diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,4462 +1,1012 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 5.2106135986733,
-  "global_step": 741,
+  "epoch": 4.397350993377484,
+  "global_step": 166,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.01,
-      "learning_rate": 4.347826086956522e-07,
-      "loss": 2.7034,
+      "epoch": 0.03,
+      "learning_rate": 8.333333333333333e-07,
+      "loss": 1.3883,
       "step": 1
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 8.695652173913044e-07,
-      "loss": 2.73,
+      "epoch": 0.05,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 1.4311,
       "step": 2
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.3043478260869566e-06,
-      "loss": 2.7564,
+      "epoch": 0.08,
+      "learning_rate": 2.5e-06,
+      "loss": 1.2563,
       "step": 3
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.7391304347826088e-06,
-      "loss": 2.6514,
+      "epoch": 0.11,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 1.2668,
       "step": 4
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 2.173913043478261e-06,
-      "loss": 2.5382,
+      "epoch": 0.13,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 1.2419,
       "step": 5
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 2.6086956521739132e-06,
-      "loss": 2.5815,
+      "epoch": 0.16,
+      "learning_rate": 5e-06,
+      "loss": 1.3009,
       "step": 6
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 3.043478260869566e-06,
-      "loss": 2.6067,
+      "epoch": 0.19,
+      "learning_rate": 5.833333333333334e-06,
+      "loss": 1.2596,
       "step": 7
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 3.4782608695652175e-06,
-      "loss": 2.6704,
+      "epoch": 0.21,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.362,
       "step": 8
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 3.91304347826087e-06,
-      "loss": 2.6642,
+      "epoch": 0.24,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 1.3293,
       "step": 9
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.347826086956522e-06,
-      "loss": 2.5732,
+      "epoch": 0.26,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 1.4663,
       "step": 10
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 4.782608695652174e-06,
-      "loss": 2.6243,
+      "epoch": 0.29,
+      "learning_rate": 9.166666666666666e-06,
+      "loss": 1.4526,
       "step": 11
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 5.2173913043478265e-06,
-      "loss": 2.5127,
+      "epoch": 0.32,
+      "learning_rate": 1e-05,
+      "loss": 1.4587,
       "step": 12
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 5.652173913043479e-06,
-      "loss": 2.6569,
+      "epoch": 0.34,
+      "learning_rate": 9.999807482189519e-06,
+      "loss": 1.2978,
       "step": 13
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 6.086956521739132e-06,
-      "loss": 2.5531,
+      "epoch": 0.37,
+      "learning_rate": 9.999229943583318e-06,
+      "loss": 1.2931,
       "step": 14
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 6.521739130434783e-06,
-      "loss": 2.6155,
+      "epoch": 0.4,
+      "learning_rate": 9.998267428655983e-06,
+      "loss": 1.4222,
       "step": 15
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 6.956521739130435e-06,
-      "loss": 2.6498,
+      "epoch": 0.42,
+      "learning_rate": 9.996920011528022e-06,
+      "loss": 1.3721,
       "step": 16
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 7.391304347826087e-06,
-      "loss": 2.5353,
+      "epoch": 0.45,
+      "learning_rate": 9.995187795960152e-06,
+      "loss": 1.2833,
       "step": 17
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 7.82608695652174e-06,
-      "loss": 2.605,
+      "epoch": 0.48,
+      "learning_rate": 9.993070915345313e-06,
+      "loss": 1.2782,
       "step": 18
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 8.260869565217392e-06,
-      "loss": 2.6735,
+      "epoch": 0.5,
+      "learning_rate": 9.990569532698393e-06,
+      "loss": 1.3794,
       "step": 19
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 8.695652173913044e-06,
-      "loss": 2.5645,
+      "epoch": 0.53,
+      "learning_rate": 9.987683840643679e-06,
+      "loss": 1.3194,
       "step": 20
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 9.130434782608697e-06,
-      "loss": 2.5904,
+      "epoch": 0.56,
+      "learning_rate": 9.984414061400011e-06,
+      "loss": 1.3508,
       "step": 21
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 9.565217391304349e-06,
-      "loss": 2.6846,
+      "epoch": 0.58,
+      "learning_rate": 9.980760446763693e-06,
+      "loss": 1.4149,
       "step": 22
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 1e-05,
-      "loss": 2.5589,
+      "epoch": 0.61,
+      "learning_rate": 9.976723278089075e-06,
+      "loss": 1.2859,
       "step": 23
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 9.999953315763929e-06,
-      "loss": 2.5329,
+      "epoch": 0.64,
+      "learning_rate": 9.97230286626691e-06,
+      "loss": 1.2876,
       "step": 24
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 9.999813263927483e-06,
-      "loss": 2.5547,
+      "epoch": 0.66,
+      "learning_rate": 9.967499551700398e-06,
+      "loss": 1.4954,
       "step": 25
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 9.999579847105947e-06,
-      "loss": 2.5863,
+      "epoch": 0.69,
+      "learning_rate": 9.962313704278981e-06,
+      "loss": 1.4377,
       "step": 26
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 9.999253069658074e-06,
-      "loss": 2.7017,
+      "epoch": 0.72,
+      "learning_rate": 9.956745723349857e-06,
+      "loss": 1.4215,
       "step": 27
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 9.99883293768601e-06,
-      "loss": 2.6934,
+      "epoch": 0.74,
+      "learning_rate": 9.950796037687224e-06,
+      "loss": 1.3041,
       "step": 28
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 9.998319459035168e-06,
-      "loss": 2.5498,
+      "epoch": 0.77,
+      "learning_rate": 9.944465105459265e-06,
+      "loss": 1.2894,
       "step": 29
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 9.997712643294093e-06,
-      "loss": 2.5468,
+      "epoch": 0.79,
+      "learning_rate": 9.937753414192862e-06,
+      "loss": 1.3224,
       "step": 30
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 9.997012501794273e-06,
-      "loss": 2.6303,
+      "epoch": 0.82,
+      "learning_rate": 9.93066148073606e-06,
+      "loss": 1.3959,
       "step": 31
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 9.996219047609943e-06,
-      "loss": 2.4883,
+      "epoch": 0.85,
+      "learning_rate": 9.923189851218259e-06,
+      "loss": 1.2608,
       "step": 32
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 9.995332295557818e-06,
-      "loss": 2.5467,
+      "epoch": 0.87,
+      "learning_rate": 9.915339101008162e-06,
+      "loss": 1.2771,
       "step": 33
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 9.994352262196839e-06,
-      "loss": 2.6186,
+      "epoch": 0.9,
+      "learning_rate": 9.907109834669465e-06,
+      "loss": 1.3177,
       "step": 34
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 9.993278965827844e-06,
-      "loss": 2.5559,
+      "epoch": 0.93,
+      "learning_rate": 9.898502685914301e-06,
+      "loss": 1.3932,
       "step": 35
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 9.992112426493247e-06,
-      "loss": 2.4594,
+      "epoch": 0.95,
+      "learning_rate": 9.889518317554446e-06,
+      "loss": 1.4338,
       "step": 36
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 9.990852665976648e-06,
-      "loss": 2.5372,
+      "epoch": 0.98,
+      "learning_rate": 9.880157421450269e-06,
+      "loss": 1.3006,
       "step": 37
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 9.989499707802424e-06,
-      "loss": 2.5247,
+      "epoch": 1.01,
+      "learning_rate": 9.870420718457458e-06,
+      "loss": 1.4079,
       "step": 38
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 9.988053577235306e-06,
-      "loss": 2.4697,
+      "epoch": 1.03,
+      "learning_rate": 9.860308958371511e-06,
+      "loss": 1.3192,
       "step": 39
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 9.986514301279894e-06,
-      "loss": 2.4605,
+      "epoch": 1.06,
+      "learning_rate": 9.849822919869993e-06,
+      "loss": 1.331,
       "step": 40
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 9.984881908680157e-06,
-      "loss": 2.5399,
+      "epoch": 1.09,
+      "learning_rate": 9.838963410452572e-06,
+      "loss": 1.2477,
       "step": 41
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 9.983156429918895e-06,
-      "loss": 2.4888,
+      "epoch": 1.11,
+      "learning_rate": 9.827731266378839e-06,
+      "loss": 1.3087,
       "step": 42
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 9.981337897217171e-06,
-      "loss": 2.4991,
+      "epoch": 1.14,
+      "learning_rate": 9.816127352603907e-06,
+      "loss": 1.2522,
       "step": 43
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 9.979426344533712e-06,
-      "loss": 2.4208,
+      "epoch": 1.17,
+      "learning_rate": 9.804152562711804e-06,
+      "loss": 1.2771,
       "step": 44
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 9.977421807564264e-06,
-      "loss": 2.5066,
+      "epoch": 1.19,
+      "learning_rate": 9.791807818846665e-06,
+      "loss": 1.4044,
       "step": 45
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 9.97532432374094e-06,
-      "loss": 2.4321,
+      "epoch": 1.22,
+      "learning_rate": 9.779094071641712e-06,
+      "loss": 1.2256,
       "step": 46
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 9.973133932231514e-06,
-      "loss": 2.5499,
+      "epoch": 1.25,
+      "learning_rate": 9.766012300146058e-06,
+      "loss": 1.3511,
       "step": 47
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 9.970850673938684e-06,
-      "loss": 2.5242,
+      "epoch": 1.27,
+      "learning_rate": 9.752563511749301e-06,
+      "loss": 1.316,
       "step": 48
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 9.96847459149932e-06,
-      "loss": 2.6053,
+      "epoch": 1.3,
+      "learning_rate": 9.738748742103961e-06,
+      "loss": 1.2006,
       "step": 49
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 9.966005729283658e-06,
-      "loss": 2.5961,
+      "epoch": 1.32,
+      "learning_rate": 9.724569055045722e-06,
+      "loss": 1.2237,
       "step": 50
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 9.963444133394478e-06,
-      "loss": 2.5773,
+      "epoch": 1.35,
+      "learning_rate": 9.710025542511503e-06,
+      "loss": 1.2012,
       "step": 51
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 9.960789851666237e-06,
-      "loss": 2.521,
+      "epoch": 1.38,
+      "learning_rate": 9.695119324455383e-06,
+      "loss": 1.3153,
       "step": 52
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 9.958042933664186e-06,
-      "loss": 2.5786,
+      "epoch": 1.4,
+      "learning_rate": 9.679851548762344e-06,
+      "loss": 1.2494,
       "step": 53
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 9.955203430683425e-06,
-      "loss": 2.4393,
+      "epoch": 1.43,
+      "learning_rate": 9.664223391159885e-06,
+      "loss": 1.3274,
       "step": 54
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 9.952271395747969e-06,
-      "loss": 2.4537,
+      "epoch": 1.46,
+      "learning_rate": 9.648236055127485e-06,
+      "loss": 1.4895,
       "step": 55
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 9.949246883609743e-06,
-      "loss": 2.5576,
+      "epoch": 1.48,
+      "learning_rate": 9.631890771803909e-06,
+      "loss": 1.2369,
       "step": 56
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 9.94612995074756e-06,
-      "loss": 2.4571,
+      "epoch": 1.51,
+      "learning_rate": 9.61518879989242e-06,
+      "loss": 1.3197,
       "step": 57
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 9.942920655366075e-06,
-      "loss": 2.4928,
+      "epoch": 1.54,
+      "learning_rate": 9.598131425563847e-06,
+      "loss": 1.3453,
       "step": 58
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 9.939619057394687e-06,
-      "loss": 2.573,
+      "epoch": 1.56,
+      "learning_rate": 9.580719962357532e-06,
+      "loss": 1.3373,
       "step": 59
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 9.936225218486428e-06,
-      "loss": 2.5714,
+      "epoch": 1.59,
+      "learning_rate": 9.562955751080183e-06,
+      "loss": 1.2564,
       "step": 60
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 9.93273920201681e-06,
-      "loss": 2.4764,
+      "epoch": 1.62,
+      "learning_rate": 9.544840159702626e-06,
+      "loss": 1.3595,
       "step": 61
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 9.929161073082636e-06,
-      "loss": 2.5248,
+      "epoch": 1.64,
+      "learning_rate": 9.526374583254454e-06,
+      "loss": 1.3072,
       "step": 62
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 9.925490898500796e-06,
-      "loss": 2.5659,
+      "epoch": 1.67,
+      "learning_rate": 9.507560443716608e-06,
+      "loss": 1.2864,
       "step": 63
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 9.921728746807008e-06,
-      "loss": 2.4022,
+      "epoch": 1.7,
+      "learning_rate": 9.488399189911866e-06,
+      "loss": 1.2683,
       "step": 64
