diff --git "a/checkpoint-3048/trainer_state.json" "b/checkpoint-3048/trainer_state.json"
--- "a/checkpoint-3048/trainer_state.json"
+++ "b/checkpoint-3048/trainer_state.json"
@@ -1,7 +1,7 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 2.998031496062992,
+  "epoch": 3.0,
   "eval_steps": 153,
   "global_step": 3048,
   "is_hyper_param_search": false,
@@ -12,23027 +12,23027 @@
       "epoch": 0.000984251968503937,
       "grad_norm": NaN,
       "learning_rate": 0.0,
-      "loss": 5.5525,
+      "loss": 18.7427,
       "step": 1
     },
     {
       "epoch": 0.001968503937007874,
       "grad_norm": NaN,
       "learning_rate": 0.0,
-      "loss": 6.2948,
+      "loss": 11.6434,
       "step": 2
     },
     {
       "epoch": 0.002952755905511811,
-      "grad_norm": 27.109525680541992,
-      "learning_rate": 9.940357852882705e-10,
-      "loss": 4.6464,
+      "grad_norm": Infinity,
+      "learning_rate": 0.0,
+      "loss": 7.4859,
       "step": 3
     },
     {
       "epoch": 0.003937007874015748,
-      "grad_norm": Infinity,
+      "grad_norm": 31.9741268157959,
       "learning_rate": 9.940357852882705e-10,
-      "loss": 5.0071,
+      "loss": 7.3779,
       "step": 4
     },
     {
       "epoch": 0.004921259842519685,
-      "grad_norm": 23.50518798828125,
-      "learning_rate": 1.988071570576541e-09,
-      "loss": 5.3338,
+      "grad_norm": Infinity,
+      "learning_rate": 9.940357852882705e-10,
+      "loss": 17.5878,
       "step": 5
     },
     {
       "epoch": 0.005905511811023622,
-      "grad_norm": 37.24928665161133,
-      "learning_rate": 2.9821073558648116e-09,
-      "loss": 5.1998,
+      "grad_norm": 40.571693420410156,
+      "learning_rate": 1.988071570576541e-09,
+      "loss": 8.4984,
       "step": 6
     },
     {
       "epoch": 0.006889763779527559,
-      "grad_norm": Infinity,
+      "grad_norm": 44.114723205566406,
       "learning_rate": 2.9821073558648116e-09,
-      "loss": 26.7171,
+      "loss": 8.375,
       "step": 7
     },
     {
       "epoch": 0.007874015748031496,
-      "grad_norm": 45.332584381103516,
+      "grad_norm": 39.95664978027344,
       "learning_rate": 3.976143141153082e-09,
-      "loss": 6.8404,
+      "loss": 7.3241,
       "step": 8
     },
     {
       "epoch": 0.008858267716535433,
-      "grad_norm": Infinity,
-      "learning_rate": 3.976143141153082e-09,
-      "loss": 26.8068,
+      "grad_norm": 54.75717544555664,
+      "learning_rate": 4.970178926441353e-09,
+      "loss": 10.3081,
       "step": 9
     },
     {
       "epoch": 0.00984251968503937,
-      "grad_norm": 37.57707595825195,
-      "learning_rate": 4.970178926441353e-09,
-      "loss": 5.7224,
+      "grad_norm": 48.129844665527344,
+      "learning_rate": 5.964214711729623e-09,
+      "loss": 8.5363,
       "step": 10
     },
     {
       "epoch": 0.010826771653543307,
-      "grad_norm": 38.55592727661133,
+      "grad_norm": Infinity,
       "learning_rate": 5.964214711729623e-09,
-      "loss": 5.9428,
+      "loss": 17.2241,
       "step": 11
     },
     {
       "epoch": 0.011811023622047244,
-      "grad_norm": 44.171085357666016,
+      "grad_norm": 37.79861068725586,
       "learning_rate": 6.9582504970178946e-09,
-      "loss": 6.7852,
+      "loss": 7.575,
       "step": 12
     },
     {
       "epoch": 0.012795275590551181,
-      "grad_norm": 33.0239372253418,
+      "grad_norm": 37.935638427734375,
       "learning_rate": 7.952286282306164e-09,
-      "loss": 6.505,
+      "loss": 9.1905,
       "step": 13
     },
     {
       "epoch": 0.013779527559055118,
-      "grad_norm": 29.830385208129883,
+      "grad_norm": 59.75738525390625,
       "learning_rate": 8.946322067594435e-09,
-      "loss": 5.4793,
+      "loss": 11.7727,
       "step": 14
     },
     {
       "epoch": 0.014763779527559055,
-      "grad_norm": 37.74555206298828,
+      "grad_norm": 46.43831253051758,
       "learning_rate": 9.940357852882705e-09,
-      "loss": 5.8316,
+      "loss": 9.5827,
       "step": 15
     },
     {
       "epoch": 0.015748031496062992,
-      "grad_norm": 26.375228881835938,
+      "grad_norm": 36.75105667114258,
       "learning_rate": 1.0934393638170978e-08,
-      "loss": 4.7716,
+      "loss": 7.4432,
       "step": 16
     },
     {
       "epoch": 0.01673228346456693,
-      "grad_norm": 27.332616806030273,
+      "grad_norm": 30.298437118530273,
       "learning_rate": 1.1928429423459246e-08,
-      "loss": 4.9198,
+      "loss": 7.1573,
       "step": 17
     },
     {
       "epoch": 0.017716535433070866,
-      "grad_norm": 19.702518463134766,
+      "grad_norm": 96.71277618408203,
       "learning_rate": 1.2922465208747517e-08,
-      "loss": 4.9205,
+      "loss": 19.8016,
       "step": 18
     },
     {
       "epoch": 0.018700787401574805,
-      "grad_norm": 37.775146484375,
+      "grad_norm": 84.41168212890625,
       "learning_rate": 1.3916500994035789e-08,
-      "loss": 6.1949,
+      "loss": 19.5118,
       "step": 19
     },
     {
       "epoch": 0.01968503937007874,
-      "grad_norm": 41.4459342956543,
+      "grad_norm": 35.94807815551758,
       "learning_rate": 1.4910536779324056e-08,
-      "loss": 6.935,
+      "loss": 7.9062,
       "step": 20
     },
     {
       "epoch": 0.02066929133858268,
-      "grad_norm": 69.52754974365234,
+      "grad_norm": 41.35914611816406,
       "learning_rate": 1.590457256461233e-08,
-      "loss": 26.2611,
+      "loss": 8.6791,
       "step": 21
     },
     {
       "epoch": 0.021653543307086614,
-      "grad_norm": 39.509559631347656,
+      "grad_norm": 37.65285873413086,
       "learning_rate": 1.68986083499006e-08,
-      "loss": 6.2476,
+      "loss": 7.7318,
       "step": 22
     },
     {
       "epoch": 0.022637795275590553,
-      "grad_norm": 32.32150650024414,
+      "grad_norm": 40.24616622924805,
       "learning_rate": 1.789264413518887e-08,
-      "loss": 5.1881,
+      "loss": 7.9319,
       "step": 23
     },
     {
       "epoch": 0.023622047244094488,
-      "grad_norm": 28.91634178161621,
+      "grad_norm": 32.05257034301758,
       "learning_rate": 1.888667992047714e-08,
-      "loss": 4.9198,
+      "loss": 7.192,
       "step": 24
     },
     {
       "epoch": 0.024606299212598427,
-      "grad_norm": 70.45647430419922,
+      "grad_norm": 87.60692596435547,
       "learning_rate": 1.988071570576541e-08,
-      "loss": 25.6461,
+      "loss": 15.5799,
       "step": 25
     },
     {
       "epoch": 0.025590551181102362,
-      "grad_norm": 69.56190490722656,
+      "grad_norm": 46.73486328125,
       "learning_rate": 2.087475149105368e-08,
-      "loss": 25.9935,
+      "loss": 9.7859,
       "step": 26
     },
     {
       "epoch": 0.0265748031496063,
-      "grad_norm": 29.483633041381836,
+      "grad_norm": 50.18130111694336,
       "learning_rate": 2.1868787276341955e-08,
-      "loss": 6.194,
+      "loss": 9.9259,
       "step": 27
     },
     {
       "epoch": 0.027559055118110236,
-      "grad_norm": 64.40437316894531,
+      "grad_norm": 29.22075653076172,
       "learning_rate": 2.2862823061630224e-08,
-      "loss": 24.9355,
+      "loss": 6.3076,
       "step": 28
     },
     {
       "epoch": 0.028543307086614175,
-      "grad_norm": 29.5191650390625,
+      "grad_norm": 37.356693267822266,
       "learning_rate": 2.3856858846918493e-08,
-      "loss": 5.2458,
+      "loss": 7.4471,
       "step": 29
     },
     {
       "epoch": 0.02952755905511811,
-      "grad_norm": 23.680418014526367,
+      "grad_norm": 35.80938720703125,
       "learning_rate": 2.4850894632206765e-08,
-      "loss": 5.487,
+      "loss": 7.1246,
       "step": 30
     },
     {
       "epoch": 0.03051181102362205,
-      "grad_norm": 20.74715805053711,
+      "grad_norm": 25.738622665405273,
       "learning_rate": 2.5844930417495034e-08,
-      "loss": 4.3883,
+      "loss": 6.5505,
       "step": 31
     },
     {
       "epoch": 0.031496062992125984,
-      "grad_norm": 37.96383285522461,
+      "grad_norm": 83.83589935302734,
       "learning_rate": 2.6838966202783303e-08,
-      "loss": 6.4174,
+      "loss": 18.5194,
       "step": 32
     },
     {
       "epoch": 0.03248031496062992,
-      "grad_norm": 27.312122344970703,
+      "grad_norm": 33.8433837890625,
       "learning_rate": 2.7833001988071578e-08,
-      "loss": 5.9618,
+      "loss": 7.0747,
       "step": 33
     },
     {
       "epoch": 0.03346456692913386,
-      "grad_norm": 27.779197692871094,
+      "grad_norm": 74.36174011230469,
       "learning_rate": 2.8827037773359847e-08,
-      "loss": 5.1205,
+      "loss": 14.9456,
       "step": 34
     },
     {
       "epoch": 0.0344488188976378,
-      "grad_norm": 29.39383316040039,
+      "grad_norm": 26.985536575317383,
       "learning_rate": 2.982107355864811e-08,
-      "loss": 5.6965,
+      "loss": 6.608,
       "step": 35
     },
     {
       "epoch": 0.03543307086614173,
-      "grad_norm": 62.016685485839844,
+      "grad_norm": 40.41023635864258,
       "learning_rate": 3.081510934393639e-08,
-      "loss": 23.4505,
+      "loss": 8.4672,
       "step": 36
     },
     {
       "epoch": 0.03641732283464567,
-      "grad_norm": 27.892751693725586,
+      "grad_norm": 33.05155944824219,
       "learning_rate": 3.180914512922466e-08,
