diff --git "a/last-checkpoint/trainer_state.json" "b/last-checkpoint/trainer_state.json"
--- "a/last-checkpoint/trainer_state.json"
+++ "b/last-checkpoint/trainer_state.json"
@@ -1,9 +1,9 @@
 {
-  "best_metric": 0.5081,
-  "best_model_checkpoint": "/kaggle/tmp/amr-tst-indo/AMRBART-id/fine-tune/../outputs/mbart-en-id-smaller-fted/checkpoint-19347",
-  "epoch": 4.999870784339062,
+  "best_metric": 11.3364,
+  "best_model_checkpoint": "/kaggle/tmp/amr-tst-indo/AMRBART-id/fine-tune/../outputs/mbart-en-id-smaller-fted/checkpoint-38692",
+  "epoch": 9.999870784339063,
   "eval_steps": 500,
-  "global_step": 19347,
+  "global_step": 38692,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -5865,6 +5865,5858 @@
       "eval_samples_per_second": 1.019,
       "eval_steps_per_second": 0.51,
       "step": 19347
+    },
+    {
+      "epoch": 5.0038764698281435,
+      "learning_rate": 1.6894852845844678e-07,
+      "loss": 2.6522,
+      "step": 19360
+    },
+    {
+      "epoch": 5.009045096265667,
+      "learning_rate": 1.6891611564890445e-07,
+      "loss": 2.6008,
+      "step": 19380
+    },
+    {
+      "epoch": 5.014213722703191,
+      "learning_rate": 1.688837028393621e-07,
+      "loss": 2.6294,
+      "step": 19400
+    },
+    {
+      "epoch": 5.019382349140716,
+      "learning_rate": 1.6885129002981977e-07,
+      "loss": 2.6519,
+      "step": 19420
+    },
+    {
+      "epoch": 5.02455097557824,
+      "learning_rate": 1.6881887722027743e-07,
+      "loss": 2.6537,
+      "step": 19440
+    },
+    {
+      "epoch": 5.029719602015764,
+      "learning_rate": 1.6878646441073512e-07,
+      "loss": 2.6256,
+      "step": 19460
+    },
+    {
+      "epoch": 5.034888228453289,
+      "learning_rate": 1.6875405160119278e-07,
+      "loss": 2.6721,
+      "step": 19480
+    },
+    {
+      "epoch": 5.040056854890813,
+      "learning_rate": 1.6872163879165044e-07,
+      "loss": 2.6894,
+      "step": 19500
+    },
+    {
+      "epoch": 5.045225481328337,
+      "learning_rate": 1.6868922598210813e-07,
+      "loss": 2.6702,
+      "step": 19520
+    },
+    {
+      "epoch": 5.050394107765861,
+      "learning_rate": 1.686568131725658e-07,
+      "loss": 2.7041,
+      "step": 19540
+    },
+    {
+      "epoch": 5.055562734203385,
+      "learning_rate": 1.6862440036302346e-07,
+      "loss": 2.7243,
+      "step": 19560
+    },
+    {
+      "epoch": 5.06073136064091,
+      "learning_rate": 1.6859198755348114e-07,
+      "loss": 2.7082,
+      "step": 19580
+    },
+    {
+      "epoch": 5.065899987078434,
+      "learning_rate": 1.685595747439388e-07,
+      "loss": 2.6755,
+      "step": 19600
+    },
+    {
+      "epoch": 5.071068613515958,
+      "learning_rate": 1.6852716193439647e-07,
+      "loss": 2.6075,
+      "step": 19620
+    },
+    {
+      "epoch": 5.076237239953483,
+      "learning_rate": 1.6849474912485413e-07,
+      "loss": 2.6402,
+      "step": 19640
+    },
+    {
+      "epoch": 5.081405866391006,
+      "learning_rate": 1.684623363153118e-07,
+      "loss": 2.6928,
+      "step": 19660
+    },
+    {
+      "epoch": 5.086574492828531,
+      "learning_rate": 1.6842992350576948e-07,
+      "loss": 2.6689,
+      "step": 19680
+    },
+    {
+      "epoch": 5.091743119266055,
+      "learning_rate": 1.6839751069622714e-07,
+      "loss": 2.6848,
+      "step": 19700
+    },
+    {
+      "epoch": 5.096911745703579,
+      "learning_rate": 1.683650978866848e-07,
+      "loss": 2.6315,
+      "step": 19720
+    },
+    {
+      "epoch": 5.102080372141104,
+      "learning_rate": 1.683326850771425e-07,
+      "loss": 2.6936,
+      "step": 19740
+    },
+    {
+      "epoch": 5.107248998578628,
+      "learning_rate": 1.6830027226760015e-07,
+      "loss": 2.6354,
+      "step": 19760
+    },
+    {
+      "epoch": 5.112417625016152,
+      "learning_rate": 1.6826785945805782e-07,
+      "loss": 2.6376,
+      "step": 19780
+    },
+    {
+      "epoch": 5.117586251453676,
+      "learning_rate": 1.6823544664851548e-07,
+      "loss": 2.7595,
+      "step": 19800
+    },
+    {
+      "epoch": 5.1227548778912,
+      "learning_rate": 1.6820303383897317e-07,
+      "loss": 2.5688,
+      "step": 19820
+    },
+    {
+      "epoch": 5.127923504328725,
+      "learning_rate": 1.6817062102943083e-07,
+      "loss": 2.6504,
+      "step": 19840
+    },
+    {
+      "epoch": 5.133092130766249,
+      "learning_rate": 1.681382082198885e-07,
+      "loss": 2.701,
+      "step": 19860
+    },
+    {
+      "epoch": 5.138260757203773,
+      "learning_rate": 1.6810579541034615e-07,
+      "loss": 2.6691,
+      "step": 19880
+    },
+    {
+      "epoch": 5.143429383641298,
+      "learning_rate": 1.6807338260080384e-07,
+      "loss": 2.6396,
+      "step": 19900
+    },
+    {
+      "epoch": 5.148598010078821,
+      "learning_rate": 1.680409697912615e-07,
+      "loss": 2.6752,
+      "step": 19920
+    },
+    {
+      "epoch": 5.1537666365163455,
+      "learning_rate": 1.6800855698171916e-07,
+      "loss": 2.567,
+      "step": 19940
+    },
+    {
+      "epoch": 5.15893526295387,
+      "learning_rate": 1.6797614417217685e-07,
+      "loss": 2.6691,
+      "step": 19960
+    },
+    {
+      "epoch": 5.164103889391394,
+      "learning_rate": 1.6794373136263451e-07,
+      "loss": 2.6172,
+      "step": 19980
+    },
+    {
+      "epoch": 5.169272515828919,
+      "learning_rate": 1.6791131855309218e-07,
+      "loss": 2.6856,
+      "step": 20000
+    },
+    {
+      "epoch": 5.174441142266443,
+      "learning_rate": 1.6787890574354984e-07,
+      "loss": 2.6874,
+      "step": 20020
+    },
+    {
+      "epoch": 5.179609768703967,
+      "learning_rate": 1.678464929340075e-07,
+      "loss": 2.6737,
+      "step": 20040
+    },
+    {
+      "epoch": 5.184778395141491,
+      "learning_rate": 1.678140801244652e-07,
+      "loss": 2.6503,
+      "step": 20060
+    },
+    {
+      "epoch": 5.189947021579015,
+      "learning_rate": 1.6778166731492285e-07,
+      "loss": 2.6155,
+      "step": 20080
+    },
+    {
+      "epoch": 5.1951156480165395,
+      "learning_rate": 1.677492545053805e-07,
+      "loss": 2.7035,
+      "step": 20100
+    },
+    {
+      "epoch": 5.200284274454064,
+      "learning_rate": 1.677168416958382e-07,
+      "loss": 2.6192,
+      "step": 20120
+    },
+    {
+      "epoch": 5.205452900891588,
+      "learning_rate": 1.6768442888629586e-07,
+      "loss": 2.5974,
+      "step": 20140
+    },
+    {
+      "epoch": 5.2106215273291125,
+      "learning_rate": 1.6765201607675352e-07,
+      "loss": 2.694,
+      "step": 20160
+    },
+    {
+      "epoch": 5.215790153766637,
+      "learning_rate": 1.676196032672112e-07,
+      "loss": 2.6391,
+      "step": 20180
+    },
+    {
+      "epoch": 5.22095878020416,
+      "learning_rate": 1.6758719045766887e-07,
+      "loss": 2.6428,
+      "step": 20200
+    },
+    {
+      "epoch": 5.226127406641685,
+      "learning_rate": 1.6755477764812654e-07,
+      "loss": 2.6796,
+      "step": 20220
+    },
+    {
+      "epoch": 5.231296033079209,
+      "learning_rate": 1.675223648385842e-07,
+      "loss": 2.6819,
+      "step": 20240
+    },
+    {
+      "epoch": 5.2364646595167335,
+      "learning_rate": 1.6748995202904186e-07,
+      "loss": 2.6714,
+      "step": 20260
+    },
+    {
+      "epoch": 5.241633285954258,
+      "learning_rate": 1.6745753921949955e-07,
+      "loss": 2.6392,
+      "step": 20280
+    },
+    {
+      "epoch": 5.246801912391782,
+      "learning_rate": 1.674251264099572e-07,
+      "loss": 2.7057,
+      "step": 20300
+    },
+    {
+      "epoch": 5.2519705388293065,
+      "learning_rate": 1.6739271360041487e-07,
+      "loss": 2.6676,
+      "step": 20320
+    },
+    {
+      "epoch": 5.25713916526683,
+      "learning_rate": 1.6736030079087256e-07,
+      "loss": 2.7037,
+      "step": 20340
+    },
+    {
+      "epoch": 5.262307791704354,
+      "learning_rate": 1.6732788798133022e-07,
+      "loss": 2.7595,
+      "step": 20360
+    },
+    {
+      "epoch": 5.267476418141879,
+      "learning_rate": 1.6729547517178788e-07,
+      "loss": 2.6665,
+      "step": 20380
+    },
+    {
+      "epoch": 5.272645044579403,
+      "learning_rate": 1.6726306236224554e-07,
+      "loss": 2.6281,
+      "step": 20400
+    },
+    {
+      "epoch": 5.277813671016927,
+      "learning_rate": 1.6723064955270323e-07,
+      "loss": 2.6047,
+      "step": 20420
+    },
+    {
+      "epoch": 5.282982297454452,
+      "learning_rate": 1.671982367431609e-07,
+      "loss": 2.6466,
+      "step": 20440
+    },
+    {
+      "epoch": 5.288150923891976,
+      "learning_rate": 1.6716582393361856e-07,
+      "loss": 2.6026,
+      "step": 20460
+    },
+    {
+      "epoch": 5.2933195503295,
+      "learning_rate": 1.6713341112407622e-07,
+      "loss": 2.6714,
+      "step": 20480
+    },
+    {
+      "epoch": 5.298488176767024,
+      "learning_rate": 1.671009983145339e-07,
+      "loss": 2.6351,
+      "step": 20500
+    },
+    {
+      "epoch": 5.303656803204548,
+      "learning_rate": 1.6706858550499157e-07,
+      "loss": 2.6293,
+      "step": 20520
+    },
+    {
+      "epoch": 5.308825429642073,
+      "learning_rate": 1.6703617269544923e-07,
+      "loss": 2.6368,
+      "step": 20540
+    },
+    {
+      "epoch": 5.313994056079597,
+      "learning_rate": 1.6700375988590692e-07,
+      "loss": 2.6963,
+      "step": 20560
+    },
+    {
+      "epoch": 5.319162682517121,
+      "learning_rate": 1.6697134707636458e-07,
+      "loss": 2.6401,
+      "step": 20580
+    },
+    {
+      "epoch": 5.324331308954645,
+      "learning_rate": 1.6693893426682224e-07,
+      "loss": 2.669,
+      "step": 20600
+    },
+    {
+      "epoch": 5.329499935392169,
+      "learning_rate": 1.669065214572799e-07,
+      "loss": 2.6384,
+      "step": 20620
+    },
+    {
+      "epoch": 5.334668561829694,
+      "learning_rate": 1.6687410864773757e-07,
+      "loss": 2.7073,
+      "step": 20640
+    },
+    {
+      "epoch": 5.339837188267218,
+      "learning_rate": 1.6684169583819526e-07,
+      "loss": 2.6507,
+      "step": 20660
+    },
+    {
+      "epoch": 5.345005814704742,
+      "learning_rate": 1.6680928302865292e-07,
+      "loss": 2.6378,
+      "step": 20680
+    },
+    {
+      "epoch": 5.350174441142267,
+      "learning_rate": 1.6677687021911058e-07,
+      "loss": 2.6756,
+      "step": 20700
+    },
+    {
+      "epoch": 5.355343067579791,
+      "learning_rate": 1.6674445740956827e-07,
+      "loss": 2.6667,
+      "step": 20720
+    },
+    {
+      "epoch": 5.3605116940173145,
+      "learning_rate": 1.6671204460002593e-07,
+      "loss": 2.6714,
+      "step": 20740
+    },
+    {
+      "epoch": 5.365680320454839,
+      "learning_rate": 1.666796317904836e-07,
+      "loss": 2.6117,
+      "step": 20760
+    },
+    {
+      "epoch": 5.370848946892363,
+      "learning_rate": 1.6664721898094128e-07,
+      "loss": 2.6513,
+      "step": 20780
+    },
+    {
+      "epoch": 5.376017573329888,
+      "learning_rate": 1.6661480617139894e-07,
+      "loss": 2.6997,
+      "step": 20800
+    },
+    {
+      "epoch": 5.381186199767412,
+      "learning_rate": 1.665823933618566e-07,
+      "loss": 2.6395,
+      "step": 20820
+    },
+    {
+      "epoch": 5.386354826204936,
+      "learning_rate": 1.6654998055231426e-07,
+      "loss": 2.6615,
+      "step": 20840
+    },
+    {
+      "epoch": 5.39152345264246,
+      "learning_rate": 1.6651756774277193e-07,
+      "loss": 2.6915,
+      "step": 20860
+    },
+    {
+      "epoch": 5.396692079079984,
+      "learning_rate": 1.6648515493322961e-07,
+      "loss": 2.6248,
+      "step": 20880
+    },
+    {
+      "epoch": 5.4018607055175085,
+      "learning_rate": 1.6645274212368728e-07,
+      "loss": 2.7061,
+      "step": 20900
+    },
+    {
+      "epoch": 5.407029331955033,
+      "learning_rate": 1.6642032931414494e-07,
+      "loss": 2.642,
+      "step": 20920
+    },
+    {
+      "epoch": 5.412197958392557,
+      "learning_rate": 1.6638791650460263e-07,
+      "loss": 2.6377,
+      "step": 20940
+    },
+    {
+      "epoch": 5.417366584830082,
+      "learning_rate": 1.663555036950603e-07,
+      "loss": 2.6452,
+      "step": 20960
+    },
+    {
+      "epoch": 5.422535211267606,
+      "learning_rate": 1.6632309088551795e-07,
+      "loss": 2.7206,
+      "step": 20980
+    },
+    {
+      "epoch": 5.427703837705129,
+      "learning_rate": 1.662906780759756e-07,
+      "loss": 2.6524,
+      "step": 21000
+    },
+    {
+      "epoch": 5.432872464142654,
+      "learning_rate": 1.662582652664333e-07,
+      "loss": 2.6398,
+      "step": 21020
+    },
+    {
+      "epoch": 5.438041090580178,
+      "learning_rate": 1.6622585245689096e-07,
+      "loss": 2.6486,
+      "step": 21040
+    },
+    {
+      "epoch": 5.4432097170177025,
+      "learning_rate": 1.6619343964734862e-07,
+      "loss": 2.6466,
+      "step": 21060
+    },
+    {
+      "epoch": 5.448378343455227,
+      "learning_rate": 1.6616102683780629e-07,
+      "loss": 2.6818,
+      "step": 21080
+    },
+    {
+      "epoch": 5.453546969892751,
+      "learning_rate": 1.6612861402826397e-07,
+      "loss": 2.6826,
+      "step": 21100
+    },
+    {
+      "epoch": 5.458715596330276,
+      "learning_rate": 1.6609620121872164e-07,
+      "loss": 2.665,
+      "step": 21120
+    },
+    {
+      "epoch": 5.463884222767799,
+      "learning_rate": 1.660637884091793e-07,
+      "loss": 2.6018,
+      "step": 21140
+    },
+    {
+      "epoch": 5.469052849205323,
+      "learning_rate": 1.66031375599637e-07,
+      "loss": 2.6867,
+      "step": 21160
+    },
+    {
+      "epoch": 5.474221475642848,
+      "learning_rate": 1.6599896279009465e-07,
+      "loss": 2.6972,
+      "step": 21180
+    },
+    {
+      "epoch": 5.479390102080372,
+      "learning_rate": 1.659665499805523e-07,
+      "loss": 2.6775,
+      "step": 21200
+    },
+    {
+      "epoch": 5.4845587285178965,
+      "learning_rate": 1.6593413717100997e-07,
+      "loss": 2.7239,
+      "step": 21220
+    },
+    {
+      "epoch": 5.489727354955421,
+      "learning_rate": 1.6590172436146763e-07,
+      "loss": 2.6349,
+      "step": 21240
+    },
+    {
+      "epoch": 5.494895981392945,
+      "learning_rate": 1.6586931155192532e-07,
+      "loss": 2.6607,
+      "step": 21260
+    },
+    {
+      "epoch": 5.500064607830469,
+      "learning_rate": 1.6583689874238298e-07,
+      "loss": 2.6806,
+      "step": 21280
+    },
+    {
+      "epoch": 5.505233234267993,
+      "learning_rate": 1.6580448593284065e-07,
+      "loss": 2.7212,
+      "step": 21300
+    },
+    {
+      "epoch": 5.510401860705517,
+      "learning_rate": 1.6577207312329833e-07,
+      "loss": 2.6997,
+      "step": 21320
+    },
+    {
+      "epoch": 5.515570487143042,
+      "learning_rate": 1.65739660313756e-07,
+      "loss": 2.6277,
+      "step": 21340
+    },
+    {
+      "epoch": 5.520739113580566,
+      "learning_rate": 1.6570724750421366e-07,
+      "loss": 2.6953,
+      "step": 21360
+    },
+    {
+      "epoch": 5.5259077400180905,
+      "learning_rate": 1.6567483469467135e-07,
+      "loss": 2.7586,
+      "step": 21380
+    },
+    {
+      "epoch": 5.531076366455615,
+      "learning_rate": 1.65642421885129e-07,
+      "loss": 2.5887,
+      "step": 21400
+    },
+    {
+      "epoch": 5.536244992893138,
+      "learning_rate": 1.6561000907558667e-07,
+      "loss": 2.6364,
+      "step": 21420
+    },
+    {
+      "epoch": 5.541413619330663,
+      "learning_rate": 1.6557759626604433e-07,
+      "loss": 2.5951,
+      "step": 21440
+    },
+    {
+      "epoch": 5.546582245768187,
+      "learning_rate": 1.65545183456502e-07,
+      "loss": 2.6595,
+      "step": 21460
+    },
+    {
+      "epoch": 5.551750872205711,
+      "learning_rate": 1.6551277064695968e-07,
+      "loss": 2.602,
+      "step": 21480
+    },
+    {
+      "epoch": 5.556919498643236,
+      "learning_rate": 1.6548035783741734e-07,
+      "loss": 2.5996,
+      "step": 21500
+    },
+    {
+      "epoch": 5.56208812508076,
+      "learning_rate": 1.65447945027875e-07,
+      "loss": 2.6529,
+      "step": 21520
+    },
+    {
+      "epoch": 5.5672567515182845,
+      "learning_rate": 1.654155322183327e-07,
+      "loss": 2.6917,
+      "step": 21540
+    },
+    {
+      "epoch": 5.572425377955808,
+      "learning_rate": 1.6538311940879036e-07,
+      "loss": 2.6509,
+      "step": 21560
+    },
+    {
+      "epoch": 5.577594004393332,
+      "learning_rate": 1.6535070659924802e-07,
+      "loss": 2.6524,
+      "step": 21580
+    },
+    {
+      "epoch": 5.582762630830857,
+      "learning_rate": 1.6531829378970568e-07,
+      "loss": 2.6894,
+      "step": 21600
+    },
+    {
+      "epoch": 5.587931257268381,
+      "learning_rate": 1.6528588098016334e-07,
+      "loss": 2.6451,
+      "step": 21620
+    },
+    {
+      "epoch": 5.593099883705905,
+      "learning_rate": 1.6525346817062103e-07,
+      "loss": 2.6636,
+      "step": 21640
+    },
+    {
+      "epoch": 5.59826851014343,
+      "learning_rate": 1.652210553610787e-07,
+      "loss": 2.6696,
+      "step": 21660
+    },
+    {
+      "epoch": 5.603437136580954,
+      "learning_rate": 1.6518864255153635e-07,
+      "loss": 2.6577,
+      "step": 21680
+    },
+    {