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 9.917874688254542e-06,
-      "loss": 2.493,
+      "epoch": 1.72,
+      "learning_rate": 9.468892297393281e-06,
+      "loss": 1.3008,
       "step": 65
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 9.913928794812909e-06,
-      "loss": 2.4454,
+      "epoch": 1.75,
+      "learning_rate": 9.449041268330549e-06,
+      "loss": 1.2022,
       "step": 66
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 9.90989114016652e-06,
-      "loss": 2.5217,
+      "epoch": 1.77,
+      "learning_rate": 9.428847631394324e-06,
+      "loss": 1.2575,
       "step": 67
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 9.905761799713302e-06,
-      "loss": 2.66,
+      "epoch": 1.8,
+      "learning_rate": 9.408312941638522e-06,
+      "loss": 1.2705,
       "step": 68
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 9.901540850563295e-06,
-      "loss": 2.4788,
+      "epoch": 1.83,
+      "learning_rate": 9.387438780380541e-06,
+      "loss": 1.2692,
       "step": 69
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 9.89722837153722e-06,
-      "loss": 2.4752,
+      "epoch": 1.85,
+      "learning_rate": 9.366226755079513e-06,
+      "loss": 1.2362,
       "step": 70
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 9.892824443164987e-06,
-      "loss": 2.5233,
+      "epoch": 1.88,
+      "learning_rate": 9.344678499212503e-06,
+      "loss": 1.2403,
       "step": 71
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 9.88832914768421e-06,
-      "loss": 2.4721,
+      "epoch": 1.91,
+      "learning_rate": 9.322795672148726e-06,
+      "loss": 1.3415,
       "step": 72
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 9.883742569038663e-06,
-      "loss": 2.5893,
+      "epoch": 1.93,
+      "learning_rate": 9.300579959021766e-06,
+      "loss": 1.4069,
       "step": 73
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 9.879064792876717e-06,
-      "loss": 2.4385,
+      "epoch": 1.96,
+      "learning_rate": 9.2780330705998e-06,
+      "loss": 1.2953,
       "step": 74
     },
     {
-      "epoch": 1.0,
-      "learning_rate": 9.874295906549728e-06,
-      "loss": 2.3598,
+      "epoch": 1.99,
+      "learning_rate": 9.255156743153867e-06,
+      "loss": 1.3696,
       "step": 75
     },
     {
-      "epoch": 1.01,
-      "learning_rate": 9.869435999110428e-06,
-      "loss": 2.3664,
+      "epoch": 2.01,
+      "learning_rate": 9.231952738324155e-06,
+      "loss": 1.3844,
       "step": 76
     },
     {
-      "epoch": 1.02,
-      "learning_rate": 9.864485161311242e-06,
-      "loss": 2.5441,
+      "epoch": 2.04,
+      "learning_rate": 9.208422842984345e-06,
+      "loss": 1.277,
       "step": 77
     },
     {
-      "epoch": 1.03,
-      "learning_rate": 9.859443485602603e-06,
-      "loss": 2.536,
+      "epoch": 2.07,
+      "learning_rate": 9.18456886910401e-06,
+      "loss": 1.2956,
       "step": 78
     },
     {
-      "epoch": 1.05,
-      "learning_rate": 9.85431106613122e-06,
-      "loss": 2.5167,
+      "epoch": 2.09,
+      "learning_rate": 9.16039265360908e-06,
+      "loss": 1.4063,
       "step": 79
     },
     {
-      "epoch": 1.06,
-      "learning_rate": 9.849087998738328e-06,
-      "loss": 2.4948,
+      "epoch": 2.12,
+      "learning_rate": 9.135896058240384e-06,
+      "loss": 1.3085,
       "step": 80
     },
     {
-      "epoch": 1.07,
-      "learning_rate": 9.84377438095789e-06,
-      "loss": 2.377,
+      "epoch": 2.15,
+      "learning_rate": 9.111080969410282e-06,
+      "loss": 1.3986,
       "step": 81
     },
     {
-      "epoch": 1.09,
-      "learning_rate": 9.838370312014783e-06,
-      "loss": 2.5684,
+      "epoch": 2.17,
+      "learning_rate": 9.085949298057402e-06,
+      "loss": 1.4697,
       "step": 82
     },
     {
-      "epoch": 1.1,
-      "learning_rate": 9.832875892822937e-06,
-      "loss": 2.4302,
+      "epoch": 2.2,
+      "learning_rate": 9.060502979499484e-06,
+      "loss": 1.239,
       "step": 83
     },
     {
-      "epoch": 1.11,
-      "learning_rate": 9.827291225983458e-06,
-      "loss": 2.4731,
+      "epoch": 2.23,
+      "learning_rate": 9.034743973284337e-06,
+      "loss": 1.2664,
       "step": 84
     },
     {
-      "epoch": 1.13,
-      "learning_rate": 9.821616415782708e-06,
-      "loss": 2.4375,
+      "epoch": 2.25,
+      "learning_rate": 9.008674263038954e-06,
+      "loss": 1.3195,
       "step": 85
     },
     {
-      "epoch": 1.14,
-      "learning_rate": 9.815851568190358e-06,
-      "loss": 2.3958,
+      "epoch": 2.28,
+      "learning_rate": 8.98229585631675e-06,
+      "loss": 1.3899,
       "step": 86
     },
     {
-      "epoch": 1.15,
-      "learning_rate": 9.80999679085741e-06,
-      "loss": 2.3743,
+      "epoch": 2.3,
+      "learning_rate": 8.955610784442967e-06,
+      "loss": 1.2493,
       "step": 87
     },
     {
-      "epoch": 1.17,
-      "learning_rate": 9.80405219311419e-06,
-      "loss": 2.4446,
+      "epoch": 2.33,
+      "learning_rate": 8.928621102358248e-06,
+      "loss": 1.2793,
       "step": 88
     },
     {
-      "epoch": 1.18,
-      "learning_rate": 9.798017885968295e-06,
-      "loss": 2.4291,
+      "epoch": 2.36,
+      "learning_rate": 8.901328888460394e-06,
+      "loss": 1.2779,
       "step": 89
     },
     {
-      "epoch": 1.19,
-      "learning_rate": 9.791893982102537e-06,
-      "loss": 2.5374,
+      "epoch": 2.38,
+      "learning_rate": 8.873736244444311e-06,
+      "loss": 1.4215,
       "step": 90
     },
     {
-      "epoch": 1.21,
-      "learning_rate": 9.785680595872824e-06,
-      "loss": 2.4543,
+      "epoch": 2.41,
+      "learning_rate": 8.845845295140163e-06,
+      "loss": 1.2914,
       "step": 91
     },
     {
-      "epoch": 1.22,
-      "learning_rate": 9.77937784330603e-06,
-      "loss": 2.3586,
+      "epoch": 2.44,
+      "learning_rate": 8.817658188349745e-06,
+      "loss": 1.2663,
       "step": 92
     },
     {
-      "epoch": 1.23,
-      "learning_rate": 9.772985842097832e-06,
-      "loss": 2.5718,
+      "epoch": 2.46,
+      "learning_rate": 8.789177094681091e-06,
+      "loss": 1.3403,
       "step": 93
     },
     {
-      "epoch": 1.25,
-      "learning_rate": 9.766504711610507e-06,
-      "loss": 2.3728,
+      "epoch": 2.49,
+      "learning_rate": 8.76040420738132e-06,
+      "loss": 1.301,
       "step": 94
     },
     {
-      "epoch": 1.26,
-      "learning_rate": 9.759934572870706e-06,
-      "loss": 2.5059,
+      "epoch": 2.52,
+      "learning_rate": 8.731341742167739e-06,
+      "loss": 1.2804,
       "step": 95
     },
     {
-      "epoch": 1.27,
-      "learning_rate": 9.753275548567192e-06,
-      "loss": 2.4287,
+      "epoch": 2.54,
+      "learning_rate": 8.701991937057211e-06,
+      "loss": 1.4168,
       "step": 96
     },
     {
-      "epoch": 1.29,
-      "learning_rate": 9.74652776304855e-06,
-      "loss": 2.3691,
+      "epoch": 2.57,
+      "learning_rate": 8.672357052193826e-06,
+      "loss": 1.4377,
       "step": 97
     },
     {
-      "epoch": 1.3,
-      "learning_rate": 9.739691342320866e-06,
-      "loss": 2.3754,
+      "epoch": 2.6,
+      "learning_rate": 8.642439369674845e-06,
+      "loss": 1.272,
       "step": 98
     },
     {
-      "epoch": 1.31,
-      "learning_rate": 9.732766414045368e-06,
-      "loss": 2.5768,
+      "epoch": 2.62,
+      "learning_rate": 8.612241193374958e-06,
+      "loss": 1.2942,
       "step": 99
     },
     {
-      "epoch": 1.33,
-      "learning_rate": 9.725753107536053e-06,
-      "loss": 2.5617,
+      "epoch": 2.65,
+      "learning_rate": 8.581764848768878e-06,
+      "loss": 1.3492,
       "step": 100
     },
     {
-      "epoch": 1.34,
-      "learning_rate": 9.718651553757266e-06,
-      "loss": 2.4548,
+      "epoch": 2.68,
+      "learning_rate": 8.551012682752262e-06,
+      "loss": 1.198,
       "step": 101
     },
     {
-      "epoch": 1.35,
-      "learning_rate": 9.711461885321247e-06,
-      "loss": 2.3576,
+      "epoch": 2.7,
+      "learning_rate": 8.519987063460973e-06,
+      "loss": 1.2902,
       "step": 102
     },
     {
-      "epoch": 1.37,
-      "learning_rate": 9.704184236485672e-06,
-      "loss": 2.4765,
+      "epoch": 2.73,
+      "learning_rate": 8.488690380088733e-06,
+      "loss": 1.2746,
       "step": 103
     },
     {
-      "epoch": 1.38,
-      "learning_rate": 9.696818743151128e-06,
-      "loss": 2.5565,
+      "epoch": 2.75,
+      "learning_rate": 8.457125042703124e-06,
+      "loss": 1.2148,
       "step": 104
     },
     {
-      "epoch": 1.39,
-      "learning_rate": 9.68936554285859e-06,
-      "loss": 2.4757,
+      "epoch": 2.78,
+      "learning_rate": 8.425293482060002e-06,
+      "loss": 1.2625,
       "step": 105
     },
     {
-      "epoch": 1.41,
-      "learning_rate": 9.68182477478684e-06,
-      "loss": 2.5012,
+      "epoch": 2.81,
+      "learning_rate": 8.393198149416311e-06,
+      "loss": 1.4119,
       "step": 106
     },
     {
-      "epoch": 1.42,
-      "learning_rate": 9.67419657974988e-06,
-      "loss": 2.4415,
+      "epoch": 2.83,
+      "learning_rate": 8.360841516341319e-06,
+      "loss": 1.1984,
       "step": 107
     },
     {
-      "epoch": 1.43,
-      "learning_rate": 9.66648110019429e-06,
-      "loss": 2.5461,
+      "epoch": 2.86,
+      "learning_rate": 8.328226074526284e-06,
+      "loss": 1.1971,
       "step": 108
     },
     {
-      "epoch": 1.45,
-      "learning_rate": 9.658678480196579e-06,
-      "loss": 2.4411,
+      "epoch": 2.89,
+      "learning_rate": 8.295354335592588e-06,
+      "loss": 1.1877,
       "step": 109
     },
     {
-      "epoch": 1.46,
-      "learning_rate": 9.650788865460487e-06,
-      "loss": 2.4137,
+      "epoch": 2.91,
+      "learning_rate": 8.262228830898313e-06,
+      "loss": 1.2444,
       "step": 110
     },
     {
-      "epoch": 1.47,
-      "learning_rate": 9.642812403314272e-06,
-      "loss": 2.4774,
+      "epoch": 2.94,
+      "learning_rate": 8.22885211134331e-06,
+      "loss": 1.288,
       "step": 111
     },
     {
-      "epoch": 1.49,
-      "learning_rate": 9.634749242707948e-06,
-      "loss": 2.3834,
+      "epoch": 2.97,
+      "learning_rate": 8.19522674717277e-06,
+      "loss": 1.1986,
       "step": 112
     },
     {
-      "epoch": 1.5,
-      "learning_rate": 9.626599534210514e-06,
-      "loss": 2.3706,
+      "epoch": 2.99,
+      "learning_rate": 8.161355327779289e-06,
+      "loss": 1.3661,
       "step": 113
     },
     {
-      "epoch": 1.51,
-      "learning_rate": 9.618363430007134e-06,
-      "loss": 2.4467,
+      "epoch": 3.02,
+      "learning_rate": 8.127240461503462e-06,
+      "loss": 1.3307,
       "step": 114
     },
     {
-      "epoch": 1.53,
-      "learning_rate": 9.610041083896304e-06,
-      "loss": 2.3918,
+      "epoch": 3.05,
+      "learning_rate": 8.092884775433037e-06,
+      "loss": 1.2406,
       "step": 115
     },
     {
-      "epoch": 1.54,
-      "learning_rate": 9.60163265128697e-06,
-      "loss": 2.4193,
+      "epoch": 3.07,
+      "learning_rate": 8.058290915200597e-06,
+      "loss": 1.3712,
       "step": 116
     },
     {
-      "epoch": 1.55,
-      "learning_rate": 9.593138289195634e-06,
-      "loss": 2.4662,
+      "epoch": 3.1,
+      "learning_rate": 8.023461544779833e-06,
+      "loss": 1.3295,
       "step": 117
     },
     {
-      "epoch": 1.57,
-      "learning_rate": 9.584558156243418e-06,
-      "loss": 2.4145,
+      "epoch": 3.13,
+      "learning_rate": 7.988399346280398e-06,
+      "loss": 1.4179,
       "step": 118
     },
     {
-      "epoch": 1.58,
-      "learning_rate": 9.575892412653102e-06,
-      "loss": 2.4683,
+      "epoch": 3.15,
+      "learning_rate": 7.953107019741366e-06,
+      "loss": 1.3459,
       "step": 119
     },
     {
-      "epoch": 1.59,
-      "learning_rate": 9.567141220246136e-06,
-      "loss": 2.4461,
+      "epoch": 3.18,
+      "learning_rate": 7.917587282923312e-06,
+      "loss": 1.3335,
       "step": 120
     },
     {
-      "epoch": 1.61,
-      "learning_rate": 9.55830474243961e-06,
-      "loss": 2.4211,
+      "epoch": 3.21,
+      "learning_rate": 7.88184287109902e-06,
+      "loss": 1.2718,
       "step": 121
     },
     {
-      "epoch": 1.62,
-      "learning_rate": 9.549383144243213e-06,
-      "loss": 2.6434,
+      "epoch": 3.23,
+      "learning_rate": 7.845876536842846e-06,
+      "loss": 1.2594,
       "step": 122
     },
     {
-      "epoch": 1.63,
-      "learning_rate": 9.540376592256142e-06,
-      "loss": 2.3675,
+      "epoch": 3.26,
+      "learning_rate": 7.809691049818766e-06,
+      "loss": 1.2333,
       "step": 123
     },
     {
-      "epoch": 1.65,
-      "learning_rate": 9.531285254663997e-06,
-      "loss": 2.4487,
+      "epoch": 3.28,
+      "learning_rate": 7.773289196567066e-06,
+      "loss": 1.1418,
       "step": 124
     },
     {
-      "epoch": 1.66,
-      "learning_rate": 9.522109301235637e-06,
-      "loss": 2.4202,
+      "epoch": 3.31,
+      "learning_rate": 7.736673780289788e-06,
+      "loss": 1.3162,
       "step": 125
     },
     {
-      "epoch": 1.67,
-      "learning_rate": 9.512848903320017e-06,
-      "loss": 2.5347,
+      "epoch": 3.34,
+      "learning_rate": 7.699847620634834e-06,
+      "loss": 1.1933,
       "step": 126
     },
     {
-      "epoch": 1.68,
-      "learning_rate": 9.503504233842973e-06,
-      "loss": 2.5286,
+      "epoch": 3.36,
+      "learning_rate": 7.662813553478857e-06,