-      "loss": 5.2556,
+      "loss": 6.8853,
       "step": 37
     },
     {
       "epoch": 0.03740157480314961,
-      "grad_norm": 24.589818954467773,
+      "grad_norm": 74.3410873413086,
       "learning_rate": 3.280318091451293e-08,
-      "loss": 4.9034,
+      "loss": 13.6063,
       "step": 38
     },
     {
       "epoch": 0.038385826771653545,
-      "grad_norm": 59.737709045410156,
+      "grad_norm": 35.997493743896484,
       "learning_rate": 3.37972166998012e-08,
-      "loss": 22.9819,
+      "loss": 7.2625,
       "step": 39
     },
     {
       "epoch": 0.03937007874015748,
-      "grad_norm": 20.44304847717285,
+      "grad_norm": 31.67265510559082,
       "learning_rate": 3.479125248508947e-08,
-      "loss": 4.5872,
+      "loss": 6.2234,
       "step": 40
     },
     {
       "epoch": 0.040354330708661415,
-      "grad_norm": 19.250782012939453,
+      "grad_norm": 78.0589599609375,
       "learning_rate": 3.578528827037774e-08,
-      "loss": 5.0072,
+      "loss": 14.9675,
       "step": 41
     },
     {
       "epoch": 0.04133858267716536,
-      "grad_norm": 18.789878845214844,
+      "grad_norm": 27.73711395263672,
       "learning_rate": 3.6779324055666005e-08,
-      "loss": 4.3917,
+      "loss": 6.6038,
       "step": 42
     },
     {
       "epoch": 0.04232283464566929,
-      "grad_norm": 15.848797798156738,
+      "grad_norm": 61.068092346191406,
       "learning_rate": 3.777335984095428e-08,
-      "loss": 4.0698,
+      "loss": 13.1173,
       "step": 43
     },
     {
       "epoch": 0.04330708661417323,
-      "grad_norm": 20.03360939025879,
+      "grad_norm": 84.73564910888672,
       "learning_rate": 3.8767395626242556e-08,
-      "loss": 4.9849,
+      "loss": 16.6992,
       "step": 44
     },
     {
       "epoch": 0.04429133858267716,
-      "grad_norm": 23.202022552490234,
+      "grad_norm": 27.289846420288086,
       "learning_rate": 3.976143141153082e-08,
-      "loss": 4.9534,
+      "loss": 6.4828,
       "step": 45
     },
     {
       "epoch": 0.045275590551181105,
-      "grad_norm": 18.500751495361328,
+      "grad_norm": 25.921550750732422,
       "learning_rate": 4.0755467196819094e-08,
-      "loss": 4.7506,
+      "loss": 5.9815,
       "step": 46
     },
     {
       "epoch": 0.04625984251968504,
-      "grad_norm": 15.16248607635498,
+      "grad_norm": 28.005834579467773,
       "learning_rate": 4.174950298210736e-08,
-      "loss": 4.6928,
+      "loss": 6.1738,
       "step": 47
     },
     {
       "epoch": 0.047244094488188976,
-      "grad_norm": 51.08195114135742,
+      "grad_norm": 33.678253173828125,
       "learning_rate": 4.274353876739563e-08,
-      "loss": 21.0191,
+      "loss": 7.134,
       "step": 48
     },
     {
       "epoch": 0.04822834645669291,
-      "grad_norm": 20.750757217407227,
+      "grad_norm": 50.81912612915039,
       "learning_rate": 4.373757455268391e-08,
-      "loss": 4.5581,
+      "loss": 9.3933,
       "step": 49
     },
     {
       "epoch": 0.04921259842519685,
-      "grad_norm": 26.147449493408203,
+      "grad_norm": 53.16002655029297,
       "learning_rate": 4.4731610337972176e-08,
-      "loss": 5.8754,
+      "loss": 10.8085,
       "step": 50
     },
     {
       "epoch": 0.05019685039370079,
-      "grad_norm": 18.754613876342773,
+      "grad_norm": 53.343692779541016,
       "learning_rate": 4.572564612326045e-08,
-      "loss": 4.4697,
+      "loss": 11.4172,
       "step": 51
     },
     {
       "epoch": 0.051181102362204724,
-      "grad_norm": 23.842952728271484,
+      "grad_norm": 29.51181983947754,
       "learning_rate": 4.6719681908548713e-08,
-      "loss": 5.1178,
+      "loss": 7.3397,
       "step": 52
     },
     {
       "epoch": 0.05216535433070866,
-      "grad_norm": 21.579486846923828,
+      "grad_norm": 28.54021644592285,
       "learning_rate": 4.7713717693836986e-08,
-      "loss": 4.7223,
+      "loss": 5.8851,
       "step": 53
     },
     {
       "epoch": 0.0531496062992126,
-      "grad_norm": 49.253963470458984,
+      "grad_norm": 29.310325622558594,
       "learning_rate": 4.870775347912525e-08,
-      "loss": 20.0171,
+      "loss": 6.8105,
       "step": 54
     },
     {
       "epoch": 0.054133858267716536,
-      "grad_norm": 18.999666213989258,
+      "grad_norm": 19.80927276611328,
       "learning_rate": 4.970178926441353e-08,
-      "loss": 4.5361,
+      "loss": 5.3637,
       "step": 55
     },
     {
       "epoch": 0.05511811023622047,
-      "grad_norm": 16.33707618713379,
+      "grad_norm": 30.965606689453125,
       "learning_rate": 5.06958250497018e-08,
-      "loss": 4.2409,
+      "loss": 6.2628,
       "step": 56
     },
     {
       "epoch": 0.05610236220472441,
-      "grad_norm": 13.874579429626465,
+      "grad_norm": 21.39442253112793,
       "learning_rate": 5.168986083499007e-08,
-      "loss": 4.8144,
+      "loss": 6.0039,
       "step": 57
     },
     {
       "epoch": 0.05708661417322835,
-      "grad_norm": 47.893150329589844,
+      "grad_norm": 38.37812805175781,
       "learning_rate": 5.268389662027834e-08,
-      "loss": 18.6889,
+      "loss": 7.5859,
       "step": 58
     },
     {
       "epoch": 0.058070866141732284,
-      "grad_norm": 18.776273727416992,
+      "grad_norm": 22.37571144104004,
       "learning_rate": 5.3677932405566605e-08,
-      "loss": 5.1869,
+      "loss": 6.0802,
       "step": 59
     },
     {
       "epoch": 0.05905511811023622,
-      "grad_norm": 14.070062637329102,
+      "grad_norm": 27.690608978271484,
       "learning_rate": 5.467196819085488e-08,
-      "loss": 3.9955,
+      "loss": 5.5822,
       "step": 60
     },
     {
       "epoch": 0.060039370078740155,
-      "grad_norm": 16.921262741088867,
+      "grad_norm": 23.312894821166992,
       "learning_rate": 5.5666003976143156e-08,
-      "loss": 4.705,
+      "loss": 5.8773,
       "step": 61
     },
     {
       "epoch": 0.0610236220472441,
-      "grad_norm": 10.885038375854492,
+      "grad_norm": 19.575923919677734,
       "learning_rate": 5.666003976143142e-08,
-      "loss": 4.0126,
+      "loss": 6.0814,
       "step": 62
     },
     {
       "epoch": 0.06200787401574803,
-      "grad_norm": 11.301884651184082,
+      "grad_norm": 20.126571655273438,
       "learning_rate": 5.7654075546719694e-08,
-      "loss": 3.3997,
+      "loss": 5.4483,
       "step": 63
     },
     {
       "epoch": 0.06299212598425197,
-      "grad_norm": 22.516740798950195,
+      "grad_norm": 57.36194610595703,
       "learning_rate": 5.864811133200796e-08,
-      "loss": 4.9795,
+      "loss": 10.2506,
       "step": 64
     },
     {
       "epoch": 0.0639763779527559,
-      "grad_norm": 21.108901977539062,
+      "grad_norm": 48.520416259765625,
       "learning_rate": 5.964214711729623e-08,
-      "loss": 5.867,
+      "loss": 10.5976,
       "step": 65
     },
     {
       "epoch": 0.06496062992125984,
-      "grad_norm": 13.245989799499512,
+      "grad_norm": 24.259727478027344,
       "learning_rate": 6.06361829025845e-08,
-      "loss": 4.2919,
+      "loss": 6.9942,
       "step": 66
     },
     {
       "epoch": 0.06594488188976377,
-      "grad_norm": 10.895593643188477,
+      "grad_norm": 18.457971572875977,
       "learning_rate": 6.163021868787278e-08,
-      "loss": 4.4194,
+      "loss": 5.4813,
       "step": 67
     },
     {
       "epoch": 0.06692913385826772,
-      "grad_norm": 13.352555274963379,
+      "grad_norm": 27.634117126464844,
       "learning_rate": 6.262425447316104e-08,
-      "loss": 4.2895,
+      "loss": 7.045,
       "step": 68
     },
     {
       "epoch": 0.06791338582677166,
-      "grad_norm": 40.99150085449219,
+      "grad_norm": 18.592876434326172,
       "learning_rate": 6.361829025844931e-08,
-      "loss": 16.8972,
+      "loss": 5.8549,
       "step": 69
     },
     {
       "epoch": 0.0688976377952756,
-      "grad_norm": 11.698071479797363,
+      "grad_norm": 37.92665481567383,
       "learning_rate": 6.461232604373759e-08,
-      "loss": 4.6964,
+      "loss": 8.8514,
       "step": 70
     },
     {
       "epoch": 0.06988188976377953,
-      "grad_norm": 11.680269241333008,
+      "grad_norm": 19.356266021728516,
       "learning_rate": 6.560636182902586e-08,
-      "loss": 3.9401,
+      "loss": 5.2557,
       "step": 71
     },
     {
       "epoch": 0.07086614173228346,
-      "grad_norm": 12.034364700317383,
+      "grad_norm": 13.698501586914062,
       "learning_rate": 6.660039761431412e-08,
-      "loss": 3.9605,
+      "loss": 5.1181,
       "step": 72
     },
     {
       "epoch": 0.0718503937007874,
-      "grad_norm": 11.984710693359375,
+      "grad_norm": 19.626392364501953,
       "learning_rate": 6.75944333996024e-08,
-      "loss": 4.1512,
+      "loss": 5.5331,
       "step": 73
     },
     {
       "epoch": 0.07283464566929133,
-      "grad_norm": 40.74935531616211,
+      "grad_norm": 15.392860412597656,
       "learning_rate": 6.858846918489067e-08,
-      "loss": 16.3885,
+      "loss": 5.5944,
       "step": 74
     },
     {
       "epoch": 0.07381889763779527,
-      "grad_norm": 11.025054931640625,
+      "grad_norm": 16.435239791870117,
       "learning_rate": 6.958250497017893e-08,
-      "loss": 3.7214,
+      "loss": 4.6332,
