diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,8 +1,8 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.9984,
-  "global_step": 468,
+  "epoch": 0.999865870405263,
+  "global_step": 1223,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -10,2824 +10,7354 @@
     {
       "epoch": 0.0,
       "learning_rate": 0.0,
-      "loss": 2.2138,
+      "loss": 2.5458,
       "step": 1
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.0,
-      "loss": 2.1559,
+      "loss": 2.5554,
       "step": 2
     },
     {
-      "epoch": 0.01,
+      "epoch": 0.0,
       "learning_rate": 0.0,
-      "loss": 2.1561,
+      "loss": 2.5621,
       "step": 3
     },
     {
-      "epoch": 0.01,
+      "epoch": 0.0,
       "learning_rate": 0.0,
-      "loss": 2.2014,
+      "loss": 2.5386,
       "step": 4
     },
     {
-      "epoch": 0.01,
+      "epoch": 0.0,
       "learning_rate": 0.0,
-      "loss": 2.1554,
+      "loss": 2.5162,
       "step": 5
     },
     {
-      "epoch": 0.01,
+      "epoch": 0.0,
       "learning_rate": 0.0,
-      "loss": 2.2069,
+      "loss": 2.5274,
       "step": 6
     },
     {
       "epoch": 0.01,
       "learning_rate": 0.0,
-      "loss": 2.1243,
+      "loss": 2.5574,
       "step": 7
     },
     {
-      "epoch": 0.02,
+      "epoch": 0.01,
       "learning_rate": 0.0,
-      "loss": 2.1486,
+      "loss": 2.5572,
       "step": 8
     },
     {
-      "epoch": 0.02,
+      "epoch": 0.01,
       "learning_rate": 0.0,
-      "loss": 2.1337,
+      "loss": 2.527,
       "step": 9
     },
     {
-      "epoch": 0.02,
+      "epoch": 0.01,
       "learning_rate": 0.0,
-      "loss": 2.1883,
+      "loss": 2.6285,
       "step": 10
     },
     {
-      "epoch": 0.02,
+      "epoch": 0.01,
       "learning_rate": 0.0,
-      "loss": 2.1896,
+      "loss": 2.5563,
       "step": 11
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.01,
       "learning_rate": 0.0,
-      "loss": 2.1696,
+      "loss": 2.5448,
       "step": 12
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.01,
       "learning_rate": 0.0,
-      "loss": 2.1555,
+      "loss": 2.5394,
       "step": 13
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.01,
       "learning_rate": 0.0,
-      "loss": 2.144,
+      "loss": 2.5701,
       "step": 14
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.01,
       "learning_rate": 0.0,
-      "loss": 2.2026,
+      "loss": 2.5555,
       "step": 15
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.01,
       "learning_rate": 0.0,
-      "loss": 2.1668,
+      "loss": 2.546,
       "step": 16
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.01,
       "learning_rate": 2e-05,
-      "loss": 2.1488,
+      "loss": 2.5434,
       "step": 17
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.999977372615812e-05,
-      "loss": 2.1886,
+      "epoch": 0.01,
+      "learning_rate": 1.999996695337603e-05,
+      "loss": 2.5695,
       "step": 18
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.9999094914872443e-05,
-      "loss": 1.9954,
+      "epoch": 0.02,
+      "learning_rate": 1.9999867813722544e-05,
+      "loss": 2.1921,
       "step": 19
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.999796359686242e-05,
-      "loss": 1.9071,
+      "epoch": 0.02,
+      "learning_rate": 1.999970258169478e-05,
+      "loss": 2.1095,
       "step": 20
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.9996379823325586e-05,
-      "loss": 1.8843,
+      "epoch": 0.02,
+      "learning_rate": 1.9999471258384817e-05,
+      "loss": 2.0806,
       "step": 21
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.999434366593524e-05,
-      "loss": 1.8831,
+      "epoch": 0.02,
+      "learning_rate": 1.999917384532154e-05,
+      "loss": 2.0858,
       "step": 22
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.9991855216837224e-05,
-      "loss": 1.8479,
+      "epoch": 0.02,
+      "learning_rate": 1.999881034447065e-05,
+      "loss": 1.9848,
       "step": 23
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.9988914588645715e-05,
-      "loss": 1.8049,
+      "epoch": 0.02,
+      "learning_rate": 1.9998380758234648e-05,
+      "loss": 2.0091,
       "step": 24
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.9985521914438167e-05,
-      "loss": 1.8064,
+      "epoch": 0.02,
+      "learning_rate": 1.99978850894528e-05,
+      "loss": 2.0208,
       "step": 25
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.998167734774926e-05,
-      "loss": 1.8176,
+      "epoch": 0.02,
+      "learning_rate": 1.999732334140115e-05,
+      "loss": 1.9977,
       "step": 26
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.9977381062563975e-05,
-      "loss": 1.7639,
+      "epoch": 0.02,
+      "learning_rate": 1.9996695517792468e-05,
+      "loss": 1.959,
       "step": 27
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.9972633253309702e-05,
-      "loss": 1.7097,
+      "epoch": 0.02,
+      "learning_rate": 1.9996001622776244e-05,
+      "loss": 1.9719,
       "step": 28
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.9967434134847443e-05,
-      "loss": 1.8038,
+      "epoch": 0.02,
+      "learning_rate": 1.999524166093866e-05,
+      "loss": 1.9741,
       "step": 29
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.9961783942462104e-05,
-      "loss": 1.6991,
+      "epoch": 0.02,
+      "learning_rate": 1.9994415637302545e-05,
+      "loss": 1.9321,
       "step": 30
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.9955682931851835e-05,
-      "loss": 1.7989,
+      "epoch": 0.03,
+      "learning_rate": 1.9993523557327365e-05,
+      "loss": 1.9157,
       "step": 31
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.9949131379116457e-05,
-      "loss": 1.7358,
+      "epoch": 0.03,
+      "learning_rate": 1.999256542690916e-05,
+      "loss": 1.9143,
       "step": 32
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.9942129580744968e-05,
-      "loss": 1.7183,
+      "epoch": 0.03,
+      "learning_rate": 1.9991541252380526e-05,
+      "loss": 1.9143,
       "step": 33
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.9934677853602133e-05,
-      "loss": 1.7071,
+      "epoch": 0.03,
+      "learning_rate": 1.9990451040510564e-05,
+      "loss": 1.8873,
       "step": 34
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.9926776534914142e-05,
-      "loss": 1.6926,
+      "epoch": 0.03,
+      "learning_rate": 1.9989294798504846e-05,
+      "loss": 1.8526,
       "step": 35
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.9918425982253335e-05,
-      "loss": 1.6736,
+      "epoch": 0.03,
+      "learning_rate": 1.998807253400534e-05,
+      "loss": 1.8974,
       "step": 36
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.9909626573522042e-05,
-      "loss": 1.7163,
+      "epoch": 0.03,
+      "learning_rate": 1.9986784255090395e-05,
+      "loss": 1.903,
       "step": 37
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.990037870693547e-05,
-      "loss": 1.7593,
+      "epoch": 0.03,
+      "learning_rate": 1.9985429970274672e-05,
+      "loss": 1.8515,
       "step": 38
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.9890682801003676e-05,
-      "loss": 1.695,
+      "epoch": 0.03,
+      "learning_rate": 1.998400968850906e-05,
+      "loss": 1.8438,
       "step": 39
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.9880539294512638e-05,
-      "loss": 1.71,
+      "epoch": 0.03,
+      "learning_rate": 1.998252341918068e-05,
+      "loss": 1.8866,
       "step": 40
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.986994864650439e-05,
-      "loss": 1.6417,
+      "epoch": 0.03,
+      "learning_rate": 1.9980971172112763e-05,
+      "loss": 1.8503,
       "step": 41
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.9858911336256258e-05,
-      "loss": 1.6575,
+      "epoch": 0.03,
+      "learning_rate": 1.997935295756461e-05,
+      "loss": 1.8564,
       "step": 42
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.9847427863259164e-05,
-      "loss": 1.6953,
+      "epoch": 0.04,
+      "learning_rate": 1.9977668786231536e-05,
+      "loss": 1.8617,
       "step": 43
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.983549874719501e-05,
-      "loss": 1.6198,
+      "epoch": 0.04,
+      "learning_rate": 1.9975918669244762e-05,
+      "loss": 1.8385,
       "step": 44
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.9823124527913185e-05,
-      "loss": 1.5939,
+      "epoch": 0.04,
+      "learning_rate": 1.9974102618171395e-05,
+      "loss": 1.8284,
       "step": 45
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.981030576540612e-05,
-      "loss": 1.6517,
+      "epoch": 0.04,
+      "learning_rate": 1.9972220645014298e-05,
+      "loss": 1.8777,
       "step": 46
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.9797043039783936e-05,
-      "loss": 1.7013,
+      "epoch": 0.04,
+      "learning_rate": 1.9970272762212045e-05,
+      "loss": 1.8068,
       "step": 47
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.978333695124821e-05,
-      "loss": 1.7184,
+      "epoch": 0.04,
+      "learning_rate": 1.996825898263882e-05,
+      "loss": 1.8234,
       "step": 48
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.9769188120064814e-05,
-      "loss": 1.6911,
+      "epoch": 0.04,
+      "learning_rate": 1.9966179319604357e-05,
+      "loss": 1.8189,
       "step": 49
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.9754597186535812e-05,
-      "loss": 1.6215,
+      "epoch": 0.04,
+      "learning_rate": 1.996403378685382e-05,
+      "loss": 1.8029,
       "step": 50
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.9739564810970534e-05,
-      "loss": 1.6675,
+      "epoch": 0.04,
+      "learning_rate": 1.9961822398567727e-05,
+      "loss": 1.8235,
       "step": 51
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.9724091673655642e-05,
-      "loss": 1.6552,
+      "epoch": 0.04,
+      "learning_rate": 1.9959545169361867e-05,
+      "loss": 1.7572,
       "step": 52
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.970817847482439e-05,
-      "loss": 1.7052,
+      "epoch": 0.04,
+      "learning_rate": 1.9957202114287186e-05,
+      "loss": 1.8396,
       "step": 53
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.96918259346249e-05,
-      "loss": 1.6832,
+      "epoch": 0.04,
+      "learning_rate": 1.9954793248829696e-05,
+      "loss": 1.8163,
       "step": 54
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.9675034793087598e-05,
-      "loss": 1.6272,
+      "epoch": 0.04,
+      "learning_rate": 1.995231858891037e-05,
+      "loss": 1.799,
       "step": 55
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.96578058100917e-05,
-      "loss": 1.6939,
+      "epoch": 0.05,
+      "learning_rate": 1.994977815088504e-05,
+      "loss": 1.7842,
       "step": 56
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.964013976533084e-05,
-      "loss": 1.6046,
+      "epoch": 0.05,
+      "learning_rate": 1.994717195154429e-05,
+      "loss": 1.7806,
       "step": 57
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.9622037458277785e-05,
-      "loss": 1.6639,
+      "epoch": 0.05,
+      "learning_rate": 1.994450000811333e-05,
+      "loss": 1.8058,
       "step": 58
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.9603499708148245e-05,
-      "loss": 1.5518,
+      "epoch": 0.05,
+      "learning_rate": 1.994176233825191e-05,
+      "loss": 1.7702,
       "step": 59
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.958452735386381e-05,
-      "loss": 1.6399,
+      "epoch": 0.05,
+      "learning_rate": 1.993895896005417e-05,
+      "loss": 1.8053,
       "step": 60
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.956512125401398e-05,
-      "loss": 1.6161,
+      "epoch": 0.05,
+      "learning_rate": 1.9936089892048554e-05,
+      "loss": 1.7719,
       "step": 61
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.9545282286817304e-05,
-      "loss": 1.6676,
+      "epoch": 0.05,
+      "learning_rate": 1.9933155153197666e-05,
+      "loss": 1.788,
       "step": 62
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.952501135008165e-05,
-      "loss": 1.6628,
+      "epoch": 0.05,
+      "learning_rate": 1.9930154762898146e-05,
+      "loss": 1.7705,
       "step": 63
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.9504309361163566e-05,
-      "loss": 1.5907,
+      "epoch": 0.05,
+      "learning_rate": 1.992708874098054e-05,
+      "loss": 1.7969,
       "step": 64
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.9483177256926767e-05,
-      "loss": 1.643,
+      "epoch": 0.05,
+      "learning_rate": 1.9923957107709194e-05,
+      "loss": 1.7531,
       "step": 65
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.946161599369973e-05,
-      "loss": 1.6137,
+      "epoch": 0.05,
+      "learning_rate": 1.9920759883782087e-05,
+      "loss": 1.7445,
       "step": 66
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.9439626547232433e-05,
-      "loss": 1.6541,
+      "epoch": 0.05,
+      "learning_rate": 1.9917497090330706e-05,
+      "loss": 1.7906,
       "step": 67
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.941720991265218e-05,
-      "loss": 1.6705,
+      "epoch": 0.06,
+      "learning_rate": 1.9914168748919914e-05,
+      "loss": 1.7894,
       "step": 68
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.9394367104418578e-05,
-      "loss": 1.5576,
+      "epoch": 0.06,
+      "learning_rate": 1.9910774881547803e-05,
+      "loss": 1.7732,
       "step": 69
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.937109915627762e-05,
-      "loss": 1.5932,
+      "epoch": 0.06,
+      "learning_rate": 1.9907315510645545e-05,
+      "loss": 1.7476,
       "step": 70
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.9347407121214917e-05,
-      "loss": 1.5397,
+      "epoch": 0.06,
+      "learning_rate": 1.990379065907724e-05,
+      "loss": 1.7845,
       "step": 71
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.9323292071408018e-05,
-      "loss": 1.6614,
+      "epoch": 0.06,
+      "learning_rate": 1.990020035013978e-05,
+      "loss": 1.722,
       "step": 72
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.9298755098177925e-05,
-      "loss": 1.5635,
+      "epoch": 0.06,
+      "learning_rate": 1.989654460756269e-05,
+      "loss": 1.7516,
       "step": 73
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.9273797311939673e-05,
-      "loss": 1.6104,
+      "epoch": 0.06,
+      "learning_rate": 1.989282345550795e-05,
+      "loss": 1.7839,
       "step": 74
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.92484198421521e-05,
-      "loss": 1.6588,
+      "epoch": 0.06,
+      "learning_rate": 1.9889036918569868e-05,
+      "loss": 1.7555,
       "step": 75
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.922262383726672e-05,
-      "loss": 1.5772,
+      "epoch": 0.06,
+      "learning_rate": 1.9885185021774896e-05,
+      "loss": 1.7675,
       "step": 76
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.9196410464675765e-05,
-      "loss": 1.5928,
+      "epoch": 0.06,
+      "learning_rate": 1.9881267790581466e-05,
+      "loss": 1.7695,
       "step": 77
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.9169780910659334e-05,
-      "loss": 1.6197,
+      "epoch": 0.06,
+      "learning_rate": 1.9877285250879835e-05,
+      "loss": 1.7549,
       "step": 78
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.914273638033173e-05,
-      "loss": 1.6234,
+      "epoch": 0.06,
+      "learning_rate": 1.9873237428991907e-05,
+      "loss": 1.7511,
       "step": 79
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.9115278097586906e-05,
-      "loss": 1.5536,
+      "epoch": 0.07,
+      "learning_rate": 1.9869124351671042e-05,
+      "loss": 1.729,
       "step": 80
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.9087407305043085e-05,
-      "loss": 1.5844,
+      "epoch": 0.07,
+      "learning_rate": 1.986494604610191e-05,
+      "loss": 1.7177,
       "step": 81
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.905912526398654e-05,
-      "loss": 1.6047,
+      "epoch": 0.07,
+      "learning_rate": 1.9860702539900288e-05,
+      "loss": 1.7459,
       "step": 82
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.9030433254314477e-05,
-      "loss": 1.521,
+      "epoch": 0.07,
+      "learning_rate": 1.9856393861112884e-05,
+      "loss": 1.7245,
       "step": 83
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.9001332574477147e-05,
-      "loss": 1.5703,
+      "epoch": 0.07,
+      "learning_rate": 1.9852020038217158e-05,
+      "loss": 1.7318,
       "step": 84
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.8971824541419085e-05,
-      "loss": 1.5954,
+      "epoch": 0.07,
+      "learning_rate": 1.9847581100121128e-05,
+      "loss": 1.704,
       "step": 85
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.8941910490519483e-05,
-      "loss": 1.6148,
+      "epoch": 0.07,
+      "learning_rate": 1.9843077076163177e-05,
+      "loss": 1.7242,
       "step": 86
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.891159177553179e-05,
-      "loss": 1.5881,
+      "epoch": 0.07,
+      "learning_rate": 1.9838507996111856e-05,
+      "loss": 1.7163,
       "step": 87
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.888086976852243e-05,
-      "loss": 1.5238,
+      "epoch": 0.07,
+      "learning_rate": 1.9833873890165707e-05,
+      "loss": 1.6952,
       "step": 88
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.8849745859808718e-05,
-      "loss": 1.6288,
+      "epoch": 0.07,
+      "learning_rate": 1.982917478895304e-05,
+      "loss": 1.7191,
       "step": 89
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.8818221457895928e-05,
-      "loss": 1.6117,
+      "epoch": 0.07,
+      "learning_rate": 1.9824410723531735e-05,
+      "loss": 1.7027,
       "step": 90
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.878629798941357e-05,
-      "loss": 1.5633,
+      "epoch": 0.07,
+      "learning_rate": 1.9819581725389055e-05,
+      "loss": 1.705,
       "step": 91
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.875397689905081e-05,
-      "loss": 1.6096,
+      "epoch": 0.08,
+      "learning_rate": 1.9814687826441413e-05,
+      "loss": 1.6849,
       "step": 92
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.8721259649491116e-05,
-      "loss": 1.5754,
+      "epoch": 0.08,
+      "learning_rate": 1.980972905903418e-05,
+      "loss": 1.689,
       "step": 93
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.868814772134603e-05,
-      "loss": 1.5741,
+      "epoch": 0.08,
+      "learning_rate": 1.9804705455941458e-05,
+      "loss": 1.7007,
       "step": 94
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.8654642613088194e-05,
-      "loss": 1.5657,
+      "epoch": 0.08,
+      "learning_rate": 1.979961705036587e-05,
+      "loss": 1.6949,
       "step": 95
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.8620745840983522e-05,
-      "loss": 1.6128,
+      "epoch": 0.08,
+      "learning_rate": 1.9794463875938343e-05,
+      "loss": 1.7183,
       "step": 96
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.8586458939022586e-05,
-      "loss": 1.5422,
+      "epoch": 0.08,
+      "learning_rate": 1.9789245966717883e-05,
+      "loss": 1.7328,
       "step": 97
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.855178345885119e-05,
-      "loss": 1.5605,
+      "epoch": 0.08,
+      "learning_rate": 1.9783963357191342e-05,
+      "loss": 1.7029,
       "step": 98
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.851672096970016e-05,
-      "loss": 1.6035,
+      "epoch": 0.08,
+      "learning_rate": 1.9778616082273202e-05,
+      "loss": 1.6737,
       "step": 99
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.8481273058314317e-05,
-      "loss": 1.5934,
+      "epoch": 0.08,
+      "learning_rate": 1.9773204177305344e-05,
+      "loss": 1.7199,
       "step": 100
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.844544132888068e-05,
-      "loss": 1.5641,
+      "epoch": 0.08,
+      "learning_rate": 1.9767727678056804e-05,
+      "loss": 1.6804,
       "step": 101
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.840922740295587e-05,
-      "loss": 1.5742,
+      "epoch": 0.08,
+      "learning_rate": 1.9762186620723545e-05,
+      "loss": 1.7044,
       "step": 102
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.8372632919392717e-05,
-      "loss": 1.5688,
+      "epoch": 0.08,
+      "learning_rate": 1.9756581041928213e-05,
+      "loss": 1.6696,
       "step": 103
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.8335659534266095e-05,
-      "loss": 1.5993,
+      "epoch": 0.09,
+      "learning_rate": 1.97509109787199e-05,
+      "loss": 1.6691,
       "step": 104
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.8298308920797986e-05,
-      "loss": 1.5762,
+      "epoch": 0.09,
+      "learning_rate": 1.9745176468573893e-05,
+      "loss": 1.7026,
       "step": 105
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.8260582769281747e-05,
-      "loss": 1.5409,
+      "epoch": 0.09,
+      "learning_rate": 1.9739377549391438e-05,
+      "loss": 1.7319,
       "step": 106
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.8222482787005628e-05,
-      "loss": 1.5436,
+      "epoch": 0.09,
+      "learning_rate": 1.9733514259499465e-05,
+      "loss": 1.6532,
       "step": 107
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.8184010698175507e-05,
-      "loss": 1.5606,
+      "epoch": 0.09,
+      "learning_rate": 1.9727586637650373e-05,
+      "loss": 1.6804,
       "step": 108
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.814516824383685e-05,
-      "loss": 1.5505,
+      "epoch": 0.09,
+      "learning_rate": 1.9721594723021735e-05,
+      "loss": 1.679,
       "step": 109
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.810595718179593e-05,
-      "loss": 1.5514,
+      "epoch": 0.09,
+      "learning_rate": 1.9715538555216055e-05,
+      "loss": 1.7028,
       "step": 110
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.8066379286540278e-05,
-      "loss": 1.4967,
+      "epoch": 0.09,
+      "learning_rate": 1.9709418174260523e-05,
+      "loss": 1.6968,
       "step": 111
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.8026436349158378e-05,
-      "loss": 1.5973,
+      "epoch": 0.09,
+      "learning_rate": 1.9703233620606717e-05,
+      "loss": 1.7381,
       "step": 112
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.7986130177258608e-05,
-      "loss": 1.5811,
+      "epoch": 0.09,
+      "learning_rate": 1.9696984935130365e-05,
+      "loss": 1.7058,
       "step": 113
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.7945462594887443e-05,
-      "loss": 1.5523,
+      "epoch": 0.09,
+      "learning_rate": 1.9690672159131058e-05,
+      "loss": 1.6795,
       "step": 114
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.79044354424469e-05,
-      "loss": 1.5572,
+      "epoch": 0.09,
+      "learning_rate": 1.9684295334331985e-05,
+      "loss": 1.666,
       "step": 115
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.7863050576611267e-05,
-      "loss": 1.5383,
+      "epoch": 0.09,