+      "epoch": 5.608605763018478,
+      "learning_rate": 1.6515622974199404e-07,
+      "loss": 2.5968,
+      "step": 21700
+    },
+    {
+      "epoch": 5.613774389456002,
+      "learning_rate": 1.651238169324517e-07,
+      "loss": 2.6492,
+      "step": 21720
+    },
+    {
+      "epoch": 5.618943015893526,
+      "learning_rate": 1.6509140412290937e-07,
+      "loss": 2.6357,
+      "step": 21740
+    },
+    {
+      "epoch": 5.624111642331051,
+      "learning_rate": 1.6505899131336705e-07,
+      "loss": 2.6401,
+      "step": 21760
+    },
+    {
+      "epoch": 5.629280268768575,
+      "learning_rate": 1.650265785038247e-07,
+      "loss": 2.5984,
+      "step": 21780
+    },
+    {
+      "epoch": 5.6344488952060985,
+      "learning_rate": 1.6499416569428238e-07,
+      "loss": 2.6678,
+      "step": 21800
+    },
+    {
+      "epoch": 5.639617521643623,
+      "learning_rate": 1.6496175288474004e-07,
+      "loss": 2.6161,
+      "step": 21820
+    },
+    {
+      "epoch": 5.644786148081147,
+      "learning_rate": 1.649293400751977e-07,
+      "loss": 2.6262,
+      "step": 21840
+    },
+    {
+      "epoch": 5.649954774518672,
+      "learning_rate": 1.648969272656554e-07,
+      "loss": 2.6514,
+      "step": 21860
+    },
+    {
+      "epoch": 5.655123400956196,
+      "learning_rate": 1.6486451445611305e-07,
+      "loss": 2.6629,
+      "step": 21880
+    },
+    {
+      "epoch": 5.66029202739372,
+      "learning_rate": 1.6483210164657071e-07,
+      "loss": 2.6764,
+      "step": 21900
+    },
+    {
+      "epoch": 5.665460653831245,
+      "learning_rate": 1.647996888370284e-07,
+      "loss": 2.6414,
+      "step": 21920
+    },
+    {
+      "epoch": 5.670629280268768,
+      "learning_rate": 1.6476727602748604e-07,
+      "loss": 2.5379,
+      "step": 21940
+    },
+    {
+      "epoch": 5.6757979067062925,
+      "learning_rate": 1.6473486321794373e-07,
+      "loss": 2.6744,
+      "step": 21960
+    },
+    {
+      "epoch": 5.680966533143817,
+      "learning_rate": 1.647024504084014e-07,
+      "loss": 2.7254,
+      "step": 21980
+    },
+    {
+      "epoch": 5.686135159581341,
+      "learning_rate": 1.6467003759885905e-07,
+      "loss": 2.6408,
+      "step": 22000
+    },
+    {
+      "epoch": 5.6913037860188656,
+      "learning_rate": 1.6463762478931674e-07,
+      "loss": 2.6751,
+      "step": 22020
+    },
+    {
+      "epoch": 5.69647241245639,
+      "learning_rate": 1.646052119797744e-07,
+      "loss": 2.6391,
+      "step": 22040
+    },
+    {
+      "epoch": 5.701641038893914,
+      "learning_rate": 1.6457279917023206e-07,
+      "loss": 2.625,
+      "step": 22060
+    },
+    {
+      "epoch": 5.706809665331438,
+      "learning_rate": 1.6454038636068975e-07,
+      "loss": 2.607,
+      "step": 22080
+    },
+    {
+      "epoch": 5.711978291768962,
+      "learning_rate": 1.6450797355114739e-07,
+      "loss": 2.6629,
+      "step": 22100
+    },
+    {
+      "epoch": 5.7171469182064865,
+      "learning_rate": 1.6447556074160507e-07,
+      "loss": 2.6358,
+      "step": 22120
+    },
+    {
+      "epoch": 5.722315544644011,
+      "learning_rate": 1.6444314793206274e-07,
+      "loss": 2.6962,
+      "step": 22140
+    },
+    {
+      "epoch": 5.727484171081535,
+      "learning_rate": 1.644107351225204e-07,
+      "loss": 2.6403,
+      "step": 22160
+    },
+    {
+      "epoch": 5.7326527975190595,
+      "learning_rate": 1.6437832231297809e-07,
+      "loss": 2.641,
+      "step": 22180
+    },
+    {
+      "epoch": 5.737821423956584,
+      "learning_rate": 1.6434590950343575e-07,
+      "loss": 2.6213,
+      "step": 22200
+    },
+    {
+      "epoch": 5.742990050394107,
+      "learning_rate": 1.643134966938934e-07,
+      "loss": 2.6508,
+      "step": 22220
+    },
+    {
+      "epoch": 5.748158676831632,
+      "learning_rate": 1.642810838843511e-07,
+      "loss": 2.6528,
+      "step": 22240
+    },
+    {
+      "epoch": 5.753327303269156,
+      "learning_rate": 1.6424867107480873e-07,
+      "loss": 2.5921,
+      "step": 22260
+    },
+    {
+      "epoch": 5.7584959297066804,
+      "learning_rate": 1.6421625826526642e-07,
+      "loss": 2.6379,
+      "step": 22280
+    },
+    {
+      "epoch": 5.763664556144205,
+      "learning_rate": 1.641838454557241e-07,
+      "loss": 2.6838,
+      "step": 22300
+    },
+    {
+      "epoch": 5.768833182581729,
+      "learning_rate": 1.6415143264618175e-07,
+      "loss": 2.6511,
+      "step": 22320
+    },
+    {
+      "epoch": 5.7740018090192535,
+      "learning_rate": 1.6411901983663943e-07,
+      "loss": 2.6547,
+      "step": 22340
+    },
+    {
+      "epoch": 5.779170435456777,
+      "learning_rate": 1.640866070270971e-07,
+      "loss": 2.5962,
+      "step": 22360
+    },
+    {
+      "epoch": 5.784339061894301,
+      "learning_rate": 1.6405419421755476e-07,
+      "loss": 2.6074,
+      "step": 22380
+    },
+    {
+      "epoch": 5.789507688331826,
+      "learning_rate": 1.6402178140801245e-07,
+      "loss": 2.5927,
+      "step": 22400
+    },
+    {
+      "epoch": 5.79467631476935,
+      "learning_rate": 1.6398936859847008e-07,
+      "loss": 2.6697,
+      "step": 22420
+    },
+    {
+      "epoch": 5.799844941206874,
+      "learning_rate": 1.6395695578892777e-07,
+      "loss": 2.6267,
+      "step": 22440
+    },
+    {
+      "epoch": 5.805013567644399,
+      "learning_rate": 1.6392454297938546e-07,
+      "loss": 2.6833,
+      "step": 22460
+    },
+    {
+      "epoch": 5.810182194081923,
+      "learning_rate": 1.638921301698431e-07,
+      "loss": 2.6647,
+      "step": 22480
+    },
+    {
+      "epoch": 5.815350820519447,
+      "learning_rate": 1.6385971736030078e-07,
+      "loss": 2.6619,
+      "step": 22500
+    },
+    {
+      "epoch": 5.820519446956971,
+      "learning_rate": 1.6382730455075847e-07,
+      "loss": 2.6341,
+      "step": 22520
+    },
+    {
+      "epoch": 5.825688073394495,
+      "learning_rate": 1.637948917412161e-07,
+      "loss": 2.604,
+      "step": 22540
+    },
+    {
+      "epoch": 5.83085669983202,
+      "learning_rate": 1.637624789316738e-07,
+      "loss": 2.6344,
+      "step": 22560
+    },
+    {
+      "epoch": 5.836025326269544,
+      "learning_rate": 1.6373006612213146e-07,
+      "loss": 2.6453,
+      "step": 22580
+    },
+    {
+      "epoch": 5.841193952707068,
+      "learning_rate": 1.6369765331258912e-07,
+      "loss": 2.6433,
+      "step": 22600
+    },
+    {
+      "epoch": 5.846362579144593,
+      "learning_rate": 1.636652405030468e-07,
+      "loss": 2.5673,
+      "step": 22620
+    },
+    {
+      "epoch": 5.851531205582116,
+      "learning_rate": 1.6363282769350444e-07,
+      "loss": 2.6455,
+      "step": 22640
+    },
+    {
+      "epoch": 5.856699832019641,
+      "learning_rate": 1.6360041488396213e-07,
+      "loss": 2.6617,
+      "step": 22660
+    },
+    {
+      "epoch": 5.861868458457165,
+      "learning_rate": 1.6356800207441982e-07,
+      "loss": 2.6395,
+      "step": 22680
+    },
+    {
+      "epoch": 5.867037084894689,
+      "learning_rate": 1.6353558926487745e-07,
+      "loss": 2.6358,
+      "step": 22700
+    },
+    {
+      "epoch": 5.872205711332214,
+      "learning_rate": 1.6350317645533514e-07,
+      "loss": 2.618,
+      "step": 22720
+    },
+    {
+      "epoch": 5.877374337769738,
+      "learning_rate": 1.634707636457928e-07,
+      "loss": 2.6219,
+      "step": 22740
+    },
+    {
+      "epoch": 5.882542964207262,
+      "learning_rate": 1.6343835083625047e-07,
+      "loss": 2.6028,
+      "step": 22760
+    },
+    {
+      "epoch": 5.887711590644786,
+      "learning_rate": 1.6340593802670815e-07,
+      "loss": 2.599,
+      "step": 22780
+    },
+    {
+      "epoch": 5.89288021708231,
+      "learning_rate": 1.633735252171658e-07,
+      "loss": 2.6062,
+      "step": 22800
+    },
+    {
+      "epoch": 5.898048843519835,
+      "learning_rate": 1.6334111240762348e-07,
+      "loss": 2.5854,
+      "step": 22820
+    },
+    {
+      "epoch": 5.903217469957359,
+      "learning_rate": 1.6330869959808117e-07,
+      "loss": 2.5416,
+      "step": 22840
+    },
+    {
+      "epoch": 5.908386096394883,
+      "learning_rate": 1.632762867885388e-07,
+      "loss": 2.6731,
+      "step": 22860
+    },
+    {
+      "epoch": 5.913554722832407,
+      "learning_rate": 1.632438739789965e-07,
+      "loss": 2.6271,
+      "step": 22880
+    },
+    {
+      "epoch": 5.918723349269931,
+      "learning_rate": 1.6321146116945418e-07,
+      "loss": 2.5896,
+      "step": 22900
+    },
+    {
+      "epoch": 5.9238919757074555,
+      "learning_rate": 1.6317904835991181e-07,
+      "loss": 2.6794,
+      "step": 22920
+    },
+    {
+      "epoch": 5.92906060214498,
+      "learning_rate": 1.631466355503695e-07,
+      "loss": 2.6051,
+      "step": 22940
+    },
+    {
+      "epoch": 5.934229228582504,
+      "learning_rate": 1.6311422274082716e-07,
+      "loss": 2.6901,
+      "step": 22960
+    },
+    {
+      "epoch": 5.939397855020029,
+      "learning_rate": 1.6308180993128483e-07,
+      "loss": 2.6288,
+      "step": 22980
+    },
+    {
+      "epoch": 5.944566481457553,
+      "learning_rate": 1.6304939712174251e-07,
+      "loss": 2.6806,
+      "step": 23000
+    },
+    {
+      "epoch": 5.949735107895076,
+      "learning_rate": 1.6301698431220015e-07,
+      "loss": 2.6003,
+      "step": 23020
+    },
+    {
+      "epoch": 5.954903734332601,
+      "learning_rate": 1.6298457150265784e-07,
+      "loss": 2.6595,
+      "step": 23040
+    },
+    {
+      "epoch": 5.960072360770125,
+      "learning_rate": 1.6295215869311553e-07,
+      "loss": 2.6563,
+      "step": 23060
+    },
+    {
+      "epoch": 5.9652409872076495,
+      "learning_rate": 1.6291974588357316e-07,
+      "loss": 2.6499,
+      "step": 23080
+    },
+    {
+      "epoch": 5.970409613645174,
+      "learning_rate": 1.6288733307403085e-07,
+      "loss": 2.6836,
+      "step": 23100
+    },
+    {
+      "epoch": 5.975578240082698,
+      "learning_rate": 1.6285492026448854e-07,
+      "loss": 2.6193,
+      "step": 23120
+    },
+    {
+      "epoch": 5.980746866520223,
+      "learning_rate": 1.6282250745494617e-07,
+      "loss": 2.6293,
+      "step": 23140
+    },
+    {
+      "epoch": 5.985915492957746,
+      "learning_rate": 1.6279009464540386e-07,
+      "loss": 2.7275,
+      "step": 23160
+    },
+    {
+      "epoch": 5.99108411939527,
+      "learning_rate": 1.6275768183586152e-07,
+      "loss": 2.6021,
+      "step": 23180
+    },
+    {
+      "epoch": 5.996252745832795,
+      "learning_rate": 1.6272526902631919e-07,
+      "loss": 2.6331,
+      "step": 23200
+    },
+    {
+      "epoch": 5.999870784339062,
+      "eval_bleu": 1.6991,
+      "eval_gen_len": 66.9245,
+      "eval_loss": 2.596095085144043,
+      "eval_runtime": 1347.5897,
+      "eval_samples_per_second": 1.278,
+      "eval_steps_per_second": 0.639,
+      "step": 23214
+    },
+    {
+      "epoch": 6.001421372270319,
+      "learning_rate": 1.6269285621677687e-07,
+      "loss": 2.6423,
+      "step": 23220
+    },
+    {
+      "epoch": 6.0065899987078435,
+      "learning_rate": 1.626604434072345e-07,
+      "loss": 2.667,
+      "step": 23240
+    },
+    {
+      "epoch": 6.011758625145368,
+      "learning_rate": 1.626280305976922e-07,
+      "loss": 2.6741,
+      "step": 23260
+    },
+    {
+      "epoch": 6.016927251582892,
+      "learning_rate": 1.6259561778814989e-07,
+      "loss": 2.6269,
+      "step": 23280
+    },
+    {
+      "epoch": 6.022095878020416,
+      "learning_rate": 1.6256320497860752e-07,
+      "loss": 2.6817,
+      "step": 23300
+    },
+    {
+      "epoch": 6.02726450445794,
+      "learning_rate": 1.625307921690652e-07,
+      "loss": 2.6804,
+      "step": 23320
+    },
+    {
+      "epoch": 6.032433130895464,
+      "learning_rate": 1.6249837935952287e-07,
+      "loss": 2.6283,
+      "step": 23340
+    },
+    {
+      "epoch": 6.037601757332989,
+      "learning_rate": 1.6246596654998053e-07,
+      "loss": 2.6344,
+      "step": 23360
+    },
+    {
+      "epoch": 6.042770383770513,
+      "learning_rate": 1.6243355374043822e-07,
+      "loss": 2.6453,
+      "step": 23380
+    },
+    {
+      "epoch": 6.0479390102080375,
+      "learning_rate": 1.6240114093089586e-07,
+      "loss": 2.5926,
+      "step": 23400
+    },
+    {
+      "epoch": 6.053107636645562,
+      "learning_rate": 1.6236872812135355e-07,
+      "loss": 2.6709,
+      "step": 23420
+    },
+    {
+      "epoch": 6.058276263083085,
+      "learning_rate": 1.6233631531181123e-07,
+      "loss": 2.6228,
+      "step": 23440
+    },
+    {
+      "epoch": 6.06344488952061,
+      "learning_rate": 1.6230390250226887e-07,
+      "loss": 2.6006,
+      "step": 23460
+    },
+    {
+      "epoch": 6.068613515958134,
+      "learning_rate": 1.6227148969272656e-07,
+      "loss": 2.6179,
+      "step": 23480
+    },
+    {
+      "epoch": 6.073782142395658,
+      "learning_rate": 1.6223907688318425e-07,
+      "loss": 2.63,
+      "step": 23500
+    },
+    {
+      "epoch": 6.078950768833183,
+      "learning_rate": 1.6220666407364188e-07,
+      "loss": 2.653,
+      "step": 23520
+    },
+    {
+      "epoch": 6.084119395270707,
+      "learning_rate": 1.6217425126409957e-07,
+      "loss": 2.6129,
+      "step": 23540
+    },
+    {
+      "epoch": 6.0892880217082315,
+      "learning_rate": 1.6214183845455723e-07,
+      "loss": 2.6008,
+      "step": 23560
+    },
+    {
+      "epoch": 6.094456648145755,
+      "learning_rate": 1.621094256450149e-07,
+      "loss": 2.6594,
+      "step": 23580
+    },
+    {
+      "epoch": 6.099625274583279,
+      "learning_rate": 1.6207701283547258e-07,
+      "loss": 2.617,
+      "step": 23600
+    },
+    {
+      "epoch": 6.104793901020804,
+      "learning_rate": 1.6204460002593022e-07,
+      "loss": 2.6392,
+      "step": 23620
+    },
+    {
+      "epoch": 6.109962527458328,
+      "learning_rate": 1.620121872163879e-07,
+      "loss": 2.6315,
+      "step": 23640
+    },
+    {
+      "epoch": 6.115131153895852,
+      "learning_rate": 1.619797744068456e-07,
+      "loss": 2.6131,
+      "step": 23660
+    },
+    {
+      "epoch": 6.120299780333377,
+      "learning_rate": 1.6194736159730323e-07,
+      "loss": 2.6512,
+      "step": 23680
+    },
+    {
+      "epoch": 6.1254684067709,
+      "learning_rate": 1.6191494878776092e-07,
+      "loss": 2.6196,
+      "step": 23700
+    },
+    {
+      "epoch": 6.130637033208425,
+      "learning_rate": 1.618825359782186e-07,
+      "loss": 2.6234,
+      "step": 23720
+    },
+    {
+      "epoch": 6.135805659645949,
+      "learning_rate": 1.6185012316867624e-07,
+      "loss": 2.5817,
+      "step": 23740
+    },
+    {
+      "epoch": 6.140974286083473,
+      "learning_rate": 1.6181771035913393e-07,
+      "loss": 2.5873,
+      "step": 23760
+    },
+    {
+      "epoch": 6.146142912520998,
+      "learning_rate": 1.617852975495916e-07,
+      "loss": 2.5957,
+      "step": 23780
+    },
+    {
+      "epoch": 6.151311538958522,
+      "learning_rate": 1.6175288474004925e-07,
+      "loss": 2.613,
+      "step": 23800
+    },
+    {
+      "epoch": 6.156480165396046,
+      "learning_rate": 1.6172047193050694e-07,
+      "loss": 2.5577,
+      "step": 23820
+    },
+    {
+      "epoch": 6.16164879183357,
+      "learning_rate": 1.6168805912096458e-07,
+      "loss": 2.6101,
+      "step": 23840
+    },
+    {
+      "epoch": 6.166817418271094,
+      "learning_rate": 1.6165564631142227e-07,
+      "loss": 2.5553,
+      "step": 23860
+    },
+    {
+      "epoch": 6.171986044708619,
+      "learning_rate": 1.6162323350187995e-07,
+      "loss": 2.6326,
+      "step": 23880
+    },
+    {
+      "epoch": 6.177154671146143,
+      "learning_rate": 1.615908206923376e-07,
+      "loss": 2.5922,
+      "step": 23900
+    },
+    {
+      "epoch": 6.182323297583667,
+      "learning_rate": 1.6155840788279528e-07,
+      "loss": 2.5913,
+      "step": 23920
+    },
+    {
+      "epoch": 6.187491924021192,
+      "learning_rate": 1.6152599507325294e-07,
+      "loss": 2.6378,
+      "step": 23940
+    },
+    {
+      "epoch": 6.192660550458716,
+      "learning_rate": 1.614935822637106e-07,
+      "loss": 2.5969,
+      "step": 23960
+    },
+    {
+      "epoch": 6.1978291768962395,
+      "learning_rate": 1.614611694541683e-07,
+      "loss": 2.5971,
+      "step": 23980
+    },
+    {
+      "epoch": 6.202997803333764,
+      "learning_rate": 1.6142875664462593e-07,
+      "loss": 2.616,
+      "step": 24000
+    },
+    {
+      "epoch": 6.208166429771288,
+      "learning_rate": 1.6139634383508361e-07,
+      "loss": 2.6352,
+      "step": 24020
+    },
+    {
+      "epoch": 6.2133350562088125,
+      "learning_rate": 1.613639310255413e-07,
+      "loss": 2.6371,
+      "step": 24040
+    },
+    {
+      "epoch": 6.218503682646337,
+      "learning_rate": 1.6133151821599894e-07,
+      "loss": 2.5946,
+      "step": 24060
+    },
+    {
+      "epoch": 6.223672309083861,
+      "learning_rate": 1.6129910540645663e-07,
+      "loss": 2.6379,
+      "step": 24080
+    },
+    {
+      "epoch": 6.228840935521385,
+      "learning_rate": 1.6126669259691431e-07,
+      "loss": 2.6046,
+      "step": 24100
+    },
+    {