+      "loss": 1.1409,
       "step": 127
     },
     {
-      "epoch": 1.7,
-      "learning_rate": 9.494075467304007e-06,
-      "loss": 2.5092,
+      "epoch": 3.39,
+      "learning_rate": 7.625574430708867e-06,
+      "loss": 1.1902,
       "step": 128
     },
     {
-      "epoch": 1.71,
-      "learning_rate": 9.484562779773027e-06,
-      "loss": 2.4088,
+      "epoch": 3.42,
+      "learning_rate": 7.588133120002612e-06,
+      "loss": 1.1772,
       "step": 129
     },
     {
-      "epoch": 1.72,
-      "learning_rate": 9.474966348887055e-06,
-      "loss": 2.4655,
+      "epoch": 3.44,
+      "learning_rate": 7.5504925046077596e-06,
+      "loss": 1.1877,
       "step": 130
     },
     {
-      "epoch": 1.74,
-      "learning_rate": 9.465286353846905e-06,
-      "loss": 2.4513,
+      "epoch": 3.47,
+      "learning_rate": 7.5126554831198506e-06,
+      "loss": 1.1943,
       "step": 131
     },
     {
-      "epoch": 1.75,
-      "learning_rate": 9.455522975413846e-06,
-      "loss": 2.4779,
+      "epoch": 3.5,
+      "learning_rate": 7.474624969259101e-06,
+      "loss": 1.3182,
       "step": 132
     },
     {
-      "epoch": 1.76,
-      "learning_rate": 9.445676395906226e-06,
-      "loss": 2.4917,
+      "epoch": 3.52,
+      "learning_rate": 7.436403891646014e-06,
+      "loss": 1.2202,
       "step": 133
     },
     {
-      "epoch": 1.78,
-      "learning_rate": 9.435746799196061e-06,
-      "loss": 2.4344,
+      "epoch": 3.55,
+      "learning_rate": 7.3979951935758596e-06,
+      "loss": 1.4638,
       "step": 134
     },
     {
-      "epoch": 1.79,
-      "learning_rate": 9.425734370705606e-06,
-      "loss": 2.3951,
+      "epoch": 3.58,
+      "learning_rate": 7.359401832792019e-06,
+      "loss": 1.2086,
       "step": 135
     },
     {
-      "epoch": 1.8,
-      "learning_rate": 9.415639297403891e-06,
-      "loss": 2.5753,
+      "epoch": 3.6,
+      "learning_rate": 7.32062678125822e-06,
+      "loss": 1.26,
       "step": 136
     },
     {
-      "epoch": 1.82,
-      "learning_rate": 9.40546176780323e-06,
-      "loss": 2.4163,
+      "epoch": 3.63,
+      "learning_rate": 7.281673024929674e-06,
+      "loss": 1.1927,
       "step": 137
     },
     {
-      "epoch": 1.83,
-      "learning_rate": 9.395201971955701e-06,
-      "loss": 2.406,
+      "epoch": 3.66,
+      "learning_rate": 7.242543563523128e-06,
+      "loss": 1.3236,
       "step": 138
     },
     {
-      "epoch": 1.84,
-      "learning_rate": 9.384860101449598e-06,
-      "loss": 2.4533,
+      "epoch": 3.68,
+      "learning_rate": 7.2032414102858795e-06,
+      "loss": 1.2632,
       "step": 139
     },
     {
-      "epoch": 1.86,
-      "learning_rate": 9.374436349405847e-06,
-      "loss": 2.4783,
+      "epoch": 3.71,
+      "learning_rate": 7.163769591763723e-06,
+      "loss": 1.2393,
       "step": 140
     },
     {
-      "epoch": 1.87,
-      "learning_rate": 9.36393091047441e-06,
-      "loss": 2.4423,
+      "epoch": 3.74,
+      "learning_rate": 7.1241311475678896e-06,
+      "loss": 1.2769,
       "step": 141
     },
     {
-      "epoch": 1.88,
-      "learning_rate": 9.353343980830644e-06,
-      "loss": 2.4606,
+      "epoch": 3.76,
+      "learning_rate": 7.084329130140972e-06,
+      "loss": 1.177,
       "step": 142
     },
     {
-      "epoch": 1.9,
-      "learning_rate": 9.342675758171638e-06,
-      "loss": 2.3636,
+      "epoch": 3.79,
+      "learning_rate": 7.044366604521874e-06,
+      "loss": 1.3389,
       "step": 143
     },
     {
-      "epoch": 1.91,
-      "learning_rate": 9.331926441712522e-06,
-      "loss": 2.3702,
+      "epoch": 3.81,
+      "learning_rate": 7.004246648109765e-06,
+      "loss": 1.2673,
       "step": 144
     },
     {
-      "epoch": 1.92,
-      "learning_rate": 9.32109623218275e-06,
-      "loss": 2.4619,
+      "epoch": 3.84,
+      "learning_rate": 6.963972350427112e-06,
+      "loss": 1.3116,
       "step": 145
     },
     {
-      "epoch": 1.94,
-      "learning_rate": 9.310185331822338e-06,
-      "loss": 2.2859,
+      "epoch": 3.87,
+      "learning_rate": 6.923546812881759e-06,
+      "loss": 1.1558,
       "step": 146
     },
     {
-      "epoch": 1.95,
-      "learning_rate": 9.299193944378112e-06,
-      "loss": 2.3135,
+      "epoch": 3.89,
+      "learning_rate": 6.882973148528096e-06,
+      "loss": 1.2577,
       "step": 147
     },
     {
-      "epoch": 1.96,
-      "learning_rate": 9.28812227509988e-06,
-      "loss": 2.3796,
+      "epoch": 3.92,
+      "learning_rate": 6.8422544818273336e-06,
+      "loss": 1.1485,
       "step": 148
     },
     {
-      "epoch": 1.98,
-      "learning_rate": 9.27697053073661e-06,
-      "loss": 2.5387,
+      "epoch": 3.95,
+      "learning_rate": 6.801393948406894e-06,
+      "loss": 1.3308,
       "step": 149
     },
     {
-      "epoch": 1.99,
-      "learning_rate": 9.26573891953257e-06,
-      "loss": 2.4443,
+      "epoch": 3.97,
+      "learning_rate": 6.760394694818949e-06,
+      "loss": 1.1488,
       "step": 150
     },
     {
-      "epoch": 2.0,
-      "learning_rate": 9.254427651223434e-06,
-      "loss": 2.3509,
+      "epoch": 4.0,
+      "learning_rate": 6.719259878298113e-06,
+      "loss": 1.2665,
       "step": 151
     },
     {
-      "epoch": 2.02,
-      "learning_rate": 9.243036937032373e-06,
-      "loss": 2.3877,
+      "epoch": 4.03,
+      "learning_rate": 6.67799266651831e-06,
+      "loss": 1.2601,
       "step": 152
     },
     {
-      "epoch": 2.03,
-      "learning_rate": 9.2315669896661e-06,
-      "loss": 2.4596,
+      "epoch": 4.05,
+      "learning_rate": 6.63659623734884e-06,
+      "loss": 1.2757,
       "step": 153
     },
     {
-      "epoch": 2.04,
-      "learning_rate": 9.220018023310908e-06,
-      "loss": 2.5048,
+      "epoch": 4.08,
+      "learning_rate": 6.59507377860967e-06,
+      "loss": 1.3131,
       "step": 154
     },
     {
-      "epoch": 2.06,
-      "learning_rate": 9.208390253628667e-06,
-      "loss": 2.3846,
+      "epoch": 4.11,
+      "learning_rate": 6.553428487825933e-06,
+      "loss": 1.4065,
       "step": 155
     },
     {
-      "epoch": 2.07,
-      "learning_rate": 9.196683897752794e-06,
-      "loss": 2.3357,
+      "epoch": 4.13,
+      "learning_rate": 6.511663571981708e-06,
+      "loss": 1.2555,
       "step": 156
     },
     {
-      "epoch": 2.08,
-      "learning_rate": 9.184899174284201e-06,
-      "loss": 2.6323,
+      "epoch": 4.16,
+      "learning_rate": 6.46978224727306e-06,
+      "loss": 1.2421,
       "step": 157
     },
     {
-      "epoch": 2.1,
-      "learning_rate": 9.173036303287215e-06,
-      "loss": 2.4945,
+      "epoch": 4.19,
+      "learning_rate": 6.42778773886036e-06,
+      "loss": 1.3245,
       "step": 158
     },
     {
-      "epoch": 2.11,
-      "learning_rate": 9.16109550628546e-06,
-      "loss": 2.3997,
+      "epoch": 4.21,
+      "learning_rate": 6.3856832806199324e-06,
+      "loss": 1.3012,
       "step": 159
     },
     {
-      "epoch": 2.12,
-      "learning_rate": 9.149077006257734e-06,
-      "loss": 2.4204,
+      "epoch": 4.24,
+      "learning_rate": 6.343472114895022e-06,
+      "loss": 1.155,
       "step": 160
     },
     {
-      "epoch": 2.14,
-      "learning_rate": 9.136981027633834e-06,
-      "loss": 2.4147,
+      "epoch": 4.26,
+      "learning_rate": 6.301157492246111e-06,
+      "loss": 1.2554,
       "step": 161
     },
     {
-      "epoch": 2.15,
-      "learning_rate": 9.124807796290366e-06,
-      "loss": 2.3096,
+      "epoch": 4.29,
+      "learning_rate": 6.2587426712006005e-06,
+      "loss": 1.348,
       "step": 162
     },
     {
-      "epoch": 2.16,
-      "learning_rate": 9.112557539546535e-06,
-      "loss": 2.3793,
+      "epoch": 4.32,
+      "learning_rate": 6.216230918001881e-06,
+      "loss": 1.2359,
       "step": 163
     },
     {
-      "epoch": 2.18,
-      "learning_rate": 9.100230486159893e-06,
-      "loss": 2.5293,
+      "epoch": 4.34,
+      "learning_rate": 6.173625506357814e-06,
+      "loss": 1.2989,
       "step": 164
     },
     {
-      "epoch": 2.19,
-      "learning_rate": 9.087826866322065e-06,
-      "loss": 2.3883,
+      "epoch": 4.37,
+      "learning_rate": 6.1309297171886225e-06,
+      "loss": 1.2298,
       "step": 165
     },
     {
-      "epoch": 2.2,
-      "learning_rate": 9.075346911654456e-06,
-      "loss": 2.321,
+      "epoch": 4.4,
+      "learning_rate": 6.088146838374247e-06,
+      "loss": 1.2104,
       "step": 166
-    },
-    {
-      "epoch": 2.22,
-      "learning_rate": 9.062790855203932e-06,
-      "loss": 2.3753,
-      "step": 167
-    },
-    {
-      "epoch": 2.23,
-      "learning_rate": 9.050158931438451e-06,
-      "loss": 2.3658,
-      "step": 168
-    },
-    {
-      "epoch": 2.24,
-      "learning_rate": 9.037451376242696e-06,
-      "loss": 2.3558,
-      "step": 169
-    },
-    {
-      "epoch": 2.26,
-      "learning_rate": 9.024668426913671e-06,
-      "loss": 2.3781,
-      "step": 170
-    },
-    {
-      "epoch": 2.27,
-      "learning_rate": 9.011810322156269e-06,
-      "loss": 2.3236,
-      "step": 171
-    },
-    {
-      "epoch": 2.28,
-      "learning_rate": 8.998877302078803e-06,
-      "loss": 2.3662,
-      "step": 172
-    },
-    {
-      "epoch": 2.3,
-      "learning_rate": 8.985869608188545e-06,
-      "loss": 2.4065,
-      "step": 173
-    },
-    {
-      "epoch": 2.31,
-      "learning_rate": 8.97278748338719e-06,
-      "loss": 2.56,
-      "step": 174
-    },
-    {
-      "epoch": 2.32,
-      "learning_rate": 8.95963117196634e-06,
-      "loss": 2.4793,
-      "step": 175
-    },
-    {
-      "epoch": 2.33,
-      "learning_rate": 8.946400919602933e-06,
-      "loss": 2.4959,
-      "step": 176
-    },
-    {
-      "epoch": 2.35,
-      "learning_rate": 8.933096973354665e-06,
-      "loss": 2.3332,
-      "step": 177
-    },
-    {
-      "epoch": 2.36,
-      "learning_rate": 8.919719581655357e-06,
-      "loss": 2.4093,
-      "step": 178
-    },
-    {
-      "epoch": 2.37,
-      "learning_rate": 8.906268994310339e-06,
-      "loss": 2.4014,
-      "step": 179
-    },
-    {
-      "epoch": 2.39,
-      "learning_rate": 8.892745462491763e-06,
-      "loss": 2.3916,
-      "step": 180
-    },
-    {
-      "epoch": 2.4,
-      "learning_rate": 8.879149238733932e-06,
-      "loss": 2.3817,
-      "step": 181
-    },
-    {
-      "epoch": 2.41,
-      "learning_rate": 8.865480576928578e-06,
-      "loss": 2.5024,
-      "step": 182
-    },
-    {
-      "epoch": 2.43,
-      "learning_rate": 8.851739732320109e-06,
-      "loss": 2.3461,
-      "step": 183
-    },
-    {
-      "epoch": 2.44,
-      "learning_rate": 8.83792696150086e-06,
-      "loss": 2.3118,
-      "step": 184
-    },
-    {
-      "epoch": 2.45,
-      "learning_rate": 8.824042522406295e-06,
-      "loss": 2.51,
-      "step": 185
-    },
-    {
-      "epoch": 2.47,
-      "learning_rate": 8.810086674310184e-06,
-      "loss": 2.3644,
-      "step": 186
-    },
-    {
-      "epoch": 2.48,
-      "learning_rate": 8.796059677819773e-06,
-      "loss": 2.434,
-      "step": 187
-    },
-    {
-      "epoch": 2.49,
-      "learning_rate": 8.781961794870903e-06,
-      "loss": 2.3205,
-      "step": 188
-    },
-    {
-      "epoch": 2.51,
-      "learning_rate": 8.767793288723137e-06,
-      "loss": 2.3496,
-      "step": 189
-    },
-    {
-      "epoch": 2.52,
-      "learning_rate": 8.753554423954828e-06,
-      "loss": 2.545,
-      "step": 190
-    },
-    {
-      "epoch": 2.53,
-      "learning_rate": 8.739245466458187e-06,
-      "loss": 2.3487,
-      "step": 191
-    },
-    {
-      "epoch": 2.55,
-      "learning_rate": 8.72486668343431e-06,
-      "loss": 2.4724,
-      "step": 192
-    },
-    {
-      "epoch": 2.56,
-      "learning_rate": 8.7104183433882e-06,
-      "loss": 2.3646,
-      "step": 193
-    },
-    {
-      "epoch": 2.57,
-      "learning_rate": 8.695900716123744e-06,
-      "loss": 2.3395,
-      "step": 194
-    },
-    {
-      "epoch": 2.59,
-      "learning_rate": 8.681314072738678e-06,
-      "loss": 2.5131,
-      "step": 195
-    },
-    {
-      "epoch": 2.6,
-      "learning_rate": 8.666658685619523e-06,
-      "loss": 2.4721,
-      "step": 196
-    },
-    {
-      "epoch": 2.61,
-      "learning_rate": 8.651934828436497e-06,
-      "loss": 2.3247,
-      "step": 197
-    },
-    {
-      "epoch": 2.63,
-      "learning_rate": 8.637142776138415e-06,
-      "loss": 2.4546,
-      "step": 198
-    },
-    {
-      "epoch": 2.64,
-      "learning_rate": 8.622282804947537e-06,
-      "loss": 2.3452,
-      "step": 199
-    },
-    {
-      "epoch": 2.65,
-      "learning_rate": 8.607355192354425e-06,
-      "loss": 2.3957,
-      "step": 200
-    },
-    {
-      "epoch": 2.67,
-      "learning_rate": 8.592360217112759e-06,
-      "loss": 2.3971,
-      "step": 201
-    },
-    {
-      "epoch": 2.68,
-      "learning_rate": 8.57729815923412e-06,
-      "loss": 2.2656,
-      "step": 202
-    },
-    {
-      "epoch": 2.69,
-      "learning_rate": 8.562169299982776e-06,
-      "loss": 2.4263,
-      "step": 203
-    },
-    {
-      "epoch": 2.71,
-      "learning_rate": 8.546973921870421e-06,
-      "loss": 2.4916,
-      "step": 204
-    },
-    {
-      "epoch": 2.72,
-      "learning_rate": 8.531712308650904e-06,
-      "loss": 2.469,
-      "step": 205
-    },
-    {
-      "epoch": 2.73,
-      "learning_rate": 8.516384745314926e-06,
-      "loss": 2.3268,
