      "step": 19481500
+    },
+    {
+      "epoch": 96.52,
+      "learning_rate": 1.7569720029924248e-06,
+      "loss": 1.7846,
+      "step": 19482000
+    },
+    {
+      "epoch": 96.52,
+      "learning_rate": 1.7557358937391934e-06,
+      "loss": 1.786,
+      "step": 19482500
+    },
+    {
+      "epoch": 96.53,
+      "learning_rate": 1.7544973073131097e-06,
+      "loss": 1.7939,
+      "step": 19483000
+    },
+    {
+      "epoch": 96.53,
+      "learning_rate": 1.753258720887026e-06,
+      "loss": 1.7561,
+      "step": 19483500
+    },
+    {
+      "epoch": 96.53,
+      "learning_rate": 1.7520201344609425e-06,
+      "loss": 1.7863,
+      "step": 19484000
+    },
+    {
+      "epoch": 96.53,
+      "learning_rate": 1.7507840252077111e-06,
+      "loss": 1.7796,
+      "step": 19484500
+    },
+    {
+      "epoch": 96.54,
+      "learning_rate": 1.7495454387816274e-06,
+      "loss": 1.7723,
+      "step": 19485000
+    },
+    {
+      "epoch": 96.54,
+      "learning_rate": 1.7483068523555437e-06,
+      "loss": 1.762,
+      "step": 19485500
+    },
+    {
+      "epoch": 96.54,
+      "learning_rate": 1.74706826592946e-06,
+      "loss": 1.7802,
+      "step": 19486000
+    },
+    {
+      "epoch": 96.54,
+      "learning_rate": 1.7458296795033763e-06,
+      "loss": 1.7831,
+      "step": 19486500
+    },
+    {
+      "epoch": 96.55,
+      "learning_rate": 1.7445910930772926e-06,
+      "loss": 1.7912,
+      "step": 19487000
+    },
+    {
+      "epoch": 96.55,
+      "learning_rate": 1.7433525066512093e-06,
+      "loss": 1.7793,
+      "step": 19487500
+    },
+    {
+      "epoch": 96.55,
+      "learning_rate": 1.7421139202251256e-06,
+      "loss": 1.7749,
+      "step": 19488000
+    },
+    {
+      "epoch": 96.55,
+      "learning_rate": 1.740875333799042e-06,
+      "loss": 1.7621,
+      "step": 19488500
+    },
+    {
+      "epoch": 96.56,
+      "learning_rate": 1.7396367473729584e-06,
+      "loss": 1.7757,
+      "step": 19489000
+    },
+    {
+      "epoch": 96.56,
+      "learning_rate": 1.7383981609468747e-06,
+      "loss": 1.7612,
+      "step": 19489500
+    },
+    {
+      "epoch": 96.56,
+      "learning_rate": 1.7371620516936433e-06,
+      "loss": 1.7701,
+      "step": 19490000
+    },
+    {
+      "epoch": 96.56,
+      "learning_rate": 1.7359234652675596e-06,
+      "loss": 1.7683,
+      "step": 19490500
+    },
+    {
+      "epoch": 96.57,
+      "learning_rate": 1.7346873560143282e-06,
+      "loss": 1.7587,
+      "step": 19491000
+    },
+    {
+      "epoch": 96.57,
+      "learning_rate": 1.7334487695882445e-06,
+      "loss": 1.7667,
+      "step": 19491500
+    },
+    {
+      "epoch": 96.57,
+      "learning_rate": 1.732210183162161e-06,
+      "loss": 1.7667,
+      "step": 19492000
+    },
+    {
+      "epoch": 96.57,
+      "learning_rate": 1.7309715967360773e-06,
+      "loss": 1.7855,
+      "step": 19492500
+    },
+    {
+      "epoch": 96.58,
+      "learning_rate": 1.7297330103099935e-06,
+      "loss": 1.7779,
+      "step": 19493000
+    },
+    {
+      "epoch": 96.58,
+      "learning_rate": 1.7284944238839098e-06,
+      "loss": 1.7761,
+      "step": 19493500
+    },
+    {
+      "epoch": 96.58,
+      "learning_rate": 1.7272583146306784e-06,
+      "loss": 1.7894,
+      "step": 19494000
+    },
+    {
+      "epoch": 96.58,
+      "learning_rate": 1.7260197282045947e-06,
+      "loss": 1.7668,
+      "step": 19494500
+    },
+    {
+      "epoch": 96.58,
+      "learning_rate": 1.724781141778511e-06,
+      "loss": 1.7783,
+      "step": 19495000
+    },
+    {
+      "epoch": 96.59,
+      "learning_rate": 1.7235425553524275e-06,
+      "loss": 1.7897,
+      "step": 19495500
+    },
+    {
+      "epoch": 96.59,
+      "learning_rate": 1.7223039689263438e-06,
+      "loss": 1.7871,
+      "step": 19496000
+    },
+    {
+      "epoch": 96.59,
+      "learning_rate": 1.7210678596731124e-06,
+      "loss": 1.7633,
+      "step": 19496500
+    },
+    {
+      "epoch": 96.59,
+      "learning_rate": 1.7198292732470287e-06,
+      "loss": 1.7783,
+      "step": 19497000
+    },
+    {
+      "epoch": 96.6,
+      "learning_rate": 1.718590686820945e-06,
+      "loss": 1.7635,
+      "step": 19497500
+    },
+    {
+      "epoch": 96.6,
+      "learning_rate": 1.7173521003948613e-06,
+      "loss": 1.7797,
+      "step": 19498000
+    },
+    {
+      "epoch": 96.6,
+      "learning_rate": 1.7161135139687778e-06,
+      "loss": 1.7683,
+      "step": 19498500
+    },
+    {
+      "epoch": 96.6,
+      "learning_rate": 1.714874927542694e-06,
+      "loss": 1.7766,
+      "step": 19499000
+    },
+    {
+      "epoch": 96.61,
+      "learning_rate": 1.7136363411166104e-06,
+      "loss": 1.7482,
+      "step": 19499500
+    },
+    {
+      "epoch": 96.61,
+      "learning_rate": 1.7123977546905271e-06,
+      "loss": 1.7579,
+      "step": 19500000
+    },
+    {
+      "epoch": 96.61,
+      "learning_rate": 1.7111591682644434e-06,
+      "loss": 1.751,
+      "step": 19500500
+    },
+    {
+      "epoch": 96.61,
+      "learning_rate": 1.7099205818383597e-06,
+      "loss": 1.7741,
+      "step": 19501000
+    },
+    {
+      "epoch": 96.62,
+      "learning_rate": 1.708681995412276e-06,
+      "loss": 1.7752,
+      "step": 19501500
+    },
+    {
+      "epoch": 96.62,
+      "learning_rate": 1.7074434089861923e-06,
+      "loss": 1.7571,
+      "step": 19502000
+    },
+    {
+      "epoch": 96.62,
+      "learning_rate": 1.7062048225601088e-06,
+      "loss": 1.7799,
+      "step": 19502500
+    },
+    {
+      "epoch": 96.62,
+      "learning_rate": 1.704966236134025e-06,
+      "loss": 1.7797,
+      "step": 19503000
+    },
+    {
+      "epoch": 96.63,
+      "learning_rate": 1.7037276497079414e-06,
+      "loss": 1.8,
+      "step": 19503500
+    },
+    {
+      "epoch": 96.63,
+      "learning_rate": 1.7024890632818577e-06,
+      "loss": 1.7566,
+      "step": 19504000
+    },
+    {
+      "epoch": 96.63,
+      "learning_rate": 1.701250476855774e-06,
+      "loss": 1.7812,
+      "step": 19504500
+    },
+    {
+      "epoch": 96.63,
+      "learning_rate": 1.7000143676025426e-06,
+      "loss": 1.7622,
+      "step": 19505000
+    },
+    {
+      "epoch": 96.64,
+      "learning_rate": 1.698775781176459e-06,
+      "loss": 1.7617,
+      "step": 19505500
+    },
+    {
+      "epoch": 96.64,
+      "learning_rate": 1.6975371947503753e-06,
+      "loss": 1.7571,
+      "step": 19506000
+    },
+    {
+      "epoch": 96.64,
+      "learning_rate": 1.696301085497144e-06,
+      "loss": 1.7746,
+      "step": 19506500
+    },
+    {
+      "epoch": 96.64,
+      "learning_rate": 1.6950624990710602e-06,
+      "loss": 1.7793,
+      "step": 19507000
+    },
+    {
+      "epoch": 96.65,
+      "learning_rate": 1.6938239126449765e-06,
+      "loss": 1.768,
+      "step": 19507500
+    },
+    {
+      "epoch": 96.65,
+      "learning_rate": 1.6925853262188928e-06,
+      "loss": 1.7824,
+      "step": 19508000
+    },
+    {
+      "epoch": 96.65,
+      "learning_rate": 1.6913467397928091e-06,
+      "loss": 1.7817,
+      "step": 19508500
+    },
+    {
+      "epoch": 96.65,
+      "learning_rate": 1.6901081533667256e-06,
+      "loss": 1.7621,
+      "step": 19509000
+    },
+    {
+      "epoch": 96.66,
+      "learning_rate": 1.6888720441134942e-06,
+      "loss": 1.7791,
+      "step": 19509500
+    },
+    {
+      "epoch": 96.66,
+      "learning_rate": 1.6876334576874105e-06,
+      "loss": 1.7719,
+      "step": 19510000
+    },
+    {
+      "epoch": 96.66,
+      "learning_rate": 1.6863948712613268e-06,
+      "loss": 1.7583,
+      "step": 19510500
+    },
+    {
+      "epoch": 96.66,
+      "learning_rate": 1.6851562848352435e-06,
+      "loss": 1.7641,
+      "step": 19511000
+    },
+    {
+      "epoch": 96.67,
+      "learning_rate": 1.6839176984091598e-06,
+      "loss": 1.7687,
+      "step": 19511500
+    },
+    {
+      "epoch": 96.67,
+      "learning_rate": 1.6826791119830761e-06,
+      "loss": 1.771,
+      "step": 19512000
+    },
+    {
+      "epoch": 96.67,
+      "learning_rate": 1.6814430027298447e-06,
+      "loss": 1.7837,
+      "step": 19512500
+    },
+    {
+      "epoch": 96.67,
+      "learning_rate": 1.680204416303761e-06,
+      "loss": 1.7891,
+      "step": 19513000
+    },
+    {
+      "epoch": 96.68,
+      "learning_rate": 1.6789683070505294e-06,
+      "loss": 1.7869,
+      "step": 19513500
+    },
+    {
+      "epoch": 96.68,
+      "learning_rate": 1.6777297206244459e-06,
+      "loss": 1.7826,
+      "step": 19514000
+    },
+    {
+      "epoch": 96.68,
+      "learning_rate": 1.6764936113712143e-06,
+      "loss": 1.771,
+      "step": 19514500
+    },
+    {
+      "epoch": 96.68,
+      "learning_rate": 1.6752550249451306e-06,
+      "loss": 1.7739,
+      "step": 19515000
+    },
+    {
+      "epoch": 96.69,
+      "learning_rate": 1.6740164385190469e-06,
+      "loss": 1.7587,
+      "step": 19515500
+    },
+    {
+      "epoch": 96.69,
+      "learning_rate": 1.6727778520929636e-06,
+      "loss": 1.7664,
+      "step": 19516000
+    },
+    {
+      "epoch": 96.69,
+      "learning_rate": 1.6715392656668799e-06,
+      "loss": 1.7853,
+      "step": 19516500
+    },
+    {
+      "epoch": 96.69,
+      "learning_rate": 1.6703031564136483e-06,
+      "loss": 1.7426,
+      "step": 19517000
+    },
+    {
+      "epoch": 96.7,
+      "learning_rate": 1.6690645699875648e-06,
+      "loss": 1.7778,
+      "step": 19517500
+    },
+    {
+      "epoch": 96.7,
+      "learning_rate": 1.6678259835614813e-06,
+      "loss": 1.7772,
+      "step": 19518000
+    },
+    {
+      "epoch": 96.7,
+      "learning_rate": 1.6665873971353976e-06,
+      "loss": 1.7556,
+      "step": 19518500
+    },
+    {
+      "epoch": 96.7,
+      "learning_rate": 1.6653488107093139e-06,
+      "loss": 1.7885,
+      "step": 19519000
+    },
+    {
+      "epoch": 96.71,
+      "learning_rate": 1.6641102242832301e-06,
+      "loss": 1.769,
+      "step": 19519500
+    },
+    {
+      "epoch": 96.71,
+      "learning_rate": 1.6628716378571464e-06,
+      "loss": 1.7393,
+      "step": 19520000
+    },
+    {
+      "epoch": 96.71,
+      "learning_rate": 1.6616330514310627e-06,
+      "loss": 1.759,
+      "step": 19520500
+    },
+    {
+      "epoch": 96.71,
+      "learning_rate": 1.6603969421778313e-06,
+      "loss": 1.7882,
+      "step": 19521000
+    },
+    {
+      "epoch": 96.72,
+      "learning_rate": 1.6591583557517478e-06,
+      "loss": 1.7764,
+      "step": 19521500
+    },
+    {
+      "epoch": 96.72,
+      "learning_rate": 1.6579197693256641e-06,
+      "loss": 1.7591,
+      "step": 19522000
+    },
+    {
+      "epoch": 96.72,
+      "learning_rate": 1.6566811828995804e-06,
+      "loss": 1.7775,
+      "step": 19522500
+    },
+    {
+      "epoch": 96.72,
+      "learning_rate": 1.6554425964734967e-06,
+      "loss": 1.7543,
+      "step": 19523000
+    },
+    {
+      "epoch": 96.73,
+      "learning_rate": 1.654204010047413e-06,
+      "loss": 1.7792,
+      "step": 19523500
+    },
+    {
+      "epoch": 96.73,
+      "learning_rate": 1.6529654236213295e-06,
+      "loss": 1.77,
+      "step": 19524000
+    },
+    {
+      "epoch": 96.73,
+      "learning_rate": 1.6517268371952458e-06,
+      "loss": 1.7766,
+      "step": 19524500
+    },
+    {
+      "epoch": 96.73,
+      "learning_rate": 1.6504907279420144e-06,
+      "loss": 1.7763,
+      "step": 19525000
+    },
+    {
+      "epoch": 96.74,
+      "learning_rate": 1.6492521415159307e-06,
+      "loss": 1.8082,
+      "step": 19525500
+    },
+    {
+      "epoch": 96.74,
+      "learning_rate": 1.6480160322626993e-06,
+      "loss": 1.8005,
+      "step": 19526000
+    },
+    {
+      "epoch": 96.74,
+      "learning_rate": 1.6467774458366156e-06,
+      "loss": 1.7744,
+      "step": 19526500
+    },
+    {
+      "epoch": 96.74,
+      "learning_rate": 1.6455388594105319e-06,
+      "loss": 1.7949,
+      "step": 19527000
+    },
+    {
+      "epoch": 96.75,
+      "learning_rate": 1.6443002729844482e-06,
+      "loss": 1.7659,
+      "step": 19527500
+    },
+    {
+      "epoch": 96.75,
+      "learning_rate": 1.6430616865583647e-06,
+      "loss": 1.7715,
+      "step": 19528000
+    },
+    {
+      "epoch": 96.75,
+      "learning_rate": 1.6418231001322812e-06,
+      "loss": 1.7802,
+      "step": 19528500
+    },
+    {
+      "epoch": 96.75,
+      "learning_rate": 1.6405869908790496e-06,
+      "loss": 1.7755,
+      "step": 19529000
+    },
+    {
+      "epoch": 96.76,
+      "learning_rate": 1.6393484044529659e-06,
+      "loss": 1.77,
+      "step": 19529500
+    },
+    {
+      "epoch": 96.76,
+      "learning_rate": 1.6381098180268826e-06,
+      "loss": 1.7753,
+      "step": 19530000
+    },
+    {
+      "epoch": 96.76,
+      "learning_rate": 1.6368712316007989e-06,
+      "loss": 1.7573,
+      "step": 19530500
+    },
+    {
+      "epoch": 96.76,
+      "learning_rate": 1.6356326451747152e-06,
+      "loss": 1.7937,
+      "step": 19531000
+    },
+    {
+      "epoch": 96.77,
+      "learning_rate": 1.6343940587486315e-06,
+      "loss": 1.7662,
+      "step": 19531500
+    },
+    {
+      "epoch": 96.77,
+      "learning_rate": 1.6331579494954e-06,
+      "loss": 1.7663,
+      "step": 19532000
+    },
+    {
+      "epoch": 96.77,
+      "learning_rate": 1.6319193630693163e-06,
+      "loss": 1.7576,
+      "step": 19532500
+    },
+    {
+      "epoch": 96.77,
+      "learning_rate": 1.6306807766432328e-06,
+      "loss": 1.7541,
+      "step": 19533000
+    },
+    {
+      "epoch": 96.78,
+      "learning_rate": 1.6294421902171491e-06,
+      "loss": 1.7779,
+      "step": 19533500
+    },
+    {
+      "epoch": 96.78,
+      "learning_rate": 1.6282060809639177e-06,
+      "loss": 1.7563,
+      "step": 19534000
+    },
+    {
+      "epoch": 96.78,
+      "learning_rate": 1.626967494537834e-06,
+      "loss": 1.7726,
+      "step": 19534500
+    },
+    {
+      "epoch": 96.78,
+      "learning_rate": 1.6257289081117503e-06,
+      "loss": 1.78,
+      "step": 19535000
+    },
+    {
+      "epoch": 96.79,
+      "learning_rate": 1.6244903216856666e-06,
+      "loss": 1.7495,
+      "step": 19535500
+    },
+    {
+      "epoch": 96.79,
+      "learning_rate": 1.6232517352595831e-06,
+      "loss": 1.7837,
+      "step": 19536000
+    },
+    {
+      "epoch": 96.79,
+      "learning_rate": 1.6220131488334994e-06,
+      "loss": 1.7874,
+      "step": 19536500
+    },
+    {
+      "epoch": 96.79,
+      "learning_rate": 1.6207745624074157e-06,
+      "loss": 1.78,
+      "step": 19537000
+    },
+    {
+      "epoch": 96.8,
+      "learning_rate": 1.6195384531541843e-06,
+      "loss": 1.7783,
+      "step": 19537500
+    },
+    {
+      "epoch": 96.8,
+      "learning_rate": 1.6182998667281006e-06,
+      "loss": 1.7689,
+      "step": 19538000
+    },
+    {
+      "epoch": 96.8,
+      "learning_rate": 1.6170612803020169e-06,
+      "loss": 1.7706,
+      "step": 19538500
+    },
+    {
+      "epoch": 96.8,
+      "learning_rate": 1.6158226938759332e-06,
+      "loss": 1.7943,
+      "step": 19539000
+    },
+    {
+      "epoch": 96.81,
+      "learning_rate": 1.6145841074498497e-06,
+      "loss": 1.7748,
+      "step": 19539500
+    },
+    {
+      "epoch": 96.81,
+      "learning_rate": 1.613345521023766e-06,
+      "loss": 1.7902,
+      "step": 19540000
+    },
+    {
+      "epoch": 96.81,
+      "learning_rate": 1.6121069345976823e-06,
+      "loss": 1.7764,
+      "step": 19540500
+    },
+    {
+      "epoch": 96.81,
+      "learning_rate": 1.610868348171599e-06,
+      "loss": 1.7821,
+      "step": 19541000
+    },
+    {
+      "epoch": 96.82,
+      "learning_rate": 1.6096297617455153e-06,
+      "loss": 1.7594,
+      "step": 19541500
+    },
+    {
+      "epoch": 96.82,
+      "learning_rate": 1.6083911753194316e-06,
+      "loss": 1.7809,
+      "step": 19542000
+    },
+    {
+      "epoch": 96.82,
+      "learning_rate": 1.607157543239052e-06,
+      "loss": 1.7691,
+      "step": 19542500
+    },
+    {
+      "epoch": 96.82,
+      "learning_rate": 1.6059214339858207e-06,
+      "loss": 1.7642,
+      "step": 19543000
+    },
+    {
+      "epoch": 96.83,
+      "learning_rate": 1.604682847559737e-06,
+      "loss": 1.7991,
+      "step": 19543500
+    },
+    {
+      "epoch": 96.83,
+      "learning_rate": 1.6034442611336534e-06,
+      "loss": 1.781,
+      "step": 19544000
+    },
+    {
+      "epoch": 96.83,
+      "learning_rate": 1.6022056747075697e-06,
+      "loss": 1.7573,
+      "step": 19544500
+    },
+    {
+      "epoch": 96.83,
+      "learning_rate": 1.6009695654543383e-06,
+      "loss": 1.7686,
+      "step": 19545000
+    },
+    {
+      "epoch": 96.84,
+      "learning_rate": 1.5997309790282546e-06,
+      "loss": 1.7765,
+      "step": 19545500
+    },
+    {
+      "epoch": 96.84,
+      "learning_rate": 1.598492392602171e-06,
+      "loss": 1.7812,
+      "step": 19546000
+    },
+    {
+      "epoch": 96.84,
+      "learning_rate": 1.5972538061760872e-06,
+      "loss": 1.7667,
+      "step": 19546500
+    },
+    {
+      "epoch": 96.84,
+      "learning_rate": 1.5960152197500037e-06,
+      "loss": 1.7856,
+      "step": 19547000
+    },
+    {
+      "epoch": 96.85,
+      "learning_rate": 1.5947766333239202e-06,
+      "loss": 1.7714,
+      "step": 19547500
+    },
+    {
+      "epoch": 96.85,
+      "learning_rate": 1.5935380468978367e-06,
+      "loss": 1.7427,
+      "step": 19548000
+    },
+    {
+      "epoch": 96.85,
+      "learning_rate": 1.592299460471753e-06,
+      "loss": 1.7997,
+      "step": 19548500
+    },
+    {
+      "epoch": 96.85,
+      "learning_rate": 1.5910608740456693e-06,
+      "loss": 1.7525,
+      "step": 19549000
+    },
+    {
+      "epoch": 96.85,
+      "learning_rate": 1.589824764792438e-06,
+      "loss": 1.7727,
+      "step": 19549500
+    },
+    {
+      "epoch": 96.86,
+      "learning_rate": 1.5885861783663542e-06,
+      "loss": 1.764,
+      "step": 19550000
+    },
+    {
+      "epoch": 96.86,
+      "learning_rate": 1.5873475919402705e-06,
+      "loss": 1.7971,
+      "step": 19550500
+    },
+    {
+      "epoch": 96.86,
+      "learning_rate": 1.5861090055141868e-06,
+      "loss": 1.7743,
+      "step": 19551000
+    },
+    {
+      "epoch": 96.86,
+      "learning_rate": 1.5848704190881033e-06,
+      "loss": 1.7605,
+      "step": 19551500
+    },
+    {
+      "epoch": 96.87,
+      "learning_rate": 1.5836318326620196e-06,
+      "loss": 1.7656,
+      "step": 19552000
+    },
+    {
+      "epoch": 96.87,
+      "learning_rate": 1.5823932462359359e-06,
+      "loss": 1.7754,
+      "step": 19552500
+    },
+    {
+      "epoch": 96.87,
+      "learning_rate": 1.5811546598098522e-06,
+      "loss": 1.7737,
+      "step": 19553000
+    },
+    {
+      "epoch": 96.87,
+      "learning_rate": 1.5799160733837685e-06,
+      "loss": 1.7744,
+      "step": 19553500
+    },
+    {
+      "epoch": 96.88,
+      "learning_rate": 1.578677486957685e-06,
+      "loss": 1.7661,
+      "step": 19554000
+    },
+    {
+      "epoch": 96.88,
+      "learning_rate": 1.5774389005316013e-06,
+      "loss": 1.7731,
+      "step": 19554500
+    },
+    {
+      "epoch": 96.88,
+      "learning_rate": 1.5762027912783699e-06,
+      "loss": 1.7722,
+      "step": 19555000
+    },
+    {
+      "epoch": 96.88,
+      "learning_rate": 1.5749642048522862e-06,
+      "loss": 1.7812,
+      "step": 19555500
+    },
+    {
+      "epoch": 96.89,
+      "learning_rate": 1.5737256184262025e-06,
+      "loss": 1.7647,
+      "step": 19556000
+    },
+    {
+      "epoch": 96.89,
+      "learning_rate": 1.5724870320001192e-06,
+      "loss": 1.7568,
+      "step": 19556500
+    },
+    {
+      "epoch": 96.89,
+      "learning_rate": 1.5712509227468873e-06,
+      "loss": 1.7795,
+      "step": 19557000
+    },
+    {
+      "epoch": 96.89,
+      "learning_rate": 1.5700123363208036e-06,
+      "loss": 1.7727,
+      "step": 19557500
+    },
+    {
+      "epoch": 96.9,
+      "learning_rate": 1.5687762270675722e-06,
+      "loss": 1.7771,
+      "step": 19558000
+    },
+    {
+      "epoch": 96.9,
+      "learning_rate": 1.5675376406414887e-06,
+      "loss": 1.7857,
+      "step": 19558500
+    },
+    {
+      "epoch": 96.9,
+      "learning_rate": 1.566299054215405e-06,
+      "loss": 1.7725,
+      "step": 19559000
+    },
+    {
+      "epoch": 96.9,
+      "learning_rate": 1.5650629449621736e-06,
+      "loss": 1.7768,
+      "step": 19559500
+    },
+    {
+      "epoch": 96.91,
+      "learning_rate": 1.56382435853609e-06,
+      "loss": 1.782,
+      "step": 19560000
+    },
+    {
+      "epoch": 96.91,
+      "learning_rate": 1.5625857721100062e-06,
+      "loss": 1.7727,
+      "step": 19560500
+    },
+    {
+      "epoch": 96.91,
+      "learning_rate": 1.5613471856839227e-06,
+      "loss": 1.7636,
+      "step": 19561000
+    },
+    {
+      "epoch": 96.91,
+      "learning_rate": 1.560108599257839e-06,
+      "loss": 1.7679,
+      "step": 19561500
+    },
+    {
+      "epoch": 96.92,
+      "learning_rate": 1.5588700128317553e-06,
+      "loss": 1.7707,
+      "step": 19562000
+    },
+    {
+      "epoch": 96.92,
+      "learning_rate": 1.5576314264056718e-06,
+      "loss": 1.7689,
+      "step": 19562500
+    },
+    {
+      "epoch": 96.92,
+      "learning_rate": 1.5563928399795883e-06,
+      "loss": 1.7666,
+      "step": 19563000
+    },
+    {
+      "epoch": 96.92,
+      "learning_rate": 1.5551542535535046e-06,
+      "loss": 1.7815,
+      "step": 19563500
+    },
+    {
+      "epoch": 96.93,
+      "learning_rate": 1.553915667127421e-06,
+      "loss": 1.7636,
+      "step": 19564000
+    },
+    {
+      "epoch": 96.93,
+      "learning_rate": 1.5526770807013372e-06,
+      "loss": 1.7943,
+      "step": 19564500
+    },
+    {
+      "epoch": 96.93,
+      "learning_rate": 1.5514384942752535e-06,
+      "loss": 1.7625,
+      "step": 19565000
+    },
+    {
+      "epoch": 96.93,
+      "learning_rate": 1.55019990784917e-06,
+      "loss": 1.7836,
+      "step": 19565500
+    },
+    {
+      "epoch": 96.94,
+      "learning_rate": 1.5489613214230865e-06,
+      "loss": 1.7826,
+      "step": 19566000
+    },
+    {
+      "epoch": 96.94,
+      "learning_rate": 1.5477227349970028e-06,
+      "loss": 1.7827,
+      "step": 19566500
+    },
+    {
+      "epoch": 96.94,
+      "learning_rate": 1.546484148570919e-06,
+      "loss": 1.7787,
+      "step": 19567000
+    },
+    {
+      "epoch": 96.94,
+      "learning_rate": 1.5452455621448354e-06,
+      "loss": 1.7784,
+      "step": 19567500
+    },
+    {
+      "epoch": 96.95,
+      "learning_rate": 1.5440069757187517e-06,
+      "loss": 1.7933,
+      "step": 19568000
+    },
+    {
+      "epoch": 96.95,
+      "learning_rate": 1.542768389292668e-06,
+      "loss": 1.7818,
+      "step": 19568500
+    },
+    {
+      "epoch": 96.95,
+      "learning_rate": 1.5415298028665845e-06,
+      "loss": 1.7579,
+      "step": 19569000
+    },
+    {
+      "epoch": 96.95,
+      "learning_rate": 1.540291216440501e-06,
+      "loss": 1.7744,
+      "step": 19569500
+    },
+    {
+      "epoch": 96.96,
+      "learning_rate": 1.5390575843601217e-06,