+      "epoch": 6.234009561958909,
+      "learning_rate": 1.6123427978737195e-07,
+      "loss": 2.653,
+      "step": 24120
+    },
+    {
+      "epoch": 6.2391781883964335,
+      "learning_rate": 1.6120186697782964e-07,
+      "loss": 2.6409,
+      "step": 24140
+    },
+    {
+      "epoch": 6.244346814833958,
+      "learning_rate": 1.611694541682873e-07,
+      "loss": 2.6077,
+      "step": 24160
+    },
+    {
+      "epoch": 6.249515441271482,
+      "learning_rate": 1.6113704135874496e-07,
+      "loss": 2.5993,
+      "step": 24180
+    },
+    {
+      "epoch": 6.2546840677090065,
+      "learning_rate": 1.6110462854920265e-07,
+      "loss": 2.6326,
+      "step": 24200
+    },
+    {
+      "epoch": 6.259852694146531,
+      "learning_rate": 1.6107221573966029e-07,
+      "loss": 2.612,
+      "step": 24220
+    },
+    {
+      "epoch": 6.265021320584054,
+      "learning_rate": 1.6103980293011797e-07,
+      "loss": 2.662,
+      "step": 24240
+    },
+    {
+      "epoch": 6.270189947021579,
+      "learning_rate": 1.6100739012057566e-07,
+      "loss": 2.6377,
+      "step": 24260
+    },
+    {
+      "epoch": 6.275358573459103,
+      "learning_rate": 1.609749773110333e-07,
+      "loss": 2.6113,
+      "step": 24280
+    },
+    {
+      "epoch": 6.280527199896627,
+      "learning_rate": 1.6094256450149099e-07,
+      "loss": 2.6094,
+      "step": 24300
+    },
+    {
+      "epoch": 6.285695826334152,
+      "learning_rate": 1.6091015169194867e-07,
+      "loss": 2.6185,
+      "step": 24320
+    },
+    {
+      "epoch": 6.290864452771676,
+      "learning_rate": 1.608777388824063e-07,
+      "loss": 2.6523,
+      "step": 24340
+    },
+    {
+      "epoch": 6.2960330792092005,
+      "learning_rate": 1.60845326072864e-07,
+      "loss": 2.6458,
+      "step": 24360
+    },
+    {
+      "epoch": 6.301201705646724,
+      "learning_rate": 1.6081291326332166e-07,
+      "loss": 2.5826,
+      "step": 24380
+    },
+    {
+      "epoch": 6.306370332084248,
+      "learning_rate": 1.6078050045377932e-07,
+      "loss": 2.6395,
+      "step": 24400
+    },
+    {
+      "epoch": 6.311538958521773,
+      "learning_rate": 1.60748087644237e-07,
+      "loss": 2.6305,
+      "step": 24420
+    },
+    {
+      "epoch": 6.316707584959297,
+      "learning_rate": 1.6071567483469465e-07,
+      "loss": 2.6493,
+      "step": 24440
+    },
+    {
+      "epoch": 6.321876211396821,
+      "learning_rate": 1.6068326202515233e-07,
+      "loss": 2.6198,
+      "step": 24460
+    },
+    {
+      "epoch": 6.327044837834346,
+      "learning_rate": 1.6065084921561002e-07,
+      "loss": 2.5705,
+      "step": 24480
+    },
+    {
+      "epoch": 6.33221346427187,
+      "learning_rate": 1.6061843640606766e-07,
+      "loss": 2.5797,
+      "step": 24500
+    },
+    {
+      "epoch": 6.337382090709394,
+      "learning_rate": 1.6058602359652535e-07,
+      "loss": 2.6152,
+      "step": 24520
+    },
+    {
+      "epoch": 6.342550717146918,
+      "learning_rate": 1.60553610786983e-07,
+      "loss": 2.6075,
+      "step": 24540
+    },
+    {
+      "epoch": 6.347719343584442,
+      "learning_rate": 1.6052119797744067e-07,
+      "loss": 2.6446,
+      "step": 24560
+    },
+    {
+      "epoch": 6.352887970021967,
+      "learning_rate": 1.6048878516789836e-07,
+      "loss": 2.6204,
+      "step": 24580
+    },
+    {
+      "epoch": 6.358056596459491,
+      "learning_rate": 1.60456372358356e-07,
+      "loss": 2.6079,
+      "step": 24600
+    },
+    {
+      "epoch": 6.363225222897015,
+      "learning_rate": 1.6042395954881368e-07,
+      "loss": 2.59,
+      "step": 24620
+    },
+    {
+      "epoch": 6.36839384933454,
+      "learning_rate": 1.6039154673927137e-07,
+      "loss": 2.6417,
+      "step": 24640
+    },
+    {
+      "epoch": 6.373562475772063,
+      "learning_rate": 1.60359133929729e-07,
+      "loss": 2.6426,
+      "step": 24660
+    },
+    {
+      "epoch": 6.378731102209588,
+      "learning_rate": 1.603267211201867e-07,
+      "loss": 2.6004,
+      "step": 24680
+    },
+    {
+      "epoch": 6.383899728647112,
+      "learning_rate": 1.6029430831064438e-07,
+      "loss": 2.6422,
+      "step": 24700
+    },
+    {
+      "epoch": 6.389068355084636,
+      "learning_rate": 1.6026189550110202e-07,
+      "loss": 2.595,
+      "step": 24720
+    },
+    {
+      "epoch": 6.394236981522161,
+      "learning_rate": 1.602294826915597e-07,
+      "loss": 2.6091,
+      "step": 24740
+    },
+    {
+      "epoch": 6.399405607959685,
+      "learning_rate": 1.6019706988201737e-07,
+      "loss": 2.5978,
+      "step": 24760
+    },
+    {
+      "epoch": 6.404574234397209,
+      "learning_rate": 1.6016465707247503e-07,
+      "loss": 2.5525,
+      "step": 24780
+    },
+    {
+      "epoch": 6.409742860834733,
+      "learning_rate": 1.6013224426293272e-07,
+      "loss": 2.6011,
+      "step": 24800
+    },
+    {
+      "epoch": 6.414911487272257,
+      "learning_rate": 1.6009983145339035e-07,
+      "loss": 2.5727,
+      "step": 24820
+    },
+    {
+      "epoch": 6.420080113709782,
+      "learning_rate": 1.6006741864384804e-07,
+      "loss": 2.6214,
+      "step": 24840
+    },
+    {
+      "epoch": 6.425248740147306,
+      "learning_rate": 1.6003500583430573e-07,
+      "loss": 2.6492,
+      "step": 24860
+    },
+    {
+      "epoch": 6.43041736658483,
+      "learning_rate": 1.6000259302476336e-07,
+      "loss": 2.628,
+      "step": 24880
+    },
+    {
+      "epoch": 6.435585993022355,
+      "learning_rate": 1.5997018021522105e-07,
+      "loss": 2.6282,
+      "step": 24900
+    },
+    {
+      "epoch": 6.440754619459878,
+      "learning_rate": 1.5993776740567874e-07,
+      "loss": 2.6483,
+      "step": 24920
+    },
+    {
+      "epoch": 6.4459232458974025,
+      "learning_rate": 1.5990535459613638e-07,
+      "loss": 2.6026,
+      "step": 24940
+    },
+    {
+      "epoch": 6.451091872334927,
+      "learning_rate": 1.5987294178659407e-07,
+      "loss": 2.6324,
+      "step": 24960
+    },
+    {
+      "epoch": 6.456260498772451,
+      "learning_rate": 1.5984052897705173e-07,
+      "loss": 2.5601,
+      "step": 24980
+    },
+    {
+      "epoch": 6.461429125209976,
+      "learning_rate": 1.598081161675094e-07,
+      "loss": 2.6437,
+      "step": 25000
+    },
+    {
+      "epoch": 6.4665977516475,
+      "learning_rate": 1.5977570335796708e-07,
+      "loss": 2.6326,
+      "step": 25020
+    },
+    {
+      "epoch": 6.471766378085024,
+      "learning_rate": 1.597432905484247e-07,
+      "loss": 2.6664,
+      "step": 25040
+    },
+    {
+      "epoch": 6.476935004522548,
+      "learning_rate": 1.597108777388824e-07,
+      "loss": 2.6057,
+      "step": 25060
+    },
+    {
+      "epoch": 6.482103630960072,
+      "learning_rate": 1.596784649293401e-07,
+      "loss": 2.6007,
+      "step": 25080
+    },
+    {
+      "epoch": 6.4872722573975965,
+      "learning_rate": 1.5964605211979772e-07,
+      "loss": 2.5935,
+      "step": 25100
+    },
+    {
+      "epoch": 6.492440883835121,
+      "learning_rate": 1.5961363931025541e-07,
+      "loss": 2.5344,
+      "step": 25120
+    },
+    {
+      "epoch": 6.497609510272645,
+      "learning_rate": 1.5958122650071308e-07,
+      "loss": 2.6249,
+      "step": 25140
+    },
+    {
+      "epoch": 6.50277813671017,
+      "learning_rate": 1.5954881369117074e-07,
+      "loss": 2.5882,
+      "step": 25160
+    },
+    {
+      "epoch": 6.507946763147693,
+      "learning_rate": 1.5951640088162843e-07,
+      "loss": 2.6219,
+      "step": 25180
+    },
+    {
+      "epoch": 6.513115389585217,
+      "learning_rate": 1.5948398807208606e-07,
+      "loss": 2.5838,
+      "step": 25200
+    },
+    {
+      "epoch": 6.518284016022742,
+      "learning_rate": 1.5945157526254375e-07,
+      "loss": 2.5943,
+      "step": 25220
+    },
+    {
+      "epoch": 6.523452642460266,
+      "learning_rate": 1.5941916245300144e-07,
+      "loss": 2.6468,
+      "step": 25240
+    },
+    {
+      "epoch": 6.5286212688977905,
+      "learning_rate": 1.5938674964345907e-07,
+      "loss": 2.6726,
+      "step": 25260
+    },
+    {
+      "epoch": 6.533789895335315,
+      "learning_rate": 1.5935433683391676e-07,
+      "loss": 2.5732,
+      "step": 25280
+    },
+    {
+      "epoch": 6.538958521772839,
+      "learning_rate": 1.5932192402437442e-07,
+      "loss": 2.5739,
+      "step": 25300
+    },
+    {
+      "epoch": 6.544127148210363,
+      "learning_rate": 1.5928951121483208e-07,
+      "loss": 2.5914,
+      "step": 25320
+    },
+    {
+      "epoch": 6.549295774647887,
+      "learning_rate": 1.5925709840528977e-07,
+      "loss": 2.6142,
+      "step": 25340
+    },
+    {
+      "epoch": 6.554464401085411,
+      "learning_rate": 1.5922468559574744e-07,
+      "loss": 2.6145,
+      "step": 25360
+    },
+    {
+      "epoch": 6.559633027522936,
+      "learning_rate": 1.591922727862051e-07,
+      "loss": 2.6449,
+      "step": 25380
+    },
+    {
+      "epoch": 6.56480165396046,
+      "learning_rate": 1.5915985997666279e-07,
+      "loss": 2.586,
+      "step": 25400
+    },
+    {
+      "epoch": 6.5699702803979845,
+      "learning_rate": 1.5912744716712042e-07,
+      "loss": 2.6136,
+      "step": 25420
+    },
+    {
+      "epoch": 6.575138906835509,
+      "learning_rate": 1.590950343575781e-07,
+      "loss": 2.6556,
+      "step": 25440
+    },
+    {
+      "epoch": 6.580307533273032,
+      "learning_rate": 1.5906262154803577e-07,
+      "loss": 2.6557,
+      "step": 25460
+    },
+    {
+      "epoch": 6.585476159710557,
+      "learning_rate": 1.5903020873849343e-07,
+      "loss": 2.5518,
+      "step": 25480
+    },
+    {
+      "epoch": 6.590644786148081,
+      "learning_rate": 1.5899779592895112e-07,
+      "loss": 2.6054,
+      "step": 25500
+    },
+    {
+      "epoch": 6.595813412585605,
+      "learning_rate": 1.5896538311940878e-07,
+      "loss": 2.5766,
+      "step": 25520
+    },
+    {
+      "epoch": 6.60098203902313,
+      "learning_rate": 1.5893297030986644e-07,
+      "loss": 2.5573,
+      "step": 25540
+    },
+    {
+      "epoch": 6.606150665460654,
+      "learning_rate": 1.5890055750032413e-07,
+      "loss": 2.6429,
+      "step": 25560
+    },
+    {
+      "epoch": 6.6113192918981785,
+      "learning_rate": 1.588681446907818e-07,
+      "loss": 2.6795,
+      "step": 25580
+    },
+    {
+      "epoch": 6.616487918335702,
+      "learning_rate": 1.5883573188123946e-07,
+      "loss": 2.6573,
+      "step": 25600
+    },
+    {
+      "epoch": 6.621656544773226,
+      "learning_rate": 1.5880331907169712e-07,
+      "loss": 2.5762,
+      "step": 25620
+    },
+    {
+      "epoch": 6.626825171210751,
+      "learning_rate": 1.5877090626215478e-07,
+      "loss": 2.6336,
+      "step": 25640
+    },
+    {
+      "epoch": 6.631993797648275,
+      "learning_rate": 1.5873849345261247e-07,
+      "loss": 2.5999,
+      "step": 25660
+    },
+    {
+      "epoch": 6.637162424085799,
+      "learning_rate": 1.5870608064307013e-07,
+      "loss": 2.6625,
+      "step": 25680
+    },
+    {
+      "epoch": 6.642331050523324,
+      "learning_rate": 1.586736678335278e-07,
+      "loss": 2.5277,
+      "step": 25700
+    },
+    {
+      "epoch": 6.647499676960848,
+      "learning_rate": 1.5864125502398548e-07,
+      "loss": 2.5688,
+      "step": 25720
+    },
+    {
+      "epoch": 6.652668303398372,
+      "learning_rate": 1.5860884221444314e-07,
+      "loss": 2.5813,
+      "step": 25740
+    },
+    {
+      "epoch": 6.657836929835896,
+      "learning_rate": 1.585764294049008e-07,
+      "loss": 2.622,
+      "step": 25760
+    },
+    {
+      "epoch": 6.66300555627342,
+      "learning_rate": 1.5854401659535847e-07,
+      "loss": 2.5956,
+      "step": 25780
+    },
+    {
+      "epoch": 6.668174182710945,
+      "learning_rate": 1.5851160378581613e-07,
+      "loss": 2.5834,
+      "step": 25800
+    },
+    {
+      "epoch": 6.673342809148469,
+      "learning_rate": 1.5847919097627382e-07,
+      "loss": 2.5737,
+      "step": 25820
+    },
+    {
+      "epoch": 6.678511435585993,
+      "learning_rate": 1.5844677816673148e-07,
+      "loss": 2.5815,
+      "step": 25840
+    },
+    {
+      "epoch": 6.683680062023518,
+      "learning_rate": 1.5841436535718914e-07,
+      "loss": 2.5868,
+      "step": 25860
+    },
+    {
+      "epoch": 6.688848688461041,
+      "learning_rate": 1.5838195254764683e-07,
+      "loss": 2.6376,
+      "step": 25880
+    },
+    {
+      "epoch": 6.6940173148985656,
+      "learning_rate": 1.583495397381045e-07,
+      "loss": 2.5857,
+      "step": 25900
+    },
+    {
+      "epoch": 6.69918594133609,
+      "learning_rate": 1.5831712692856215e-07,
+      "loss": 2.6078,
+      "step": 25920
+    },
+    {
+      "epoch": 6.704354567773614,
+      "learning_rate": 1.5828471411901981e-07,
+      "loss": 2.6262,
+      "step": 25940
+    },
+    {
+      "epoch": 6.709523194211139,
+      "learning_rate": 1.582523013094775e-07,
+      "loss": 2.5676,
+      "step": 25960
+    },
+    {
+      "epoch": 6.714691820648663,
+      "learning_rate": 1.5821988849993516e-07,
+      "loss": 2.6036,
+      "step": 25980
+    },
+    {
+      "epoch": 6.7198604470861865,
+      "learning_rate": 1.5818747569039283e-07,
+      "loss": 2.5575,
+      "step": 26000
+    },
+    {
+      "epoch": 6.725029073523711,
+      "learning_rate": 1.581550628808505e-07,
+      "loss": 2.5708,
+      "step": 26020
+    },
+    {
+      "epoch": 6.730197699961235,
+      "learning_rate": 1.5812265007130818e-07,
+      "loss": 2.6489,
+      "step": 26040
+    },
+    {
+      "epoch": 6.7353663263987595,
+      "learning_rate": 1.5809023726176584e-07,
+      "loss": 2.6338,
+      "step": 26060
+    },
+    {
+      "epoch": 6.740534952836284,
+      "learning_rate": 1.580578244522235e-07,
+      "loss": 2.6506,
+      "step": 26080
+    },
+    {
+      "epoch": 6.745703579273808,
+      "learning_rate": 1.5802541164268116e-07,
+      "loss": 2.541,
+      "step": 26100
+    },
+    {
+      "epoch": 6.750872205711332,
+      "learning_rate": 1.5799299883313885e-07,
+      "loss": 2.5367,
+      "step": 26120
+    },
+    {
+      "epoch": 6.756040832148856,
+      "learning_rate": 1.579605860235965e-07,
+      "loss": 2.5891,
+      "step": 26140
+    },
+    {
+      "epoch": 6.7612094585863804,
+      "learning_rate": 1.5792817321405417e-07,
+      "loss": 2.6226,
+      "step": 26160
+    },
+    {
+      "epoch": 6.766378085023905,
+      "learning_rate": 1.5789576040451186e-07,
+      "loss": 2.6313,
+      "step": 26180
+    },
+    {
+      "epoch": 6.771546711461429,
+      "learning_rate": 1.5786334759496952e-07,
+      "loss": 2.5891,
+      "step": 26200
+    },
+    {
+      "epoch": 6.7767153378989535,
+      "learning_rate": 1.5783093478542719e-07,
+      "loss": 2.5806,
+      "step": 26220
+    },
+    {
+      "epoch": 6.781883964336478,
+      "learning_rate": 1.5779852197588485e-07,
+      "loss": 2.5518,
+      "step": 26240
+    },
+    {
+      "epoch": 6.787052590774001,
+      "learning_rate": 1.5776610916634254e-07,
+      "loss": 2.6104,
+      "step": 26260
+    },
+    {
+      "epoch": 6.792221217211526,
+      "learning_rate": 1.577336963568002e-07,
+      "loss": 2.6052,
+      "step": 26280
+    },
+    {
+      "epoch": 6.79738984364905,
+      "learning_rate": 1.5770128354725786e-07,
+      "loss": 2.5845,
+      "step": 26300
+    },
+    {
+      "epoch": 6.802558470086574,
+      "learning_rate": 1.5766887073771552e-07,
+      "loss": 2.6096,
+      "step": 26320
+    },
+    {
+      "epoch": 6.807727096524099,
+      "learning_rate": 1.576364579281732e-07,
+      "loss": 2.5916,
+      "step": 26340
+    },
+    {
+      "epoch": 6.812895722961623,
+      "learning_rate": 1.5760404511863087e-07,
+      "loss": 2.584,
+      "step": 26360
+    },
+    {
+      "epoch": 6.8180643493991475,
+      "learning_rate": 1.5757163230908853e-07,
+      "loss": 2.6194,
+      "step": 26380
+    },
+    {
+      "epoch": 6.823232975836671,
+      "learning_rate": 1.575392194995462e-07,
+      "loss": 2.6581,
+      "step": 26400
+    },
+    {
+      "epoch": 6.828401602274195,
+      "learning_rate": 1.5750680669000388e-07,
+      "loss": 2.6636,
+      "step": 26420
+    },
+    {
+      "epoch": 6.83357022871172,
+      "learning_rate": 1.5747439388046155e-07,
+      "loss": 2.6135,
+      "step": 26440
+    },
+    {
+      "epoch": 6.838738855149244,
+      "learning_rate": 1.574419810709192e-07,
+      "loss": 2.609,
+      "step": 26460
+    },
+    {
+      "epoch": 6.843907481586768,
+      "learning_rate": 1.574095682613769e-07,
+      "loss": 2.6168,
+      "step": 26480
+    },
+    {
+      "epoch": 6.849076108024293,
+      "learning_rate": 1.5737715545183456e-07,
+      "loss": 2.5408,
+      "step": 26500
+    },
+    {
+      "epoch": 6.854244734461817,
+      "learning_rate": 1.5734474264229222e-07,
+      "loss": 2.6252,
+      "step": 26520
+    },
+    {
+      "epoch": 6.859413360899341,
+      "learning_rate": 1.5731232983274988e-07,
+      "loss": 2.5935,
+      "step": 26540
+    },
+    {
+      "epoch": 6.864581987336865,
+      "learning_rate": 1.5727991702320757e-07,
+      "loss": 2.5767,
+      "step": 26560
+    },
+    {