-      "step": 206
-    },
-    {
-      "epoch": 2.75,
-      "learning_rate": 8.50099151808472e-06,
-      "loss": 2.3219,
-      "step": 207
-    },
-    {
-      "epoch": 2.76,
-      "learning_rate": 8.485532914408712e-06,
-      "loss": 2.3158,
-      "step": 208
-    },
-    {
-      "epoch": 2.77,
-      "learning_rate": 8.470009222956138e-06,
-      "loss": 2.3291,
-      "step": 209
-    },
-    {
-      "epoch": 2.79,
-      "learning_rate": 8.45442073361167e-06,
-      "loss": 2.4595,
-      "step": 210
-    },
-    {
-      "epoch": 2.8,
-      "learning_rate": 8.438767737469995e-06,
-      "loss": 2.3873,
-      "step": 211
-    },
-    {
-      "epoch": 2.81,
-      "learning_rate": 8.42305052683038e-06,
-      "loss": 2.3981,
-      "step": 212
-    },
-    {
-      "epoch": 2.83,
-      "learning_rate": 8.407269395191216e-06,
-      "loss": 2.3886,
-      "step": 213
-    },
-    {
-      "epoch": 2.84,
-      "learning_rate": 8.391424637244528e-06,
-      "loss": 2.3819,
-      "step": 214
-    },
-    {
-      "epoch": 2.85,
-      "learning_rate": 8.375516548870489e-06,
-      "loss": 2.3791,
-      "step": 215
-    },
-    {
-      "epoch": 2.87,
-      "learning_rate": 8.359545427131876e-06,
-      "loss": 2.3164,
-      "step": 216
-    },
-    {
-      "epoch": 2.88,
-      "learning_rate": 8.343511570268541e-06,
-      "loss": 2.3921,
-      "step": 217
-    },
-    {
-      "epoch": 2.89,
-      "learning_rate": 8.327415277691824e-06,
-      "loss": 2.3502,
-      "step": 218
-    },
-    {
-      "epoch": 2.91,
-      "learning_rate": 8.311256849978974e-06,
-      "loss": 2.2597,
-      "step": 219
-    },
-    {
-      "epoch": 2.92,
-      "learning_rate": 8.295036588867533e-06,
-      "loss": 2.3938,
-      "step": 220
-    },
-    {
-      "epoch": 2.93,
-      "learning_rate": 8.278754797249702e-06,
-      "loss": 2.4966,
-      "step": 221
-    },
-    {
-      "epoch": 2.95,
-      "learning_rate": 8.262411779166681e-06,
-      "loss": 2.3562,
-      "step": 222
-    },
-    {
-      "epoch": 2.96,
-      "learning_rate": 8.246007839802997e-06,
-      "loss": 2.3949,
-      "step": 223
-    },
-    {
-      "epoch": 2.97,
-      "learning_rate": 8.229543285480797e-06,
-      "loss": 2.3203,
-      "step": 224
-    },
-    {
-      "epoch": 2.99,
-      "learning_rate": 8.213018423654144e-06,
-      "loss": 2.3606,
-      "step": 225
-    },
-    {
-      "epoch": 3.0,
-      "learning_rate": 8.196433562903252e-06,
-      "loss": 2.3521,
-      "step": 226
-    },
-    {
-      "epoch": 3.01,
-      "learning_rate": 8.179789012928747e-06,
-      "loss": 2.4479,
-      "step": 227
-    },
-    {
-      "epoch": 3.02,
-      "learning_rate": 8.163085084545867e-06,
-      "loss": 2.4099,
-      "step": 228
-    },
-    {
-      "epoch": 3.04,
-      "learning_rate": 8.146322089678668e-06,
-      "loss": 2.4377,
-      "step": 229
-    },
-    {
-      "epoch": 3.05,
-      "learning_rate": 8.129500341354192e-06,
-      "loss": 2.4691,
-      "step": 230
-    },
-    {
-      "epoch": 3.06,
-      "learning_rate": 8.11262015369663e-06,
-      "loss": 2.2523,
-      "step": 231
-    },
-    {
-      "epoch": 3.08,
-      "learning_rate": 8.095681841921441e-06,
-      "loss": 2.4388,
-      "step": 232
-    },
-    {
-      "epoch": 3.09,
-      "learning_rate": 8.07868572232949e-06,
-      "loss": 2.408,
-      "step": 233
-    },
-    {
-      "epoch": 3.1,
-      "learning_rate": 8.061632112301122e-06,
-      "loss": 2.5245,
-      "step": 234
-    },
-    {
-      "epoch": 3.12,
-      "learning_rate": 8.044521330290235e-06,
-      "loss": 2.3994,
-      "step": 235
-    },
-    {
-      "epoch": 3.13,
-      "learning_rate": 8.027353695818345e-06,
-      "loss": 2.2924,
-      "step": 236
-    },
-    {
-      "epoch": 3.14,
-      "learning_rate": 8.010129529468614e-06,
-      "loss": 2.2718,
-      "step": 237
-    },
-    {
-      "epoch": 3.16,
-      "learning_rate": 7.992849152879857e-06,
-      "loss": 2.4062,
-      "step": 238
-    },
-    {
-      "epoch": 3.17,
-      "learning_rate": 7.97551288874055e-06,
-      "loss": 2.3462,
-      "step": 239
-    },
-    {
-      "epoch": 3.18,
-      "learning_rate": 7.95812106078279e-06,
-      "loss": 2.3439,
-      "step": 240
-    },
-    {
-      "epoch": 3.2,
-      "learning_rate": 7.940673993776258e-06,
-      "loss": 2.4067,
-      "step": 241
-    },
-    {
-      "epoch": 3.21,
-      "learning_rate": 7.923172013522153e-06,
-      "loss": 2.4119,
-      "step": 242
-    },
-    {
-      "epoch": 3.22,
-      "learning_rate": 7.905615446847107e-06,
-      "loss": 2.3155,
-      "step": 243
-    },
-    {
-      "epoch": 3.24,
-      "learning_rate": 7.888004621597079e-06,
-      "loss": 2.5228,
-      "step": 244
-    },
-    {
-      "epoch": 3.25,
-      "learning_rate": 7.87033986663124e-06,
-      "loss": 2.4585,
-      "step": 245
-    },
-    {
-      "epoch": 3.26,
-      "learning_rate": 7.852621511815825e-06,
-      "loss": 2.4616,
-      "step": 246
-    },
-    {
-      "epoch": 3.28,
-      "learning_rate": 7.834849888017979e-06,
-      "loss": 2.4858,
-      "step": 247
-    },
-    {
-      "epoch": 3.29,
-      "learning_rate": 7.817025327099574e-06,
-      "loss": 2.4023,
-      "step": 248
-    },
-    {
-      "epoch": 3.3,
-      "learning_rate": 7.799148161911013e-06,
-      "loss": 2.279,
-      "step": 249
-    },
-    {
-      "epoch": 1.01,
-      "learning_rate": 7.781218726285014e-06,
-      "loss": 2.3382,
-      "step": 250
-    },
-    {
-      "epoch": 1.02,
-      "learning_rate": 7.763237355030384e-06,
-      "loss": 2.2992,
-      "step": 251
-    },
-    {
-      "epoch": 1.03,
-      "learning_rate": 7.745204383925753e-06,
-      "loss": 2.3759,
-      "step": 252
-    },
-    {
-      "epoch": 1.05,
-      "learning_rate": 7.727120149713313e-06,
-      "loss": 2.3034,
-      "step": 253
-    },
-    {
-      "epoch": 1.06,
-      "learning_rate": 7.708984990092528e-06,
-      "loss": 2.3704,
-      "step": 254
-    },
-    {
-      "epoch": 1.07,
-      "learning_rate": 7.690799243713825e-06,
-      "loss": 2.2834,
-      "step": 255
-    },
-    {
-      "epoch": 1.09,
-      "learning_rate": 7.672563250172278e-06,
-      "loss": 2.3202,
-      "step": 256
-    },
-    {
-      "epoch": 1.1,
-      "learning_rate": 7.654277350001255e-06,
-      "loss": 2.3212,
-      "step": 257
-    },
-    {
-      "epoch": 1.11,
-      "learning_rate": 7.635941884666072e-06,
-      "loss": 2.3898,
-      "step": 258
-    },
-    {
-      "epoch": 1.13,
-      "learning_rate": 7.617557196557601e-06,
-      "loss": 2.4211,
-      "step": 259
-    },
-    {
-      "epoch": 1.14,
-      "learning_rate": 7.599123628985894e-06,
-      "loss": 2.4041,
-      "step": 260
-    },
-    {
-      "epoch": 1.15,
-      "learning_rate": 7.580641526173758e-06,
-      "loss": 2.3458,
-      "step": 261
-    },
-    {
-      "epoch": 1.17,
-      "learning_rate": 7.5621112332503325e-06,
-      "loss": 2.3498,
-      "step": 262
-    },
-    {
-      "epoch": 1.18,
-      "learning_rate": 7.543533096244644e-06,
-      "loss": 2.3767,
-      "step": 263
-    },
-    {
-      "epoch": 1.19,
-      "learning_rate": 7.524907462079149e-06,
-      "loss": 2.529,
-      "step": 264
-    },
-    {
-      "epoch": 1.21,
-      "learning_rate": 7.506234678563248e-06,
-      "loss": 2.3742,
-      "step": 265
-    },
-    {
-      "epoch": 1.22,
-      "learning_rate": 7.487515094386792e-06,
-      "loss": 2.3069,
-      "step": 266
-    },
-    {
-      "epoch": 1.23,
-      "learning_rate": 7.468749059113578e-06,
-      "loss": 2.3369,
-      "step": 267
-    },
-    {
-      "epoch": 1.25,
-      "learning_rate": 7.449936923174813e-06,
-      "loss": 2.3661,
-      "step": 268
-    },
-    {
-      "epoch": 1.26,
-      "learning_rate": 7.431079037862575e-06,
-      "loss": 2.3691,
-      "step": 269
-    },
-    {
-      "epoch": 1.27,
-      "learning_rate": 7.412175755323254e-06,
-      "loss": 2.361,
-      "step": 270
-    },
-    {
-      "epoch": 1.29,
-      "learning_rate": 7.39322742855097e-06,
-      "loss": 2.3486,
-      "step": 271
-    },
-    {
-      "epoch": 1.3,
-      "learning_rate": 7.374234411380987e-06,
-      "loss": 2.3527,
-      "step": 272
-    },
-    {
-      "epoch": 1.31,
-      "learning_rate": 7.355197058483103e-06,
-      "loss": 2.3877,
-      "step": 273
-    },
-    {
-      "epoch": 1.33,
-      "learning_rate": 7.336115725355033e-06,
-      "loss": 2.4065,
-      "step": 274
-    },
-    {
-      "epoch": 1.34,
-      "learning_rate": 7.316990768315757e-06,
-      "loss": 2.2512,
-      "step": 275
-    },
-    {
-      "epoch": 1.35,
-      "learning_rate": 7.297822544498879e-06,
-      "loss": 2.3204,
-      "step": 276
-    },
-    {
-      "epoch": 1.36,
-      "learning_rate": 7.2786114118459564e-06,
-      "loss": 2.2819,
-      "step": 277
-    },
-    {
-      "epoch": 1.38,
-      "learning_rate": 7.259357729099805e-06,
-      "loss": 2.3463,
-      "step": 278
-    },
-    {
-      "epoch": 1.39,
-      "learning_rate": 7.240061855797818e-06,
-      "loss": 2.4142,
-      "step": 279
-    },
-    {
-      "epoch": 1.4,
-      "learning_rate": 7.220724152265234e-06,
-      "loss": 2.4426,
-      "step": 280
-    },
-    {
-      "epoch": 1.42,
-      "learning_rate": 7.201344979608423e-06,
-      "loss": 2.3442,
-      "step": 281
-    },
-    {
-      "epoch": 1.43,
-      "learning_rate": 7.181924699708127e-06,
-      "loss": 2.4645,
-      "step": 282
-    },
-    {
-      "epoch": 1.44,
-      "learning_rate": 7.162463675212726e-06,
-      "loss": 2.2494,
-      "step": 283
-    },
-    {
-      "epoch": 1.46,
-      "learning_rate": 7.142962269531439e-06,
-      "loss": 2.3921,
-      "step": 284
-    },
-    {
-      "epoch": 1.47,
-      "learning_rate": 7.12342084682756e-06,
-      "loss": 2.3148,
-      "step": 285
-    },
-    {
-      "epoch": 1.48,
-      "learning_rate": 7.1038397720116445e-06,
-      "loss": 2.3584,
-      "step": 286
-    },
-    {
-      "epoch": 1.5,
-      "learning_rate": 7.084219410734701e-06,
-      "loss": 2.2597,
-      "step": 287
-    },
-    {
-      "epoch": 1.51,
-      "learning_rate": 7.064560129381359e-06,
-      "loss": 2.3235,
-      "step": 288
-    },
-    {
-      "epoch": 1.52,
-      "learning_rate": 7.0448622950630305e-06,
-      "loss": 2.345,
-      "step": 289
-    },
-    {
-      "epoch": 1.54,
-      "learning_rate": 7.025126275611058e-06,
-      "loss": 2.4172,
-      "step": 290
-    },
-    {
-      "epoch": 1.55,
-      "learning_rate": 7.0053524395698345e-06,
-      "loss": 2.3613,
-      "step": 291
-    },
-    {
-      "epoch": 1.56,
-      "learning_rate": 6.985541156189932e-06,
-      "loss": 2.3586,
-      "step": 292
-    },
-    {
-      "epoch": 1.58,
-      "learning_rate": 6.965692795421206e-06,
-      "loss": 2.4496,
-      "step": 293
-    },
-    {
-      "epoch": 1.59,
-      "learning_rate": 6.945807727905876e-06,
-      "loss": 2.3308,
-      "step": 294
-    },
-    {
-      "epoch": 1.6,
-      "learning_rate": 6.925886324971619e-06,
-      "loss": 2.415,
-      "step": 295
-    },
-    {
-      "epoch": 1.62,
-      "learning_rate": 6.905928958624627e-06,
-      "loss": 2.3855,
-      "step": 296
-    },
-    {
-      "epoch": 1.63,
-      "learning_rate": 6.885936001542658e-06,
-      "loss": 2.4297,
-      "step": 297
-    },
-    {
-      "epoch": 1.64,
-      "learning_rate": 6.865907827068085e-06,
-      "loss": 2.3936,
-      "step": 298
-    },
-    {
-      "epoch": 1.66,
-      "learning_rate": 6.845844809200918e-06,
-      "loss": 2.3937,
-      "step": 299
-    },
-    {
-      "epoch": 1.67,
-      "learning_rate": 6.82574732259182e-06,
-      "loss": 2.3423,
-      "step": 300
-    },
-    {
-      "epoch": 1.68,
-      "learning_rate": 6.805615742535117e-06,
-      "loss": 2.317,
-      "step": 301
-    },
-    {
-      "epoch": 1.7,
-      "learning_rate": 6.785450444961783e-06,
-      "loss": 2.4252,
-      "step": 302
-    },
-    {
-      "epoch": 1.71,
-      "learning_rate": 6.765251806432423e-06,
-      "loss": 2.2925,
-      "step": 303
-    },
-    {
-      "epoch": 1.72,
-      "learning_rate": 6.7450202041302404e-06,
-      "loss": 2.3396,
-      "step": 304
-    },
-    {
-      "epoch": 1.74,
-      "learning_rate": 6.724756015853994e-06,
-      "loss": 2.4506,
-      "step": 305
-    },
-    {
-      "epoch": 1.75,
-      "learning_rate": 6.704459620010945e-06,
-      "loss": 2.3618,
-      "step": 306
-    },
-    {
-      "epoch": 1.76,
-      "learning_rate": 6.684131395609784e-06,
-      "loss": 2.2474,
-      "step": 307
-    },
-    {
-      "epoch": 1.78,
-      "learning_rate": 6.663771722253567e-06,
-      "loss": 2.2596,
-      "step": 308
-    },
-    {
-      "epoch": 1.79,
-      "learning_rate": 6.643380980132608e-06,
-      "loss": 2.3431,
-      "step": 309
-    },
-    {
-      "epoch": 1.8,
-      "learning_rate": 6.622959550017397e-06,
-      "loss": 2.2327,
-      "step": 310
-    },
-    {
-      "epoch": 1.82,
-      "learning_rate": 6.602507813251478e-06,
-      "loss": 2.2748,
-      "step": 311
-    },
-    {
-      "epoch": 1.83,
-      "learning_rate": 6.5820261517443365e-06,
-      "loss": 2.2585,
-      "step": 312
-    },
-    {
-      "epoch": 1.84,
-      "learning_rate": 6.561514947964258e-06,
-      "loss": 2.3711,
-      "step": 313
-    },
-    {
-      "epoch": 1.86,