+      "learning_rate": 1.9677854502879647e-05,
+      "loss": 1.7037,
       "step": 116
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.7821309870243056e-05,
-      "loss": 1.5965,
+      "epoch": 0.1,
+      "learning_rate": 1.9671349707343594e-05,
+      "loss": 1.6822,
       "step": 117
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.7779215212308265e-05,
-      "loss": 1.5478,
+      "epoch": 0.1,
+      "learning_rate": 1.9664780990716133e-05,
+      "loss": 1.6369,
       "step": 118
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.773676850779089e-05,
-      "loss": 1.5333,
+      "epoch": 0.1,
+      "learning_rate": 1.9658148396412048e-05,
+      "loss": 1.6697,
       "step": 119
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.7693971677606714e-05,
-      "loss": 1.5609,
+      "epoch": 0.1,
+      "learning_rate": 1.96514519682683e-05,
+      "loss": 1.6856,
       "step": 120
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.7650826658516375e-05,
-      "loss": 1.7049,
+      "epoch": 0.1,
+      "learning_rate": 1.964469175054377e-05,
+      "loss": 1.6896,
       "step": 121
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.7607335403037713e-05,
-      "loss": 1.6295,
+      "epoch": 0.1,
+      "learning_rate": 1.963786778791892e-05,
+      "loss": 1.6605,
       "step": 122
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.7563499879357425e-05,
-      "loss": 1.5529,
+      "epoch": 0.1,
+      "learning_rate": 1.9630980125495545e-05,
+      "loss": 1.6322,
       "step": 123
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.7519322071241983e-05,
-      "loss": 1.5082,
+      "epoch": 0.1,
+      "learning_rate": 1.9624028808796437e-05,
+      "loss": 1.715,
       "step": 124
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.747480397794786e-05,
-      "loss": 1.5974,
+      "epoch": 0.1,
+      "learning_rate": 1.961701388376511e-05,
+      "loss": 1.6332,
       "step": 125
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.742994761413105e-05,
-      "loss": 1.5395,
+      "epoch": 0.1,
+      "learning_rate": 1.960993539676548e-05,
+      "loss": 1.6746,
       "step": 126
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.738475500975592e-05,
-      "loss": 1.5854,
+      "epoch": 0.1,
+      "learning_rate": 1.9602793394581573e-05,
+      "loss": 1.6904,
       "step": 127
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.7339228210003305e-05,
-      "loss": 1.5478,
+      "epoch": 0.1,
+      "learning_rate": 1.9595587924417186e-05,
+      "loss": 1.6706,
       "step": 128
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 1.7293369275177984e-05,
-      "loss": 1.5289,
+      "epoch": 0.11,
+      "learning_rate": 1.9588319033895625e-05,
+      "loss": 1.6783,
       "step": 129
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 1.724718028061543e-05,
-      "loss": 1.618,
+      "epoch": 0.11,
+      "learning_rate": 1.9580986771059343e-05,
+      "loss": 1.6417,
       "step": 130
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 1.7200663316587897e-05,
-      "loss": 1.5981,
+      "epoch": 0.11,
+      "learning_rate": 1.9573591184369647e-05,
+      "loss": 1.6743,
       "step": 131
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 1.715382048820981e-05,
-      "loss": 1.6268,
+      "epoch": 0.11,
+      "learning_rate": 1.9566132322706376e-05,
+      "loss": 1.6824,
       "step": 132
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 1.7106653915342527e-05,
-      "loss": 1.5983,
+      "epoch": 0.11,
+      "learning_rate": 1.9558610235367563e-05,
+      "loss": 1.6913,
       "step": 133
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 1.7059165732498374e-05,
-      "loss": 1.5532,
+      "epoch": 0.11,
+      "learning_rate": 1.9551024972069127e-05,
+      "loss": 1.685,
       "step": 134
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 1.701135808874406e-05,
-      "loss": 1.4959,
+      "epoch": 0.11,
+      "learning_rate": 1.9543376582944538e-05,
+      "loss": 1.6655,
       "step": 135
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 1.696323314760344e-05,
-      "loss": 1.5412,
+      "epoch": 0.11,
+      "learning_rate": 1.9535665118544488e-05,
+      "loss": 1.6743,
       "step": 136
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 1.6914793086959574e-05,
-      "loss": 1.4874,
+      "epoch": 0.11,
+      "learning_rate": 1.952789062983654e-05,
+      "loss": 1.6702,
       "step": 137
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 1.6866040098956186e-05,
-      "loss": 1.4976,
+      "epoch": 0.11,
+      "learning_rate": 1.9520053168204827e-05,
+      "loss": 1.6522,
       "step": 138
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 1.6816976389898457e-05,
-      "loss": 1.5949,
+      "epoch": 0.11,
+      "learning_rate": 1.9512152785449672e-05,
+      "loss": 1.7291,
       "step": 139
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 1.6767604180153177e-05,
-      "loss": 1.5491,
+      "epoch": 0.11,
+      "learning_rate": 1.950418953378727e-05,
+      "loss": 1.6736,
       "step": 140
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 1.6717925704048256e-05,
-      "loss": 1.5164,
+      "epoch": 0.12,
+      "learning_rate": 1.949616346584934e-05,
+      "loss": 1.6277,
       "step": 141
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 1.6667943209771634e-05,
-      "loss": 1.5603,
+      "epoch": 0.12,
+      "learning_rate": 1.9488074634682766e-05,
+      "loss": 1.6564,
       "step": 142
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 1.6617658959269506e-05,
-      "loss": 1.5746,
+      "epoch": 0.12,
+      "learning_rate": 1.9479923093749274e-05,
+      "loss": 1.6754,
       "step": 143
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 1.6567075228143975e-05,
-      "loss": 1.5456,
+      "epoch": 0.12,
+      "learning_rate": 1.947170889692503e-05,
+      "loss": 1.6514,
       "step": 144
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 1.651619430555009e-05,
-      "loss": 1.5626,
+      "epoch": 0.12,
+      "learning_rate": 1.9463432098500338e-05,
+      "loss": 1.6156,
       "step": 145
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 1.6465018494092213e-05,
-      "loss": 1.5366,
+      "epoch": 0.12,
+      "learning_rate": 1.9455092753179245e-05,
+      "loss": 1.6607,
       "step": 146
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 1.6413550109719828e-05,
-      "loss": 1.5409,
+      "epoch": 0.12,
+      "learning_rate": 1.944669091607919e-05,
+      "loss": 1.6355,
       "step": 147
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 1.6361791481622763e-05,
-      "loss": 1.6488,
+      "epoch": 0.12,
+      "learning_rate": 1.9438226642730648e-05,
+      "loss": 1.6938,
       "step": 148
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 1.6309744952125736e-05,
-      "loss": 1.4961,
+      "epoch": 0.12,
+      "learning_rate": 1.9429699989076746e-05,
+      "loss": 1.6579,
       "step": 149
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 1.6257412876582387e-05,
-      "loss": 1.529,
+      "epoch": 0.12,
+      "learning_rate": 1.9421111011472912e-05,
+      "loss": 1.6352,
       "step": 150
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 1.6204797623268675e-05,
-      "loss": 1.5248,
+      "epoch": 0.12,
+      "learning_rate": 1.9412459766686484e-05,
+      "loss": 1.6207,
       "step": 151
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 1.6151901573275693e-05,
-      "loss": 1.5768,
+      "epoch": 0.12,
+      "learning_rate": 1.9403746311896353e-05,
+      "loss": 1.6526,
       "step": 152
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 1.6098727120401945e-05,
-      "loss": 1.4936,
+      "epoch": 0.13,
+      "learning_rate": 1.9394970704692565e-05,
+      "loss": 1.6738,
       "step": 153
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 1.6045276671044967e-05,
-      "loss": 1.5577,
+      "epoch": 0.13,
+      "learning_rate": 1.9386133003075967e-05,
+      "loss": 1.6581,
       "step": 154
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 1.5991552644092472e-05,
-      "loss": 1.4967,
+      "epoch": 0.13,
+      "learning_rate": 1.9377233265457796e-05,
+      "loss": 1.6377,
       "step": 155
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 1.5937557470812852e-05,
-      "loss": 1.5754,
+      "epoch": 0.13,
+      "learning_rate": 1.9368271550659307e-05,
+      "loss": 1.6432,
       "step": 156
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 1.5883293594745168e-05,
-      "loss": 1.5351,
+      "epoch": 0.13,
+      "learning_rate": 1.9359247917911385e-05,
+      "loss": 1.661,
       "step": 157
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 1.5828763471588564e-05,
-      "loss": 1.5232,
+      "epoch": 0.13,
+      "learning_rate": 1.9350162426854152e-05,
+      "loss": 1.6666,
       "step": 158
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 1.577396956909113e-05,
-      "loss": 1.5244,
+      "epoch": 0.13,
+      "learning_rate": 1.9341015137536562e-05,
+      "loss": 1.6555,
       "step": 159
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 1.5718914366938238e-05,
-      "loss": 1.5161,
+      "epoch": 0.13,
+      "learning_rate": 1.9331806110416027e-05,
+      "loss": 1.6748,
       "step": 160
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 1.5663600356640306e-05,
-      "loss": 1.5548,
+      "epoch": 0.13,
+      "learning_rate": 1.9322535406357998e-05,
+      "loss": 1.6546,
       "step": 161
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 1.5608030041420056e-05,
-      "loss": 1.5389,
+      "epoch": 0.13,
+      "learning_rate": 1.9313203086635568e-05,
+      "loss": 1.6231,
       "step": 162
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 1.555220593609923e-05,
-      "loss": 1.51,
+      "epoch": 0.13,
+      "learning_rate": 1.930380921292907e-05,
+      "loss": 1.6791,
       "step": 163
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 1.5496130566984794e-05,
-      "loss": 1.5267,
+      "epoch": 0.13,
+      "learning_rate": 1.9294353847325666e-05,
+      "loss": 1.6399,
       "step": 164
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 1.5439806471754577e-05,
-      "loss": 1.5266,
+      "epoch": 0.13,
+      "learning_rate": 1.9284837052318935e-05,
+      "loss": 1.6412,
       "step": 165
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 1.538323619934247e-05,
-      "loss": 1.5357,
+      "epoch": 0.14,
+      "learning_rate": 1.9275258890808474e-05,
+      "loss": 1.6032,
       "step": 166
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 1.5326422309823046e-05,
-      "loss": 1.6172,
+      "epoch": 0.14,
+      "learning_rate": 1.9265619426099452e-05,
+      "loss": 1.6022,
       "step": 167
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 1.5269367374295714e-05,
-      "loss": 1.4937,
+      "epoch": 0.14,
+      "learning_rate": 1.9255918721902236e-05,
+      "loss": 1.6609,
       "step": 168
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 1.521207397476837e-05,
-      "loss": 1.5166,
+      "epoch": 0.14,
+      "learning_rate": 1.9246156842331916e-05,
+      "loss": 1.6414,
       "step": 169
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 1.5154544704040537e-05,
-      "loss": 1.5722,
+      "epoch": 0.14,
+      "learning_rate": 1.9236333851907938e-05,
+      "loss": 1.6695,
       "step": 170
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 1.5096782165586037e-05,
-      "loss": 1.5171,
+      "epoch": 0.14,
+      "learning_rate": 1.922644981555363e-05,
+      "loss": 1.6274,
       "step": 171
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 1.5038788973435165e-05,
-      "loss": 1.5285,
+      "epoch": 0.14,
+      "learning_rate": 1.9216504798595797e-05,
+      "loss": 1.6009,
       "step": 172
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 1.4980567752056405e-05,
-      "loss": 1.5144,
+      "epoch": 0.14,
+      "learning_rate": 1.920649886676429e-05,
+      "loss": 1.612,
       "step": 173
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 1.4922121136237645e-05,
-      "loss": 1.5198,
+      "epoch": 0.14,
+      "learning_rate": 1.9196432086191558e-05,
+      "loss": 1.6351,
       "step": 174
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 1.486345177096694e-05,
-      "loss": 1.5095,
+      "epoch": 0.14,
+      "learning_rate": 1.9186304523412226e-05,
+      "loss": 1.6606,
       "step": 175
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 1.480456231131283e-05,
-      "loss": 1.5243,
+      "epoch": 0.14,
+      "learning_rate": 1.9176116245362647e-05,
+      "loss": 1.6228,
       "step": 176
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 1.4745455422304173e-05,
-      "loss": 1.5131,
+      "epoch": 0.14,
+      "learning_rate": 1.9165867319380456e-05,
+      "loss": 1.6527,
       "step": 177
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 1.4686133778809538e-05,
-      "loss": 1.5437,
+      "epoch": 0.15,
+      "learning_rate": 1.9155557813204138e-05,
+      "loss": 1.6437,
       "step": 178
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 1.4626600065416161e-05,
-      "loss": 1.5402,
+      "epoch": 0.15,
+      "learning_rate": 1.9145187794972566e-05,
+      "loss": 1.6606,
       "step": 179
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 1.4566856976308455e-05,
-      "loss": 1.4677,
+      "epoch": 0.15,
+      "learning_rate": 1.9134757333224558e-05,
+      "loss": 1.6435,
       "step": 180
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.4506907215146075e-05,
-      "loss": 1.5523,
+      "epoch": 0.15,
+      "learning_rate": 1.912426649689842e-05,
+      "loss": 1.6718,
       "step": 181
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.4446753494941583e-05,
-      "loss": 1.5167,
+      "epoch": 0.15,
+      "learning_rate": 1.9113715355331505e-05,
+      "loss": 1.67,
       "step": 182
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.438639853793765e-05,
-      "loss": 1.512,
+      "epoch": 0.15,
+      "learning_rate": 1.9103103978259728e-05,
+      "loss": 1.6288,
       "step": 183
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.4325845075483874e-05,
-      "loss": 1.4698,
+      "epoch": 0.15,
+      "learning_rate": 1.9092432435817128e-05,
+      "loss": 1.6611,
       "step": 184
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.4265095847913175e-05,
-      "loss": 1.5465,
+      "epoch": 0.15,
+      "learning_rate": 1.9081700798535395e-05,
+      "loss": 1.6282,
       "step": 185
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.4204153604417775e-05,
-      "loss": 1.5565,
+      "epoch": 0.15,
+      "learning_rate": 1.907090913734341e-05,
+      "loss": 1.6323,
       "step": 186
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.414302110292479e-05,
-      "loss": 1.5588,
+      "epoch": 0.15,
+      "learning_rate": 1.9060057523566758e-05,
+      "loss": 1.6307,
       "step": 187
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.4081701109971411e-05,
-      "loss": 1.5708,
+      "epoch": 0.15,
+      "learning_rate": 1.9049146028927286e-05,
+      "loss": 1.6285,
       "step": 188
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.402019640057972e-05,
-      "loss": 1.5777,
+      "epoch": 0.15,
+      "learning_rate": 1.9038174725542604e-05,
+      "loss": 1.6585,
       "step": 189
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.3958509758131093e-05,
-      "loss": 1.529,
+      "epoch": 0.16,
+      "learning_rate": 1.902714368592562e-05,
+      "loss": 1.6359,
       "step": 190
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.3896643974240245e-05,
-      "loss": 1.5666,
+      "epoch": 0.16,
+      "learning_rate": 1.9016052982984054e-05,
+      "loss": 1.6446,
       "step": 191
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.3834601848628897e-05,
-      "loss": 1.5468,
+      "epoch": 0.16,
+      "learning_rate": 1.9004902690019968e-05,
+      "loss": 1.6536,
       "step": 192
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.3772386188999074e-05,
-      "loss": 1.5441,
+      "epoch": 0.16,
+      "learning_rate": 1.899369288072927e-05,
+      "loss": 1.6578,
       "step": 193
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.3709999810906043e-05,
-      "loss": 1.5623,
+      "epoch": 0.16,
+      "learning_rate": 1.8982423629201226e-05,
+      "loss": 1.6091,
       "step": 194
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.3647445537630892e-05,
-      "loss": 1.5692,
+      "epoch": 0.16,
+      "learning_rate": 1.8971095009917985e-05,
+      "loss": 1.6318,
       "step": 195
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.3584726200052767e-05,
-      "loss": 1.5123,
+      "epoch": 0.16,
+      "learning_rate": 1.8959707097754067e-05,
+      "loss": 1.6396,
       "step": 196
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.3521844636520767e-05,
-      "loss": 1.5064,
+      "epoch": 0.16,
+      "learning_rate": 1.8948259967975887e-05,
+      "loss": 1.606,
       "step": 197
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.3458803692725486e-05,
-      "loss": 1.5287,
+      "epoch": 0.16,
+      "learning_rate": 1.8936753696241238e-05,
+      "loss": 1.6607,
       "step": 198
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.339560622157023e-05,
-      "loss": 1.4452,
+      "epoch": 0.16,
+      "learning_rate": 1.8925188358598815e-05,
+      "loss": 1.6244,
       "step": 199
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.3332255083041915e-05,
-      "loss": 1.5381,
+      "epoch": 0.16,
+      "learning_rate": 1.891356403148768e-05,
+      "loss": 1.6414,
       "step": 200
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.3268753144081652e-05,
-      "loss": 1.4764,
+      "epoch": 0.16,
+      "learning_rate": 1.8901880791736795e-05,
+      "loss": 1.6298,
       "step": 201
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.3205103278454971e-05,
-      "loss": 1.4674,
+      "epoch": 0.17,
+      "learning_rate": 1.8890138716564484e-05,
+      "loss": 1.6163,
       "step": 202
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.3141308366621802e-05,
-      "loss": 1.5338,
+      "epoch": 0.17,
+      "learning_rate": 1.887833788357793e-05,
+      "loss": 1.648,
       "step": 203
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.3077371295606098e-05,
-      "loss": 1.4538,
+      "epoch": 0.17,
+      "learning_rate": 1.886647837077268e-05,
+      "loss": 1.6144,
       "step": 204
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.3013294958865208e-05,
-      "loss": 1.5016,
+      "epoch": 0.17,
+      "learning_rate": 1.8854560256532098e-05,
+      "loss": 1.6285,
       "step": 205
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.2949082256158904e-05,
-      "loss": 1.5428,
+      "epoch": 0.17,
+      "learning_rate": 1.884258361962688e-05,
+      "loss": 1.6183,
       "step": 206
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.2884736093418173e-05,
-      "loss": 1.5386,
+      "epoch": 0.17,
+      "learning_rate": 1.8830548539214506e-05,
+      "loss": 1.5969,
       "step": 207
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.282025938261371e-05,
-      "loss": 1.4571,
+      "epoch": 0.17,
+      "learning_rate": 1.8818455094838728e-05,
+      "loss": 1.6097,
       "step": 208
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.2755655041624121e-05,
-      "loss": 1.469,
+      "epoch": 0.17,
+      "learning_rate": 1.880630336642905e-05,
+      "loss": 1.6035,
       "step": 209
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.26909259941039e-05,
-      "loss": 1.4959,
+      "epoch": 0.17,
+      "learning_rate": 1.879409343430019e-05,
+      "loss": 1.5975,
       "step": 210
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.26260751693511e-05,
-      "loss": 1.4292,
+      "epoch": 0.17,
+      "learning_rate": 1.878182537915156e-05,
+      "loss": 1.5975,
       "step": 211
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.2561105502174768e-05,
-      "loss": 1.499,
+      "epoch": 0.17,
+      "learning_rate": 1.8769499282066716e-05,
+      "loss": 1.6287,
       "step": 212
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.2496019932762149e-05,
-      "loss": 1.4792,
+      "epoch": 0.17,
+      "learning_rate": 1.875711522451284e-05,
+      "loss": 1.6185,
       "step": 213
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.243082140654561e-05,
-      "loss": 1.4519,
+      "epoch": 0.17,
+      "learning_rate": 1.8744673288340193e-05,
+      "loss": 1.5915,
       "step": 214
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.2365512874069355e-05,
-      "loss": 1.5228,
+      "epoch": 0.18,
+      "learning_rate": 1.873217355578157e-05,
+      "loss": 1.6246,
       "step": 215
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.2300097290855887e-05,
-      "loss": 1.5358,
+      "epoch": 0.18,
+      "learning_rate": 1.8719616109451754e-05,
+      "loss": 1.5973,
       "step": 216
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.2234577617272282e-05,
-      "loss": 1.5408,
+      "epoch": 0.18,
+      "learning_rate": 1.8707001032347e-05,
+      "loss": 1.6421,
       "step": 217
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.2168956818396191e-05,
-      "loss": 1.5011,
+      "epoch": 0.18,
+      "learning_rate": 1.8694328407844448e-05,
+      "loss": 1.5918,
       "step": 218
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.210323786388166e-05,
-      "loss": 1.5249,
+      "epoch": 0.18,
+      "learning_rate": 1.868159831970158e-05,
+      "loss": 1.6338,
       "step": 219
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.2037423727824761e-05,
-      "loss": 1.5088,
+      "epoch": 0.18,
+      "learning_rate": 1.866881085205569e-05,
+      "loss": 1.662,
       "step": 220
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.1971517388628972e-05,
-      "loss": 1.492,
+      "epoch": 0.18,
+      "learning_rate": 1.865596608942331e-05,
+      "loss": 1.6196,
       "step": 221
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.1905521828870415e-05,
-      "loss": 1.4988,
+      "epoch": 0.18,
+      "learning_rate": 1.8643064116699635e-05,
+      "loss": 1.6176,
       "step": 222
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.1839440035162855e-05,
-      "loss": 1.5146,
+      "epoch": 0.18,
+      "learning_rate": 1.863010501915801e-05,
+      "loss": 1.6206,
       "step": 223
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.1773274998022557e-05,
-      "loss": 1.5127,
+      "epoch": 0.18,
+      "learning_rate": 1.86170888824493e-05,
+      "loss": 1.6259,
       "step": 224
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.1707029711732962e-05,
-      "loss": 1.5316,
+      "epoch": 0.18,
+      "learning_rate": 1.8604015792601395e-05,
+      "loss": 1.6659,
       "step": 225
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.1640707174209147e-05,
-      "loss": 1.5447,
+      "epoch": 0.18,
+      "learning_rate": 1.859088583601859e-05,
+      "loss": 1.6342,
       "step": 226
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.157431038686219e-05,