+      "loss": 1.7896,
+      "step": 19570000
+    },
+    {
+      "epoch": 96.96,
+      "learning_rate": 1.537818997934038e-06,
+      "loss": 1.7505,
+      "step": 19570500
+    },
+    {
+      "epoch": 96.96,
+      "learning_rate": 1.5365804115079542e-06,
+      "loss": 1.806,
+      "step": 19571000
+    },
+    {
+      "epoch": 96.96,
+      "learning_rate": 1.5353443022547228e-06,
+      "loss": 1.7725,
+      "step": 19571500
+    },
+    {
+      "epoch": 96.97,
+      "learning_rate": 1.5341057158286391e-06,
+      "loss": 1.7695,
+      "step": 19572000
+    },
+    {
+      "epoch": 96.97,
+      "learning_rate": 1.5328671294025554e-06,
+      "loss": 1.7656,
+      "step": 19572500
+    },
+    {
+      "epoch": 96.97,
+      "learning_rate": 1.531628542976472e-06,
+      "loss": 1.7648,
+      "step": 19573000
+    },
+    {
+      "epoch": 96.97,
+      "learning_rate": 1.5303899565503884e-06,
+      "loss": 1.7489,
+      "step": 19573500
+    },
+    {
+      "epoch": 96.98,
+      "learning_rate": 1.5291513701243047e-06,
+      "loss": 1.7772,
+      "step": 19574000
+    },
+    {
+      "epoch": 96.98,
+      "learning_rate": 1.5279152608710731e-06,
+      "loss": 1.7718,
+      "step": 19574500
+    },
+    {
+      "epoch": 96.98,
+      "learning_rate": 1.5266766744449894e-06,
+      "loss": 1.7679,
+      "step": 19575000
+    },
+    {
+      "epoch": 96.98,
+      "learning_rate": 1.525438088018906e-06,
+      "loss": 1.785,
+      "step": 19575500
+    },
+    {
+      "epoch": 96.99,
+      "learning_rate": 1.5241995015928222e-06,
+      "loss": 1.7767,
+      "step": 19576000
+    },
+    {
+      "epoch": 96.99,
+      "learning_rate": 1.5229633923395906e-06,
+      "loss": 1.7805,
+      "step": 19576500
+    },
+    {
+      "epoch": 96.99,
+      "learning_rate": 1.521724805913507e-06,
+      "loss": 1.7484,
+      "step": 19577000
+    },
+    {
+      "epoch": 96.99,
+      "learning_rate": 1.5204862194874236e-06,
+      "loss": 1.755,
+      "step": 19577500
+    },
+    {
+      "epoch": 97.0,
+      "learning_rate": 1.51924763306134e-06,
+      "loss": 1.7707,
+      "step": 19578000
+    },
+    {
+      "epoch": 97.0,
+      "learning_rate": 1.5180090466352562e-06,
+      "loss": 1.7597,
+      "step": 19578500
+    },
+    {
+      "epoch": 97.0,
+      "eval_accuracy": 0.6871512097341254,
+      "eval_accuracy_mlm": 0.6487664185147802,
+      "eval_accuracy_nsp": 0.8682297938099851,
+      "eval_loss": 2.3128037452697754,
+      "eval_runtime": 147.1489,
+      "eval_samples_per_second": 1732.66,
+      "eval_steps_per_second": 72.199,
+      "step": 19578771
+    },
+    {
+      "epoch": 97.0,
+      "learning_rate": 1.5167704602091725e-06,
+      "loss": 1.7546,
+      "step": 19579000
+    },
+    {
+      "epoch": 97.0,
+      "learning_rate": 1.5155318737830888e-06,
+      "loss": 1.7672,
+      "step": 19579500
+    },
+    {
+      "epoch": 97.01,
+      "learning_rate": 1.5142932873570053e-06,
+      "loss": 1.7602,
+      "step": 19580000
+    },
+    {
+      "epoch": 97.01,
+      "learning_rate": 1.5130547009309216e-06,
+      "loss": 1.7658,
+      "step": 19580500
+    },
+    {
+      "epoch": 97.01,
+      "learning_rate": 1.511816114504838e-06,
+      "loss": 1.7547,
+      "step": 19581000
+    },
+    {
+      "epoch": 97.01,
+      "learning_rate": 1.5105800052516067e-06,
+      "loss": 1.7727,
+      "step": 19581500
+    },
+    {
+      "epoch": 97.02,
+      "learning_rate": 1.509343895998375e-06,
+      "loss": 1.7694,
+      "step": 19582000
+    },
+    {
+      "epoch": 97.02,
+      "learning_rate": 1.5081053095722914e-06,
+      "loss": 1.7775,
+      "step": 19582500
+    },
+    {
+      "epoch": 97.02,
+      "learning_rate": 1.5068667231462079e-06,
+      "loss": 1.7942,
+      "step": 19583000
+    },
+    {
+      "epoch": 97.02,
+      "learning_rate": 1.5056281367201241e-06,
+      "loss": 1.7681,
+      "step": 19583500
+    },
+    {
+      "epoch": 97.03,
+      "learning_rate": 1.5043920274668925e-06,
+      "loss": 1.756,
+      "step": 19584000
+    },
+    {
+      "epoch": 97.03,
+      "learning_rate": 1.503153441040809e-06,
+      "loss": 1.7519,
+      "step": 19584500
+    },
+    {
+      "epoch": 97.03,
+      "learning_rate": 1.5019148546147255e-06,
+      "loss": 1.779,
+      "step": 19585000
+    },
+    {
+      "epoch": 97.03,
+      "learning_rate": 1.5006762681886418e-06,
+      "loss": 1.7867,
+      "step": 19585500
+    },
+    {
+      "epoch": 97.04,
+      "learning_rate": 1.4994376817625581e-06,
+      "loss": 1.7509,
+      "step": 19586000
+    },
+    {
+      "epoch": 97.04,
+      "learning_rate": 1.4981990953364744e-06,
+      "loss": 1.7484,
+      "step": 19586500
+    },
+    {
+      "epoch": 97.04,
+      "learning_rate": 1.4969605089103907e-06,
+      "loss": 1.7661,
+      "step": 19587000
+    },
+    {
+      "epoch": 97.04,
+      "learning_rate": 1.495721922484307e-06,
+      "loss": 1.7693,
+      "step": 19587500
+    },
+    {
+      "epoch": 97.05,
+      "learning_rate": 1.4944833360582235e-06,
+      "loss": 1.7697,
+      "step": 19588000
+    },
+    {
+      "epoch": 97.05,
+      "learning_rate": 1.49324474963214e-06,
+      "loss": 1.7784,
+      "step": 19588500
+    },
+    {
+      "epoch": 97.05,
+      "learning_rate": 1.4920061632060563e-06,
+      "loss": 1.7567,
+      "step": 19589000
+    },
+    {
+      "epoch": 97.05,
+      "learning_rate": 1.4907675767799726e-06,
+      "loss": 1.7582,
+      "step": 19589500
+    },
+    {
+      "epoch": 97.06,
+      "learning_rate": 1.489528990353889e-06,
+      "loss": 1.7645,
+      "step": 19590000
+    },
+    {
+      "epoch": 97.06,
+      "learning_rate": 1.4882904039278052e-06,
+      "loss": 1.7531,
+      "step": 19590500
+    },
+    {
+      "epoch": 97.06,
+      "learning_rate": 1.4870518175017217e-06,
+      "loss": 1.7213,
+      "step": 19591000
+    },
+    {
+      "epoch": 97.06,
+      "learning_rate": 1.4858132310756382e-06,
+      "loss": 1.7696,
+      "step": 19591500
+    },
+    {
+      "epoch": 97.07,
+      "learning_rate": 1.4845771218224066e-06,
+      "loss": 1.779,
+      "step": 19592000
+    },
+    {
+      "epoch": 97.07,
+      "learning_rate": 1.483338535396323e-06,
+      "loss": 1.7905,
+      "step": 19592500
+    },
+    {
+      "epoch": 97.07,
+      "learning_rate": 1.4820999489702394e-06,
+      "loss": 1.7474,
+      "step": 19593000
+    },
+    {
+      "epoch": 97.07,
+      "learning_rate": 1.4808613625441557e-06,
+      "loss": 1.7516,
+      "step": 19593500
+    },
+    {
+      "epoch": 97.08,
+      "learning_rate": 1.479622776118072e-06,
+      "loss": 1.7688,
+      "step": 19594000
+    },
+    {
+      "epoch": 97.08,
+      "learning_rate": 1.4783841896919883e-06,
+      "loss": 1.7769,
+      "step": 19594500
+    },
+    {
+      "epoch": 97.08,
+      "learning_rate": 1.4771456032659048e-06,
+      "loss": 1.7458,
+      "step": 19595000
+    },
+    {
+      "epoch": 97.08,
+      "learning_rate": 1.4759070168398213e-06,
+      "loss": 1.7846,
+      "step": 19595500
+    },
+    {
+      "epoch": 97.09,
+      "learning_rate": 1.4746684304137376e-06,
+      "loss": 1.7676,
+      "step": 19596000
+    },
+    {
+      "epoch": 97.09,
+      "learning_rate": 1.473432321160506e-06,
+      "loss": 1.7624,
+      "step": 19596500
+    },
+    {
+      "epoch": 97.09,
+      "learning_rate": 1.4721937347344225e-06,
+      "loss": 1.7679,
+      "step": 19597000
+    },
+    {
+      "epoch": 97.09,
+      "learning_rate": 1.4709551483083387e-06,
+      "loss": 1.8029,
+      "step": 19597500
+    },
+    {
+      "epoch": 97.1,
+      "learning_rate": 1.4697190390551071e-06,
+      "loss": 1.7733,
+      "step": 19598000
+    },
+    {
+      "epoch": 97.1,
+      "learning_rate": 1.4684804526290236e-06,
+      "loss": 1.7623,
+      "step": 19598500
+    },
+    {
+      "epoch": 97.1,
+      "learning_rate": 1.46724186620294e-06,
+      "loss": 1.7845,
+      "step": 19599000
+    },
+    {
+      "epoch": 97.1,
+      "learning_rate": 1.4660032797768564e-06,
+      "loss": 1.768,
+      "step": 19599500
+    },
+    {
+      "epoch": 97.11,
+      "learning_rate": 1.4647646933507727e-06,
+      "loss": 1.762,
+      "step": 19600000
+    },
+    {
+      "epoch": 97.11,
+      "learning_rate": 1.463526106924689e-06,
+      "loss": 1.7689,
+      "step": 19600500
+    },
+    {
+      "epoch": 97.11,
+      "learning_rate": 1.4622899976714576e-06,
+      "loss": 1.7591,
+      "step": 19601000
+    },
+    {
+      "epoch": 97.11,
+      "learning_rate": 1.461051411245374e-06,
+      "loss": 1.7873,
+      "step": 19601500
+    },
+    {
+      "epoch": 97.12,
+      "learning_rate": 1.4598153019921425e-06,
+      "loss": 1.7869,
+      "step": 19602000
+    },
+    {
+      "epoch": 97.12,
+      "learning_rate": 1.4585767155660588e-06,
+      "loss": 1.7642,
+      "step": 19602500
+    },
+    {
+      "epoch": 97.12,
+      "learning_rate": 1.4573381291399753e-06,
+      "loss": 1.7792,
+      "step": 19603000
+    },
+    {
+      "epoch": 97.12,
+      "learning_rate": 1.4560995427138916e-06,
+      "loss": 1.7682,
+      "step": 19603500
+    },
+    {
+      "epoch": 97.12,
+      "learning_rate": 1.4548609562878079e-06,
+      "loss": 1.8008,
+      "step": 19604000
+    },
+    {
+      "epoch": 97.13,
+      "learning_rate": 1.4536223698617242e-06,
+      "loss": 1.755,
+      "step": 19604500
+    },
+    {
+      "epoch": 97.13,
+      "learning_rate": 1.4523837834356407e-06,
+      "loss": 1.7603,
+      "step": 19605000
+    },
+    {
+      "epoch": 97.13,
+      "learning_rate": 1.451145197009557e-06,
+      "loss": 1.7713,
+      "step": 19605500
+    },
+    {
+      "epoch": 97.13,
+      "learning_rate": 1.4499090877563254e-06,
+      "loss": 1.7513,
+      "step": 19606000
+    },
+    {
+      "epoch": 97.14,
+      "learning_rate": 1.4486705013302419e-06,
+      "loss": 1.7599,
+      "step": 19606500
+    },
+    {
+      "epoch": 97.14,
+      "learning_rate": 1.4474319149041584e-06,
+      "loss": 1.7784,
+      "step": 19607000
+    },
+    {
+      "epoch": 97.14,
+      "learning_rate": 1.4461958056509268e-06,
+      "loss": 1.7667,
+      "step": 19607500
+    },
+    {
+      "epoch": 97.14,
+      "learning_rate": 1.4449572192248433e-06,
+      "loss": 1.7528,
+      "step": 19608000
+    },
+    {
+      "epoch": 97.15,
+      "learning_rate": 1.4437186327987596e-06,
+      "loss": 1.7417,
+      "step": 19608500
+    },
+    {
+      "epoch": 97.15,
+      "learning_rate": 1.442482523545528e-06,
+      "loss": 1.7629,
+      "step": 19609000
+    },
+    {
+      "epoch": 97.15,
+      "learning_rate": 1.4412439371194442e-06,
+      "loss": 1.7657,
+      "step": 19609500
+    },
+    {
+      "epoch": 97.15,
+      "learning_rate": 1.4400053506933607e-06,
+      "loss": 1.7576,
+      "step": 19610000
+    },
+    {
+      "epoch": 97.16,
+      "learning_rate": 1.438766764267277e-06,
+      "loss": 1.7852,
+      "step": 19610500
+    },
+    {
+      "epoch": 97.16,
+      "learning_rate": 1.4375281778411935e-06,
+      "loss": 1.7767,
+      "step": 19611000
+    },
+    {
+      "epoch": 97.16,
+      "learning_rate": 1.4362895914151098e-06,
+      "loss": 1.7851,
+      "step": 19611500
+    },
+    {
+      "epoch": 97.16,
+      "learning_rate": 1.4350510049890261e-06,
+      "loss": 1.7677,
+      "step": 19612000
+    },
+    {
+      "epoch": 97.17,
+      "learning_rate": 1.4338124185629424e-06,
+      "loss": 1.7673,
+      "step": 19612500
+    },
+    {
+      "epoch": 97.17,
+      "learning_rate": 1.432573832136859e-06,
+      "loss": 1.7776,
+      "step": 19613000
+    },
+    {
+      "epoch": 97.17,
+      "learning_rate": 1.4313352457107752e-06,
+      "loss": 1.7592,
+      "step": 19613500
+    },
+    {
+      "epoch": 97.17,
+      "learning_rate": 1.4300991364575438e-06,
+      "loss": 1.765,
+      "step": 19614000
+    },
+    {
+      "epoch": 97.18,
+      "learning_rate": 1.4288605500314603e-06,
+      "loss": 1.7632,
+      "step": 19614500
+    },
+    {
+      "epoch": 97.18,
+      "learning_rate": 1.4276219636053766e-06,
+      "loss": 1.7646,
+      "step": 19615000
+    },
+    {
+      "epoch": 97.18,
+      "learning_rate": 1.426383377179293e-06,
+      "loss": 1.7751,
+      "step": 19615500
+    },
+    {
+      "epoch": 97.18,
+      "learning_rate": 1.4251447907532092e-06,
+      "loss": 1.7708,
+      "step": 19616000
+    },
+    {
+      "epoch": 97.19,
+      "learning_rate": 1.4239062043271255e-06,
+      "loss": 1.7755,
+      "step": 19616500
+    },
+    {
+      "epoch": 97.19,
+      "learning_rate": 1.4226676179010418e-06,
+      "loss": 1.7566,
+      "step": 19617000
+    },
+    {
+      "epoch": 97.19,
+      "learning_rate": 1.4214290314749583e-06,
+      "loss": 1.7606,
+      "step": 19617500
+    },
+    {
+      "epoch": 97.19,
+      "learning_rate": 1.4201904450488748e-06,
+      "loss": 1.7396,
+      "step": 19618000
+    },
+    {
+      "epoch": 97.2,
+      "learning_rate": 1.418951858622791e-06,
+      "loss": 1.7663,
+      "step": 19618500
+    },
+    {
+      "epoch": 97.2,
+      "learning_rate": 1.4177132721967074e-06,
+      "loss": 1.7751,
+      "step": 19619000
+    },
+    {
+      "epoch": 97.2,
+      "learning_rate": 1.416477162943476e-06,
+      "loss": 1.7858,
+      "step": 19619500
+    },
+    {
+      "epoch": 97.2,
+      "learning_rate": 1.4152385765173923e-06,
+      "loss": 1.7663,
+      "step": 19620000
+    },
+    {
+      "epoch": 97.21,
+      "learning_rate": 1.4139999900913086e-06,
+      "loss": 1.7891,
+      "step": 19620500
+    },
+    {
+      "epoch": 97.21,
+      "learning_rate": 1.412761403665225e-06,
+      "loss": 1.7552,
+      "step": 19621000
+    },
+    {
+      "epoch": 97.21,
+      "learning_rate": 1.4115228172391414e-06,
+      "loss": 1.7719,
+      "step": 19621500
+    },
+    {
+      "epoch": 97.21,
+      "learning_rate": 1.4102842308130579e-06,
+      "loss": 1.7584,
+      "step": 19622000
+    },
+    {
+      "epoch": 97.22,
+      "learning_rate": 1.4090456443869742e-06,
+      "loss": 1.7691,
+      "step": 19622500
+    },
+    {
+      "epoch": 97.22,
+      "learning_rate": 1.4078070579608905e-06,
+      "loss": 1.7712,
+      "step": 19623000
+    },
+    {
+      "epoch": 97.22,
+      "learning_rate": 1.406570948707659e-06,
+      "loss": 1.775,
+      "step": 19623500
+    },
+    {
+      "epoch": 97.22,
+      "learning_rate": 1.4053323622815753e-06,
+      "loss": 1.7804,
+      "step": 19624000
+    },
+    {
+      "epoch": 97.23,
+      "learning_rate": 1.4040962530283437e-06,
+      "loss": 1.7761,
+      "step": 19624500
+    },
+    {
+      "epoch": 97.23,
+      "learning_rate": 1.4028576666022602e-06,
+      "loss": 1.7656,
+      "step": 19625000
+    },
+    {
+      "epoch": 97.23,
+      "learning_rate": 1.4016190801761767e-06,
+      "loss": 1.7517,
+      "step": 19625500
+    },
+    {
+      "epoch": 97.23,
+      "learning_rate": 1.400380493750093e-06,
+      "loss": 1.7781,
+      "step": 19626000
+    },
+    {
+      "epoch": 97.24,
+      "learning_rate": 1.3991419073240093e-06,
+      "loss": 1.7756,
+      "step": 19626500
+    },
+    {
+      "epoch": 97.24,
+      "learning_rate": 1.3979033208979256e-06,
+      "loss": 1.7576,
+      "step": 19627000
+    },
+    {
+      "epoch": 97.24,
+      "learning_rate": 1.396664734471842e-06,
+      "loss": 1.7815,
+      "step": 19627500
+    },
+    {
+      "epoch": 97.24,
+      "learning_rate": 1.3954261480457584e-06,
+      "loss": 1.7735,
+      "step": 19628000
+    },
+    {
+      "epoch": 97.25,
+      "learning_rate": 1.394190038792527e-06,
+      "loss": 1.7687,
+      "step": 19628500
+    },
+    {
+      "epoch": 97.25,
+      "learning_rate": 1.3929514523664433e-06,
+      "loss": 1.7684,
+      "step": 19629000
+    },
+    {
+      "epoch": 97.25,
+      "learning_rate": 1.391715343113212e-06,
+      "loss": 1.7791,
+      "step": 19629500
+    },
+    {
+      "epoch": 97.25,
+      "learning_rate": 1.3904767566871282e-06,
+      "loss": 1.7797,
+      "step": 19630000
+    },
+    {
+      "epoch": 97.26,
+      "learning_rate": 1.3892406474338968e-06,
+      "loss": 1.7826,
+      "step": 19630500
+    },
+    {
+      "epoch": 97.26,
+      "learning_rate": 1.388002061007813e-06,
+      "loss": 1.7884,
+      "step": 19631000
+    },
+    {
+      "epoch": 97.26,
+      "learning_rate": 1.3867634745817294e-06,
+      "loss": 1.7643,
+      "step": 19631500
+    },
+    {
+      "epoch": 97.26,
+      "learning_rate": 1.3855248881556457e-06,
+      "loss": 1.7788,
+      "step": 19632000
+    },
+    {
+      "epoch": 97.27,
+      "learning_rate": 1.3842863017295622e-06,
+      "loss": 1.7674,
+      "step": 19632500
+    },
+    {
+      "epoch": 97.27,
+      "learning_rate": 1.3830477153034787e-06,
+      "loss": 1.7901,
+      "step": 19633000
+    },
+    {
+      "epoch": 97.27,
+      "learning_rate": 1.381809128877395e-06,
+      "loss": 1.778,
+      "step": 19633500
+    },
+    {
+      "epoch": 97.27,
+      "learning_rate": 1.3805705424513113e-06,
+      "loss": 1.7617,
+      "step": 19634000
+    },
+    {
+      "epoch": 97.28,
+      "learning_rate": 1.3793319560252276e-06,
+      "loss": 1.7737,
+      "step": 19634500
+    },
+    {
+      "epoch": 97.28,
+      "learning_rate": 1.3780933695991439e-06,
+      "loss": 1.7886,
+      "step": 19635000
+    },
+    {
+      "epoch": 97.28,
+      "learning_rate": 1.3768547831730601e-06,
+      "loss": 1.7649,
+      "step": 19635500
+    },
+    {
+      "epoch": 97.28,
+      "learning_rate": 1.3756161967469769e-06,
+      "loss": 1.7973,
+      "step": 19636000
+    },
+    {
+      "epoch": 97.29,
+      "learning_rate": 1.3743776103208932e-06,
+      "loss": 1.7982,
+      "step": 19636500
+    },
+    {
+      "epoch": 97.29,
+      "learning_rate": 1.3731415010676615e-06,
+      "loss": 1.7675,
+      "step": 19637000
+    },
+    {
+      "epoch": 97.29,
+      "learning_rate": 1.3719053918144301e-06,
+      "loss": 1.7869,
+      "step": 19637500
+    },
+    {
+      "epoch": 97.29,
+      "learning_rate": 1.3706692825611987e-06,
+      "loss": 1.782,
+      "step": 19638000
+    },
+    {
+      "epoch": 97.3,
+      "learning_rate": 1.369430696135115e-06,
+      "loss": 1.7823,
+      "step": 19638500
+    },
+    {
+      "epoch": 97.3,
+      "learning_rate": 1.3681921097090313e-06,
+      "loss": 1.7464,
+      "step": 19639000
+    },
+    {
+      "epoch": 97.3,
+      "learning_rate": 1.3669535232829476e-06,
+      "loss": 1.7815,
+      "step": 19639500
+    },
+    {
+      "epoch": 97.3,
+      "learning_rate": 1.3657149368568641e-06,
+      "loss": 1.7678,
+      "step": 19640000
+    },
+    {
+      "epoch": 97.31,
+      "learning_rate": 1.3644763504307804e-06,
+      "loss": 1.7916,
+      "step": 19640500
+    },
+    {
+      "epoch": 97.31,
+      "learning_rate": 1.363240241177549e-06,
+      "loss": 1.7716,
+      "step": 19641000
+    },
+    {
+      "epoch": 97.31,
+      "learning_rate": 1.3620016547514653e-06,
+      "loss": 1.7713,
+      "step": 19641500
+    },
+    {
+      "epoch": 97.31,
+      "learning_rate": 1.3607630683253816e-06,
+      "loss": 1.7461,
+      "step": 19642000
+    },
+    {
+      "epoch": 97.32,
+      "learning_rate": 1.359524481899298e-06,
+      "loss": 1.7682,
+      "step": 19642500
+    },
+    {
+      "epoch": 97.32,
+      "learning_rate": 1.3582858954732144e-06,
+      "loss": 1.7682,
+      "step": 19643000
+    },
+    {
+      "epoch": 97.32,
+      "learning_rate": 1.3570497862199828e-06,
+      "loss": 1.8012,
+      "step": 19643500
+    },
+    {
+      "epoch": 97.32,
+      "learning_rate": 1.3558111997938993e-06,
+      "loss": 1.7251,
+      "step": 19644000
+    },
+    {
+      "epoch": 97.33,
+      "learning_rate": 1.3545726133678158e-06,
+      "loss": 1.7824,
+      "step": 19644500
+    },
+    {
+      "epoch": 97.33,
+      "learning_rate": 1.353334026941732e-06,
+      "loss": 1.7649,
+      "step": 19645000
+    },
+    {
+      "epoch": 97.33,
+      "learning_rate": 1.3520954405156484e-06,
+      "loss": 1.7681,
+      "step": 19645500
+    },
+    {
+      "epoch": 97.33,
+      "learning_rate": 1.3508568540895647e-06,
+      "loss": 1.7781,
+      "step": 19646000
+    },
+    {
+      "epoch": 97.34,
+      "learning_rate": 1.349618267663481e-06,
+      "loss": 1.7733,
+      "step": 19646500
+    },
+    {
+      "epoch": 97.34,
+      "learning_rate": 1.3483796812373975e-06,
+      "loss": 1.7433,
+      "step": 19647000
+    },
+    {
+      "epoch": 97.34,
+      "learning_rate": 1.347141094811314e-06,
+      "loss": 1.7692,
+      "step": 19647500
+    },
+    {
+      "epoch": 97.34,
+      "learning_rate": 1.3459025083852303e-06,
+      "loss": 1.7837,
+      "step": 19648000
+    },
+    {
+      "epoch": 97.35,
+      "learning_rate": 1.3446639219591466e-06,
+      "loss": 1.7578,
+      "step": 19648500
+    },
+    {
+      "epoch": 97.35,
+      "learning_rate": 1.3434253355330628e-06,
+      "loss": 1.761,
+      "step": 19649000
+    },
+    {
+      "epoch": 97.35,
+      "learning_rate": 1.3421867491069791e-06,
+      "loss": 1.7524,
+      "step": 19649500
+    },
+    {
+      "epoch": 97.35,
+      "learning_rate": 1.3409481626808956e-06,
+      "loss": 1.7609,
+      "step": 19650000
+    },
+    {
+      "epoch": 97.36,
+      "learning_rate": 1.339712053427664e-06,
+      "loss": 1.7622,
+      "step": 19650500
+    },
+    {
+      "epoch": 97.36,
+      "learning_rate": 1.3384734670015805e-06,
+      "loss": 1.7524,
+      "step": 19651000
+    },
+    {
+      "epoch": 97.36,
+      "learning_rate": 1.3372348805754968e-06,
+      "loss": 1.7559,
+      "step": 19651500
+    },
+    {
+      "epoch": 97.36,
+      "learning_rate": 1.3359987713222654e-06,
+      "loss": 1.7725,
+      "step": 19652000
+    },
+    {
+      "epoch": 97.37,
+      "learning_rate": 1.3347601848961817e-06,
+      "loss": 1.763,
+      "step": 19652500
+    },
+    {
+      "epoch": 97.37,
+      "learning_rate": 1.333521598470098e-06,
+      "loss": 1.771,
+      "step": 19653000
+    },
+    {
+      "epoch": 97.37,
+      "learning_rate": 1.3322830120440145e-06,
+      "loss": 1.771,
+      "step": 19653500
+    },
+    {
+      "epoch": 97.37,
+      "learning_rate": 1.3310444256179308e-06,
+      "loss": 1.7806,
+      "step": 19654000
+    },
+    {
+      "epoch": 97.38,
+      "learning_rate": 1.329805839191847e-06,
+      "loss": 1.7713,
+      "step": 19654500
+    },
+    {
+      "epoch": 97.38,
+      "learning_rate": 1.3285672527657636e-06,
+      "loss": 1.7857,
+      "step": 19655000
+    },
+    {
+      "epoch": 97.38,
+      "learning_rate": 1.32732866633968e-06,
+      "loss": 1.7563,
+      "step": 19655500
+    },
+    {
+      "epoch": 97.38,
+      "learning_rate": 1.3260925570864485e-06,
+      "loss": 1.7745,
+      "step": 19656000
+    },
+    {
+      "epoch": 97.39,
+      "learning_rate": 1.3248539706603648e-06,
+      "loss": 1.7793,
+      "step": 19656500
+    },
+    {
+      "epoch": 97.39,
+      "learning_rate": 1.323615384234281e-06,