+      "epoch": 6.869750613774389,
+      "learning_rate": 1.5724750421366523e-07,
+      "loss": 2.6351,
+      "step": 26580
+    },
+    {
+      "epoch": 6.874919240211914,
+      "learning_rate": 1.572150914041229e-07,
+      "loss": 2.5793,
+      "step": 26600
+    },
+    {
+      "epoch": 6.880087866649438,
+      "learning_rate": 1.5718267859458056e-07,
+      "loss": 2.6026,
+      "step": 26620
+    },
+    {
+      "epoch": 6.885256493086962,
+      "learning_rate": 1.5715026578503824e-07,
+      "loss": 2.5796,
+      "step": 26640
+    },
+    {
+      "epoch": 6.890425119524487,
+      "learning_rate": 1.571178529754959e-07,
+      "loss": 2.6488,
+      "step": 26660
+    },
+    {
+      "epoch": 6.89559374596201,
+      "learning_rate": 1.5708544016595357e-07,
+      "loss": 2.6461,
+      "step": 26680
+    },
+    {
+      "epoch": 6.900762372399535,
+      "learning_rate": 1.5705302735641123e-07,
+      "loss": 2.5738,
+      "step": 26700
+    },
+    {
+      "epoch": 6.905930998837059,
+      "learning_rate": 1.5702061454686892e-07,
+      "loss": 2.5407,
+      "step": 26720
+    },
+    {
+      "epoch": 6.911099625274583,
+      "learning_rate": 1.5698820173732658e-07,
+      "loss": 2.6304,
+      "step": 26740
+    },
+    {
+      "epoch": 6.916268251712108,
+      "learning_rate": 1.5695578892778424e-07,
+      "loss": 2.576,
+      "step": 26760
+    },
+    {
+      "epoch": 6.921436878149632,
+      "learning_rate": 1.5692337611824193e-07,
+      "loss": 2.5784,
+      "step": 26780
+    },
+    {
+      "epoch": 6.926605504587156,
+      "learning_rate": 1.568909633086996e-07,
+      "loss": 2.6124,
+      "step": 26800
+    },
+    {
+      "epoch": 6.93177413102468,
+      "learning_rate": 1.5685855049915725e-07,
+      "loss": 2.5992,
+      "step": 26820
+    },
+    {
+      "epoch": 6.936942757462204,
+      "learning_rate": 1.5682613768961492e-07,
+      "loss": 2.5961,
+      "step": 26840
+    },
+    {
+      "epoch": 6.942111383899729,
+      "learning_rate": 1.567937248800726e-07,
+      "loss": 2.5989,
+      "step": 26860
+    },
+    {
+      "epoch": 6.947280010337253,
+      "learning_rate": 1.5676131207053027e-07,
+      "loss": 2.6514,
+      "step": 26880
+    },
+    {
+      "epoch": 6.952448636774777,
+      "learning_rate": 1.5672889926098793e-07,
+      "loss": 2.5921,
+      "step": 26900
+    },
+    {
+      "epoch": 6.957617263212302,
+      "learning_rate": 1.566964864514456e-07,
+      "loss": 2.5907,
+      "step": 26920
+    },
+    {
+      "epoch": 6.962785889649826,
+      "learning_rate": 1.5666407364190328e-07,
+      "loss": 2.5183,
+      "step": 26940
+    },
+    {
+      "epoch": 6.9679545160873495,
+      "learning_rate": 1.5663166083236094e-07,
+      "loss": 2.6151,
+      "step": 26960
+    },
+    {
+      "epoch": 6.973123142524874,
+      "learning_rate": 1.565992480228186e-07,
+      "loss": 2.5454,
+      "step": 26980
+    },
+    {
+      "epoch": 6.978291768962398,
+      "learning_rate": 1.5656683521327626e-07,
+      "loss": 2.5349,
+      "step": 27000
+    },
+    {
+      "epoch": 6.983460395399923,
+      "learning_rate": 1.5653442240373395e-07,
+      "loss": 2.6422,
+      "step": 27020
+    },
+    {
+      "epoch": 6.988629021837447,
+      "learning_rate": 1.5650200959419161e-07,
+      "loss": 2.6273,
+      "step": 27040
+    },
+    {
+      "epoch": 6.993797648274971,
+      "learning_rate": 1.5646959678464928e-07,
+      "loss": 2.5851,
+      "step": 27060
+    },
+    {
+      "epoch": 6.998966274712495,
+      "learning_rate": 1.5643718397510696e-07,
+      "loss": 2.5716,
+      "step": 27080
+    },
+    {
+      "epoch": 7.0,
+      "eval_bleu": 5.2201,
+      "eval_gen_len": 46.1405,
+      "eval_loss": 2.561117172241211,
+      "eval_runtime": 958.6568,
+      "eval_samples_per_second": 1.796,
+      "eval_steps_per_second": 0.898,
+      "step": 27084
+    },
+    {
+      "epoch": 7.004134901150019,
+      "learning_rate": 1.5640477116556463e-07,
+      "loss": 2.5946,
+      "step": 27100
+    },
+    {
+      "epoch": 7.0093035275875435,
+      "learning_rate": 1.563723583560223e-07,
+      "loss": 2.5941,
+      "step": 27120
+    },
+    {
+      "epoch": 7.014472154025068,
+      "learning_rate": 1.5633994554647995e-07,
+      "loss": 2.5895,
+      "step": 27140
+    },
+    {
+      "epoch": 7.019640780462592,
+      "learning_rate": 1.5630753273693764e-07,
+      "loss": 2.5711,
+      "step": 27160
+    },
+    {
+      "epoch": 7.024809406900117,
+      "learning_rate": 1.562751199273953e-07,
+      "loss": 2.6074,
+      "step": 27180
+    },
+    {
+      "epoch": 7.029978033337641,
+      "learning_rate": 1.5624270711785296e-07,
+      "loss": 2.5763,
+      "step": 27200
+    },
+    {
+      "epoch": 7.035146659775164,
+      "learning_rate": 1.5621029430831062e-07,
+      "loss": 2.5656,
+      "step": 27220
+    },
+    {
+      "epoch": 7.040315286212689,
+      "learning_rate": 1.561778814987683e-07,
+      "loss": 2.576,
+      "step": 27240
+    },
+    {
+      "epoch": 7.045483912650213,
+      "learning_rate": 1.5614546868922597e-07,
+      "loss": 2.6323,
+      "step": 27260
+    },
+    {
+      "epoch": 7.0506525390877375,
+      "learning_rate": 1.5611305587968364e-07,
+      "loss": 2.6121,
+      "step": 27280
+    },
+    {
+      "epoch": 7.055821165525262,
+      "learning_rate": 1.560806430701413e-07,
+      "loss": 2.5962,
+      "step": 27300
+    },
+    {
+      "epoch": 7.060989791962786,
+      "learning_rate": 1.5604823026059899e-07,
+      "loss": 2.614,
+      "step": 27320
+    },
+    {
+      "epoch": 7.06615841840031,
+      "learning_rate": 1.5601581745105665e-07,
+      "loss": 2.5828,
+      "step": 27340
+    },
+    {
+      "epoch": 7.071327044837834,
+      "learning_rate": 1.559834046415143e-07,
+      "loss": 2.6366,
+      "step": 27360
+    },
+    {
+      "epoch": 7.076495671275358,
+      "learning_rate": 1.55950991831972e-07,
+      "loss": 2.5704,
+      "step": 27380
+    },
+    {
+      "epoch": 7.081664297712883,
+      "learning_rate": 1.5591857902242966e-07,
+      "loss": 2.6038,
+      "step": 27400
+    },
+    {
+      "epoch": 7.086832924150407,
+      "learning_rate": 1.5588616621288732e-07,
+      "loss": 2.5829,
+      "step": 27420
+    },
+    {
+      "epoch": 7.0920015505879315,
+      "learning_rate": 1.5585375340334498e-07,
+      "loss": 2.5535,
+      "step": 27440
+    },
+    {
+      "epoch": 7.097170177025456,
+      "learning_rate": 1.5582134059380267e-07,
+      "loss": 2.6527,
+      "step": 27460
+    },
+    {
+      "epoch": 7.102338803462979,
+      "learning_rate": 1.5578892778426033e-07,
+      "loss": 2.5602,
+      "step": 27480
+    },
+    {
+      "epoch": 7.107507429900504,
+      "learning_rate": 1.55756514974718e-07,
+      "loss": 2.5519,
+      "step": 27500
+    },
+    {
+      "epoch": 7.112676056338028,
+      "learning_rate": 1.5572410216517566e-07,
+      "loss": 2.6072,
+      "step": 27520
+    },
+    {
+      "epoch": 7.117844682775552,
+      "learning_rate": 1.5569168935563335e-07,
+      "loss": 2.5692,
+      "step": 27540
+    },
+    {
+      "epoch": 7.123013309213077,
+      "learning_rate": 1.55659276546091e-07,
+      "loss": 2.5865,
+      "step": 27560
+    },
+    {
+      "epoch": 7.128181935650601,
+      "learning_rate": 1.5562686373654867e-07,
+      "loss": 2.5513,
+      "step": 27580
+    },
+    {
+      "epoch": 7.1333505620881255,
+      "learning_rate": 1.5559445092700633e-07,
+      "loss": 2.5622,
+      "step": 27600
+    },
+    {
+      "epoch": 7.138519188525649,
+      "learning_rate": 1.5556203811746402e-07,
+      "loss": 2.5787,
+      "step": 27620
+    },
+    {
+      "epoch": 7.143687814963173,
+      "learning_rate": 1.5552962530792168e-07,
+      "loss": 2.5697,
+      "step": 27640
+    },
+    {
+      "epoch": 7.148856441400698,
+      "learning_rate": 1.5549721249837934e-07,
+      "loss": 2.5477,
+      "step": 27660
+    },
+    {
+      "epoch": 7.154025067838222,
+      "learning_rate": 1.5546479968883703e-07,
+      "loss": 2.5206,
+      "step": 27680
+    },
+    {
+      "epoch": 7.159193694275746,
+      "learning_rate": 1.554323868792947e-07,
+      "loss": 2.6325,
+      "step": 27700
+    },
+    {
+      "epoch": 7.164362320713271,
+      "learning_rate": 1.5539997406975236e-07,
+      "loss": 2.5897,
+      "step": 27720
+    },
+    {
+      "epoch": 7.169530947150795,
+      "learning_rate": 1.5536756126021002e-07,
+      "loss": 2.5669,
+      "step": 27740
+    },
+    {
+      "epoch": 7.174699573588319,
+      "learning_rate": 1.553351484506677e-07,
+      "loss": 2.5391,
+      "step": 27760
+    },
+    {
+      "epoch": 7.179868200025843,
+      "learning_rate": 1.5530273564112537e-07,
+      "loss": 2.5715,
+      "step": 27780
+    },
+    {
+      "epoch": 7.185036826463367,
+      "learning_rate": 1.5527032283158303e-07,
+      "loss": 2.6015,
+      "step": 27800
+    },
+    {
+      "epoch": 7.190205452900892,
+      "learning_rate": 1.552379100220407e-07,
+      "loss": 2.6043,
+      "step": 27820
+    },
+    {
+      "epoch": 7.195374079338416,
+      "learning_rate": 1.5520549721249838e-07,
+      "loss": 2.5723,
+      "step": 27840
+    },
+    {
+      "epoch": 7.20054270577594,
+      "learning_rate": 1.5517308440295604e-07,
+      "loss": 2.5877,
+      "step": 27860
+    },
+    {
+      "epoch": 7.205711332213465,
+      "learning_rate": 1.551406715934137e-07,
+      "loss": 2.5823,
+      "step": 27880
+    },
+    {
+      "epoch": 7.210879958650988,
+      "learning_rate": 1.5510825878387137e-07,
+      "loss": 2.5435,
+      "step": 27900
+    },
+    {
+      "epoch": 7.2160485850885125,
+      "learning_rate": 1.5507584597432905e-07,
+      "loss": 2.5416,
+      "step": 27920
+    },
+    {
+      "epoch": 7.221217211526037,
+      "learning_rate": 1.5504343316478672e-07,
+      "loss": 2.6001,
+      "step": 27940
+    },
+    {
+      "epoch": 7.226385837963561,
+      "learning_rate": 1.5501102035524438e-07,
+      "loss": 2.6136,
+      "step": 27960
+    },
+    {
+      "epoch": 7.231554464401086,
+      "learning_rate": 1.5497860754570204e-07,
+      "loss": 2.6079,
+      "step": 27980
+    },
+    {
+      "epoch": 7.23672309083861,
+      "learning_rate": 1.5494619473615973e-07,
+      "loss": 2.6051,
+      "step": 28000
+    },
+    {
+      "epoch": 7.2418917172761335,
+      "learning_rate": 1.549137819266174e-07,
+      "loss": 2.6275,
+      "step": 28020
+    },
+    {
+      "epoch": 7.247060343713658,
+      "learning_rate": 1.5488136911707505e-07,
+      "loss": 2.6114,
+      "step": 28040
+    },
+    {
+      "epoch": 7.252228970151182,
+      "learning_rate": 1.5484895630753274e-07,
+      "loss": 2.6532,
+      "step": 28060
+    },
+    {
+      "epoch": 7.2573975965887065,
+      "learning_rate": 1.548165434979904e-07,
+      "loss": 2.5954,
+      "step": 28080
+    },
+    {
+      "epoch": 7.262566223026231,
+      "learning_rate": 1.5478413068844806e-07,
+      "loss": 2.555,
+      "step": 28100
+    },
+    {
+      "epoch": 7.267734849463755,
+      "learning_rate": 1.5475171787890573e-07,
+      "loss": 2.601,
+      "step": 28120
+    },
+    {
+      "epoch": 7.27290347590128,
+      "learning_rate": 1.547193050693634e-07,
+      "loss": 2.5666,
+      "step": 28140
+    },
+    {
+      "epoch": 7.278072102338803,
+      "learning_rate": 1.5468689225982108e-07,
+      "loss": 2.563,
+      "step": 28160
+    },
+    {
+      "epoch": 7.283240728776327,
+      "learning_rate": 1.5465447945027874e-07,
+      "loss": 2.5919,
+      "step": 28180
+    },
+    {
+      "epoch": 7.288409355213852,
+      "learning_rate": 1.546220666407364e-07,
+      "loss": 2.5042,
+      "step": 28200
+    },
+    {
+      "epoch": 7.293577981651376,
+      "learning_rate": 1.545896538311941e-07,
+      "loss": 2.614,
+      "step": 28220
+    },
+    {
+      "epoch": 7.2987466080889005,
+      "learning_rate": 1.5455724102165175e-07,
+      "loss": 2.5763,
+      "step": 28240
+    },
+    {
+      "epoch": 7.303915234526425,
+      "learning_rate": 1.545248282121094e-07,
+      "loss": 2.5852,
+      "step": 28260
+    },
+    {
+      "epoch": 7.309083860963949,
+      "learning_rate": 1.544924154025671e-07,
+      "loss": 2.638,
+      "step": 28280
+    },
+    {
+      "epoch": 7.314252487401473,
+      "learning_rate": 1.5446000259302476e-07,
+      "loss": 2.5642,
+      "step": 28300
+    },
+    {
+      "epoch": 7.319421113838997,
+      "learning_rate": 1.5442758978348242e-07,
+      "loss": 2.5905,
+      "step": 28320
+    },
+    {
+      "epoch": 7.324589740276521,
+      "learning_rate": 1.5439517697394009e-07,
+      "loss": 2.5641,
+      "step": 28340
+    },
+    {
+      "epoch": 7.329758366714046,
+      "learning_rate": 1.5436276416439775e-07,
+      "loss": 2.5369,
+      "step": 28360
+    },
+    {
+      "epoch": 7.33492699315157,
+      "learning_rate": 1.5433035135485544e-07,
+      "loss": 2.5828,
+      "step": 28380
+    },
+    {
+      "epoch": 7.3400956195890945,
+      "learning_rate": 1.542979385453131e-07,
+      "loss": 2.6183,
+      "step": 28400
+    },
+    {
+      "epoch": 7.345264246026618,
+      "learning_rate": 1.5426552573577076e-07,
+      "loss": 2.5041,
+      "step": 28420
+    },
+    {
+      "epoch": 7.350432872464142,
+      "learning_rate": 1.5423311292622845e-07,
+      "loss": 2.6381,
+      "step": 28440
+    },
+    {
+      "epoch": 7.355601498901667,
+      "learning_rate": 1.542007001166861e-07,
+      "loss": 2.607,
+      "step": 28460
+    },
+    {
+      "epoch": 7.360770125339191,
+      "learning_rate": 1.5416828730714377e-07,
+      "loss": 2.5654,
+      "step": 28480
+    },
+    {
+      "epoch": 7.365938751776715,
+      "learning_rate": 1.5413587449760143e-07,
+      "loss": 2.5333,
+      "step": 28500
+    },
+    {
+      "epoch": 7.37110737821424,
+      "learning_rate": 1.5410346168805912e-07,
+      "loss": 2.5771,
+      "step": 28520
+    },
+    {
+      "epoch": 7.376276004651764,
+      "learning_rate": 1.5407104887851678e-07,
+      "loss": 2.5225,
+      "step": 28540
+    },
+    {
+      "epoch": 7.381444631089288,
+      "learning_rate": 1.5403863606897445e-07,
+      "loss": 2.6159,
+      "step": 28560
+    },
+    {
+      "epoch": 7.386613257526812,
+      "learning_rate": 1.540062232594321e-07,
+      "loss": 2.6229,
+      "step": 28580
+    },
+    {
+      "epoch": 7.391781883964336,
+      "learning_rate": 1.539738104498898e-07,
+      "loss": 2.5326,
+      "step": 28600
+    },
+    {
+      "epoch": 7.396950510401861,
+      "learning_rate": 1.5394139764034746e-07,
+      "loss": 2.5116,
+      "step": 28620
+    },
+    {
+      "epoch": 7.402119136839385,
+      "learning_rate": 1.5390898483080512e-07,
+      "loss": 2.5276,
+      "step": 28640
+    },
+    {
+      "epoch": 7.407287763276909,
+      "learning_rate": 1.538765720212628e-07,
+      "loss": 2.5662,
+      "step": 28660
+    },
+    {
+      "epoch": 7.412456389714434,
+      "learning_rate": 1.5384415921172047e-07,
+      "loss": 2.5843,
+      "step": 28680
+    },
+    {
+      "epoch": 7.417625016151957,
+      "learning_rate": 1.5381174640217813e-07,
+      "loss": 2.6341,
+      "step": 28700
+    },
+    {
+      "epoch": 7.422793642589482,
+      "learning_rate": 1.537793335926358e-07,
+      "loss": 2.5741,
+      "step": 28720
+    },
+    {
+      "epoch": 7.427962269027006,
+      "learning_rate": 1.5374692078309346e-07,
+      "loss": 2.5658,
+      "step": 28740
+    },
+    {
+      "epoch": 7.43313089546453,
+      "learning_rate": 1.5371450797355114e-07,
+      "loss": 2.5909,
+      "step": 28760
+    },
+    {
+      "epoch": 7.438299521902055,
+      "learning_rate": 1.536820951640088e-07,
+      "loss": 2.5543,
+      "step": 28780
+    },
+    {
+      "epoch": 7.443468148339579,
+      "learning_rate": 1.5364968235446647e-07,
+      "loss": 2.5546,
+      "step": 28800
+    },
+    {
+      "epoch": 7.448636774777103,
+      "learning_rate": 1.5361726954492416e-07,
+      "loss": 2.5236,
+      "step": 28820
+    },
+    {
+      "epoch": 7.453805401214627,
+      "learning_rate": 1.5358485673538182e-07,
+      "loss": 2.5981,
+      "step": 28840
+    },
+    {
+      "epoch": 7.458974027652151,
+      "learning_rate": 1.5355244392583948e-07,
+      "loss": 2.5322,
+      "step": 28860
+    },
+    {
+      "epoch": 7.464142654089676,
+      "learning_rate": 1.5352003111629717e-07,
+      "loss": 2.6326,
+      "step": 28880
+    },
+    {
+      "epoch": 7.4693112805272,
+      "learning_rate": 1.5348761830675483e-07,
+      "loss": 2.5647,
+      "step": 28900
+    },
+    {
+      "epoch": 7.474479906964724,
+      "learning_rate": 1.534552054972125e-07,
+      "loss": 2.6036,
+      "step": 28920
+    },
+    {
+      "epoch": 7.479648533402249,
+      "learning_rate": 1.5342279268767015e-07,
+      "loss": 2.5672,
+      "step": 28940
+    },
+    {
+      "epoch": 7.484817159839773,
+      "learning_rate": 1.5339037987812782e-07,
+      "loss": 2.5102,
+      "step": 28960
+    },
+    {
+      "epoch": 7.4899857862772965,
+      "learning_rate": 1.533579670685855e-07,
+      "loss": 2.6278,
+      "step": 28980
+    },
+    {