-      "loss": 1.5086,
+      "epoch": 0.19,
+      "learning_rate": 1.8577699099481025e-05,
+      "loss": 1.6232,
       "step": 227
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.150784235446332e-05,
-      "loss": 1.5467,
+      "epoch": 0.19,
+      "learning_rate": 1.8564455670144128e-05,
+      "loss": 1.5443,
       "step": 228
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.1441306085007957e-05,
-      "loss": 1.5022,
+      "epoch": 0.19,
+      "learning_rate": 1.855115563553803e-05,
+      "loss": 1.5705,
       "step": 229
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.1374704589579552e-05,
-      "loss": 1.4693,
+      "epoch": 0.19,
+      "learning_rate": 1.8537799083566974e-05,
+      "loss": 1.5783,
       "step": 230
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.1308040882213363e-05,
-      "loss": 1.4585,
+      "epoch": 0.19,
+      "learning_rate": 1.8524386102508754e-05,
+      "loss": 1.5572,
       "step": 231
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.1241317979760022e-05,
-      "loss": 1.5255,
+      "epoch": 0.19,
+      "learning_rate": 1.8510916781014118e-05,
+      "loss": 1.6019,
       "step": 232
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.1174538901749026e-05,
-      "loss": 1.5313,
+      "epoch": 0.19,
+      "learning_rate": 1.8497391208106182e-05,
+      "loss": 1.6203,
       "step": 233
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.1107706670252079e-05,
-      "loss": 1.5279,
+      "epoch": 0.19,
+      "learning_rate": 1.8483809473179854e-05,
+      "loss": 1.6123,
       "step": 234
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.104082430974634e-05,
-      "loss": 1.5155,
+      "epoch": 0.19,
+      "learning_rate": 1.8470171666001226e-05,
+      "loss": 1.612,
       "step": 235
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.0973894846977548e-05,
-      "loss": 1.4693,
+      "epoch": 0.19,
+      "learning_rate": 1.8456477876707004e-05,
+      "loss": 1.5885,
       "step": 236
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.090692131082303e-05,
-      "loss": 1.5077,
+      "epoch": 0.19,
+      "learning_rate": 1.8442728195803882e-05,
+      "loss": 1.622,
       "step": 237
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.0839906732154655e-05,
-      "loss": 1.4728,
+      "epoch": 0.19,
+      "learning_rate": 1.842892271416797e-05,
+      "loss": 1.5859,
       "step": 238
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.0772854143701661e-05,
-      "loss": 1.4855,
+      "epoch": 0.2,
+      "learning_rate": 1.8415061523044186e-05,
+      "loss": 1.6028,
       "step": 239
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.0705766579913401e-05,
-      "loss": 1.5412,
+      "epoch": 0.2,
+      "learning_rate": 1.840114471404563e-05,
+      "loss": 1.5905,
       "step": 240
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.0638647076822041e-05,
-      "loss": 1.5136,
+      "epoch": 0.2,
+      "learning_rate": 1.838717237915302e-05,
+      "loss": 1.5952,
       "step": 241
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 1.0571498671905146e-05,
-      "loss": 1.5312,
+      "epoch": 0.2,
+      "learning_rate": 1.837314461071406e-05,
+      "loss": 1.6294,
       "step": 242
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 1.0504324403948225e-05,
-      "loss": 1.5057,
+      "epoch": 0.2,
+      "learning_rate": 1.8359061501442826e-05,
+      "loss": 1.5569,
       "step": 243
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 1.0437127312907218e-05,
-      "loss": 1.5462,
+      "epoch": 0.2,
+      "learning_rate": 1.8344923144419156e-05,
+      "loss": 1.5726,
       "step": 244
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 1.0369910439770913e-05,
-      "loss": 1.4872,
+      "epoch": 0.2,
+      "learning_rate": 1.8330729633088044e-05,
+      "loss": 1.6316,
       "step": 245
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 1.030267682642334e-05,
-      "loss": 1.4561,
+      "epoch": 0.2,
+      "learning_rate": 1.8316481061259023e-05,
+      "loss": 1.6277,
       "step": 246
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 1.023542951550609e-05,
-      "loss": 1.5295,
+      "epoch": 0.2,
+      "learning_rate": 1.8302177523105528e-05,
+      "loss": 1.6231,
       "step": 247
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 1.0168171550280647e-05,
-      "loss": 1.4817,
+      "epoch": 0.2,
+      "learning_rate": 1.8287819113164286e-05,
+      "loss": 1.6207,
       "step": 248
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 1.010090597449065e-05,
-      "loss": 1.5396,
+      "epoch": 0.2,
+      "learning_rate": 1.8273405926334697e-05,
+      "loss": 1.559,
       "step": 249
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 1.003363583222415e-05,
-      "loss": 1.491,
+      "epoch": 0.2,
+      "learning_rate": 1.8258938057878188e-05,
+      "loss": 1.5918,
       "step": 250
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 9.966364167775851e-06,
-      "loss": 1.4951,
+      "epoch": 0.21,
+      "learning_rate": 1.8244415603417603e-05,
+      "loss": 1.5889,
       "step": 251
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 9.899094025509353e-06,
-      "loss": 1.4492,
+      "epoch": 0.21,
+      "learning_rate": 1.8229838658936566e-05,
+      "loss": 1.6143,
       "step": 252
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 9.831828449719354e-06,
-      "loss": 1.4906,
+      "epoch": 0.21,
+      "learning_rate": 1.8215207320778828e-05,
+      "loss": 1.6087,
       "step": 253
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 9.764570484493916e-06,
-      "loss": 1.5193,
+      "epoch": 0.21,
+      "learning_rate": 1.8200521685647662e-05,
+      "loss": 1.6031,
       "step": 254
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 9.697323173576667e-06,
-      "loss": 1.5142,
+      "epoch": 0.21,
+      "learning_rate": 1.8185781850605203e-05,
+      "loss": 1.5853,
       "step": 255
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 9.630089560229088e-06,
-      "loss": 1.5225,
+      "epoch": 0.21,
+      "learning_rate": 1.8170987913071797e-05,
+      "loss": 1.589,
       "step": 256
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 9.562872687092784e-06,
-      "loss": 1.481,
+      "epoch": 0.21,
+      "learning_rate": 1.8156139970825393e-05,
+      "loss": 1.599,
       "step": 257
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 9.495675596051777e-06,
-      "loss": 1.4661,
+      "epoch": 0.21,
+      "learning_rate": 1.8141238122000858e-05,
+      "loss": 1.554,
       "step": 258
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 9.428501328094855e-06,
-      "loss": 1.5027,
+      "epoch": 0.21,
+      "learning_rate": 1.8126282465089352e-05,
+      "loss": 1.5801,
       "step": 259
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 9.36135292317796e-06,
-      "loss": 1.5151,
+      "epoch": 0.21,
+      "learning_rate": 1.8111273098937666e-05,
+      "loss": 1.5895,
       "step": 260
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 9.294233420086604e-06,
-      "loss": 1.4931,
+      "epoch": 0.21,
+      "learning_rate": 1.8096210122747585e-05,
+      "loss": 1.59,
       "step": 261
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 9.227145856298344e-06,
-      "loss": 1.5098,
+      "epoch": 0.21,
+      "learning_rate": 1.8081093636075204e-05,
+      "loss": 1.5676,
       "step": 262
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 9.16009326784535e-06,
-      "loss": 1.5395,
+      "epoch": 0.22,
+      "learning_rate": 1.8065923738830294e-05,
+      "loss": 1.5716,
       "step": 263
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 9.093078689176973e-06,
-      "loss": 1.4412,
+      "epoch": 0.22,
+      "learning_rate": 1.8050700531275632e-05,
+      "loss": 1.6131,
       "step": 264
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 9.026105153022453e-06,
-      "loss": 1.4393,
+      "epoch": 0.22,
+      "learning_rate": 1.803542411402634e-05,
+      "loss": 1.6058,
       "step": 265
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 8.95917569025366e-06,
-      "loss": 1.4435,
+      "epoch": 0.22,
+      "learning_rate": 1.802009458804923e-05,
+      "loss": 1.5639,
       "step": 266
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 8.892293329747923e-06,
-      "loss": 1.5213,
+      "epoch": 0.22,
+      "learning_rate": 1.800471205466211e-05,
+      "loss": 1.6021,
       "step": 267
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 8.82546109825098e-06,
-      "loss": 1.4944,
+      "epoch": 0.22,
+      "learning_rate": 1.798927661553314e-05,
+      "loss": 1.6076,
       "step": 268
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 8.758682020239983e-06,
-      "loss": 1.4446,
+      "epoch": 0.22,
+      "learning_rate": 1.797378837268015e-05,
+      "loss": 1.6016,
       "step": 269
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 8.69195911778664e-06,
-      "loss": 1.5148,
+      "epoch": 0.22,
+      "learning_rate": 1.7958247428469972e-05,
+      "loss": 1.605,
       "step": 270
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 8.625295410420451e-06,
-      "loss": 1.546,
+      "epoch": 0.22,
+      "learning_rate": 1.7942653885617753e-05,
+      "loss": 1.5578,
       "step": 271
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 8.558693914992046e-06,
-      "loss": 1.5003,
+      "epoch": 0.22,
+      "learning_rate": 1.7927007847186276e-05,
+      "loss": 1.5959,
       "step": 272
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 8.492157645536678e-06,
-      "loss": 1.4644,
+      "epoch": 0.22,
+      "learning_rate": 1.7911309416585295e-05,
+      "loss": 1.5958,
       "step": 273
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 8.425689613137812e-06,
-      "loss": 1.4522,
+      "epoch": 0.22,
+      "learning_rate": 1.789555869757084e-05,
+      "loss": 1.6015,
       "step": 274
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 8.35929282579086e-06,
-      "loss": 1.5279,
+      "epoch": 0.22,
+      "learning_rate": 1.7879755794244524e-05,
+      "loss": 1.5896,
       "step": 275
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 8.292970288267043e-06,
-      "loss": 1.5479,
+      "epoch": 0.23,
+      "learning_rate": 1.786390081105287e-05,
+      "loss": 1.5939,
       "step": 276
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 8.226725001977445e-06,
-      "loss": 1.4698,
+      "epoch": 0.23,
+      "learning_rate": 1.7847993852786612e-05,
+      "loss": 1.5938,
       "step": 277
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 8.160559964837148e-06,
-      "loss": 1.4243,
+      "epoch": 0.23,
+      "learning_rate": 1.7832035024580002e-05,
+      "loss": 1.592,
       "step": 278
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 8.094478171129588e-06,
-      "loss": 1.4687,
+      "epoch": 0.23,
+      "learning_rate": 1.781602443191012e-05,
+      "loss": 1.5881,
       "step": 279
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 8.028482611371028e-06,
-      "loss": 1.4989,
+      "epoch": 0.23,
+      "learning_rate": 1.7799962180596167e-05,
+      "loss": 1.5726,
       "step": 280
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 7.96257627217524e-06,
-      "loss": 1.5403,
+      "epoch": 0.23,
+      "learning_rate": 1.778384837679879e-05,
+      "loss": 1.5781,
       "step": 281
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 7.896762136118342e-06,
-      "loss": 1.4832,
+      "epoch": 0.23,
+      "learning_rate": 1.7767683127019342e-05,
+      "loss": 1.5997,
       "step": 282
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 7.831043181603814e-06,
-      "loss": 1.4891,
+      "epoch": 0.23,
+      "learning_rate": 1.775146653809922e-05,
+      "loss": 1.5959,
       "step": 283
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 7.76542238272772e-06,
-      "loss": 1.4889,
+      "epoch": 0.23,
+      "learning_rate": 1.7735198717219114e-05,
+      "loss": 1.5975,
       "step": 284
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 7.699902709144114e-06,
-      "loss": 1.5028,
+      "epoch": 0.23,
+      "learning_rate": 1.771887977189835e-05,
+      "loss": 1.5867,
       "step": 285
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 7.634487125930649e-06,
-      "loss": 1.4516,
+      "epoch": 0.23,
+      "learning_rate": 1.7702509809994124e-05,
+      "loss": 1.5665,
       "step": 286
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 7.5691785934543926e-06,
-      "loss": 1.469,
+      "epoch": 0.23,
+      "learning_rate": 1.768608893970084e-05,
+      "loss": 1.5753,
       "step": 287
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 7.503980067237853e-06,
-      "loss": 1.4795,
+      "epoch": 0.24,
+      "learning_rate": 1.766961726954936e-05,
+      "loss": 1.6099,
       "step": 288
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 7.438894497825235e-06,
-      "loss": 1.604,
+      "epoch": 0.24,
+      "learning_rate": 1.7653094908406304e-05,
+      "loss": 1.6077,
       "step": 289
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 7.373924830648904e-06,
-      "loss": 1.4822,
+      "epoch": 0.24,
+      "learning_rate": 1.7636521965473324e-05,
+      "loss": 1.5996,
       "step": 290
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 7.309074005896103e-06,
-      "loss": 1.4996,
+      "epoch": 0.24,
+      "learning_rate": 1.7619898550286377e-05,
+      "loss": 1.5869,
       "step": 291
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 7.244344958375881e-06,
-      "loss": 1.4586,
+      "epoch": 0.24,
+      "learning_rate": 1.7603224772715015e-05,
+      "loss": 1.575,
       "step": 292
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 7.179740617386295e-06,
-      "loss": 1.4558,
+      "epoch": 0.24,
+      "learning_rate": 1.7586500742961653e-05,
+      "loss": 1.5547,
       "step": 293
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 7.115263906581828e-06,
-      "loss": 1.4616,
+      "epoch": 0.24,
+      "learning_rate": 1.7569726571560838e-05,
+      "loss": 1.58,
       "step": 294
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 7.050917743841099e-06,
-      "loss": 1.4783,
+      "epoch": 0.24,
+      "learning_rate": 1.7552902369378507e-05,
+      "loss": 1.5877,
       "step": 295
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 6.9867050411347955e-06,
-      "loss": 1.5111,
+      "epoch": 0.24,
+      "learning_rate": 1.7536028247611284e-05,
+      "loss": 1.5849,
       "step": 296
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 6.922628704393904e-06,
-      "loss": 1.5296,
+      "epoch": 0.24,
+      "learning_rate": 1.751910431778572e-05,
+      "loss": 1.6247,
       "step": 297
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 6.858691633378202e-06,
-      "loss": 1.3932,
+      "epoch": 0.24,
+      "learning_rate": 1.7502130691757556e-05,
+      "loss": 1.5345,
       "step": 298
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 6.794896721545032e-06,
-      "loss": 1.5553,
+      "epoch": 0.24,
+      "learning_rate": 1.7485107481711014e-05,
+      "loss": 1.5816,
       "step": 299
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 6.73124685591835e-06,
-      "loss": 1.4674,
+      "epoch": 0.25,
+      "learning_rate": 1.7468034800158004e-05,
+      "loss": 1.6096,
       "step": 300
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 6.667744916958085e-06,
-      "loss": 1.49,
+      "epoch": 0.25,
+      "learning_rate": 1.7450912759937434e-05,
+      "loss": 1.5818,
       "step": 301
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 6.604393778429772e-06,
-      "loss": 1.5332,
+      "epoch": 0.25,
+      "learning_rate": 1.7433741474214422e-05,
+      "loss": 1.5764,
       "step": 302
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 6.541196307274518e-06,
-      "loss": 1.3804,
+      "epoch": 0.25,
+      "learning_rate": 1.7416521056479577e-05,
+      "loss": 1.6029,
       "step": 303
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 6.478155363479236e-06,
-      "loss": 1.4536,
+      "epoch": 0.25,
+      "learning_rate": 1.739925162054823e-05,
+      "loss": 1.5636,
       "step": 304
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 6.415273799947235e-06,
-      "loss": 1.5034,
+      "epoch": 0.25,
+      "learning_rate": 1.738193328055969e-05,
+      "loss": 1.596,
       "step": 305
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 6.352554462369112e-06,
-      "loss": 1.4668,
+      "epoch": 0.25,
+      "learning_rate": 1.7364566150976498e-05,
+      "loss": 1.552,
       "step": 306
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 6.29000018909396e-06,
-      "loss": 1.4662,
+      "epoch": 0.25,
+      "learning_rate": 1.734715034658365e-05,
+      "loss": 1.5304,
       "step": 307
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 6.2276138110009255e-06,
-      "loss": 1.5042,
+      "epoch": 0.25,
+      "learning_rate": 1.7329685982487854e-05,
+      "loss": 1.5709,
       "step": 308
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 6.1653981513711055e-06,
-      "loss": 1.5066,
+      "epoch": 0.25,
+      "learning_rate": 1.7312173174116763e-05,
+      "loss": 1.5789,
       "step": 309
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 6.103356025759759e-06,
-      "loss": 1.4631,
+      "epoch": 0.25,
+      "learning_rate": 1.7294612037218216e-05,
+      "loss": 1.5734,
       "step": 310
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 6.04149024186891e-06,
-      "loss": 1.4443,
+      "epoch": 0.25,
+      "learning_rate": 1.727700268785947e-05,
+      "loss": 1.5888,
       "step": 311
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 5.979803599420284e-06,
-      "loss": 1.434,
+      "epoch": 0.26,
+      "learning_rate": 1.7259345242426443e-05,
+      "loss": 1.6017,
       "step": 312
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 5.918298890028591e-06,
-      "loss": 1.4716,
+      "epoch": 0.26,
+      "learning_rate": 1.724163981762291e-05,
+      "loss": 1.5575,
       "step": 313
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 5.8569788970752114e-06,
-      "loss": 1.5051,
+      "epoch": 0.26,
+      "learning_rate": 1.722388653046979e-05,
+      "loss": 1.6164,
       "step": 314
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 5.795846395582225e-06,
-      "loss": 1.4497,
+      "epoch": 0.26,
+      "learning_rate": 1.720608549830432e-05,
+      "loss": 1.5519,
       "step": 315
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 5.734904152086829e-06,
-      "loss": 1.4782,
+      "epoch": 0.26,
+      "learning_rate": 1.7188236838779297e-05,
+      "loss": 1.5376,
       "step": 316
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 5.674154924516129e-06,
-      "loss": 1.4782,
+      "epoch": 0.26,
+      "learning_rate": 1.717034066986231e-05,
+      "loss": 1.5968,
       "step": 317
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 5.613601462062352e-06,
-      "loss": 1.4425,
+      "epoch": 0.26,
+      "learning_rate": 1.715239710983496e-05,
+      "loss": 1.5931,
       "step": 318
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 5.553246505058421e-06,
-      "loss": 1.4392,
+      "epoch": 0.26,
+      "learning_rate": 1.7134406277292054e-05,
+      "loss": 1.5761,
       "step": 319
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 5.493092784853926e-06,
-      "loss": 1.4346,
+      "epoch": 0.26,
+      "learning_rate": 1.7116368291140854e-05,
+      "loss": 1.5768,
       "step": 320
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 5.433143023691547e-06,
-      "loss": 1.4843,
+      "epoch": 0.26,
+      "learning_rate": 1.709828327060027e-05,
+      "loss": 1.5913,
       "step": 321
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 5.3733999345838395e-06,
-      "loss": 1.4629,
+      "epoch": 0.26,
+      "learning_rate": 1.7080151335200067e-05,
+      "loss": 1.5935,
       "step": 322
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 5.313866221190466e-06,
-      "loss": 1.4557,
+      "epoch": 0.26,
+      "learning_rate": 1.7061972604780105e-05,
+      "loss": 1.5581,
       "step": 323
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 5.2545445776958304e-06,
-      "loss": 1.4436,
+      "epoch": 0.26,
+      "learning_rate": 1.7043747199489516e-05,
+      "loss": 1.5535,
       "step": 324
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 5.195437688687175e-06,
-      "loss": 1.4592,
+      "epoch": 0.27,
+      "learning_rate": 1.7025475239785917e-05,
+      "loss": 1.6057,
       "step": 325
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 5.136548229033065e-06,
-      "loss": 1.4946,
+      "epoch": 0.27,
+      "learning_rate": 1.7007156846434632e-05,
+      "loss": 1.5988,
       "step": 326
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 5.07787886376236e-06,
-      "loss": 1.4053,
+      "epoch": 0.27,
+      "learning_rate": 1.698879214050787e-05,
+      "loss": 1.5762,
       "step": 327
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 5.019432247943595e-06,
-      "loss": 1.4686,
+      "epoch": 0.27,
+      "learning_rate": 1.6970381243383934e-05,
+      "loss": 1.5708,
       "step": 328
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 4.961211026564837e-06,
-      "loss": 1.4728,
+      "epoch": 0.27,
+      "learning_rate": 1.6951924276746425e-05,
+      "loss": 1.5827,
       "step": 329
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 4.9032178344139704e-06,
-      "loss": 1.4095,
+      "epoch": 0.27,
+      "learning_rate": 1.6933421362583428e-05,
+      "loss": 1.5688,
       "step": 330
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 4.845455295959468e-06,
-      "loss": 1.4501,
+      "epoch": 0.27,
+      "learning_rate": 1.6914872623186716e-05,
+      "loss": 1.5523,
       "step": 331
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 4.787926025231634e-06,
-      "loss": 1.46,
+      "epoch": 0.27,
+      "learning_rate": 1.6896278181150934e-05,
+      "loss": 1.5762,
       "step": 332
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 4.730632625704288e-06,
-      "loss": 1.4633,
+      "epoch": 0.27,
+      "learning_rate": 1.6877638159372783e-05,
+      "loss": 1.5363,
       "step": 333
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 4.673577690176956e-06,
-      "loss": 1.4805,
+      "epoch": 0.27,
+      "learning_rate": 1.6858952681050222e-05,
+      "loss": 1.5774,
       "step": 334
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 4.616763800657534e-06,
-      "loss": 1.4563,
+      "epoch": 0.27,
+      "learning_rate": 1.6840221869681652e-05,
+      "loss": 1.5293,
       "step": 335
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 4.5601935282454255e-06,
-      "loss": 1.4409,
+      "epoch": 0.27,
+      "learning_rate": 1.6821445849065082e-05,
+      "loss": 1.6277,
       "step": 336
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 4.503869433015213e-06,
-      "loss": 1.454,
+      "epoch": 0.28,
+      "learning_rate": 1.6802624743297333e-05,
+      "loss": 1.543,
       "step": 337