+      "loss": 1.7389,
+      "step": 19657000
+    },
+    {
+      "epoch": 97.39,
+      "learning_rate": 1.3223792749810497e-06,
+      "loss": 1.7671,
+      "step": 19657500
+    },
+    {
+      "epoch": 97.39,
+      "learning_rate": 1.321140688554966e-06,
+      "loss": 1.768,
+      "step": 19658000
+    },
+    {
+      "epoch": 97.4,
+      "learning_rate": 1.3199045793017346e-06,
+      "loss": 1.7535,
+      "step": 19658500
+    },
+    {
+      "epoch": 97.4,
+      "learning_rate": 1.318665992875651e-06,
+      "loss": 1.7656,
+      "step": 19659000
+    },
+    {
+      "epoch": 97.4,
+      "learning_rate": 1.3174274064495674e-06,
+      "loss": 1.7822,
+      "step": 19659500
+    },
+    {
+      "epoch": 97.4,
+      "learning_rate": 1.316191297196336e-06,
+      "loss": 1.7471,
+      "step": 19660000
+    },
+    {
+      "epoch": 97.4,
+      "learning_rate": 1.3149527107702523e-06,
+      "loss": 1.7435,
+      "step": 19660500
+    },
+    {
+      "epoch": 97.41,
+      "learning_rate": 1.3137141243441686e-06,
+      "loss": 1.7685,
+      "step": 19661000
+    },
+    {
+      "epoch": 97.41,
+      "learning_rate": 1.3124755379180848e-06,
+      "loss": 1.7598,
+      "step": 19661500
+    },
+    {
+      "epoch": 97.41,
+      "learning_rate": 1.3112369514920011e-06,
+      "loss": 1.7538,
+      "step": 19662000
+    },
+    {
+      "epoch": 97.41,
+      "learning_rate": 1.3099983650659176e-06,
+      "loss": 1.7734,
+      "step": 19662500
+    },
+    {
+      "epoch": 97.42,
+      "learning_rate": 1.3087597786398341e-06,
+      "loss": 1.774,
+      "step": 19663000
+    },
+    {
+      "epoch": 97.42,
+      "learning_rate": 1.3075211922137504e-06,
+      "loss": 1.7541,
+      "step": 19663500
+    },
+    {
+      "epoch": 97.42,
+      "learning_rate": 1.3062826057876667e-06,
+      "loss": 1.7711,
+      "step": 19664000
+    },
+    {
+      "epoch": 97.42,
+      "learning_rate": 1.305044019361583e-06,
+      "loss": 1.7749,
+      "step": 19664500
+    },
+    {
+      "epoch": 97.43,
+      "learning_rate": 1.3038054329354993e-06,
+      "loss": 1.792,
+      "step": 19665000
+    },
+    {
+      "epoch": 97.43,
+      "learning_rate": 1.3025668465094158e-06,
+      "loss": 1.7684,
+      "step": 19665500
+    },
+    {
+      "epoch": 97.43,
+      "learning_rate": 1.3013282600833323e-06,
+      "loss": 1.7717,
+      "step": 19666000
+    },
+    {
+      "epoch": 97.43,
+      "learning_rate": 1.3000896736572486e-06,
+      "loss": 1.7568,
+      "step": 19666500
+    },
+    {
+      "epoch": 97.44,
+      "learning_rate": 1.298853564404017e-06,
+      "loss": 1.7658,
+      "step": 19667000
+    },
+    {
+      "epoch": 97.44,
+      "learning_rate": 1.2976174551507856e-06,
+      "loss": 1.7915,
+      "step": 19667500
+    },
+    {
+      "epoch": 97.44,
+      "learning_rate": 1.296378868724702e-06,
+      "loss": 1.7448,
+      "step": 19668000
+    },
+    {
+      "epoch": 97.44,
+      "learning_rate": 1.2951402822986182e-06,
+      "loss": 1.7846,
+      "step": 19668500
+    },
+    {
+      "epoch": 97.45,
+      "learning_rate": 1.2939016958725347e-06,
+      "loss": 1.7807,
+      "step": 19669000
+    },
+    {
+      "epoch": 97.45,
+      "learning_rate": 1.292663109446451e-06,
+      "loss": 1.753,
+      "step": 19669500
+    },
+    {
+      "epoch": 97.45,
+      "learning_rate": 1.2914245230203675e-06,
+      "loss": 1.7772,
+      "step": 19670000
+    },
+    {
+      "epoch": 97.45,
+      "learning_rate": 1.2901859365942838e-06,
+      "loss": 1.7613,
+      "step": 19670500
+    },
+    {
+      "epoch": 97.46,
+      "learning_rate": 1.2889473501682e-06,
+      "loss": 1.7823,
+      "step": 19671000
+    },
+    {
+      "epoch": 97.46,
+      "learning_rate": 1.2877087637421164e-06,
+      "loss": 1.7761,
+      "step": 19671500
+    },
+    {
+      "epoch": 97.46,
+      "learning_rate": 1.2864701773160329e-06,
+      "loss": 1.7496,
+      "step": 19672000
+    },
+    {
+      "epoch": 97.46,
+      "learning_rate": 1.2852315908899492e-06,
+      "loss": 1.7696,
+      "step": 19672500
+    },
+    {
+      "epoch": 97.47,
+      "learning_rate": 1.2839930044638655e-06,
+      "loss": 1.7918,
+      "step": 19673000
+    },
+    {
+      "epoch": 97.47,
+      "learning_rate": 1.282754418037782e-06,
+      "loss": 1.7457,
+      "step": 19673500
+    },
+    {
+      "epoch": 97.47,
+      "learning_rate": 1.2815158316116983e-06,
+      "loss": 1.7561,
+      "step": 19674000
+    },
+    {
+      "epoch": 97.47,
+      "learning_rate": 1.2802797223584669e-06,
+      "loss": 1.7666,
+      "step": 19674500
+    },
+    {
+      "epoch": 97.48,
+      "learning_rate": 1.2790411359323831e-06,
+      "loss": 1.7688,
+      "step": 19675000
+    },
+    {
+      "epoch": 97.48,
+      "learning_rate": 1.2778050266791517e-06,
+      "loss": 1.7644,
+      "step": 19675500
+    },
+    {
+      "epoch": 97.48,
+      "learning_rate": 1.276566440253068e-06,
+      "loss": 1.7311,
+      "step": 19676000
+    },
+    {
+      "epoch": 97.48,
+      "learning_rate": 1.2753278538269843e-06,
+      "loss": 1.7753,
+      "step": 19676500
+    },
+    {
+      "epoch": 97.49,
+      "learning_rate": 1.2740892674009008e-06,
+      "loss": 1.7833,
+      "step": 19677000
+    },
+    {
+      "epoch": 97.49,
+      "learning_rate": 1.2728506809748171e-06,
+      "loss": 1.7539,
+      "step": 19677500
+    },
+    {
+      "epoch": 97.49,
+      "learning_rate": 1.2716120945487334e-06,
+      "loss": 1.7617,
+      "step": 19678000
+    },
+    {
+      "epoch": 97.49,
+      "learning_rate": 1.27037350812265e-06,
+      "loss": 1.7662,
+      "step": 19678500
+    },
+    {
+      "epoch": 97.5,
+      "learning_rate": 1.2691349216965662e-06,
+      "loss": 1.7592,
+      "step": 19679000
+    },
+    {
+      "epoch": 97.5,
+      "learning_rate": 1.2678963352704825e-06,
+      "loss": 1.759,
+      "step": 19679500
+    },
+    {
+      "epoch": 97.5,
+      "learning_rate": 1.2666602260172511e-06,
+      "loss": 1.7664,
+      "step": 19680000
+    },
+    {
+      "epoch": 97.5,
+      "learning_rate": 1.2654241167640195e-06,
+      "loss": 1.7703,
+      "step": 19680500
+    },
+    {
+      "epoch": 97.51,
+      "learning_rate": 1.264188007510788e-06,
+      "loss": 1.7597,
+      "step": 19681000
+    },
+    {
+      "epoch": 97.51,
+      "learning_rate": 1.2629494210847046e-06,
+      "loss": 1.7917,
+      "step": 19681500
+    },
+    {
+      "epoch": 97.51,
+      "learning_rate": 1.2617108346586209e-06,
+      "loss": 1.754,
+      "step": 19682000
+    },
+    {
+      "epoch": 97.51,
+      "learning_rate": 1.2604722482325372e-06,
+      "loss": 1.7945,
+      "step": 19682500
+    },
+    {
+      "epoch": 97.52,
+      "learning_rate": 1.2592336618064537e-06,
+      "loss": 1.7634,
+      "step": 19683000
+    },
+    {
+      "epoch": 97.52,
+      "learning_rate": 1.25799507538037e-06,
+      "loss": 1.7538,
+      "step": 19683500
+    },
+    {
+      "epoch": 97.52,
+      "learning_rate": 1.2567564889542863e-06,
+      "loss": 1.7468,
+      "step": 19684000
+    },
+    {
+      "epoch": 97.52,
+      "learning_rate": 1.2555179025282026e-06,
+      "loss": 1.7567,
+      "step": 19684500
+    },
+    {
+      "epoch": 97.53,
+      "learning_rate": 1.254279316102119e-06,
+      "loss": 1.7433,
+      "step": 19685000
+    },
+    {
+      "epoch": 97.53,
+      "learning_rate": 1.2530407296760354e-06,
+      "loss": 1.7718,
+      "step": 19685500
+    },
+    {
+      "epoch": 97.53,
+      "learning_rate": 1.2518021432499519e-06,
+      "loss": 1.7742,
+      "step": 19686000
+    },
+    {
+      "epoch": 97.53,
+      "learning_rate": 1.2505635568238682e-06,
+      "loss": 1.7599,
+      "step": 19686500
+    },
+    {
+      "epoch": 97.54,
+      "learning_rate": 1.2493274475706365e-06,
+      "loss": 1.7744,
+      "step": 19687000
+    },
+    {
+      "epoch": 97.54,
+      "learning_rate": 1.2480913383174051e-06,
+      "loss": 1.7805,
+      "step": 19687500
+    },
+    {
+      "epoch": 97.54,
+      "learning_rate": 1.2468527518913214e-06,
+      "loss": 1.7652,
+      "step": 19688000
+    },
+    {
+      "epoch": 97.54,
+      "learning_rate": 1.245614165465238e-06,
+      "loss": 1.7571,
+      "step": 19688500
+    },
+    {
+      "epoch": 97.55,
+      "learning_rate": 1.2443755790391542e-06,
+      "loss": 1.7606,
+      "step": 19689000
+    },
+    {
+      "epoch": 97.55,
+      "learning_rate": 1.2431369926130707e-06,
+      "loss": 1.767,
+      "step": 19689500
+    },
+    {
+      "epoch": 97.55,
+      "learning_rate": 1.241898406186987e-06,
+      "loss": 1.8022,
+      "step": 19690000
+    },
+    {
+      "epoch": 97.55,
+      "learning_rate": 1.2406622969337554e-06,
+      "loss": 1.7593,
+      "step": 19690500
+    },
+    {
+      "epoch": 97.56,
+      "learning_rate": 1.239423710507672e-06,
+      "loss": 1.7545,
+      "step": 19691000
+    },
+    {
+      "epoch": 97.56,
+      "learning_rate": 1.2381851240815882e-06,
+      "loss": 1.7776,
+      "step": 19691500
+    },
+    {
+      "epoch": 97.56,
+      "learning_rate": 1.2369465376555045e-06,
+      "loss": 1.7478,
+      "step": 19692000
+    },
+    {
+      "epoch": 97.56,
+      "learning_rate": 1.2357104284022731e-06,
+      "loss": 1.7958,
+      "step": 19692500
+    },
+    {
+      "epoch": 97.57,
+      "learning_rate": 1.2344718419761896e-06,
+      "loss": 1.7311,
+      "step": 19693000
+    },
+    {
+      "epoch": 97.57,
+      "learning_rate": 1.233233255550106e-06,
+      "loss": 1.7466,
+      "step": 19693500
+    },
+    {
+      "epoch": 97.57,
+      "learning_rate": 1.2319946691240222e-06,
+      "loss": 1.7474,
+      "step": 19694000
+    },
+    {
+      "epoch": 97.57,
+      "learning_rate": 1.2307560826979385e-06,
+      "loss": 1.752,
+      "step": 19694500
+    },
+    {
+      "epoch": 97.58,
+      "learning_rate": 1.2295174962718548e-06,
+      "loss": 1.7535,
+      "step": 19695000
+    },
+    {
+      "epoch": 97.58,
+      "learning_rate": 1.2282789098457713e-06,
+      "loss": 1.7494,
+      "step": 19695500
+    },
+    {
+      "epoch": 97.58,
+      "learning_rate": 1.2270403234196878e-06,
+      "loss": 1.7415,
+      "step": 19696000
+    },
+    {
+      "epoch": 97.58,
+      "learning_rate": 1.225801736993604e-06,
+      "loss": 1.7925,
+      "step": 19696500
+    },
+    {
+      "epoch": 97.59,
+      "learning_rate": 1.2245631505675204e-06,
+      "loss": 1.7631,
+      "step": 19697000
+    },
+    {
+      "epoch": 97.59,
+      "learning_rate": 1.2233245641414367e-06,
+      "loss": 1.7752,
+      "step": 19697500
+    },
+    {
+      "epoch": 97.59,
+      "learning_rate": 1.2220884548882053e-06,
+      "loss": 1.7872,
+      "step": 19698000
+    },
+    {
+      "epoch": 97.59,
+      "learning_rate": 1.2208498684621216e-06,
+      "loss": 1.7462,
+      "step": 19698500
+    },
+    {
+      "epoch": 97.6,
+      "learning_rate": 1.2196112820360379e-06,
+      "loss": 1.7791,
+      "step": 19699000
+    },
+    {
+      "epoch": 97.6,
+      "learning_rate": 1.2183726956099544e-06,
+      "loss": 1.7638,
+      "step": 19699500
+    },
+    {
+      "epoch": 97.6,
+      "learning_rate": 1.2171341091838707e-06,
+      "loss": 1.746,
+      "step": 19700000
+    },
+    {
+      "epoch": 97.6,
+      "learning_rate": 1.2158955227577872e-06,
+      "loss": 1.762,
+      "step": 19700500
+    },
+    {
+      "epoch": 97.61,
+      "learning_rate": 1.2146594135045555e-06,
+      "loss": 1.7906,
+      "step": 19701000
+    },
+    {
+      "epoch": 97.61,
+      "learning_rate": 1.2134208270784718e-06,
+      "loss": 1.769,
+      "step": 19701500
+    },
+    {
+      "epoch": 97.61,
+      "learning_rate": 1.2121822406523883e-06,
+      "loss": 1.7781,
+      "step": 19702000
+    },
+    {
+      "epoch": 97.61,
+      "learning_rate": 1.2109436542263046e-06,
+      "loss": 1.7548,
+      "step": 19702500
+    },
+    {
+      "epoch": 97.62,
+      "learning_rate": 1.209707544973073e-06,
+      "loss": 1.7679,
+      "step": 19703000
+    },
+    {
+      "epoch": 97.62,
+      "learning_rate": 1.2084689585469895e-06,
+      "loss": 1.7793,
+      "step": 19703500
+    },
+    {
+      "epoch": 97.62,
+      "learning_rate": 1.207230372120906e-06,
+      "loss": 1.7655,
+      "step": 19704000
+    },
+    {
+      "epoch": 97.62,
+      "learning_rate": 1.2059917856948223e-06,
+      "loss": 1.7691,
+      "step": 19704500
+    },
+    {
+      "epoch": 97.63,
+      "learning_rate": 1.2047531992687386e-06,
+      "loss": 1.7806,
+      "step": 19705000
+    },
+    {
+      "epoch": 97.63,
+      "learning_rate": 1.203514612842655e-06,
+      "loss": 1.7869,
+      "step": 19705500
+    },
+    {
+      "epoch": 97.63,
+      "learning_rate": 1.2022760264165712e-06,
+      "loss": 1.7528,
+      "step": 19706000
+    },
+    {
+      "epoch": 97.63,
+      "learning_rate": 1.2010399171633398e-06,
+      "loss": 1.7799,
+      "step": 19706500
+    },
+    {
+      "epoch": 97.64,
+      "learning_rate": 1.1998013307372563e-06,
+      "loss": 1.7762,
+      "step": 19707000
+    },
+    {
+      "epoch": 97.64,
+      "learning_rate": 1.1985627443111726e-06,
+      "loss": 1.789,
+      "step": 19707500
+    },
+    {
+      "epoch": 97.64,
+      "learning_rate": 1.197324157885089e-06,
+      "loss": 1.7784,
+      "step": 19708000
+    },
+    {
+      "epoch": 97.64,
+      "learning_rate": 1.1960855714590054e-06,
+      "loss": 1.7741,
+      "step": 19708500
+    },
+    {
+      "epoch": 97.65,
+      "learning_rate": 1.1948494622057738e-06,
+      "loss": 1.7642,
+      "step": 19709000
+    },
+    {
+      "epoch": 97.65,
+      "learning_rate": 1.19361087577969e-06,
+      "loss": 1.7523,
+      "step": 19709500
+    },
+    {
+      "epoch": 97.65,
+      "learning_rate": 1.1923722893536066e-06,
+      "loss": 1.7572,
+      "step": 19710000
+    },
+    {
+      "epoch": 97.65,
+      "learning_rate": 1.1911337029275229e-06,
+      "loss": 1.7739,
+      "step": 19710500
+    },
+    {
+      "epoch": 97.66,
+      "learning_rate": 1.1898951165014394e-06,
+      "loss": 1.7775,
+      "step": 19711000
+    },
+    {
+      "epoch": 97.66,
+      "learning_rate": 1.1886565300753557e-06,
+      "loss": 1.747,
+      "step": 19711500
+    },
+    {
+      "epoch": 97.66,
+      "learning_rate": 1.187417943649272e-06,
+      "loss": 1.7653,
+      "step": 19712000
+    },
+    {
+      "epoch": 97.66,
+      "learning_rate": 1.1861793572231883e-06,
+      "loss": 1.7619,
+      "step": 19712500
+    },
+    {
+      "epoch": 97.67,
+      "learning_rate": 1.1849407707971048e-06,
+      "loss": 1.7837,
+      "step": 19713000
+    },
+    {
+      "epoch": 97.67,
+      "learning_rate": 1.1837046615438731e-06,
+      "loss": 1.7601,
+      "step": 19713500
+    },
+    {
+      "epoch": 97.67,
+      "learning_rate": 1.1824660751177896e-06,
+      "loss": 1.7663,
+      "step": 19714000
+    },
+    {
+      "epoch": 97.67,
+      "learning_rate": 1.1812274886917062e-06,
+      "loss": 1.7622,
+      "step": 19714500
+    },
+    {
+      "epoch": 97.67,
+      "learning_rate": 1.1799889022656224e-06,
+      "loss": 1.7498,
+      "step": 19715000
+    },
+    {
+      "epoch": 97.68,
+      "learning_rate": 1.1787503158395387e-06,
+      "loss": 1.7819,
+      "step": 19715500
+    },
+    {
+      "epoch": 97.68,
+      "learning_rate": 1.1775142065863073e-06,
+      "loss": 1.7446,
+      "step": 19716000
+    },
+    {
+      "epoch": 97.68,
+      "learning_rate": 1.1762756201602236e-06,
+      "loss": 1.7676,
+      "step": 19716500
+    },
+    {
+      "epoch": 97.68,
+      "learning_rate": 1.175039510906992e-06,
+      "loss": 1.7614,
+      "step": 19717000
+    },
+    {
+      "epoch": 97.69,
+      "learning_rate": 1.1738009244809085e-06,
+      "loss": 1.784,
+      "step": 19717500
+    },
+    {
+      "epoch": 97.69,
+      "learning_rate": 1.1725623380548248e-06,
+      "loss": 1.7791,
+      "step": 19718000
+    },
+    {
+      "epoch": 97.69,
+      "learning_rate": 1.1713237516287413e-06,
+      "loss": 1.7531,
+      "step": 19718500
+    },
+    {
+      "epoch": 97.69,
+      "learning_rate": 1.1700851652026576e-06,
+      "loss": 1.7695,
+      "step": 19719000
+    },
+    {
+      "epoch": 97.7,
+      "learning_rate": 1.1688490559494262e-06,
+      "loss": 1.7614,
+      "step": 19719500
+    },
+    {
+      "epoch": 97.7,
+      "learning_rate": 1.1676129466961946e-06,
+      "loss": 1.7822,
+      "step": 19720000
+    },
+    {
+      "epoch": 97.7,
+      "learning_rate": 1.1663743602701109e-06,
+      "loss": 1.7676,
+      "step": 19720500
+    },
+    {
+      "epoch": 97.7,
+      "learning_rate": 1.1651357738440274e-06,
+      "loss": 1.7657,
+      "step": 19721000
+    },
+    {
+      "epoch": 97.71,
+      "learning_rate": 1.1638971874179437e-06,
+      "loss": 1.7669,
+      "step": 19721500
+    },
+    {
+      "epoch": 97.71,
+      "learning_rate": 1.16265860099186e-06,
+      "loss": 1.7787,
+      "step": 19722000
+    },
+    {
+      "epoch": 97.71,
+      "learning_rate": 1.1614200145657765e-06,
+      "loss": 1.7682,
+      "step": 19722500
+    },
+    {
+      "epoch": 97.71,
+      "learning_rate": 1.160183905312545e-06,
+      "loss": 1.7712,
+      "step": 19723000
+    },
+    {
+      "epoch": 97.72,
+      "learning_rate": 1.1589453188864614e-06,
+      "loss": 1.7657,
+      "step": 19723500
+    },
+    {
+      "epoch": 97.72,
+      "learning_rate": 1.15770920963323e-06,
+      "loss": 1.7565,
+      "step": 19724000
+    },
+    {
+      "epoch": 97.72,
+      "learning_rate": 1.1564706232071463e-06,
+      "loss": 1.7937,
+      "step": 19724500
+    },
+    {
+      "epoch": 97.72,
+      "learning_rate": 1.1552320367810626e-06,
+      "loss": 1.7822,
+      "step": 19725000
+    },
+    {
+      "epoch": 97.73,
+      "learning_rate": 1.1539934503549788e-06,
+      "loss": 1.7577,
+      "step": 19725500
+    },
+    {
+      "epoch": 97.73,
+      "learning_rate": 1.1527548639288954e-06,
+      "loss": 1.7604,
+      "step": 19726000
+    },
+    {
+      "epoch": 97.73,
+      "learning_rate": 1.1515162775028116e-06,
+      "loss": 1.7542,
+      "step": 19726500
+    },
+    {
+      "epoch": 97.73,
+      "learning_rate": 1.1502776910767282e-06,
+      "loss": 1.7883,
+      "step": 19727000
+    },
+    {
+      "epoch": 97.74,
+      "learning_rate": 1.1490391046506444e-06,
+      "loss": 1.7631,
+      "step": 19727500
+    },
+    {
+      "epoch": 97.74,
+      "learning_rate": 1.1478005182245607e-06,
+      "loss": 1.7772,
+      "step": 19728000
+    },
+    {
+      "epoch": 97.74,
+      "learning_rate": 1.146561931798477e-06,
+      "loss": 1.7695,
+      "step": 19728500
+    },
+    {
+      "epoch": 97.74,
+      "learning_rate": 1.1453233453723933e-06,
+      "loss": 1.7594,
+      "step": 19729000
+    },
+    {
+      "epoch": 97.75,
+      "learning_rate": 1.1440847589463098e-06,
+      "loss": 1.7847,
+      "step": 19729500
+    },
+    {
+      "epoch": 97.75,
+      "learning_rate": 1.1428461725202263e-06,
+      "loss": 1.7682,
+      "step": 19730000
+    },
+    {
+      "epoch": 97.75,
+      "learning_rate": 1.1416075860941426e-06,
+      "loss": 1.7534,
+      "step": 19730500
+    },
+    {
+      "epoch": 97.75,
+      "learning_rate": 1.140371476840911e-06,
+      "loss": 1.755,
+      "step": 19731000
+    },
+    {
+      "epoch": 97.76,
+      "learning_rate": 1.1391328904148275e-06,
+      "loss": 1.7571,
+      "step": 19731500
+    },
+    {
+      "epoch": 97.76,
+      "learning_rate": 1.1378943039887438e-06,
+      "loss": 1.7834,
+      "step": 19732000
+    },
+    {
+      "epoch": 97.76,
+      "learning_rate": 1.1366581947355122e-06,
+      "loss": 1.7811,
+      "step": 19732500
+    },
+    {
+      "epoch": 97.76,
+      "learning_rate": 1.1354196083094285e-06,
+      "loss": 1.7719,
+      "step": 19733000
+    },
+    {
+      "epoch": 97.77,
+      "learning_rate": 1.134181021883345e-06,
+      "loss": 1.7946,
+      "step": 19733500
+    },
+    {
+      "epoch": 97.77,
+      "learning_rate": 1.1329424354572615e-06,
+      "loss": 1.7884,
+      "step": 19734000
+    },
+    {
+      "epoch": 97.77,
+      "learning_rate": 1.1317038490311778e-06,
+      "loss": 1.7604,
+      "step": 19734500
+    },
+    {
+      "epoch": 97.77,
+      "learning_rate": 1.130465262605094e-06,
+      "loss": 1.7796,
+      "step": 19735000
+    },
+    {
+      "epoch": 97.78,
+      "learning_rate": 1.1292266761790104e-06,
+      "loss": 1.7702,
+      "step": 19735500
+    },
+    {
+      "epoch": 97.78,
+      "learning_rate": 1.127990566925779e-06,
+      "loss": 1.7932,
+      "step": 19736000
+    },
+    {
+      "epoch": 97.78,
+      "learning_rate": 1.1267519804996953e-06,
+      "loss": 1.7591,
+      "step": 19736500
+    },
+    {
+      "epoch": 97.78,
+      "learning_rate": 1.1255133940736118e-06,
+      "loss": 1.7742,
+      "step": 19737000
+    },
+    {
+      "epoch": 97.79,
+      "learning_rate": 1.124274807647528e-06,
+      "loss": 1.7724,
+      "step": 19737500
+    },
+    {
+      "epoch": 97.79,
+      "learning_rate": 1.1230362212214446e-06,
+      "loss": 1.7509,
+      "step": 19738000
+    },
+    {
+      "epoch": 97.79,
+      "learning_rate": 1.1217976347953609e-06,
+      "loss": 1.7613,
+      "step": 19738500
+    },
+    {
+      "epoch": 97.79,
+      "learning_rate": 1.1205590483692772e-06,
+      "loss": 1.7828,
+      "step": 19739000
+    },
+    {
+      "epoch": 97.8,
+      "learning_rate": 1.1193204619431934e-06,
+      "loss": 1.7732,
+      "step": 19739500
+    },
+    {
+      "epoch": 97.8,
+      "learning_rate": 1.1180818755171097e-06,
+      "loss": 1.7737,
+      "step": 19740000
+    },
+    {
+      "epoch": 97.8,
+      "learning_rate": 1.1168432890910262e-06,
+      "loss": 1.786,
+      "step": 19740500
+    },
+    {
+      "epoch": 97.8,
+      "learning_rate": 1.1156071798377948e-06,
+      "loss": 1.7374,
+      "step": 19741000
+    },
+    {
+      "epoch": 97.81,
+      "learning_rate": 1.1143685934117111e-06,
+      "loss": 1.788,
+      "step": 19741500
+    },
+    {
+      "epoch": 97.81,
+      "learning_rate": 1.1131324841584797e-06,
+      "loss": 1.7598,
+      "step": 19742000
+    },
+    {
+      "epoch": 97.81,
+      "learning_rate": 1.111893897732396e-06,
+      "loss": 1.7746,
+      "step": 19742500
+    },
+    {
+      "epoch": 97.81,
+      "learning_rate": 1.1106553113063123e-06,
+      "loss": 1.7589,
+      "step": 19743000
+    },
+    {
+      "epoch": 97.82,
+      "learning_rate": 1.1094167248802286e-06,
+      "loss": 1.7645,
+      "step": 19743500
+    },
+    {
+      "epoch": 97.82,
+      "learning_rate": 1.1081781384541451e-06,
+      "loss": 1.7569,
+      "step": 19744000
+    },
+    {
+      "epoch": 97.82,
+      "learning_rate": 1.1069395520280616e-06,
+      "loss": 1.7848,
+      "step": 19744500
+    },
+    {
+      "epoch": 97.82,
+      "learning_rate": 1.105700965601978e-06,
+      "loss": 1.7475,
+      "step": 19745000
+    },
+    {
+      "epoch": 97.83,