       "step": 75
     },
     {
       "epoch": 0.07480314960629922,
-      "grad_norm": 11.088836669921875,
+      "grad_norm": 16.286462783813477,
       "learning_rate": 7.057654075546721e-08,
-      "loss": 3.6038,
+      "loss": 4.9532,
       "step": 76
     },
     {
       "epoch": 0.07578740157480315,
-      "grad_norm": 11.580099105834961,
+      "grad_norm": 17.462007522583008,
       "learning_rate": 7.157057654075548e-08,
-      "loss": 4.1485,
+      "loss": 5.055,
       "step": 77
     },
     {
       "epoch": 0.07677165354330709,
-      "grad_norm": 10.454608917236328,
+      "grad_norm": 15.096108436584473,
       "learning_rate": 7.256461232604374e-08,
-      "loss": 3.7848,
+      "loss": 4.5005,
       "step": 78
     },
     {
       "epoch": 0.07775590551181102,
-      "grad_norm": 10.972256660461426,
+      "grad_norm": 14.233656883239746,
       "learning_rate": 7.355864811133201e-08,
-      "loss": 4.679,
+      "loss": 5.1997,
       "step": 79
     },
     {
       "epoch": 0.07874015748031496,
-      "grad_norm": 42.540199279785156,
+      "grad_norm": 19.416706085205078,
       "learning_rate": 7.455268389662029e-08,
-      "loss": 14.9568,
+      "loss": 5.1479,
       "step": 80
     },
     {
       "epoch": 0.0797244094488189,
-      "grad_norm": 11.498126029968262,
+      "grad_norm": 18.06542205810547,
       "learning_rate": 7.554671968190855e-08,
-      "loss": 4.1834,
+      "loss": 5.1777,
       "step": 81
     },
     {
       "epoch": 0.08070866141732283,
-      "grad_norm": 11.960761070251465,
+      "grad_norm": 16.027755737304688,
       "learning_rate": 7.654075546719683e-08,
-      "loss": 4.6593,
+      "loss": 5.5565,
       "step": 82
     },
     {
       "epoch": 0.08169291338582677,
-      "grad_norm": 10.582605361938477,
+      "grad_norm": 16.291139602661133,
       "learning_rate": 7.753479125248511e-08,
-      "loss": 3.5834,
+      "loss": 4.6999,
       "step": 83
     },
     {
       "epoch": 0.08267716535433071,
-      "grad_norm": 10.250467300415039,
+      "grad_norm": 16.005495071411133,
       "learning_rate": 7.852882703777338e-08,
-      "loss": 3.9454,
+      "loss": 5.0681,
       "step": 84
     },
     {
       "epoch": 0.08366141732283465,
-      "grad_norm": 13.08010482788086,
+      "grad_norm": 19.505733489990234,
       "learning_rate": 7.952286282306164e-08,
-      "loss": 4.1069,
+      "loss": 5.2208,
       "step": 85
     },
     {
       "epoch": 0.08464566929133858,
-      "grad_norm": 11.765287399291992,
+      "grad_norm": 18.256221771240234,
       "learning_rate": 8.051689860834992e-08,
-      "loss": 3.8427,
+      "loss": 4.56,
       "step": 86
     },
     {
       "epoch": 0.08562992125984252,
-      "grad_norm": 11.19498062133789,
+      "grad_norm": 14.99673843383789,
       "learning_rate": 8.151093439363819e-08,
-      "loss": 4.1385,
+      "loss": 4.6793,
       "step": 87
     },
     {
       "epoch": 0.08661417322834646,
-      "grad_norm": 9.966585159301758,
+      "grad_norm": 13.485681533813477,
       "learning_rate": 8.250497017892645e-08,
-      "loss": 3.678,
+      "loss": 4.4611,
       "step": 88
     },
     {
       "epoch": 0.08759842519685039,
-      "grad_norm": 10.848848342895508,
+      "grad_norm": 52.90532302856445,
       "learning_rate": 8.349900596421472e-08,
-      "loss": 4.6092,
+      "loss": 9.623,
       "step": 89
     },
     {
       "epoch": 0.08858267716535433,
-      "grad_norm": 42.184574127197266,
+      "grad_norm": 18.402536392211914,
       "learning_rate": 8.4493041749503e-08,
-      "loss": 13.5522,
+      "loss": 5.0316,
       "step": 90
     },
     {
       "epoch": 0.08956692913385826,
-      "grad_norm": 10.029223442077637,
+      "grad_norm": 12.668858528137207,
       "learning_rate": 8.548707753479126e-08,
-      "loss": 3.091,
+      "loss": 4.1771,
       "step": 91
     },
     {
       "epoch": 0.09055118110236221,
-      "grad_norm": 13.168388366699219,
+      "grad_norm": 17.02667236328125,
       "learning_rate": 8.648111332007953e-08,
-      "loss": 4.1669,
+      "loss": 4.9652,
       "step": 92
     },
     {
       "epoch": 0.09153543307086615,
-      "grad_norm": 9.128764152526855,
+      "grad_norm": 51.03838348388672,
       "learning_rate": 8.747514910536782e-08,
-      "loss": 3.9135,
+      "loss": 8.7432,
       "step": 93
     },
     {
       "epoch": 0.09251968503937008,
-      "grad_norm": 41.40175247192383,
+      "grad_norm": 15.051873207092285,
       "learning_rate": 8.846918489065609e-08,
-      "loss": 13.7737,
+      "loss": 4.6234,
       "step": 94
     },
     {
       "epoch": 0.09350393700787402,
-      "grad_norm": 10.502514839172363,
+      "grad_norm": 15.075822830200195,
       "learning_rate": 8.946322067594435e-08,
-      "loss": 3.7046,
+      "loss": 4.4016,
       "step": 95
     },
     {
       "epoch": 0.09448818897637795,
-      "grad_norm": 10.247618675231934,
+      "grad_norm": 14.73617172241211,
       "learning_rate": 9.045725646123262e-08,
-      "loss": 4.2823,
+      "loss": 4.9903,
       "step": 96
     },
     {
       "epoch": 0.09547244094488189,
-      "grad_norm": 9.737785339355469,
+      "grad_norm": 12.564230918884277,
       "learning_rate": 9.14512922465209e-08,
-      "loss": 3.5464,
+      "loss": 4.5606,
       "step": 97
     },
     {
       "epoch": 0.09645669291338582,
-      "grad_norm": 10.351418495178223,
+      "grad_norm": 12.062474250793457,
       "learning_rate": 9.244532803180916e-08,
-      "loss": 4.3904,
+      "loss": 4.9534,
       "step": 98
     },
     {
       "epoch": 0.09744094488188976,
-      "grad_norm": 10.671058654785156,
+      "grad_norm": 42.591461181640625,
       "learning_rate": 9.343936381709743e-08,
-      "loss": 4.3634,
+      "loss": 8.1838,
       "step": 99
     },
     {
       "epoch": 0.0984251968503937,
-      "grad_norm": 8.69608211517334,
+      "grad_norm": 12.18688678741455,
       "learning_rate": 9.44333996023857e-08,
-      "loss": 3.974,
+      "loss": 4.9736,
       "step": 100
     },
     {
       "epoch": 0.09940944881889764,
-      "grad_norm": 9.65053939819336,
+      "grad_norm": 14.301939964294434,
       "learning_rate": 9.542743538767397e-08,
-      "loss": 3.2875,
+      "loss": 4.4733,
       "step": 101
     },
     {
       "epoch": 0.10039370078740158,
-      "grad_norm": 9.247035026550293,
+      "grad_norm": 15.062311172485352,
       "learning_rate": 9.642147117296224e-08,
-      "loss": 3.7778,
+      "loss": 4.9725,
       "step": 102
     },
     {
       "epoch": 0.10137795275590551,
-      "grad_norm": 9.037029266357422,
+      "grad_norm": 11.707622528076172,
       "learning_rate": 9.74155069582505e-08,
-      "loss": 3.6514,
+      "loss": 4.5861,
       "step": 103
     },
     {
       "epoch": 0.10236220472440945,
-      "grad_norm": 12.288069725036621,
+      "grad_norm": 42.08045196533203,
       "learning_rate": 9.840954274353878e-08,
-      "loss": 4.0938,
+      "loss": 7.7634,
       "step": 104
     },
     {
       "epoch": 0.10334645669291338,
-      "grad_norm": 10.267972946166992,
+      "grad_norm": 13.918876647949219,
       "learning_rate": 9.940357852882706e-08,
-      "loss": 3.7483,
+      "loss": 4.9915,
       "step": 105
     },
     {
       "epoch": 0.10433070866141732,
-      "grad_norm": 12.335816383361816,
+      "grad_norm": 12.0372953414917,
       "learning_rate": 1.0039761431411533e-07,
-      "loss": 3.857,
+      "loss": 5.1391,
       "step": 106
     },
     {
       "epoch": 0.10531496062992125,
-      "grad_norm": 10.682831764221191,
+      "grad_norm": 16.739591598510742,
       "learning_rate": 1.013916500994036e-07,
-      "loss": 4.01,
+      "loss": 5.0157,
       "step": 107
     },
     {
       "epoch": 0.1062992125984252,
-      "grad_norm": 9.247139930725098,
+      "grad_norm": 16.80640983581543,
       "learning_rate": 1.0238568588469187e-07,
-      "loss": 3.0828,
+      "loss": 4.0982,
       "step": 108
     },
     {
       "epoch": 0.10728346456692914,
-      "grad_norm": 9.375234603881836,
+      "grad_norm": 10.70914077758789,
       "learning_rate": 1.0337972166998014e-07,
-      "loss": 3.5559,
+      "loss": 4.2178,
       "step": 109
     },
     {
       "epoch": 0.10826771653543307,
-      "grad_norm": 47.49584197998047,
+      "grad_norm": 11.523154258728027,
       "learning_rate": 1.043737574552684e-07,
-      "loss": 13.298,
+      "loss": 4.6193,
       "step": 110
     },
     {
       "epoch": 0.10925196850393701,
-      "grad_norm": 10.562832832336426,
+      "grad_norm": 14.624194145202637,
       "learning_rate": 1.0536779324055668e-07,
-      "loss": 3.8423,
+      "loss": 4.7638,
       "step": 111
     },
     {
       "epoch": 0.11023622047244094,
-      "grad_norm": 9.938478469848633,
+      "grad_norm": 11.593772888183594,
       "learning_rate": 1.0636182902584495e-07,
-      "loss": 3.7067,
+      "loss": 4.1207,
       "step": 112
     },
     {
       "epoch": 0.11122047244094488,
-      "grad_norm": 11.084013938903809,
+      "grad_norm": 18.1988468170166,
       "learning_rate": 1.0735586481113321e-07,
-      "loss": 3.6174,
+      "loss": 5.2034,
       "step": 113
     },
     {
       "epoch": 0.11220472440944881,