-      "learning_rate": 6.540974584931199e-06,
-      "loss": 2.3061,
-      "step": 314
-    },
-    {
-      "epoch": 1.87,
-      "learning_rate": 6.520405446209615e-06,
-      "loss": 2.2768,
-      "step": 315
-    },
-    {
-      "epoch": 1.88,
-      "learning_rate": 6.4998079159013236e-06,
-      "loss": 2.3563,
-      "step": 316
-    },
-    {
-      "epoch": 1.9,
-      "learning_rate": 6.479182378638308e-06,
-      "loss": 2.3836,
-      "step": 317
-    },
-    {
-      "epoch": 1.91,
-      "learning_rate": 6.458529219575551e-06,
-      "loss": 2.3632,
-      "step": 318
-    },
-    {
-      "epoch": 1.92,
-      "learning_rate": 6.437848824383832e-06,
-      "loss": 2.3514,
-      "step": 319
-    },
-    {
-      "epoch": 1.94,
-      "learning_rate": 6.417141579242532e-06,
-      "loss": 2.5659,
-      "step": 320
-    },
-    {
-      "epoch": 1.95,
-      "learning_rate": 6.396407870832419e-06,
-      "loss": 2.346,
-      "step": 321
-    },
-    {
-      "epoch": 1.96,
-      "learning_rate": 6.375648086328431e-06,
-      "loss": 2.3057,
-      "step": 322
-    },
-    {
-      "epoch": 1.98,
-      "learning_rate": 6.354862613392436e-06,
-      "loss": 2.3792,
-      "step": 323
-    },
-    {
-      "epoch": 1.99,
-      "learning_rate": 6.334051840166006e-06,
-      "loss": 2.3504,
-      "step": 324
-    },
-    {
-      "epoch": 2.0,
-      "learning_rate": 6.313216155263161e-06,
-      "loss": 2.356,
-      "step": 325
-    },
-    {
-      "epoch": 2.01,
-      "learning_rate": 6.292355947763114e-06,
-      "loss": 2.3169,
-      "step": 326
-    },
-    {
-      "epoch": 2.03,
-      "learning_rate": 6.271471607203006e-06,
-      "loss": 2.315,
-      "step": 327
-    },
-    {
-      "epoch": 2.04,
-      "learning_rate": 6.25056352357063e-06,
-      "loss": 2.3571,
-      "step": 328
-    },
-    {
-      "epoch": 2.05,
-      "learning_rate": 6.2296320872971515e-06,
-      "loss": 2.2367,
-      "step": 329
-    },
-    {
-      "epoch": 2.07,
-      "learning_rate": 6.208677689249816e-06,
-      "loss": 2.4162,
-      "step": 330
-    },
-    {
-      "epoch": 2.08,
-      "learning_rate": 6.187700720724648e-06,
-      "loss": 2.3897,
-      "step": 331
-    },
-    {
-      "epoch": 2.09,
-      "learning_rate": 6.16670157343915e-06,
-      "loss": 2.3532,
-      "step": 332
-    },
-    {
-      "epoch": 2.11,
-      "learning_rate": 6.14568063952498e-06,
-      "loss": 2.345,
-      "step": 333
-    },
-    {
-      "epoch": 2.12,
-      "learning_rate": 6.124638311520634e-06,
-      "loss": 2.3358,
-      "step": 334
-    },
-    {
-      "epoch": 2.13,
-      "learning_rate": 6.103574982364118e-06,
-      "loss": 2.3113,
-      "step": 335
-    },
-    {
-      "epoch": 2.15,
-      "learning_rate": 6.082491045385601e-06,
-      "loss": 2.2876,
-      "step": 336
-    },
-    {
-      "epoch": 2.16,
-      "learning_rate": 6.061386894300082e-06,
-      "loss": 2.3047,
-      "step": 337
-    },
-    {
-      "epoch": 2.17,
-      "learning_rate": 6.0402629232000275e-06,
-      "loss": 2.2679,
-      "step": 338
-    },
-    {
-      "epoch": 2.19,
-      "learning_rate": 6.01911952654802e-06,
-      "loss": 2.2512,
-      "step": 339
-    },
-    {
-      "epoch": 2.2,
-      "learning_rate": 5.997957099169388e-06,
-      "loss": 2.3577,
-      "step": 340
-    },
-    {
-      "epoch": 2.21,
-      "learning_rate": 5.976776036244833e-06,
-      "loss": 2.3674,
-      "step": 341
-    },
-    {
-      "epoch": 2.23,
-      "learning_rate": 5.955576733303053e-06,
-      "loss": 2.1205,
-      "step": 342
-    },
-    {
-      "epoch": 2.24,
-      "learning_rate": 5.9343595862133515e-06,
-      "loss": 2.2911,
-      "step": 343
-    },
-    {
-      "epoch": 2.25,
-      "learning_rate": 5.91312499117825e-06,
-      "loss": 2.247,
-      "step": 344
-    },
-    {
-      "epoch": 2.27,
-      "learning_rate": 5.891873344726089e-06,
-      "loss": 2.3012,
-      "step": 345
-    },
-    {
-      "epoch": 2.28,
-      "learning_rate": 5.87060504370362e-06,
-      "loss": 2.2817,
-      "step": 346
-    },
-    {
-      "epoch": 2.29,
-      "learning_rate": 5.849320485268597e-06,
-      "loss": 2.3684,
-      "step": 347
-    },
-    {
-      "epoch": 2.31,
-      "learning_rate": 5.828020066882361e-06,
-      "loss": 2.4574,
-      "step": 348
-    },
-    {
-      "epoch": 2.32,
-      "learning_rate": 5.806704186302413e-06,
-      "loss": 2.2969,
-      "step": 349
-    },
-    {
-      "epoch": 2.33,
-      "learning_rate": 5.7853732415749985e-06,
-      "loss": 2.3386,
-      "step": 350
-    },
-    {
-      "epoch": 2.35,
-      "learning_rate": 5.764027631027659e-06,
-      "loss": 2.2298,
-      "step": 351
-    },
-    {
-      "epoch": 2.36,
-      "learning_rate": 5.7426677532618e-06,
-      "loss": 2.4084,
-      "step": 352
-    },
-    {
-      "epoch": 2.37,
-      "learning_rate": 5.721294007145256e-06,
-      "loss": 2.3175,
-      "step": 353
-    },
-    {
-      "epoch": 2.39,
-      "learning_rate": 5.69990679180483e-06,
-      "loss": 2.4387,
-      "step": 354
-    },
-    {
-      "epoch": 2.4,
-      "learning_rate": 5.678506506618845e-06,
-      "loss": 2.4227,
-      "step": 355
-    },
-    {
-      "epoch": 2.41,
-      "learning_rate": 5.657093551209687e-06,
-      "loss": 2.235,
-      "step": 356
-    },
-    {
-      "epoch": 2.43,
-      "learning_rate": 5.635668325436343e-06,
-      "loss": 2.2519,
-      "step": 357
-    },
-    {
-      "epoch": 2.44,
-      "learning_rate": 5.614231229386933e-06,
-      "loss": 2.3856,
-      "step": 358
-    },
-    {
-      "epoch": 2.45,
-      "learning_rate": 5.592782663371237e-06,
-      "loss": 2.3068,
-      "step": 359
-    },
-    {
-      "epoch": 2.47,
-      "learning_rate": 5.571323027913221e-06,
-      "loss": 2.2521,
-      "step": 360
-    },
-    {
-      "epoch": 2.48,
-      "learning_rate": 5.549852723743564e-06,
-      "loss": 2.1542,
-      "step": 361
-    },
-    {
-      "epoch": 2.49,
-      "learning_rate": 5.528372151792161e-06,
-      "loss": 2.3091,
-      "step": 362
-    },
-    {
-      "epoch": 2.51,
-      "learning_rate": 5.506881713180652e-06,
-      "loss": 2.4098,
-      "step": 363
-    },
-    {
-      "epoch": 2.52,
-      "learning_rate": 5.485381809214921e-06,
-      "loss": 2.3544,
-      "step": 364
-    },
-    {
-      "epoch": 2.53,
-      "learning_rate": 5.463872841377601e-06,
-      "loss": 2.4103,
-      "step": 365
-    },
-    {
-      "epoch": 2.55,
-      "learning_rate": 5.44235521132059e-06,
-      "loss": 2.3074,
-      "step": 366
-    },
-    {
-      "epoch": 2.56,
-      "learning_rate": 5.420829320857532e-06,
-      "loss": 2.3675,
-      "step": 367
-    },
-    {
-      "epoch": 2.57,
-      "learning_rate": 5.39929557195633e-06,
-      "loss": 2.3678,
-      "step": 368
-    },
-    {
-      "epoch": 2.59,
-      "learning_rate": 5.377754366731633e-06,
-      "loss": 2.2966,
-      "step": 369
-    },
-    {
-      "epoch": 2.6,
-      "learning_rate": 5.35620610743732e-06,
-      "loss": 2.3583,
-      "step": 370
-    },
-    {
-      "epoch": 2.61,
-      "learning_rate": 5.334651196459003e-06,
-      "loss": 2.2363,
-      "step": 371
-    },
-    {
-      "epoch": 2.63,
-      "learning_rate": 5.3130900363065055e-06,
-      "loss": 2.2912,
-      "step": 372
-    },
-    {
-      "epoch": 2.64,
-      "learning_rate": 5.291523029606339e-06,
-      "loss": 2.3355,
-      "step": 373
-    },
-    {
-      "epoch": 2.65,
-      "learning_rate": 5.269950579094199e-06,
-      "loss": 2.3922,
-      "step": 374
-    },
-    {
-      "epoch": 2.67,
-      "learning_rate": 5.248373087607434e-06,
-      "loss": 2.2934,
-      "step": 375
-    },
-    {
-      "epoch": 2.68,
-      "learning_rate": 5.22679095807753e-06,
-      "loss": 2.2972,
-      "step": 376
-    },
-    {
-      "epoch": 2.69,
-      "learning_rate": 5.2052045935225725e-06,
-      "loss": 2.3647,
-      "step": 377
-    },
-    {
-      "epoch": 2.7,
-      "learning_rate": 5.183614397039741e-06,
-      "loss": 2.2244,
-      "step": 378
-    },
-    {
-      "epoch": 2.72,
-      "learning_rate": 5.162020771797768e-06,
-      "loss": 2.4884,
-      "step": 379
-    },
-    {
-      "epoch": 2.73,
-      "learning_rate": 5.1404241210294095e-06,
-      "loss": 2.3397,
-      "step": 380
-    },
-    {
-      "epoch": 2.74,
-      "learning_rate": 5.118824848023926e-06,
-      "loss": 2.261,
-      "step": 381
-    },
-    {
-      "epoch": 2.76,
-      "learning_rate": 5.097223356119538e-06,
-      "loss": 2.4023,
-      "step": 382
-    },
-    {
-      "epoch": 2.77,
-      "learning_rate": 5.07562004869591e-06,
-      "loss": 2.3712,
-      "step": 383
-    },
-    {
-      "epoch": 2.78,
-      "learning_rate": 5.054015329166596e-06,
-      "loss": 2.298,
-      "step": 384
-    },
-    {
-      "epoch": 2.8,
-      "learning_rate": 5.032409600971533e-06,
-      "loss": 2.3549,
-      "step": 385
-    },
-    {
-      "epoch": 2.81,
-      "learning_rate": 5.010803267569483e-06,
-      "loss": 2.4179,
-      "step": 386
-    },
-    {
-      "epoch": 2.82,
-      "learning_rate": 4.989196732430518e-06,
-      "loss": 2.361,
-      "step": 387
-    },
-    {
-      "epoch": 2.84,
-      "learning_rate": 4.967590399028468e-06,
-      "loss": 2.3038,
-      "step": 388
-    },
-    {
-      "epoch": 2.85,
-      "learning_rate": 4.9459846708334044e-06,
-      "loss": 2.3168,
-      "step": 389
-    },
-    {
-      "epoch": 2.86,
-      "learning_rate": 4.924379951304094e-06,
-      "loss": 2.2561,
-      "step": 390
-    },
-    {
-      "epoch": 2.88,
-      "learning_rate": 4.902776643880461e-06,
-      "loss": 2.3989,
-      "step": 391
-    },
-    {
-      "epoch": 2.89,
-      "learning_rate": 4.881175151976075e-06,
-      "loss": 2.1956,
-      "step": 392
-    },
-    {
-      "epoch": 2.9,
-      "learning_rate": 4.859575878970592e-06,
-      "loss": 2.4416,
-      "step": 393
-    },
-    {
-      "epoch": 2.92,
-      "learning_rate": 4.837979228202234e-06,
-      "loss": 2.2633,
-      "step": 394
-    },
-    {
-      "epoch": 2.93,
-      "learning_rate": 4.81638560296026e-06,
-      "loss": 2.2302,
-      "step": 395
-    },
-    {
-      "epoch": 2.94,
-      "learning_rate": 4.794795406477429e-06,
-      "loss": 2.2946,
-      "step": 396
-    },
-    {
-      "epoch": 2.96,
-      "learning_rate": 4.773209041922472e-06,
-      "loss": 2.365,
-      "step": 397
-    },
-    {
-      "epoch": 2.97,
-      "learning_rate": 4.7516269123925665e-06,
-      "loss": 2.2991,
-      "step": 398
-    },
-    {
-      "epoch": 2.98,
-      "learning_rate": 4.730049420905801e-06,
-      "loss": 2.3378,
-      "step": 399
-    },
-    {
-      "epoch": 3.0,
-      "learning_rate": 4.708476970393662e-06,
-      "loss": 2.3307,
-      "step": 400
-    },
-    {
-      "epoch": 3.01,
-      "learning_rate": 4.686909963693498e-06,
-      "loss": 2.3909,
-      "step": 401
-    },
-    {
-      "epoch": 3.02,
-      "learning_rate": 4.6653488035409975e-06,
-      "loss": 2.3368,
-      "step": 402
-    },
-    {
-      "epoch": 3.04,
-      "learning_rate": 4.643793892562682e-06,
-      "loss": 2.2934,
-      "step": 403
-    },
-    {
-      "epoch": 3.05,
-      "learning_rate": 4.622245633268371e-06,
-      "loss": 2.2731,
-      "step": 404
-    },
-    {
-      "epoch": 3.06,
-      "learning_rate": 4.60070442804367e-06,
-      "loss": 2.3552,
-      "step": 405
-    },
-    {
-      "epoch": 3.08,
-      "learning_rate": 4.5791706791424694e-06,
-      "loss": 2.3491,
-      "step": 406
-    },
-    {
-      "epoch": 3.09,
-      "learning_rate": 4.557644788679413e-06,
-      "loss": 2.2096,
-      "step": 407
-    },
-    {
-      "epoch": 3.1,
-      "learning_rate": 4.536127158622401e-06,
-      "loss": 2.2307,
-      "step": 408
-    },
-    {
-      "epoch": 3.12,
-      "learning_rate": 4.514618190785081e-06,
-      "loss": 2.2926,
-      "step": 409
-    },
-    {
-      "epoch": 3.13,
-      "learning_rate": 4.493118286819348e-06,
-      "loss": 2.2682,
-      "step": 410
-    },
-    {
-      "epoch": 3.14,
-      "learning_rate": 4.47162784820784e-06,
-      "loss": 2.3075,
-      "step": 411
-    },
-    {
-      "epoch": 3.16,
-      "learning_rate": 4.450147276256439e-06,
-      "loss": 2.3367,
-      "step": 412
-    },
-    {
-      "epoch": 3.17,
-      "learning_rate": 4.42867697208678e-06,
-      "loss": 2.276,
-      "step": 413
-    },
-    {
-      "epoch": 3.18,
-      "learning_rate": 4.407217336628765e-06,
-      "loss": 2.3126,
-      "step": 414
-    },
-    {
-      "epoch": 3.2,
-      "learning_rate": 4.385768770613069e-06,
-      "loss": 2.1602,
-      "step": 415
-    },
-    {
-      "epoch": 3.21,
-      "learning_rate": 4.3643316745636574e-06,
-      "loss": 2.3406,
-      "step": 416
-    },
-    {
-      "epoch": 3.22,
-      "learning_rate": 4.342906448790315e-06,
-      "loss": 2.2879,
-      "step": 417
-    },
-    {
-      "epoch": 3.24,
-      "learning_rate": 4.321493493381157e-06,
-      "loss": 2.333,
-      "step": 418
-    },
-    {
-      "epoch": 3.25,
-      "learning_rate": 4.300093208195171e-06,
-      "loss": 2.3582,
-      "step": 419
-    },
-    {
-      "epoch": 3.26,
-      "learning_rate": 4.278705992854745e-06,
-      "loss": 2.3028,
-      "step": 420
-    },
-    {
-      "epoch": 3.28,
-      "learning_rate": 4.257332246738201e-06,
-      "loss": 2.3173,
-      "step": 421