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 4.447794063900772e-06,
-      "loss": 1.4864,
+      "epoch": 0.28,
+      "learning_rate": 1.6783758676773208e-05,
+      "loss": 1.5296,
       "step": 338
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 4.391969958579948e-06,
-      "loss": 1.4723,
+      "epoch": 0.28,
+      "learning_rate": 1.6764847774184666e-05,
+      "loss": 1.5622,
       "step": 339
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 4.336399643359695e-06,
-      "loss": 1.5223,
+      "epoch": 0.28,
+      "learning_rate": 1.6745892160520004e-05,
+      "loss": 1.5588,
       "step": 340
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 4.281085633061764e-06,
-      "loss": 1.4559,
+      "epoch": 0.28,
+      "learning_rate": 1.672689196106303e-05,
+      "loss": 1.5447,
       "step": 341
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 4.226030430908869e-06,
-      "loss": 1.443,
+      "epoch": 0.28,
+      "learning_rate": 1.6707847301392237e-05,
+      "loss": 1.5592,
       "step": 342
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 4.1712365284114364e-06,
-      "loss": 1.4478,
+      "epoch": 0.28,
+      "learning_rate": 1.668875830737996e-05,
+      "loss": 1.5749,
       "step": 343
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 4.116706405254834e-06,
-      "loss": 1.4457,
+      "epoch": 0.28,
+      "learning_rate": 1.6669625105191564e-05,
+      "loss": 1.5794,
       "step": 344
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 4.062442529187151e-06,
-      "loss": 1.5377,
+      "epoch": 0.28,
+      "learning_rate": 1.6650447821284593e-05,
+      "loss": 1.5471,
       "step": 345
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 4.0084473559075335e-06,
-      "loss": 1.4432,
+      "epoch": 0.28,
+      "learning_rate": 1.6631226582407954e-05,
+      "loss": 1.5166,
       "step": 346
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 3.954723328955036e-06,
-      "loss": 1.4507,
+      "epoch": 0.28,
+      "learning_rate": 1.661196151560105e-05,
+      "loss": 1.5485,
       "step": 347
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 3.901272879598058e-06,
-      "loss": 1.4577,
+      "epoch": 0.28,
+      "learning_rate": 1.6592652748192964e-05,
+      "loss": 1.5649,
       "step": 348
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 3.848098426724306e-06,
-      "loss": 1.5067,
+      "epoch": 0.29,
+      "learning_rate": 1.6573300407801616e-05,
+      "loss": 1.5667,
       "step": 349
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 3.7952023767313264e-06,
-      "loss": 1.441,
+      "epoch": 0.29,
+      "learning_rate": 1.6553904622332903e-05,
+      "loss": 1.5397,
       "step": 350
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 3.7425871234176134e-06,
-      "loss": 1.476,
+      "epoch": 0.29,
+      "learning_rate": 1.6534465519979875e-05,
+      "loss": 1.5435,
       "step": 351
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 3.6902550478742673e-06,
-      "loss": 1.4455,
+      "epoch": 0.29,
+      "learning_rate": 1.6514983229221878e-05,
+      "loss": 1.5857,
       "step": 352
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 3.6382085183772396e-06,
-      "loss": 1.4195,
+      "epoch": 0.29,
+      "learning_rate": 1.649545787882369e-05,
+      "loss": 1.5859,
       "step": 353
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 3.5864498902801727e-06,
-      "loss": 1.4927,
+      "epoch": 0.29,
+      "learning_rate": 1.6475889597834695e-05,
+      "loss": 1.5436,
       "step": 354
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 3.534981505907792e-06,
-      "loss": 1.4858,
+      "epoch": 0.29,
+      "learning_rate": 1.6456278515588023e-05,
+      "loss": 1.5477,
       "step": 355
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 3.483805694449913e-06,
-      "loss": 1.4873,
+      "epoch": 0.29,
+      "learning_rate": 1.6436624761699687e-05,
+      "loss": 1.5404,
       "step": 356
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 3.432924771856029e-06,
-      "loss": 1.4472,
+      "epoch": 0.29,
+      "learning_rate": 1.6416928466067725e-05,
+      "loss": 1.5854,
       "step": 357
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 3.382341040730501e-06,
-      "loss": 1.4626,
+      "epoch": 0.29,
+      "learning_rate": 1.6397189758871345e-05,
+      "loss": 1.5782,
       "step": 358
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 3.3320567902283694e-06,
-      "loss": 1.4504,
+      "epoch": 0.29,
+      "learning_rate": 1.6377408770570092e-05,
+      "loss": 1.5157,
       "step": 359
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 3.282074295951744e-06,
-      "loss": 1.5434,
+      "epoch": 0.29,
+      "learning_rate": 1.635758563190293e-05,
+      "loss": 1.5664,
       "step": 360
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 3.232395819846824e-06,
-      "loss": 1.4466,
+      "epoch": 0.3,
+      "learning_rate": 1.6337720473887418e-05,
+      "loss": 1.5691,
       "step": 361
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 3.183023610101544e-06,
-      "loss": 1.4407,
+      "epoch": 0.3,
+      "learning_rate": 1.631781342781885e-05,
+      "loss": 1.5844,
       "step": 362
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 3.1339599010438137e-06,
-      "loss": 1.4824,
+      "epoch": 0.3,
+      "learning_rate": 1.6297864625269347e-05,
+      "loss": 1.5679,
       "step": 363
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 3.085206913040428e-06,
-      "loss": 1.4635,
+      "epoch": 0.3,
+      "learning_rate": 1.627787419808703e-05,
+      "loss": 1.5586,
       "step": 364
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 3.036766852396561e-06,
-      "loss": 1.4263,
+      "epoch": 0.3,
+      "learning_rate": 1.6257842278395125e-05,
+      "loss": 1.575,
       "step": 365
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 2.9886419112559396e-06,
-      "loss": 1.4619,
+      "epoch": 0.3,
+      "learning_rate": 1.6237768998591097e-05,
+      "loss": 1.5624,
       "step": 366
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 2.9408342675016288e-06,
-      "loss": 1.4485,
+      "epoch": 0.3,
+      "learning_rate": 1.6217654491345767e-05,
+      "loss": 1.5384,
       "step": 367
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 2.893346084657477e-06,
-      "loss": 1.443,
+      "epoch": 0.3,
+      "learning_rate": 1.619749888960245e-05,
+      "loss": 1.5608,
       "step": 368
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 2.84617951179019e-06,
-      "loss": 1.4452,
+      "epoch": 0.3,
+      "learning_rate": 1.6177302326576064e-05,
+      "loss": 1.5779,
       "step": 369
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 2.799336683412106e-06,
-      "loss": 1.4306,
+      "epoch": 0.3,
+      "learning_rate": 1.6157064935752252e-05,
+      "loss": 1.541,
       "step": 370
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 2.752819719384573e-06,
-      "loss": 1.4571,
+      "epoch": 0.3,
+      "learning_rate": 1.6136786850886506e-05,
+      "loss": 1.5876,
       "step": 371
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 2.7066307248220193e-06,
-      "loss": 1.4602,
+      "epoch": 0.3,
+      "learning_rate": 1.6116468206003268e-05,
+      "loss": 1.5222,
       "step": 372
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 2.660771789996699e-06,
-      "loss": 1.4771,
+      "epoch": 0.3,
+      "learning_rate": 1.6096109135395074e-05,
+      "loss": 1.5336,
       "step": 373
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 2.6152449902440826e-06,
-      "loss": 1.5166,
+      "epoch": 0.31,
+      "learning_rate": 1.6075709773621625e-05,
+      "loss": 1.6012,
       "step": 374
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 2.5700523858689496e-06,
-      "loss": 1.4976,
+      "epoch": 0.31,
+      "learning_rate": 1.6055270255508925e-05,
+      "loss": 1.5729,
       "step": 375
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 2.5251960220521422e-06,
-      "loss": 1.4903,
+      "epoch": 0.31,
+      "learning_rate": 1.6034790716148397e-05,
+      "loss": 1.555,
       "step": 376
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 2.480677928758017e-06,
-      "loss": 1.4761,
+      "epoch": 0.31,
+      "learning_rate": 1.6014271290895963e-05,
+      "loss": 1.5126,
       "step": 377
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 2.4365001206425765e-06,
-      "loss": 1.471,
+      "epoch": 0.31,
+      "learning_rate": 1.5993712115371175e-05,
+      "loss": 1.6053,
       "step": 378
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 2.3926645969622907e-06,
-      "loss": 1.3977,
+      "epoch": 0.31,
+      "learning_rate": 1.597311332545629e-05,
+      "loss": 1.5432,
       "step": 379
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 2.349173341483629e-06,
-      "loss": 1.4499,
+      "epoch": 0.31,
+      "learning_rate": 1.595247505729541e-05,
+      "loss": 1.5364,
       "step": 380
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 2.3060283223932876e-06,
-      "loss": 1.4965,
+      "epoch": 0.31,
+      "learning_rate": 1.5931797447293553e-05,
+      "loss": 1.5321,
       "step": 381
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 2.2632314922091105e-06,
-      "loss": 1.4997,
+      "epoch": 0.31,
+      "learning_rate": 1.5911080632115755e-05,
+      "loss": 1.5476,
       "step": 382
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 2.2207847876917353e-06,
-      "loss": 1.4241,
+      "epoch": 0.31,
+      "learning_rate": 1.589032474868618e-05,
+      "loss": 1.5652,
       "step": 383
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 2.1786901297569473e-06,
-      "loss": 1.4439,
+      "epoch": 0.31,
+      "learning_rate": 1.58695299341872e-05,
+      "loss": 1.563,
       "step": 384
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 2.1369494233887365e-06,
-      "loss": 1.4172,
+      "epoch": 0.31,
+      "learning_rate": 1.5848696326058498e-05,
+      "loss": 1.552,
       "step": 385
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 2.0955645575531e-06,
-      "loss": 1.3756,
+      "epoch": 0.32,
+      "learning_rate": 1.5827824061996158e-05,
+      "loss": 1.555,
       "step": 386
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 2.054537405112559e-06,
-      "loss": 1.5035,
+      "epoch": 0.32,
+      "learning_rate": 1.580691327995175e-05,
+      "loss": 1.5695,
       "step": 387
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 2.013869822741392e-06,
-      "loss": 1.4611,
+      "epoch": 0.32,
+      "learning_rate": 1.578596411813143e-05,
+      "loss": 1.5342,
       "step": 388
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 1.9735636508416243e-06,
-      "loss": 1.408,
+      "epoch": 0.32,
+      "learning_rate": 1.576497671499501e-05,
+      "loss": 1.5439,
       "step": 389
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 1.933620713459723e-06,
-      "loss": 1.4531,
+      "epoch": 0.32,
+      "learning_rate": 1.5743951209255047e-05,
+      "loss": 1.5162,
       "step": 390
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 1.8940428182040715e-06,
-      "loss": 1.515,
+      "epoch": 0.32,
+      "learning_rate": 1.5722887739875947e-05,
+      "loss": 1.5501,
       "step": 391
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 1.8548317561631534e-06,
-      "loss": 1.3971,
+      "epoch": 0.32,
+      "learning_rate": 1.5701786446073016e-05,
+      "loss": 1.5259,
       "step": 392
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 1.8159893018244946e-06,
-      "loss": 1.4291,
+      "epoch": 0.32,
+      "learning_rate": 1.568064746731156e-05,
+      "loss": 1.5575,
       "step": 393
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 1.7775172129943719e-06,
-      "loss": 1.4801,
+      "epoch": 0.32,
+      "learning_rate": 1.5659470943305956e-05,
+      "loss": 1.554,
       "step": 394
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 1.7394172307182577e-06,
-      "loss": 1.4615,
+      "epoch": 0.32,
+      "learning_rate": 1.5638257014018722e-05,
+      "loss": 1.5814,
       "step": 395
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 1.7016910792020191e-06,
-      "loss": 1.4804,
+      "epoch": 0.32,
+      "learning_rate": 1.5617005819659614e-05,
+      "loss": 1.5256,
       "step": 396
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 1.6643404657339069e-06,
-      "loss": 1.4691,
+      "epoch": 0.32,
+      "learning_rate": 1.559571750068468e-05,
+      "loss": 1.5752,
       "step": 397
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 1.6273670806072827e-06,
-      "loss": 1.4577,
+      "epoch": 0.33,
+      "learning_rate": 1.5574392197795328e-05,
+      "loss": 1.566,
       "step": 398
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 1.59077259704413e-06,
-      "loss": 1.4457,
+      "epoch": 0.33,
+      "learning_rate": 1.5553030051937415e-05,
+      "loss": 1.5455,
       "step": 399
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 1.5545586711193194e-06,
-      "loss": 1.3774,
+      "epoch": 0.33,
+      "learning_rate": 1.55316312043003e-05,
+      "loss": 1.5944,
       "step": 400
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 1.5187269416856875e-06,
-      "loss": 1.4894,
+      "epoch": 0.33,
+      "learning_rate": 1.5510195796315913e-05,
+      "loss": 1.5543,
       "step": 401
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 1.4832790302998446e-06,
-      "loss": 1.4736,
+      "epoch": 0.33,
+      "learning_rate": 1.548872396965784e-05,
+      "loss": 1.5155,
       "step": 402
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 1.4482165411488113e-06,
-      "loss": 1.4095,
+      "epoch": 0.33,
+      "learning_rate": 1.546721586624034e-05,
+      "loss": 1.5782,
       "step": 403
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 1.4135410609774148e-06,
-      "loss": 1.4903,
+      "epoch": 0.33,
+      "learning_rate": 1.5445671628217467e-05,
+      "loss": 1.5515,
       "step": 404
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 1.3792541590164787e-06,
-      "loss": 1.4789,
+      "epoch": 0.33,
+      "learning_rate": 1.5424091397982085e-05,
+      "loss": 1.5528,
       "step": 405
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 1.3453573869118097e-06,
-      "loss": 1.4282,
+      "epoch": 0.33,
+      "learning_rate": 1.540247531816494e-05,
+      "loss": 1.5429,
       "step": 406
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 1.3118522786539744e-06,
-      "loss": 1.5141,
+      "epoch": 0.33,
+      "learning_rate": 1.5380823531633727e-05,
+      "loss": 1.5486,
       "step": 407
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 1.278740350508888e-06,
-      "loss": 1.5421,
+      "epoch": 0.33,
+      "learning_rate": 1.535913618149214e-05,
+      "loss": 1.5708,
       "step": 408
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 1.2460231009491908e-06,
-      "loss": 1.4952,
+      "epoch": 0.33,
+      "learning_rate": 1.5337413411078915e-05,
+      "loss": 1.5113,
       "step": 409
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 1.213702010586434e-06,
-      "loss": 1.4756,
+      "epoch": 0.34,
+      "learning_rate": 1.5315655363966896e-05,
+      "loss": 1.5773,
       "step": 410
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 1.181778542104075e-06,
-      "loss": 1.4918,
+      "epoch": 0.34,
+      "learning_rate": 1.529386218396209e-05,
+      "loss": 1.5463,
       "step": 411
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 1.150254140191286e-06,
-      "loss": 1.4367,
+      "epoch": 0.34,
+      "learning_rate": 1.5272034015102698e-05,
+      "loss": 1.5272,
       "step": 412
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 1.119130231477571e-06,
-      "loss": 1.4663,
+      "epoch": 0.34,
+      "learning_rate": 1.5250171001658171e-05,
+      "loss": 1.5479,
       "step": 413
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 1.0884082244682114e-06,
-      "loss": 1.5101,
+      "epoch": 0.34,
+      "learning_rate": 1.5228273288128273e-05,
+      "loss": 1.5214,
       "step": 414
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 1.0580895094805189e-06,
-      "loss": 1.4738,
+      "epoch": 0.34,
+      "learning_rate": 1.5206341019242108e-05,
+      "loss": 1.5527,
       "step": 415
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 1.028175458580918e-06,
-      "loss": 1.4324,
+      "epoch": 0.34,
+      "learning_rate": 1.518437433995716e-05,
+      "loss": 1.5576,
       "step": 416
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 9.986674255228546e-07,
-      "loss": 1.4551,
+      "epoch": 0.34,
+      "learning_rate": 1.5162373395458345e-05,
+      "loss": 1.5843,
       "step": 417
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 9.695667456855272e-07,
-      "loss": 1.4711,
+      "epoch": 0.34,
+      "learning_rate": 1.5140338331157053e-05,
+      "loss": 1.5552,
       "step": 418
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 9.408747360134651e-07,
-      "loss": 1.4835,
+      "epoch": 0.34,
+      "learning_rate": 1.5118269292690184e-05,
+      "loss": 1.5856,
       "step": 419
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 9.125926949569153e-07,
-      "loss": 1.4696,
+      "epoch": 0.34,
+      "learning_rate": 1.5096166425919176e-05,
+      "loss": 1.5227,
       "step": 420
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 8.84721902413097e-07,
-      "loss": 1.462,
+      "epoch": 0.34,
+      "learning_rate": 1.5074029876929057e-05,
+      "loss": 1.5702,
       "step": 421
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 8.572636196682749e-07,
-      "loss": 1.467,
+      "epoch": 0.35,
+      "learning_rate": 1.505185979202747e-05,
+      "loss": 1.5476,
       "step": 422
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 8.302190893406691e-07,
-      "loss": 1.467,
+      "epoch": 0.35,
+      "learning_rate": 1.5029656317743704e-05,
+      "loss": 1.5583,
       "step": 423
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 8.035895353242373e-07,
-      "loss": 1.5104,
+      "epoch": 0.35,
+      "learning_rate": 1.5007419600827735e-05,
+      "loss": 1.5613,
       "step": 424
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 7.773761627332799e-07,
-      "loss": 1.4397,
+      "epoch": 0.35,
+      "learning_rate": 1.4985149788249248e-05,
+      "loss": 1.5411,
       "step": 425
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 7.515801578479032e-07,
-      "loss": 1.4398,
+      "epoch": 0.35,
+      "learning_rate": 1.4962847027196661e-05,
+      "loss": 1.5052,
       "step": 426
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 7.26202688060329e-07,
-      "loss": 1.4161,
+      "epoch": 0.35,
+      "learning_rate": 1.4940511465076173e-05,
+      "loss": 1.5789,
       "step": 427
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 7.01244901822078e-07,
-      "loss": 1.5453,
+      "epoch": 0.35,
+      "learning_rate": 1.4918143249510764e-05,
+      "loss": 1.5514,
       "step": 428
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 6.76707928591983e-07,
-      "loss": 1.4451,
+      "epoch": 0.35,
+      "learning_rate": 1.4895742528339239e-05,
+      "loss": 1.5699,
       "step": 429
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 6.525928787850866e-07,
-      "loss": 1.4359,
+      "epoch": 0.35,
+      "learning_rate": 1.487330944961524e-05,
+      "loss": 1.5552,
       "step": 430
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 6.289008437223798e-07,
-      "loss": 1.4948,
+      "epoch": 0.35,
+      "learning_rate": 1.4850844161606264e-05,
+      "loss": 1.5239,
       "step": 431
     },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4828346812792702e-05,
+      "loss": 1.5541,
+      "step": 432
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4805817551866839e-05,
+      "loss": 1.5264,
+      "step": 433
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4783256527731878e-05,
+      "loss": 1.5321,
+      "step": 434
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4760663889500953e-05,
+      "loss": 1.5413,
+      "step": 435
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4738039786496148e-05,
+      "loss": 1.5131,
+      "step": 436
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4715384368247507e-05,
+      "loss": 1.5229,
+      "step": 437
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4692697784492052e-05,
+      "loss": 1.5143,
+      "step": 438
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.466998018517278e-05,
+      "loss": 1.554,
+      "step": 439
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4647231720437687e-05,
+      "loss": 1.5265,
+      "step": 440
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.462445254063876e-05,
+      "loss": 1.5723,
+      "step": 441
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4601642796331001e-05,
+      "loss": 1.5362,
+      "step": 442
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4578802638271414e-05,
+      "loss": 1.5337,
+      "step": 443
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4555932217418028e-05,
+      "loss": 1.5464,
+      "step": 444
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4533031684928876e-05,
+      "loss": 1.5602,
+      "step": 445
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.451010119216102e-05,
+      "loss": 1.537,
+      "step": 446
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4487140890669533e-05,
+      "loss": 1.5097,
+      "step": 447
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4464150932206507e-05,
+      "loss": 1.5563,
+      "step": 448
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4441131468720038e-05,
+      "loss": 1.5616,
+      "step": 449
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.441808265235324e-05,
+      "loss": 1.5334,
+      "step": 450
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4395004635443232e-05,
+      "loss": 1.525,
+      "step": 451
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4371897570520116e-05,
+      "loss": 1.5306,
+      "step": 452
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4348761610305994e-05,
+      "loss": 1.5039,
+      "step": 453
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4325596907713938e-05,
+      "loss": 1.5359,
+      "step": 454
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4302403615846994e-05,
+      "loss": 1.5154,
+      "step": 455
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4279181887997158e-05,
+      "loss": 1.5771,
+      "step": 456
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4255931877644374e-05,
+      "loss": 1.5744,
+      "step": 457
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.423265373845551e-05,
+      "loss": 1.5382,
+      "step": 458
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4209347624283352e-05,
+      "loss": 1.55,
+      "step": 459
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4186013689165574e-05,
+      "loss": 1.525,
+      "step": 460
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4162652087323735e-05,
+      "loss": 1.5166,
+      "step": 461