-      "grad_norm": 11.869967460632324,
+      "grad_norm": 14.054649353027344,
       "learning_rate": 1.0834990059642149e-07,
-      "loss": 4.1381,
+      "loss": 5.0693,
       "step": 114
     },
     {
       "epoch": 0.11318897637795275,
-      "grad_norm": 37.53839111328125,
+      "grad_norm": 12.262086868286133,
       "learning_rate": 1.0934393638170976e-07,
-      "loss": 5.5301,
+      "loss": 4.7895,
       "step": 115
     },
     {
       "epoch": 0.1141732283464567,
-      "grad_norm": 10.753473281860352,
+      "grad_norm": 12.955352783203125,
       "learning_rate": 1.1033797216699802e-07,
-      "loss": 3.8449,
+      "loss": 4.9486,
       "step": 116
     },
     {
       "epoch": 0.11515748031496063,
-      "grad_norm": 10.626410484313965,
+      "grad_norm": 14.151504516601562,
       "learning_rate": 1.1133200795228631e-07,
-      "loss": 3.2364,
+      "loss": 4.6552,
       "step": 117
     },
     {
       "epoch": 0.11614173228346457,
-      "grad_norm": 11.42529296875,
+      "grad_norm": 11.60208511352539,
       "learning_rate": 1.1232604373757458e-07,
-      "loss": 3.875,
+      "loss": 4.4555,
       "step": 118
     },
     {
       "epoch": 0.1171259842519685,
-      "grad_norm": 50.451011657714844,
+      "grad_norm": 11.05080509185791,
       "learning_rate": 1.1332007952286284e-07,
-      "loss": 14.9283,
+      "loss": 4.8977,
       "step": 119
     },
     {
       "epoch": 0.11811023622047244,
-      "grad_norm": 13.83673095703125,
+      "grad_norm": 43.464908599853516,
       "learning_rate": 1.1431411530815111e-07,
-      "loss": 3.4704,
+      "loss": 7.6836,
       "step": 120
     },
     {
       "epoch": 0.11909448818897637,
-      "grad_norm": 10.942705154418945,
+      "grad_norm": 11.060967445373535,
       "learning_rate": 1.1530815109343939e-07,
-      "loss": 3.6903,
+      "loss": 4.8106,
       "step": 121
     },
     {
       "epoch": 0.12007874015748031,
-      "grad_norm": 12.80109691619873,
+      "grad_norm": 11.406394004821777,
       "learning_rate": 1.1630218687872765e-07,
-      "loss": 3.6518,
+      "loss": 4.9958,
       "step": 122
     },
     {
       "epoch": 0.12106299212598425,
-      "grad_norm": 11.662338256835938,
+      "grad_norm": 12.630412101745605,
       "learning_rate": 1.1729622266401592e-07,
-      "loss": 3.6259,
+      "loss": 4.4585,
       "step": 123
     },
     {
       "epoch": 0.1220472440944882,
-      "grad_norm": 53.99936294555664,
+      "grad_norm": 42.50083541870117,
       "learning_rate": 1.182902584493042e-07,
-      "loss": 14.6407,
+      "loss": 7.5559,
       "step": 124
     },
     {
       "epoch": 0.12303149606299213,
-      "grad_norm": 9.35714340209961,
+      "grad_norm": 12.229009628295898,
       "learning_rate": 1.1928429423459245e-07,
-      "loss": 3.6638,
+      "loss": 4.2636,
       "step": 125
     },
     {
       "epoch": 0.12401574803149606,
-      "grad_norm": 10.13734245300293,
+      "grad_norm": 9.794486999511719,
       "learning_rate": 1.2027833001988073e-07,
-      "loss": 3.5541,
+      "loss": 4.0436,
       "step": 126
     },
     {
       "epoch": 0.125,
-      "grad_norm": 45.38465118408203,
+      "grad_norm": 11.505509376525879,
       "learning_rate": 1.21272365805169e-07,
-      "loss": 13.1452,
+      "loss": 4.7416,
       "step": 127
     },
     {
       "epoch": 0.12598425196850394,
-      "grad_norm": 9.38457202911377,
+      "grad_norm": 12.046993255615234,
       "learning_rate": 1.222664015904573e-07,
-      "loss": 3.2458,
+      "loss": 4.2215,
       "step": 128
     },
     {
       "epoch": 0.12696850393700787,
-      "grad_norm": 44.96736526489258,
+      "grad_norm": 33.795265197753906,
       "learning_rate": 1.2326043737574557e-07,
-      "loss": 12.2948,
+      "loss": 6.3561,
       "step": 129
     },
     {
       "epoch": 0.1279527559055118,
-      "grad_norm": 9.680144309997559,
+      "grad_norm": 33.15379333496094,
       "learning_rate": 1.2425447316103382e-07,
-      "loss": 3.1747,
+      "loss": 6.2299,
       "step": 130
     },
     {
       "epoch": 0.12893700787401574,
-      "grad_norm": 10.51301097869873,
+      "grad_norm": 11.174860000610352,
       "learning_rate": 1.2524850894632207e-07,
-      "loss": 3.2047,
+      "loss": 4.3492,
       "step": 131
     },
     {
       "epoch": 0.12992125984251968,
-      "grad_norm": 10.548707008361816,
+      "grad_norm": 10.696345329284668,
       "learning_rate": 1.2624254473161035e-07,
-      "loss": 3.3273,
+      "loss": 4.0216,
       "step": 132
     },
     {
       "epoch": 0.1309055118110236,
-      "grad_norm": 10.188769340515137,
+      "grad_norm": 35.72112274169922,
       "learning_rate": 1.2723658051689863e-07,
-      "loss": 3.7355,
+      "loss": 6.963,
       "step": 133
     },
     {
       "epoch": 0.13188976377952755,
-      "grad_norm": 9.908833503723145,
+      "grad_norm": 9.074990272521973,
       "learning_rate": 1.282306163021869e-07,
-      "loss": 3.4228,
+      "loss": 3.9474,
       "step": 134
     },
     {
       "epoch": 0.1328740157480315,
-      "grad_norm": 11.374958038330078,
+      "grad_norm": 11.409770965576172,
       "learning_rate": 1.2922465208747519e-07,
-      "loss": 3.5149,
+      "loss": 4.3437,
       "step": 135
     },
     {
       "epoch": 0.13385826771653545,
-      "grad_norm": 10.55904483795166,
+      "grad_norm": 10.008737564086914,
       "learning_rate": 1.3021868787276344e-07,
-      "loss": 3.1236,
+      "loss": 3.6267,
       "step": 136
     },
     {
       "epoch": 0.13484251968503938,
-      "grad_norm": 55.37184524536133,
+      "grad_norm": 11.481470108032227,
       "learning_rate": 1.3121272365805172e-07,
-      "loss": 12.2603,
+      "loss": 3.9896,
       "step": 137
     },
     {
       "epoch": 0.13582677165354332,
-      "grad_norm": 77.44969940185547,
+      "grad_norm": 12.856668472290039,
       "learning_rate": 1.3220675944333997e-07,
-      "loss": 12.36,
+      "loss": 4.8156,
       "step": 138
     },
     {
       "epoch": 0.13681102362204725,
-      "grad_norm": 10.958036422729492,
+      "grad_norm": 12.462326049804688,
       "learning_rate": 1.3320079522862825e-07,
-      "loss": 3.4953,
+      "loss": 4.9751,
       "step": 139
     },
     {
       "epoch": 0.1377952755905512,
-      "grad_norm": 10.853155136108398,
+      "grad_norm": 13.430492401123047,
       "learning_rate": 1.3419483101391653e-07,
-      "loss": 3.7484,
+      "loss": 4.4144,
       "step": 140
     },
     {
       "epoch": 0.13877952755905512,
-      "grad_norm": 10.227787017822266,
+      "grad_norm": 24.038984298706055,
       "learning_rate": 1.351888667992048e-07,
-      "loss": 3.6234,
+      "loss": 4.7213,
       "step": 141
     },
     {
       "epoch": 0.13976377952755906,
-      "grad_norm": 37.7499885559082,
+      "grad_norm": 39.7520637512207,
       "learning_rate": 1.3618290258449306e-07,
-      "loss": 11.2821,
+      "loss": 6.6081,
       "step": 142
     },
     {
       "epoch": 0.140748031496063,
-      "grad_norm": 9.38892650604248,
+      "grad_norm": 10.726151466369629,
       "learning_rate": 1.3717693836978134e-07,
-      "loss": 3.0984,
+      "loss": 4.2929,
       "step": 143
     },
     {
       "epoch": 0.14173228346456693,
-      "grad_norm": 10.987704277038574,
+      "grad_norm": 11.658100128173828,
       "learning_rate": 1.381709741550696e-07,
-      "loss": 3.5274,
+      "loss": 4.2537,
       "step": 144
     },
     {
       "epoch": 0.14271653543307086,
-      "grad_norm": 11.594810485839844,
+      "grad_norm": 9.662211418151855,
       "learning_rate": 1.3916500994035787e-07,
-      "loss": 3.3123,
+      "loss": 4.0647,
       "step": 145
     },
     {
       "epoch": 0.1437007874015748,
-      "grad_norm": 10.136295318603516,
+      "grad_norm": 10.004919052124023,
       "learning_rate": 1.4015904572564615e-07,
-      "loss": 3.4145,
+      "loss": 3.937,
       "step": 146
     },
     {
       "epoch": 0.14468503937007873,
-      "grad_norm": 37.52333068847656,
+      "grad_norm": 35.38102340698242,
       "learning_rate": 1.4115308151093443e-07,
-      "loss": 10.6589,
+      "loss": 5.6582,
       "step": 147
     },
     {
       "epoch": 0.14566929133858267,
-      "grad_norm": 9.571504592895508,
+      "grad_norm": 11.280171394348145,
       "learning_rate": 1.421471172962227e-07,
-      "loss": 3.1544,
+      "loss": 4.2648,
       "step": 148
     },
     {
       "epoch": 0.1466535433070866,
-      "grad_norm": 9.913719177246094,
+      "grad_norm": 10.569673538208008,
       "learning_rate": 1.4314115308151096e-07,
-      "loss": 3.5877,
+      "loss": 4.4429,
       "step": 149
     },
     {
       "epoch": 0.14763779527559054,
-      "grad_norm": 10.658916473388672,
+      "grad_norm": 14.092275619506836,
       "learning_rate": 1.4413518886679924e-07,
-      "loss": 3.2683,
+      "loss": 3.6197,
       "step": 150
     },
     {
       "epoch": 0.1486220472440945,
-      "grad_norm": 11.442712783813477,
+      "grad_norm": 8.960445404052734,
       "learning_rate": 1.451292246520875e-07,
-      "loss": 3.5216,
+      "loss": 3.7953,
       "step": 151
     },
     {
       "epoch": 0.14960629921259844,
-      "grad_norm": 51.636688232421875,