-    },
-    {
-      "epoch": 3.29,
-      "learning_rate": 4.235972368972343e-06,
-      "loss": 2.3594,
-      "step": 422
-    },
-    {
-      "epoch": 3.3,
-      "learning_rate": 4.214626758425003e-06,
-      "loss": 2.267,
-      "step": 423
-    },
-    {
-      "epoch": 3.32,
-      "learning_rate": 4.193295813697587e-06,
-      "loss": 2.2458,
-      "step": 424
-    },
-    {
-      "epoch": 3.33,
-      "learning_rate": 4.171979933117641e-06,
-      "loss": 2.2125,
-      "step": 425
-    },
-    {
-      "epoch": 3.34,
-      "learning_rate": 4.150679514731405e-06,
-      "loss": 2.1231,
-      "step": 426
-    },
-    {
-      "epoch": 3.35,
-      "learning_rate": 4.12939495629638e-06,
-      "loss": 2.2268,
-      "step": 427
-    },
-    {
-      "epoch": 3.37,
-      "learning_rate": 4.108126655273912e-06,
-      "loss": 2.2652,
-      "step": 428
-    },
-    {
-      "epoch": 3.38,
-      "learning_rate": 4.086875008821752e-06,
-      "loss": 2.2586,
-      "step": 429
-    },
-    {
-      "epoch": 3.39,
-      "learning_rate": 4.06564041378665e-06,
-      "loss": 2.3749,
-      "step": 430
-    },
-    {
-      "epoch": 3.41,
-      "learning_rate": 4.04442326669695e-06,
-      "loss": 2.2593,
-      "step": 431
-    },
-    {
-      "epoch": 3.42,
-      "learning_rate": 4.023223963755168e-06,
-      "loss": 2.3175,
-      "step": 432
-    },
-    {
-      "epoch": 3.43,
-      "learning_rate": 4.002042900830613e-06,
-      "loss": 2.3238,
-      "step": 433
-    },
-    {
-      "epoch": 3.45,
-      "learning_rate": 3.980880473451982e-06,
-      "loss": 2.4295,
-      "step": 434
-    },
-    {
-      "epoch": 3.46,
-      "learning_rate": 3.959737076799974e-06,
-      "loss": 2.2343,
-      "step": 435
-    },
-    {
-      "epoch": 3.47,
-      "learning_rate": 3.93861310569992e-06,
-      "loss": 2.339,
-      "step": 436
-    },
-    {
-      "epoch": 3.49,
-      "learning_rate": 3.917508954614401e-06,
-      "loss": 2.3416,
-      "step": 437
-    },
-    {
-      "epoch": 3.5,
-      "learning_rate": 3.896425017635884e-06,
-      "loss": 2.3733,
-      "step": 438
-    },
-    {
-      "epoch": 3.51,
-      "learning_rate": 3.875361688479367e-06,
-      "loss": 2.3012,
-      "step": 439
-    },
-    {
-      "epoch": 3.53,
-      "learning_rate": 3.854319360475022e-06,
-      "loss": 2.296,
-      "step": 440
-    },
-    {
-      "epoch": 3.54,
-      "learning_rate": 3.833298426560851e-06,
-      "loss": 2.3398,
-      "step": 441
-    },
-    {
-      "epoch": 3.55,
-      "learning_rate": 3.8122992792753534e-06,
-      "loss": 2.2432,
-      "step": 442
-    },
-    {
-      "epoch": 3.57,
-      "learning_rate": 3.7913223107501847e-06,
-      "loss": 2.2624,
-      "step": 443
-    },
-    {
-      "epoch": 3.58,
-      "learning_rate": 3.7703679127028497e-06,
-      "loss": 2.3152,
-      "step": 444
-    },
-    {
-      "epoch": 3.59,
-      "learning_rate": 3.7494364764293722e-06,
-      "loss": 2.2709,
-      "step": 445
-    },
-    {
-      "epoch": 3.61,
-      "learning_rate": 3.728528392796995e-06,
-      "loss": 2.4161,
-      "step": 446
-    },
-    {
-      "epoch": 3.62,
-      "learning_rate": 3.707644052236887e-06,
-      "loss": 2.3116,
-      "step": 447
-    },
-    {
-      "epoch": 3.63,
-      "learning_rate": 3.6867838447368414e-06,
-      "loss": 2.159,
-      "step": 448
-    },
-    {
-      "epoch": 3.65,
-      "learning_rate": 3.6659481598339952e-06,
-      "loss": 2.2846,
-      "step": 449
-    },
-    {
-      "epoch": 3.66,
-      "learning_rate": 3.6451373866075657e-06,
-      "loss": 2.337,
-      "step": 450
-    },
-    {
-      "epoch": 3.67,
-      "learning_rate": 3.624351913671571e-06,
-      "loss": 2.3568,
-      "step": 451
-    },
-    {
-      "epoch": 3.69,
-      "learning_rate": 3.6035921291675815e-06,
-      "loss": 2.3092,
-      "step": 452
-    },
-    {
-      "epoch": 3.7,
-      "learning_rate": 3.5828584207574698e-06,
-      "loss": 2.3266,
-      "step": 453
-    },
-    {
-      "epoch": 3.71,
-      "learning_rate": 3.5621511756161686e-06,
-      "loss": 2.3007,
-      "step": 454
-    },
-    {
-      "epoch": 3.73,
-      "learning_rate": 3.54147078042445e-06,
-      "loss": 2.3106,
-      "step": 455
-    },
-    {
-      "epoch": 3.74,
-      "learning_rate": 3.520817621361693e-06,
-      "loss": 2.2919,
-      "step": 456
-    },
-    {
-      "epoch": 3.75,
-      "learning_rate": 3.500192084098677e-06,
-      "loss": 2.3618,
-      "step": 457
-    },
-    {
-      "epoch": 3.77,
-      "learning_rate": 3.4795945537903852e-06,
-      "loss": 2.2895,
-      "step": 458
-    },
-    {
-      "epoch": 3.78,
-      "learning_rate": 3.459025415068804e-06,
-      "loss": 2.3116,
-      "step": 459
-    },
-    {
-      "epoch": 3.79,
-      "learning_rate": 3.4384850520357416e-06,
-      "loss": 2.2887,
-      "step": 460
-    },
-    {
-      "epoch": 3.81,
-      "learning_rate": 3.4179738482556648e-06,
-      "loss": 2.2725,
-      "step": 461
-    },
-    {
-      "epoch": 3.82,
-      "learning_rate": 3.3974921867485238e-06,
-      "loss": 2.3094,
-      "step": 462
-    },
-    {
-      "epoch": 3.83,
-      "learning_rate": 3.377040449982604e-06,
-      "loss": 2.3289,
-      "step": 463
-    },
-    {
-      "epoch": 3.85,
-      "learning_rate": 3.356619019867394e-06,
-      "loss": 2.2469,
-      "step": 464
-    },
-    {
-      "epoch": 3.86,
-      "learning_rate": 3.336228277746435e-06,
-      "loss": 2.3668,
-      "step": 465
-    },
-    {
-      "epoch": 3.87,
-      "learning_rate": 3.3158686043902166e-06,
-      "loss": 2.4661,
-      "step": 466
-    },
-    {
-      "epoch": 3.89,
-      "learning_rate": 3.2955403799890567e-06,
-      "loss": 2.2158,
-      "step": 467
-    },
-    {
-      "epoch": 3.9,
-      "learning_rate": 3.2752439841460063e-06,
-      "loss": 2.1988,
-      "step": 468
-    },
-    {
-      "epoch": 3.91,
-      "learning_rate": 3.254979795869761e-06,
-      "loss": 2.3208,
-      "step": 469
-    },
-    {
-      "epoch": 3.93,
-      "learning_rate": 3.234748193567579e-06,
-      "loss": 2.2562,
-      "step": 470
-    },
-    {
-      "epoch": 3.94,
-      "learning_rate": 3.214549555038218e-06,
-      "loss": 2.2584,
-      "step": 471
-    },
-    {
-      "epoch": 3.95,
-      "learning_rate": 3.194384257464884e-06,
-      "loss": 2.1852,
-      "step": 472
-    },
-    {
-      "epoch": 3.97,
-      "learning_rate": 3.1742526774081822e-06,
-      "loss": 2.2364,
-      "step": 473
-    },
-    {
-      "epoch": 3.98,
-      "learning_rate": 3.154155190799084e-06,
-      "loss": 2.3356,
-      "step": 474
-    },
-    {
-      "epoch": 3.99,
-      "learning_rate": 3.1340921729319173e-06,
-      "loss": 2.2179,
-      "step": 475
-    },
-    {
-      "epoch": 4.0,
-      "learning_rate": 3.1140639984573428e-06,
-      "loss": 2.2551,
-      "step": 476
-    },
-    {
-      "epoch": 4.02,
-      "learning_rate": 3.094071041375375e-06,
-      "loss": 2.4366,
-      "step": 477
-    },
-    {
-      "epoch": 4.03,
-      "learning_rate": 3.0741136750283816e-06,
-      "loss": 2.4265,
-      "step": 478
-    },
-    {
-      "epoch": 4.04,
-      "learning_rate": 3.054192272094125e-06,
-      "loss": 2.3015,
-      "step": 479
-    },
-    {
-      "epoch": 4.06,
-      "learning_rate": 3.0343072045787956e-06,
-      "loss": 2.3268,
-      "step": 480
-    },
-    {
-      "epoch": 4.07,
-      "learning_rate": 3.0144588438100693e-06,
-      "loss": 2.2294,
-      "step": 481
-    },
-    {
-      "epoch": 4.08,
-      "learning_rate": 2.994647560430167e-06,
-      "loss": 2.3442,
-      "step": 482
-    },
-    {
-      "epoch": 4.1,
-      "learning_rate": 2.974873724388945e-06,
-      "loss": 2.2479,
-      "step": 483
-    },
-    {
-      "epoch": 4.11,
-      "learning_rate": 2.955137704936971e-06,
-      "loss": 2.1724,
-      "step": 484
-    },
-    {
-      "epoch": 4.12,
-      "learning_rate": 2.9354398706186427e-06,
-      "loss": 2.4248,
-      "step": 485
-    },
-    {
-      "epoch": 4.14,
-      "learning_rate": 2.915780589265301e-06,
-      "loss": 2.3454,
-      "step": 486
-    },
-    {
-      "epoch": 4.15,
-      "learning_rate": 2.896160227988357e-06,
-      "loss": 2.3336,
-      "step": 487
-    },
-    {
-      "epoch": 4.16,
-      "learning_rate": 2.876579153172441e-06,
-      "loss": 2.3406,
-      "step": 488
-    },
-    {
-      "epoch": 4.18,
-      "learning_rate": 2.8570377304685627e-06,
-      "loss": 2.2548,
-      "step": 489
-    },
-    {
-      "epoch": 4.19,
-      "learning_rate": 2.8375363247872756e-06,
-      "loss": 2.3472,
-      "step": 490
-    },
-    {
-      "epoch": 4.2,
-      "learning_rate": 2.8180753002918735e-06,
-      "loss": 2.3438,
-      "step": 491
-    },
-    {
-      "epoch": 4.22,
-      "learning_rate": 2.7986550203915807e-06,
-      "loss": 2.3328,
-      "step": 492
-    },
-    {
-      "epoch": 4.23,
-      "learning_rate": 2.779275847734766e-06,
-      "loss": 2.3456,
-      "step": 493
-    },
-    {
-      "epoch": 4.24,
-      "learning_rate": 2.7599381442021833e-06,
-      "loss": 2.3872,
-      "step": 494
-    },
-    {
-      "epoch": 4.26,
-      "learning_rate": 2.7406422709001956e-06,
-      "loss": 2.3587,
-      "step": 495
-    },
-    {
-      "epoch": 4.27,
-      "learning_rate": 2.721388588154045e-06,
-      "loss": 2.3089,
-      "step": 496
-    },
-    {
-      "epoch": 4.28,
-      "learning_rate": 2.7021774555011214e-06,
-      "loss": 2.303,
-      "step": 497
-    },
-    {
-      "epoch": 4.3,
-      "learning_rate": 2.6830092316842448e-06,
-      "loss": 2.2772,
-      "step": 498
-    },
-    {
-      "epoch": 4.31,
-      "learning_rate": 2.6638842746449672e-06,
-      "loss": 2.3411,
-      "step": 499
-    },
-    {
-      "epoch": 2.01,
-      "learning_rate": 2.6448029415168964e-06,
-      "loss": 2.2603,
-      "step": 500
-    },
-    {
-      "epoch": 2.03,
-      "learning_rate": 2.6257655886190147e-06,
-      "loss": 2.2749,
-      "step": 501
-    },
-    {
-      "epoch": 2.04,
-      "learning_rate": 2.6067725714490307e-06,
-      "loss": 2.2677,
-      "step": 502
-    },
-    {
-      "epoch": 2.05,
-      "learning_rate": 2.5878242446767466e-06,
-      "loss": 2.3006,
-      "step": 503
-    },
-    {
-      "epoch": 2.07,
-      "learning_rate": 2.5689209621374257e-06,
-      "loss": 2.2082,
-      "step": 504
-    },
-    {
-      "epoch": 2.08,
-      "learning_rate": 2.5500630768251895e-06,
-      "loss": 2.2563,
-      "step": 505
-    },
-    {
-      "epoch": 2.09,
-      "learning_rate": 2.5312509408864248e-06,
-      "loss": 2.3765,
-      "step": 506
-    },
-    {
-      "epoch": 2.11,
-      "learning_rate": 2.5124849056132094e-06,
-      "loss": 2.3231,
-      "step": 507
-    },
-    {
-      "epoch": 2.12,
-      "learning_rate": 2.493765321436755e-06,
-      "loss": 2.2649,
-      "step": 508
-    },
-    {
-      "epoch": 2.13,
-      "learning_rate": 2.475092537920853e-06,
-      "loss": 2.3013,
-      "step": 509
-    },
-    {
-      "epoch": 2.15,
-      "learning_rate": 2.456466903755357e-06,
-      "loss": 2.3028,
-      "step": 510
-    },
-    {
-      "epoch": 2.16,
-      "learning_rate": 2.4378887667496696e-06,
-      "loss": 2.3475,
-      "step": 511
-    },
-    {
-      "epoch": 2.17,
-      "learning_rate": 2.4193584738262426e-06,
-      "loss": 2.2435,
-      "step": 512
-    },
-    {
-      "epoch": 2.19,
-      "learning_rate": 2.400876371014107e-06,
-      "loss": 2.2967,
-      "step": 513
-    },
-    {
-      "epoch": 2.2,
-      "learning_rate": 2.3824428034424e-06,
-      "loss": 2.2928,
-      "step": 514
-    },
-    {
-      "epoch": 2.21,
-      "learning_rate": 2.3640581153339293e-06,
-      "loss": 2.3718,
-      "step": 515
-    },
-    {
-      "epoch": 2.23,
-      "learning_rate": 2.3457226499987456e-06,
-      "loss": 2.2232,
-      "step": 516
-    },
-    {
-      "epoch": 2.24,
-      "learning_rate": 2.3274367498277246e-06,
-      "loss": 2.3189,
-      "step": 517
-    },
-    {
-      "epoch": 2.25,
-      "learning_rate": 2.3092007562861756e-06,
-      "loss": 2.3843,
-      "step": 518
-    },
-    {
-      "epoch": 2.27,
-      "learning_rate": 2.291015009907474e-06,
-      "loss": 2.2295,
-      "step": 519
-    },
-    {
-      "epoch": 2.28,
-      "learning_rate": 2.2728798502866887e-06,
-      "loss": 2.2856,
-      "step": 520
-    },
-    {
-      "epoch": 2.29,
-      "learning_rate": 2.2547956160742473e-06,
-      "loss": 2.2164,
-      "step": 521
-    },
-    {
-      "epoch": 2.31,
-      "learning_rate": 2.2367626449696168e-06,
-      "loss": 2.2529,
-      "step": 522
-    },
-    {
-      "epoch": 2.32,
-      "learning_rate": 2.2187812737149856e-06,
-      "loss": 2.2924,
-      "step": 523
-    },
-    {
-      "epoch": 2.33,
-      "learning_rate": 2.2008518380889892e-06,
-      "loss": 2.3298,
-      "step": 524
-    },
-    {
-      "epoch": 2.34,
-      "learning_rate": 2.182974672900428e-06,
-      "loss": 2.2495,
-      "step": 525
-    },
-    {
-      "epoch": 2.36,
-      "learning_rate": 2.1651501119820212e-06,
-      "loss": 2.3201,
-      "step": 526
-    },
-    {
-      "epoch": 2.37,
-      "learning_rate": 2.1473784881841753e-06,
-      "loss": 2.2512,
-      "step": 527
-    },
-    {
-      "epoch": 2.38,
-      "learning_rate": 2.129660133368761e-06,
-      "loss": 2.3506,
-      "step": 528
-    },
-    {