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4139262973162246e-05,
+      "loss": 1.5254,
+      "step": 462
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4115846501267361e-05,
+      "loss": 1.5826,
+      "step": 463
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4092402826406151e-05,
+      "loss": 1.5803,
+      "step": 464
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4068932103525472e-05,
+      "loss": 1.5332,
+      "step": 465
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.404543448775096e-05,
+      "loss": 1.5448,
+      "step": 466
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4021910134385983e-05,
+      "loss": 1.5654,
+      "step": 467
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3998359198910642e-05,
+      "loss": 1.5371,
+      "step": 468
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3974781836980713e-05,
+      "loss": 1.5361,
+      "step": 469
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3951178204426636e-05,
+      "loss": 1.5748,
+      "step": 470
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3927548457252491e-05,
+      "loss": 1.5647,
+      "step": 471
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3903892751634949e-05,
+      "loss": 1.5231,
+      "step": 472
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3880211243922251e-05,
+      "loss": 1.5684,
+      "step": 473
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3856504090633175e-05,
+      "loss": 1.5472,
+      "step": 474
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3832771448455996e-05,
+      "loss": 1.5587,
+      "step": 475
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3809013474247457e-05,
+      "loss": 1.5091,
+      "step": 476
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.378523032503172e-05,
+      "loss": 1.5243,
+      "step": 477
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3761422157999346e-05,
+      "loss": 1.5217,
+      "step": 478
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3737589130506246e-05,
+      "loss": 1.528,
+      "step": 479
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3713731400072635e-05,
+      "loss": 1.5345,
+      "step": 480
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3689849124382006e-05,
+      "loss": 1.5176,
+      "step": 481
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3665942461280074e-05,
+      "loss": 1.5613,
+      "step": 482
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3642011568773743e-05,
+      "loss": 1.4981,
+      "step": 483
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3618056605030048e-05,
+      "loss": 1.5381,
+      "step": 484
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3594077728375129e-05,
+      "loss": 1.5159,
+      "step": 485
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3570075097293171e-05,
+      "loss": 1.5436,
+      "step": 486
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3546048870425356e-05,
+      "loss": 1.5194,
+      "step": 487
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3521999206568826e-05,
+      "loss": 1.4828,
+      "step": 488
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3497926264675613e-05,
+      "loss": 1.5387,
+      "step": 489
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3473830203851615e-05,
+      "loss": 1.5005,
+      "step": 490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3449711183355519e-05,
+      "loss": 1.5352,
+      "step": 491
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.342556936259777e-05,
+      "loss": 1.5321,
+      "step": 492
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.34014049011395e-05,
+      "loss": 1.5403,
+      "step": 493
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3377217958691484e-05,
+      "loss": 1.5865,
+      "step": 494
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3353008695113081e-05,
+      "loss": 1.5507,
+      "step": 495
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3328777270411174e-05,
+      "loss": 1.5404,
+      "step": 496
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3304523844739123e-05,
+      "loss": 1.5474,
+      "step": 497
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.328024857839569e-05,
+      "loss": 1.5206,
+      "step": 498
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3255951631824004e-05,
+      "loss": 1.5451,
+      "step": 499
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.323163316561047e-05,
+      "loss": 1.5084,
+      "step": 500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3207293340483727e-05,
+      "loss": 1.509,
+      "step": 501
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.318293231731359e-05,
+      "loss": 1.5512,
+      "step": 502
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3158550257109973e-05,
+      "loss": 1.5205,
+      "step": 503
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3134147321021828e-05,
+      "loss": 1.5494,
+      "step": 504
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.310972367033609e-05,
+      "loss": 1.5237,
+      "step": 505
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3085279466476595e-05,
+      "loss": 1.5019,
+      "step": 506
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3060814871003026e-05,
+      "loss": 1.532,
+      "step": 507
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3036330045609842e-05,
+      "loss": 1.5421,
+      "step": 508
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3011825152125204e-05,
+      "loss": 1.5403,
+      "step": 509
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2987300352509915e-05,
+      "loss": 1.5125,
+      "step": 510
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2962755808856341e-05,
+      "loss": 1.5301,
+      "step": 511
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2938191683387342e-05,
+      "loss": 1.5196,
+      "step": 512
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2913608138455202e-05,
+      "loss": 1.5312,
+      "step": 513
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2889005336540552e-05,
+      "loss": 1.5319,
+      "step": 514
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2864383440251302e-05,
+      "loss": 1.5171,
+      "step": 515
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2839742612321564e-05,
+      "loss": 1.517,
+      "step": 516
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2815083015610572e-05,
+      "loss": 1.5774,
+      "step": 517
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2790404813101607e-05,
+      "loss": 1.5382,
+      "step": 518
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2765708167900925e-05,
+      "loss": 1.5212,
+      "step": 519
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2740993243236676e-05,
+      "loss": 1.5092,
+      "step": 520
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2716260202457828e-05,
+      "loss": 1.525,
+      "step": 521
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2691509209033078e-05,
+      "loss": 1.5068,
+      "step": 522
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.266674042654978e-05,
+      "loss": 1.5253,
+      "step": 523
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2641954018712863e-05,
+      "loss": 1.5444,
+      "step": 524
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2617150149343745e-05,
+      "loss": 1.5144,
+      "step": 525
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2592328982379256e-05,
+      "loss": 1.5106,
+      "step": 526
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2567490681870554e-05,
+      "loss": 1.5135,
+      "step": 527
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2542635411982028e-05,
+      "loss": 1.5339,
+      "step": 528
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2517763336990231e-05,
+      "loss": 1.5207,
+      "step": 529
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.249287462128279e-05,
+      "loss": 1.5188,
+      "step": 530
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2467969429357303e-05,
+      "loss": 1.4879,
+      "step": 531
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2443047925820277e-05,
+      "loss": 1.5045,
+      "step": 532
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2418110275386028e-05,
+      "loss": 1.5512,
+      "step": 533
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2393156642875579e-05,
+      "loss": 1.5395,
+      "step": 534
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2368187193215597e-05,
+      "loss": 1.4893,
+      "step": 535
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2343202091437279e-05,
+      "loss": 1.5357,
+      "step": 536
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2318201502675285e-05,
+      "loss": 1.5113,
+      "step": 537
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.229318559216662e-05,
+      "loss": 1.5082,
+      "step": 538
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2268154525249565e-05,
+      "loss": 1.4945,
+      "step": 539
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2243108467362572e-05,
+      "loss": 1.5286,
+      "step": 540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2218047584043172e-05,
+      "loss": 1.4984,
+      "step": 541
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.219297204092688e-05,
+      "loss": 1.565,
+      "step": 542
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2167882003746103e-05,
+      "loss": 1.5193,
+      "step": 543
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.214277763832905e-05,
+      "loss": 1.5345,
+      "step": 544
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2117659110598629e-05,
+      "loss": 1.5158,
+      "step": 545
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.209252658657134e-05,
+      "loss": 1.5305,
+      "step": 546
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2067380232356202e-05,
+      "loss": 1.5081,
+      "step": 547
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2042220214153637e-05,
+      "loss": 1.5055,
+      "step": 548
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2017046698254376e-05,
+      "loss": 1.4961,
+      "step": 549
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1991859851038362e-05,
+      "loss": 1.4647,
+      "step": 550
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1966659838973649e-05,
+      "loss": 1.5288,
+      "step": 551
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1941446828615305e-05,
+      "loss": 1.5291,
+      "step": 552
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1916220986604296e-05,
+      "loss": 1.5273,
+      "step": 553
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1890982479666412e-05,
+      "loss": 1.5122,
+      "step": 554
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1865731474611141e-05,
+      "loss": 1.5419,
+      "step": 555
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1840468138330579e-05,
+      "loss": 1.5139,
+      "step": 556
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1815192637798314e-05,
+      "loss": 1.4909,
+      "step": 557
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1789905140068347e-05,
+      "loss": 1.5141,
+      "step": 558
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1764605812273957e-05,
+      "loss": 1.5062,
+      "step": 559
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.173929482162662e-05,
+      "loss": 1.5417,
+      "step": 560
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1713972335414896e-05,
+      "loss": 1.5351,
+      "step": 561
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1688638521003321e-05,
+      "loss": 1.5173,
+      "step": 562
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1663293545831302e-05,
+      "loss": 1.5466,
+      "step": 563
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1637937577412011e-05,
+      "loss": 1.4859,
+      "step": 564
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1612570783331278e-05,
+      "loss": 1.5246,
+      "step": 565
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1587193331246487e-05,
+      "loss": 1.5134,
+      "step": 566
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1561805388885454e-05,
+      "loss": 1.5183,
+      "step": 567
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1536407124045349e-05,
+      "loss": 1.5665,
+      "step": 568
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1510998704591543e-05,
+      "loss": 1.5153,
+      "step": 569
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1485580298456536e-05,
+      "loss": 1.558,
+      "step": 570
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1460152073638828e-05,
+      "loss": 1.5233,
+      "step": 571
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.143471419820182e-05,
+      "loss": 1.5311,
+      "step": 572
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1409266840272685e-05,
+      "loss": 1.5001,
+      "step": 573
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1383810168041284e-05,
+      "loss": 1.513,
+      "step": 574
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1358344349759031e-05,
+      "loss": 1.4718,
+      "step": 575
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.133286955373779e-05,
+      "loss": 1.4876,
+      "step": 576
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.130738594834876e-05,
+      "loss": 1.5036,
+      "step": 577
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1281893702021364e-05,
+      "loss": 1.5068,
+      "step": 578
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1256392983242143e-05,
+      "loss": 1.5037,
+      "step": 579
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1230883960553627e-05,
+      "loss": 1.4809,
+      "step": 580
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1205366802553231e-05,
+      "loss": 1.5242,
+      "step": 581
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1179841677892143e-05,
+      "loss": 1.5166,
+      "step": 582
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1154308755274201e-05,
+      "loss": 1.5167,
+      "step": 583
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1128768203454783e-05,
+      "loss": 1.5383,
+      "step": 584
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1103220191239689e-05,
+      "loss": 1.5197,
+      "step": 585
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1077664887484036e-05,
+      "loss": 1.5139,
+      "step": 586
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.105210246109112e-05,
+      "loss": 1.5296,
+      "step": 587
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1026533081011324e-05,
+      "loss": 1.5002,
+      "step": 588
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1000956916240985e-05,
+      "loss": 1.5186,
+      "step": 589
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.097537413582128e-05,
+      "loss": 1.5046,
+      "step": 590
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0949784908837117e-05,
+      "loss": 1.5066,
+      "step": 591
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0924189404416007e-05,
+      "loss": 1.5024,
+      "step": 592
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0898587791726956e-05,
+      "loss": 1.484,
+      "step": 593
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0872980239979331e-05,
+      "loss": 1.5228,
+      "step": 594
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.084736691842176e-05,
+      "loss": 1.5109,
+      "step": 595
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0821747996341005e-05,
+      "loss": 1.4942,
+      "step": 596
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0796123643060847e-05,
+      "loss": 1.4914,
+      "step": 597
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0770494027940957e-05,
+      "loss": 1.4946,
+      "step": 598
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.074485932037578e-05,
+      "loss": 1.514,
+      "step": 599
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0719219689793435e-05,
+      "loss": 1.5256,
+      "step": 600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0693575305654558e-05,
+      "loss": 1.5331,
+      "step": 601
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0667926337451217e-05,
+      "loss": 1.5268,
+      "step": 602
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0642272954705775e-05,
+      "loss": 1.4878,
+      "step": 603
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0616615326969768e-05,
+      "loss": 1.513,
+      "step": 604
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0590953623822795e-05,
+      "loss": 1.5174,
+      "step": 605
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.056528801487138e-05,
+      "loss": 1.5498,
+      "step": 606
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0539618669747878e-05,
+      "loss": 1.5225,
+      "step": 607
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0513945758109319e-05,
+      "loss": 1.5104,
+      "step": 608
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0488269449636316e-05,
+      "loss": 1.5471,
+      "step": 609
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0462589914031937e-05,
+      "loss": 1.4924,
+      "step": 610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0436907321020573e-05,
+      "loss": 1.5351,
+      "step": 611
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0411221840346817e-05,
+      "loss": 1.5236,
+      "step": 612
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0385533641774354e-05,
+      "loss": 1.492,
+      "step": 613
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0359842895084833e-05,
+      "loss": 1.5294,
+      "step": 614
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0334149770076747e-05,
+      "loss": 1.502,
+      "step": 615
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0308454436564298e-05,
+      "loss": 1.503,
+      "step": 616
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0282757064376296e-05,
+      "loss": 1.5607,
+      "step": 617
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0257057823355018e-05,
+      "loss": 1.5096,
+      "step": 618
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0231356883355092e-05,
+      "loss": 1.5303,
+      "step": 619
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0205654414242383e-05,
+      "loss": 1.4697,
+      "step": 620
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0179950585892855e-05,
+      "loss": 1.5153,
+      "step": 621
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.015424556819146e-05,
+      "loss": 1.518,
+      "step": 622
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0128539531031e-05,
+      "loss": 1.5318,
+      "step": 623
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0102832644311038e-05,
+      "loss": 1.5339,
+      "step": 624
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0077125077936727e-05,
+      "loss": 1.5286,
+      "step": 625
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0051417001817727e-05,
+      "loss": 1.5,
+      "step": 626
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.002570858586706e-05,
+      "loss": 1.4887,
+      "step": 627
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1e-05,
+      "loss": 1.5339,
+      "step": 628
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.97429141413294e-06,
+      "loss": 1.5158,
+      "step": 629
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.94858299818228e-06,
+      "loss": 1.4884,
+      "step": 630
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.922874922063277e-06,
+      "loss": 1.5124,
+      "step": 631
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.897167355688964e-06,
+      "loss": 1.5286,
+      "step": 632
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.871460468969001e-06,
+      "loss": 1.5335,
+      "step": 633
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.845754431808544e-06,
+      "loss": 1.5615,
+      "step": 634
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.820049414107145e-06,
+      "loss": 1.4869,
+      "step": 635
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.79434558575762e-06,
+      "loss": 1.492,
+      "step": 636
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.76864311664491e-06,
+      "loss": 1.5013,
+      "step": 637
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.742942176644987e-06,
+      "loss": 1.5183,
+      "step": 638
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.717242935623707e-06,
+      "loss": 1.4852,
+      "step": 639
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.691545563435703e-06,
+      "loss": 1.5282,
+      "step": 640
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.665850229923258e-06,
+      "loss": 1.5197,
+      "step": 641
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.640157104915168e-06,
+      "loss": 1.4608,
+      "step": 642
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.614466358225649e-06,
+      "loss": 1.5153,
+      "step": 643
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.588778159653188e-06,
+      "loss": 1.4592,
+      "step": 644
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.563092678979432e-06,
+      "loss": 1.5082,
+      "step": 645
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.537410085968063e-06,
+      "loss": 1.4705,
+      "step": 646
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.511730550363685e-06,
+      "loss": 1.4846,
+      "step": 647
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.486054241890685e-06,
+      "loss": 1.521,
+      "step": 648
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.460381330252127e-06,
+      "loss": 1.5349,
+      "step": 649
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.434711985128621e-06,
+      "loss": 1.5277,
+      "step": 650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.409046376177208e-06,
+      "loss": 1.47,
+      "step": 651
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.383384673030237e-06,
+      "loss": 1.5058,
+      "step": 652
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.357727045294229e-06,
+      "loss": 1.5427,
+      "step": 653
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.332073662548785e-06,
+      "loss": 1.4975,
+      "step": 654
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.306424694345447e-06,
+      "loss": 1.5062,
+      "step": 655
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.28078031020657e-06,
+      "loss": 1.511,
+      "step": 656
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.255140679624219e-06,
+      "loss": 1.5649,
+      "step": 657
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.229505972059048e-06,
+      "loss": 1.4819,
+      "step": 658
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.203876356939155e-06,
+      "loss": 1.5152,
+      "step": 659
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.178252003658995e-06,
+      "loss": 1.4639,
+      "step": 660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.152633081578244e-06,
+      "loss": 1.457,
+      "step": 661
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.127019760020672e-06,