+      "epoch": 7.495154412714821,
+      "learning_rate": 1.5332555425904317e-07,
+      "loss": 2.6092,
+      "step": 29000
+    },
+    {
+      "epoch": 7.500323039152345,
+      "learning_rate": 1.5329314144950083e-07,
+      "loss": 2.5685,
+      "step": 29020
+    },
+    {
+      "epoch": 7.50549166558987,
+      "learning_rate": 1.5326072863995852e-07,
+      "loss": 2.5203,
+      "step": 29040
+    },
+    {
+      "epoch": 7.510660292027394,
+      "learning_rate": 1.5322831583041618e-07,
+      "loss": 2.6086,
+      "step": 29060
+    },
+    {
+      "epoch": 7.515828918464918,
+      "learning_rate": 1.5319590302087384e-07,
+      "loss": 2.556,
+      "step": 29080
+    },
+    {
+      "epoch": 7.520997544902443,
+      "learning_rate": 1.531634902113315e-07,
+      "loss": 2.6068,
+      "step": 29100
+    },
+    {
+      "epoch": 7.526166171339966,
+      "learning_rate": 1.531310774017892e-07,
+      "loss": 2.5915,
+      "step": 29120
+    },
+    {
+      "epoch": 7.5313347977774905,
+      "learning_rate": 1.5309866459224685e-07,
+      "loss": 2.5853,
+      "step": 29140
+    },
+    {
+      "epoch": 7.536503424215015,
+      "learning_rate": 1.5306625178270451e-07,
+      "loss": 2.5933,
+      "step": 29160
+    },
+    {
+      "epoch": 7.541672050652539,
+      "learning_rate": 1.5303383897316218e-07,
+      "loss": 2.5732,
+      "step": 29180
+    },
+    {
+      "epoch": 7.546840677090064,
+      "learning_rate": 1.5300142616361986e-07,
+      "loss": 2.5925,
+      "step": 29200
+    },
+    {
+      "epoch": 7.552009303527588,
+      "learning_rate": 1.5296901335407753e-07,
+      "loss": 2.6029,
+      "step": 29220
+    },
+    {
+      "epoch": 7.557177929965111,
+      "learning_rate": 1.529366005445352e-07,
+      "loss": 2.6053,
+      "step": 29240
+    },
+    {
+      "epoch": 7.562346556402636,
+      "learning_rate": 1.5290418773499288e-07,
+      "loss": 2.5341,
+      "step": 29260
+    },
+    {
+      "epoch": 7.56751518284016,
+      "learning_rate": 1.5287177492545054e-07,
+      "loss": 2.4981,
+      "step": 29280
+    },
+    {
+      "epoch": 7.5726838092776845,
+      "learning_rate": 1.528393621159082e-07,
+      "loss": 2.5551,
+      "step": 29300
+    },
+    {
+      "epoch": 7.577852435715209,
+      "learning_rate": 1.5280694930636586e-07,
+      "loss": 2.5029,
+      "step": 29320
+    },
+    {
+      "epoch": 7.583021062152733,
+      "learning_rate": 1.5277453649682352e-07,
+      "loss": 2.6108,
+      "step": 29340
+    },
+    {
+      "epoch": 7.588189688590257,
+      "learning_rate": 1.527421236872812e-07,
+      "loss": 2.5652,
+      "step": 29360
+    },
+    {
+      "epoch": 7.593358315027781,
+      "learning_rate": 1.5270971087773887e-07,
+      "loss": 2.5423,
+      "step": 29380
+    },
+    {
+      "epoch": 7.598526941465305,
+      "learning_rate": 1.5267729806819654e-07,
+      "loss": 2.5123,
+      "step": 29400
+    },
+    {
+      "epoch": 7.60369556790283,
+      "learning_rate": 1.5264488525865422e-07,
+      "loss": 2.5787,
+      "step": 29420
+    },
+    {
+      "epoch": 7.608864194340354,
+      "learning_rate": 1.5261247244911189e-07,
+      "loss": 2.5894,
+      "step": 29440
+    },
+    {
+      "epoch": 7.6140328207778785,
+      "learning_rate": 1.5258005963956955e-07,
+      "loss": 2.5427,
+      "step": 29460
+    },
+    {
+      "epoch": 7.619201447215403,
+      "learning_rate": 1.5254764683002724e-07,
+      "loss": 2.6234,
+      "step": 29480
+    },
+    {
+      "epoch": 7.624370073652926,
+      "learning_rate": 1.525152340204849e-07,
+      "loss": 2.5201,
+      "step": 29500
+    },
+    {
+      "epoch": 7.629538700090451,
+      "learning_rate": 1.5248282121094256e-07,
+      "loss": 2.5933,
+      "step": 29520
+    },
+    {
+      "epoch": 7.634707326527975,
+      "learning_rate": 1.5245040840140022e-07,
+      "loss": 2.555,
+      "step": 29540
+    },
+    {
+      "epoch": 7.639875952965499,
+      "learning_rate": 1.5241799559185788e-07,
+      "loss": 2.5297,
+      "step": 29560
+    },
+    {
+      "epoch": 7.645044579403024,
+      "learning_rate": 1.5238558278231557e-07,
+      "loss": 2.574,
+      "step": 29580
+    },
+    {
+      "epoch": 7.650213205840548,
+      "learning_rate": 1.5235316997277323e-07,
+      "loss": 2.6062,
+      "step": 29600
+    },
+    {
+      "epoch": 7.6553818322780725,
+      "learning_rate": 1.523207571632309e-07,
+      "loss": 2.5437,
+      "step": 29620
+    },
+    {
+      "epoch": 7.660550458715596,
+      "learning_rate": 1.5228834435368858e-07,
+      "loss": 2.599,
+      "step": 29640
+    },
+    {
+      "epoch": 7.66571908515312,
+      "learning_rate": 1.5225593154414625e-07,
+      "loss": 2.5211,
+      "step": 29660
+    },
+    {
+      "epoch": 7.670887711590645,
+      "learning_rate": 1.522235187346039e-07,
+      "loss": 2.538,
+      "step": 29680
+    },
+    {
+      "epoch": 7.676056338028169,
+      "learning_rate": 1.5219110592506157e-07,
+      "loss": 2.5887,
+      "step": 29700
+    },
+    {
+      "epoch": 7.681224964465693,
+      "learning_rate": 1.5215869311551926e-07,
+      "loss": 2.5519,
+      "step": 29720
+    },
+    {
+      "epoch": 7.686393590903218,
+      "learning_rate": 1.5212628030597692e-07,
+      "loss": 2.5509,
+      "step": 29740
+    },
+    {
+      "epoch": 7.691562217340742,
+      "learning_rate": 1.5209386749643458e-07,
+      "loss": 2.5353,
+      "step": 29760
+    },
+    {
+      "epoch": 7.6967308437782656,
+      "learning_rate": 1.5206145468689224e-07,
+      "loss": 2.6387,
+      "step": 29780
+    },
+    {
+      "epoch": 7.70189947021579,
+      "learning_rate": 1.5202904187734993e-07,
+      "loss": 2.5547,
+      "step": 29800
+    },
+    {
+      "epoch": 7.707068096653314,
+      "learning_rate": 1.519966290678076e-07,
+      "loss": 2.65,
+      "step": 29820
+    },
+    {
+      "epoch": 7.712236723090839,
+      "learning_rate": 1.5196421625826526e-07,
+      "loss": 2.5733,
+      "step": 29840
+    },
+    {
+      "epoch": 7.717405349528363,
+      "learning_rate": 1.5193180344872294e-07,
+      "loss": 2.5608,
+      "step": 29860
+    },
+    {
+      "epoch": 7.722573975965887,
+      "learning_rate": 1.518993906391806e-07,
+      "loss": 2.5819,
+      "step": 29880
+    },
+    {
+      "epoch": 7.727742602403412,
+      "learning_rate": 1.5186697782963827e-07,
+      "loss": 2.6017,
+      "step": 29900
+    },
+    {
+      "epoch": 7.732911228840935,
+      "learning_rate": 1.5183456502009593e-07,
+      "loss": 2.5467,
+      "step": 29920
+    },
+    {
+      "epoch": 7.7380798552784595,
+      "learning_rate": 1.518021522105536e-07,
+      "loss": 2.558,
+      "step": 29940
+    },
+    {
+      "epoch": 7.743248481715984,
+      "learning_rate": 1.5176973940101128e-07,
+      "loss": 2.5716,
+      "step": 29960
+    },
+    {
+      "epoch": 7.748417108153508,
+      "learning_rate": 1.5173732659146894e-07,
+      "loss": 2.5765,
+      "step": 29980
+    },
+    {
+      "epoch": 7.753585734591033,
+      "learning_rate": 1.517049137819266e-07,
+      "loss": 2.6051,
+      "step": 30000
+    },
+    {
+      "epoch": 7.758754361028557,
+      "learning_rate": 1.516725009723843e-07,
+      "loss": 2.6049,
+      "step": 30020
+    },
+    {
+      "epoch": 7.763922987466081,
+      "learning_rate": 1.5164008816284195e-07,
+      "loss": 2.5621,
+      "step": 30040
+    },
+    {
+      "epoch": 7.769091613903605,
+      "learning_rate": 1.5160767535329962e-07,
+      "loss": 2.5449,
+      "step": 30060
+    },
+    {
+      "epoch": 7.774260240341129,
+      "learning_rate": 1.515752625437573e-07,
+      "loss": 2.5652,
+      "step": 30080
+    },
+    {
+      "epoch": 7.7794288667786535,
+      "learning_rate": 1.5154284973421497e-07,
+      "loss": 2.5216,
+      "step": 30100
+    },
+    {
+      "epoch": 7.784597493216178,
+      "learning_rate": 1.5151043692467263e-07,
+      "loss": 2.521,
+      "step": 30120
+    },
+    {
+      "epoch": 7.789766119653702,
+      "learning_rate": 1.514780241151303e-07,
+      "loss": 2.5561,
+      "step": 30140
+    },
+    {
+      "epoch": 7.794934746091227,
+      "learning_rate": 1.5144561130558795e-07,
+      "loss": 2.5601,
+      "step": 30160
+    },
+    {
+      "epoch": 7.800103372528751,
+      "learning_rate": 1.5141319849604564e-07,
+      "loss": 2.5599,
+      "step": 30180
+    },
+    {
+      "epoch": 7.805271998966274,
+      "learning_rate": 1.513807856865033e-07,
+      "loss": 2.6198,
+      "step": 30200
+    },
+    {
+      "epoch": 7.810440625403799,
+      "learning_rate": 1.5134837287696096e-07,
+      "loss": 2.6023,
+      "step": 30220
+    },
+    {
+      "epoch": 7.815609251841323,
+      "learning_rate": 1.5131596006741865e-07,
+      "loss": 2.5248,
+      "step": 30240
+    },
+    {
+      "epoch": 7.8207778782788475,
+      "learning_rate": 1.5128354725787631e-07,
+      "loss": 2.5971,
+      "step": 30260
+    },
+    {
+      "epoch": 7.825946504716372,
+      "learning_rate": 1.5125113444833398e-07,
+      "loss": 2.5428,
+      "step": 30280
+    },
+    {
+      "epoch": 7.831115131153896,
+      "learning_rate": 1.5121872163879164e-07,
+      "loss": 2.5757,
+      "step": 30300
+    },
+    {
+      "epoch": 7.83628375759142,
+      "learning_rate": 1.5118630882924933e-07,
+      "loss": 2.561,
+      "step": 30320
+    },
+    {
+      "epoch": 7.841452384028944,
+      "learning_rate": 1.51153896019707e-07,
+      "loss": 2.5586,
+      "step": 30340
+    },
+    {
+      "epoch": 7.846621010466468,
+      "learning_rate": 1.5112148321016465e-07,
+      "loss": 2.612,
+      "step": 30360
+    },
+    {
+      "epoch": 7.851789636903993,
+      "learning_rate": 1.510890704006223e-07,
+      "loss": 2.5565,
+      "step": 30380
+    },
+    {
+      "epoch": 7.856958263341517,
+      "learning_rate": 1.5105665759108e-07,
+      "loss": 2.5571,
+      "step": 30400
+    },
+    {
+      "epoch": 7.8621268897790415,
+      "learning_rate": 1.5102424478153766e-07,
+      "loss": 2.5895,
+      "step": 30420
+    },
+    {
+      "epoch": 7.867295516216565,
+      "learning_rate": 1.5099183197199532e-07,
+      "loss": 2.5987,
+      "step": 30440
+    },
+    {
+      "epoch": 7.872464142654089,
+      "learning_rate": 1.5095941916245298e-07,
+      "loss": 2.5516,
+      "step": 30460
+    },
+    {
+      "epoch": 7.877632769091614,
+      "learning_rate": 1.5092700635291067e-07,
+      "loss": 2.5649,
+      "step": 30480
+    },
+    {
+      "epoch": 7.882801395529138,
+      "learning_rate": 1.5089459354336834e-07,
+      "loss": 2.5498,
+      "step": 30500
+    },
+    {
+      "epoch": 7.887970021966662,
+      "learning_rate": 1.50862180733826e-07,
+      "loss": 2.6242,
+      "step": 30520
+    },
+    {
+      "epoch": 7.893138648404187,
+      "learning_rate": 1.5082976792428366e-07,
+      "loss": 2.5774,
+      "step": 30540
+    },
+    {
+      "epoch": 7.898307274841711,
+      "learning_rate": 1.5079735511474135e-07,
+      "loss": 2.5298,
+      "step": 30560
+    },
+    {
+      "epoch": 7.903475901279235,
+      "learning_rate": 1.50764942305199e-07,
+      "loss": 2.55,
+      "step": 30580
+    },
+    {
+      "epoch": 7.908644527716759,
+      "learning_rate": 1.5073252949565667e-07,
+      "loss": 2.5924,
+      "step": 30600
+    },
+    {
+      "epoch": 7.913813154154283,
+      "learning_rate": 1.5070011668611436e-07,
+      "loss": 2.515,
+      "step": 30620
+    },
+    {
+      "epoch": 7.918981780591808,
+      "learning_rate": 1.5066770387657202e-07,
+      "loss": 2.5605,
+      "step": 30640
+    },
+    {
+      "epoch": 7.924150407029332,
+      "learning_rate": 1.5063529106702968e-07,
+      "loss": 2.6101,
+      "step": 30660
+    },
+    {
+      "epoch": 7.929319033466856,
+      "learning_rate": 1.5060287825748734e-07,
+      "loss": 2.5612,
+      "step": 30680
+    },
+    {
+      "epoch": 7.934487659904381,
+      "learning_rate": 1.5057046544794503e-07,
+      "loss": 2.6157,
+      "step": 30700
+    },
+    {
+      "epoch": 7.939656286341904,
+      "learning_rate": 1.505380526384027e-07,
+      "loss": 2.5566,
+      "step": 30720
+    },
+    {
+      "epoch": 7.944824912779429,
+      "learning_rate": 1.5050563982886036e-07,
+      "loss": 2.5501,
+      "step": 30740
+    },
+    {
+      "epoch": 7.949993539216953,
+      "learning_rate": 1.5047322701931802e-07,
+      "loss": 2.5918,
+      "step": 30760
+    },
+    {
+      "epoch": 7.955162165654477,
+      "learning_rate": 1.504408142097757e-07,
+      "loss": 2.5722,
+      "step": 30780
+    },
+    {
+      "epoch": 7.960330792092002,
+      "learning_rate": 1.5040840140023337e-07,
+      "loss": 2.5233,
+      "step": 30800
+    },
+    {
+      "epoch": 7.965499418529526,
+      "learning_rate": 1.5037598859069103e-07,
+      "loss": 2.5999,
+      "step": 30820
+    },
+    {
+      "epoch": 7.97066804496705,
+      "learning_rate": 1.503435757811487e-07,
+      "loss": 2.5356,
+      "step": 30840
+    },
+    {
+      "epoch": 7.975836671404574,
+      "learning_rate": 1.5031116297160638e-07,
+      "loss": 2.5524,
+      "step": 30860
+    },
+    {
+      "epoch": 7.981005297842098,
+      "learning_rate": 1.5027875016206404e-07,
+      "loss": 2.6066,
+      "step": 30880
+    },
+    {
+      "epoch": 7.986173924279623,
+      "learning_rate": 1.502463373525217e-07,
+      "loss": 2.5563,
+      "step": 30900
+    },
+    {
+      "epoch": 7.991342550717147,
+      "learning_rate": 1.502139245429794e-07,
+      "loss": 2.5131,
+      "step": 30920
+    },
+    {
+      "epoch": 7.996511177154671,
+      "learning_rate": 1.5018151173343705e-07,
+      "loss": 2.5943,
+      "step": 30940
+    },
+    {
+      "epoch": 7.999870784339062,
+      "eval_bleu": 8.0263,
+      "eval_gen_len": 40.7538,
+      "eval_loss": 2.5299909114837646,
+      "eval_runtime": 863.7677,
+      "eval_samples_per_second": 1.994,
+      "eval_steps_per_second": 0.997,
+      "step": 30953
+    },
+    {
+      "epoch": 8.001679803592195,
+      "learning_rate": 1.5014909892389472e-07,
+      "loss": 2.5393,
+      "step": 30960
+    },
+    {
+      "epoch": 8.00684843002972,
+      "learning_rate": 1.5011668611435238e-07,
+      "loss": 2.4841,
+      "step": 30980
+    },
+    {
+      "epoch": 8.012017056467243,
+      "learning_rate": 1.5008427330481007e-07,
+      "loss": 2.5913,
+      "step": 31000
+    },
+    {
+      "epoch": 8.017185682904769,
+      "learning_rate": 1.5005186049526773e-07,
+      "loss": 2.5053,
+      "step": 31020
+    },
+    {
+      "epoch": 8.022354309342292,
+      "learning_rate": 1.500194476857254e-07,
+      "loss": 2.5612,
+      "step": 31040
+    },
+    {
+      "epoch": 8.027522935779816,
+      "learning_rate": 1.4998703487618305e-07,
+      "loss": 2.5478,
+      "step": 31060
+    },
+    {
+      "epoch": 8.032691562217341,
+      "learning_rate": 1.4995462206664074e-07,
+      "loss": 2.5783,
+      "step": 31080
+    },
+    {
+      "epoch": 8.037860188654864,
+      "learning_rate": 1.499222092570984e-07,
+      "loss": 2.5484,
+      "step": 31100
+    },
+    {
+      "epoch": 8.04302881509239,
+      "learning_rate": 1.4988979644755606e-07,
+      "loss": 2.6095,
+      "step": 31120
+    },
+    {
+      "epoch": 8.048197441529913,
+      "learning_rate": 1.4985738363801373e-07,
+      "loss": 2.5052,
+      "step": 31140
+    },
+    {
+      "epoch": 8.053366067967438,
+      "learning_rate": 1.4982497082847141e-07,
+      "loss": 2.4753,
+      "step": 31160
+    },
+    {
+      "epoch": 8.058534694404962,
+      "learning_rate": 1.4979255801892908e-07,
+      "loss": 2.5393,
+      "step": 31180
+    },
+    {
+      "epoch": 8.063703320842485,
+      "learning_rate": 1.4976014520938674e-07,
+      "loss": 2.5935,
+      "step": 31200
+    },
+    {
+      "epoch": 8.06887194728001,
+      "learning_rate": 1.497277323998444e-07,
+      "loss": 2.5911,
+      "step": 31220
+    },
+    {
+      "epoch": 8.074040573717534,
+      "learning_rate": 1.496953195903021e-07,
+      "loss": 2.572,
+      "step": 31240
+    },
+    {
+      "epoch": 8.07920920015506,
+      "learning_rate": 1.4966290678075975e-07,
+      "loss": 2.5707,
+      "step": 31260
+    },
+    {
+      "epoch": 8.084377826592583,
+      "learning_rate": 1.496304939712174e-07,
+      "loss": 2.5847,
+      "step": 31280
+    },
+    {
+      "epoch": 8.089546453030108,
+      "learning_rate": 1.4959808116167507e-07,
+      "loss": 2.5486,
+      "step": 31300
+    },
+    {
+      "epoch": 8.094715079467631,
+      "learning_rate": 1.4956566835213276e-07,
+      "loss": 2.5432,
+      "step": 31320
+    },
+    {
+      "epoch": 8.099883705905155,
+      "learning_rate": 1.4953325554259042e-07,
+      "loss": 2.5547,
+      "step": 31340
+    },
+    {
+      "epoch": 8.10505233234268,
+      "learning_rate": 1.4950084273304809e-07,
+      "loss": 2.5349,
+      "step": 31360
+    },
+    {
+      "epoch": 8.110220958780204,
+      "learning_rate": 1.4946842992350577e-07,
+      "loss": 2.5603,
+      "step": 31380
+    },
+    {
+      "epoch": 8.115389585217729,
+      "learning_rate": 1.4943601711396344e-07,
+      "loss": 2.5763,
+      "step": 31400
+    },
+    {