+      "grad_norm": 11.712779998779297,
       "learning_rate": 1.4612326043737577e-07,
-      "loss": 10.2844,
+      "loss": 3.8175,
       "step": 152
     },
     {
       "epoch": 0.15059055118110237,
-      "grad_norm": 9.432707786560059,
+      "grad_norm": 11.68183708190918,
       "learning_rate": 1.4711729622266402e-07,
-      "loss": 3.6084,
+      "loss": 4.5137,
       "step": 153
     },
     {
       "epoch": 0.15059055118110237,
-      "eval_Qnli-dev_cosine_accuracy": 0.5859375,
-      "eval_Qnli-dev_cosine_accuracy_threshold": 0.9280097484588623,
-      "eval_Qnli-dev_cosine_ap": 0.5559868473974543,
-      "eval_Qnli-dev_cosine_f1": 0.629878869448183,
-      "eval_Qnli-dev_cosine_f1_threshold": 0.8130671977996826,
-      "eval_Qnli-dev_cosine_precision": 0.46153846153846156,
-      "eval_Qnli-dev_cosine_recall": 0.9915254237288136,
-      "eval_Qnli-dev_dot_accuracy": 0.5859375,
-      "eval_Qnli-dev_dot_accuracy_threshold": 712.7769775390625,
-      "eval_Qnli-dev_dot_ap": 0.5559079473320865,
-      "eval_Qnli-dev_dot_f1": 0.629878869448183,
-      "eval_Qnli-dev_dot_f1_threshold": 624.4847412109375,
-      "eval_Qnli-dev_dot_precision": 0.46153846153846156,
-      "eval_Qnli-dev_dot_recall": 0.9915254237288136,
-      "eval_Qnli-dev_euclidean_accuracy": 0.5859375,
-      "eval_Qnli-dev_euclidean_accuracy_threshold": 10.516043663024902,
-      "eval_Qnli-dev_euclidean_ap": 0.5559637033258996,
-      "eval_Qnli-dev_euclidean_f1": 0.629878869448183,
-      "eval_Qnli-dev_euclidean_f1_threshold": 16.945524215698242,
-      "eval_Qnli-dev_euclidean_precision": 0.46153846153846156,
-      "eval_Qnli-dev_euclidean_recall": 0.9915254237288136,
-      "eval_Qnli-dev_manhattan_accuracy": 0.60546875,
-      "eval_Qnli-dev_manhattan_accuracy_threshold": 191.7781982421875,
-      "eval_Qnli-dev_manhattan_ap": 0.5815151188080563,
-      "eval_Qnli-dev_manhattan_f1": 0.6331658291457287,
-      "eval_Qnli-dev_manhattan_f1_threshold": 215.1607666015625,
-      "eval_Qnli-dev_manhattan_precision": 0.5235457063711911,
-      "eval_Qnli-dev_manhattan_recall": 0.8008474576271186,
-      "eval_Qnli-dev_max_accuracy": 0.60546875,
-      "eval_Qnli-dev_max_accuracy_threshold": 712.7769775390625,
-      "eval_Qnli-dev_max_ap": 0.5815151188080563,
-      "eval_Qnli-dev_max_f1": 0.6331658291457287,
-      "eval_Qnli-dev_max_f1_threshold": 624.4847412109375,
-      "eval_Qnli-dev_max_precision": 0.5235457063711911,
-      "eval_Qnli-dev_max_recall": 0.9915254237288136,
-      "eval_allNLI-dev_cosine_accuracy": 0.669921875,
-      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9783592820167542,
-      "eval_allNLI-dev_cosine_ap": 0.41204201087169523,
-      "eval_allNLI-dev_cosine_f1": 0.5058479532163743,
-      "eval_allNLI-dev_cosine_f1_threshold": 0.8176803588867188,
-      "eval_allNLI-dev_cosine_precision": 0.3385518590998043,
-      "eval_allNLI-dev_cosine_recall": 1.0,
-      "eval_allNLI-dev_dot_accuracy": 0.669921875,
-      "eval_allNLI-dev_dot_accuracy_threshold": 751.4491577148438,
-      "eval_allNLI-dev_dot_ap": 0.4121114399865795,
-      "eval_allNLI-dev_dot_f1": 0.5058479532163743,
-      "eval_allNLI-dev_dot_f1_threshold": 628.0350952148438,
-      "eval_allNLI-dev_dot_precision": 0.3385518590998043,
-      "eval_allNLI-dev_dot_recall": 1.0,
-      "eval_allNLI-dev_euclidean_accuracy": 0.669921875,
-      "eval_allNLI-dev_euclidean_accuracy_threshold": 5.765616416931152,
-      "eval_allNLI-dev_euclidean_ap": 0.41201664900644336,
-      "eval_allNLI-dev_euclidean_f1": 0.5058479532163743,
-      "eval_allNLI-dev_euclidean_f1_threshold": 16.7352294921875,
-      "eval_allNLI-dev_euclidean_precision": 0.3385518590998043,
-      "eval_allNLI-dev_euclidean_recall": 1.0,
-      "eval_allNLI-dev_manhattan_accuracy": 0.669921875,
-      "eval_allNLI-dev_manhattan_accuracy_threshold": 128.160400390625,
-      "eval_allNLI-dev_manhattan_ap": 0.410947097259041,
-      "eval_allNLI-dev_manhattan_f1": 0.5071942446043165,
-      "eval_allNLI-dev_manhattan_f1_threshold": 212.47366333007812,
-      "eval_allNLI-dev_manhattan_precision": 0.3681462140992167,
-      "eval_allNLI-dev_manhattan_recall": 0.815028901734104,
-      "eval_allNLI-dev_max_accuracy": 0.669921875,
-      "eval_allNLI-dev_max_accuracy_threshold": 751.4491577148438,
-      "eval_allNLI-dev_max_ap": 0.4121114399865795,
-      "eval_allNLI-dev_max_f1": 0.5071942446043165,
-      "eval_allNLI-dev_max_f1_threshold": 628.0350952148438,
-      "eval_allNLI-dev_max_precision": 0.3681462140992167,
-      "eval_allNLI-dev_max_recall": 1.0,
-      "eval_loss": 3.319310426712036,
-      "eval_runtime": 54.6245,
-      "eval_samples_per_second": 30.463,
-      "eval_sequential_score": 0.5815151188080563,
-      "eval_steps_per_second": 0.128,
-      "eval_sts-test_pearson_cosine": 0.19018447741866837,
-      "eval_sts-test_pearson_dot": 0.19002495383670479,
-      "eval_sts-test_pearson_euclidean": 0.2209959418841952,
-      "eval_sts-test_pearson_manhattan": 0.24668057542280933,
-      "eval_sts-test_pearson_max": 0.24668057542280933,
-      "eval_sts-test_spearman_cosine": 0.25171604480037757,
-      "eval_sts-test_spearman_dot": 0.25161920307072655,
-      "eval_sts-test_spearman_euclidean": 0.2517444171000997,
-      "eval_sts-test_spearman_manhattan": 0.27227065716753845,
-      "eval_sts-test_spearman_max": 0.27227065716753845,
+      "eval_Qnli-dev_cosine_accuracy": 0.58203125,
+      "eval_Qnli-dev_cosine_accuracy_threshold": 0.9368094801902771,
+      "eval_Qnli-dev_cosine_ap": 0.5484497034083067,
+      "eval_Qnli-dev_cosine_f1": 0.6300268096514745,
+      "eval_Qnli-dev_cosine_f1_threshold": 0.802739143371582,
+      "eval_Qnli-dev_cosine_precision": 0.46078431372549017,
+      "eval_Qnli-dev_cosine_recall": 0.9957627118644068,
+      "eval_Qnli-dev_dot_accuracy": 0.58203125,
+      "eval_Qnli-dev_dot_accuracy_threshold": 719.7518310546875,
+      "eval_Qnli-dev_dot_ap": 0.548461685358088,
+      "eval_Qnli-dev_dot_f1": 0.6300268096514745,
+      "eval_Qnli-dev_dot_f1_threshold": 616.7227783203125,
+      "eval_Qnli-dev_dot_precision": 0.46078431372549017,
+      "eval_Qnli-dev_dot_recall": 0.9957627118644068,
+      "eval_Qnli-dev_euclidean_accuracy": 0.58203125,
+      "eval_Qnli-dev_euclidean_accuracy_threshold": 9.853867530822754,
+      "eval_Qnli-dev_euclidean_ap": 0.5484497034083067,
+      "eval_Qnli-dev_euclidean_f1": 0.6300268096514745,
+      "eval_Qnli-dev_euclidean_f1_threshold": 17.40953254699707,
+      "eval_Qnli-dev_euclidean_precision": 0.46078431372549017,
+      "eval_Qnli-dev_euclidean_recall": 0.9957627118644068,
+      "eval_Qnli-dev_manhattan_accuracy": 0.607421875,
+      "eval_Qnli-dev_manhattan_accuracy_threshold": 182.1275177001953,
+      "eval_Qnli-dev_manhattan_ap": 0.5750034744442096,
+      "eval_Qnli-dev_manhattan_f1": 0.6303724928366763,
+      "eval_Qnli-dev_manhattan_f1_threshold": 230.0565185546875,
+      "eval_Qnli-dev_manhattan_precision": 0.47619047619047616,
+      "eval_Qnli-dev_manhattan_recall": 0.9322033898305084,
+      "eval_Qnli-dev_max_accuracy": 0.607421875,
+      "eval_Qnli-dev_max_accuracy_threshold": 719.7518310546875,
+      "eval_Qnli-dev_max_ap": 0.5750034744442096,
+      "eval_Qnli-dev_max_f1": 0.6303724928366763,
+      "eval_Qnli-dev_max_f1_threshold": 616.7227783203125,
+      "eval_Qnli-dev_max_precision": 0.47619047619047616,
+      "eval_Qnli-dev_max_recall": 0.9957627118644068,
+      "eval_allNLI-dev_cosine_accuracy": 0.66796875,
+      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9721524119377136,
+      "eval_allNLI-dev_cosine_ap": 0.3857994503224615,
+      "eval_allNLI-dev_cosine_f1": 0.5029239766081871,
+      "eval_allNLI-dev_cosine_f1_threshold": 0.821484386920929,
+      "eval_allNLI-dev_cosine_precision": 0.33659491193737767,
+      "eval_allNLI-dev_cosine_recall": 0.9942196531791907,
+      "eval_allNLI-dev_dot_accuracy": 0.66796875,
+      "eval_allNLI-dev_dot_accuracy_threshold": 746.914794921875,
+      "eval_allNLI-dev_dot_ap": 0.38572844452312516,
+      "eval_allNLI-dev_dot_f1": 0.5029239766081871,
+      "eval_allNLI-dev_dot_f1_threshold": 631.138916015625,
+      "eval_allNLI-dev_dot_precision": 0.33659491193737767,
+      "eval_allNLI-dev_dot_recall": 0.9942196531791907,
+      "eval_allNLI-dev_euclidean_accuracy": 0.66796875,
+      "eval_allNLI-dev_euclidean_accuracy_threshold": 6.541449546813965,
+      "eval_allNLI-dev_euclidean_ap": 0.3858031188548441,
+      "eval_allNLI-dev_euclidean_f1": 0.5029239766081871,
+      "eval_allNLI-dev_euclidean_f1_threshold": 16.558998107910156,
+      "eval_allNLI-dev_euclidean_precision": 0.33659491193737767,
+      "eval_allNLI-dev_euclidean_recall": 0.9942196531791907,
+      "eval_allNLI-dev_manhattan_accuracy": 0.666015625,
+      "eval_allNLI-dev_manhattan_accuracy_threshold": 95.24527740478516,
+      "eval_allNLI-dev_manhattan_ap": 0.39193409293721965,
+      "eval_allNLI-dev_manhattan_f1": 0.5045317220543807,
+      "eval_allNLI-dev_manhattan_f1_threshold": 254.973388671875,
+      "eval_allNLI-dev_manhattan_precision": 0.34151329243353784,
+      "eval_allNLI-dev_manhattan_recall": 0.9653179190751445,
+      "eval_allNLI-dev_max_accuracy": 0.66796875,
+      "eval_allNLI-dev_max_accuracy_threshold": 746.914794921875,