+      "learning_rate": 1.1044623791758942e-06,
+      "loss": 1.7384,
+      "step": 19745500
+    },
+    {
+      "epoch": 97.83,
+      "learning_rate": 1.1032237927498105e-06,
+      "loss": 1.767,
+      "step": 19746000
+    },
+    {
+      "epoch": 97.83,
+      "learning_rate": 1.1019852063237268e-06,
+      "loss": 1.7835,
+      "step": 19746500
+    },
+    {
+      "epoch": 97.83,
+      "learning_rate": 1.1007515742433475e-06,
+      "loss": 1.7716,
+      "step": 19747000
+    },
+    {
+      "epoch": 97.84,
+      "learning_rate": 1.099512987817264e-06,
+      "loss": 1.7766,
+      "step": 19747500
+    },
+    {
+      "epoch": 97.84,
+      "learning_rate": 1.0982744013911803e-06,
+      "loss": 1.761,
+      "step": 19748000
+    },
+    {
+      "epoch": 97.84,
+      "learning_rate": 1.0970358149650968e-06,
+      "loss": 1.7613,
+      "step": 19748500
+    },
+    {
+      "epoch": 97.84,
+      "learning_rate": 1.095797228539013e-06,
+      "loss": 1.7795,
+      "step": 19749000
+    },
+    {
+      "epoch": 97.85,
+      "learning_rate": 1.0945586421129294e-06,
+      "loss": 1.7898,
+      "step": 19749500
+    },
+    {
+      "epoch": 97.85,
+      "learning_rate": 1.0933200556868457e-06,
+      "loss": 1.7451,
+      "step": 19750000
+    },
+    {
+      "epoch": 97.85,
+      "learning_rate": 1.0920814692607622e-06,
+      "loss": 1.773,
+      "step": 19750500
+    },
+    {
+      "epoch": 97.85,
+      "learning_rate": 1.0908428828346785e-06,
+      "loss": 1.7571,
+      "step": 19751000
+    },
+    {
+      "epoch": 97.86,
+      "learning_rate": 1.0896067735814468e-06,
+      "loss": 1.7707,
+      "step": 19751500
+    },
+    {
+      "epoch": 97.86,
+      "learning_rate": 1.0883681871553634e-06,
+      "loss": 1.7527,
+      "step": 19752000
+    },
+    {
+      "epoch": 97.86,
+      "learning_rate": 1.0871296007292799e-06,
+      "loss": 1.7704,
+      "step": 19752500
+    },
+    {
+      "epoch": 97.86,
+      "learning_rate": 1.0858934914760482e-06,
+      "loss": 1.763,
+      "step": 19753000
+    },
+    {
+      "epoch": 97.87,
+      "learning_rate": 1.0846549050499647e-06,
+      "loss": 1.7554,
+      "step": 19753500
+    },
+    {
+      "epoch": 97.87,
+      "learning_rate": 1.083416318623881e-06,
+      "loss": 1.7598,
+      "step": 19754000
+    },
+    {
+      "epoch": 97.87,
+      "learning_rate": 1.0821777321977973e-06,
+      "loss": 1.7554,
+      "step": 19754500
+    },
+    {
+      "epoch": 97.87,
+      "learning_rate": 1.0809391457717136e-06,
+      "loss": 1.7897,
+      "step": 19755000
+    },
+    {
+      "epoch": 97.88,
+      "learning_rate": 1.0797005593456301e-06,
+      "loss": 1.7763,
+      "step": 19755500
+    },
+    {
+      "epoch": 97.88,
+      "learning_rate": 1.0784619729195464e-06,
+      "loss": 1.7729,
+      "step": 19756000
+    },
+    {
+      "epoch": 97.88,
+      "learning_rate": 1.077223386493463e-06,
+      "loss": 1.7708,
+      "step": 19756500
+    },
+    {
+      "epoch": 97.88,
+      "learning_rate": 1.0759848000673792e-06,
+      "loss": 1.7835,
+      "step": 19757000
+    },
+    {
+      "epoch": 97.89,
+      "learning_rate": 1.0747486908141476e-06,
+      "loss": 1.7729,
+      "step": 19757500
+    },
+    {
+      "epoch": 97.89,
+      "learning_rate": 1.073510104388064e-06,
+      "loss": 1.7566,
+      "step": 19758000
+    },
+    {
+      "epoch": 97.89,
+      "learning_rate": 1.0722715179619804e-06,
+      "loss": 1.7864,
+      "step": 19758500
+    },
+    {
+      "epoch": 97.89,
+      "learning_rate": 1.0710329315358967e-06,
+      "loss": 1.7725,
+      "step": 19759000
+    },
+    {
+      "epoch": 97.9,
+      "learning_rate": 1.0697943451098132e-06,
+      "loss": 1.7539,
+      "step": 19759500
+    },
+    {
+      "epoch": 97.9,
+      "learning_rate": 1.0685557586837295e-06,
+      "loss": 1.7797,
+      "step": 19760000
+    },
+    {
+      "epoch": 97.9,
+      "learning_rate": 1.0673171722576458e-06,
+      "loss": 1.7668,
+      "step": 19760500
+    },
+    {
+      "epoch": 97.9,
+      "learning_rate": 1.0660810630044144e-06,
+      "loss": 1.7602,
+      "step": 19761000
+    },
+    {
+      "epoch": 97.91,
+      "learning_rate": 1.064844953751183e-06,
+      "loss": 1.7635,
+      "step": 19761500
+    },
+    {
+      "epoch": 97.91,
+      "learning_rate": 1.0636063673250993e-06,
+      "loss": 1.7799,
+      "step": 19762000
+    },
+    {
+      "epoch": 97.91,
+      "learning_rate": 1.0623677808990156e-06,
+      "loss": 1.7903,
+      "step": 19762500
+    },
+    {
+      "epoch": 97.91,
+      "learning_rate": 1.061129194472932e-06,
+      "loss": 1.7749,
+      "step": 19763000
+    },
+    {
+      "epoch": 97.92,
+      "learning_rate": 1.0598906080468484e-06,
+      "loss": 1.7641,
+      "step": 19763500
+    },
+    {
+      "epoch": 97.92,
+      "learning_rate": 1.0586520216207647e-06,
+      "loss": 1.761,
+      "step": 19764000
+    },
+    {
+      "epoch": 97.92,
+      "learning_rate": 1.0574134351946812e-06,
+      "loss": 1.7756,
+      "step": 19764500
+    },
+    {
+      "epoch": 97.92,
+      "learning_rate": 1.0561748487685975e-06,
+      "loss": 1.7639,
+      "step": 19765000
+    },
+    {
+      "epoch": 97.93,
+      "learning_rate": 1.0549362623425137e-06,
+      "loss": 1.7746,
+      "step": 19765500
+    },
+    {
+      "epoch": 97.93,
+      "learning_rate": 1.05369767591643e-06,
+      "loss": 1.7728,
+      "step": 19766000
+    },
+    {
+      "epoch": 97.93,
+      "learning_rate": 1.0524615666631986e-06,
+      "loss": 1.7662,
+      "step": 19766500
+    },
+    {
+      "epoch": 97.93,
+      "learning_rate": 1.0512229802371151e-06,
+      "loss": 1.7825,
+      "step": 19767000
+    },
+    {
+      "epoch": 97.94,
+      "learning_rate": 1.0499843938110314e-06,
+      "loss": 1.7757,
+      "step": 19767500
+    },
+    {
+      "epoch": 97.94,
+      "learning_rate": 1.0487458073849477e-06,
+      "loss": 1.7639,
+      "step": 19768000
+    },
+    {
+      "epoch": 97.94,
+      "learning_rate": 1.047507220958864e-06,
+      "loss": 1.748,
+      "step": 19768500
+    },
+    {
+      "epoch": 97.94,
+      "learning_rate": 1.0462686345327805e-06,
+      "loss": 1.7769,
+      "step": 19769000
+    },
+    {
+      "epoch": 97.94,
+      "learning_rate": 1.0450300481066968e-06,
+      "loss": 1.7762,
+      "step": 19769500
+    },
+    {
+      "epoch": 97.95,
+      "learning_rate": 1.0437914616806133e-06,
+      "loss": 1.7631,
+      "step": 19770000
+    },
+    {
+      "epoch": 97.95,
+      "learning_rate": 1.0425528752545296e-06,
+      "loss": 1.7642,
+      "step": 19770500
+    },
+    {
+      "epoch": 97.95,
+      "learning_rate": 1.041314288828446e-06,
+      "loss": 1.7826,
+      "step": 19771000
+    },
+    {
+      "epoch": 97.95,
+      "learning_rate": 1.0400781795752145e-06,
+      "loss": 1.7514,
+      "step": 19771500
+    },
+    {
+      "epoch": 97.96,
+      "learning_rate": 1.0388395931491308e-06,
+      "loss": 1.8017,
+      "step": 19772000
+    },
+    {
+      "epoch": 97.96,
+      "learning_rate": 1.0376034838958994e-06,
+      "loss": 1.7718,
+      "step": 19772500
+    },
+    {
+      "epoch": 97.96,
+      "learning_rate": 1.0363648974698157e-06,
+      "loss": 1.7455,
+      "step": 19773000
+    },
+    {
+      "epoch": 97.96,
+      "learning_rate": 1.035126311043732e-06,
+      "loss": 1.7603,
+      "step": 19773500
+    },
+    {
+      "epoch": 97.97,
+      "learning_rate": 1.0338877246176485e-06,
+      "loss": 1.7651,
+      "step": 19774000
+    },
+    {
+      "epoch": 97.97,
+      "learning_rate": 1.0326491381915648e-06,
+      "loss": 1.7796,
+      "step": 19774500
+    },
+    {
+      "epoch": 97.97,
+      "learning_rate": 1.031410551765481e-06,
+      "loss": 1.7474,
+      "step": 19775000
+    },
+    {
+      "epoch": 97.97,
+      "learning_rate": 1.0301719653393976e-06,
+      "loss": 1.7553,
+      "step": 19775500
+    },
+    {
+      "epoch": 97.98,
+      "learning_rate": 1.0289333789133139e-06,
+      "loss": 1.769,
+      "step": 19776000
+    },
+    {
+      "epoch": 97.98,
+      "learning_rate": 1.0276947924872302e-06,
+      "loss": 1.7556,
+      "step": 19776500
+    },
+    {
+      "epoch": 97.98,
+      "learning_rate": 1.0264562060611467e-06,
+      "loss": 1.7668,
+      "step": 19777000
+    },
+    {
+      "epoch": 97.98,
+      "learning_rate": 1.025217619635063e-06,
+      "loss": 1.7572,
+      "step": 19777500
+    },
+    {
+      "epoch": 97.99,
+      "learning_rate": 1.0239815103818316e-06,
+      "loss": 1.7613,
+      "step": 19778000
+    },
+    {
+      "epoch": 97.99,
+      "learning_rate": 1.0227429239557479e-06,
+      "loss": 1.7745,
+      "step": 19778500
+    },
+    {
+      "epoch": 97.99,
+      "learning_rate": 1.0215043375296641e-06,
+      "loss": 1.7424,
+      "step": 19779000
+    },
+    {
+      "epoch": 97.99,
+      "learning_rate": 1.0202682282764327e-06,
+      "loss": 1.782,
+      "step": 19779500
+    },
+    {
+      "epoch": 98.0,
+      "learning_rate": 1.019029641850349e-06,
+      "loss": 1.7797,
+      "step": 19780000
+    },
+    {
+      "epoch": 98.0,
+      "learning_rate": 1.0177910554242653e-06,
+      "loss": 1.7705,
+      "step": 19780500
+    },
+    {
+      "epoch": 98.0,
+      "eval_accuracy": 0.6875432970204784,
+      "eval_accuracy_mlm": 0.6491573384378511,
+      "eval_accuracy_nsp": 0.8684455147690413,
+      "eval_loss": 2.312778949737549,
+      "eval_runtime": 147.1591,
+      "eval_samples_per_second": 1732.539,
+      "eval_steps_per_second": 72.194,
+      "step": 19780614
+    },
+    {
+      "epoch": 98.0,
+      "learning_rate": 1.0165524689981818e-06,
+      "loss": 1.7802,
+      "step": 19781000
+    },
+    {
+      "epoch": 98.0,
+      "learning_rate": 1.0153138825720983e-06,
+      "loss": 1.7516,
+      "step": 19781500
+    },
+    {
+      "epoch": 98.01,
+      "learning_rate": 1.0140752961460146e-06,
+      "loss": 1.7575,
+      "step": 19782000
+    },
+    {
+      "epoch": 98.01,
+      "learning_rate": 1.012836709719931e-06,
+      "loss": 1.7681,
+      "step": 19782500
+    },
+    {
+      "epoch": 98.01,
+      "learning_rate": 1.0116006004666993e-06,
+      "loss": 1.773,
+      "step": 19783000
+    },
+    {
+      "epoch": 98.01,
+      "learning_rate": 1.010364491213468e-06,
+      "loss": 1.7569,
+      "step": 19783500
+    },
+    {
+      "epoch": 98.02,
+      "learning_rate": 1.0091259047873842e-06,
+      "loss": 1.7638,
+      "step": 19784000
+    },
+    {
+      "epoch": 98.02,
+      "learning_rate": 1.0078873183613005e-06,
+      "loss": 1.7566,
+      "step": 19784500
+    },
+    {
+      "epoch": 98.02,
+      "learning_rate": 1.006648731935217e-06,
+      "loss": 1.7487,
+      "step": 19785000
+    },
+    {
+      "epoch": 98.02,
+      "learning_rate": 1.0054101455091335e-06,
+      "loss": 1.7702,
+      "step": 19785500
+    },
+    {
+      "epoch": 98.03,
+      "learning_rate": 1.0041715590830498e-06,
+      "loss": 1.7679,
+      "step": 19786000
+    },
+    {
+      "epoch": 98.03,
+      "learning_rate": 1.002932972656966e-06,
+      "loss": 1.7697,
+      "step": 19786500
+    },
+    {
+      "epoch": 98.03,
+      "learning_rate": 1.0016943862308824e-06,
+      "loss": 1.7545,
+      "step": 19787000
+    },
+    {
+      "epoch": 98.03,
+      "learning_rate": 1.0004557998047987e-06,
+      "loss": 1.7625,
+      "step": 19787500
+    },
+    {
+      "epoch": 98.04,
+      "learning_rate": 9.992196905515673e-07,
+      "loss": 1.7565,
+      "step": 19788000
+    },
+    {
+      "epoch": 98.04,
+      "learning_rate": 9.979811041254838e-07,
+      "loss": 1.7763,
+      "step": 19788500
+    },
+    {
+      "epoch": 98.04,
+      "learning_rate": 9.967425176994e-07,
+      "loss": 1.771,
+      "step": 19789000
+    },
+    {
+      "epoch": 98.04,
+      "learning_rate": 9.955039312733166e-07,
+      "loss": 1.7735,
+      "step": 19789500
+    },
+    {
+      "epoch": 98.05,
+      "learning_rate": 9.94267822020085e-07,
+      "loss": 1.7723,
+      "step": 19790000
+    },
+    {
+      "epoch": 98.05,
+      "learning_rate": 9.930292355940013e-07,
+      "loss": 1.7826,
+      "step": 19790500
+    },
+    {
+      "epoch": 98.05,
+      "learning_rate": 9.917906491679178e-07,
+      "loss": 1.7642,
+      "step": 19791000
+    },
+    {
+      "epoch": 98.05,
+      "learning_rate": 9.90552062741834e-07,
+      "loss": 1.765,
+      "step": 19791500
+    },
+    {
+      "epoch": 98.06,
+      "learning_rate": 9.893134763157503e-07,
+      "loss": 1.7722,
+      "step": 19792000
+    },
+    {
+      "epoch": 98.06,
+      "learning_rate": 9.880748898896668e-07,
+      "loss": 1.77,
+      "step": 19792500
+    },
+    {
+      "epoch": 98.06,
+      "learning_rate": 9.868363034635831e-07,
+      "loss": 1.7674,
+      "step": 19793000
+    },
+    {
+      "epoch": 98.06,
+      "learning_rate": 9.855977170374994e-07,
+      "loss": 1.757,
+      "step": 19793500
+    },
+    {
+      "epoch": 98.07,
+      "learning_rate": 9.84359130611416e-07,
+      "loss": 1.788,
+      "step": 19794000
+    },
+    {
+      "epoch": 98.07,
+      "learning_rate": 9.831205441853322e-07,
+      "loss": 1.7798,
+      "step": 19794500
+    },
+    {
+      "epoch": 98.07,
+      "learning_rate": 9.818819577592485e-07,
+      "loss": 1.7642,
+      "step": 19795000
+    },
+    {
+      "epoch": 98.07,
+      "learning_rate": 9.80643371333165e-07,
+      "loss": 1.7519,
+      "step": 19795500
+    },
+    {
+      "epoch": 98.08,
+      "learning_rate": 9.794047849070813e-07,
+      "loss": 1.7755,
+      "step": 19796000
+    },
+    {
+      "epoch": 98.08,
+      "learning_rate": 9.7816867565385e-07,
+      "loss": 1.7439,
+      "step": 19796500
+    },
+    {
+      "epoch": 98.08,
+      "learning_rate": 9.769300892277662e-07,
+      "loss": 1.7529,
+      "step": 19797000
+    },
+    {
+      "epoch": 98.08,
+      "learning_rate": 9.756939799745348e-07,
+      "loss": 1.7851,
+      "step": 19797500
+    },
+    {
+      "epoch": 98.09,
+      "learning_rate": 9.74455393548451e-07,
+      "loss": 1.7637,
+      "step": 19798000
+    },
+    {
+      "epoch": 98.09,
+      "learning_rate": 9.732168071223674e-07,
+      "loss": 1.7256,
+      "step": 19798500
+    },
+    {
+      "epoch": 98.09,
+      "learning_rate": 9.719782206962837e-07,
+      "loss": 1.739,
+      "step": 19799000
+    },
+    {
+      "epoch": 98.09,
+      "learning_rate": 9.707396342702002e-07,
+      "loss": 1.7437,
+      "step": 19799500
+    },
+    {
+      "epoch": 98.1,
+      "learning_rate": 9.695010478441165e-07,
+      "loss": 1.761,
+      "step": 19800000
+    },
+    {
+      "epoch": 98.1,
+      "learning_rate": 9.68262461418033e-07,
+      "loss": 1.7472,
+      "step": 19800500
+    },
+    {
+      "epoch": 98.1,
+      "learning_rate": 9.670238749919493e-07,
+      "loss": 1.7556,
+      "step": 19801000
+    },
+    {
+      "epoch": 98.1,
+      "learning_rate": 9.657877657387177e-07,
+      "loss": 1.7687,
+      "step": 19801500
+    },
+    {
+      "epoch": 98.11,
+      "learning_rate": 9.645491793126342e-07,
+      "loss": 1.7789,
+      "step": 19802000
+    },
+    {
+      "epoch": 98.11,
+      "learning_rate": 9.633105928865505e-07,
+      "loss": 1.7704,
+      "step": 19802500
+    },
+    {
+      "epoch": 98.11,
+      "learning_rate": 9.620720064604668e-07,
+      "loss": 1.7474,
+      "step": 19803000
+    },
+    {
+      "epoch": 98.11,
+      "learning_rate": 9.608334200343833e-07,
+      "loss": 1.7386,
+      "step": 19803500
+    },
+    {
+      "epoch": 98.12,
+      "learning_rate": 9.595948336082996e-07,
+      "loss": 1.7541,
+      "step": 19804000
+    },
+    {
+      "epoch": 98.12,
+      "learning_rate": 9.583562471822159e-07,
+      "loss": 1.7396,
+      "step": 19804500
+    },
+    {
+      "epoch": 98.12,
+      "learning_rate": 9.571201379289844e-07,
+      "loss": 1.7542,
+      "step": 19805000
+    },
+    {
+      "epoch": 98.12,
+      "learning_rate": 9.558815515029007e-07,
+      "loss": 1.7921,
+      "step": 19805500
+    },
+    {
+      "epoch": 98.13,
+      "learning_rate": 9.54642965076817e-07,
+      "loss": 1.7591,
+      "step": 19806000
+    },
+    {
+      "epoch": 98.13,
+      "learning_rate": 9.534043786507336e-07,
+      "loss": 1.7668,
+      "step": 19806500
+    },
+    {
+      "epoch": 98.13,
+      "learning_rate": 9.521657922246499e-07,
+      "loss": 1.7636,
+      "step": 19807000
+    },
+    {
+      "epoch": 98.13,
+      "learning_rate": 9.509272057985662e-07,
+      "loss": 1.7654,
+      "step": 19807500
+    },
+    {
+      "epoch": 98.14,
+      "learning_rate": 9.496886193724826e-07,
+      "loss": 1.7505,
+      "step": 19808000
+    },
+    {
+      "epoch": 98.14,
+      "learning_rate": 9.484500329463989e-07,
+      "loss": 1.7647,
+      "step": 19808500
+    },
+    {
+      "epoch": 98.14,
+      "learning_rate": 9.472139236931675e-07,
+      "loss": 1.7379,
+      "step": 19809000
+    },
+    {
+      "epoch": 98.14,
+      "learning_rate": 9.459753372670839e-07,
+      "loss": 1.7504,
+      "step": 19809500
+    },
+    {
+      "epoch": 98.15,
+      "learning_rate": 9.447367508410002e-07,
+      "loss": 1.7559,
+      "step": 19810000
+    },
+    {
+      "epoch": 98.15,
+      "learning_rate": 9.435006415877688e-07,
+      "loss": 1.7443,
+      "step": 19810500
+    },
+    {
+      "epoch": 98.15,
+      "learning_rate": 9.422620551616851e-07,
+      "loss": 1.7548,
+      "step": 19811000
+    },
+    {
+      "epoch": 98.15,
+      "learning_rate": 9.410234687356015e-07,
+      "loss": 1.7367,
+      "step": 19811500
+    },
+    {
+      "epoch": 98.16,
+      "learning_rate": 9.397848823095178e-07,
+      "loss": 1.7747,
+      "step": 19812000
+    },
+    {
+      "epoch": 98.16,
+      "learning_rate": 9.385462958834341e-07,
+      "loss": 1.7316,
+      "step": 19812500
+    },
+    {
+      "epoch": 98.16,
+      "learning_rate": 9.373077094573506e-07,
+      "loss": 1.7559,
+      "step": 19813000
+    },
+    {
+      "epoch": 98.16,
+      "learning_rate": 9.36069123031267e-07,
+      "loss": 1.7578,
+      "step": 19813500
+    },
+    {
+      "epoch": 98.17,
+      "learning_rate": 9.348305366051833e-07,
+      "loss": 1.7442,
+      "step": 19814000
+    },
+    {
+      "epoch": 98.17,
+      "learning_rate": 9.335919501790996e-07,
+      "loss": 1.7707,
+      "step": 19814500
+    },
+    {
+      "epoch": 98.17,
+      "learning_rate": 9.32353363753016e-07,
+      "loss": 1.758,
+      "step": 19815000
+    },
+    {
+      "epoch": 98.17,
+      "learning_rate": 9.311147773269323e-07,
+      "loss": 1.7589,
+      "step": 19815500
+    },
+    {
+      "epoch": 98.18,
+      "learning_rate": 9.298761909008488e-07,
+      "loss": 1.7761,
+      "step": 19816000
+    },
+    {
+      "epoch": 98.18,
+      "learning_rate": 9.286400816476172e-07,
+      "loss": 1.7828,
+      "step": 19816500
+    },
+    {
+      "epoch": 98.18,
+      "learning_rate": 9.274014952215336e-07,
+      "loss": 1.7531,
+      "step": 19817000
+    },
+    {
+      "epoch": 98.18,
+      "learning_rate": 9.261629087954501e-07,
+      "loss": 1.7362,
+      "step": 19817500
+    },
+    {
+      "epoch": 98.19,
+      "learning_rate": 9.249243223693664e-07,
+      "loss": 1.7804,
+      "step": 19818000
+    },
+    {
+      "epoch": 98.19,
+      "learning_rate": 9.236857359432828e-07,
+      "loss": 1.7474,
+      "step": 19818500
+    },
+    {
+      "epoch": 98.19,
+      "learning_rate": 9.22447149517199e-07,
+      "loss": 1.7407,
+      "step": 19819000
+    },
+    {
+      "epoch": 98.19,
+      "learning_rate": 9.212085630911153e-07,
+      "loss": 1.7506,
+      "step": 19819500
+    },
+    {
+      "epoch": 98.2,
+      "learning_rate": 9.199699766650317e-07,
+      "loss": 1.7572,
+      "step": 19820000
+    },
+    {
+      "epoch": 98.2,
+      "learning_rate": 9.187313902389482e-07,
+      "loss": 1.7541,
+      "step": 19820500
+    },
+    {
+      "epoch": 98.2,
+      "learning_rate": 9.174928038128645e-07,
+      "loss": 1.7984,
+      "step": 19821000
+    },
+    {
+      "epoch": 98.2,
+      "learning_rate": 9.162542173867808e-07,
+      "loss": 1.7555,
+      "step": 19821500
+    },
+    {
+      "epoch": 98.21,
+      "learning_rate": 9.150156309606972e-07,
+      "loss": 1.7673,
+      "step": 19822000
+    },
+    {
+      "epoch": 98.21,
+      "learning_rate": 9.137795217074658e-07,
+      "loss": 1.7601,
+      "step": 19822500
+    },
+    {
+      "epoch": 98.21,
+      "learning_rate": 9.125434124542342e-07,
+      "loss": 1.754,
+      "step": 19823000
+    },
+    {
+      "epoch": 98.21,
+      "learning_rate": 9.113048260281507e-07,
+      "loss": 1.7612,
+      "step": 19823500
+    },
+    {
+      "epoch": 98.21,
+      "learning_rate": 9.100711939477714e-07,
+      "loss": 1.7789,
+      "step": 19824000
+    },
+    {
+      "epoch": 98.22,
+      "learning_rate": 9.088326075216877e-07,
+      "loss": 1.7769,
+      "step": 19824500
+    },
+    {
+      "epoch": 98.22,
+      "learning_rate": 9.075940210956041e-07,
+      "loss": 1.7737,
+      "step": 19825000
+    },
+    {
+      "epoch": 98.22,
+      "learning_rate": 9.063554346695204e-07,
+      "loss": 1.7564,
+      "step": 19825500
+    },
+    {
+      "epoch": 98.22,
+      "learning_rate": 9.051168482434367e-07,
+      "loss": 1.751,
+      "step": 19826000
+    },
+    {
+      "epoch": 98.23,
+      "learning_rate": 9.038782618173531e-07,
+      "loss": 1.7562,
+      "step": 19826500
+    },
+    {
+      "epoch": 98.23,
+      "learning_rate": 9.026396753912696e-07,
+      "loss": 1.7619,
+      "step": 19827000
+    },
+    {
+      "epoch": 98.23,
+      "learning_rate": 9.014010889651859e-07,
+      "loss": 1.7527,
+      "step": 19827500
+    },
+    {
+      "epoch": 98.23,
+      "learning_rate": 9.001625025391023e-07,
+      "loss": 1.7428,
+      "step": 19828000
+    },
+    {
+      "epoch": 98.24,
+      "learning_rate": 8.989239161130186e-07,
+      "loss": 1.7683,
+      "step": 19828500
+    },
+    {
+      "epoch": 98.24,
+      "learning_rate": 8.976853296869349e-07,
+      "loss": 1.7528,
+      "step": 19829000
+    },
+    {
+      "epoch": 98.24,
+      "learning_rate": 8.964467432608513e-07,
+      "loss": 1.7577,
+      "step": 19829500
+    },
+    {
+      "epoch": 98.24,
+      "learning_rate": 8.952081568347678e-07,
+      "loss": 1.7656,
+      "step": 19830000
+    },
+    {
+      "epoch": 98.25,
+      "learning_rate": 8.939695704086841e-07,
+      "loss": 1.7471,
+      "step": 19830500
+    },
+    {
+      "epoch": 98.25,
+      "learning_rate": 8.927309839826004e-07,
+      "loss": 1.7794,
+      "step": 19831000
+    },
+    {
+      "epoch": 98.25,
+      "learning_rate": 8.91494874729369e-07,
+      "loss": 1.7637,
+      "step": 19831500
+    },
+    {
+      "epoch": 98.25,
+      "learning_rate": 8.902587654761374e-07,
+      "loss": 1.7511,
+      "step": 19832000
+    },
+    {
+      "epoch": 98.26,
+      "learning_rate": 8.890201790500537e-07,
+      "loss": 1.7753,