+      "epoch": 8.120558211655252,
+      "learning_rate": 1.494036043044211e-07,
+      "loss": 2.5194,
+      "step": 31420
+    },
+    {
+      "epoch": 8.125726838092778,
+      "learning_rate": 1.4937119149487876e-07,
+      "loss": 2.5414,
+      "step": 31440
+    },
+    {
+      "epoch": 8.130895464530301,
+      "learning_rate": 1.4933877868533645e-07,
+      "loss": 2.5459,
+      "step": 31460
+    },
+    {
+      "epoch": 8.136064090967825,
+      "learning_rate": 1.493063658757941e-07,
+      "loss": 2.5307,
+      "step": 31480
+    },
+    {
+      "epoch": 8.14123271740535,
+      "learning_rate": 1.4927395306625177e-07,
+      "loss": 2.4837,
+      "step": 31500
+    },
+    {
+      "epoch": 8.146401343842873,
+      "learning_rate": 1.4924154025670943e-07,
+      "loss": 2.5149,
+      "step": 31520
+    },
+    {
+      "epoch": 8.151569970280399,
+      "learning_rate": 1.4920912744716712e-07,
+      "loss": 2.4708,
+      "step": 31540
+    },
+    {
+      "epoch": 8.156738596717922,
+      "learning_rate": 1.4917671463762478e-07,
+      "loss": 2.5476,
+      "step": 31560
+    },
+    {
+      "epoch": 8.161907223155447,
+      "learning_rate": 1.4914430182808245e-07,
+      "loss": 2.5492,
+      "step": 31580
+    },
+    {
+      "epoch": 8.16707584959297,
+      "learning_rate": 1.491118890185401e-07,
+      "loss": 2.5697,
+      "step": 31600
+    },
+    {
+      "epoch": 8.172244476030494,
+      "learning_rate": 1.490794762089978e-07,
+      "loss": 2.5333,
+      "step": 31620
+    },
+    {
+      "epoch": 8.17741310246802,
+      "learning_rate": 1.4904706339945546e-07,
+      "loss": 2.5197,
+      "step": 31640
+    },
+    {
+      "epoch": 8.182581728905543,
+      "learning_rate": 1.4901465058991312e-07,
+      "loss": 2.4984,
+      "step": 31660
+    },
+    {
+      "epoch": 8.187750355343068,
+      "learning_rate": 1.489822377803708e-07,
+      "loss": 2.5087,
+      "step": 31680
+    },
+    {
+      "epoch": 8.192918981780592,
+      "learning_rate": 1.4894982497082847e-07,
+      "loss": 2.5842,
+      "step": 31700
+    },
+    {
+      "epoch": 8.198087608218117,
+      "learning_rate": 1.4891741216128613e-07,
+      "loss": 2.5249,
+      "step": 31720
+    },
+    {
+      "epoch": 8.20325623465564,
+      "learning_rate": 1.488849993517438e-07,
+      "loss": 2.5729,
+      "step": 31740
+    },
+    {
+      "epoch": 8.208424861093164,
+      "learning_rate": 1.4885258654220148e-07,
+      "loss": 2.5038,
+      "step": 31760
+    },
+    {
+      "epoch": 8.213593487530689,
+      "learning_rate": 1.4882017373265914e-07,
+      "loss": 2.512,
+      "step": 31780
+    },
+    {
+      "epoch": 8.218762113968213,
+      "learning_rate": 1.487877609231168e-07,
+      "loss": 2.6194,
+      "step": 31800
+    },
+    {
+      "epoch": 8.223930740405738,
+      "learning_rate": 1.4875534811357447e-07,
+      "loss": 2.5354,
+      "step": 31820
+    },
+    {
+      "epoch": 8.229099366843261,
+      "learning_rate": 1.4872293530403216e-07,
+      "loss": 2.577,
+      "step": 31840
+    },
+    {
+      "epoch": 8.234267993280785,
+      "learning_rate": 1.4869052249448982e-07,
+      "loss": 2.5439,
+      "step": 31860
+    },
+    {
+      "epoch": 8.23943661971831,
+      "learning_rate": 1.4865810968494748e-07,
+      "loss": 2.4729,
+      "step": 31880
+    },
+    {
+      "epoch": 8.244605246155833,
+      "learning_rate": 1.4862569687540514e-07,
+      "loss": 2.5979,
+      "step": 31900
+    },
+    {
+      "epoch": 8.249773872593359,
+      "learning_rate": 1.4859328406586283e-07,
+      "loss": 2.5676,
+      "step": 31920
+    },
+    {
+      "epoch": 8.254942499030882,
+      "learning_rate": 1.485608712563205e-07,
+      "loss": 2.5859,
+      "step": 31940
+    },
+    {
+      "epoch": 8.260111125468407,
+      "learning_rate": 1.4852845844677815e-07,
+      "loss": 2.558,
+      "step": 31960
+    },
+    {
+      "epoch": 8.265279751905931,
+      "learning_rate": 1.4849604563723584e-07,
+      "loss": 2.5678,
+      "step": 31980
+    },
+    {
+      "epoch": 8.270448378343456,
+      "learning_rate": 1.484636328276935e-07,
+      "loss": 2.5299,
+      "step": 32000
+    },
+    {
+      "epoch": 8.27561700478098,
+      "learning_rate": 1.4843122001815117e-07,
+      "loss": 2.5439,
+      "step": 32020
+    },
+    {
+      "epoch": 8.280785631218503,
+      "learning_rate": 1.4839880720860883e-07,
+      "loss": 2.4766,
+      "step": 32040
+    },
+    {
+      "epoch": 8.285954257656028,
+      "learning_rate": 1.4836639439906652e-07,
+      "loss": 2.5289,
+      "step": 32060
+    },
+    {
+      "epoch": 8.291122884093552,
+      "learning_rate": 1.4833398158952418e-07,
+      "loss": 2.5708,
+      "step": 32080
+    },
+    {
+      "epoch": 8.296291510531077,
+      "learning_rate": 1.4830156877998184e-07,
+      "loss": 2.4981,
+      "step": 32100
+    },
+    {
+      "epoch": 8.3014601369686,
+      "learning_rate": 1.482691559704395e-07,
+      "loss": 2.5352,
+      "step": 32120
+    },
+    {
+      "epoch": 8.306628763406124,
+      "learning_rate": 1.482367431608972e-07,
+      "loss": 2.558,
+      "step": 32140
+    },
+    {
+      "epoch": 8.31179738984365,
+      "learning_rate": 1.4820433035135485e-07,
+      "loss": 2.503,
+      "step": 32160
+    },
+    {
+      "epoch": 8.316966016281173,
+      "learning_rate": 1.4817191754181251e-07,
+      "loss": 2.5792,
+      "step": 32180
+    },
+    {
+      "epoch": 8.322134642718698,
+      "learning_rate": 1.4813950473227018e-07,
+      "loss": 2.5312,
+      "step": 32200
+    },
+    {
+      "epoch": 8.327303269156221,
+      "learning_rate": 1.4810709192272786e-07,
+      "loss": 2.5503,
+      "step": 32220
+    },
+    {
+      "epoch": 8.332471895593747,
+      "learning_rate": 1.4807467911318553e-07,
+      "loss": 2.5442,
+      "step": 32240
+    },
+    {
+      "epoch": 8.33764052203127,
+      "learning_rate": 1.480422663036432e-07,
+      "loss": 2.5578,
+      "step": 32260
+    },
+    {
+      "epoch": 8.342809148468794,
+      "learning_rate": 1.4800985349410088e-07,
+      "loss": 2.5253,
+      "step": 32280
+    },
+    {
+      "epoch": 8.347977774906319,
+      "learning_rate": 1.4797744068455854e-07,
+      "loss": 2.4777,
+      "step": 32300
+    },
+    {
+      "epoch": 8.353146401343842,
+      "learning_rate": 1.479450278750162e-07,
+      "loss": 2.525,
+      "step": 32320
+    },
+    {
+      "epoch": 8.358315027781368,
+      "learning_rate": 1.4791261506547386e-07,
+      "loss": 2.5804,
+      "step": 32340
+    },
+    {
+      "epoch": 8.363483654218891,
+      "learning_rate": 1.4788020225593155e-07,
+      "loss": 2.5387,
+      "step": 32360
+    },
+    {
+      "epoch": 8.368652280656416,
+      "learning_rate": 1.478477894463892e-07,
+      "loss": 2.5867,
+      "step": 32380
+    },
+    {
+      "epoch": 8.37382090709394,
+      "learning_rate": 1.4781537663684687e-07,
+      "loss": 2.5073,
+      "step": 32400
+    },
+    {
+      "epoch": 8.378989533531463,
+      "learning_rate": 1.4778296382730454e-07,
+      "loss": 2.5671,
+      "step": 32420
+    },
+    {
+      "epoch": 8.384158159968988,
+      "learning_rate": 1.4775055101776222e-07,
+      "loss": 2.5598,
+      "step": 32440
+    },
+    {
+      "epoch": 8.389326786406512,
+      "learning_rate": 1.4771813820821989e-07,
+      "loss": 2.552,
+      "step": 32460
+    },
+    {
+      "epoch": 8.394495412844037,
+      "learning_rate": 1.4768572539867755e-07,
+      "loss": 2.5682,
+      "step": 32480
+    },
+    {
+      "epoch": 8.39966403928156,
+      "learning_rate": 1.476533125891352e-07,
+      "loss": 2.5659,
+      "step": 32500
+    },
+    {
+      "epoch": 8.404832665719086,
+      "learning_rate": 1.476208997795929e-07,
+      "loss": 2.5365,
+      "step": 32520
+    },
+    {
+      "epoch": 8.41000129215661,
+      "learning_rate": 1.4758848697005056e-07,
+      "loss": 2.5348,
+      "step": 32540
+    },
+    {
+      "epoch": 8.415169918594133,
+      "learning_rate": 1.4755607416050822e-07,
+      "loss": 2.5808,
+      "step": 32560
+    },
+    {
+      "epoch": 8.420338545031658,
+      "learning_rate": 1.475236613509659e-07,
+      "loss": 2.5285,
+      "step": 32580
+    },
+    {
+      "epoch": 8.425507171469182,
+      "learning_rate": 1.4749124854142357e-07,
+      "loss": 2.5493,
+      "step": 32600
+    },
+    {
+      "epoch": 8.430675797906707,
+      "learning_rate": 1.4745883573188123e-07,
+      "loss": 2.5594,
+      "step": 32620
+    },
+    {
+      "epoch": 8.43584442434423,
+      "learning_rate": 1.474264229223389e-07,
+      "loss": 2.5149,
+      "step": 32640
+    },
+    {
+      "epoch": 8.441013050781756,
+      "learning_rate": 1.4739401011279656e-07,
+      "loss": 2.5238,
+      "step": 32660
+    },
+    {
+      "epoch": 8.446181677219279,
+      "learning_rate": 1.4736159730325425e-07,
+      "loss": 2.5444,
+      "step": 32680
+    },
+    {
+      "epoch": 8.451350303656803,
+      "learning_rate": 1.473291844937119e-07,
+      "loss": 2.5498,
+      "step": 32700
+    },
+    {
+      "epoch": 8.456518930094328,
+      "learning_rate": 1.4729677168416957e-07,
+      "loss": 2.5182,
+      "step": 32720
+    },
+    {
+      "epoch": 8.461687556531851,
+      "learning_rate": 1.4726435887462726e-07,
+      "loss": 2.5588,
+      "step": 32740
+    },
+    {
+      "epoch": 8.466856182969376,
+      "learning_rate": 1.4723194606508492e-07,
+      "loss": 2.5302,
+      "step": 32760
+    },
+    {
+      "epoch": 8.4720248094069,
+      "learning_rate": 1.4719953325554258e-07,
+      "loss": 2.5296,
+      "step": 32780
+    },
+    {
+      "epoch": 8.477193435844425,
+      "learning_rate": 1.4716712044600024e-07,
+      "loss": 2.5676,
+      "step": 32800
+    },
+    {
+      "epoch": 8.482362062281949,
+      "learning_rate": 1.471347076364579e-07,
+      "loss": 2.567,
+      "step": 32820
+    },
+    {
+      "epoch": 8.487530688719472,
+      "learning_rate": 1.471022948269156e-07,
+      "loss": 2.5531,
+      "step": 32840
+    },
+    {
+      "epoch": 8.492699315156997,
+      "learning_rate": 1.4706988201737326e-07,
+      "loss": 2.6131,
+      "step": 32860
+    },
+    {
+      "epoch": 8.49786794159452,
+      "learning_rate": 1.4703746920783092e-07,
+      "loss": 2.5818,
+      "step": 32880
+    },
+    {
+      "epoch": 8.503036568032046,
+      "learning_rate": 1.470050563982886e-07,
+      "loss": 2.542,
+      "step": 32900
+    },
+    {
+      "epoch": 8.50820519446957,
+      "learning_rate": 1.4697264358874627e-07,
+      "loss": 2.5656,
+      "step": 32920
+    },
+    {
+      "epoch": 8.513373820907095,
+      "learning_rate": 1.4694023077920393e-07,
+      "loss": 2.5548,
+      "step": 32940
+    },
+    {
+      "epoch": 8.518542447344618,
+      "learning_rate": 1.4690781796966162e-07,
+      "loss": 2.5295,
+      "step": 32960
+    },
+    {
+      "epoch": 8.523711073782142,
+      "learning_rate": 1.4687540516011925e-07,
+      "loss": 2.4961,
+      "step": 32980
+    },
+    {
+      "epoch": 8.528879700219667,
+      "learning_rate": 1.4684299235057694e-07,
+      "loss": 2.5361,
+      "step": 33000
+    },
+    {
+      "epoch": 8.53404832665719,
+      "learning_rate": 1.468105795410346e-07,
+      "loss": 2.5244,
+      "step": 33020
+    },
+    {
+      "epoch": 8.539216953094716,
+      "learning_rate": 1.4677816673149227e-07,
+      "loss": 2.4984,
+      "step": 33040
+    },
+    {
+      "epoch": 8.54438557953224,
+      "learning_rate": 1.4674575392194995e-07,
+      "loss": 2.5107,
+      "step": 33060
+    },
+    {
+      "epoch": 8.549554205969763,
+      "learning_rate": 1.4671334111240762e-07,
+      "loss": 2.6018,
+      "step": 33080
+    },
+    {
+      "epoch": 8.554722832407288,
+      "learning_rate": 1.4668092830286528e-07,
+      "loss": 2.5481,
+      "step": 33100
+    },
+    {
+      "epoch": 8.559891458844811,
+      "learning_rate": 1.4664851549332297e-07,
+      "loss": 2.5671,
+      "step": 33120
+    },
+    {
+      "epoch": 8.565060085282337,
+      "learning_rate": 1.466161026837806e-07,
+      "loss": 2.5587,
+      "step": 33140
+    },
+    {
+      "epoch": 8.57022871171986,
+      "learning_rate": 1.465836898742383e-07,
+      "loss": 2.4905,
+      "step": 33160
+    },
+    {
+      "epoch": 8.575397338157385,
+      "learning_rate": 1.4655127706469598e-07,
+      "loss": 2.5159,
+      "step": 33180
+    },
+    {
+      "epoch": 8.580565964594909,
+      "learning_rate": 1.4651886425515361e-07,
+      "loss": 2.5419,
+      "step": 33200
+    },
+    {
+      "epoch": 8.585734591032432,
+      "learning_rate": 1.464864514456113e-07,
+      "loss": 2.5811,
+      "step": 33220
+    },
+    {
+      "epoch": 8.590903217469958,
+      "learning_rate": 1.4645403863606896e-07,
+      "loss": 2.5175,
+      "step": 33240
+    },
+    {
+      "epoch": 8.596071843907481,
+      "learning_rate": 1.4642162582652663e-07,
+      "loss": 2.553,
+      "step": 33260
+    },
+    {
+      "epoch": 8.601240470345006,
+      "learning_rate": 1.4638921301698431e-07,
+      "loss": 2.5869,
+      "step": 33280
+    },
+    {
+      "epoch": 8.60640909678253,
+      "learning_rate": 1.4635680020744198e-07,
+      "loss": 2.5046,
+      "step": 33300
+    },
+    {
+      "epoch": 8.611577723220055,
+      "learning_rate": 1.4632438739789964e-07,
+      "loss": 2.5389,
+      "step": 33320
+    },
+    {
+      "epoch": 8.616746349657578,
+      "learning_rate": 1.4629197458835733e-07,
+      "loss": 2.5497,
+      "step": 33340
+    },
+    {
+      "epoch": 8.621914976095102,
+      "learning_rate": 1.4625956177881496e-07,
+      "loss": 2.5344,
+      "step": 33360
+    },
+    {
+      "epoch": 8.627083602532627,
+      "learning_rate": 1.4622714896927265e-07,
+      "loss": 2.6,
+      "step": 33380
+    },
+    {
+      "epoch": 8.63225222897015,
+      "learning_rate": 1.461947361597303e-07,
+      "loss": 2.5658,
+      "step": 33400
+    },
+    {
+      "epoch": 8.637420855407676,
+      "learning_rate": 1.4616232335018797e-07,
+      "loss": 2.539,
+      "step": 33420
+    },
+    {
+      "epoch": 8.6425894818452,
+      "learning_rate": 1.4612991054064566e-07,
+      "loss": 2.5898,
+      "step": 33440
+    },
+    {
+      "epoch": 8.647758108282725,
+      "learning_rate": 1.4609749773110332e-07,
+      "loss": 2.4808,
+      "step": 33460
+    },
+    {
+      "epoch": 8.652926734720248,
+      "learning_rate": 1.4606508492156099e-07,
+      "loss": 2.4885,
+      "step": 33480
+    },
+    {
+      "epoch": 8.658095361157772,
+      "learning_rate": 1.4603267211201867e-07,
+      "loss": 2.5437,
+      "step": 33500
+    },
+    {
+      "epoch": 8.663263987595297,
+      "learning_rate": 1.4600025930247634e-07,
+      "loss": 2.5715,
+      "step": 33520
+    },
+    {
+      "epoch": 8.66843261403282,
+      "learning_rate": 1.45967846492934e-07,
+      "loss": 2.5828,
+      "step": 33540
+    },
+    {
+      "epoch": 8.673601240470346,
+      "learning_rate": 1.4593543368339169e-07,
+      "loss": 2.4961,
+      "step": 33560
+    },
+    {
+      "epoch": 8.678769866907869,
+      "learning_rate": 1.4590302087384932e-07,
+      "loss": 2.5212,
+      "step": 33580
+    },
+    {
+      "epoch": 8.683938493345394,
+      "learning_rate": 1.45870608064307e-07,
+      "loss": 2.5321,
+      "step": 33600
+    },
+    {
+      "epoch": 8.689107119782918,
+      "learning_rate": 1.4583819525476467e-07,
+      "loss": 2.5146,
+      "step": 33620
+    },
+    {
+      "epoch": 8.694275746220441,
+      "learning_rate": 1.4580578244522233e-07,
+      "loss": 2.5531,
+      "step": 33640
+    },
+    {
+      "epoch": 8.699444372657966,
+      "learning_rate": 1.4577336963568002e-07,
+      "loss": 2.4989,
+      "step": 33660
+    },
+    {
+      "epoch": 8.70461299909549,
+      "learning_rate": 1.4574095682613768e-07,
+      "loss": 2.5502,
+      "step": 33680
+    },
+    {
+      "epoch": 8.709781625533015,
+      "learning_rate": 1.4570854401659535e-07,
+      "loss": 2.4926,
+      "step": 33700
+    },
+    {
+      "epoch": 8.714950251970539,
+      "learning_rate": 1.4567613120705303e-07,
+      "loss": 2.5462,
+      "step": 33720
+    },
+    {
+      "epoch": 8.720118878408064,
+      "learning_rate": 1.4564371839751067e-07,
+      "loss": 2.5373,
+      "step": 33740
+    },
+    {
+      "epoch": 8.725287504845587,
+      "learning_rate": 1.4561130558796836e-07,
+      "loss": 2.5745,
+      "step": 33760
+    },
+    {
+      "epoch": 8.73045613128311,
+      "learning_rate": 1.4557889277842605e-07,
+      "loss": 2.5468,
+      "step": 33780
+    },
+    {
+      "epoch": 8.735624757720636,
+      "learning_rate": 1.4554647996888368e-07,
+      "loss": 2.5654,
+      "step": 33800
+    },
+    {
+      "epoch": 8.74079338415816,
+      "learning_rate": 1.4551406715934137e-07,
+      "loss": 2.5433,
+      "step": 33820
+    },
+    {
+      "epoch": 8.745962010595685,
+      "learning_rate": 1.4548165434979903e-07,
+      "loss": 2.5865,
+      "step": 33840
+    },
+    {
+      "epoch": 8.751130637033208,
+      "learning_rate": 1.454492415402567e-07,
+      "loss": 2.5631,
+      "step": 33860
+    },
+    {
+      "epoch": 8.756299263470734,