+      "loss": 1.5217,
+      "step": 662
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.101412208273045e-06,
+      "loss": 1.4973,
+      "step": 663
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.075810595583994e-06,
+      "loss": 1.5108,
+      "step": 664
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.050215091162884e-06,
+      "loss": 1.5186,
+      "step": 665
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.024625864178722e-06,
+      "loss": 1.5456,
+      "step": 666
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.999043083759016e-06,
+      "loss": 1.5038,
+      "step": 667
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.973466918988676e-06,
+      "loss": 1.5569,
+      "step": 668
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.947897538908883e-06,
+      "loss": 1.5544,
+      "step": 669
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.922335112515967e-06,
+      "loss": 1.5157,
+      "step": 670
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.896779808760313e-06,
+      "loss": 1.4928,
+      "step": 671
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.87123179654522e-06,
+      "loss": 1.5128,
+      "step": 672
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.845691244725802e-06,
+      "loss": 1.4782,
+      "step": 673
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.820158322107857e-06,
+      "loss": 1.4956,
+      "step": 674
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.79463319744677e-06,
+      "loss": 1.5242,
+      "step": 675
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.769116039446376e-06,
+      "loss": 1.4854,
+      "step": 676
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.743607016757859e-06,
+      "loss": 1.552,
+      "step": 677
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.718106297978637e-06,
+      "loss": 1.4937,
+      "step": 678
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.692614051651243e-06,
+      "loss": 1.5348,
+      "step": 679
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.667130446262214e-06,
+      "loss": 1.4935,
+      "step": 680
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.64165565024097e-06,
+      "loss": 1.5037,
+      "step": 681
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.616189831958717e-06,
+      "loss": 1.5422,
+      "step": 682
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.590733159727316e-06,
+      "loss": 1.5013,
+      "step": 683
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.565285801798184e-06,
+      "loss": 1.4945,
+      "step": 684
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.539847926361174e-06,
+      "loss": 1.4727,
+      "step": 685
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.514419701543468e-06,
+      "loss": 1.486,
+      "step": 686
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.48900129540846e-06,
+      "loss": 1.4953,
+      "step": 687
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.463592875954653e-06,
+      "loss": 1.5129,
+      "step": 688
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.438194611114548e-06,
+      "loss": 1.5252,
+      "step": 689
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.412806668753515e-06,
+      "loss": 1.5052,
+      "step": 690
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.387429216668722e-06,
+      "loss": 1.4768,
+      "step": 691
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.362062422587994e-06,
+      "loss": 1.5244,
+      "step": 692
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.336706454168701e-06,
+      "loss": 1.5109,
+      "step": 693
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.311361478996684e-06,
+      "loss": 1.4886,
+      "step": 694
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.286027664585107e-06,
+      "loss": 1.4735,
+      "step": 695
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.260705178373381e-06,
+      "loss": 1.4719,
+      "step": 696
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.235394187726046e-06,
+      "loss": 1.5143,
+      "step": 697
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.210094859931656e-06,
+      "loss": 1.5149,
+      "step": 698
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.184807362201687e-06,
+      "loss": 1.5029,
+      "step": 699
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.159531861669428e-06,
+      "loss": 1.4877,
+      "step": 700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.134268525388862e-06,
+      "loss": 1.5006,
+      "step": 701
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.10901752033359e-06,
+      "loss": 1.4649,
+      "step": 702
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.083779013395707e-06,
+      "loss": 1.5278,
+      "step": 703
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.058553171384699e-06,
+      "loss": 1.4807,
+      "step": 704
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.033340161026351e-06,
+      "loss": 1.507,
+      "step": 705
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.008140148961642e-06,
+      "loss": 1.4854,
+      "step": 706
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.982953301745626e-06,
+      "loss": 1.4593,
+      "step": 707
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.957779785846366e-06,
+      "loss": 1.4996,
+      "step": 708
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.932619767643801e-06,
+      "loss": 1.5045,
+      "step": 709
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.907473413428661e-06,
+      "loss": 1.5039,
+      "step": 710
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.882340889401376e-06,
+      "loss": 1.4655,
+      "step": 711
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.857222361670953e-06,
+      "loss": 1.4736,
+      "step": 712
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.832117996253899e-06,
+      "loss": 1.5249,
+      "step": 713
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.807027959073126e-06,
+      "loss": 1.4742,
+      "step": 714
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.781952415956831e-06,
+      "loss": 1.497,
+      "step": 715
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.756891532637428e-06,
+      "loss": 1.4684,
+      "step": 716
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.731845474750438e-06,
+      "loss": 1.4712,
+      "step": 717
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.706814407833383e-06,
+      "loss": 1.4959,
+      "step": 718
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.681798497324717e-06,
+      "loss": 1.4895,
+      "step": 719
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.656797908562723e-06,
+      "loss": 1.5182,
+      "step": 720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.631812806784406e-06,
+      "loss": 1.4932,
+      "step": 721
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.606843357124426e-06,
+      "loss": 1.4802,
+      "step": 722
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.581889724613976e-06,
+      "loss": 1.4572,
+      "step": 723
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.556952074179722e-06,
+      "loss": 1.5025,
+      "step": 724
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.532030570642699e-06,
+      "loss": 1.5279,
+      "step": 725
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5071253787172125e-06,
+      "loss": 1.4678,
+      "step": 726
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4822366630097695e-06,
+      "loss": 1.4903,
+      "step": 727
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.457364588017976e-06,
+      "loss": 1.522,
+      "step": 728
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.432509318129451e-06,
+      "loss": 1.517,
+      "step": 729
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.407671017620745e-06,
+      "loss": 1.4934,
+      "step": 730
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.382849850656258e-06,
+      "loss": 1.4617,
+      "step": 731
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.358045981287141e-06,
+      "loss": 1.5171,
+      "step": 732
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.333259573450222e-06,
+      "loss": 1.4506,
+      "step": 733
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3084907909669265e-06,
+      "loss": 1.4695,
+      "step": 734
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2837397975421744e-06,
+      "loss": 1.4734,
+      "step": 735
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.259006756763327e-06,
+      "loss": 1.4495,
+      "step": 736
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.234291832099079e-06,
+      "loss": 1.4909,
+      "step": 737
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.209595186898397e-06,
+      "loss": 1.5015,
+      "step": 738
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.184916984389434e-06,
+      "loss": 1.4922,
+      "step": 739
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.160257387678439e-06,
+      "loss": 1.5028,
+      "step": 740
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1356165597486996e-06,
+      "loss": 1.4723,
+      "step": 741
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.110994663459451e-06,
+      "loss": 1.4615,
+      "step": 742
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0863918615448e-06,
+      "loss": 1.5286,
+      "step": 743
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.061808316612658e-06,
+      "loss": 1.4474,
+      "step": 744
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.037244191143662e-06,
+      "loss": 1.5005,
+      "step": 745
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.012699647490086e-06,
+      "loss": 1.4831,
+      "step": 746
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.988174847874798e-06,
+      "loss": 1.4932,
+      "step": 747
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.963669954390162e-06,
+      "loss": 1.4578,
+      "step": 748
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.939185128996977e-06,
+      "loss": 1.4829,
+      "step": 749
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.914720533523411e-06,
+      "loss": 1.5128,
+      "step": 750
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8902763296639145e-06,
+      "loss": 1.5171,
+      "step": 751
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.865852678978173e-06,
+      "loss": 1.4876,
+      "step": 752
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.841449742890031e-06,
+      "loss": 1.4936,
+      "step": 753
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.817067682686413e-06,
+      "loss": 1.4912,
+      "step": 754
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.792706659516275e-06,
+      "loss": 1.4854,
+      "step": 755
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.768366834389534e-06,
+      "loss": 1.4645,
+      "step": 756
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7440483681759985e-06,
+      "loss": 1.5392,
+      "step": 757
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.719751421604309e-06,
+      "loss": 1.4595,
+      "step": 758
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6954761552608824e-06,
+      "loss": 1.4636,
+      "step": 759
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.671222729588829e-06,
+      "loss": 1.5119,
+      "step": 760
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.646991304886923e-06,
+      "loss": 1.5072,
+      "step": 761
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.622782041308519e-06,
+      "loss": 1.4949,
+      "step": 762
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5985950988605e-06,
+      "loss": 1.5071,
+      "step": 763
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.574430637402235e-06,
+      "loss": 1.5156,
+      "step": 764
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.5502888166444835e-06,
+      "loss": 1.4994,
+      "step": 765
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.526169796148388e-06,
+      "loss": 1.4984,
+      "step": 766
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.50207373532439e-06,
+      "loss": 1.4642,
+      "step": 767
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.478000793431177e-06,
+      "loss": 1.4458,
+      "step": 768
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.453951129574644e-06,
+      "loss": 1.4577,
+      "step": 769
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.429924902706832e-06,
+      "loss": 1.4708,
+      "step": 770
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.405922271624874e-06,
+      "loss": 1.4926,
+      "step": 771
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.381943394969954e-06,
+      "loss": 1.5007,
+      "step": 772
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.357988431226262e-06,
+      "loss": 1.5075,
+      "step": 773
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.334057538719928e-06,
+      "loss": 1.4804,
+      "step": 774
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.310150875617995e-06,
+      "loss": 1.4649,
+      "step": 775
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.286268599927368e-06,
+      "loss": 1.5092,
+      "step": 776
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.262410869493758e-06,
+      "loss": 1.5112,
+      "step": 777
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.238577842000657e-06,
+      "loss": 1.5036,
+      "step": 778
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.214769674968283e-06,
+      "loss": 1.5185,
+      "step": 779
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1909865257525445e-06,
+      "loss": 1.5063,
+      "step": 780
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.167228551544007e-06,
+      "loss": 1.5102,
+      "step": 781
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.143495909366827e-06,
+      "loss": 1.4845,
+      "step": 782
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.119788756077749e-06,
+      "loss": 1.4679,
+      "step": 783
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0961072483650526e-06,
+      "loss": 1.4548,
+      "step": 784
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0724515427475105e-06,
+      "loss": 1.4871,
+      "step": 785
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.048821795573364e-06,
+      "loss": 1.4905,
+      "step": 786
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.025218163019293e-06,
+      "loss": 1.4674,
+      "step": 787
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0016408010893615e-06,
+      "loss": 1.4697,
+      "step": 788
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.978089865614016e-06,
+      "loss": 1.4725,
+      "step": 789
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.954565512249045e-06,
+      "loss": 1.4913,
+      "step": 790
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.931067896474529e-06,
+      "loss": 1.4639,
+      "step": 791
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9075971735938554e-06,
+      "loss": 1.516,
+      "step": 792
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.884153498732642e-06,
+      "loss": 1.4964,
+      "step": 793
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.860737026837757e-06,
+      "loss": 1.5188,
+      "step": 794
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.837347912676272e-06,
+      "loss": 1.48,
+      "step": 795
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.813986310834431e-06,
+      "loss": 1.506,
+      "step": 796
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.790652375716653e-06,
+      "loss": 1.4669,
+      "step": 797
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.767346261544491e-06,
+      "loss": 1.4757,
+      "step": 798
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.744068122355627e-06,
+      "loss": 1.5137,
+      "step": 799
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.720818112002842e-06,
+      "loss": 1.4895,
+      "step": 800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.697596384153009e-06,
+      "loss": 1.504,
+      "step": 801
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6744030922860625e-06,
+      "loss": 1.5133,
+      "step": 802
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.651238389694007e-06,
+      "loss": 1.4883,
+      "step": 803
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.628102429479887e-06,
+      "loss": 1.483,
+      "step": 804
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.60499536455677e-06,
+      "loss": 1.4806,
+      "step": 805
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.581917347646763e-06,
+      "loss": 1.4874,
+      "step": 806
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.558868531279967e-06,
+      "loss": 1.4918,
+      "step": 807
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.535849067793499e-06,
+      "loss": 1.504,
+      "step": 808
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.512859109330473e-06,
+      "loss": 1.4576,
+      "step": 809
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.48989880783898e-06,
+      "loss": 1.5117,
+      "step": 810
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4669683150711235e-06,
+      "loss": 1.4818,
+      "step": 811
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.444067782581976e-06,
+      "loss": 1.4856,
+      "step": 812
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.421197361728587e-06,
+      "loss": 1.479,
+      "step": 813
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.398357203669e-06,
+      "loss": 1.4429,
+      "step": 814
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.375547459361243e-06,
+      "loss": 1.5004,
+      "step": 815
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.352768279562315e-06,
+      "loss": 1.4646,
+      "step": 816
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.330019814827219e-06,
+      "loss": 1.4968,
+      "step": 817
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.307302215507951e-06,
+      "loss": 1.5241,
+      "step": 818
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.284615631752495e-06,
+      "loss": 1.4951,
+      "step": 819
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2619602135038586e-06,
+      "loss": 1.4876,
+      "step": 820
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.239336110499054e-06,
+      "loss": 1.4786,
+      "step": 821
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.216743472268126e-06,
+      "loss": 1.4258,
+      "step": 822
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.194182448133163e-06,
+      "loss": 1.5021,
+      "step": 823
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.171653187207299e-06,
+      "loss": 1.5046,
+      "step": 824
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.149155838393737e-06,
+      "loss": 1.477,
+      "step": 825
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.126690550384765e-06,
+      "loss": 1.4718,
+      "step": 826
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.104257471660763e-06,
+      "loss": 1.4797,
+      "step": 827
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0818567504892365e-06,
+      "loss": 1.4911,
+      "step": 828
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.059488534923831e-06,
+      "loss": 1.5326,
+      "step": 829
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.037152972803342e-06,
+      "loss": 1.466,
+      "step": 830
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.014850211750757e-06,
+      "loss": 1.5072,
+      "step": 831
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9925803991722696e-06,
+      "loss": 1.4716,
+      "step": 832
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.970343682256299e-06,
+      "loss": 1.4813,
+      "step": 833
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.948140207972533e-06,
+      "loss": 1.5041,
+      "step": 834
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.925970123070944e-06,
+      "loss": 1.4816,
+      "step": 835
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.903833574080825e-06,
+      "loss": 1.5077,
+      "step": 836
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.881730707309821e-06,
+      "loss": 1.552,
+      "step": 837
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.859661668842949e-06,
+      "loss": 1.5323,
+      "step": 838
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.837626604541658e-06,
+      "loss": 1.4837,
+      "step": 839
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.815625660042845e-06,
+      "loss": 1.526,
+      "step": 840
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.793658980757895e-06,
+      "loss": 1.4679,
+      "step": 841
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.771726711871727e-06,
+      "loss": 1.4888,
+      "step": 842
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.749828998341835e-06,
+      "loss": 1.4469,
+      "step": 843
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.727965984897309e-06,
+      "loss": 1.4914,
+      "step": 844
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.706137816037914e-06,
+      "loss": 1.4481,
+      "step": 845
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.684344636033104e-06,
+      "loss": 1.4505,
+      "step": 846
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.662586588921087e-06,
+      "loss": 1.4454,
+      "step": 847
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.64086381850786e-06,
+      "loss": 1.4984,
+      "step": 848
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.619176468366274e-06,
+      "loss": 1.4956,
+      "step": 849
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.597524681835062e-06,
+      "loss": 1.4527,
+      "step": 850
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5759086020179225e-06,
+      "loss": 1.4658,
+      "step": 851
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.554328371782537e-06,
+      "loss": 1.5114,
+      "step": 852
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.532784133759663e-06,
+      "loss": 1.5165,
+      "step": 853
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.51127603034217e-06,
+      "loss": 1.5028,
+      "step": 854
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.489804203684091e-06,
+      "loss": 1.4622,
+      "step": 855
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.468368795699705e-06,
+      "loss": 1.467,
+      "step": 856
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.446969948062588e-06,
+      "loss": 1.4885,
+      "step": 857
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.425607802204673e-06,
+      "loss": 1.4953,
+      "step": 858
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4042824993153195e-06,
+      "loss": 1.5061,
+      "step": 859
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.382994180340386e-06,
+      "loss": 1.5067,
+      "step": 860
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.361742985981279e-06,
+      "loss": 1.4638,
+      "step": 861
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3405290566940475e-06,