+      "eval_allNLI-dev_max_ap": 0.39193409293721965,
+      "eval_allNLI-dev_max_f1": 0.5045317220543807,
+      "eval_allNLI-dev_max_f1_threshold": 631.138916015625,
+      "eval_allNLI-dev_max_precision": 0.34151329243353784,
+      "eval_allNLI-dev_max_recall": 0.9942196531791907,
+      "eval_loss": 3.3210110664367676,
+      "eval_runtime": 53.0927,
+      "eval_samples_per_second": 31.341,
+      "eval_sequential_score": 0.5750034744442096,
+      "eval_steps_per_second": 0.132,
+      "eval_sts-test_pearson_cosine": 0.12009124140478655,
+      "eval_sts-test_pearson_dot": 0.11997652374043644,
+      "eval_sts-test_pearson_euclidean": 0.15529980522625675,
+      "eval_sts-test_pearson_manhattan": 0.18492770691981375,
+      "eval_sts-test_pearson_max": 0.18492770691981375,
+      "eval_sts-test_spearman_cosine": 0.180573622028628,
+      "eval_sts-test_spearman_dot": 0.18041242798509616,
+      "eval_sts-test_spearman_euclidean": 0.18058248277838349,
+      "eval_sts-test_spearman_manhattan": 0.21139381574888486,
+      "eval_sts-test_spearman_max": 0.21139381574888486,
       "step": 153
     },
     {
       "epoch": 0.1515748031496063,
-      "grad_norm": 10.549654960632324,
+      "grad_norm": 12.480842590332031,
       "learning_rate": 1.4811133200795232e-07,
-      "loss": 3.7025,
+      "loss": 4.3528,
       "step": 154
     },
     {
       "epoch": 0.15255905511811024,
-      "grad_norm": 10.626328468322754,
+      "grad_norm": 10.134323120117188,
       "learning_rate": 1.4910536779324058e-07,
-      "loss": 3.2625,
+      "loss": 3.6573,
       "step": 155
     },
     {
       "epoch": 0.15354330708661418,
-      "grad_norm": 10.456741333007812,
+      "grad_norm": 11.572721481323242,
       "learning_rate": 1.5009940357852886e-07,
-      "loss": 3.0366,
+      "loss": 3.5248,
       "step": 156
     },
     {
       "epoch": 0.1545275590551181,
-      "grad_norm": 38.44537353515625,
+      "grad_norm": 12.186054229736328,
       "learning_rate": 1.510934393638171e-07,
-      "loss": 9.183,
+      "loss": 3.9275,
       "step": 157
     },
     {
       "epoch": 0.15551181102362205,
-      "grad_norm": 10.580129623413086,
+      "grad_norm": 42.491851806640625,
       "learning_rate": 1.5208747514910539e-07,
-      "loss": 3.5048,
+      "loss": 7.1868,
       "step": 158
     },
     {
       "epoch": 0.15649606299212598,
-      "grad_norm": 42.2885627746582,
+      "grad_norm": 10.867408752441406,
       "learning_rate": 1.5308151093439367e-07,
-      "loss": 9.8654,
+      "loss": 3.6294,
       "step": 159
     },
     {
       "epoch": 0.15748031496062992,
-      "grad_norm": 8.579136848449707,
+      "grad_norm": 11.463488578796387,
       "learning_rate": 1.5407554671968192e-07,
-      "loss": 3.0458,
+      "loss": 3.6886,
       "step": 160
     },
     {
       "epoch": 0.15846456692913385,
-      "grad_norm": 12.40068244934082,
+      "grad_norm": 10.748167991638184,
       "learning_rate": 1.5506958250497022e-07,
-      "loss": 2.7602,
+      "loss": 3.1873,
       "step": 161
     },
     {
       "epoch": 0.1594488188976378,
-      "grad_norm": 10.646333694458008,
+      "grad_norm": 38.69701385498047,
       "learning_rate": 1.5606361829025848e-07,
-      "loss": 3.4081,
+      "loss": 6.1951,
       "step": 162
     },
     {
       "epoch": 0.16043307086614172,
-      "grad_norm": 12.419439315795898,
+      "grad_norm": 11.077349662780762,
       "learning_rate": 1.5705765407554675e-07,
-      "loss": 3.6212,
+      "loss": 3.9747,
       "step": 163
     },
     {
       "epoch": 0.16141732283464566,
-      "grad_norm": 11.597086906433105,
+      "grad_norm": 47.21928405761719,
       "learning_rate": 1.58051689860835e-07,
-      "loss": 3.625,
+      "loss": 7.004,
       "step": 164
     },
     {
       "epoch": 0.1624015748031496,
-      "grad_norm": 15.256538391113281,
+      "grad_norm": 15.23488712310791,
       "learning_rate": 1.5904572564612329e-07,
-      "loss": 3.9519,
+      "loss": 4.3221,
       "step": 165
     },
     {
       "epoch": 0.16338582677165353,
-      "grad_norm": 69.25948333740234,
+      "grad_norm": 10.329791069030762,
       "learning_rate": 1.6003976143141154e-07,
-      "loss": 9.3491,
+      "loss": 3.5963,
       "step": 166
     },
     {
       "epoch": 0.1643700787401575,
-      "grad_norm": 10.0520658493042,
+      "grad_norm": 9.624051094055176,
       "learning_rate": 1.6103379721669984e-07,
-      "loss": 2.7756,
+      "loss": 3.1988,
       "step": 167
     },
     {
       "epoch": 0.16535433070866143,
-      "grad_norm": 12.943922996520996,
+      "grad_norm": 11.511842727661133,
       "learning_rate": 1.620278330019881e-07,
-      "loss": 2.9741,
+      "loss": 3.8236,
       "step": 168
     },
     {
       "epoch": 0.16633858267716536,
-      "grad_norm": 13.181682586669922,
+      "grad_norm": 10.053628921508789,
       "learning_rate": 1.6302186878727637e-07,
-      "loss": 3.6422,
+      "loss": 3.5063,
       "step": 169
     },
     {
       "epoch": 0.1673228346456693,
-      "grad_norm": 11.436388969421387,
+      "grad_norm": 30.96078109741211,
       "learning_rate": 1.6401590457256465e-07,
-      "loss": 3.3038,
+      "loss": 5.9843,
       "step": 170
     },
     {
       "epoch": 0.16830708661417323,
-      "grad_norm": 10.708134651184082,
+      "grad_norm": 27.598793029785156,
       "learning_rate": 1.650099403578529e-07,
-      "loss": 2.8385,
+      "loss": 5.884,
       "step": 171
     },
     {
       "epoch": 0.16929133858267717,
-      "grad_norm": 13.435418128967285,
+      "grad_norm": 10.295365333557129,
       "learning_rate": 1.6600397614314118e-07,
-      "loss": 3.1081,
+      "loss": 4.1317,
       "step": 172
     },
     {
       "epoch": 0.1702755905511811,
-      "grad_norm": 11.510961532592773,
+      "grad_norm": 10.183586120605469,
       "learning_rate": 1.6699801192842944e-07,
-      "loss": 3.4009,
+      "loss": 3.9255,
       "step": 173
     },
     {
       "epoch": 0.17125984251968504,
-      "grad_norm": 14.482342720031738,
+      "grad_norm": 11.306191444396973,
       "learning_rate": 1.6799204771371774e-07,
-      "loss": 2.7715,
+      "loss": 4.1121,
       "step": 174
     },
     {
       "epoch": 0.17224409448818898,
-      "grad_norm": 11.370920181274414,
+      "grad_norm": 9.866985321044922,
       "learning_rate": 1.68986083499006e-07,
-      "loss": 3.1816,
+      "loss": 3.7748,
       "step": 175
     },
     {
       "epoch": 0.1732283464566929,
-      "grad_norm": 14.27595329284668,
+      "grad_norm": 25.779844284057617,
       "learning_rate": 1.6998011928429427e-07,
-      "loss": 3.0486,
+      "loss": 5.1602,
       "step": 176
     },
     {
       "epoch": 0.17421259842519685,
-      "grad_norm": 11.878148078918457,
+      "grad_norm": 22.994916915893555,
       "learning_rate": 1.7097415506958253e-07,
-      "loss": 2.8444,
+      "loss": 4.8807,
       "step": 177
     },
     {
       "epoch": 0.17519685039370078,
-      "grad_norm": 115.54975128173828,
+      "grad_norm": 9.323945045471191,
       "learning_rate": 1.719681908548708e-07,
-      "loss": 12.6448,
+      "loss": 3.4643,
       "step": 178
     },
     {
       "epoch": 0.17618110236220472,
-      "grad_norm": 12.08621597290039,
+      "grad_norm": 9.101490020751953,
       "learning_rate": 1.7296222664015906e-07,
-      "loss": 3.0867,
+      "loss": 3.4937,
       "step": 179
     },
     {
       "epoch": 0.17716535433070865,
-      "grad_norm": 11.200992584228516,
+      "grad_norm": 18.585084915161133,
       "learning_rate": 1.7395626242544734e-07,
-      "loss": 2.8874,
+      "loss": 5.2731,
       "step": 180
     },
     {
       "epoch": 0.1781496062992126,
-      "grad_norm": 10.824020385742188,
+      "grad_norm": 17.947126388549805,
       "learning_rate": 1.7495029821073564e-07,
-      "loss": 2.8767,
+      "loss": 4.6416,
       "step": 181
     },
     {
       "epoch": 0.17913385826771652,
-      "grad_norm": 12.792362213134766,
+      "grad_norm": 9.495512008666992,
       "learning_rate": 1.759443339960239e-07,
-      "loss": 3.1107,
+      "loss": 3.5226,
       "step": 182
     },
     {
       "epoch": 0.18011811023622049,
-      "grad_norm": 12.027408599853516,
+      "grad_norm": 21.095630645751953,
       "learning_rate": 1.7693836978131217e-07,
-      "loss": 2.3596,
+      "loss": 4.7794,
       "step": 183
     },
     {
       "epoch": 0.18110236220472442,
-      "grad_norm": 12.46939468383789,
+      "grad_norm": 9.708288192749023,
       "learning_rate": 1.7793240556660042e-07,
-      "loss": 2.6352,
+      "loss": 3.8504,
       "step": 184
     },
     {
       "epoch": 0.18208661417322836,
-      "grad_norm": 12.76395320892334,
+      "grad_norm": 10.588545799255371,
       "learning_rate": 1.789264413518887e-07,
-      "loss": 2.8008,
+      "loss": 3.5391,
       "step": 185
     },
     {
       "epoch": 0.1830708661417323,
-      "grad_norm": 10.95465087890625,
+      "grad_norm": 10.238434791564941,
       "learning_rate": 1.7992047713717695e-07,
-      "loss": 2.7295,
+      "loss": 4.0291,
       "step": 186
     },
     {
       "epoch": 0.18405511811023623,
-      "grad_norm": 75.06632995605469,