+      "learning_rate": 1.4541682873071438e-07,
+      "loss": 2.5857,
+      "step": 33880
+    },
+    {
+      "epoch": 8.761467889908257,
+      "learning_rate": 1.4538441592117204e-07,
+      "loss": 2.5168,
+      "step": 33900
+    },
+    {
+      "epoch": 8.76663651634578,
+      "learning_rate": 1.453520031116297e-07,
+      "loss": 2.5316,
+      "step": 33920
+    },
+    {
+      "epoch": 8.771805142783306,
+      "learning_rate": 1.453195903020874e-07,
+      "loss": 2.5464,
+      "step": 33940
+    },
+    {
+      "epoch": 8.77697376922083,
+      "learning_rate": 1.4528717749254503e-07,
+      "loss": 2.5508,
+      "step": 33960
+    },
+    {
+      "epoch": 8.782142395658354,
+      "learning_rate": 1.4525476468300272e-07,
+      "loss": 2.5082,
+      "step": 33980
+    },
+    {
+      "epoch": 8.787311022095878,
+      "learning_rate": 1.4522235187346038e-07,
+      "loss": 2.502,
+      "step": 34000
+    },
+    {
+      "epoch": 8.792479648533401,
+      "learning_rate": 1.4518993906391804e-07,
+      "loss": 2.5586,
+      "step": 34020
+    },
+    {
+      "epoch": 8.797648274970927,
+      "learning_rate": 1.4515752625437573e-07,
+      "loss": 2.5555,
+      "step": 34040
+    },
+    {
+      "epoch": 8.80281690140845,
+      "learning_rate": 1.451251134448334e-07,
+      "loss": 2.5683,
+      "step": 34060
+    },
+    {
+      "epoch": 8.807985527845975,
+      "learning_rate": 1.4509270063529105e-07,
+      "loss": 2.5462,
+      "step": 34080
+    },
+    {
+      "epoch": 8.813154154283499,
+      "learning_rate": 1.4506028782574874e-07,
+      "loss": 2.4885,
+      "step": 34100
+    },
+    {
+      "epoch": 8.818322780721024,
+      "learning_rate": 1.450278750162064e-07,
+      "loss": 2.4856,
+      "step": 34120
+    },
+    {
+      "epoch": 8.823491407158548,
+      "learning_rate": 1.4499546220666407e-07,
+      "loss": 2.538,
+      "step": 34140
+    },
+    {
+      "epoch": 8.828660033596073,
+      "learning_rate": 1.4496304939712175e-07,
+      "loss": 2.5263,
+      "step": 34160
+    },
+    {
+      "epoch": 8.833828660033596,
+      "learning_rate": 1.449306365875794e-07,
+      "loss": 2.5397,
+      "step": 34180
+    },
+    {
+      "epoch": 8.83899728647112,
+      "learning_rate": 1.4489822377803708e-07,
+      "loss": 2.5295,
+      "step": 34200
+    },
+    {
+      "epoch": 8.844165912908645,
+      "learning_rate": 1.4486581096849474e-07,
+      "loss": 2.5853,
+      "step": 34220
+    },
+    {
+      "epoch": 8.849334539346168,
+      "learning_rate": 1.448333981589524e-07,
+      "loss": 2.5566,
+      "step": 34240
+    },
+    {
+      "epoch": 8.854503165783694,
+      "learning_rate": 1.448009853494101e-07,
+      "loss": 2.5319,
+      "step": 34260
+    },
+    {
+      "epoch": 8.859671792221217,
+      "learning_rate": 1.4476857253986775e-07,
+      "loss": 2.4863,
+      "step": 34280
+    },
+    {
+      "epoch": 8.86484041865874,
+      "learning_rate": 1.4473615973032541e-07,
+      "loss": 2.5584,
+      "step": 34300
+    },
+    {
+      "epoch": 8.870009045096266,
+      "learning_rate": 1.447037469207831e-07,
+      "loss": 2.5321,
+      "step": 34320
+    },
+    {
+      "epoch": 8.87517767153379,
+      "learning_rate": 1.4467133411124074e-07,
+      "loss": 2.5304,
+      "step": 34340
+    },
+    {
+      "epoch": 8.880346297971315,
+      "learning_rate": 1.4463892130169843e-07,
+      "loss": 2.5564,
+      "step": 34360
+    },
+    {
+      "epoch": 8.885514924408838,
+      "learning_rate": 1.4460650849215611e-07,
+      "loss": 2.5027,
+      "step": 34380
+    },
+    {
+      "epoch": 8.890683550846363,
+      "learning_rate": 1.4457409568261375e-07,
+      "loss": 2.4922,
+      "step": 34400
+    },
+    {
+      "epoch": 8.895852177283887,
+      "learning_rate": 1.4454168287307144e-07,
+      "loss": 2.5166,
+      "step": 34420
+    },
+    {
+      "epoch": 8.90102080372141,
+      "learning_rate": 1.445092700635291e-07,
+      "loss": 2.527,
+      "step": 34440
+    },
+    {
+      "epoch": 8.906189430158935,
+      "learning_rate": 1.4447685725398676e-07,
+      "loss": 2.5665,
+      "step": 34460
+    },
+    {
+      "epoch": 8.911358056596459,
+      "learning_rate": 1.4444444444444445e-07,
+      "loss": 2.5615,
+      "step": 34480
+    },
+    {
+      "epoch": 8.916526683033984,
+      "learning_rate": 1.444120316349021e-07,
+      "loss": 2.4899,
+      "step": 34500
+    },
+    {
+      "epoch": 8.921695309471508,
+      "learning_rate": 1.4437961882535977e-07,
+      "loss": 2.5012,
+      "step": 34520
+    },
+    {
+      "epoch": 8.926863935909033,
+      "learning_rate": 1.4434720601581746e-07,
+      "loss": 2.5526,
+      "step": 34540
+    },
+    {
+      "epoch": 8.932032562346556,
+      "learning_rate": 1.443147932062751e-07,
+      "loss": 2.5167,
+      "step": 34560
+    },
+    {
+      "epoch": 8.93720118878408,
+      "learning_rate": 1.4428238039673279e-07,
+      "loss": 2.5177,
+      "step": 34580
+    },
+    {
+      "epoch": 8.942369815221605,
+      "learning_rate": 1.4424996758719045e-07,
+      "loss": 2.5548,
+      "step": 34600
+    },
+    {
+      "epoch": 8.947538441659129,
+      "learning_rate": 1.442175547776481e-07,
+      "loss": 2.5742,
+      "step": 34620
+    },
+    {
+      "epoch": 8.952707068096654,
+      "learning_rate": 1.441851419681058e-07,
+      "loss": 2.5149,
+      "step": 34640
+    },
+    {
+      "epoch": 8.957875694534177,
+      "learning_rate": 1.4415272915856346e-07,
+      "loss": 2.5042,
+      "step": 34660
+    },
+    {
+      "epoch": 8.963044320971703,
+      "learning_rate": 1.4412031634902112e-07,
+      "loss": 2.5076,
+      "step": 34680
+    },
+    {
+      "epoch": 8.968212947409226,
+      "learning_rate": 1.440879035394788e-07,
+      "loss": 2.5282,
+      "step": 34700
+    },
+    {
+      "epoch": 8.97338157384675,
+      "learning_rate": 1.4405549072993647e-07,
+      "loss": 2.5373,
+      "step": 34720
+    },
+    {
+      "epoch": 8.978550200284275,
+      "learning_rate": 1.4402307792039413e-07,
+      "loss": 2.5556,
+      "step": 34740
+    },
+    {
+      "epoch": 8.983718826721798,
+      "learning_rate": 1.4399066511085182e-07,
+      "loss": 2.5413,
+      "step": 34760
+    },
+    {
+      "epoch": 8.988887453159323,
+      "learning_rate": 1.4395825230130946e-07,
+      "loss": 2.5563,
+      "step": 34780
+    },
+    {
+      "epoch": 8.994056079596847,
+      "learning_rate": 1.4392583949176715e-07,
+      "loss": 2.5267,
+      "step": 34800
+    },
+    {
+      "epoch": 8.999224706034372,
+      "learning_rate": 1.438934266822248e-07,
+      "loss": 2.5622,
+      "step": 34820
+    },
+    {
+      "epoch": 9.0,
+      "eval_bleu": 10.2353,
+      "eval_gen_len": 38.2607,
+      "eval_loss": 2.504971742630005,
+      "eval_runtime": 830.1656,
+      "eval_samples_per_second": 2.074,
+      "eval_steps_per_second": 1.037,
+      "step": 34823
+    },
+    {
+      "epoch": 9.004393332471896,
+      "learning_rate": 1.4386101387268247e-07,
+      "loss": 2.5627,
+      "step": 34840
+    },
+    {
+      "epoch": 9.00956195890942,
+      "learning_rate": 1.4382860106314016e-07,
+      "loss": 2.538,
+      "step": 34860
+    },
+    {
+      "epoch": 9.014730585346944,
+      "learning_rate": 1.4379618825359782e-07,
+      "loss": 2.5561,
+      "step": 34880
+    },
+    {
+      "epoch": 9.019899211784468,
+      "learning_rate": 1.4376377544405548e-07,
+      "loss": 2.5216,
+      "step": 34900
+    },
+    {
+      "epoch": 9.025067838221993,
+      "learning_rate": 1.4373136263451317e-07,
+      "loss": 2.4699,
+      "step": 34920
+    },
+    {
+      "epoch": 9.030236464659517,
+      "learning_rate": 1.436989498249708e-07,
+      "loss": 2.5566,
+      "step": 34940
+    },
+    {
+      "epoch": 9.035405091097042,
+      "learning_rate": 1.436665370154285e-07,
+      "loss": 2.4698,
+      "step": 34960
+    },
+    {
+      "epoch": 9.040573717534565,
+      "learning_rate": 1.4363412420588616e-07,
+      "loss": 2.4995,
+      "step": 34980
+    },
+    {
+      "epoch": 9.045742343972089,
+      "learning_rate": 1.4360171139634382e-07,
+      "loss": 2.522,
+      "step": 35000
+    },
+    {
+      "epoch": 9.050910970409614,
+      "learning_rate": 1.435692985868015e-07,
+      "loss": 2.5507,
+      "step": 35020
+    },
+    {
+      "epoch": 9.056079596847137,
+      "learning_rate": 1.4353688577725917e-07,
+      "loss": 2.4685,
+      "step": 35040
+    },
+    {
+      "epoch": 9.061248223284663,
+      "learning_rate": 1.4350447296771683e-07,
+      "loss": 2.4728,
+      "step": 35060
+    },
+    {
+      "epoch": 9.066416849722186,
+      "learning_rate": 1.4347206015817452e-07,
+      "loss": 2.6299,
+      "step": 35080
+    },
+    {
+      "epoch": 9.071585476159711,
+      "learning_rate": 1.4343964734863218e-07,
+      "loss": 2.5499,
+      "step": 35100
+    },
+    {
+      "epoch": 9.076754102597235,
+      "learning_rate": 1.4340723453908984e-07,
+      "loss": 2.5478,
+      "step": 35120
+    },
+    {
+      "epoch": 9.081922729034758,
+      "learning_rate": 1.433748217295475e-07,
+      "loss": 2.4566,
+      "step": 35140
+    },
+    {
+      "epoch": 9.087091355472284,
+      "learning_rate": 1.4334240892000516e-07,
+      "loss": 2.4931,
+      "step": 35160
+    },
+    {
+      "epoch": 9.092259981909807,
+      "learning_rate": 1.4330999611046285e-07,
+      "loss": 2.5343,
+      "step": 35180
+    },
+    {
+      "epoch": 9.097428608347332,
+      "learning_rate": 1.4327758330092052e-07,
+      "loss": 2.5651,
+      "step": 35200
+    },
+    {
+      "epoch": 9.102597234784856,
+      "learning_rate": 1.4324517049137818e-07,
+      "loss": 2.5226,
+      "step": 35220
+    },
+    {
+      "epoch": 9.10776586122238,
+      "learning_rate": 1.4321275768183587e-07,
+      "loss": 2.5504,
+      "step": 35240
+    },
+    {
+      "epoch": 9.112934487659905,
+      "learning_rate": 1.4318034487229353e-07,
+      "loss": 2.5127,
+      "step": 35260
+    },
+    {
+      "epoch": 9.118103114097428,
+      "learning_rate": 1.431479320627512e-07,
+      "loss": 2.4964,
+      "step": 35280
+    },
+    {
+      "epoch": 9.123271740534953,
+      "learning_rate": 1.4311551925320885e-07,
+      "loss": 2.5396,
+      "step": 35300
+    },
+    {
+      "epoch": 9.128440366972477,
+      "learning_rate": 1.4308310644366654e-07,
+      "loss": 2.5303,
+      "step": 35320
+    },
+    {
+      "epoch": 9.133608993410002,
+      "learning_rate": 1.430506936341242e-07,
+      "loss": 2.5574,
+      "step": 35340
+    },
+    {
+      "epoch": 9.138777619847525,
+      "learning_rate": 1.4301828082458186e-07,
+      "loss": 2.5649,
+      "step": 35360
+    },
+    {
+      "epoch": 9.143946246285049,
+      "learning_rate": 1.4298586801503952e-07,
+      "loss": 2.5113,
+      "step": 35380
+    },
+    {
+      "epoch": 9.149114872722574,
+      "learning_rate": 1.429534552054972e-07,
+      "loss": 2.4826,
+      "step": 35400
+    },
+    {
+      "epoch": 9.154283499160098,
+      "learning_rate": 1.4292104239595488e-07,
+      "loss": 2.5148,
+      "step": 35420
+    },
+    {
+      "epoch": 9.159452125597623,
+      "learning_rate": 1.4288862958641254e-07,
+      "loss": 2.5568,
+      "step": 35440
+    },
+    {
+      "epoch": 9.164620752035146,
+      "learning_rate": 1.428562167768702e-07,
+      "loss": 2.5475,
+      "step": 35460
+    },
+    {
+      "epoch": 9.169789378472672,
+      "learning_rate": 1.428238039673279e-07,
+      "loss": 2.5335,
+      "step": 35480
+    },
+    {
+      "epoch": 9.174958004910195,
+      "learning_rate": 1.4279139115778555e-07,
+      "loss": 2.494,
+      "step": 35500
+    },
+    {
+      "epoch": 9.180126631347719,
+      "learning_rate": 1.427589783482432e-07,
+      "loss": 2.5568,
+      "step": 35520
+    },
+    {
+      "epoch": 9.185295257785244,
+      "learning_rate": 1.4272656553870087e-07,
+      "loss": 2.5425,
+      "step": 35540
+    },
+    {
+      "epoch": 9.190463884222767,
+      "learning_rate": 1.4269415272915856e-07,
+      "loss": 2.5672,
+      "step": 35560
+    },
+    {
+      "epoch": 9.195632510660293,
+      "learning_rate": 1.4266173991961622e-07,
+      "loss": 2.5532,
+      "step": 35580
+    },
+    {
+      "epoch": 9.200801137097816,
+      "learning_rate": 1.4262932711007388e-07,
+      "loss": 2.4934,
+      "step": 35600
+    },
+    {
+      "epoch": 9.205969763535341,
+      "learning_rate": 1.4259691430053157e-07,
+      "loss": 2.5009,
+      "step": 35620
+    },
+    {
+      "epoch": 9.211138389972865,
+      "learning_rate": 1.4256450149098924e-07,
+      "loss": 2.4935,
+      "step": 35640
+    },
+    {
+      "epoch": 9.216307016410388,
+      "learning_rate": 1.425320886814469e-07,
+      "loss": 2.5217,
+      "step": 35660
+    },
+    {
+      "epoch": 9.221475642847913,
+      "learning_rate": 1.4249967587190456e-07,
+      "loss": 2.5246,
+      "step": 35680
+    },
+    {
+      "epoch": 9.226644269285437,
+      "learning_rate": 1.4246726306236225e-07,
+      "loss": 2.4851,
+      "step": 35700
+    },
+    {
+      "epoch": 9.231812895722962,
+      "learning_rate": 1.424348502528199e-07,
+      "loss": 2.585,
+      "step": 35720
+    },
+    {
+      "epoch": 9.236981522160486,
+      "learning_rate": 1.4240243744327757e-07,
+      "loss": 2.5602,
+      "step": 35740
+    },
+    {
+      "epoch": 9.24215014859801,
+      "learning_rate": 1.4237002463373523e-07,
+      "loss": 2.4916,
+      "step": 35760
+    },
+    {
+      "epoch": 9.247318775035534,
+      "learning_rate": 1.4233761182419292e-07,
+      "loss": 2.584,
+      "step": 35780
+    },
+    {
+      "epoch": 9.252487401473058,
+      "learning_rate": 1.4230519901465058e-07,
+      "loss": 2.5671,
+      "step": 35800
+    },
+    {
+      "epoch": 9.257656027910583,
+      "learning_rate": 1.4227278620510824e-07,
+      "loss": 2.5741,
+      "step": 35820
+    },
+    {
+      "epoch": 9.262824654348107,
+      "learning_rate": 1.422403733955659e-07,
+      "loss": 2.544,
+      "step": 35840
+    },
+    {
+      "epoch": 9.267993280785632,
+      "learning_rate": 1.422079605860236e-07,
+      "loss": 2.5616,
+      "step": 35860
+    },
+    {
+      "epoch": 9.273161907223155,
+      "learning_rate": 1.4217554777648126e-07,
+      "loss": 2.5236,
+      "step": 35880
+    },
+    {
+      "epoch": 9.27833053366068,
+      "learning_rate": 1.4214313496693892e-07,
+      "loss": 2.4909,
+      "step": 35900
+    },
+    {
+      "epoch": 9.283499160098204,
+      "learning_rate": 1.421107221573966e-07,
+      "loss": 2.514,
+      "step": 35920
+    },
+    {
+      "epoch": 9.288667786535727,
+      "learning_rate": 1.4207830934785427e-07,
+      "loss": 2.6055,
+      "step": 35940
+    },
+    {
+      "epoch": 9.293836412973253,
+      "learning_rate": 1.4204589653831193e-07,
+      "loss": 2.5403,
+      "step": 35960
+    },
+    {
+      "epoch": 9.299005039410776,
+      "learning_rate": 1.420134837287696e-07,
+      "loss": 2.4817,
+      "step": 35980
+    },
+    {
+      "epoch": 9.304173665848301,
+      "learning_rate": 1.4198107091922728e-07,
+      "loss": 2.4968,
+      "step": 36000
+    },
+    {
+      "epoch": 9.309342292285825,
+      "learning_rate": 1.4194865810968494e-07,
+      "loss": 2.5205,
+      "step": 36020
+    },
+    {
+      "epoch": 9.31451091872335,
+      "learning_rate": 1.419162453001426e-07,
+      "loss": 2.4925,
+      "step": 36040
+    },
+    {
+      "epoch": 9.319679545160874,
+      "learning_rate": 1.4188383249060027e-07,
+      "loss": 2.4745,
+      "step": 36060
+    },
+    {
+      "epoch": 9.324848171598397,
+      "learning_rate": 1.4185141968105795e-07,
+      "loss": 2.5256,
+      "step": 36080
+    },
+    {
+      "epoch": 9.330016798035922,
+      "learning_rate": 1.4181900687151562e-07,
+      "loss": 2.5286,
+      "step": 36100
+    },
+    {
+      "epoch": 9.335185424473446,
+      "learning_rate": 1.4178659406197328e-07,
+      "loss": 2.5655,
+      "step": 36120
+    },
+    {
+      "epoch": 9.340354050910971,
+      "learning_rate": 1.4175418125243094e-07,
+      "loss": 2.5093,
+      "step": 36140
+    },
+    {
+      "epoch": 9.345522677348495,
+      "learning_rate": 1.4172176844288863e-07,
+      "loss": 2.5257,
+      "step": 36160
+    },
+    {
+      "epoch": 9.350691303786018,
+      "learning_rate": 1.416893556333463e-07,
+      "loss": 2.5267,
+      "step": 36180
+    },
+    {
+      "epoch": 9.355859930223543,
+      "learning_rate": 1.4165694282380395e-07,
+      "loss": 2.4716,
+      "step": 36200
+    },
+    {
+      "epoch": 9.361028556661067,
+      "learning_rate": 1.4162453001426164e-07,
+      "loss": 2.5255,
+      "step": 36220
+    },
+    {
+      "epoch": 9.366197183098592,
+      "learning_rate": 1.415921172047193e-07,
+      "loss": 2.5087,
+      "step": 36240
+    },
+    {
+      "epoch": 9.371365809536115,
+      "learning_rate": 1.4155970439517696e-07,
+      "loss": 2.5494,
+      "step": 36260
+    },
+    {
+      "epoch": 9.37653443597364,
+      "learning_rate": 1.4152729158563463e-07,
+      "loss": 2.5466,
+      "step": 36280
+    },
+    {
+      "epoch": 9.381703062411164,
+      "learning_rate": 1.4149487877609231e-07,