+      "loss": 1.4669,
+      "step": 862
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.319352532688444e-06,
+      "loss": 1.4824,
+      "step": 863
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.298213553926985e-06,
+      "loss": 1.4599,
+      "step": 864
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.277112260124057e-06,
+      "loss": 1.5134,
+      "step": 865
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.256048790744957e-06,
+      "loss": 1.4654,
+      "step": 866
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2350232850049956e-06,
+      "loss": 1.5097,
+      "step": 867
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.214035881868572e-06,
+      "loss": 1.4803,
+      "step": 868
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.193086720048249e-06,
+      "loss": 1.4775,
+      "step": 869
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.172175938003843e-06,
+      "loss": 1.4807,
+      "step": 870
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1513036739415055e-06,
+      "loss": 1.4631,
+      "step": 871
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1304700658128025e-06,
+      "loss": 1.5057,
+      "step": 872
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.109675251313821e-06,
+      "loss": 1.4515,
+      "step": 873
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.088919367884248e-06,
+      "loss": 1.5013,
+      "step": 874
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.0682025527064486e-06,
+      "loss": 1.4812,
+      "step": 875
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.047524942704589e-06,
+      "loss": 1.451,
+      "step": 876
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.026886674543713e-06,
+      "loss": 1.4876,
+      "step": 877
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.00628788462883e-06,
+      "loss": 1.4566,
+      "step": 878
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.985728709104041e-06,
+      "loss": 1.486,
+      "step": 879
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.965209283851607e-06,
+      "loss": 1.5009,
+      "step": 880
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.944729744491078e-06,
+      "loss": 1.4668,
+      "step": 881
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.92429022637838e-06,
+      "loss": 1.4911,
+      "step": 882
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.903890864604929e-06,
+      "loss": 1.4734,
+      "step": 883
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.88353179399673e-06,
+      "loss": 1.4751,
+      "step": 884
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.863213149113498e-06,
+      "loss": 1.4848,
+      "step": 885
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.842935064247751e-06,
+      "loss": 1.514,
+      "step": 886
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.822697673423938e-06,
+      "loss": 1.4657,
+      "step": 887
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.802501110397553e-06,
+      "loss": 1.4796,
+      "step": 888
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.782345508654235e-06,
+      "loss": 1.4974,
+      "step": 889
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7622310014089046e-06,
+      "loss": 1.4794,
+      "step": 890
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.742157721604878e-06,
+      "loss": 1.4633,
+      "step": 891
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7221258019129726e-06,
+      "loss": 1.4871,
+      "step": 892
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7021353747306556e-06,
+      "loss": 1.5102,
+      "step": 893
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6821865721811533e-06,
+      "loss": 1.4658,
+      "step": 894
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.662279526112581e-06,
+      "loss": 1.4831,
+      "step": 895
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.642414368097076e-06,
+      "loss": 1.4384,
+      "step": 896
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6225912294299117e-06,
+      "loss": 1.467,
+      "step": 897
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6028102411286535e-06,
+      "loss": 1.4989,
+      "step": 898
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5830715339322817e-06,
+      "loss": 1.4905,
+      "step": 899
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.5633752383003175e-06,
+      "loss": 1.4961,
+      "step": 900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.543721484411976e-06,
+      "loss": 1.4832,
+      "step": 901
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.524110402165307e-06,
+      "loss": 1.4535,
+      "step": 902
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.504542121176314e-06,
+      "loss": 1.4819,
+      "step": 903
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4850167707781257e-06,
+      "loss": 1.5111,
+      "step": 904
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.465534480020124e-06,
+      "loss": 1.4983,
+      "step": 905
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.446095377667097e-06,
+      "loss": 1.5001,
+      "step": 906
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.426699592198389e-06,
+      "loss": 1.4873,
+      "step": 907
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4073472518070394e-06,
+      "loss": 1.4836,
+      "step": 908
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.388038484398953e-06,
+      "loss": 1.4849,
+      "step": 909
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3687734175920505e-06,
+      "loss": 1.4598,
+      "step": 910
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3495521787154075e-06,
+      "loss": 1.4681,
+      "step": 911
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.330374894808438e-06,
+      "loss": 1.4681,
+      "step": 912
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.3112416926200453e-06,
+      "loss": 1.4906,
+      "step": 913
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.292152698607768e-06,
+      "loss": 1.4761,
+      "step": 914
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2731080389369707e-06,
+      "loss": 1.4528,
+      "step": 915
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2541078394799975e-06,
+      "loss": 1.5102,
+      "step": 916
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2351522258153346e-06,
+      "loss": 1.4994,
+      "step": 917
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.216241323226792e-06,
+      "loss": 1.4762,
+      "step": 918
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.197375256702668e-06,
+      "loss": 1.4718,
+      "step": 919
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1785541509349192e-06,
+      "loss": 1.4823,
+      "step": 920
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1597781303183517e-06,
+      "loss": 1.5027,
+      "step": 921
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.141047318949778e-06,
+      "loss": 1.4856,
+      "step": 922
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.122361840627218e-06,
+      "loss": 1.4914,
+      "step": 923
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1037218188490703e-06,
+      "loss": 1.4732,
+      "step": 924
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.085127376813285e-06,
+      "loss": 1.4807,
+      "step": 925
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.066578637416573e-06,
+      "loss": 1.4771,
+      "step": 926
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0480757232535773e-06,
+      "loss": 1.4518,
+      "step": 927
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0296187566160674e-06,
+      "loss": 1.4861,
+      "step": 928
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.011207859492131e-06,
+      "loss": 1.5088,
+      "step": 929
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.992843153565369e-06,
+      "loss": 1.4648,
+      "step": 930
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9745247602140826e-06,
+      "loss": 1.4458,
+      "step": 931
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.956252800510485e-06,
+      "loss": 1.5177,
+      "step": 932
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9380273952198957e-06,
+      "loss": 1.4825,
+      "step": 933
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9198486647999336e-06,
+      "loss": 1.4637,
+      "step": 934
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.901716729399735e-06,
+      "loss": 1.4598,
+      "step": 935
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.883631708859147e-06,
+      "loss": 1.4957,
+      "step": 936
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8655937227079468e-06,
+      "loss": 1.4678,
+      "step": 937
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8476028901650454e-06,
+      "loss": 1.4799,
+      "step": 938
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.829659330137693e-06,
+      "loss": 1.4756,
+      "step": 939
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8117631612207084e-06,
+      "loss": 1.5086,
+      "step": 940
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7939145016956847e-06,
+      "loss": 1.4428,
+      "step": 941
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7761134695302105e-06,
+      "loss": 1.4967,
+      "step": 942
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.758360182377089e-06,
+      "loss": 1.4854,
+      "step": 943
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.740654757573562e-06,
+      "loss": 1.4703,
+      "step": 944
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7229973121405295e-06,
+      "loss": 1.4722,
+      "step": 945
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7053879627817857e-06,
+      "loss": 1.4592,
+      "step": 946
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.687826825883242e-06,
+      "loss": 1.4834,
+      "step": 947
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.67031401751215e-06,
+      "loss": 1.4943,
+      "step": 948
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.652849653416354e-06,
+      "loss": 1.4528,
+      "step": 949
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6354338490235044e-06,
+      "loss": 1.478,
+      "step": 950
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.61806671944031e-06,
+      "loss": 1.4515,
+      "step": 951
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.600748379451773e-06,
+      "loss": 1.454,
+      "step": 952
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5834789435204245e-06,
+      "loss": 1.4803,
+      "step": 953
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.566258525785578e-06,
+      "loss": 1.4786,
+      "step": 954
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.549087240062569e-06,
+      "loss": 1.4797,
+      "step": 955
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.531965199841997e-06,
+      "loss": 1.4721,
+      "step": 956
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.514892518288988e-06,
+      "loss": 1.4868,
+      "step": 957
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4978693082424453e-06,
+      "loss": 1.512,
+      "step": 958
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.480895682214286e-06,
+      "loss": 1.4883,
+      "step": 959
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4639717523887196e-06,
+      "loss": 1.4874,
+      "step": 960
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4470976306214976e-06,
+      "loss": 1.4963,
+      "step": 961
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4302734284391684e-06,
+      "loss": 1.4564,
+      "step": 962
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4134992570383486e-06,
+      "loss": 1.4658,
+      "step": 963
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.396775227284985e-06,
+      "loss": 1.5031,
+      "step": 964
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3801014497136256e-06,
+      "loss": 1.4992,
+      "step": 965
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3634780345266805e-06,
+      "loss": 1.5132,
+      "step": 966
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3469050915936975e-06,
+      "loss": 1.4746,
+      "step": 967
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3303827304506397e-06,
+      "loss": 1.4821,
+      "step": 968
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3139110602991634e-06,
+      "loss": 1.4866,
+      "step": 969
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.297490190005878e-06,
+      "loss": 1.4974,
+      "step": 970
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.281120228101653e-06,
+      "loss": 1.4743,
+      "step": 971
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2648012827808874e-06,
+      "loss": 1.4587,
+      "step": 972
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.248533461900784e-06,
+      "loss": 1.4846,
+      "step": 973
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2323168729806575e-06,
+      "loss": 1.4768,
+      "step": 974
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2161516232012114e-06,
+      "loss": 1.4711,
+      "step": 975
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.200037819403833e-06,
+      "loss": 1.443,
+      "step": 976
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1839755680898855e-06,
+      "loss": 1.4517,
+      "step": 977
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1679649754200016e-06,
+      "loss": 1.4972,
+      "step": 978
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1520061472133903e-06,
+      "loss": 1.4599,
+      "step": 979
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.136099188947133e-06,
+      "loss": 1.4681,
+      "step": 980
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.120244205755478e-06,
+      "loss": 1.4867,
+      "step": 981
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.104441302429162e-06,
+      "loss": 1.4827,
+      "step": 982
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0886905834147076e-06,
+      "loss": 1.4903,
+      "step": 983
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0729921528137287e-06,
+      "loss": 1.4744,
+      "step": 984
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.057346114382253e-06,
+      "loss": 1.4598,
+      "step": 985
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.04175257153003e-06,
+      "loss": 1.4511,
+      "step": 986
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0262116273198508e-06,
+      "loss": 1.4945,
+      "step": 987
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0107233844668615e-06,
+      "loss": 1.4843,
+      "step": 988
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9952879453378937e-06,
+      "loss": 1.4685,
+      "step": 989
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.979905411950772e-06,
+      "loss": 1.5035,
+      "step": 990
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.964575885973661e-06,
+      "loss": 1.4911,
+      "step": 991
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9492994687243715e-06,
+      "loss": 1.4781,
+      "step": 992
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9340762611697094e-06,
+      "loss": 1.486,
+      "step": 993
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9189063639248006e-06,
+      "loss": 1.4826,
+      "step": 994
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9037898772524176e-06,
+      "loss": 1.4507,
+      "step": 995
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8887269010623343e-06,
+      "loss": 1.4604,
+      "step": 996
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.873717534910653e-06,
+      "loss": 1.5126,
+      "step": 997
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8587618779991467e-06,
+      "loss": 1.4802,
+      "step": 998
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8438600291746101e-06,
+      "loss": 1.502,
+      "step": 999
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8290120869282035e-06,
+      "loss": 1.4721,
+      "step": 1000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8142181493947997e-06,
+      "loss": 1.4653,
+      "step": 1001
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7994783143523365e-06,
+      "loss": 1.4814,
+      "step": 1002
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7847926792211722e-06,
+      "loss": 1.4807,
+      "step": 1003
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7701613410634367e-06,
+      "loss": 1.5036,
+      "step": 1004
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7555843965823992e-06,
+      "loss": 1.4944,
+      "step": 1005
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7410619421218156e-06,
+      "loss": 1.4984,
+      "step": 1006
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7265940736653074e-06,
+      "loss": 1.528,
+      "step": 1007
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.712180886835718e-06,
+      "loss": 1.4487,
+      "step": 1008
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6978224768944772e-06,
+      "loss": 1.47,
+      "step": 1009
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6835189387409801e-06,
+      "loss": 1.4834,
+      "step": 1010
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6692703669119569e-06,
+      "loss": 1.4584,
+      "step": 1011
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6550768555808461e-06,
+      "loss": 1.4727,
+      "step": 1012
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6409384985571752e-06,
+      "loss": 1.5025,
+      "step": 1013
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6268553892859406e-06,
+      "loss": 1.5049,
+      "step": 1014
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6128276208469794e-06,
+      "loss": 1.4821,
+      "step": 1015
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5988552859543715e-06,
+      "loss": 1.4763,
+      "step": 1016
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5849384769558196e-06,
+      "loss": 1.4761,
+      "step": 1017
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.57107728583203e-06,
+      "loss": 1.4933,
+      "step": 1018
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.557271804196121e-06,
+      "loss": 1.4478,
+      "step": 1019
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5435221232929997e-06,
+      "loss": 1.4945,
+      "step": 1020
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5298283339987763e-06,
+      "loss": 1.5044,
+      "step": 1021
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5161905268201516e-06,
+      "loss": 1.4953,
+      "step": 1022
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5026087918938214e-06,
+      "loss": 1.4592,
+      "step": 1023
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4890832189858851e-06,
+      "loss": 1.4548,
+      "step": 1024
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4756138974912482e-06,
+      "loss": 1.5126,
+      "step": 1025
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4622009164330264e-06,
+      "loss": 1.4886,
+      "step": 1026
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.448844364461971e-06,
+      "loss": 1.4704,
+      "step": 1027
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4355443298558735e-06,
+      "loss": 1.499,
+      "step": 1028
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4223009005189792e-06,
+      "loss": 1.4707,
+      "step": 1029
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4091141639814143e-06,
+      "loss": 1.4761,
+      "step": 1030
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3959842073986085e-06,
+      "loss": 1.4603,
+      "step": 1031
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3829111175507016e-06,
+      "loss": 1.4696,
+      "step": 1032
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3698949808419959e-06,
+      "loss": 1.4737,
+      "step": 1033
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3569358833003644e-06,
+      "loss": 1.481,
+      "step": 1034
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.344033910576692e-06,
+      "loss": 1.4565,
+      "step": 1035
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3311891479443096e-06,
+      "loss": 1.4353,
+      "step": 1036
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3184016802984213e-06,
+      "loss": 1.4765,
+      "step": 1037
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3056715921555552e-06,
+      "loss": 1.4603,
+      "step": 1038
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.292998967653002e-06,
+      "loss": 1.4922,
+      "step": 1039
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2803838905482468e-06,
+      "loss": 1.442,
+      "step": 1040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2678264442184362e-06,
+      "loss": 1.505,
+      "step": 1041
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2553267116598112e-06,
+      "loss": 1.4905,
+      "step": 1042
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.242884775487161e-06,
+      "loss": 1.513,
+      "step": 1043
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2305007179332851e-06,
+      "loss": 1.4441,
+      "step": 1044
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2181746208484425e-06,
+      "loss": 1.472,
+      "step": 1045
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2059065656998103e-06,
+      "loss": 1.4889,
+      "step": 1046
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1936966335709522e-06,
+      "loss": 1.4735,
+      "step": 1047
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.181544905161276e-06,
+      "loss": 1.4716,
+      "step": 1048
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1694514607854967e-06,
+      "loss": 1.497,
+      "step": 1049
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.157416380373122e-06,
+      "loss": 1.4833,
+      "step": 1050
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1454397434679022e-06,
+      "loss": 1.44,
+      "step": 1051
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.133521629227322e-06,
+      "loss": 1.4568,
+      "step": 1052
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.121662116422072e-06,
+      "loss": 1.5146,
+      "step": 1053
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1098612834355204e-06,
+      "loss": 1.5,
+      "step": 1054
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0981192082632076e-06,
+      "loss": 1.451,
+      "step": 1055
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0864359685123227e-06,
+      "loss": 1.4711,
+      "step": 1056
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.074811641401189e-06,
+      "loss": 1.4436,
+      "step": 1057
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0632463037587604e-06,
+      "loss": 1.5397,
+      "step": 1058
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0517400320241133e-06,
+      "loss": 1.4662,
+      "step": 1059
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0402929022459319e-06,
+      "loss": 1.4883,
+      "step": 1060
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0289049900820147e-06,
+      "loss": 1.4553,
+      "step": 1061
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0175763707987752e-06,