-      "epoch": 2.4,
-      "learning_rate": 2.1119953784029207e-06,
-      "loss": 2.3026,
-      "step": 529
-    },
-    {
-      "epoch": 2.41,
-      "learning_rate": 2.0943845531528932e-06,
-      "loss": 2.3814,
-      "step": 530
-    },
-    {
-      "epoch": 2.42,
-      "learning_rate": 2.0768279864778475e-06,
-      "loss": 2.2394,
-      "step": 531
-    },
-    {
-      "epoch": 2.44,
-      "learning_rate": 2.059326006223743e-06,
-      "loss": 2.3631,
-      "step": 532
-    },
-    {
-      "epoch": 2.45,
-      "learning_rate": 2.0418789392172113e-06,
-      "loss": 2.2676,
-      "step": 533
-    },
-    {
-      "epoch": 2.46,
-      "learning_rate": 2.0244871112594523e-06,
-      "loss": 2.3143,
-      "step": 534
-    },
-    {
-      "epoch": 2.48,
-      "learning_rate": 2.007150847120145e-06,
-      "loss": 2.164,
-      "step": 535
-    },
-    {
-      "epoch": 2.49,
-      "learning_rate": 1.98987047053139e-06,
-      "loss": 2.332,
-      "step": 536
-    },
-    {
-      "epoch": 2.5,
-      "learning_rate": 1.972646304181656e-06,
-      "loss": 2.2384,
-      "step": 537
-    },
-    {
-      "epoch": 2.52,
-      "learning_rate": 1.9554786697097668e-06,
-      "loss": 2.3065,
-      "step": 538
-    },
-    {
-      "epoch": 2.53,
-      "learning_rate": 1.9383678876988797e-06,
-      "loss": 2.3406,
-      "step": 539
-    },
-    {
-      "epoch": 2.54,
-      "learning_rate": 1.921314277670509e-06,
-      "loss": 2.2071,
-      "step": 540
-    },
-    {
-      "epoch": 2.56,
-      "learning_rate": 1.9043181580785597e-06,
-      "loss": 2.4211,
-      "step": 541
-    },
-    {
-      "epoch": 2.57,
-      "learning_rate": 1.8873798463033742e-06,
-      "loss": 2.2528,
-      "step": 542
-    },
-    {
-      "epoch": 2.58,
-      "learning_rate": 1.870499658645809e-06,
-      "loss": 2.2526,
-      "step": 543
-    },
-    {
-      "epoch": 2.6,
-      "learning_rate": 1.8536779103213336e-06,
-      "loss": 2.1434,
-      "step": 544
-    },
-    {
-      "epoch": 2.61,
-      "learning_rate": 1.8369149154541333e-06,
-      "loss": 2.2012,
-      "step": 545
-    },
-    {
-      "epoch": 2.62,
-      "learning_rate": 1.8202109870712542e-06,
-      "loss": 2.2621,
-      "step": 546
-    },
-    {
-      "epoch": 2.64,
-      "learning_rate": 1.8035664370967493e-06,
-      "loss": 2.327,
-      "step": 547
-    },
-    {
-      "epoch": 2.65,
-      "learning_rate": 1.7869815763458576e-06,
-      "loss": 2.2459,
-      "step": 548
-    },
-    {
-      "epoch": 2.66,
-      "learning_rate": 1.7704567145192036e-06,
-      "loss": 2.3164,
-      "step": 549
-    },
-    {
-      "epoch": 2.68,
-      "learning_rate": 1.753992160197006e-06,
-      "loss": 2.2619,
-      "step": 550
-    },
-    {
-      "epoch": 2.69,
-      "learning_rate": 1.73758822083332e-06,
-      "loss": 2.3484,
-      "step": 551
-    },
-    {
-      "epoch": 2.7,
-      "learning_rate": 1.721245202750299e-06,
-      "loss": 2.2519,
-      "step": 552
-    },
-    {
-      "epoch": 2.72,
-      "learning_rate": 1.7049634111324687e-06,
-      "loss": 2.3655,
-      "step": 553
-    },
-    {
-      "epoch": 2.73,
-      "learning_rate": 1.6887431500210272e-06,
-      "loss": 2.4345,
-      "step": 554
-    },
-    {
-      "epoch": 2.74,
-      "learning_rate": 1.6725847223081776e-06,
-      "loss": 2.3152,
-      "step": 555
-    },
-    {
-      "epoch": 2.76,
-      "learning_rate": 1.6564884297314593e-06,
-      "loss": 2.3824,
-      "step": 556
-    },
-    {
-      "epoch": 2.77,
-      "learning_rate": 1.6404545728681232e-06,
-      "loss": 2.3621,
-      "step": 557
-    },
-    {
-      "epoch": 2.78,
-      "learning_rate": 1.624483451129512e-06,
-      "loss": 2.2405,
-      "step": 558
-    },
-    {
-      "epoch": 2.8,
-      "learning_rate": 1.6085753627554728e-06,
-      "loss": 2.2798,
-      "step": 559
-    },
-    {
-      "epoch": 2.81,
-      "learning_rate": 1.5927306048087855e-06,
-      "loss": 2.2497,
-      "step": 560
-    },
-    {
-      "epoch": 2.82,
-      "learning_rate": 1.5769494731696206e-06,
-      "loss": 2.25,
-      "step": 561
-    },
-    {
-      "epoch": 2.84,
-      "learning_rate": 1.5612322625300064e-06,
-      "loss": 2.4691,
-      "step": 562
-    },
-    {
-      "epoch": 2.85,
-      "learning_rate": 1.5455792663883329e-06,
-      "loss": 2.3415,
-      "step": 563
-    },
-    {
-      "epoch": 2.86,
-      "learning_rate": 1.529990777043866e-06,
-      "loss": 2.23,
-      "step": 564
-    },
-    {
-      "epoch": 2.88,
-      "learning_rate": 1.5144670855912908e-06,
-      "loss": 2.2422,
-      "step": 565
-    },
-    {
-      "epoch": 2.89,
-      "learning_rate": 1.499008481915281e-06,
-      "loss": 2.2791,
-      "step": 566
-    },
-    {
-      "epoch": 2.9,
-      "learning_rate": 1.483615254685075e-06,
-      "loss": 2.3216,
-      "step": 567
-    },
-    {
-      "epoch": 2.92,
-      "learning_rate": 1.4682876913490973e-06,
-      "loss": 2.4383,
-      "step": 568
-    },
-    {
-      "epoch": 2.93,
-      "learning_rate": 1.4530260781295813e-06,
-      "loss": 2.3001,
-      "step": 569
-    },
-    {
-      "epoch": 2.94,
-      "learning_rate": 1.437830700017226e-06,
-      "loss": 2.3399,
-      "step": 570
-    },
-    {
-      "epoch": 2.96,
-      "learning_rate": 1.4227018407658822e-06,
-      "loss": 2.324,
-      "step": 571
-    },
-    {
-      "epoch": 2.97,
-      "learning_rate": 1.4076397828872441e-06,
-      "loss": 2.2255,
-      "step": 572
-    },
-    {
-      "epoch": 2.98,
-      "learning_rate": 1.392644807645575e-06,
-      "loss": 2.4436,
-      "step": 573
-    },
-    {
-      "epoch": 3.0,
-      "learning_rate": 1.3777171950524648e-06,
-      "loss": 2.2523,
-      "step": 574
-    },
-    {
-      "epoch": 3.01,
-      "learning_rate": 1.3628572238615878e-06,
-      "loss": 2.3668,
-      "step": 575
-    },
-    {
-      "epoch": 3.02,
-      "learning_rate": 1.3480651715635035e-06,
-      "loss": 2.3376,
-      "step": 576
-    },
-    {
-      "epoch": 3.03,
-      "learning_rate": 1.333341314380479e-06,
-      "loss": 2.2861,
-      "step": 577
-    },
-    {
-      "epoch": 3.05,
-      "learning_rate": 1.3186859272613222e-06,
-      "loss": 2.2756,
-      "step": 578
-    },
-    {
-      "epoch": 3.06,
-      "learning_rate": 1.3040992838762562e-06,
-      "loss": 2.1302,
-      "step": 579
-    },
-    {
-      "epoch": 3.07,
-      "learning_rate": 1.2895816566118014e-06,
-      "loss": 2.4016,
-      "step": 580
-    },
-    {
-      "epoch": 3.09,
-      "learning_rate": 1.275133316565691e-06,
-      "loss": 2.2167,
-      "step": 581
-    },
-    {
-      "epoch": 3.1,
-      "learning_rate": 1.2607545335418154e-06,
-      "loss": 2.2321,
-      "step": 582
-    },
-    {
-      "epoch": 3.11,
-      "learning_rate": 1.2464455760451733e-06,
-      "loss": 2.3397,
-      "step": 583
-    },
-    {
-      "epoch": 3.13,
-      "learning_rate": 1.2322067112768632e-06,
-      "loss": 2.3304,
-      "step": 584
-    },
-    {
-      "epoch": 3.14,
-      "learning_rate": 1.2180382051290974e-06,
-      "loss": 2.3833,
-      "step": 585
-    },
-    {
-      "epoch": 3.15,
-      "learning_rate": 1.2039403221802297e-06,
-      "loss": 2.4011,
-      "step": 586
-    },
-    {
-      "epoch": 3.17,
-      "learning_rate": 1.189913325689816e-06,
-      "loss": 2.2326,
-      "step": 587
-    },
-    {
-      "epoch": 3.18,
-      "learning_rate": 1.175957477593706e-06,
-      "loss": 2.2893,
-      "step": 588
-    },
-    {
-      "epoch": 3.19,
-      "learning_rate": 1.1620730384991407e-06,
-      "loss": 2.2846,
-      "step": 589
-    },
-    {
-      "epoch": 3.21,
-      "learning_rate": 1.1482602676798933e-06,
-      "loss": 2.2759,
-      "step": 590
-    },
-    {
-      "epoch": 3.22,
-      "learning_rate": 1.1345194230714235e-06,
-      "loss": 2.2681,
-      "step": 591
-    },
-    {
-      "epoch": 3.23,
-      "learning_rate": 1.120850761266068e-06,
-      "loss": 2.2402,
-      "step": 592
-    },
-    {
-      "epoch": 3.25,
-      "learning_rate": 1.107254537508239e-06,
-      "loss": 2.2913,
-      "step": 593
-    },
-    {
-      "epoch": 3.26,
-      "learning_rate": 1.0937310056896643e-06,
-      "loss": 2.3218,
-      "step": 594
-    },
-    {
-      "epoch": 3.27,
-      "learning_rate": 1.080280418344643e-06,
-      "loss": 2.2079,
-      "step": 595
-    },
-    {
-      "epoch": 3.29,
-      "learning_rate": 1.0669030266453367e-06,
-      "loss": 2.2507,
-      "step": 596
-    },
-    {
-      "epoch": 3.3,
-      "learning_rate": 1.053599080397068e-06,
-      "loss": 2.3211,
-      "step": 597
-    },
-    {
-      "epoch": 3.31,
-      "learning_rate": 1.0403688280336626e-06,
-      "loss": 2.3367,
-      "step": 598
-    },
-    {
-      "epoch": 3.33,
-      "learning_rate": 1.027212516612814e-06,
-      "loss": 2.2297,
-      "step": 599
-    },
-    {
-      "epoch": 3.34,
-      "learning_rate": 1.014130391811457e-06,
-      "loss": 2.3259,
-      "step": 600
-    },
-    {
-      "epoch": 3.35,
-      "learning_rate": 1.001122697921197e-06,
-      "loss": 2.2548,
-      "step": 601
-    },
-    {
-      "epoch": 3.37,
-      "learning_rate": 9.881896778437328e-07,
-      "loss": 2.2746,
-      "step": 602
-    },
-    {
-      "epoch": 3.38,
-      "learning_rate": 9.753315730863284e-07,
-      "loss": 2.3727,
-      "step": 603
-    },
-    {
-      "epoch": 3.39,
-      "learning_rate": 9.625486237573046e-07,
-      "loss": 2.2859,
-      "step": 604
-    },
-    {
-      "epoch": 3.41,
-      "learning_rate": 9.498410685615511e-07,
-      "loss": 2.2621,
-      "step": 605
-    },
-    {
-      "epoch": 3.42,
-      "learning_rate": 9.372091447960685e-07,
-      "loss": 2.3176,
-      "step": 606
-    },
-    {
-      "epoch": 3.43,
-      "learning_rate": 9.24653088345544e-07,
-      "loss": 2.4216,
-      "step": 607
-    },
-    {
-      "epoch": 3.45,
-      "learning_rate": 9.121731336779377e-07,
-      "loss": 2.261,
-      "step": 608
-    },
-    {
-      "epoch": 3.46,
-      "learning_rate": 8.99769513840108e-07,
-      "loss": 2.3639,
-      "step": 609
-    },
-    {
-      "epoch": 3.47,
-      "learning_rate": 8.874424604534643e-07,
-      "loss": 2.2011,
-      "step": 610
-    },
-    {
-      "epoch": 3.49,
-      "learning_rate": 8.751922037096328e-07,
-      "loss": 2.3312,
-      "step": 611
-    },
-    {
-      "epoch": 3.5,
-      "learning_rate": 8.630189723661663e-07,
-      "loss": 2.2176,
-      "step": 612
-    },
-    {
-      "epoch": 3.51,
-      "learning_rate": 8.509229937422664e-07,
-      "loss": 2.1899,
-      "step": 613
-    },
-    {
-      "epoch": 3.53,
-      "learning_rate": 8.389044937145397e-07,
-      "loss": 2.2521,
-      "step": 614
-    },
-    {
-      "epoch": 3.54,
-      "learning_rate": 8.269636967127864e-07,
-      "loss": 2.4843,
-      "step": 615
-    },
-    {
-      "epoch": 3.55,
-      "learning_rate": 8.151008257158e-07,
-      "loss": 2.1477,
-      "step": 616
-    },
-    {
-      "epoch": 3.57,
-      "learning_rate": 8.033161022472063e-07,
-      "loss": 2.3578,
-      "step": 617
-    },
-    {
-      "epoch": 3.58,
-      "learning_rate": 7.916097463713335e-07,
-      "loss": 2.3254,
-      "step": 618
-    },
-    {
-      "epoch": 3.59,
-      "learning_rate": 7.799819766890926e-07,
-      "loss": 2.3148,
-      "step": 619
-    },
-    {
-      "epoch": 3.61,
-      "learning_rate": 7.684330103339016e-07,
-      "loss": 2.2215,
-      "step": 620
-    },
-    {
-      "epoch": 3.62,
-      "learning_rate": 7.569630629676294e-07,
-      "loss": 2.2728,
-      "step": 621
-    },
-    {
-      "epoch": 3.63,
-      "learning_rate": 7.455723487765664e-07,
-      "loss": 2.3129,
-      "step": 622
-    },
-    {
-      "epoch": 3.65,
-      "learning_rate": 7.342610804674316e-07,
-      "loss": 2.3301,
-      "step": 623
-    },
-    {
-      "epoch": 3.66,
-      "learning_rate": 7.230294692633922e-07,
-      "loss": 2.2177,
-      "step": 624
-    },
-    {
-      "epoch": 3.67,
-      "learning_rate": 7.118777249001213e-07,
-      "loss": 2.2484,
-      "step": 625
-    },
-    {
-      "epoch": 3.68,
-      "learning_rate": 7.008060556218893e-07,
-      "loss": 2.3895,
-      "step": 626
-    },
-    {
-      "epoch": 3.7,
-      "learning_rate": 6.898146681776629e-07,
-      "loss": 2.2685,
-      "step": 627
-    },
-    {
-      "epoch": 3.71,
-      "learning_rate": 6.789037678172522e-07,
-      "loss": 2.2515,
-      "step": 628
-    },
-    {
-      "epoch": 3.72,
-      "learning_rate": 6.680735582874781e-07,
-      "loss": 2.2984,
-      "step": 629
-    },
-    {
-      "epoch": 3.74,
-      "learning_rate": 6.573242418283632e-07,
-      "loss": 2.3604,
-      "step": 630
-    },
-    {
-      "epoch": 3.75,
-      "learning_rate": 6.466560191693566e-07,
-      "loss": 2.2527,
-      "step": 631
-    },
-    {
-      "epoch": 3.76,
-      "learning_rate": 6.360690895255916e-07,
-      "loss": 2.187,
-      "step": 632
-    },
-    {
-      "epoch": 3.78,
-      "learning_rate": 6.255636505941548e-07,
-      "loss": 2.3555,
-      "step": 633
-    },
-    {
-      "epoch": 3.79,
-      "learning_rate": 6.151398985504043e-07,
-      "loss": 2.3284,
-      "step": 634
-    },
-    {
-      "epoch": 3.8,
-      "learning_rate": 6.047980280443e-07,
-      "loss": 2.2503,
-      "step": 635
-    },
-    {
-      "epoch": 3.82,