+      "grad_norm": 12.710492134094238,
       "learning_rate": 1.8091451292246523e-07,
-      "loss": 12.2772,
+      "loss": 3.5606,
       "step": 187
     },
     {
       "epoch": 0.18503937007874016,
-      "grad_norm": 10.998300552368164,
+      "grad_norm": 13.008979797363281,
       "learning_rate": 1.819085487077535e-07,
-      "loss": 2.6023,
+      "loss": 3.8957,
       "step": 188
     },
     {
       "epoch": 0.1860236220472441,
-      "grad_norm": 13.164251327514648,
+      "grad_norm": 16.594804763793945,
       "learning_rate": 1.829025844930418e-07,
-      "loss": 3.27,
+      "loss": 4.3657,
       "step": 189
     },
     {
       "epoch": 0.18700787401574803,
-      "grad_norm": 12.379849433898926,
+      "grad_norm": 25.25087547302246,
       "learning_rate": 1.8389662027833004e-07,
-      "loss": 3.1023,
+      "loss": 5.0173,
       "step": 190
     },
     {
       "epoch": 0.18799212598425197,
-      "grad_norm": 123.68094635009766,
+      "grad_norm": 18.24939727783203,
       "learning_rate": 1.8489065606361832e-07,
-      "loss": 15.7886,
+      "loss": 4.3915,
       "step": 191
     },
     {
       "epoch": 0.1889763779527559,
-      "grad_norm": 12.211271286010742,
+      "grad_norm": 11.081143379211426,
       "learning_rate": 1.8588469184890657e-07,
-      "loss": 2.8468,
+      "loss": 3.4613,
       "step": 192
     },
     {
       "epoch": 0.18996062992125984,
-      "grad_norm": 13.436681747436523,
+      "grad_norm": 10.244654655456543,
       "learning_rate": 1.8687872763419485e-07,
-      "loss": 2.1157,
+      "loss": 3.2005,
       "step": 193
     },
     {
       "epoch": 0.19094488188976377,
-      "grad_norm": 71.8902816772461,
+      "grad_norm": 10.305020332336426,
       "learning_rate": 1.8787276341948313e-07,
-      "loss": 12.1402,
+      "loss": 3.3986,
       "step": 194
     },
     {
       "epoch": 0.1919291338582677,
-      "grad_norm": 59.65647888183594,
+      "grad_norm": 12.405226707458496,
       "learning_rate": 1.888667992047714e-07,
-      "loss": 10.5477,
+      "loss": 3.7937,
       "step": 195
     },
     {
       "epoch": 0.19291338582677164,
-      "grad_norm": 11.768324851989746,
+      "grad_norm": 10.247261047363281,
       "learning_rate": 1.898608349900597e-07,
-      "loss": 2.8112,
+      "loss": 3.8981,
       "step": 196
     },
     {
       "epoch": 0.19389763779527558,
-      "grad_norm": 76.99214172363281,
+      "grad_norm": 10.535922050476074,
       "learning_rate": 1.9085487077534794e-07,
-      "loss": 11.2557,
+      "loss": 3.7051,
       "step": 197
     },
     {
       "epoch": 0.19488188976377951,
-      "grad_norm": 13.789910316467285,
+      "grad_norm": 9.828015327453613,
       "learning_rate": 1.9184890656063622e-07,
-      "loss": 2.629,
+      "loss": 3.8028,
       "step": 198
     },
     {
       "epoch": 0.19586614173228348,
-      "grad_norm": 13.719752311706543,
+      "grad_norm": 9.550689697265625,
       "learning_rate": 1.9284294234592447e-07,
-      "loss": 3.1383,
+      "loss": 3.3294,
       "step": 199
     },
     {
       "epoch": 0.1968503937007874,
-      "grad_norm": 58.55073547363281,
+      "grad_norm": 16.766008377075195,
       "learning_rate": 1.9383697813121275e-07,
-      "loss": 9.7592,
+      "loss": 4.1252,
       "step": 200
     },
     {
       "epoch": 0.19783464566929135,
-      "grad_norm": 10.811251640319824,
+      "grad_norm": 16.56619644165039,
       "learning_rate": 1.94831013916501e-07,
-      "loss": 2.9253,
+      "loss": 4.2564,
       "step": 201
     },
     {
       "epoch": 0.19881889763779528,
-      "grad_norm": 13.063098907470703,
+      "grad_norm": 17.413206100463867,
       "learning_rate": 1.958250497017893e-07,
-      "loss": 2.9712,
+      "loss": 3.8258,
       "step": 202
     },
     {
       "epoch": 0.19980314960629922,
-      "grad_norm": 12.683143615722656,
+      "grad_norm": 10.175580978393555,
       "learning_rate": 1.9681908548707756e-07,
-      "loss": 2.8585,
+      "loss": 3.1025,
       "step": 203
     },
     {
       "epoch": 0.20078740157480315,
-      "grad_norm": 11.156440734863281,
+      "grad_norm": 11.58014965057373,
       "learning_rate": 1.9781312127236584e-07,
-      "loss": 3.082,
+      "loss": 3.5038,
       "step": 204
     },
     {
       "epoch": 0.2017716535433071,
-      "grad_norm": 11.65755844116211,
+      "grad_norm": 9.23029613494873,
       "learning_rate": 1.9880715705765412e-07,
-      "loss": 3.2078,
+      "loss": 3.6021,
       "step": 205
     },
     {
       "epoch": 0.20275590551181102,
-      "grad_norm": 11.212764739990234,
+      "grad_norm": 11.87929630279541,
       "learning_rate": 1.9980119284294237e-07,
-      "loss": 2.4837,
+      "loss": 3.7637,
       "step": 206
     },
     {
       "epoch": 0.20374015748031496,
-      "grad_norm": 47.730979919433594,
+      "grad_norm": 10.55318546295166,
       "learning_rate": 2.0079522862823065e-07,
-      "loss": 7.7377,
+      "loss": 3.2563,
       "step": 207
     },
     {
       "epoch": 0.2047244094488189,
-      "grad_norm": 12.969826698303223,
+      "grad_norm": 15.530010223388672,
       "learning_rate": 2.017892644135189e-07,
-      "loss": 2.3949,
+      "loss": 3.9323,
       "step": 208
     },
     {
       "epoch": 0.20570866141732283,
-      "grad_norm": 10.695167541503906,
+      "grad_norm": 12.805971145629883,
       "learning_rate": 2.027833001988072e-07,
-      "loss": 2.7936,
+      "loss": 3.489,
       "step": 209
     },
     {
       "epoch": 0.20669291338582677,
-      "grad_norm": 11.578596115112305,
+      "grad_norm": 11.042547225952148,
       "learning_rate": 2.0377733598409546e-07,
-      "loss": 2.5481,
+      "loss": 3.6549,
       "step": 210
     },
     {
       "epoch": 0.2076771653543307,
-      "grad_norm": 12.902640342712402,
+      "grad_norm": 13.565040588378906,
       "learning_rate": 2.0477137176938374e-07,
-      "loss": 2.8108,
+      "loss": 3.1609,
       "step": 211
     },
     {
       "epoch": 0.20866141732283464,
-      "grad_norm": 13.556937217712402,
+      "grad_norm": 10.574054718017578,
       "learning_rate": 2.05765407554672e-07,
-      "loss": 2.6916,
+      "loss": 3.2467,
       "step": 212
     },
     {
       "epoch": 0.20964566929133857,
-      "grad_norm": 14.155546188354492,
+      "grad_norm": 9.994040489196777,
       "learning_rate": 2.0675944333996027e-07,
-      "loss": 3.3112,
+      "loss": 3.4514,
       "step": 213
     },
     {
       "epoch": 0.2106299212598425,
-      "grad_norm": 12.20670223236084,
+      "grad_norm": 11.422471046447754,
       "learning_rate": 2.0775347912524852e-07,
-      "loss": 2.6884,
+      "loss": 3.4945,
       "step": 214
     },
     {
       "epoch": 0.21161417322834647,
-      "grad_norm": 13.744552612304688,
+      "grad_norm": 13.696518898010254,
       "learning_rate": 2.087475149105368e-07,
-      "loss": 2.4656,
+      "loss": 3.5932,
       "step": 215
     },
     {
       "epoch": 0.2125984251968504,
-      "grad_norm": 12.515267372131348,
+      "grad_norm": 9.801803588867188,
       "learning_rate": 2.097415506958251e-07,
-      "loss": 2.3028,
+      "loss": 3.2289,
       "step": 216
     },
     {
       "epoch": 0.21358267716535434,
-      "grad_norm": 12.110260009765625,
+      "grad_norm": 10.710221290588379,
       "learning_rate": 2.1073558648111336e-07,
-      "loss": 2.4515,
+      "loss": 3.3279,
       "step": 217
     },
     {
       "epoch": 0.21456692913385828,
-      "grad_norm": 13.38983154296875,
+      "grad_norm": 12.974212646484375,
       "learning_rate": 2.1172962226640164e-07,
-      "loss": 2.7888,
+      "loss": 3.8141,
       "step": 218
     },
     {
       "epoch": 0.2155511811023622,
-      "grad_norm": 12.12717342376709,
+      "grad_norm": 10.196144104003906,
       "learning_rate": 2.127236580516899e-07,
-      "loss": 2.7798,
+      "loss": 3.1171,
       "step": 219
     },
     {
       "epoch": 0.21653543307086615,
-      "grad_norm": 11.084267616271973,
+      "grad_norm": 12.518829345703125,
       "learning_rate": 2.1371769383697817e-07,
-      "loss": 2.2994,
+      "loss": 3.6287,
       "step": 220
     },
     {
       "epoch": 0.21751968503937008,
-      "grad_norm": 13.082998275756836,
+      "grad_norm": 18.030723571777344,
       "learning_rate": 2.1471172962226642e-07,
-      "loss": 2.4806,
+      "loss": 3.8517,
       "step": 221
     },
     {
       "epoch": 0.21850393700787402,
-      "grad_norm": 13.084571838378906,
+      "grad_norm": 11.197093963623047,
       "learning_rate": 2.1570576540755473e-07,
-      "loss": 2.6355,
+      "loss": 3.3836,
       "step": 222
     },
     {
       "epoch": 0.21948818897637795,
-      "grad_norm": 12.001659393310547,
+      "grad_norm": 10.613561630249023,
       "learning_rate": 2.1669980119284298e-07,
-      "loss": 2.4232,
+      "loss": 3.425,
       "step": 223
     },
     {
       "epoch": 0.2204724409448819,
-      "grad_norm": 12.470407485961914,
+      "grad_norm": 11.70770263671875,
       "learning_rate": 2.1769383697813126e-07,
-      "loss": 2.6152,
+      "loss": 3.6246,
       "step": 224
     },
     {
       "epoch": 0.22145669291338582,
-      "grad_norm": 11.995153427124023,
+      "grad_norm": 10.802409172058105,