+      "loss": 1.4732,
+      "step": 1062
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.006307119270733e-06,
+      "loss": 1.4619,
+      "step": 1063
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.950973099800354e-07,
+      "loss": 1.4882,
+      "step": 1064
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.839470170159492e-07,
+      "loss": 1.4761,
+      "step": 1065
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.72856314074384e-07,
+      "loss": 1.4862,
+      "step": 1066
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.618252744573997e-07,
+      "loss": 1.4642,
+      "step": 1067
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.508539710727171e-07,
+      "loss": 1.4809,
+      "step": 1068
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.399424764332432e-07,
+      "loss": 1.5006,
+      "step": 1069
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.290908626565931e-07,
+      "loss": 1.4993,
+      "step": 1070
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.182992014646042e-07,
+      "loss": 1.4929,
+      "step": 1071
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.075675641828718e-07,
+      "loss": 1.4862,
+      "step": 1072
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.968960217402744e-07,
+      "loss": 1.4525,
+      "step": 1073
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.862846446684969e-07,
+      "loss": 1.4811,
+      "step": 1074
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.75733503101579e-07,
+      "loss": 1.4714,
+      "step": 1075
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.652426667754454e-07,
+      "loss": 1.4927,
+      "step": 1076
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.548122050274365e-07,
+      "loss": 1.4673,
+      "step": 1077
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.444421867958652e-07,
+      "loss": 1.4939,
+      "step": 1078
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.34132680619546e-07,
+      "loss": 1.4373,
+      "step": 1079
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.238837546373557e-07,
+      "loss": 1.4275,
+      "step": 1080
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.136954765877747e-07,
+      "loss": 1.4708,
+      "step": 1081
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.035679138084429e-07,
+      "loss": 1.4546,
+      "step": 1082
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.935011332357113e-07,
+      "loss": 1.4467,
+      "step": 1083
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.834952014042041e-07,
+      "loss": 1.4444,
+      "step": 1084
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.735501844463711e-07,
+      "loss": 1.4411,
+      "step": 1085
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.636661480920615e-07,
+      "loss": 1.4534,
+      "step": 1086
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.538431576680838e-07,
+      "loss": 1.4524,
+      "step": 1087
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.440812780977668e-07,
+      "loss": 1.4508,
+      "step": 1088
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.343805739005472e-07,
+      "loss": 1.4384,
+      "step": 1089
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.247411091915302e-07,
+      "loss": 1.4806,
+      "step": 1090
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.151629476810662e-07,
+      "loss": 1.4378,
+      "step": 1091
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.056461526743374e-07,
+      "loss": 1.4236,
+      "step": 1092
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.961907870709316e-07,
+      "loss": 1.496,
+      "step": 1093
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.867969133644326e-07,
+      "loss": 1.485,
+      "step": 1094
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.774645936420044e-07,
+      "loss": 1.4526,
+      "step": 1095
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.681938895839746e-07,
+      "loss": 1.4605,
+      "step": 1096
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.589848624634399e-07,
+      "loss": 1.481,
+      "step": 1097
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.498375731458529e-07,
+      "loss": 1.4648,
+      "step": 1098
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.40752082088616e-07,
+      "loss": 1.4926,
+      "step": 1099
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.317284493406939e-07,
+      "loss": 1.4581,
+      "step": 1100
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.227667345422061e-07,
+      "loss": 1.4898,
+      "step": 1101
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.13866996924034e-07,
+      "loss": 1.4699,
+      "step": 1102
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.050292953074355e-07,
+      "loss": 1.5129,
+      "step": 1103
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.962536881036507e-07,
+      "loss": 1.4732,
+      "step": 1104
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.875402333135183e-07,
+      "loss": 1.4736,
+      "step": 1105
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.78888988527092e-07,
+      "loss": 1.4427,
+      "step": 1106
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.70300010923256e-07,
+      "loss": 1.4744,
+      "step": 1107
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.617733572693551e-07,
+      "loss": 1.4824,
+      "step": 1108
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.533090839208133e-07,
+      "loss": 1.4383,
+      "step": 1109
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.449072468207584e-07,
+      "loss": 1.4233,
+      "step": 1110
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.365679014996639e-07,
+      "loss": 1.4417,
+      "step": 1111
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.282911030749727e-07,
+      "loss": 1.4793,
+      "step": 1112
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.200769062507305e-07,
+      "loss": 1.4878,
+      "step": 1113
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.119253653172329e-07,
+      "loss": 1.5076,
+      "step": 1114
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.038365341506635e-07,
+      "loss": 1.5378,
+      "step": 1115
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.958104662127316e-07,
+      "loss": 1.4577,
+      "step": 1116
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.878472145503288e-07,
+      "loss": 1.4676,
+      "step": 1117
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.799468317951739e-07,
+      "loss": 1.4805,
+      "step": 1118
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7210937016345936e-07,
+      "loss": 1.4311,
+      "step": 1119
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6433488145551595e-07,
+      "loss": 1.4862,
+      "step": 1120
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.5662341705546396e-07,
+      "loss": 1.4706,
+      "step": 1121
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.4897502793087576e-07,
+      "loss": 1.4878,
+      "step": 1122
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.4138976463244165e-07,
+      "loss": 1.4921,
+      "step": 1123
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.338676772936279e-07,
+      "loss": 1.4882,
+      "step": 1124
+    },
     {
       "epoch": 0.92,
-      "learning_rate": 6.05632895581425e-07,
-      "loss": 1.4227,
-      "step": 432
+      "learning_rate": 4.264088156303536e-07,
+      "loss": 1.4931,
+      "step": 1125
     },
     {
       "epoch": 0.92,
-      "learning_rate": 5.827900873478232e-07,
-      "loss": 1.4464,
-      "step": 433
+      "learning_rate": 4.19013228940659e-07,
+      "loss": 1.4843,
+      "step": 1126
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.116809661043775e-07,
+      "loss": 1.4885,
+      "step": 1127
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0441207558281517e-07,
+      "loss": 1.4979,
+      "step": 1128
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.97206605418432e-07,
+      "loss": 1.4859,
+      "step": 1129
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.90064603234519e-07,
+      "loss": 1.4578,
+      "step": 1130
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8298611623488913e-07,
+      "loss": 1.4758,
+      "step": 1131
     },
     {
       "epoch": 0.93,
-      "learning_rate": 5.603734527675697e-07,
-      "loss": 1.3927,
-      "step": 434
+      "learning_rate": 3.759711912035635e-07,
+      "loss": 1.5247,
+      "step": 1132
     },
     {
       "epoch": 0.93,
-      "learning_rate": 5.383840063002721e-07,
-      "loss": 1.4894,
-      "step": 435
+      "learning_rate": 3.690198745044571e-07,
+      "loss": 1.4811,
+      "step": 1133
     },
     {
       "epoch": 0.93,
-      "learning_rate": 5.168227430732353e-07,
-      "loss": 1.4976,
-      "step": 436
+      "learning_rate": 3.621322120810822e-07,
+      "loss": 1.473,
+      "step": 1134
     },
     {
       "epoch": 0.93,
-      "learning_rate": 4.956906388364335e-07,
-      "loss": 1.4203,
-      "step": 437
+      "learning_rate": 3.553082494562354e-07,
+      "loss": 1.4959,
+      "step": 1135
     },
     {
       "epoch": 0.93,
-      "learning_rate": 4.749886499183509e-07,
-      "loss": 1.4502,
-      "step": 438
+      "learning_rate": 3.4854803173170135e-07,
+      "loss": 1.5119,
+      "step": 1136
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.418516035879571e-07,
+      "loss": 1.4836,
+      "step": 1137
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.352190092838692e-07,
+      "loss": 1.4934,
+      "step": 1138
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.2865029265640725e-07,
+      "loss": 1.4714,
+      "step": 1139
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.2214549712035526e-07,
+      "loss": 1.4937,
+      "step": 1140
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.157046656680174e-07,
+      "loss": 1.4935,
+      "step": 1141
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0932784086894265e-07,
+      "loss": 1.506,
+      "step": 1142
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.030150648696362e-07,
+      "loss": 1.5073,
+      "step": 1143
     },
     {
       "epoch": 0.94,
-      "learning_rate": 4.5471771318269855e-07,
-      "loss": 1.4525,
-      "step": 439
+      "learning_rate": 2.96766379393284e-07,
+      "loss": 1.4763,
+      "step": 1144
     },
     {
       "epoch": 0.94,
-      "learning_rate": 4.3487874598602485e-07,
-      "loss": 1.436,
-      "step": 440
+      "learning_rate": 2.905818257394799e-07,
+      "loss": 1.4864,
+      "step": 1145
     },
     {
       "epoch": 0.94,
-      "learning_rate": 4.1547264613619243e-07,
-      "loss": 1.3797,
-      "step": 441
+      "learning_rate": 2.8446144478394667e-07,
+      "loss": 1.4535,
+      "step": 1146
     },
     {
       "epoch": 0.94,
-      "learning_rate": 3.965002918517569e-07,
-      "loss": 1.4255,
-      "step": 442
+      "learning_rate": 2.7840527697826967e-07,
+      "loss": 1.5243,
+      "step": 1147
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.7241336234962943e-07,
+      "loss": 1.4252,
+      "step": 1148
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.66485740500535e-07,
+      "loss": 1.4871,
+      "step": 1149
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.606224506085653e-07,
+      "loss": 1.49,
+      "step": 1150
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.548235314261072e-07,
+      "loss": 1.5302,
+      "step": 1151
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.4908902128010115e-07,
+      "loss": 1.4879,
+      "step": 1152
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.434189580717872e-07,
+      "loss": 1.4539,
+      "step": 1153
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3781337927645587e-07,
+      "loss": 1.4534,
+      "step": 1154
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.322723219431955e-07,
+      "loss": 1.5072,
+      "step": 1155
     },
     {
       "epoch": 0.95,
-      "learning_rate": 3.779625417222166e-07,
-      "loss": 1.4805,
-      "step": 443
+      "learning_rate": 2.2679582269465538e-07,
+      "loss": 1.439,
+      "step": 1156
     },
     {
       "epoch": 0.95,
-      "learning_rate": 3.5986023466916153e-07,
-      "loss": 1.3812,
-      "step": 444
+      "learning_rate": 2.2138391772679846e-07,
+      "loss": 1.5096,
+      "step": 1157
     },
     {
       "epoch": 0.95,
-      "learning_rate": 3.4219418990830146e-07,
-      "loss": 1.4686,
-      "step": 445
+      "learning_rate": 2.160366428086613e-07,
+      "loss": 1.4868,
+      "step": 1158
     },
     {
       "epoch": 0.95,
-      "learning_rate": 3.249652069124032e-07,
-      "loss": 1.3986,
-      "step": 446
+      "learning_rate": 2.1075403328212007e-07,
+      "loss": 1.4674,
+      "step": 1159
     },
     {
       "epoch": 0.95,
-      "learning_rate": 3.0817406537509977e-07,
-      "loss": 1.4392,
-      "step": 447
+      "learning_rate": 2.0553612406165933e-07,
+      "loss": 1.4574,
+      "step": 1160
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.0038294963413251e-07,
+      "loss": 1.471,
+      "step": 1161
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9529454405854519e-07,
+      "loss": 1.4696,
+      "step": 1162
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.90270940965821e-07,
+      "loss": 1.438,
+      "step": 1163
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.853121735585872e-07,
+      "loss": 1.4319,
+      "step": 1164
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8041827461094619e-07,
+      "loss": 1.4839,
+      "step": 1165
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.755892764682654e-07,
+      "loss": 1.4501,
+      "step": 1166
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7082521104696215e-07,
+      "loss": 1.5273,
+      "step": 1167
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6612610983429366e-07,
+      "loss": 1.5343,
+      "step": 1168
     },
     {
       "epoch": 0.96,
-      "learning_rate": 2.918215251756118e-07,
-      "loss": 1.4771,
-      "step": 448
+      "learning_rate": 1.6149200388814513e-07,
+      "loss": 1.4308,
+      "step": 1169
     },
     {
       "epoch": 0.96,
-      "learning_rate": 2.759083263443607e-07,
-      "loss": 1.4234,
-      "step": 449
+      "learning_rate": 1.5692292383682638e-07,
+      "loss": 1.4743,
+      "step": 1170
     },
     {
       "epoch": 0.96,
-      "learning_rate": 2.6043518902947096e-07,
-      "loss": 1.4375,
-      "step": 450
+      "learning_rate": 1.5241889987887338e-07,
+      "loss": 1.464,
+      "step": 1171
     },
     {
       "epoch": 0.96,
-      "learning_rate": 2.4540281346418946e-07,
-      "loss": 1.4974,
-      "step": 451
+      "learning_rate": 1.4797996178284256e-07,
+      "loss": 1.4413,
+      "step": 1172
     },
     {
       "epoch": 0.96,
-      "learning_rate": 2.3081187993518883e-07,
-      "loss": 1.4981,
-      "step": 452
+      "learning_rate": 1.4360613888711793e-07,
+      "loss": 1.4782,
+      "step": 1173
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3929746009971434e-07,
+      "loss": 1.4581,
+      "step": 1174
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3505395389809106e-07,
+      "loss": 1.4802,
+      "step": 1175
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.308756483289586e-07,
+      "loss": 1.4607,
+      "step": 1176
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2676257100809442e-07,
+      "loss": 1.488,
+      "step": 1177
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2271474912016413e-07,
+      "loss": 1.4427,
+      "step": 1178
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1873220941853503e-07,
+      "loss": 1.4987,
+      "step": 1179
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1481497822510734e-07,
+      "loss": 1.47,
+      "step": 1180
     },
     {
       "epoch": 0.97,
-      "learning_rate": 2.1666304875178846e-07,
-      "loss": 1.4874,
-      "step": 453
+      "learning_rate": 1.1096308143013323e-07,
+      "loss": 1.4791,
+      "step": 1181
     },
     {
       "epoch": 0.97,
-      "learning_rate": 2.0295696021606637e-07,
-      "loss": 1.5138,
-      "step": 454
+      "learning_rate": 1.071765444920514e-07,
+      "loss": 1.4965,
+      "step": 1182
     },
     {
       "epoch": 0.97,
-      "learning_rate": 1.8969423459388458e-07,
-      "loss": 1.4575,
-      "step": 455
+      "learning_rate": 1.0345539243731274e-07,
+      "loss": 1.4757,
+      "step": 1183
     },
     {
       "epoch": 0.97,
-      "learning_rate": 1.76875472086816e-07,
-      "loss": 1.4458,
-      "step": 456
+      "learning_rate": 9.979964986021939e-08,
+      "loss": 1.502,
+      "step": 1184
     },
     {
       "epoch": 0.97,
-      "learning_rate": 1.6450125280499273e-07,
-      "loss": 1.5189,
-      "step": 457
+      "learning_rate": 9.620934092276268e-08,
+      "loss": 1.4856,
+      "step": 1185
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.268448935445983e-08,
+      "loss": 1.5055,
+      "step": 1186
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.922511845219972e-08,
+      "loss": 1.4923,
+      "step": 1187
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.583125108008738e-08,
+      "loss": 1.4665,
+      "step": 1188
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.250290966929642e-08,
+      "loss": 1.5095,
+      "step": 1189
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.924011621791571e-08,
+      "loss": 1.4373,
+      "step": 1190
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.60428922908063e-08,
+      "loss": 1.4454,
+      "step": 1191
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.291125901946027e-08,
+      "loss": 1.4819,
+      "step": 1192
     },
     {
       "epoch": 0.98,
-      "learning_rate": 1.525721367408395e-07,
-      "loss": 1.4563,
-      "step": 458
+      "learning_rate": 6.984523710185876e-08,
+      "loss": 1.4787,
+      "step": 1193
     },
     {
       "epoch": 0.98,
-      "learning_rate": 1.410886637437414e-07,
-      "loss": 1.5629,
-      "step": 459
+      "learning_rate": 6.684484680233527e-08,
+      "loss": 1.4911,
+      "step": 1194
     },
     {
       "epoch": 0.98,
-      "learning_rate": 1.3005135349561048e-07,
-      "loss": 1.4554,
-      "step": 460
+      "learning_rate": 6.39101079514448e-08,
+      "loss": 1.4773,
+      "step": 1195
     },
     {
       "epoch": 0.98,
-      "learning_rate": 1.1946070548736532e-07,
-      "loss": 1.4735,
-      "step": 461
+      "learning_rate": 6.104103994583055e-08,
+      "loss": 1.5064,
+      "step": 1196
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.8237661748093976e-08,
+      "loss": 1.4913,
+      "step": 1197
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.5499991886670543e-08,
+      "loss": 1.4605,
+      "step": 1198
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.282804845571199e-08,
+      "loss": 1.4541,
+      "step": 1199
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.022184911495864e-08,
+      "loss": 1.4862,
+      "step": 1200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.768141108963065e-08,
+      "loss": 1.4534,
+      "step": 1201
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.520675117030582e-08,
+      "loss": 1.5035,
+      "step": 1202
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.279788571281529e-08,
+      "loss": 1.4906,
+      "step": 1203
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.045483063813471e-08,
+      "loss": 1.4779,
+      "step": 1204
     },
     {
       "epoch": 0.99,
-      "learning_rate": 1.0931719899632597e-07,
-      "loss": 1.4703,
-      "step": 462
+      "learning_rate": 3.817760143227434e-08,
+      "loss": 1.4881,
+      "step": 1205
     },
     {
       "epoch": 0.99,
-      "learning_rate": 9.962129306453128e-08,
-      "loss": 1.4669,
-      "step": 463
+      "learning_rate": 3.596621314618354e-08,
+      "loss": 1.4464,
+      "step": 1206
     },
     {
       "epoch": 0.99,
-      "learning_rate": 9.037342647795877e-08,
-      "loss": 1.4213,
-      "step": 464
+      "learning_rate": 3.382068039564423e-08,
+      "loss": 1.5057,
+      "step": 1207
     },
     {
       "epoch": 0.99,
-      "learning_rate": 8.15740177466673e-08,
-      "loss": 1.4501,
-      "step": 465
+      "learning_rate": 3.1741017361179846e-08,
+      "loss": 1.4333,
+      "step": 1208
     },
     {
       "epoch": 0.99,
-      "learning_rate": 7.322346508586209e-08,
-      "loss": 1.475,
-      "step": 466
+      "learning_rate": 2.9727237787958717e-08,
+      "loss": 1.4769,
+      "step": 1209
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7779354985704166e-08,
+      "loss": 1.4589,
+      "step": 1210
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5897381828606794e-08,
+      "loss": 1.4706,
+      "step": 1211
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4081330755237887e-08,
+      "loss": 1.4797,
+      "step": 1212
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2331213768468363e-08,
+      "loss": 1.4751,
+      "step": 1213
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.064704243539106e-08,
+      "loss": 1.4778,
+      "step": 1214
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9028827887239698e-08,
+      "loss": 1.5184,
+      "step": 1215
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7476580819321133e-08,
+      "loss": 1.4679,
+      "step": 1216
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.599031149093988e-08,
+      "loss": 1.4753,
+      "step": 1217
     },
     {
       "epoch": 1.0,
-      "learning_rate": 6.532214639786926e-08,
-      "loss": 1.4629,
-      "step": 467
+      "learning_rate": 1.4570029725333723e-08,
+      "loss": 1.4828,
+      "step": 1218
     },
     {
       "epoch": 1.0,
-      "learning_rate": 5.7870419255035e-08,
-      "loss": 1.436,
-      "step": 468
+      "learning_rate": 1.3215744909603756e-08,
+      "loss": 1.4947,
+      "step": 1219
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1927465994659993e-08,
+      "loss": 1.4694,
+      "step": 1220
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0705201495155859e-08,
+      "loss": 1.4471,
+      "step": 1221
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.548959489434905e-09,
+      "loss": 1.4873,
+      "step": 1222
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.458747619475293e-09,
+      "loss": 1.4522,
+      "step": 1223
     },
     {
       "epoch": 1.0,
-      "step": 468,
-      "total_flos": 1.7585883807377e+17,
-      "train_loss": 1.5493416032220564,
-      "train_runtime": 6511.0109,
-      "train_samples_per_second": 4.608,
-      "train_steps_per_second": 0.072
+      "step": 1223,
+      "total_flos": 7.383809426907464e+17,
+      "train_loss": 1.5605670542182828,
+      "train_runtime": 29774.8409,
+      "train_samples_per_second": 5.258,
+      "train_steps_per_second": 0.041
     }
   ],
-  "max_steps": 468,
+  "max_steps": 1223,
   "num_train_epochs": 1,
-  "total_flos": 1.7585883807377e+17,
+  "total_flos": 7.383809426907464e+17,
   "trial_name": null,
   "trial_params": null
 }