+      "step": 19832500
+    },
+    {
+      "epoch": 98.26,
+      "learning_rate": 8.877840697968223e-07,
+      "loss": 1.7634,
+      "step": 19833000
+    },
+    {
+      "epoch": 98.26,
+      "learning_rate": 8.865454833707386e-07,
+      "loss": 1.783,
+      "step": 19833500
+    },
+    {
+      "epoch": 98.26,
+      "learning_rate": 8.85306896944655e-07,
+      "loss": 1.7684,
+      "step": 19834000
+    },
+    {
+      "epoch": 98.27,
+      "learning_rate": 8.840683105185713e-07,
+      "loss": 1.7591,
+      "step": 19834500
+    },
+    {
+      "epoch": 98.27,
+      "learning_rate": 8.828297240924878e-07,
+      "loss": 1.764,
+      "step": 19835000
+    },
+    {
+      "epoch": 98.27,
+      "learning_rate": 8.815911376664041e-07,
+      "loss": 1.7483,
+      "step": 19835500
+    },
+    {
+      "epoch": 98.27,
+      "learning_rate": 8.803525512403205e-07,
+      "loss": 1.7387,
+      "step": 19836000
+    },
+    {
+      "epoch": 98.28,
+      "learning_rate": 8.791139648142368e-07,
+      "loss": 1.7695,
+      "step": 19836500
+    },
+    {
+      "epoch": 98.28,
+      "learning_rate": 8.778753783881532e-07,
+      "loss": 1.7549,
+      "step": 19837000
+    },
+    {
+      "epoch": 98.28,
+      "learning_rate": 8.766367919620695e-07,
+      "loss": 1.7789,
+      "step": 19837500
+    },
+    {
+      "epoch": 98.28,
+      "learning_rate": 8.754006827088381e-07,
+      "loss": 1.7495,
+      "step": 19838000
+    },
+    {
+      "epoch": 98.29,
+      "learning_rate": 8.741620962827544e-07,
+      "loss": 1.7448,
+      "step": 19838500
+    },
+    {
+      "epoch": 98.29,
+      "learning_rate": 8.729235098566708e-07,
+      "loss": 1.7539,
+      "step": 19839000
+    },
+    {
+      "epoch": 98.29,
+      "learning_rate": 8.716849234305873e-07,
+      "loss": 1.7722,
+      "step": 19839500
+    },
+    {
+      "epoch": 98.29,
+      "learning_rate": 8.704463370045036e-07,
+      "loss": 1.75,
+      "step": 19840000
+    },
+    {
+      "epoch": 98.3,
+      "learning_rate": 8.692077505784199e-07,
+      "loss": 1.746,
+      "step": 19840500
+    },
+    {
+      "epoch": 98.3,
+      "learning_rate": 8.679691641523363e-07,
+      "loss": 1.7555,
+      "step": 19841000
+    },
+    {
+      "epoch": 98.3,
+      "learning_rate": 8.667305777262526e-07,
+      "loss": 1.772,
+      "step": 19841500
+    },
+    {
+      "epoch": 98.3,
+      "learning_rate": 8.654919913001689e-07,
+      "loss": 1.767,
+      "step": 19842000
+    },
+    {
+      "epoch": 98.31,
+      "learning_rate": 8.642534048740854e-07,
+      "loss": 1.7683,
+      "step": 19842500
+    },
+    {
+      "epoch": 98.31,
+      "learning_rate": 8.630148184480018e-07,
+      "loss": 1.727,
+      "step": 19843000
+    },
+    {
+      "epoch": 98.31,
+      "learning_rate": 8.617762320219181e-07,
+      "loss": 1.7596,
+      "step": 19843500
+    },
+    {
+      "epoch": 98.31,
+      "learning_rate": 8.605376455958345e-07,
+      "loss": 1.7415,
+      "step": 19844000
+    },
+    {
+      "epoch": 98.32,
+      "learning_rate": 8.59301536342603e-07,
+      "loss": 1.7687,
+      "step": 19844500
+    },
+    {
+      "epoch": 98.32,
+      "learning_rate": 8.580654270893714e-07,
+      "loss": 1.778,
+      "step": 19845000
+    },
+    {
+      "epoch": 98.32,
+      "learning_rate": 8.5682931783614e-07,
+      "loss": 1.7664,
+      "step": 19845500
+    },
+    {
+      "epoch": 98.32,
+      "learning_rate": 8.555907314100563e-07,
+      "loss": 1.7437,
+      "step": 19846000
+    },
+    {
+      "epoch": 98.33,
+      "learning_rate": 8.543521449839726e-07,
+      "loss": 1.7531,
+      "step": 19846500
+    },
+    {
+      "epoch": 98.33,
+      "learning_rate": 8.531135585578891e-07,
+      "loss": 1.7711,
+      "step": 19847000
+    },
+    {
+      "epoch": 98.33,
+      "learning_rate": 8.518749721318055e-07,
+      "loss": 1.7628,
+      "step": 19847500
+    },
+    {
+      "epoch": 98.33,
+      "learning_rate": 8.506363857057218e-07,
+      "loss": 1.7677,
+      "step": 19848000
+    },
+    {
+      "epoch": 98.34,
+      "learning_rate": 8.493977992796382e-07,
+      "loss": 1.7678,
+      "step": 19848500
+    },
+    {
+      "epoch": 98.34,
+      "learning_rate": 8.481616900264068e-07,
+      "loss": 1.7668,
+      "step": 19849000
+    },
+    {
+      "epoch": 98.34,
+      "learning_rate": 8.469231036003231e-07,
+      "loss": 1.7717,
+      "step": 19849500
+    },
+    {
+      "epoch": 98.34,
+      "learning_rate": 8.456869943470915e-07,
+      "loss": 1.7698,
+      "step": 19850000
+    },
+    {
+      "epoch": 98.35,
+      "learning_rate": 8.44448407921008e-07,
+      "loss": 1.7801,
+      "step": 19850500
+    },
+    {
+      "epoch": 98.35,
+      "learning_rate": 8.432098214949244e-07,
+      "loss": 1.7777,
+      "step": 19851000
+    },
+    {
+      "epoch": 98.35,
+      "learning_rate": 8.419712350688407e-07,
+      "loss": 1.766,
+      "step": 19851500
+    },
+    {
+      "epoch": 98.35,
+      "learning_rate": 8.40732648642757e-07,
+      "loss": 1.7746,
+      "step": 19852000
+    },
+    {
+      "epoch": 98.36,
+      "learning_rate": 8.394940622166734e-07,
+      "loss": 1.7569,
+      "step": 19852500
+    },
+    {
+      "epoch": 98.36,
+      "learning_rate": 8.382554757905897e-07,
+      "loss": 1.7748,
+      "step": 19853000
+    },
+    {
+      "epoch": 98.36,
+      "learning_rate": 8.370168893645062e-07,
+      "loss": 1.7529,
+      "step": 19853500
+    },
+    {
+      "epoch": 98.36,
+      "learning_rate": 8.357783029384225e-07,
+      "loss": 1.7736,
+      "step": 19854000
+    },
+    {
+      "epoch": 98.37,
+      "learning_rate": 8.345397165123389e-07,
+      "loss": 1.7537,
+      "step": 19854500
+    },
+    {
+      "epoch": 98.37,
+      "learning_rate": 8.333011300862552e-07,
+      "loss": 1.7423,
+      "step": 19855000
+    },
+    {
+      "epoch": 98.37,
+      "learning_rate": 8.320625436601716e-07,
+      "loss": 1.7377,
+      "step": 19855500
+    },
+    {
+      "epoch": 98.37,
+      "learning_rate": 8.308239572340879e-07,
+      "loss": 1.758,
+      "step": 19856000
+    },
+    {
+      "epoch": 98.38,
+      "learning_rate": 8.295878479808565e-07,
+      "loss": 1.7618,
+      "step": 19856500
+    },
+    {
+      "epoch": 98.38,
+      "learning_rate": 8.283492615547727e-07,
+      "loss": 1.772,
+      "step": 19857000
+    },
+    {
+      "epoch": 98.38,
+      "learning_rate": 8.271106751286891e-07,
+      "loss": 1.7681,
+      "step": 19857500
+    },
+    {
+      "epoch": 98.38,
+      "learning_rate": 8.258720887026057e-07,
+      "loss": 1.743,
+      "step": 19858000
+    },
+    {
+      "epoch": 98.39,
+      "learning_rate": 8.24635979449374e-07,
+      "loss": 1.7709,
+      "step": 19858500
+    },
+    {
+      "epoch": 98.39,
+      "learning_rate": 8.233998701961426e-07,
+      "loss": 1.7771,
+      "step": 19859000
+    },
+    {
+      "epoch": 98.39,
+      "learning_rate": 8.221612837700589e-07,
+      "loss": 1.7787,
+      "step": 19859500
+    },
+    {
+      "epoch": 98.39,
+      "learning_rate": 8.209226973439753e-07,
+      "loss": 1.7447,
+      "step": 19860000
+    },
+    {
+      "epoch": 98.4,
+      "learning_rate": 8.196841109178916e-07,
+      "loss": 1.7661,
+      "step": 19860500
+    },
+    {
+      "epoch": 98.4,
+      "learning_rate": 8.184455244918079e-07,
+      "loss": 1.7668,
+      "step": 19861000
+    },
+    {
+      "epoch": 98.4,
+      "learning_rate": 8.172069380657244e-07,
+      "loss": 1.7754,
+      "step": 19861500
+    },
+    {
+      "epoch": 98.4,
+      "learning_rate": 8.159683516396408e-07,
+      "loss": 1.7673,
+      "step": 19862000
+    },
+    {
+      "epoch": 98.41,
+      "learning_rate": 8.147322423864092e-07,
+      "loss": 1.7533,
+      "step": 19862500
+    },
+    {
+      "epoch": 98.41,
+      "learning_rate": 8.134936559603257e-07,
+      "loss": 1.7664,
+      "step": 19863000
+    },
+    {
+      "epoch": 98.41,
+      "learning_rate": 8.12255069534242e-07,
+      "loss": 1.7806,
+      "step": 19863500
+    },
+    {
+      "epoch": 98.41,
+      "learning_rate": 8.110164831081584e-07,
+      "loss": 1.7439,
+      "step": 19864000
+    },
+    {
+      "epoch": 98.42,
+      "learning_rate": 8.097778966820747e-07,
+      "loss": 1.7396,
+      "step": 19864500
+    },
+    {
+      "epoch": 98.42,
+      "learning_rate": 8.08539310255991e-07,
+      "loss": 1.7486,
+      "step": 19865000
+    },
+    {
+      "epoch": 98.42,
+      "learning_rate": 8.073007238299074e-07,
+      "loss": 1.7563,
+      "step": 19865500
+    },
+    {
+      "epoch": 98.42,
+      "learning_rate": 8.060621374038239e-07,
+      "loss": 1.7816,
+      "step": 19866000
+    },
+    {
+      "epoch": 98.43,
+      "learning_rate": 8.048235509777402e-07,
+      "loss": 1.7581,
+      "step": 19866500
+    },
+    {
+      "epoch": 98.43,
+      "learning_rate": 8.035849645516566e-07,
+      "loss": 1.7542,
+      "step": 19867000
+    },
+    {
+      "epoch": 98.43,
+      "learning_rate": 8.023463781255729e-07,
+      "loss": 1.7493,
+      "step": 19867500
+    },
+    {
+      "epoch": 98.43,
+      "learning_rate": 8.011077916994892e-07,
+      "loss": 1.7595,
+      "step": 19868000
+    },
+    {
+      "epoch": 98.44,
+      "learning_rate": 7.998692052734056e-07,
+      "loss": 1.7722,
+      "step": 19868500
+    },
+    {
+      "epoch": 98.44,
+      "learning_rate": 7.986330960201742e-07,
+      "loss": 1.7789,
+      "step": 19869000
+    },
+    {
+      "epoch": 98.44,
+      "learning_rate": 7.973945095940905e-07,
+      "loss": 1.7632,
+      "step": 19869500
+    },
+    {
+      "epoch": 98.44,
+      "learning_rate": 7.96158400340859e-07,
+      "loss": 1.7494,
+      "step": 19870000
+    },
+    {
+      "epoch": 98.45,
+      "learning_rate": 7.949198139147753e-07,
+      "loss": 1.7418,
+      "step": 19870500
+    },
+    {
+      "epoch": 98.45,
+      "learning_rate": 7.936812274886917e-07,
+      "loss": 1.7656,
+      "step": 19871000
+    },
+    {
+      "epoch": 98.45,
+      "learning_rate": 7.92442641062608e-07,
+      "loss": 1.749,
+      "step": 19871500
+    },
+    {
+      "epoch": 98.45,
+      "learning_rate": 7.912040546365245e-07,
+      "loss": 1.7481,
+      "step": 19872000
+    },
+    {
+      "epoch": 98.46,
+      "learning_rate": 7.899679453832929e-07,
+      "loss": 1.7642,
+      "step": 19872500
+    },
+    {
+      "epoch": 98.46,
+      "learning_rate": 7.887293589572093e-07,
+      "loss": 1.7562,
+      "step": 19873000
+    },
+    {
+      "epoch": 98.46,
+      "learning_rate": 7.874907725311256e-07,
+      "loss": 1.731,
+      "step": 19873500
+    },
+    {
+      "epoch": 98.46,
+      "learning_rate": 7.862521861050421e-07,
+      "loss": 1.7488,
+      "step": 19874000
+    },
+    {
+      "epoch": 98.47,
+      "learning_rate": 7.850135996789585e-07,
+      "loss": 1.753,
+      "step": 19874500
+    },
+    {
+      "epoch": 98.47,
+      "learning_rate": 7.837750132528748e-07,
+      "loss": 1.7585,
+      "step": 19875000
+    },
+    {
+      "epoch": 98.47,
+      "learning_rate": 7.825364268267911e-07,
+      "loss": 1.7577,
+      "step": 19875500
+    },
+    {
+      "epoch": 98.47,
+      "learning_rate": 7.812978404007075e-07,
+      "loss": 1.7818,
+      "step": 19876000
+    },
+    {
+      "epoch": 98.48,
+      "learning_rate": 7.800617311474761e-07,
+      "loss": 1.7708,
+      "step": 19876500
+    },
+    {
+      "epoch": 98.48,
+      "learning_rate": 7.788231447213924e-07,
+      "loss": 1.7725,
+      "step": 19877000
+    },
+    {
+      "epoch": 98.48,
+      "learning_rate": 7.775845582953088e-07,
+      "loss": 1.7589,
+      "step": 19877500
+    },
+    {
+      "epoch": 98.48,
+      "learning_rate": 7.763459718692251e-07,
+      "loss": 1.783,
+      "step": 19878000
+    },
+    {
+      "epoch": 98.48,
+      "learning_rate": 7.751073854431415e-07,
+      "loss": 1.7505,
+      "step": 19878500
+    },
+    {
+      "epoch": 98.49,
+      "learning_rate": 7.7387127618991e-07,
+      "loss": 1.7474,
+      "step": 19879000
+    },
+    {
+      "epoch": 98.49,
+      "learning_rate": 7.726351669366786e-07,
+      "loss": 1.7412,
+      "step": 19879500
+    },
+    {
+      "epoch": 98.49,
+      "learning_rate": 7.713990576834471e-07,
+      "loss": 1.7456,
+      "step": 19880000
+    },
+    {
+      "epoch": 98.49,
+      "learning_rate": 7.701604712573634e-07,
+      "loss": 1.745,
+      "step": 19880500
+    },
+    {
+      "epoch": 98.5,
+      "learning_rate": 7.689218848312799e-07,
+      "loss": 1.765,
+      "step": 19881000
+    },
+    {
+      "epoch": 98.5,
+      "learning_rate": 7.676832984051962e-07,
+      "loss": 1.7472,
+      "step": 19881500
+    },
+    {
+      "epoch": 98.5,
+      "learning_rate": 7.664447119791125e-07,
+      "loss": 1.794,
+      "step": 19882000
+    },
+    {
+      "epoch": 98.5,
+      "learning_rate": 7.65206125553029e-07,
+      "loss": 1.7685,
+      "step": 19882500
+    },
+    {
+      "epoch": 98.51,
+      "learning_rate": 7.639675391269452e-07,
+      "loss": 1.7655,
+      "step": 19883000
+    },
+    {
+      "epoch": 98.51,
+      "learning_rate": 7.627289527008615e-07,
+      "loss": 1.7721,
+      "step": 19883500
+    },
+    {
+      "epoch": 98.51,
+      "learning_rate": 7.614903662747779e-07,
+      "loss": 1.7591,
+      "step": 19884000
+    },
+    {
+      "epoch": 98.51,
+      "learning_rate": 7.602517798486943e-07,
+      "loss": 1.7528,
+      "step": 19884500
+    },
+    {
+      "epoch": 98.52,
+      "learning_rate": 7.590131934226106e-07,
+      "loss": 1.7471,
+      "step": 19885000
+    },
+    {
+      "epoch": 98.52,
+      "learning_rate": 7.57774606996527e-07,
+      "loss": 1.7757,
+      "step": 19885500
+    },
+    {
+      "epoch": 98.52,
+      "learning_rate": 7.565360205704434e-07,
+      "loss": 1.7342,
+      "step": 19886000
+    },
+    {
+      "epoch": 98.52,
+      "learning_rate": 7.552974341443597e-07,
+      "loss": 1.7749,
+      "step": 19886500
+    },
+    {
+      "epoch": 98.53,
+      "learning_rate": 7.540613248911283e-07,
+      "loss": 1.7655,
+      "step": 19887000
+    },
+    {
+      "epoch": 98.53,
+      "learning_rate": 7.528252156378968e-07,
+      "loss": 1.7489,
+      "step": 19887500
+    },
+    {
+      "epoch": 98.53,
+      "learning_rate": 7.515866292118132e-07,
+      "loss": 1.7582,
+      "step": 19888000
+    },
+    {
+      "epoch": 98.53,
+      "learning_rate": 7.503480427857295e-07,
+      "loss": 1.7692,
+      "step": 19888500
+    },
+    {
+      "epoch": 98.54,
+      "learning_rate": 7.491094563596459e-07,
+      "loss": 1.769,
+      "step": 19889000
+    },
+    {
+      "epoch": 98.54,
+      "learning_rate": 7.478708699335622e-07,
+      "loss": 1.7609,
+      "step": 19889500
+    },
+    {
+      "epoch": 98.54,
+      "learning_rate": 7.466347606803308e-07,
+      "loss": 1.7562,
+      "step": 19890000
+    },
+    {
+      "epoch": 98.54,
+      "learning_rate": 7.453961742542472e-07,
+      "loss": 1.7585,
+      "step": 19890500
+    },
+    {
+      "epoch": 98.55,
+      "learning_rate": 7.441600650010157e-07,
+      "loss": 1.7718,
+      "step": 19891000
+    },
+    {
+      "epoch": 98.55,
+      "learning_rate": 7.42921478574932e-07,
+      "loss": 1.7594,
+      "step": 19891500
+    },
+    {
+      "epoch": 98.55,
+      "learning_rate": 7.416828921488484e-07,
+      "loss": 1.7583,
+      "step": 19892000
+    },
+    {
+      "epoch": 98.55,
+      "learning_rate": 7.404443057227648e-07,
+      "loss": 1.7754,
+      "step": 19892500
+    },
+    {
+      "epoch": 98.56,
+      "learning_rate": 7.392081964695333e-07,
+      "loss": 1.7598,
+      "step": 19893000
+    },
+    {
+      "epoch": 98.56,
+      "learning_rate": 7.379720872163018e-07,
+      "loss": 1.7743,
+      "step": 19893500
+    },
+    {
+      "epoch": 98.56,
+      "learning_rate": 7.367335007902182e-07,
+      "loss": 1.7669,
+      "step": 19894000
+    },
+    {
+      "epoch": 98.56,
+      "learning_rate": 7.354949143641346e-07,
+      "loss": 1.7569,
+      "step": 19894500
+    },
+    {
+      "epoch": 98.57,
+      "learning_rate": 7.34256327938051e-07,
+      "loss": 1.7755,
+      "step": 19895000
+    },
+    {
+      "epoch": 98.57,
+      "learning_rate": 7.330202186848194e-07,
+      "loss": 1.7562,
+      "step": 19895500
+    },
+    {
+      "epoch": 98.57,
+      "learning_rate": 7.317816322587357e-07,
+      "loss": 1.7859,
+      "step": 19896000
+    },
+    {
+      "epoch": 98.57,
+      "learning_rate": 7.305430458326521e-07,
+      "loss": 1.7473,
+      "step": 19896500
+    },
+    {
+      "epoch": 98.58,
+      "learning_rate": 7.293044594065685e-07,
+      "loss": 1.7524,
+      "step": 19897000
+    },
+    {
+      "epoch": 98.58,
+      "learning_rate": 7.280658729804848e-07,
+      "loss": 1.7527,
+      "step": 19897500
+    },
+    {
+      "epoch": 98.58,
+      "learning_rate": 7.268272865544012e-07,
+      "loss": 1.7752,
+      "step": 19898000
+    },
+    {
+      "epoch": 98.58,
+      "learning_rate": 7.255887001283176e-07,
+      "loss": 1.763,
+      "step": 19898500
+    },
+    {
+      "epoch": 98.59,
+      "learning_rate": 7.243525908750861e-07,
+      "loss": 1.7595,
+      "step": 19899000
+    },
+    {
+      "epoch": 98.59,
+      "learning_rate": 7.231140044490024e-07,
+      "loss": 1.7701,
+      "step": 19899500
+    },
+    {
+      "epoch": 98.59,
+      "learning_rate": 7.218754180229189e-07,
+      "loss": 1.7829,
+      "step": 19900000
+    },
+    {
+      "epoch": 98.59,
+      "learning_rate": 7.206368315968352e-07,
+      "loss": 1.7343,
+      "step": 19900500
+    },
+    {
+      "epoch": 98.6,
+      "learning_rate": 7.193982451707515e-07,
+      "loss": 1.7697,
+      "step": 19901000
+    },
+    {
+      "epoch": 98.6,
+      "learning_rate": 7.181596587446679e-07,
+      "loss": 1.773,
+      "step": 19901500
+    },
+    {
+      "epoch": 98.6,
+      "learning_rate": 7.169210723185843e-07,
+      "loss": 1.7916,
+      "step": 19902000
+    },
+    {
+      "epoch": 98.6,
+      "learning_rate": 7.156824858925006e-07,
+      "loss": 1.7715,
+      "step": 19902500
+    },
+    {
+      "epoch": 98.61,
+      "learning_rate": 7.14443899466417e-07,
+      "loss": 1.759,
+      "step": 19903000
+    },
+    {
+      "epoch": 98.61,
+      "learning_rate": 7.132053130403334e-07,
+      "loss": 1.7443,
+      "step": 19903500
+    },
+    {
+      "epoch": 98.61,
+      "learning_rate": 7.119667266142497e-07,
+      "loss": 1.7597,
+      "step": 19904000
+    },
+    {
+      "epoch": 98.61,
+      "learning_rate": 7.107281401881661e-07,
+      "loss": 1.7537,
+      "step": 19904500
+    },
+    {
+      "epoch": 98.62,
+      "learning_rate": 7.094895537620825e-07,
+      "loss": 1.763,
+      "step": 19905000
+    },
+    {
+      "epoch": 98.62,
+      "learning_rate": 7.08253444508851e-07,
+      "loss": 1.7714,
+      "step": 19905500
+    },
+    {
+      "epoch": 98.62,
+      "learning_rate": 7.070148580827674e-07,
+      "loss": 1.7592,
+      "step": 19906000
+    },
+    {
+      "epoch": 98.62,
+      "learning_rate": 7.057762716566837e-07,
+      "loss": 1.756,
+      "step": 19906500
+    },
+    {
+      "epoch": 98.63,
+      "learning_rate": 7.045376852306001e-07,
+      "loss": 1.7714,
+      "step": 19907000
+    },
+    {
+      "epoch": 98.63,
+      "learning_rate": 7.032990988045165e-07,
+      "loss": 1.771,
+      "step": 19907500
+    },
+    {
+      "epoch": 98.63,
+      "learning_rate": 7.02062989551285e-07,
+      "loss": 1.7527,
+      "step": 19908000
+    },
+    {
+      "epoch": 98.63,
+      "learning_rate": 7.008268802980534e-07,
+      "loss": 1.764,
+      "step": 19908500
+    },
+    {
+      "epoch": 98.64,
+      "learning_rate": 6.995882938719698e-07,
+      "loss": 1.7534,
+      "step": 19909000
+    },
+    {
+      "epoch": 98.64,
+      "learning_rate": 6.983497074458862e-07,
+      "loss": 1.7711,
+      "step": 19909500
+    },
+    {
+      "epoch": 98.64,
+      "learning_rate": 6.971111210198025e-07,
+      "loss": 1.7525,
+      "step": 19910000
+    },
+    {
+      "epoch": 98.64,
+      "learning_rate": 6.958725345937188e-07,
+      "loss": 1.7637,
+      "step": 19910500
+    },
+    {
+      "epoch": 98.65,
+      "learning_rate": 6.946339481676353e-07,
+      "loss": 1.7835,
+      "step": 19911000
+    },
+    {
+      "epoch": 98.65,
+      "learning_rate": 6.933953617415516e-07,
+      "loss": 1.7604,
+      "step": 19911500
+    },
+    {
+      "epoch": 98.65,
+      "learning_rate": 6.921592524883201e-07,
+      "loss": 1.76,
+      "step": 19912000
+    },
+    {
+      "epoch": 98.65,
+      "learning_rate": 6.909206660622365e-07,
+      "loss": 1.7501,
+      "step": 19912500
+    },
+    {
+      "epoch": 98.66,
+      "learning_rate": 6.896820796361529e-07,
+      "loss": 1.7673,
+      "step": 19913000
+    },
+    {
+      "epoch": 98.66,
+      "learning_rate": 6.884434932100692e-07,
+      "loss": 1.7479,
+      "step": 19913500
+    },
+    {
+      "epoch": 98.66,
+      "learning_rate": 6.872049067839856e-07,
+      "loss": 1.756,
+      "step": 19914000
+    },
+    {
+      "epoch": 98.66,
+      "learning_rate": 6.85966320357902e-07,
+      "loss": 1.7843,
+      "step": 19914500
+    },
+    {
+      "epoch": 98.67,
+      "learning_rate": 6.847277339318183e-07,
+      "loss": 1.7636,
+      "step": 19915000
+    },
+    {
+      "epoch": 98.67,
+      "learning_rate": 6.834891475057347e-07,
+      "loss": 1.7613,
+      "step": 19915500
+    },
+    {
+      "epoch": 98.67,
+      "learning_rate": 6.822505610796511e-07,
+      "loss": 1.7562,
+      "step": 19916000
+    },
+    {
+      "epoch": 98.67,
+      "learning_rate": 6.810144518264196e-07,
+      "loss": 1.7684,
+      "step": 19916500
+    },
+    {
+      "epoch": 98.68,
+      "learning_rate": 6.79775865400336e-07,
+      "loss": 1.768,
+      "step": 19917000
+    },
+    {
+      "epoch": 98.68,
+      "learning_rate": 6.785372789742523e-07,
+      "loss": 1.7716,
+      "step": 19917500
+    },
+    {
+      "epoch": 98.68,
+      "learning_rate": 6.773011697210208e-07,
+      "loss": 1.7659,
+      "step": 19918000
+    },
+    {
+      "epoch": 98.68,
+      "learning_rate": 6.760625832949373e-07,
+      "loss": 1.767,
+      "step": 19918500
+    },
+    {
+      "epoch": 98.69,
+      "learning_rate": 6.748239968688536e-07,
+      "loss": 1.7554,
+      "step": 19919000
+    },
+    {
+      "epoch": 98.69,
+      "learning_rate": 6.735854104427699e-07,
+      "loss": 1.774,
+      "step": 19919500
+    },
+    {
+      "epoch": 98.69,
+      "learning_rate": 6.723468240166863e-07,
+      "loss": 1.7621,
+      "step": 19920000
+    },
+    {
+      "epoch": 98.69,
+      "learning_rate": 6.71113191936307e-07,
+      "loss": 1.7607,
+      "step": 19920500
+    },
+    {
+      "epoch": 98.7,
+      "learning_rate": 6.698746055102233e-07,
+      "loss": 1.7459,
+      "step": 19921000
+    },
+    {
+      "epoch": 98.7,
+      "learning_rate": 6.686360190841396e-07,
+      "loss": 1.777,