-      "learning_rate": 5.945382321967696e-07,
-      "loss": 2.2688,
-      "step": 636
-    },
-    {
-      "epoch": 3.83,
-      "learning_rate": 5.84360702596109e-07,
-      "loss": 2.2249,
-      "step": 637
-    },
-    {
-      "epoch": 3.84,
-      "learning_rate": 5.742656292943943e-07,
-      "loss": 2.2609,
-      "step": 638
-    },
-    {
-      "epoch": 3.86,
-      "learning_rate": 5.642532008039392e-07,
-      "loss": 2.2697,
-      "step": 639
-    },
-    {
-      "epoch": 3.87,
-      "learning_rate": 5.543236040937744e-07,
-      "loss": 2.3562,
-      "step": 640
-    },
-    {
-      "epoch": 3.88,
-      "learning_rate": 5.444770245861553e-07,
-      "loss": 2.1866,
-      "step": 641
-    },
-    {
-      "epoch": 3.9,
-      "learning_rate": 5.347136461530966e-07,
-      "loss": 2.2249,
-      "step": 642
-    },
-    {
-      "epoch": 3.91,
-      "learning_rate": 5.250336511129462e-07,
-      "loss": 2.2293,
-      "step": 643
-    },
-    {
-      "epoch": 3.92,
-      "learning_rate": 5.15437220226972e-07,
-      "loss": 2.2144,
-      "step": 644
-    },
-    {
-      "epoch": 3.94,
-      "learning_rate": 5.059245326959927e-07,
-      "loss": 2.2897,
-      "step": 645
-    },
-    {
-      "epoch": 3.95,
-      "learning_rate": 4.964957661570285e-07,
-      "loss": 2.2324,
-      "step": 646
-    },
-    {
-      "epoch": 3.96,
-      "learning_rate": 4.871510966799847e-07,
-      "loss": 2.3129,
-      "step": 647
-    },
-    {
-      "epoch": 3.98,
-      "learning_rate": 4.778906987643633e-07,
-      "loss": 2.3131,
-      "step": 648
-    },
-    {
-      "epoch": 3.99,
-      "learning_rate": 4.6871474533600413e-07,
-      "loss": 2.2208,
-      "step": 649
-    },
-    {
-      "epoch": 4.0,
-      "learning_rate": 4.5962340774385936e-07,
-      "loss": 2.3414,
-      "step": 650
-    },
-    {
-      "epoch": 4.02,
-      "learning_rate": 4.506168557567886e-07,
-      "loss": 2.2689,
-      "step": 651
-    },
-    {
-      "epoch": 4.03,
-      "learning_rate": 4.4169525756039164e-07,
-      "loss": 2.2039,
-      "step": 652
-    },
-    {
-      "epoch": 4.04,
-      "learning_rate": 4.328587797538658e-07,
-      "loss": 2.311,
-      "step": 653
-    },
-    {
-      "epoch": 4.06,
-      "learning_rate": 4.2410758734689915e-07,
-      "loss": 2.1353,
-      "step": 654
-    },
-    {
-      "epoch": 4.07,
-      "learning_rate": 4.1544184375658326e-07,
-      "loss": 2.2208,
-      "step": 655
-    },
-    {
-      "epoch": 4.08,
-      "learning_rate": 4.0686171080436767e-07,
-      "loss": 2.2835,
-      "step": 656
-    },
-    {
-      "epoch": 4.1,
-      "learning_rate": 3.983673487130313e-07,
-      "loss": 2.2034,
-      "step": 657
-    },
-    {
-      "epoch": 4.11,
-      "learning_rate": 3.8995891610369707e-07,
-      "loss": 2.2443,
-      "step": 658
-    },
-    {
-      "epoch": 4.12,
-      "learning_rate": 3.8163656999286647e-07,
-      "loss": 2.2348,
-      "step": 659
-    },
-    {
-      "epoch": 4.14,
-      "learning_rate": 3.734004657894874e-07,
-      "loss": 2.2864,
-      "step": 660
-    },
-    {
-      "epoch": 4.15,
-      "learning_rate": 3.6525075729205274e-07,
-      "loss": 2.22,
-      "step": 661
-    },
-    {
-      "epoch": 4.16,
-      "learning_rate": 3.5718759668572913e-07,
-      "loss": 2.2842,
-      "step": 662
-    },
-    {
-      "epoch": 4.18,
-      "learning_rate": 3.4921113453951385e-07,
-      "loss": 2.2445,
-      "step": 663
-    },
-    {
-      "epoch": 4.19,
-      "learning_rate": 3.4132151980342255e-07,
-      "loss": 2.3406,
-      "step": 664
-    },
-    {
-      "epoch": 4.2,
-      "learning_rate": 3.335188998057115e-07,
-      "loss": 2.1633,
-      "step": 665
-    },
-    {
-      "epoch": 4.22,
-      "learning_rate": 3.2580342025012204e-07,
-      "loss": 2.3446,
-      "step": 666
-    },
-    {
-      "epoch": 4.23,
-      "learning_rate": 3.1817522521316034e-07,
-      "loss": 2.2757,
-      "step": 667
-    },
-    {
-      "epoch": 4.24,
-      "learning_rate": 3.106344571414116e-07,
-      "loss": 2.3542,
-      "step": 668
-    },
-    {
-      "epoch": 4.26,
-      "learning_rate": 3.0318125684887233e-07,
-      "loss": 2.2571,
-      "step": 669
-    },
-    {
-      "epoch": 4.27,
-      "learning_rate": 2.958157635143294e-07,
-      "loss": 2.3593,
-      "step": 670
-    },
-    {
-      "epoch": 4.28,
-      "learning_rate": 2.8853811467875413e-07,
-      "loss": 2.2509,
-      "step": 671
-    },
-    {
-      "epoch": 4.3,
-      "learning_rate": 2.813484462427357e-07,
-      "loss": 2.2306,
-      "step": 672
-    },
-    {
-      "epoch": 4.31,
-      "learning_rate": 2.7424689246394685e-07,
-      "loss": 2.3121,
-      "step": 673
-    },
-    {
-      "epoch": 4.32,
-      "learning_rate": 2.672335859546332e-07,
-      "loss": 2.3072,
-      "step": 674
-    },
-    {
-      "epoch": 4.33,
-      "learning_rate": 2.6030865767913527e-07,
-      "loss": 2.171,
-      "step": 675
-    },
-    {
-      "epoch": 4.35,
-      "learning_rate": 2.534722369514503e-07,
-      "loss": 2.2241,
-      "step": 676
-    },
-    {
-      "epoch": 4.36,
-      "learning_rate": 2.467244514328082e-07,
-      "loss": 2.3027,
-      "step": 677
-    },
-    {
-      "epoch": 4.37,
-      "learning_rate": 2.400654271292946e-07,
-      "loss": 2.2732,
-      "step": 678
-    },
-    {
-      "epoch": 4.39,
-      "learning_rate": 2.334952883894942e-07,
-      "loss": 2.2501,
-      "step": 679
-    },
-    {
-      "epoch": 4.4,
-      "learning_rate": 2.270141579021695e-07,
-      "loss": 2.4042,
-      "step": 680
-    },
-    {
-      "epoch": 4.41,
-      "learning_rate": 2.2062215669397201e-07,
-      "loss": 2.2304,
-      "step": 681
-    },
-    {
-      "epoch": 4.43,
-      "learning_rate": 2.1431940412717843e-07,
-      "loss": 2.2965,
-      "step": 682
-    },
-    {
-      "epoch": 4.44,
-      "learning_rate": 2.081060178974642e-07,
-      "loss": 2.2902,
-      "step": 683
-    },
-    {
-      "epoch": 4.45,
-      "learning_rate": 2.019821140317052e-07,
-      "loss": 2.3338,
-      "step": 684
-    },
-    {
-      "epoch": 4.47,
-      "learning_rate": 1.9594780688581172e-07,
-      "loss": 2.2142,
-      "step": 685
-    },
-    {
-      "epoch": 4.48,
-      "learning_rate": 1.900032091425902e-07,
-      "loss": 2.4194,
-      "step": 686
-    },
-    {
-      "epoch": 4.49,
-      "learning_rate": 1.8414843180964316e-07,
-      "loss": 2.2878,
-      "step": 687
-    },
-    {
-      "epoch": 4.51,
-      "learning_rate": 1.7838358421729375e-07,
-      "loss": 2.2304,
-      "step": 688
-    },
-    {
-      "epoch": 4.52,
-      "learning_rate": 1.7270877401654283e-07,
-      "loss": 2.3066,
-      "step": 689
-    },
-    {
-      "epoch": 4.53,
-      "learning_rate": 1.6712410717706406e-07,
-      "loss": 2.2243,
-      "step": 690
-    },
-    {
-      "epoch": 4.55,
-      "learning_rate": 1.616296879852175e-07,
-      "loss": 2.2178,
-      "step": 691
-    },
-    {
-      "epoch": 4.56,
-      "learning_rate": 1.562256190421102e-07,
-      "loss": 2.2442,
-      "step": 692
-    },
-    {
-      "epoch": 4.57,
-      "learning_rate": 1.5091200126167328e-07,
-      "loss": 2.2656,
-      "step": 693
-    },
-    {
-      "epoch": 4.59,
-      "learning_rate": 1.4568893386878057e-07,
-      "loss": 2.2923,
-      "step": 694
-    },
-    {
-      "epoch": 4.6,
-      "learning_rate": 1.405565143973986e-07,
-      "loss": 2.486,
-      "step": 695
-    },
-    {
-      "epoch": 4.61,
-      "learning_rate": 1.3551483868875836e-07,
-      "loss": 2.2223,
-      "step": 696
-    },
-    {
-      "epoch": 4.63,
-      "learning_rate": 1.30564000889572e-07,
-      "loss": 2.267,
-      "step": 697
-    },
-    {
-      "epoch": 4.64,
-      "learning_rate": 1.257040934502729e-07,
-      "loss": 2.2644,
-      "step": 698
-    },
-    {
-      "epoch": 4.65,
-      "learning_rate": 1.209352071232861e-07,
-      "loss": 2.2235,
-      "step": 699
-    },
-    {
-      "epoch": 4.67,
-      "learning_rate": 1.162574309613379e-07,
-      "loss": 2.2914,
-      "step": 700
-    },
-    {
-      "epoch": 4.68,
-      "learning_rate": 1.1167085231579111e-07,
-      "loss": 2.1447,
-      "step": 701
-    },
-    {
-      "epoch": 4.69,
-      "learning_rate": 1.0717555683501413e-07,
-      "loss": 2.1958,
-      "step": 702
-    },
-    {
-      "epoch": 4.71,
-      "learning_rate": 1.027716284627811e-07,
-      "loss": 2.3153,
-      "step": 703
-    },
-    {
-      "epoch": 4.72,
-      "learning_rate": 9.845914943670432e-08,
-      "loss": 2.2076,
-      "step": 704
-    },
-    {
-      "epoch": 4.73,
-      "learning_rate": 9.423820028669983e-08,
-      "loss": 2.2937,
-      "step": 705
-    },
-    {
-      "epoch": 4.75,
-      "learning_rate": 9.010885983348094e-08,
-      "loss": 2.3275,
-      "step": 706
-    },
-    {
-      "epoch": 4.76,
-      "learning_rate": 8.607120518709156e-08,
-      "loss": 2.2495,
-      "step": 707
-    },
-    {
-      "epoch": 4.77,
-      "learning_rate": 8.212531174545957e-08,
-      "loss": 2.3742,
-      "step": 708
-    },
-    {
-      "epoch": 4.79,
-      "learning_rate": 7.827125319299301e-08,
-      "loss": 2.3173,
-      "step": 709
-    },
-    {
-      "epoch": 4.8,
-      "learning_rate": 7.450910149920499e-08,
-      "loss": 2.3841,
-      "step": 710
-    },
-    {
-      "epoch": 4.81,
-      "learning_rate": 7.083892691736428e-08,
-      "loss": 2.3203,
-      "step": 711
-    },
-    {
-      "epoch": 4.83,
-      "learning_rate": 6.726079798319185e-08,
-      "loss": 2.372,
-      "step": 712
-    },
-    {
-      "epoch": 4.84,
-      "learning_rate": 6.377478151357308e-08,
-      "loss": 2.2576,
-      "step": 713
-    },
-    {
-      "epoch": 4.85,
-      "learning_rate": 6.038094260531425e-08,
-      "loss": 2.2702,
-      "step": 714
-    },
-    {
-      "epoch": 4.87,
-      "learning_rate": 5.707934463392628e-08,
-      "loss": 2.2772,
-      "step": 715
-    },
-    {
-      "epoch": 4.88,
-      "learning_rate": 5.387004925244077e-08,
-      "loss": 2.2446,
-      "step": 716
-    },
-    {
-      "epoch": 4.89,
-      "learning_rate": 5.0753116390258594e-08,
-      "loss": 2.3236,
-      "step": 717
-    },
-    {
-      "epoch": 4.91,
-      "learning_rate": 4.772860425203252e-08,
-      "loss": 2.2049,
-      "step": 718
-    },
-    {
-      "epoch": 4.92,
-      "learning_rate": 4.479656931657694e-08,
-      "loss": 2.3853,
-      "step": 719
-    },
-    {
-      "epoch": 4.93,
-      "learning_rate": 4.195706633581709e-08,
-      "loss": 2.3235,
-      "step": 720
-    },
-    {
-      "epoch": 4.95,
-      "learning_rate": 3.9210148333763135e-08,
-      "loss": 2.2853,
-      "step": 721
-    },
-    {
-      "epoch": 4.96,
-      "learning_rate": 3.655586660552324e-08,
-      "loss": 2.2421,
-      "step": 722
-    },
-    {
-      "epoch": 4.97,
-      "learning_rate": 3.39942707163432e-08,
-      "loss": 2.3028,
-      "step": 723
-    },
-    {
-      "epoch": 4.99,
-      "learning_rate": 3.152540850068164e-08,
-      "loss": 2.3869,
-      "step": 724
-    },
-    {
-      "epoch": 5.0,
-      "learning_rate": 2.9149326061317373e-08,
-      "loss": 2.3606,
-      "step": 725
-    },
-    {
-      "epoch": 5.01,
-      "learning_rate": 2.686606776848788e-08,
-      "loss": 2.1782,
-      "step": 726
-    },
-    {
-      "epoch": 5.02,
-      "learning_rate": 2.4675676259059976e-08,
-      "loss": 2.3123,
-      "step": 727
-    },
-    {
-      "epoch": 5.04,
-      "learning_rate": 2.2578192435736555e-08,
-      "loss": 2.2398,
-      "step": 728
-    },
-    {
-      "epoch": 5.05,
-      "learning_rate": 2.0573655466289423e-08,
-      "loss": 2.3347,
-      "step": 729
-    },
-    {
-      "epoch": 5.06,
-      "learning_rate": 1.866210278282876e-08,
-      "loss": 2.2827,
-      "step": 730
-    },
-    {
-      "epoch": 5.08,
-      "learning_rate": 1.684357008110593e-08,
-      "loss": 2.2904,
-      "step": 731
-    },
-    {
-      "epoch": 5.09,
-      "learning_rate": 1.5118091319843985e-08,
-      "loss": 2.2559,
-      "step": 732
-    },
-    {
-      "epoch": 5.1,
-      "learning_rate": 1.3485698720107077e-08,
-      "loss": 2.3674,
-      "step": 733
-    },
-    {
-      "epoch": 5.12,
-      "learning_rate": 1.1946422764695376e-08,
-      "loss": 2.257,
-      "step": 734
-    },
-    {
-      "epoch": 5.13,
-      "learning_rate": 1.0500292197577756e-08,
-      "loss": 2.343,
-      "step": 735
-    },
-    {
-      "epoch": 5.14,
-      "learning_rate": 9.147334023354437e-09,
-      "loss": 2.1531,
-      "step": 736
-    },
-    {
-      "epoch": 5.16,
-      "learning_rate": 7.887573506752954e-09,
-      "loss": 2.3412,
-      "step": 737
-    },
-    {
-      "epoch": 5.17,
-      "learning_rate": 6.7210341721563044e-09,
-      "loss": 2.1971,
-      "step": 738
-    },
-    {
-      "epoch": 5.18,
-      "learning_rate": 5.647737803163855e-09,
-      "loss": 2.1668,
-      "step": 739
-    },
-    {
-      "epoch": 5.2,
-      "learning_rate": 4.667704442183341e-09,
-      "loss": 2.2644,
-      "step": 740
-    },
-    {
-      "epoch": 5.21,
-      "learning_rate": 3.780952390058379e-09,
-      "loss": 2.2436,
-      "step": 741
     }
   ],
-  "max_steps": 750,
+  "max_steps": 370,
   "num_train_epochs": 10,
-  "total_flos": 1.195119998926848e+16,
+  "total_flos": 1.3814061758152704e+16,
   "trial_name": null,
   "trial_params": null
 }