+      "loss": 2.4686,
+      "step": 36300
+    },
+    {
+      "epoch": 9.386871688848688,
+      "learning_rate": 1.4146246596654998e-07,
+      "loss": 2.5361,
+      "step": 36320
+    },
+    {
+      "epoch": 9.392040315286213,
+      "learning_rate": 1.4143005315700764e-07,
+      "loss": 2.4786,
+      "step": 36340
+    },
+    {
+      "epoch": 9.397208941723736,
+      "learning_rate": 1.413976403474653e-07,
+      "loss": 2.5012,
+      "step": 36360
+    },
+    {
+      "epoch": 9.402377568161262,
+      "learning_rate": 1.41365227537923e-07,
+      "loss": 2.5728,
+      "step": 36380
+    },
+    {
+      "epoch": 9.407546194598785,
+      "learning_rate": 1.4133281472838065e-07,
+      "loss": 2.513,
+      "step": 36400
+    },
+    {
+      "epoch": 9.41271482103631,
+      "learning_rate": 1.413004019188383e-07,
+      "loss": 2.5833,
+      "step": 36420
+    },
+    {
+      "epoch": 9.417883447473834,
+      "learning_rate": 1.4126798910929597e-07,
+      "loss": 2.5436,
+      "step": 36440
+    },
+    {
+      "epoch": 9.423052073911357,
+      "learning_rate": 1.4123557629975366e-07,
+      "loss": 2.5017,
+      "step": 36460
+    },
+    {
+      "epoch": 9.428220700348882,
+      "learning_rate": 1.4120316349021132e-07,
+      "loss": 2.4708,
+      "step": 36480
+    },
+    {
+      "epoch": 9.433389326786406,
+      "learning_rate": 1.4117075068066899e-07,
+      "loss": 2.5594,
+      "step": 36500
+    },
+    {
+      "epoch": 9.438557953223931,
+      "learning_rate": 1.4113833787112667e-07,
+      "loss": 2.5095,
+      "step": 36520
+    },
+    {
+      "epoch": 9.443726579661455,
+      "learning_rate": 1.4110592506158434e-07,
+      "loss": 2.5153,
+      "step": 36540
+    },
+    {
+      "epoch": 9.44889520609898,
+      "learning_rate": 1.41073512252042e-07,
+      "loss": 2.4579,
+      "step": 36560
+    },
+    {
+      "epoch": 9.454063832536503,
+      "learning_rate": 1.4104109944249966e-07,
+      "loss": 2.5145,
+      "step": 36580
+    },
+    {
+      "epoch": 9.459232458974027,
+      "learning_rate": 1.4100868663295735e-07,
+      "loss": 2.5117,
+      "step": 36600
+    },
+    {
+      "epoch": 9.464401085411552,
+      "learning_rate": 1.40976273823415e-07,
+      "loss": 2.571,
+      "step": 36620
+    },
+    {
+      "epoch": 9.469569711849076,
+      "learning_rate": 1.4094386101387267e-07,
+      "loss": 2.5494,
+      "step": 36640
+    },
+    {
+      "epoch": 9.4747383382866,
+      "learning_rate": 1.4091144820433033e-07,
+      "loss": 2.4598,
+      "step": 36660
+    },
+    {
+      "epoch": 9.479906964724124,
+      "learning_rate": 1.4087903539478802e-07,
+      "loss": 2.5034,
+      "step": 36680
+    },
+    {
+      "epoch": 9.48507559116165,
+      "learning_rate": 1.4084662258524568e-07,
+      "loss": 2.5229,
+      "step": 36700
+    },
+    {
+      "epoch": 9.490244217599173,
+      "learning_rate": 1.4081420977570335e-07,
+      "loss": 2.5249,
+      "step": 36720
+    },
+    {
+      "epoch": 9.495412844036696,
+      "learning_rate": 1.40781796966161e-07,
+      "loss": 2.4883,
+      "step": 36740
+    },
+    {
+      "epoch": 9.500581470474222,
+      "learning_rate": 1.407493841566187e-07,
+      "loss": 2.5082,
+      "step": 36760
+    },
+    {
+      "epoch": 9.505750096911745,
+      "learning_rate": 1.4071697134707636e-07,
+      "loss": 2.5209,
+      "step": 36780
+    },
+    {
+      "epoch": 9.51091872334927,
+      "learning_rate": 1.4068455853753402e-07,
+      "loss": 2.5263,
+      "step": 36800
+    },
+    {
+      "epoch": 9.516087349786794,
+      "learning_rate": 1.406521457279917e-07,
+      "loss": 2.4904,
+      "step": 36820
+    },
+    {
+      "epoch": 9.52125597622432,
+      "learning_rate": 1.4061973291844937e-07,
+      "loss": 2.511,
+      "step": 36840
+    },
+    {
+      "epoch": 9.526424602661843,
+      "learning_rate": 1.4058732010890703e-07,
+      "loss": 2.4821,
+      "step": 36860
+    },
+    {
+      "epoch": 9.531593229099366,
+      "learning_rate": 1.405549072993647e-07,
+      "loss": 2.5053,
+      "step": 36880
+    },
+    {
+      "epoch": 9.536761855536891,
+      "learning_rate": 1.4052249448982238e-07,
+      "loss": 2.518,
+      "step": 36900
+    },
+    {
+      "epoch": 9.541930481974415,
+      "learning_rate": 1.4049008168028004e-07,
+      "loss": 2.4693,
+      "step": 36920
+    },
+    {
+      "epoch": 9.54709910841194,
+      "learning_rate": 1.404576688707377e-07,
+      "loss": 2.5722,
+      "step": 36940
+    },
+    {
+      "epoch": 9.552267734849464,
+      "learning_rate": 1.4042525606119537e-07,
+      "loss": 2.4733,
+      "step": 36960
+    },
+    {
+      "epoch": 9.557436361286989,
+      "learning_rate": 1.4039284325165306e-07,
+      "loss": 2.5035,
+      "step": 36980
+    },
+    {
+      "epoch": 9.562604987724512,
+      "learning_rate": 1.4036043044211072e-07,
+      "loss": 2.5271,
+      "step": 37000
+    },
+    {
+      "epoch": 9.567773614162036,
+      "learning_rate": 1.4032801763256838e-07,
+      "loss": 2.5335,
+      "step": 37020
+    },
+    {
+      "epoch": 9.572942240599561,
+      "learning_rate": 1.4029560482302604e-07,
+      "loss": 2.4517,
+      "step": 37040
+    },
+    {
+      "epoch": 9.578110867037084,
+      "learning_rate": 1.4026319201348373e-07,
+      "loss": 2.5086,
+      "step": 37060
+    },
+    {
+      "epoch": 9.58327949347461,
+      "learning_rate": 1.402307792039414e-07,
+      "loss": 2.5409,
+      "step": 37080
+    },
+    {
+      "epoch": 9.588448119912133,
+      "learning_rate": 1.4019836639439905e-07,
+      "loss": 2.512,
+      "step": 37100
+    },
+    {
+      "epoch": 9.593616746349657,
+      "learning_rate": 1.4016595358485674e-07,
+      "loss": 2.5074,
+      "step": 37120
+    },
+    {
+      "epoch": 9.598785372787182,
+      "learning_rate": 1.401335407753144e-07,
+      "loss": 2.5078,
+      "step": 37140
+    },
+    {
+      "epoch": 9.603953999224705,
+      "learning_rate": 1.4010112796577207e-07,
+      "loss": 2.5574,
+      "step": 37160
+    },
+    {
+      "epoch": 9.60912262566223,
+      "learning_rate": 1.4006871515622973e-07,
+      "loss": 2.5459,
+      "step": 37180
+    },
+    {
+      "epoch": 9.614291252099754,
+      "learning_rate": 1.4003630234668742e-07,
+      "loss": 2.4473,
+      "step": 37200
+    },
+    {
+      "epoch": 9.61945987853728,
+      "learning_rate": 1.4000388953714508e-07,
+      "loss": 2.5423,
+      "step": 37220
+    },
+    {
+      "epoch": 9.624628504974803,
+      "learning_rate": 1.3997147672760274e-07,
+      "loss": 2.4988,
+      "step": 37240
+    },
+    {
+      "epoch": 9.629797131412328,
+      "learning_rate": 1.399390639180604e-07,
+      "loss": 2.6236,
+      "step": 37260
+    },
+    {
+      "epoch": 9.634965757849852,
+      "learning_rate": 1.399066511085181e-07,
+      "loss": 2.5718,
+      "step": 37280
+    },
+    {
+      "epoch": 9.640134384287375,
+      "learning_rate": 1.3987423829897575e-07,
+      "loss": 2.5012,
+      "step": 37300
+    },
+    {
+      "epoch": 9.6453030107249,
+      "learning_rate": 1.3984182548943341e-07,
+      "loss": 2.5536,
+      "step": 37320
+    },
+    {
+      "epoch": 9.650471637162424,
+      "learning_rate": 1.3980941267989108e-07,
+      "loss": 2.5416,
+      "step": 37340
+    },
+    {
+      "epoch": 9.655640263599949,
+      "learning_rate": 1.3977699987034876e-07,
+      "loss": 2.4864,
+      "step": 37360
+    },
+    {
+      "epoch": 9.660808890037472,
+      "learning_rate": 1.3974458706080643e-07,
+      "loss": 2.5365,
+      "step": 37380
+    },
+    {
+      "epoch": 9.665977516474996,
+      "learning_rate": 1.397121742512641e-07,
+      "loss": 2.5111,
+      "step": 37400
+    },
+    {
+      "epoch": 9.671146142912521,
+      "learning_rate": 1.3967976144172178e-07,
+      "loss": 2.4737,
+      "step": 37420
+    },
+    {
+      "epoch": 9.676314769350045,
+      "learning_rate": 1.3964734863217944e-07,
+      "loss": 2.5373,
+      "step": 37440
+    },
+    {
+      "epoch": 9.68148339578757,
+      "learning_rate": 1.396149358226371e-07,
+      "loss": 2.4991,
+      "step": 37460
+    },
+    {
+      "epoch": 9.686652022225093,
+      "learning_rate": 1.3958252301309476e-07,
+      "loss": 2.5186,
+      "step": 37480
+    },
+    {
+      "epoch": 9.691820648662619,
+      "learning_rate": 1.3955011020355242e-07,
+      "loss": 2.4701,
+      "step": 37500
+    },
+    {
+      "epoch": 9.696989275100142,
+      "learning_rate": 1.395176973940101e-07,
+      "loss": 2.5169,
+      "step": 37520
+    },
+    {
+      "epoch": 9.702157901537666,
+      "learning_rate": 1.3948528458446777e-07,
+      "loss": 2.4691,
+      "step": 37540
+    },
+    {
+      "epoch": 9.70732652797519,
+      "learning_rate": 1.3945287177492544e-07,
+      "loss": 2.5278,
+      "step": 37560
+    },
+    {
+      "epoch": 9.712495154412714,
+      "learning_rate": 1.3942045896538312e-07,
+      "loss": 2.4778,
+      "step": 37580
+    },
+    {
+      "epoch": 9.71766378085024,
+      "learning_rate": 1.3938804615584079e-07,
+      "loss": 2.5367,
+      "step": 37600
+    },
+    {
+      "epoch": 9.722832407287763,
+      "learning_rate": 1.3935563334629845e-07,
+      "loss": 2.5249,
+      "step": 37620
+    },
+    {
+      "epoch": 9.728001033725288,
+      "learning_rate": 1.393232205367561e-07,
+      "loss": 2.503,
+      "step": 37640
+    },
+    {
+      "epoch": 9.733169660162812,
+      "learning_rate": 1.3929080772721377e-07,
+      "loss": 2.514,
+      "step": 37660
+    },
+    {
+      "epoch": 9.738338286600335,
+      "learning_rate": 1.3925839491767146e-07,
+      "loss": 2.5158,
+      "step": 37680
+    },
+    {
+      "epoch": 9.74350691303786,
+      "learning_rate": 1.3922598210812912e-07,
+      "loss": 2.527,
+      "step": 37700
+    },
+    {
+      "epoch": 9.748675539475384,
+      "learning_rate": 1.3919356929858678e-07,
+      "loss": 2.5435,
+      "step": 37720
+    },
+    {
+      "epoch": 9.75384416591291,
+      "learning_rate": 1.3916115648904447e-07,
+      "loss": 2.5464,
+      "step": 37740
+    },
+    {
+      "epoch": 9.759012792350433,
+      "learning_rate": 1.3912874367950213e-07,
+      "loss": 2.5952,
+      "step": 37760
+    },
+    {
+      "epoch": 9.764181418787958,
+      "learning_rate": 1.390963308699598e-07,
+      "loss": 2.5066,
+      "step": 37780
+    },
+    {
+      "epoch": 9.769350045225481,
+      "learning_rate": 1.3906391806041748e-07,
+      "loss": 2.4973,
+      "step": 37800
+    },
+    {
+      "epoch": 9.774518671663005,
+      "learning_rate": 1.3903150525087512e-07,
+      "loss": 2.5344,
+      "step": 37820
+    },
+    {
+      "epoch": 9.77968729810053,
+      "learning_rate": 1.389990924413328e-07,
+      "loss": 2.4959,
+      "step": 37840
+    },
+    {
+      "epoch": 9.784855924538054,
+      "learning_rate": 1.3896667963179047e-07,
+      "loss": 2.5006,
+      "step": 37860
+    },
+    {
+      "epoch": 9.790024550975579,
+      "learning_rate": 1.3893426682224813e-07,
+      "loss": 2.5334,
+      "step": 37880
+    },
+    {
+      "epoch": 9.795193177413102,
+      "learning_rate": 1.3890185401270582e-07,
+      "loss": 2.5255,
+      "step": 37900
+    },
+    {
+      "epoch": 9.800361803850627,
+      "learning_rate": 1.3886944120316348e-07,
+      "loss": 2.5221,
+      "step": 37920
+    },
+    {
+      "epoch": 9.805530430288151,
+      "learning_rate": 1.3883702839362114e-07,
+      "loss": 2.5832,
+      "step": 37940
+    },
+    {
+      "epoch": 9.810699056725674,
+      "learning_rate": 1.3880461558407883e-07,
+      "loss": 2.5268,
+      "step": 37960
+    },
+    {
+      "epoch": 9.8158676831632,
+      "learning_rate": 1.387722027745365e-07,
+      "loss": 2.6145,
+      "step": 37980
+    },
+    {
+      "epoch": 9.821036309600723,
+      "learning_rate": 1.3873978996499416e-07,
+      "loss": 2.5483,
+      "step": 38000
+    },
+    {
+      "epoch": 9.826204936038248,
+      "learning_rate": 1.3870737715545184e-07,
+      "loss": 2.4529,
+      "step": 38020
+    },
+    {
+      "epoch": 9.831373562475772,
+      "learning_rate": 1.3867496434590948e-07,
+      "loss": 2.5547,
+      "step": 38040
+    },
+    {
+      "epoch": 9.836542188913295,
+      "learning_rate": 1.3864255153636717e-07,
+      "loss": 2.4675,
+      "step": 38060
+    },
+    {
+      "epoch": 9.84171081535082,
+      "learning_rate": 1.3861013872682483e-07,
+      "loss": 2.4782,
+      "step": 38080
+    },
+    {
+      "epoch": 9.846879441788344,
+      "learning_rate": 1.385777259172825e-07,
+      "loss": 2.527,
+      "step": 38100
+    },
+    {
+      "epoch": 9.85204806822587,
+      "learning_rate": 1.3854531310774018e-07,
+      "loss": 2.5285,
+      "step": 38120
+    },
+    {
+      "epoch": 9.857216694663393,
+      "learning_rate": 1.3851290029819784e-07,
+      "loss": 2.5057,
+      "step": 38140
+    },
+    {
+      "epoch": 9.862385321100918,
+      "learning_rate": 1.384804874886555e-07,
+      "loss": 2.4832,
+      "step": 38160
+    },
+    {
+      "epoch": 9.867553947538442,
+      "learning_rate": 1.384480746791132e-07,
+      "loss": 2.4826,
+      "step": 38180
+    },
+    {
+      "epoch": 9.872722573975967,
+      "learning_rate": 1.3841566186957083e-07,
+      "loss": 2.4912,
+      "step": 38200
+    },
+    {
+      "epoch": 9.87789120041349,
+      "learning_rate": 1.3838324906002852e-07,
+      "loss": 2.5356,
+      "step": 38220
+    },
+    {
+      "epoch": 9.883059826851014,
+      "learning_rate": 1.3835083625048618e-07,
+      "loss": 2.4311,
+      "step": 38240
+    },
+    {
+      "epoch": 9.888228453288539,
+      "learning_rate": 1.3831842344094384e-07,
+      "loss": 2.4831,
+      "step": 38260
+    },
+    {
+      "epoch": 9.893397079726062,
+      "learning_rate": 1.3828601063140153e-07,
+      "loss": 2.4975,
+      "step": 38280
+    },
+    {
+      "epoch": 9.898565706163588,
+      "learning_rate": 1.382535978218592e-07,
+      "loss": 2.5113,
+      "step": 38300
+    },
+    {
+      "epoch": 9.903734332601111,
+      "learning_rate": 1.3822118501231685e-07,
+      "loss": 2.4731,
+      "step": 38320
+    },
+    {
+      "epoch": 9.908902959038635,
+      "learning_rate": 1.3818877220277454e-07,
+      "loss": 2.5272,
+      "step": 38340
+    },
+    {
+      "epoch": 9.91407158547616,
+      "learning_rate": 1.381563593932322e-07,
+      "loss": 2.5113,
+      "step": 38360
+    },
+    {
+      "epoch": 9.919240211913683,
+      "learning_rate": 1.3812394658368986e-07,
+      "loss": 2.5481,
+      "step": 38380
+    },
+    {
+      "epoch": 9.924408838351209,
+      "learning_rate": 1.3809153377414755e-07,
+      "loss": 2.4876,
+      "step": 38400
+    },
+    {
+      "epoch": 9.929577464788732,
+      "learning_rate": 1.380591209646052e-07,
+      "loss": 2.4875,
+      "step": 38420
+    },
+    {
+      "epoch": 9.934746091226257,
+      "learning_rate": 1.3802670815506288e-07,
+      "loss": 2.4732,
+      "step": 38440
+    },
+    {
+      "epoch": 9.93991471766378,
+      "learning_rate": 1.3799429534552054e-07,
+      "loss": 2.5036,
+      "step": 38460
+    },
+    {
+      "epoch": 9.945083344101306,
+      "learning_rate": 1.379618825359782e-07,
+      "loss": 2.4979,
+      "step": 38480
+    },
+    {
+      "epoch": 9.95025197053883,
+      "learning_rate": 1.379294697264359e-07,
+      "loss": 2.4916,
+      "step": 38500
+    },
+    {
+      "epoch": 9.955420596976353,
+      "learning_rate": 1.3789705691689355e-07,
+      "loss": 2.4518,
+      "step": 38520
+    },
+    {
+      "epoch": 9.960589223413878,
+      "learning_rate": 1.378646441073512e-07,
+      "loss": 2.4727,
+      "step": 38540
+    },
+    {
+      "epoch": 9.965757849851402,
+      "learning_rate": 1.378322312978089e-07,
+      "loss": 2.5166,
+      "step": 38560
+    },
+    {
+      "epoch": 9.970926476288927,
+      "learning_rate": 1.3779981848826656e-07,
+      "loss": 2.5162,
+      "step": 38580
+    },
+    {
+      "epoch": 9.97609510272645,
+      "learning_rate": 1.3776740567872422e-07,
+      "loss": 2.4999,
+      "step": 38600
+    },
+    {
+      "epoch": 9.981263729163974,
+      "learning_rate": 1.377349928691819e-07,
+      "loss": 2.4778,
+      "step": 38620
+    },
+    {
+      "epoch": 9.986432355601499,
+      "learning_rate": 1.3770258005963955e-07,
+      "loss": 2.4741,
+      "step": 38640
+    },
+    {
+      "epoch": 9.991600982039023,
+      "learning_rate": 1.3767016725009724e-07,
+      "loss": 2.5227,
+      "step": 38660
+    },
+    {
+      "epoch": 9.996769608476548,
+      "learning_rate": 1.376377544405549e-07,
+      "loss": 2.537,
+      "step": 38680
+    },
+    {
+      "epoch": 9.999870784339063,
+      "eval_bleu": 11.3364,
+      "eval_gen_len": 36.0732,
+      "eval_loss": 2.484008550643921,
+      "eval_runtime": 789.2845,
+      "eval_samples_per_second": 2.182,
+      "eval_steps_per_second": 1.091,
+      "step": 38692
     }
   ],
   "logging_steps": 20,
@@ -5884,7 +11736,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 1.2701974646813491e+17,
+  "total_flos": 2.540236796612444e+17,
   "train_batch_size": 2,
   "trial_name": null,
   "trial_params": null