+      "step": 19921500
+    },
+    {
+      "epoch": 98.7,
+      "learning_rate": 6.67397432658056e-07,
+      "loss": 1.7674,
+      "step": 19922000
+    },
+    {
+      "epoch": 98.7,
+      "learning_rate": 6.661588462319724e-07,
+      "loss": 1.7706,
+      "step": 19922500
+    },
+    {
+      "epoch": 98.71,
+      "learning_rate": 6.649202598058887e-07,
+      "loss": 1.7572,
+      "step": 19923000
+    },
+    {
+      "epoch": 98.71,
+      "learning_rate": 6.636816733798051e-07,
+      "loss": 1.754,
+      "step": 19923500
+    },
+    {
+      "epoch": 98.71,
+      "learning_rate": 6.624430869537215e-07,
+      "loss": 1.7912,
+      "step": 19924000
+    },
+    {
+      "epoch": 98.71,
+      "learning_rate": 6.612045005276378e-07,
+      "loss": 1.7453,
+      "step": 19924500
+    },
+    {
+      "epoch": 98.72,
+      "learning_rate": 6.599659141015542e-07,
+      "loss": 1.7433,
+      "step": 19925000
+    },
+    {
+      "epoch": 98.72,
+      "learning_rate": 6.587273276754705e-07,
+      "loss": 1.7627,
+      "step": 19925500
+    },
+    {
+      "epoch": 98.72,
+      "learning_rate": 6.574912184222391e-07,
+      "loss": 1.7769,
+      "step": 19926000
+    },
+    {
+      "epoch": 98.72,
+      "learning_rate": 6.562526319961555e-07,
+      "loss": 1.7326,
+      "step": 19926500
+    },
+    {
+      "epoch": 98.73,
+      "learning_rate": 6.550140455700718e-07,
+      "loss": 1.7575,
+      "step": 19927000
+    },
+    {
+      "epoch": 98.73,
+      "learning_rate": 6.537754591439882e-07,
+      "loss": 1.7729,
+      "step": 19927500
+    },
+    {
+      "epoch": 98.73,
+      "learning_rate": 6.525368727179046e-07,
+      "loss": 1.7656,
+      "step": 19928000
+    },
+    {
+      "epoch": 98.73,
+      "learning_rate": 6.512982862918209e-07,
+      "loss": 1.7689,
+      "step": 19928500
+    },
+    {
+      "epoch": 98.74,
+      "learning_rate": 6.500596998657372e-07,
+      "loss": 1.7773,
+      "step": 19929000
+    },
+    {
+      "epoch": 98.74,
+      "learning_rate": 6.488211134396537e-07,
+      "loss": 1.7602,
+      "step": 19929500
+    },
+    {
+      "epoch": 98.74,
+      "learning_rate": 6.4758252701357e-07,
+      "loss": 1.7582,
+      "step": 19930000
+    },
+    {
+      "epoch": 98.74,
+      "learning_rate": 6.463439405874863e-07,
+      "loss": 1.7863,
+      "step": 19930500
+    },
+    {
+      "epoch": 98.75,
+      "learning_rate": 6.451053541614028e-07,
+      "loss": 1.7684,
+      "step": 19931000
+    },
+    {
+      "epoch": 98.75,
+      "learning_rate": 6.438667677353191e-07,
+      "loss": 1.7703,
+      "step": 19931500
+    },
+    {
+      "epoch": 98.75,
+      "learning_rate": 6.426331356549398e-07,
+      "loss": 1.7741,
+      "step": 19932000
+    },
+    {
+      "epoch": 98.75,
+      "learning_rate": 6.413945492288561e-07,
+      "loss": 1.7631,
+      "step": 19932500
+    },
+    {
+      "epoch": 98.75,
+      "learning_rate": 6.401559628027725e-07,
+      "loss": 1.7686,
+      "step": 19933000
+    },
+    {
+      "epoch": 98.76,
+      "learning_rate": 6.389198535495411e-07,
+      "loss": 1.7607,
+      "step": 19933500
+    },
+    {
+      "epoch": 98.76,
+      "learning_rate": 6.376837442963095e-07,
+      "loss": 1.7537,
+      "step": 19934000
+    },
+    {
+      "epoch": 98.76,
+      "learning_rate": 6.364451578702259e-07,
+      "loss": 1.7738,
+      "step": 19934500
+    },
+    {
+      "epoch": 98.76,
+      "learning_rate": 6.352065714441422e-07,
+      "loss": 1.7772,
+      "step": 19935000
+    },
+    {
+      "epoch": 98.77,
+      "learning_rate": 6.339679850180586e-07,
+      "loss": 1.7415,
+      "step": 19935500
+    },
+    {
+      "epoch": 98.77,
+      "learning_rate": 6.32729398591975e-07,
+      "loss": 1.7646,
+      "step": 19936000
+    },
+    {
+      "epoch": 98.77,
+      "learning_rate": 6.314908121658913e-07,
+      "loss": 1.7591,
+      "step": 19936500
+    },
+    {
+      "epoch": 98.77,
+      "learning_rate": 6.302522257398076e-07,
+      "loss": 1.7587,
+      "step": 19937000
+    },
+    {
+      "epoch": 98.78,
+      "learning_rate": 6.290161164865762e-07,
+      "loss": 1.7769,
+      "step": 19937500
+    },
+    {
+      "epoch": 98.78,
+      "learning_rate": 6.277775300604926e-07,
+      "loss": 1.7386,
+      "step": 19938000
+    },
+    {
+      "epoch": 98.78,
+      "learning_rate": 6.265389436344089e-07,
+      "loss": 1.7738,
+      "step": 19938500
+    },
+    {
+      "epoch": 98.78,
+      "learning_rate": 6.253003572083253e-07,
+      "loss": 1.7834,
+      "step": 19939000
+    },
+    {
+      "epoch": 98.79,
+      "learning_rate": 6.240642479550939e-07,
+      "loss": 1.7553,
+      "step": 19939500
+    },
+    {
+      "epoch": 98.79,
+      "learning_rate": 6.228256615290102e-07,
+      "loss": 1.7858,
+      "step": 19940000
+    },
+    {
+      "epoch": 98.79,
+      "learning_rate": 6.215870751029265e-07,
+      "loss": 1.7736,
+      "step": 19940500
+    },
+    {
+      "epoch": 98.79,
+      "learning_rate": 6.203484886768429e-07,
+      "loss": 1.7616,
+      "step": 19941000
+    },
+    {
+      "epoch": 98.8,
+      "learning_rate": 6.191099022507593e-07,
+      "loss": 1.7615,
+      "step": 19941500
+    },
+    {
+      "epoch": 98.8,
+      "learning_rate": 6.178713158246756e-07,
+      "loss": 1.7848,
+      "step": 19942000
+    },
+    {
+      "epoch": 98.8,
+      "learning_rate": 6.16632729398592e-07,
+      "loss": 1.7685,
+      "step": 19942500
+    },
+    {
+      "epoch": 98.8,
+      "learning_rate": 6.153941429725084e-07,
+      "loss": 1.7641,
+      "step": 19943000
+    },
+    {
+      "epoch": 98.81,
+      "learning_rate": 6.141555565464247e-07,
+      "loss": 1.7567,
+      "step": 19943500
+    },
+    {
+      "epoch": 98.81,
+      "learning_rate": 6.129169701203411e-07,
+      "loss": 1.7338,
+      "step": 19944000
+    },
+    {
+      "epoch": 98.81,
+      "learning_rate": 6.116783836942575e-07,
+      "loss": 1.7559,
+      "step": 19944500
+    },
+    {
+      "epoch": 98.81,
+      "learning_rate": 6.104397972681738e-07,
+      "loss": 1.7743,
+      "step": 19945000
+    },
+    {
+      "epoch": 98.82,
+      "learning_rate": 6.092012108420902e-07,
+      "loss": 1.7761,
+      "step": 19945500
+    },
+    {
+      "epoch": 98.82,
+      "learning_rate": 6.079651015888587e-07,
+      "loss": 1.767,
+      "step": 19946000
+    },
+    {
+      "epoch": 98.82,
+      "learning_rate": 6.067265151627751e-07,
+      "loss": 1.774,
+      "step": 19946500
+    },
+    {
+      "epoch": 98.82,
+      "learning_rate": 6.054879287366915e-07,
+      "loss": 1.7439,
+      "step": 19947000
+    },
+    {
+      "epoch": 98.83,
+      "learning_rate": 6.042493423106077e-07,
+      "loss": 1.7601,
+      "step": 19947500
+    },
+    {
+      "epoch": 98.83,
+      "learning_rate": 6.030132330573762e-07,
+      "loss": 1.7519,
+      "step": 19948000
+    },
+    {
+      "epoch": 98.83,
+      "learning_rate": 6.017746466312927e-07,
+      "loss": 1.7706,
+      "step": 19948500
+    },
+    {
+      "epoch": 98.83,
+      "learning_rate": 6.00536060205209e-07,
+      "loss": 1.7587,
+      "step": 19949000
+    },
+    {
+      "epoch": 98.84,
+      "learning_rate": 5.992974737791253e-07,
+      "loss": 1.7624,
+      "step": 19949500
+    },
+    {
+      "epoch": 98.84,
+      "learning_rate": 5.980613645258939e-07,
+      "loss": 1.7623,
+      "step": 19950000
+    },
+    {
+      "epoch": 98.84,
+      "learning_rate": 5.968227780998103e-07,
+      "loss": 1.7582,
+      "step": 19950500
+    },
+    {
+      "epoch": 98.84,
+      "learning_rate": 5.955841916737266e-07,
+      "loss": 1.78,
+      "step": 19951000
+    },
+    {
+      "epoch": 98.85,
+      "learning_rate": 5.94345605247643e-07,
+      "loss": 1.7852,
+      "step": 19951500
+    },
+    {
+      "epoch": 98.85,
+      "learning_rate": 5.931070188215594e-07,
+      "loss": 1.7913,
+      "step": 19952000
+    },
+    {
+      "epoch": 98.85,
+      "learning_rate": 5.918684323954757e-07,
+      "loss": 1.747,
+      "step": 19952500
+    },
+    {
+      "epoch": 98.85,
+      "learning_rate": 5.906298459693921e-07,
+      "loss": 1.7544,
+      "step": 19953000
+    },
+    {
+      "epoch": 98.86,
+      "learning_rate": 5.893912595433084e-07,
+      "loss": 1.762,
+      "step": 19953500
+    },
+    {
+      "epoch": 98.86,
+      "learning_rate": 5.881526731172248e-07,
+      "loss": 1.7411,
+      "step": 19954000
+    },
+    {
+      "epoch": 98.86,
+      "learning_rate": 5.869165638639934e-07,
+      "loss": 1.7551,
+      "step": 19954500
+    },
+    {
+      "epoch": 98.86,
+      "learning_rate": 5.856779774379097e-07,
+      "loss": 1.7655,
+      "step": 19955000
+    },
+    {
+      "epoch": 98.87,
+      "learning_rate": 5.844393910118261e-07,
+      "loss": 1.7744,
+      "step": 19955500
+    },
+    {
+      "epoch": 98.87,
+      "learning_rate": 5.832008045857425e-07,
+      "loss": 1.7486,
+      "step": 19956000
+    },
+    {
+      "epoch": 98.87,
+      "learning_rate": 5.819622181596588e-07,
+      "loss": 1.773,
+      "step": 19956500
+    },
+    {
+      "epoch": 98.87,
+      "learning_rate": 5.807236317335751e-07,
+      "loss": 1.7752,
+      "step": 19957000
+    },
+    {
+      "epoch": 98.88,
+      "learning_rate": 5.794850453074916e-07,
+      "loss": 1.7395,
+      "step": 19957500
+    },
+    {
+      "epoch": 98.88,
+      "learning_rate": 5.782464588814079e-07,
+      "loss": 1.7629,
+      "step": 19958000
+    },
+    {
+      "epoch": 98.88,
+      "learning_rate": 5.770078724553242e-07,
+      "loss": 1.7464,
+      "step": 19958500
+    },
+    {
+      "epoch": 98.88,
+      "learning_rate": 5.757717632020927e-07,
+      "loss": 1.7574,
+      "step": 19959000
+    },
+    {
+      "epoch": 98.89,
+      "learning_rate": 5.745331767760092e-07,
+      "loss": 1.741,
+      "step": 19959500
+    },
+    {
+      "epoch": 98.89,
+      "learning_rate": 5.732945903499255e-07,
+      "loss": 1.781,
+      "step": 19960000
+    },
+    {
+      "epoch": 98.89,
+      "learning_rate": 5.720560039238417e-07,
+      "loss": 1.7535,
+      "step": 19960500
+    },
+    {
+      "epoch": 98.89,
+      "learning_rate": 5.708174174977582e-07,
+      "loss": 1.7668,
+      "step": 19961000
+    },
+    {
+      "epoch": 98.9,
+      "learning_rate": 5.695788310716745e-07,
+      "loss": 1.7651,
+      "step": 19961500
+    },
+    {
+      "epoch": 98.9,
+      "learning_rate": 5.683402446455908e-07,
+      "loss": 1.7678,
+      "step": 19962000
+    },
+    {
+      "epoch": 98.9,
+      "learning_rate": 5.671016582195073e-07,
+      "loss": 1.7579,
+      "step": 19962500
+    },
+    {
+      "epoch": 98.9,
+      "learning_rate": 5.658630717934236e-07,
+      "loss": 1.7715,
+      "step": 19963000
+    },
+    {
+      "epoch": 98.91,
+      "learning_rate": 5.646244853673399e-07,
+      "loss": 1.7803,
+      "step": 19963500
+    },
+    {
+      "epoch": 98.91,
+      "learning_rate": 5.633858989412563e-07,
+      "loss": 1.7687,
+      "step": 19964000
+    },
+    {
+      "epoch": 98.91,
+      "learning_rate": 5.621473125151727e-07,
+      "loss": 1.7868,
+      "step": 19964500
+    },
+    {
+      "epoch": 98.91,
+      "learning_rate": 5.609112032619412e-07,
+      "loss": 1.7646,
+      "step": 19965000
+    },
+    {
+      "epoch": 98.92,
+      "learning_rate": 5.596726168358576e-07,
+      "loss": 1.7605,
+      "step": 19965500
+    },
+    {
+      "epoch": 98.92,
+      "learning_rate": 5.584340304097739e-07,
+      "loss": 1.7478,
+      "step": 19966000
+    },
+    {
+      "epoch": 98.92,
+      "learning_rate": 5.571954439836903e-07,
+      "loss": 1.7787,
+      "step": 19966500
+    },
+    {
+      "epoch": 98.92,
+      "learning_rate": 5.559593347304589e-07,
+      "loss": 1.7529,
+      "step": 19967000
+    },
+    {
+      "epoch": 98.93,
+      "learning_rate": 5.547207483043752e-07,
+      "loss": 1.735,
+      "step": 19967500
+    },
+    {
+      "epoch": 98.93,
+      "learning_rate": 5.534871162239959e-07,
+      "loss": 1.7507,
+      "step": 19968000
+    },
+    {
+      "epoch": 98.93,
+      "learning_rate": 5.522485297979122e-07,
+      "loss": 1.7483,
+      "step": 19968500
+    },
+    {
+      "epoch": 98.93,
+      "learning_rate": 5.510099433718287e-07,
+      "loss": 1.7543,
+      "step": 19969000
+    },
+    {
+      "epoch": 98.94,
+      "learning_rate": 5.49771356945745e-07,
+      "loss": 1.7729,
+      "step": 19969500
+    },
+    {
+      "epoch": 98.94,
+      "learning_rate": 5.485327705196613e-07,
+      "loss": 1.7646,
+      "step": 19970000
+    },
+    {
+      "epoch": 98.94,
+      "learning_rate": 5.472941840935778e-07,
+      "loss": 1.7822,
+      "step": 19970500
+    },
+    {
+      "epoch": 98.94,
+      "learning_rate": 5.460580748403463e-07,
+      "loss": 1.7671,
+      "step": 19971000
+    },
+    {
+      "epoch": 98.95,
+      "learning_rate": 5.448194884142626e-07,
+      "loss": 1.7559,
+      "step": 19971500
+    },
+    {
+      "epoch": 98.95,
+      "learning_rate": 5.43580901988179e-07,
+      "loss": 1.7706,
+      "step": 19972000
+    },
+    {
+      "epoch": 98.95,
+      "learning_rate": 5.423447927349474e-07,
+      "loss": 1.7557,
+      "step": 19972500
+    },
+    {
+      "epoch": 98.95,
+      "learning_rate": 5.411062063088638e-07,
+      "loss": 1.7689,
+      "step": 19973000
+    },
+    {
+      "epoch": 98.96,
+      "learning_rate": 5.398700970556324e-07,
+      "loss": 1.7753,
+      "step": 19973500
+    },
+    {
+      "epoch": 98.96,
+      "learning_rate": 5.386315106295487e-07,
+      "loss": 1.77,
+      "step": 19974000
+    },
+    {
+      "epoch": 98.96,
+      "learning_rate": 5.37392924203465e-07,
+      "loss": 1.7785,
+      "step": 19974500
+    },
+    {
+      "epoch": 98.96,
+      "learning_rate": 5.361543377773815e-07,
+      "loss": 1.7781,
+      "step": 19975000
+    },
+    {
+      "epoch": 98.97,
+      "learning_rate": 5.349157513512978e-07,
+      "loss": 1.7744,
+      "step": 19975500
+    },
+    {
+      "epoch": 98.97,
+      "learning_rate": 5.336771649252141e-07,
+      "loss": 1.7619,
+      "step": 19976000
+    },
+    {
+      "epoch": 98.97,
+      "learning_rate": 5.324385784991306e-07,
+      "loss": 1.7535,
+      "step": 19976500
+    },
+    {
+      "epoch": 98.97,
+      "learning_rate": 5.311999920730469e-07,
+      "loss": 1.7915,
+      "step": 19977000
+    },
+    {
+      "epoch": 98.98,
+      "learning_rate": 5.299614056469632e-07,
+      "loss": 1.777,
+      "step": 19977500
+    },
+    {
+      "epoch": 98.98,
+      "learning_rate": 5.287228192208796e-07,
+      "loss": 1.7638,
+      "step": 19978000
+    },
+    {
+      "epoch": 98.98,
+      "learning_rate": 5.27484232794796e-07,
+      "loss": 1.7584,
+      "step": 19978500
+    },
+    {
+      "epoch": 98.98,
+      "learning_rate": 5.262456463687123e-07,
+      "loss": 1.7711,
+      "step": 19979000
+    },
+    {
+      "epoch": 98.99,
+      "learning_rate": 5.250070599426287e-07,
+      "loss": 1.7525,
+      "step": 19979500
+    },
+    {
+      "epoch": 98.99,
+      "learning_rate": 5.237709506893973e-07,
+      "loss": 1.7502,
+      "step": 19980000
+    },
+    {
+      "epoch": 98.99,
+      "learning_rate": 5.225323642633136e-07,
+      "loss": 1.7675,
+      "step": 19980500
+    },
+    {
+      "epoch": 98.99,
+      "learning_rate": 5.212937778372299e-07,
+      "loss": 1.7547,
+      "step": 19981000
+    },
+    {
+      "epoch": 99.0,
+      "learning_rate": 5.200551914111463e-07,
+      "loss": 1.763,
+      "step": 19981500
+    },
+    {
+      "epoch": 99.0,
+      "learning_rate": 5.188166049850627e-07,
+      "loss": 1.7384,
+      "step": 19982000
+    },
+    {
+      "epoch": 99.0,
+      "eval_accuracy": 0.6875362239222096,
+      "eval_accuracy_mlm": 0.6489239224448309,
+      "eval_accuracy_nsp": 0.869461364376233,
+      "eval_loss": 2.321455478668213,
+      "eval_runtime": 147.1717,
+      "eval_samples_per_second": 1732.392,
+      "eval_steps_per_second": 72.188,
+      "step": 19982457
+    },
+    {
+      "epoch": 99.0,
+      "learning_rate": 5.175804957318312e-07,
+      "loss": 1.7502,
+      "step": 19982500
+    },
+    {
+      "epoch": 99.0,
+      "learning_rate": 5.163419093057476e-07,
+      "loss": 1.7359,
+      "step": 19983000
+    },
+    {
+      "epoch": 99.01,
+      "learning_rate": 5.151033228796639e-07,
+      "loss": 1.7604,
+      "step": 19983500
+    },
+    {
+      "epoch": 99.01,
+      "learning_rate": 5.138647364535803e-07,
+      "loss": 1.7585,
+      "step": 19984000
+    },
+    {
+      "epoch": 99.01,
+      "learning_rate": 5.126261500274967e-07,
+      "loss": 1.76,
+      "step": 19984500
+    },
+    {
+      "epoch": 99.01,
+      "learning_rate": 5.11387563601413e-07,
+      "loss": 1.7339,
+      "step": 19985000
+    },
+    {
+      "epoch": 99.02,
+      "learning_rate": 5.101489771753294e-07,
+      "loss": 1.7593,
+      "step": 19985500
+    },
+    {
+      "epoch": 99.02,
+      "learning_rate": 5.089103907492458e-07,
+      "loss": 1.7356,
+      "step": 19986000
+    },
+    {
+      "epoch": 99.02,
+      "learning_rate": 5.076742814960142e-07,
+      "loss": 1.7545,
+      "step": 19986500
+    },
+    {
+      "epoch": 99.02,
+      "learning_rate": 5.064356950699305e-07,
+      "loss": 1.7492,
+      "step": 19987000
+    },
+    {
+      "epoch": 99.02,
+      "learning_rate": 5.05197108643847e-07,
+      "loss": 1.7506,
+      "step": 19987500
+    },
+    {
+      "epoch": 99.03,
+      "learning_rate": 5.039585222177633e-07,
+      "loss": 1.7505,
+      "step": 19988000
+    },
+    {
+      "epoch": 99.03,
+      "learning_rate": 5.027199357916796e-07,
+      "loss": 1.744,
+      "step": 19988500
+    },
+    {
+      "epoch": 99.03,
+      "learning_rate": 5.014813493655961e-07,
+      "loss": 1.7645,
+      "step": 19989000
+    },
+    {
+      "epoch": 99.03,
+      "learning_rate": 5.002427629395124e-07,
+      "loss": 1.7681,
+      "step": 19989500
+    },
+    {
+      "epoch": 99.04,
+      "learning_rate": 4.990041765134287e-07,
+      "loss": 1.7629,
+      "step": 19990000
+    },
+    {
+      "epoch": 99.04,
+      "learning_rate": 4.977655900873451e-07,
+      "loss": 1.7545,
+      "step": 19990500
+    },
+    {
+      "epoch": 99.04,
+      "learning_rate": 4.965270036612615e-07,
+      "loss": 1.7557,
+      "step": 19991000
+    },
+    {
+      "epoch": 99.04,
+      "learning_rate": 4.952884172351778e-07,
+      "loss": 1.762,
+      "step": 19991500
+    },
+    {
+      "epoch": 99.05,
+      "learning_rate": 4.940523079819464e-07,
+      "loss": 1.7716,
+      "step": 19992000
+    },
+    {
+      "epoch": 99.05,
+      "learning_rate": 4.928137215558628e-07,
+      "loss": 1.7834,
+      "step": 19992500
+    },
+    {
+      "epoch": 99.05,
+      "learning_rate": 4.915751351297791e-07,
+      "loss": 1.7628,
+      "step": 19993000
+    },
+    {
+      "epoch": 99.05,
+      "learning_rate": 4.903365487036955e-07,
+      "loss": 1.769,
+      "step": 19993500
+    },
+    {
+      "epoch": 99.06,
+      "learning_rate": 4.890979622776118e-07,
+      "loss": 1.7895,
+      "step": 19994000
+    },
+    {
+      "epoch": 99.06,
+      "learning_rate": 4.878593758515282e-07,
+      "loss": 1.751,
+      "step": 19994500
+    },
+    {
+      "epoch": 99.06,
+      "learning_rate": 4.866207894254446e-07,
+      "loss": 1.7813,
+      "step": 19995000
+    },
+    {
+      "epoch": 99.06,
+      "learning_rate": 4.853822029993609e-07,
+      "loss": 1.7295,
+      "step": 19995500
+    },
+    {
+      "epoch": 99.07,
+      "learning_rate": 4.841436165732773e-07,
+      "loss": 1.7715,
+      "step": 19996000
+    },
+    {
+      "epoch": 99.07,
+      "learning_rate": 4.829075073200458e-07,
+      "loss": 1.7592,
+      "step": 19996500
+    },
+    {
+      "epoch": 99.07,
+      "learning_rate": 4.816689208939622e-07,
+      "loss": 1.7537,
+      "step": 19997000
+    },
+    {
+      "epoch": 99.07,
+      "learning_rate": 4.804352888135829e-07,
+      "loss": 1.7537,
+      "step": 19997500
+    },
+    {
+      "epoch": 99.08,
+      "learning_rate": 4.791967023874992e-07,
+      "loss": 1.7841,
+      "step": 19998000
+    },
+    {
+      "epoch": 99.08,
+      "learning_rate": 4.779581159614157e-07,
+      "loss": 1.7582,
+      "step": 19998500
+    },
+    {
+      "epoch": 99.08,
+      "learning_rate": 4.7671952953533195e-07,
+      "loss": 1.777,
+      "step": 19999000
+    },
+    {
+      "epoch": 99.08,
+      "learning_rate": 4.7548342028210044e-07,
+      "loss": 1.7739,
+      "step": 19999500
+    },
+    {
+      "epoch": 99.09,
+      "learning_rate": 4.7424483385601684e-07,
+      "loss": 1.7789,
+      "step": 20000000
+    },
+    {
+      "epoch": 99.09,
+      "learning_rate": 4.730062474299332e-07,
+      "loss": 1.7705,
+      "step": 20000500
+    },
+    {
+      "epoch": 99.09,
+      "learning_rate": 4.7176766100384953e-07,
+      "loss": 1.7505,
+      "step": 20001000
+    },
+    {
+      "epoch": 99.09,
+      "learning_rate": 4.7052907457776593e-07,
+      "loss": 1.7712,
+      "step": 20001500
+    },
+    {
+      "epoch": 99.1,
+      "learning_rate": 4.692929653245345e-07,
+      "loss": 1.761,
+      "step": 20002000
+    },
+    {
+      "epoch": 99.1,
+      "learning_rate": 4.6805437889845077e-07,
+      "loss": 1.7761,
+      "step": 20002500
+    },
+    {
+      "epoch": 99.1,
+      "learning_rate": 4.668157924723671e-07,
+      "loss": 1.7707,
+      "step": 20003000
+    },
+    {
+      "epoch": 99.1,
+      "learning_rate": 4.655772060462835e-07,
+      "loss": 1.7418,
+      "step": 20003500
+    },
+    {
+      "epoch": 99.11,
+      "learning_rate": 4.6434109679305206e-07,
+      "loss": 1.7611,
+      "step": 20004000
+    },
+    {
+      "epoch": 99.11,
+      "learning_rate": 4.6310251036696835e-07,
+      "loss": 1.7658,
+      "step": 20004500
+    },
+    {
+      "epoch": 99.11,
+      "learning_rate": 4.618639239408848e-07,
+      "loss": 1.7399,
+      "step": 20005000
+    },
+    {
+      "epoch": 99.11,
+      "learning_rate": 4.606253375148011e-07,
+      "loss": 1.7433,
+      "step": 20005500
+    },
+    {
+      "epoch": 99.12,
+      "learning_rate": 4.5938675108871744e-07,
+      "loss": 1.7513,
+      "step": 20006000
+    },
+    {
+      "epoch": 99.12,
+      "learning_rate": 4.581481646626339e-07,
+      "loss": 1.7238,
+      "step": 20006500
+    },
+    {
+      "epoch": 99.12,
+      "learning_rate": 4.569095782365502e-07,
+      "loss": 1.778,
+      "step": 20007000
+    },
+    {
+      "epoch": 99.12,
+      "learning_rate": 4.5567099181046653e-07,
+      "loss": 1.7623,
+      "step": 20007500
+    },
+    {
+      "epoch": 99.13,
+      "learning_rate": 4.5443240538438293e-07,
+      "loss": 1.7499,
+      "step": 20008000
+    },
+    {
+      "epoch": 99.13,
+      "learning_rate": 4.531962961311515e-07,
+      "loss": 1.7473,
+      "step": 20008500
+    },
+    {
+      "epoch": 99.13,
+      "learning_rate": 4.5195770970506777e-07,
+      "loss": 1.7626,
+      "step": 20009000
+    },
+    {
+      "epoch": 99.13,
+      "learning_rate": 4.5072160045183637e-07,
+      "loss": 1.7558,
+      "step": 20009500
+    },
+    {
+      "epoch": 99.14,
+      "learning_rate": 4.494830140257527e-07,
+      "loss": 1.788,
+      "step": 20010000
+    },
+    {
+      "epoch": 99.14,
+      "learning_rate": 4.4824442759966906e-07,
+      "loss": 1.7517,
+      "step": 20010500
+    },
+    {
+      "epoch": 99.14,
+      "learning_rate": 4.4700584117358546e-07,
+      "loss": 1.7667,
+      "step": 20011000
+    },
+    {
+      "epoch": 99.14,
+      "learning_rate": 4.457672547475018e-07,
+      "loss": 1.7572,
+      "step": 20011500
+    },
+    {
+      "epoch": 99.15,
+      "learning_rate": 4.4452866832141815e-07,
+      "loss": 1.7448,
+      "step": 20012000
+    },
+    {
+      "epoch": 99.15,
+      "learning_rate": 4.4329008189533455e-07,
+      "loss": 1.7616,
+      "step": 20012500
+    },
+    {
+      "epoch": 99.15,
+      "learning_rate": 4.420514954692509e-07,
+      "loss": 1.763,
+      "step": 20013000
+    },
+    {
+      "epoch": 99.15,
+      "learning_rate": 4.408129090431672e-07,
+      "loss": 1.7487,
+      "step": 20013500
+    },
+    {
+      "epoch": 99.16,
+      "learning_rate": 4.3957432261708364e-07,
+      "loss": 1.7589,
+      "step": 20014000
+    },
+    {
+      "epoch": 99.16,
+      "learning_rate": 4.3833821336385213e-07,
+      "loss": 1.7659,
+      "step": 20014500
+    },
+    {
+      "epoch": 99.16,
+      "learning_rate": 4.371021041106206e-07,
+      "loss": 1.7691,
+      "step": 20015000
+    },
+    {
+      "epoch": 99.16,
+      "learning_rate": 4.3586351768453697e-07,
+      "loss": 1.7606,
+      "step": 20015500
+    },
+    {
+      "epoch": 99.17,
+      "learning_rate": 4.3462493125845337e-07,
+      "loss": 1.7476,
+      "step": 20016000
+    },
+    {
+      "epoch": 99.17,
+      "learning_rate": 4.333863448323697e-07,
+      "loss": 1.78,
+      "step": 20016500
+    },
+    {
+      "epoch": 99.17,
+      "learning_rate": 4.3214775840628606e-07,
+      "loss": 1.7635,
+      "step": 20017000
+    },
+    {
+      "epoch": 99.17,
+      "learning_rate": 4.3091164915305466e-07,
+      "loss": 1.753,
+      "step": 20017500
+    },
+    {
+      "epoch": 99.18,
+      "learning_rate": 4.29673062726971e-07,
+      "loss": 1.7885,
+      "step": 20018000
+    },
+    {
+      "epoch": 99.18,
+      "learning_rate": 4.284344763008873e-07,
+      "loss": 1.7733,
+      "step": 20018500
+    },
+    {
+      "epoch": 99.18,
+      "learning_rate": 4.2719588987480375e-07,
+      "loss": 1.7718,
+      "step": 20019000
+    },
+    {
+      "epoch": 99.18,
+      "learning_rate": 4.2595730344872004e-07,
+      "loss": 1.7541,
+      "step": 20019500
+    },
+    {
+      "epoch": 99.19,
+      "learning_rate": 4.247187170226364e-07,
+      "loss": 1.7578,
+      "step": 20020000
+    },
+    {
+      "epoch": 99.19,
+      "learning_rate": 4.2348013059655284e-07,
+      "loss": 1.7662,
+      "step": 20020500
+    },
+    {
+      "epoch": 99.19,
+      "learning_rate": 4.2224154417046913e-07,
+      "loss": 1.7545,
+      "step": 20021000
+    },
+    {
+      "epoch": 99.19,
+      "learning_rate": 4.210054349172377e-07,
+      "loss": 1.7658,
+      "step": 20021500
+    },
+    {
+      "epoch": 99.2,
+      "learning_rate": 4.1976684849115407e-07,
+      "loss": 1.7709,
+      "step": 20022000
+    },
+    {
+      "epoch": 99.2,
+      "learning_rate": 4.185282620650704e-07,
+      "loss": 1.7873,
+      "step": 20022500
+    },
+    {
+      "epoch": 99.2,
+      "learning_rate": 4.172896756389867e-07,
+      "loss": 1.7605,
+      "step": 20023000
+    },
+    {
+      "epoch": 99.2,
+      "learning_rate": 4.160535663857553e-07,
+      "loss": 1.7724,
+      "step": 20023500
+    },
+    {
+      "epoch": 99.21,
+      "learning_rate": 4.1481497995967166e-07,
+      "loss": 1.7591,
+      "step": 20024000
+    },
+    {
+      "epoch": 99.21,
+      "learning_rate": 4.13576393533588e-07,
+      "loss": 1.7519,
+      "step": 20024500
+    },
+    {
+      "epoch": 99.21,
+      "learning_rate": 4.123378071075044e-07,
+      "loss": 1.7457,
+      "step": 20025000
+    },
+    {
+      "epoch": 99.21,
+      "learning_rate": 4.1109922068142075e-07,
+      "loss": 1.7402,
+      "step": 20025500
+    },
+    {
+      "epoch": 99.22,
+      "learning_rate": 4.0986311142818924e-07,
+      "loss": 1.7733,
+      "step": 20026000
+    },
+    {
+      "epoch": 99.22,
+      "learning_rate": 4.086245250021056e-07,
+      "loss": 1.7563,
+      "step": 20026500
+    },
+    {
+      "epoch": 99.22,
+      "learning_rate": 4.073884157488742e-07,
+      "loss": 1.7647,
+      "step": 20027000
+    },
+    {
+      "epoch": 99.22,
+      "learning_rate": 4.0614982932279053e-07,
+      "loss": 1.7521,
+      "step": 20027500
+    },
+    {
+      "epoch": 99.23,
+      "learning_rate": 4.049112428967068e-07,
+      "loss": 1.7564,
+      "step": 20028000
+    },
+    {
+      "epoch": 99.23,
+      "learning_rate": 4.0367265647062327e-07,
+      "loss": 1.7449,
+      "step": 20028500
+    },
+    {
+      "epoch": 99.23,
+      "learning_rate": 4.0243407004453956e-07,
+      "loss": 1.7851,
+      "step": 20029000
+    },
+    {
+      "epoch": 99.23,
+      "learning_rate": 4.011979607913081e-07,
+      "loss": 1.7702,
+      "step": 20029500
+    },
+    {
+      "epoch": 99.24,
+      "learning_rate": 3.999618515380767e-07,
+      "loss": 1.7508,
+      "step": 20030000
+    },
+    {
+      "epoch": 99.24,
+      "learning_rate": 3.98723265111993e-07,
+      "loss": 1.7612,
+      "step": 20030500
+    },
+    {
+      "epoch": 99.24,
+      "learning_rate": 3.9748467868590935e-07,
+      "loss": 1.7614,
+      "step": 20031000
+    },
+    {
+      "epoch": 99.24,
+      "learning_rate": 3.9624609225982574e-07,
+      "loss": 1.7661,
+      "step": 20031500
+    },
+    {
+      "epoch": 99.25,
+      "learning_rate": 3.950075058337421e-07,
+      "loss": 1.774,
+      "step": 20032000
+    },
+    {
+      "epoch": 99.25,
+      "learning_rate": 3.9376891940765844e-07,
+      "loss": 1.7541,
+      "step": 20032500
+    },
+    {
+      "epoch": 99.25,
+      "learning_rate": 3.9253033298157484e-07,
+      "loss": 1.7802,
+      "step": 20033000
+    },
+    {
+      "epoch": 99.25,
+      "learning_rate": 3.9129422372834333e-07,
+      "loss": 1.7884,
+      "step": 20033500
+    },
+    {
+      "epoch": 99.26,
+      "learning_rate": 3.9005563730225967e-07,
+      "loss": 1.7513,
+      "step": 20034000
+    },
+    {
+      "epoch": 99.26,
+      "learning_rate": 3.8881705087617607e-07,
+      "loss": 1.7693,
+      "step": 20034500
+    },
+    {
+      "epoch": 99.26,
+      "learning_rate": 3.875784644500924e-07,
+      "loss": 1.7877,
+      "step": 20035000
+    },
+    {
+      "epoch": 99.26,
+      "learning_rate": 3.8633987802400876e-07,
+      "loss": 1.7771,
+      "step": 20035500
+    },
+    {
+      "epoch": 99.27,
+      "learning_rate": 3.8510129159792516e-07,
+      "loss": 1.7684,
+      "step": 20036000
+    },
+    {
+      "epoch": 99.27,
+      "learning_rate": 3.838627051718415e-07,
+      "loss": 1.7438,
+      "step": 20036500
+    },
+    {
+      "epoch": 99.27,
+      "learning_rate": 3.8262411874575785e-07,
+      "loss": 1.7421,
+      "step": 20037000
+    },
+    {
+      "epoch": 99.27,
+      "learning_rate": 3.813855323196742e-07,
+      "loss": 1.7691,
+      "step": 20037500
+    },
+    {
+      "epoch": 99.28,
+      "learning_rate": 3.801469458935906e-07,
+      "loss": 1.7866,
+      "step": 20038000
+    },
+    {
+      "epoch": 99.28,
+      "learning_rate": 3.7890835946750694e-07,
+      "loss": 1.7576,
+      "step": 20038500
+    },
+    {
+      "epoch": 99.28,
+      "learning_rate": 3.776697730414233e-07,
+      "loss": 1.7428,
+      "step": 20039000
+    },
+    {
+      "epoch": 99.28,
+      "learning_rate": 3.764311866153397e-07,
+      "loss": 1.7614,
+      "step": 20039500
+    },
+    {
+      "epoch": 99.29,
+      "learning_rate": 3.751950773621082e-07,
+      "loss": 1.7815,
+      "step": 20040000
+    },
+    {
+      "epoch": 99.29,
+      "learning_rate": 3.739564909360246e-07,
+      "loss": 1.7932,
+      "step": 20040500
+    },
+    {
+      "epoch": 99.29,
+      "learning_rate": 3.7271790450994087e-07,
+      "loss": 1.7593,
+      "step": 20041000
+    },
+    {
+      "epoch": 99.29,
+      "learning_rate": 3.714817952567094e-07,
+      "loss": 1.7716,
+      "step": 20041500
+    },
+    {
+      "epoch": 99.29,
+      "learning_rate": 3.702432088306258e-07,
+      "loss": 1.7567,
+      "step": 20042000
+    },
+    {
+      "epoch": 99.3,
+      "learning_rate": 3.6900462240454216e-07,
+      "loss": 1.7753,
+      "step": 20042500
+    },
+    {
+      "epoch": 99.3,
+      "learning_rate": 3.677660359784585e-07,
+      "loss": 1.7546,
+      "step": 20043000
+    },
+    {
+      "epoch": 99.3,
+      "learning_rate": 3.665274495523749e-07,
+      "loss": 1.7611,
+      "step": 20043500
+    },
+    {
+      "epoch": 99.3,
+      "learning_rate": 3.6528886312629125e-07,
+      "loss": 1.7545,
+      "step": 20044000
+    },
+    {
+      "epoch": 99.31,
+      "learning_rate": 3.640502767002076e-07,
+      "loss": 1.7596,
+      "step": 20044500
+    },
+    {
+      "epoch": 99.31,
+      "learning_rate": 3.62811690274124e-07,
+      "loss": 1.764,
+      "step": 20045000
+    },
+    {
+      "epoch": 99.31,
+      "learning_rate": 3.615755810208925e-07,
+      "loss": 1.753,
+      "step": 20045500
+    },
+    {
+      "epoch": 99.31,
+      "learning_rate": 3.6033947176766103e-07,
+      "loss": 1.7552,
+      "step": 20046000
+    },
+    {
+      "epoch": 99.32,
+      "learning_rate": 3.591008853415774e-07,
+      "loss": 1.7547,
+      "step": 20046500
+    },
+    {
+      "epoch": 99.32,
+      "learning_rate": 3.578622989154937e-07,
+      "loss": 1.7551,
+      "step": 20047000
+    },
+    {
+      "epoch": 99.32,
+      "learning_rate": 3.566237124894101e-07,
+      "loss": 1.7625,
+      "step": 20047500
+    },
+    {
+      "epoch": 99.32,
+      "learning_rate": 3.553876032361786e-07,
+      "loss": 1.7741,
+      "step": 20048000
+    },
+    {
+      "epoch": 99.33,
+      "learning_rate": 3.54149016810095e-07,
+      "loss": 1.7803,
+      "step": 20048500
+    },
+    {
+      "epoch": 99.33,
+      "learning_rate": 3.529104303840113e-07,
+      "loss": 1.7555,
+      "step": 20049000
+    },
+    {
+      "epoch": 99.33,
+      "learning_rate": 3.516718439579277e-07,
+      "loss": 1.7856,
+      "step": 20049500
+    },
+    {
+      "epoch": 99.33,
+      "learning_rate": 3.5043573470469625e-07,
+      "loss": 1.7469,
+      "step": 20050000
+    },
+    {
+      "epoch": 99.34,
+      "learning_rate": 3.491971482786126e-07,
+      "loss": 1.7675,
+      "step": 20050500
+    },
+    {
+      "epoch": 99.34,
+      "learning_rate": 3.4795856185252894e-07,
+      "loss": 1.768,
+      "step": 20051000
+    },
+    {
+      "epoch": 99.34,
+      "learning_rate": 3.4671997542644534e-07,
+      "loss": 1.7626,
+      "step": 20051500
+    },
+    {
+      "epoch": 99.34,
+      "learning_rate": 3.454813890003617e-07,
+      "loss": 1.7625,
+      "step": 20052000
+    },
+    {
+      "epoch": 99.35,
+      "learning_rate": 3.4424280257427803e-07,
+      "loss": 1.7512,
+      "step": 20052500
+    },
+    {
+      "epoch": 99.35,
+      "learning_rate": 3.430066933210466e-07,
+      "loss": 1.7495,
+      "step": 20053000
+    },
+    {
+      "epoch": 99.35,
+      "learning_rate": 3.417681068949629e-07,
+      "loss": 1.7496,
+      "step": 20053500
+    },
+    {
+      "epoch": 99.35,
+      "learning_rate": 3.405295204688793e-07,
+      "loss": 1.7495,
+      "step": 20054000
+    },
+    {
+      "epoch": 99.36,
+      "learning_rate": 3.392909340427956e-07,
+      "loss": 1.7537,
+      "step": 20054500
+    },
+    {
+      "epoch": 99.36,
+      "learning_rate": 3.38052347616712e-07,
+      "loss": 1.7431,
+      "step": 20055000
+    },
+    {
+      "epoch": 99.36,
+      "learning_rate": 3.3681376119062836e-07,
+      "loss": 1.7594,
+      "step": 20055500
+    },
+    {
+      "epoch": 99.36,
+      "learning_rate": 3.355776519373969e-07,
+      "loss": 1.7632,
+      "step": 20056000
+    },
+    {
+      "epoch": 99.37,
+      "learning_rate": 3.3433906551131325e-07,
+      "loss": 1.7621,
+      "step": 20056500
+    },
+    {
+      "epoch": 99.37,
+      "learning_rate": 3.3310047908522965e-07,
+      "loss": 1.7439,
+      "step": 20057000
+    },
+    {
+      "epoch": 99.37,
+      "learning_rate": 3.31861892659146e-07,
+      "loss": 1.7869,
+      "step": 20057500
+    },
+    {
+      "epoch": 99.37,
+      "learning_rate": 3.3062330623306234e-07,
+      "loss": 1.7646,
+      "step": 20058000
+    },
+    {
+      "epoch": 99.38,
+      "learning_rate": 3.293871969798309e-07,
+      "loss": 1.7497,
+      "step": 20058500
+    },
+    {
+      "epoch": 99.38,
+      "learning_rate": 3.281510877265994e-07,
+      "loss": 1.7577,
+      "step": 20059000
+    },
+    {
+      "epoch": 99.38,
+      "learning_rate": 3.269125013005158e-07,
+      "loss": 1.7738,
+      "step": 20059500
+    },
+    {
+      "epoch": 99.38,
+      "learning_rate": 3.256739148744321e-07,
+      "loss": 1.7513,
+      "step": 20060000
+    },
+    {
+      "epoch": 99.39,
+      "learning_rate": 3.2443532844834847e-07,
+      "loss": 1.7542,
+      "step": 20060500
+    },
+    {
+      "epoch": 99.39,
+      "learning_rate": 3.2319674202226487e-07,
+      "loss": 1.764,
+      "step": 20061000
+    },
+    {
+      "epoch": 99.39,
+      "learning_rate": 3.219581555961812e-07,
+      "loss": 1.7361,
+      "step": 20061500
+    },
+    {
+      "epoch": 99.39,
+      "learning_rate": 3.2071956917009756e-07,
+      "loss": 1.7692,
+      "step": 20062000
+    },
+    {
+      "epoch": 99.4,
+      "learning_rate": 3.1948098274401396e-07,
+      "loss": 1.7767,
+      "step": 20062500
+    },
+    {
+      "epoch": 99.4,
+      "learning_rate": 3.1824487349078245e-07,
+      "loss": 1.7569,
+      "step": 20063000
+    },
+    {
+      "epoch": 99.4,
+      "learning_rate": 3.1700628706469885e-07,
+      "loss": 1.7694,
+      "step": 20063500
+    },
+    {
+      "epoch": 99.4,
+      "learning_rate": 3.157677006386152e-07,
+      "loss": 1.7717,
+      "step": 20064000
+    },
+    {
+      "epoch": 99.41,
+      "learning_rate": 3.1452911421253154e-07,
+      "loss": 1.7699,
+      "step": 20064500
+    },
+    {
+      "epoch": 99.41,
+      "learning_rate": 3.132905277864479e-07,
+      "loss": 1.7529,
+      "step": 20065000
+    },
+    {
+      "epoch": 99.41,
+      "learning_rate": 3.120519413603643e-07,
+      "loss": 1.7645,
+      "step": 20065500
+    },
+    {
+      "epoch": 99.41,
+      "learning_rate": 3.1081335493428063e-07,
+      "loss": 1.7648,
+      "step": 20066000
+    },
+    {
+      "epoch": 99.42,
+      "learning_rate": 3.09574768508197e-07,
+      "loss": 1.766,
+      "step": 20066500
+    },
+    {
+      "epoch": 99.42,
+      "learning_rate": 3.0833865925496547e-07,
+      "loss": 1.7797,
+      "step": 20067000
+    },
+    {
+      "epoch": 99.42,
+      "learning_rate": 3.0710007282888187e-07,
+      "loss": 1.7671,
+      "step": 20067500
+    },
+    {
+      "epoch": 99.42,
+      "learning_rate": 3.0586148640279826e-07,
+      "loss": 1.796,
+      "step": 20068000
+    },
+    {
+      "epoch": 99.43,
+      "learning_rate": 3.0462289997671456e-07,
+      "loss": 1.7632,
+      "step": 20068500
+    },
+    {
+      "epoch": 99.43,
+      "learning_rate": 3.0338431355063096e-07,
+      "loss": 1.7642,
+      "step": 20069000
+    },
+    {
+      "epoch": 99.43,
+      "learning_rate": 3.021482042973995e-07,
+      "loss": 1.7517,
+      "step": 20069500
+    },
+    {
+      "epoch": 99.43,
+      "learning_rate": 3.0090961787131585e-07,
+      "loss": 1.7633,
+      "step": 20070000
+    },
+    {
+      "epoch": 99.44,
+      "learning_rate": 2.996710314452322e-07,
+      "loss": 1.7589,
+      "step": 20070500
+    },
+    {
+      "epoch": 99.44,
+      "learning_rate": 2.984324450191486e-07,
+      "loss": 1.7413,
+      "step": 20071000
+    },
+    {
+      "epoch": 99.44,
+      "learning_rate": 2.9719385859306494e-07,
+      "loss": 1.75,
+      "step": 20071500
+    },
+    {
+      "epoch": 99.44,
+      "learning_rate": 2.959552721669813e-07,
+      "loss": 1.7844,
+      "step": 20072000
+    },
+    {
+      "epoch": 99.45,
+      "learning_rate": 2.947166857408977e-07,
+      "loss": 1.7742,
+      "step": 20072500
+    },
+    {
+      "epoch": 99.45,
+      "learning_rate": 2.93478099314814e-07,
+      "loss": 1.7607,
+      "step": 20073000
+    },
+    {
+      "epoch": 99.45,
+      "learning_rate": 2.9223951288873037e-07,
+      "loss": 1.7645,
+      "step": 20073500
+    },
+    {
+      "epoch": 99.45,
+      "learning_rate": 2.9100340363549886e-07,
+      "loss": 1.771,
+      "step": 20074000
+    },
+    {
+      "epoch": 99.46,
+      "learning_rate": 2.8976481720941526e-07,
+      "loss": 1.7646,
+      "step": 20074500
+    },
+    {
+      "epoch": 99.46,
+      "learning_rate": 2.885262307833316e-07,
+      "loss": 1.7641,
+      "step": 20075000
+    },
+    {
+      "epoch": 99.46,
+      "learning_rate": 2.8728764435724795e-07,
+      "loss": 1.7665,
+      "step": 20075500
+    },
+    {
+      "epoch": 99.46,
+      "learning_rate": 2.8604905793116435e-07,
+      "loss": 1.7667,
+      "step": 20076000
+    },
+    {
+      "epoch": 99.47,
+      "learning_rate": 2.848129486779329e-07,
+      "loss": 1.7599,
+      "step": 20076500
+    },
+    {
+      "epoch": 99.47,
+      "learning_rate": 2.835743622518492e-07,
+      "loss": 1.7873,
+      "step": 20077000
+    },
+    {
+      "epoch": 99.47,
+      "learning_rate": 2.823357758257656e-07,
+      "loss": 1.7583,
+      "step": 20077500
+    },
+    {
+      "epoch": 99.47,
+      "learning_rate": 2.81097189399682e-07,
+      "loss": 1.7618,
+      "step": 20078000
+    },
+    {
+      "epoch": 99.48,
+      "learning_rate": 2.798586029735983e-07,
+      "loss": 1.7548,
+      "step": 20078500
+    },
+    {
+      "epoch": 99.48,
+      "learning_rate": 2.786200165475147e-07,
+      "loss": 1.7596,
+      "step": 20079000
+    },
+    {
+      "epoch": 99.48,
+      "learning_rate": 2.77381430121431e-07,
+      "loss": 1.754,
+      "step": 20079500
+    },
+    {
+      "epoch": 99.48,
+      "learning_rate": 2.7614284369534737e-07,
+      "loss": 1.7708,
+      "step": 20080000
+    },
+    {
+      "epoch": 99.49,
+      "learning_rate": 2.7490425726926377e-07,
+      "loss": 1.7676,
+      "step": 20080500
+    },
+    {
+      "epoch": 99.49,
+      "learning_rate": 2.736656708431801e-07,
+      "loss": 1.7794,
+      "step": 20081000
+    },
+    {
+      "epoch": 99.49,
+      "learning_rate": 2.7242708441709646e-07,
+      "loss": 1.7516,
+      "step": 20081500
+    },
+    {
+      "epoch": 99.49,
+      "learning_rate": 2.711884979910128e-07,
+      "loss": 1.7753,
+      "step": 20082000
+    },
+    {
+      "epoch": 99.5,
+      "learning_rate": 2.699548659106335e-07,
+      "loss": 1.7582,
+      "step": 20082500
+    },
+    {
+      "epoch": 99.5,
+      "learning_rate": 2.687162794845499e-07,
+      "loss": 1.7626,
+      "step": 20083000
+    },
+    {
+      "epoch": 99.5,
+      "learning_rate": 2.6747769305846624e-07,
+      "loss": 1.7579,
+      "step": 20083500
+    },
+    {
+      "epoch": 99.5,
+      "learning_rate": 2.662391066323826e-07,
+      "loss": 1.7674,
+      "step": 20084000
+    },
+    {
+      "epoch": 99.51,
+      "learning_rate": 2.65000520206299e-07,
+      "loss": 1.7805,
+      "step": 20084500
+    },
+    {
+      "epoch": 99.51,
+      "learning_rate": 2.6376193378021533e-07,
+      "loss": 1.7448,
+      "step": 20085000
+    },
+    {
+      "epoch": 99.51,
+      "learning_rate": 2.625233473541317e-07,
+      "loss": 1.79,
+      "step": 20085500
+    },
+    {
+      "epoch": 99.51,
+      "learning_rate": 2.612847609280481e-07,
+      "loss": 1.7547,
+      "step": 20086000
+    },
+    {
+      "epoch": 99.52,
+      "learning_rate": 2.6004865167481657e-07,
+      "loss": 1.7662,
+      "step": 20086500
+    },
+    {
+      "epoch": 99.52,
+      "learning_rate": 2.588100652487329e-07,
+      "loss": 1.781,
+      "step": 20087000
+    },
+    {
+      "epoch": 99.52,
+      "learning_rate": 2.575714788226493e-07,
+      "loss": 1.764,
+      "step": 20087500
+    },
+    {
+      "epoch": 99.52,
+      "learning_rate": 2.5633289239656566e-07,
+      "loss": 1.7545,
+      "step": 20088000
+    },
+    {
+      "epoch": 99.53,
+      "learning_rate": 2.550967831433342e-07,
+      "loss": 1.7659,
+      "step": 20088500
+    },
+    {
+      "epoch": 99.53,
+      "learning_rate": 2.5385819671725055e-07,
+      "loss": 1.7731,
+      "step": 20089000
+    },
+    {
+      "epoch": 99.53,
+      "learning_rate": 2.526196102911669e-07,
+      "loss": 1.7792,
+      "step": 20089500
+    },
+    {
+      "epoch": 99.53,
+      "learning_rate": 2.513810238650833e-07,
+      "loss": 1.7788,
+      "step": 20090000
+    },
+    {
+      "epoch": 99.54,
+      "learning_rate": 2.5014243743899964e-07,
+      "loss": 1.7857,
+      "step": 20090500
+    },
+    {
+      "epoch": 99.54,
+      "learning_rate": 2.48903851012916e-07,
+      "loss": 1.7738,
+      "step": 20091000
+    },
+    {
+      "epoch": 99.54,
+      "learning_rate": 2.4766774175968453e-07,
+      "loss": 1.7782,
+      "step": 20091500
+    },
+    {
+      "epoch": 99.54,
+      "learning_rate": 2.464291553336009e-07,
+      "loss": 1.7558,
+      "step": 20092000
+    },
+    {
+      "epoch": 99.55,
+      "learning_rate": 2.451905689075172e-07,
+      "loss": 1.7793,
+      "step": 20092500
+    },
+    {
+      "epoch": 99.55,
+      "learning_rate": 2.439519824814336e-07,
+      "loss": 1.7905,
+      "step": 20093000
+    },
+    {
+      "epoch": 99.55,
+      "learning_rate": 2.427158732282021e-07,
+      "loss": 1.7464,
+      "step": 20093500
+    },
+    {
+      "epoch": 99.55,
+      "learning_rate": 2.414772868021185e-07,
+      "loss": 1.7328,
+      "step": 20094000
+    },
+    {
+      "epoch": 99.56,
+      "learning_rate": 2.4023870037603486e-07,
+      "loss": 1.7717,
+      "step": 20094500
+    },
+    {
+      "epoch": 99.56,
+      "learning_rate": 2.390001139499512e-07,
+      "loss": 1.7595,
+      "step": 20095000
+    },
+    {
+      "epoch": 99.56,
+      "learning_rate": 2.3776152752386758e-07,
+      "loss": 1.7994,
+      "step": 20095500
+    },
+    {
+      "epoch": 99.56,
+      "learning_rate": 2.3652294109778395e-07,
+      "loss": 1.7726,
+      "step": 20096000
+    },
+    {
+      "epoch": 99.57,
+      "learning_rate": 2.3528435467170027e-07,
+      "loss": 1.7683,
+      "step": 20096500
+    },
+    {
+      "epoch": 99.57,
+      "learning_rate": 2.3404576824561667e-07,
+      "loss": 1.7662,
+      "step": 20097000
+    },
+    {
+      "epoch": 99.57,
+      "learning_rate": 2.3280718181953304e-07,
+      "loss": 1.7738,
+      "step": 20097500
+    },
+    {
+      "epoch": 99.57,
+      "learning_rate": 2.3156859539344936e-07,
+      "loss": 1.7796,
+      "step": 20098000
+    },
+    {
+      "epoch": 99.57,
+      "learning_rate": 2.303324861402179e-07,
+      "loss": 1.7583,
+      "step": 20098500
+    },
+    {
+      "epoch": 99.58,
+      "learning_rate": 2.2909389971413425e-07,
+      "loss": 1.779,
+      "step": 20099000
+    },
+    {
+      "epoch": 99.58,
+      "learning_rate": 2.2785531328805062e-07,
+      "loss": 1.737,
+      "step": 20099500
+    },
+    {
+      "epoch": 99.58,
+      "learning_rate": 2.26616726861967e-07,
+      "loss": 1.7509,
+      "step": 20100000
+    },
+    {
+      "epoch": 99.58,
+      "learning_rate": 2.2537814043588331e-07,
+      "loss": 1.747,
+      "step": 20100500
+    },
+    {
+      "epoch": 99.59,
+      "learning_rate": 2.2414203118265188e-07,
+      "loss": 1.7728,
+      "step": 20101000
+    },
+    {
+      "epoch": 99.59,
+      "learning_rate": 2.2290344475656826e-07,
+      "loss": 1.7587,
+      "step": 20101500
+    },
+    {
+      "epoch": 99.59,
+      "learning_rate": 2.2166485833048458e-07,
+      "loss": 1.751,
+      "step": 20102000
+    },
+    {
+      "epoch": 99.59,
+      "learning_rate": 2.2042627190440095e-07,
+      "loss": 1.7498,
+      "step": 20102500
+    },
+    {
+      "epoch": 99.6,
+      "learning_rate": 2.1918768547831732e-07,
+      "loss": 1.7849,
+      "step": 20103000
+    },
+    {
+      "epoch": 99.6,
+      "learning_rate": 2.1794909905223367e-07,
+      "loss": 1.79,
+      "step": 20103500
+    },
+    {
+      "epoch": 99.6,
+      "learning_rate": 2.1671051262615004e-07,
+      "loss": 1.759,
+      "step": 20104000
+    },
+    {
+      "epoch": 99.6,
+      "learning_rate": 2.1547688054577073e-07,
+      "loss": 1.7795,
+      "step": 20104500
+    },
+    {
+      "epoch": 99.61,
+      "learning_rate": 2.142382941196871e-07,
+      "loss": 1.7455,
+      "step": 20105000
+    },
+    {
+      "epoch": 99.61,
+      "learning_rate": 2.1299970769360347e-07,
+      "loss": 1.7395,
+      "step": 20105500
+    },
+    {
+      "epoch": 99.61,
+      "learning_rate": 2.117611212675198e-07,
+      "loss": 1.7665,
+      "step": 20106000
+    },
+    {
+      "epoch": 99.61,
+      "learning_rate": 2.1052253484143617e-07,
+      "loss": 1.783,
+      "step": 20106500
+    },
+    {
+      "epoch": 99.62,
+      "learning_rate": 2.0928394841535254e-07,
+      "loss": 1.7773,
+      "step": 20107000
+    },
+    {
+      "epoch": 99.62,
+      "learning_rate": 2.0804536198926888e-07,
+      "loss": 1.773,
+      "step": 20107500
+    },
+    {
+      "epoch": 99.62,
+      "learning_rate": 2.0680677556318526e-07,
+      "loss": 1.7648,
+      "step": 20108000
+    },
+    {
+      "epoch": 99.62,
+      "learning_rate": 2.0556818913710163e-07,
+      "loss": 1.7763,
+      "step": 20108500
+    },
+    {
+      "epoch": 99.63,
+      "learning_rate": 2.0433207988387015e-07,
+      "loss": 1.7559,
+      "step": 20109000
+    },
+    {
+      "epoch": 99.63,
+      "learning_rate": 2.0309349345778652e-07,
+      "loss": 1.762,
+      "step": 20109500
+    },
+    {
+      "epoch": 99.63,
+      "learning_rate": 2.01857384204555e-07,
+      "loss": 1.7631,
+      "step": 20110000
+    },
+    {
+      "epoch": 99.63,
+      "learning_rate": 2.0061879777847138e-07,
+      "loss": 1.7684,
+      "step": 20110500
+    },
+    {
+      "epoch": 99.64,
+      "learning_rate": 1.9938021135238778e-07,
+      "loss": 1.766,
+      "step": 20111000
+    },
+    {
+      "epoch": 99.64,
+      "learning_rate": 1.981416249263041e-07,
+      "loss": 1.7544,
+      "step": 20111500
+    },
+    {
+      "epoch": 99.64,
+      "learning_rate": 1.9690303850022047e-07,
+      "loss": 1.78,
+      "step": 20112000
+    },
+    {
+      "epoch": 99.64,
+      "learning_rate": 1.9566445207413685e-07,
+      "loss": 1.7712,
+      "step": 20112500
+    },
+    {
+      "epoch": 99.65,
+      "learning_rate": 1.9442586564805322e-07,
+      "loss": 1.7714,
+      "step": 20113000
+    },
+    {
+      "epoch": 99.65,
+      "learning_rate": 1.9318975639482174e-07,
+      "loss": 1.7633,
+      "step": 20113500
+    },
+    {
+      "epoch": 99.65,
+      "learning_rate": 1.9195116996873808e-07,
+      "loss": 1.7561,
+      "step": 20114000
+    },
+    {
+      "epoch": 99.65,
+      "learning_rate": 1.9071258354265443e-07,
+      "loss": 1.7692,
+      "step": 20114500
+    },
+    {
+      "epoch": 99.66,
+      "learning_rate": 1.8947399711657083e-07,
+      "loss": 1.7868,
+      "step": 20115000
+    },
+    {
+      "epoch": 99.66,
+      "learning_rate": 1.8823541069048717e-07,
+      "loss": 1.7764,
+      "step": 20115500
+    },
+    {
+      "epoch": 99.66,
+      "learning_rate": 1.8699682426440352e-07,
+      "loss": 1.7471,
+      "step": 20116000
+    },
+    {
+      "epoch": 99.66,
+      "learning_rate": 1.857582378383199e-07,
+      "loss": 1.7777,
+      "step": 20116500
+    },
+    {
+      "epoch": 99.67,
+      "learning_rate": 1.8451965141223626e-07,
+      "loss": 1.7683,
+      "step": 20117000
+    },
+    {
+      "epoch": 99.67,
+      "learning_rate": 1.832810649861526e-07,
+      "loss": 1.7457,
+      "step": 20117500
+    },
+    {
+      "epoch": 99.67,
+      "learning_rate": 1.8204247856006898e-07,
+      "loss": 1.7618,
+      "step": 20118000
+    },
+    {
+      "epoch": 99.67,
+      "learning_rate": 1.8080389213398533e-07,
+      "loss": 1.7755,
+      "step": 20118500
+    },
+    {
+      "epoch": 99.68,
+      "learning_rate": 1.795653057079017e-07,
+      "loss": 1.7782,
+      "step": 20119000
+    },
+    {
+      "epoch": 99.68,
+      "learning_rate": 1.7832671928181807e-07,
+      "loss": 1.7722,
+      "step": 20119500
+    },
+    {
+      "epoch": 99.68,
+      "learning_rate": 1.770906100285866e-07,
+      "loss": 1.7751,
+      "step": 20120000
+    },
+    {
+      "epoch": 99.68,
+      "learning_rate": 1.7585202360250294e-07,
+      "loss": 1.777,
+      "step": 20120500
+    },
+    {
+      "epoch": 99.69,
+      "learning_rate": 1.746134371764193e-07,
+      "loss": 1.776,
+      "step": 20121000
+    },
+    {
+      "epoch": 99.69,
+      "learning_rate": 1.7337732792318783e-07,
+      "loss": 1.7826,
+      "step": 20121500
+    },
+    {
+      "epoch": 99.69,
+      "learning_rate": 1.721387414971042e-07,
+      "loss": 1.7823,
+      "step": 20122000
+    },
+    {
+      "epoch": 99.69,
+      "learning_rate": 1.7090015507102054e-07,
+      "loss": 1.7658,
+      "step": 20122500
+    },
+    {
+      "epoch": 99.7,
+      "learning_rate": 1.6966156864493692e-07,
+      "loss": 1.8001,
+      "step": 20123000
+    },
+    {
+      "epoch": 99.7,
+      "learning_rate": 1.684229822188533e-07,
+      "loss": 1.7756,
+      "step": 20123500
+    },
+    {
+      "epoch": 99.7,
+      "learning_rate": 1.6718439579276963e-07,
+      "loss": 1.7707,
+      "step": 20124000
+    },
+    {
+      "epoch": 99.7,
+      "learning_rate": 1.6594580936668598e-07,
+      "loss": 1.7665,
+      "step": 20124500
+    },
+    {
+      "epoch": 99.71,
+      "learning_rate": 1.6470970011345453e-07,
+      "loss": 1.757,
+      "step": 20125000
+    },
+    {
+      "epoch": 99.71,
+      "learning_rate": 1.634711136873709e-07,
+      "loss": 1.7583,
+      "step": 20125500
+    },
+    {
+      "epoch": 99.71,
+      "learning_rate": 1.6223252726128724e-07,
+      "loss": 1.801,
+      "step": 20126000
+    },
+    {
+      "epoch": 99.71,
+      "learning_rate": 1.609939408352036e-07,
+      "loss": 1.7775,
+      "step": 20126500
+    },
+    {
+      "epoch": 99.72,
+      "learning_rate": 1.5975535440911996e-07,
+      "loss": 1.7803,
+      "step": 20127000
+    },
+    {
+      "epoch": 99.72,
+      "learning_rate": 1.5851676798303633e-07,
+      "loss": 1.7802,
+      "step": 20127500
+    },
+    {
+      "epoch": 99.72,
+      "learning_rate": 1.5727818155695268e-07,
+      "loss": 1.7662,
+      "step": 20128000
+    },
+    {
+      "epoch": 99.72,
+      "learning_rate": 1.5603959513086905e-07,
+      "loss": 1.7612,
+      "step": 20128500
+    },
+    {
+      "epoch": 99.73,
+      "learning_rate": 1.548010087047854e-07,
+      "loss": 1.7407,
+      "step": 20129000
+    },
+    {
+      "epoch": 99.73,
+      "learning_rate": 1.5356489945155394e-07,
+      "loss": 1.758,
+      "step": 20129500
+    },
+    {
+      "epoch": 99.73,
+      "learning_rate": 1.5232879019832246e-07,
+      "loss": 1.7757,
+      "step": 20130000
+    },
+    {
+      "epoch": 99.73,
+      "learning_rate": 1.510902037722388e-07,
+      "loss": 1.7679,
+      "step": 20130500
+    },
+    {
+      "epoch": 99.74,
+      "learning_rate": 1.4985161734615518e-07,
+      "loss": 1.79,
+      "step": 20131000
+    },
+    {
+      "epoch": 99.74,
+      "learning_rate": 1.4861303092007155e-07,
+      "loss": 1.7687,
+      "step": 20131500
+    },
+    {
+      "epoch": 99.74,
+      "learning_rate": 1.473744444939879e-07,
+      "loss": 1.758,
+      "step": 20132000
+    },
+    {
+      "epoch": 99.74,
+      "learning_rate": 1.4613585806790427e-07,
+      "loss": 1.7623,
+      "step": 20132500
+    },
+    {
+      "epoch": 99.75,
+      "learning_rate": 1.4490222598752496e-07,
+      "loss": 1.7405,
+      "step": 20133000
+    },
+    {
+      "epoch": 99.75,
+      "learning_rate": 1.4366363956144133e-07,
+      "loss": 1.757,
+      "step": 20133500
+    },
+    {
+      "epoch": 99.75,
+      "learning_rate": 1.4242505313535768e-07,
+      "loss": 1.7806,
+      "step": 20134000
+    },
+    {
+      "epoch": 99.75,
+      "learning_rate": 1.4118646670927402e-07,
+      "loss": 1.7711,
+      "step": 20134500
+    },
+    {
+      "epoch": 99.76,
+      "learning_rate": 1.3994788028319042e-07,
+      "loss": 1.7627,
+      "step": 20135000
+    },
+    {
+      "epoch": 99.76,
+      "learning_rate": 1.3870929385710677e-07,
+      "loss": 1.7867,
+      "step": 20135500
+    },
+    {
+      "epoch": 99.76,
+      "learning_rate": 1.3747070743102311e-07,
+      "loss": 1.7448,
+      "step": 20136000
+    },
+    {
+      "epoch": 99.76,
+      "learning_rate": 1.3623212100493949e-07,
+      "loss": 1.7571,
+      "step": 20136500
+    },
+    {
+      "epoch": 99.77,
+      "learning_rate": 1.3499353457885586e-07,
+      "loss": 1.7337,
+      "step": 20137000
+    },
+    {
+      "epoch": 99.77,
+      "learning_rate": 1.337549481527722e-07,
+      "loss": 1.7697,
+      "step": 20137500
+    },
+    {
+      "epoch": 99.77,
+      "learning_rate": 1.3251636172668858e-07,
+      "loss": 1.773,
+      "step": 20138000
+    },
+    {
+      "epoch": 99.77,
+      "learning_rate": 1.3127777530060492e-07,
+      "loss": 1.7551,
+      "step": 20138500
+    },
+    {
+      "epoch": 99.78,
+      "learning_rate": 1.300391888745213e-07,
+      "loss": 1.7702,
+      "step": 20139000
+    },
+    {
+      "epoch": 99.78,
+      "learning_rate": 1.2880307962128984e-07,
+      "loss": 1.7797,
+      "step": 20139500
+    },
+    {
+      "epoch": 99.78,
+      "learning_rate": 1.2756449319520619e-07,
+      "loss": 1.7749,
+      "step": 20140000
+    },
+    {
+      "epoch": 99.78,
+      "learning_rate": 1.2632590676912253e-07,
+      "loss": 1.7616,
+      "step": 20140500
+    },
+    {
+      "epoch": 99.79,
+      "learning_rate": 1.250873203430389e-07,
+      "loss": 1.7777,
+      "step": 20141000
+    },
+    {
+      "epoch": 99.79,
+      "learning_rate": 1.2384873391695528e-07,
+      "loss": 1.7897,
+      "step": 20141500
+    },
+    {
+      "epoch": 99.79,
+      "learning_rate": 1.2261014749087162e-07,
+      "loss": 1.7495,
+      "step": 20142000
+    },
+    {
+      "epoch": 99.79,
+      "learning_rate": 1.2137403823764014e-07,
+      "loss": 1.7902,
+      "step": 20142500
+    },
+    {
+      "epoch": 99.8,
+      "learning_rate": 1.2014040615726086e-07,
+      "loss": 1.7714,
+      "step": 20143000
+    },
+    {
+      "epoch": 99.8,
+      "learning_rate": 1.189018197311772e-07,
+      "loss": 1.7471,
+      "step": 20143500
+    },
+    {
+      "epoch": 99.8,
+      "learning_rate": 1.1766323330509358e-07,
+      "loss": 1.781,
+      "step": 20144000
+    },
+    {
+      "epoch": 99.8,
+      "learning_rate": 1.1642464687900993e-07,
+      "loss": 1.7569,
+      "step": 20144500
+    },
+    {
+      "epoch": 99.81,
+      "learning_rate": 1.1518606045292628e-07,
+      "loss": 1.7655,
+      "step": 20145000
+    },
+    {
+      "epoch": 99.81,
+      "learning_rate": 1.1394747402684267e-07,
+      "loss": 1.7711,
+      "step": 20145500
+    },
+    {
+      "epoch": 99.81,
+      "learning_rate": 1.1270888760075901e-07,
+      "loss": 1.7743,
+      "step": 20146000
+    },
+    {
+      "epoch": 99.81,
+      "learning_rate": 1.1147277834752754e-07,
+      "loss": 1.7382,
+      "step": 20146500
+    },
+    {
+      "epoch": 99.82,
+      "learning_rate": 1.1023419192144389e-07,
+      "loss": 1.7757,
+      "step": 20147000
+    },
+    {
+      "epoch": 99.82,
+      "learning_rate": 1.0899560549536027e-07,
+      "loss": 1.8014,
+      "step": 20147500
+    },
+    {
+      "epoch": 99.82,
+      "learning_rate": 1.0775701906927662e-07,
+      "loss": 1.7693,
+      "step": 20148000
+    },
+    {
+      "epoch": 99.82,
+      "learning_rate": 1.0651843264319298e-07,
+      "loss": 1.7625,
+      "step": 20148500
+    },
+    {
+      "epoch": 99.83,
+      "learning_rate": 1.0527984621710935e-07,
+      "loss": 1.7743,
+      "step": 20149000
+    },
+    {
+      "epoch": 99.83,
+      "learning_rate": 1.0404125979102571e-07,
+      "loss": 1.774,
+      "step": 20149500
+    },
+    {
+      "epoch": 99.83,
+      "learning_rate": 1.0280267336494206e-07,
+      "loss": 1.7929,
+      "step": 20150000
+    },
+    {
+      "epoch": 99.83,
+      "learning_rate": 1.0156408693885843e-07,
+      "loss": 1.7587,
+      "step": 20150500
+    },
+    {
+      "epoch": 99.84,
+      "learning_rate": 1.0032550051277479e-07,
+      "loss": 1.7666,
+      "step": 20151000
+    },
+    {
+      "epoch": 99.84,
+      "learning_rate": 9.908691408669115e-08,
+      "loss": 1.7434,
+      "step": 20151500
+    },
+    {
+      "epoch": 99.84,
+      "learning_rate": 9.784832766060752e-08,
+      "loss": 1.7647,
+      "step": 20152000
+    },
+    {
+      "epoch": 99.84,
+      "learning_rate": 9.660974123452386e-08,
+      "loss": 1.7685,
+      "step": 20152500
+    },
+    {
+      "epoch": 99.84,
+      "learning_rate": 9.537610915414456e-08,
+      "loss": 1.765,
+      "step": 20153000
+    },
+    {
+      "epoch": 99.85,
+      "learning_rate": 9.413752272806093e-08,
+      "loss": 1.7778,
+      "step": 20153500
+    },
+    {
+      "epoch": 99.85,
+      "learning_rate": 9.289893630197729e-08,
+      "loss": 1.7707,
+      "step": 20154000
+    },
+    {
+      "epoch": 99.85,
+      "learning_rate": 9.166034987589365e-08,
+      "loss": 1.7639,
+      "step": 20154500
+    },
+    {
+      "epoch": 99.85,
+      "learning_rate": 9.042176344981e-08,
+      "loss": 1.77,
+      "step": 20155000
+    },
+    {
+      "epoch": 99.86,
+      "learning_rate": 8.918317702372636e-08,
+      "loss": 1.7732,
+      "step": 20155500
+    },
+    {
+      "epoch": 99.86,
+      "learning_rate": 8.794459059764272e-08,
+      "loss": 1.7903,
+      "step": 20156000
+    },
+    {
+      "epoch": 99.86,
+      "learning_rate": 8.670848134441126e-08,
+      "loss": 1.7838,
+      "step": 20156500
+    },
+    {
+      "epoch": 99.86,
+      "learning_rate": 8.546989491832761e-08,
+      "loss": 1.7762,
+      "step": 20157000
+    },
+    {
+      "epoch": 99.87,
+      "learning_rate": 8.423130849224399e-08,
+      "loss": 1.7784,
+      "step": 20157500
+    },
+    {
+      "epoch": 99.87,
+      "learning_rate": 8.299272206616033e-08,
+      "loss": 1.7695,
+      "step": 20158000
+    },
+    {
+      "epoch": 99.87,
+      "learning_rate": 8.17541356400767e-08,
+      "loss": 1.7672,
+      "step": 20158500
+    },
+    {
+      "epoch": 99.87,
+      "learning_rate": 8.051554921399306e-08,
+      "loss": 1.7813,
+      "step": 20159000
+    },
+    {
+      "epoch": 99.88,
+      "learning_rate": 7.927696278790942e-08,
+      "loss": 1.7601,
+      "step": 20159500
+    },
+    {
+      "epoch": 99.88,
+      "learning_rate": 7.803837636182578e-08,
+      "loss": 1.7801,
+      "step": 20160000
+    },
+    {
+      "epoch": 99.88,
+      "learning_rate": 7.680226710859431e-08,
+      "loss": 1.7729,
+      "step": 20160500
+    },
+    {
+      "epoch": 99.88,
+      "learning_rate": 7.556368068251067e-08,
+      "loss": 1.7913,
+      "step": 20161000
+    },
+    {
+      "epoch": 99.89,
+      "learning_rate": 7.432509425642703e-08,
+      "loss": 1.7757,
+      "step": 20161500
+    },
+    {
+      "epoch": 99.89,
+      "learning_rate": 7.308898500319555e-08,
+      "loss": 1.7638,
+      "step": 20162000
+    },
+    {
+      "epoch": 99.89,
+      "learning_rate": 7.185287574996408e-08,
+      "loss": 1.7614,
+      "step": 20162500
+    },
+    {
+      "epoch": 99.89,
+      "learning_rate": 7.061428932388044e-08,
+      "loss": 1.7469,
+      "step": 20163000
+    },
+    {
+      "epoch": 99.9,
+      "learning_rate": 6.937570289779681e-08,
+      "loss": 1.7518,
+      "step": 20163500
+    },
+    {
+      "epoch": 99.9,
+      "learning_rate": 6.813711647171316e-08,
+      "loss": 1.7711,
+      "step": 20164000
+    },
+    {
+      "epoch": 99.9,
+      "learning_rate": 6.690100721848169e-08,
+      "loss": 1.7831,
+      "step": 20164500
+    },
+    {
+      "epoch": 99.9,
+      "learning_rate": 6.566489796525022e-08,
+      "loss": 1.8008,
+      "step": 20165000
+    },
+    {
+      "epoch": 99.91,
+      "learning_rate": 6.442631153916658e-08,
+      "loss": 1.754,
+      "step": 20165500
+    },
+    {
+      "epoch": 99.91,
+      "learning_rate": 6.318772511308295e-08,
+      "loss": 1.801,
+      "step": 20166000
+    },
+    {
+      "epoch": 99.91,
+      "learning_rate": 6.19491386869993e-08,
+      "loss": 1.7811,
+      "step": 20166500
+    },
+    {
+      "epoch": 99.91,
+      "learning_rate": 6.071055226091566e-08,
+      "loss": 1.7753,
+      "step": 20167000
+    },
+    {
+      "epoch": 99.92,
+      "learning_rate": 5.947196583483203e-08,
+      "loss": 1.7909,
+      "step": 20167500
+    },
+    {
+      "epoch": 99.92,
+      "learning_rate": 5.823337940874838e-08,
+      "loss": 1.752,
+      "step": 20168000
+    },
+    {
+      "epoch": 99.92,
+      "learning_rate": 5.699727015551691e-08,
+      "loss": 1.7863,
+      "step": 20168500
+    },
+    {
+      "epoch": 99.92,
+      "learning_rate": 5.575868372943327e-08,
+      "loss": 1.7745,
+      "step": 20169000
+    },
+    {
+      "epoch": 99.93,
+      "learning_rate": 5.452009730334964e-08,
+      "loss": 1.7339,
+      "step": 20169500
+    },
+    {
+      "epoch": 99.93,
+      "learning_rate": 5.328151087726599e-08,
+      "loss": 1.7639,
+      "step": 20170000
+    },
+    {
+      "epoch": 99.93,
+      "learning_rate": 5.204292445118236e-08,
+      "loss": 1.7692,
+      "step": 20170500
+    },
+    {
+      "epoch": 99.93,
+      "learning_rate": 5.080433802509872e-08,
+      "loss": 1.7791,
+      "step": 20171000
+    },
+    {
+      "epoch": 99.94,
+      "learning_rate": 4.9565751599015075e-08,
+      "loss": 1.7527,
+      "step": 20171500
+    },
+    {
+      "epoch": 99.94,
+      "learning_rate": 4.832716517293144e-08,
+      "loss": 1.7679,
+      "step": 20172000
+    },
+    {
+      "epoch": 99.94,
+      "learning_rate": 4.70885787468478e-08,
+      "loss": 1.7747,
+      "step": 20172500
+    },
+    {
+      "epoch": 99.94,
+      "learning_rate": 4.584999232076416e-08,
+      "loss": 1.762,
+      "step": 20173000
+    },
+    {
+      "epoch": 99.95,
+      "learning_rate": 4.461140589468052e-08,
+      "loss": 1.7717,
+      "step": 20173500
+    },
+    {
+      "epoch": 99.95,
+      "learning_rate": 4.337529664144905e-08,
+      "loss": 1.7703,
+      "step": 20174000
+    },
+    {
+      "epoch": 99.95,
+      "learning_rate": 4.2136710215365415e-08,
+      "loss": 1.7513,
+      "step": 20174500
+    },
+    {
+      "epoch": 99.95,
+      "learning_rate": 4.0898123789281774e-08,
+      "loss": 1.7769,
+      "step": 20175000
+    },
+    {
+      "epoch": 99.96,
+      "learning_rate": 3.965953736319813e-08,
+      "loss": 1.7845,
+      "step": 20175500
+    },
+    {
+      "epoch": 99.96,
+      "learning_rate": 3.842095093711449e-08,
+      "loss": 1.7709,
+      "step": 20176000
+    },
+    {
+      "epoch": 99.96,
+      "learning_rate": 3.718236451103085e-08,
+      "loss": 1.7733,
+      "step": 20176500
+    },
+    {
+      "epoch": 99.96,
+      "learning_rate": 3.594377808494721e-08,
+      "loss": 1.7737,
+      "step": 20177000
+    },
+    {
+      "epoch": 99.97,
+      "learning_rate": 3.4705191658863576e-08,
+      "loss": 1.7951,
+      "step": 20177500
+    },
+    {
+      "epoch": 99.97,
+      "learning_rate": 3.3466605232779935e-08,
+      "loss": 1.7738,
+      "step": 20178000
+    },
+    {
+      "epoch": 99.97,
+      "learning_rate": 3.223049597954846e-08,
+      "loss": 1.7832,
+      "step": 20178500
+    },
+    {
+      "epoch": 99.97,
+      "learning_rate": 3.099190955346482e-08,
+      "loss": 1.7636,
+      "step": 20179000
+    },
+    {
+      "epoch": 99.98,
+      "learning_rate": 2.9753323127381184e-08,
+      "loss": 1.7692,
+      "step": 20179500
+    },
+    {
+      "epoch": 99.98,
+      "learning_rate": 2.8514736701297543e-08,
+      "loss": 1.7525,
+      "step": 20180000
+    },
+    {
+      "epoch": 99.98,
+      "learning_rate": 2.7276150275213902e-08,
+      "loss": 1.7682,
+      "step": 20180500
+    },
+    {
+      "epoch": 99.98,
+      "learning_rate": 2.6037563849130268e-08,
+      "loss": 1.7772,
+      "step": 20181000
+    },
+    {
+      "epoch": 99.99,
+      "learning_rate": 2.4801454595898796e-08,
+      "loss": 1.7848,
+      "step": 20181500
+    },
+    {
+      "epoch": 99.99,
+      "learning_rate": 2.356534534266732e-08,
+      "loss": 1.7652,
+      "step": 20182000
+    },
+    {
+      "epoch": 99.99,
+      "learning_rate": 2.232675891658368e-08,
+      "loss": 1.758,
+      "step": 20182500
+    },
+    {
+      "epoch": 99.99,
+      "learning_rate": 2.1088172490500043e-08,
+      "loss": 1.7551,
+      "step": 20183000
+    },
+    {
+      "epoch": 100.0,
+      "learning_rate": 1.9849586064416405e-08,
+      "loss": 1.7859,
+      "step": 20183500
+    },
+    {
+      "epoch": 100.0,
+      "learning_rate": 1.8610999638332764e-08,
+      "loss": 1.785,
+      "step": 20184000
+    },
+    {
+      "epoch": 100.0,
+      "eval_accuracy": 0.6880450819643994,
+      "eval_accuracy_mlm": 0.6495895431712034,
+      "eval_accuracy_nsp": 0.869280943210477,
+      "eval_loss": 2.31811785697937,
+      "eval_runtime": 147.0596,
+      "eval_samples_per_second": 1733.712,
+      "eval_steps_per_second": 72.243,
+      "step": 20184300
     }
   ],
   "max_steps": 20184300,
   "num_train_epochs": 100,
-  "total_flos": 2.3780546655767814e+19,
+  "total_flos": 2.6132554735491883e+19,
   "trial_name": null,
   "trial_params": null
 }