       "learning_rate": 2.186878727634195e-07,
-      "loss": 2.6337,
+      "loss": 3.5682,
       "step": 225
     },
     {
       "epoch": 0.22244094488188976,
-      "grad_norm": 14.227544784545898,
+      "grad_norm": 10.422605514526367,
       "learning_rate": 2.196819085487078e-07,
-      "loss": 2.4903,
+      "loss": 3.3034,
       "step": 226
     },
     {
       "epoch": 0.2234251968503937,
-      "grad_norm": 12.475507736206055,
+      "grad_norm": 17.00844383239746,
       "learning_rate": 2.2067594433399604e-07,
-      "loss": 2.3831,
+      "loss": 3.9251,
       "step": 227
     },
     {
       "epoch": 0.22440944881889763,
-      "grad_norm": 13.871504783630371,
+      "grad_norm": 11.708518028259277,
       "learning_rate": 2.2166998011928432e-07,
-      "loss": 2.1024,
+      "loss": 3.146,
       "step": 228
     },
     {
       "epoch": 0.22539370078740156,
-      "grad_norm": 14.166851043701172,
+      "grad_norm": 11.237220764160156,
       "learning_rate": 2.2266401590457263e-07,
-      "loss": 2.7021,
+      "loss": 3.8859,
       "step": 229
     },
     {
       "epoch": 0.2263779527559055,
-      "grad_norm": 13.170698165893555,
+      "grad_norm": 10.216686248779297,
       "learning_rate": 2.2365805168986088e-07,
-      "loss": 2.3359,
+      "loss": 3.2977,
       "step": 230
     },
     {
       "epoch": 0.22736220472440946,
-      "grad_norm": 12.220829963684082,
+      "grad_norm": 11.36294937133789,
       "learning_rate": 2.2465208747514916e-07,
-      "loss": 1.9508,
+      "loss": 3.2664,
       "step": 231
     },
     {
       "epoch": 0.2283464566929134,
-      "grad_norm": 14.62851619720459,
+      "grad_norm": 9.619734764099121,
       "learning_rate": 2.256461232604374e-07,
-      "loss": 2.061,
+      "loss": 3.1275,
       "step": 232
     },
     {
       "epoch": 0.22933070866141733,
-      "grad_norm": 117.03069305419922,
+      "grad_norm": 10.229348182678223,
       "learning_rate": 2.266401590457257e-07,
-      "loss": 15.199,
+      "loss": 3.2408,
       "step": 233
     },
     {
       "epoch": 0.23031496062992127,
-      "grad_norm": 111.26533508300781,
+      "grad_norm": 11.304025650024414,
       "learning_rate": 2.2763419483101394e-07,
-      "loss": 14.708,
+      "loss": 2.907,
       "step": 234
     },
     {
       "epoch": 0.2312992125984252,
-      "grad_norm": 107.53176879882812,
+      "grad_norm": 11.743476867675781,
       "learning_rate": 2.2862823061630222e-07,
-      "loss": 15.589,
+      "loss": 2.9178,
       "step": 235
     },
     {
       "epoch": 0.23228346456692914,
-      "grad_norm": 11.968624114990234,
+      "grad_norm": 12.260607719421387,
       "learning_rate": 2.296222664015905e-07,
-      "loss": 2.4485,
+      "loss": 3.324,
       "step": 236
     },
     {
       "epoch": 0.23326771653543307,
-      "grad_norm": 14.77911376953125,
+      "grad_norm": 12.115522384643555,
       "learning_rate": 2.3061630218687878e-07,
-      "loss": 2.1559,
+      "loss": 2.9172,
       "step": 237
     },
     {
       "epoch": 0.234251968503937,
-      "grad_norm": 17.22453498840332,
+      "grad_norm": 13.81004810333252,
       "learning_rate": 2.3161033797216703e-07,
-      "loss": 3.0109,
+      "loss": 3.4324,
       "step": 238
     },
     {
       "epoch": 0.23523622047244094,
-      "grad_norm": 15.654526710510254,
+      "grad_norm": 28.633853912353516,
       "learning_rate": 2.326043737574553e-07,
-      "loss": 2.2779,
+      "loss": 4.0563,
       "step": 239
     },
     {
       "epoch": 0.23622047244094488,
-      "grad_norm": 79.38912200927734,
+      "grad_norm": 10.444801330566406,
       "learning_rate": 2.335984095427436e-07,
-      "loss": 12.1054,
+      "loss": 2.8736,
       "step": 240
     },
     {
       "epoch": 0.2372047244094488,
-      "grad_norm": 13.8206787109375,
+      "grad_norm": 30.760478973388672,
       "learning_rate": 2.3459244532803184e-07,
-      "loss": 2.7905,
+      "loss": 4.7174,
       "step": 241
     },
     {
       "epoch": 0.23818897637795275,
-      "grad_norm": 13.186335563659668,
+      "grad_norm": 11.556520462036133,
       "learning_rate": 2.3558648111332012e-07,
-      "loss": 2.2739,
+      "loss": 3.2025,
       "step": 242
     },
     {
       "epoch": 0.23917322834645668,
-      "grad_norm": 10.965237617492676,
+      "grad_norm": 10.850605010986328,
       "learning_rate": 2.365805168986084e-07,
-      "loss": 2.0147,
+      "loss": 2.7835,
       "step": 243
     },
     {
       "epoch": 0.24015748031496062,
-      "grad_norm": 65.3853530883789,
+      "grad_norm": 26.828187942504883,
       "learning_rate": 2.3757455268389668e-07,
-      "loss": 10.8463,
+      "loss": 4.3158,
       "step": 244
     },
     {
       "epoch": 0.24114173228346455,
-      "grad_norm": 11.911761283874512,
+      "grad_norm": 10.565204620361328,
       "learning_rate": 2.385685884691849e-07,
-      "loss": 2.4448,
+      "loss": 2.8619,
       "step": 245
     },
     {
       "epoch": 0.2421259842519685,
-      "grad_norm": 11.165079116821289,
+      "grad_norm": 11.693890571594238,
       "learning_rate": 2.395626242544732e-07,
-      "loss": 2.1458,
+      "loss": 2.5156,
       "step": 246
     },
     {
       "epoch": 0.24311023622047245,
-      "grad_norm": 83.87309265136719,
+      "grad_norm": 14.487508773803711,
       "learning_rate": 2.4055666003976146e-07,
-      "loss": 11.9076,
+      "loss": 3.2144,
       "step": 247
     },
     {
       "epoch": 0.2440944881889764,
-      "grad_norm": 48.82333755493164,
+      "grad_norm": 22.58393669128418,
       "learning_rate": 2.4155069582504976e-07,
-      "loss": 4.2321,
+      "loss": 3.5927,
       "step": 248
     },
     {
       "epoch": 0.24507874015748032,
-      "grad_norm": 62.72069549560547,
+      "grad_norm": 10.880494117736816,
       "learning_rate": 2.42544731610338e-07,
-      "loss": 9.6525,
+      "loss": 2.6059,
       "step": 249
     },
     {
       "epoch": 0.24606299212598426,
-      "grad_norm": 54.45147705078125,
+      "grad_norm": 10.982386589050293,
       "learning_rate": 2.4353876739562627e-07,
-      "loss": 9.5191,
+      "loss": 2.9758,
       "step": 250
     },
     {
       "epoch": 0.2470472440944882,
-      "grad_norm": 13.214591979980469,
+      "grad_norm": 20.552265167236328,
       "learning_rate": 2.445328031809146e-07,
-      "loss": 3.0668,
+      "loss": 3.9214,
       "step": 251
     },
     {
       "epoch": 0.24803149606299213,
-      "grad_norm": 36.288726806640625,
+      "grad_norm": 11.680928230285645,
       "learning_rate": 2.4552683896620283e-07,
-      "loss": 7.0423,
+      "loss": 3.2892,
       "step": 252
     },
     {
       "epoch": 0.24901574803149606,
-      "grad_norm": 11.362079620361328,
+      "grad_norm": 10.980914115905762,
       "learning_rate": 2.4652087475149113e-07,
-      "loss": 2.4391,
+      "loss": 2.9503,
       "step": 253
     },
     {
       "epoch": 0.25,
-      "grad_norm": 11.57868766784668,
+      "grad_norm": 11.629937171936035,
       "learning_rate": 2.475149105367794e-07,
-      "loss": 2.0855,
+      "loss": 2.5969,
       "step": 254
     },
     {
       "epoch": 0.25098425196850394,
-      "grad_norm": 10.305681228637695,
+      "grad_norm": 11.606547355651855,
       "learning_rate": 2.4850894632206764e-07,
-      "loss": 2.5628,
+      "loss": 2.9908,
       "step": 255
     },
     {
       "epoch": 0.25196850393700787,
-      "grad_norm": 34.71120834350586,
+      "grad_norm": 10.955612182617188,
       "learning_rate": 2.495029821073559e-07,
-      "loss": 6.4129,
+      "loss": 2.8995,
       "step": 256
     },
     {
       "epoch": 0.2529527559055118,
-      "grad_norm": 12.710351943969727,
+      "grad_norm": 11.561877250671387,
       "learning_rate": 2.5049701789264414e-07,
-      "loss": 3.0284,
+      "loss": 3.124,
       "step": 257
     },
     {
       "epoch": 0.25393700787401574,
-      "grad_norm": 11.934942245483398,
+      "grad_norm": 12.040234565734863,
       "learning_rate": 2.5149105367793245e-07,
-      "loss": 3.0393,
+      "loss": 3.1197,
       "step": 258
     },
     {
       "epoch": 0.2549212598425197,
-      "grad_norm": 33.399269104003906,
+      "grad_norm": 12.25498104095459,
       "learning_rate": 2.524850894632207e-07,
-      "loss": 5.7205,
+      "loss": 2.3073,
       "step": 259
     },
     {
       "epoch": 0.2559055118110236,
-      "grad_norm": 10.524690628051758,
+      "grad_norm": 9.674198150634766,
       "learning_rate": 2.53479125248509e-07,
-      "loss": 2.8864,
+      "loss": 2.8441,
       "step": 260
     },
     {
       "epoch": 0.25688976377952755,
-      "grad_norm": 10.373811721801758,
+      "grad_norm": 11.55859661102295,
       "learning_rate": 2.5447316103379726e-07,
-      "loss": 2.1274,
+      "loss": 1.9788,
       "step": 261
     },
     {
       "epoch": 0.2578740157480315,
-      "grad_norm": 10.531494140625,
+      "grad_norm": 12.343245506286621,
       "learning_rate": 2.554671968190855e-07,
-      "loss": 2.3253,
+      "loss": 2.1442,
       "step": 262
     },
     {
       "epoch": 0.2588582677165354,
-      "grad_norm": 12.033869743347168,
+      "grad_norm": 35.242835998535156,
       "learning_rate": 2.564612326043738e-07,
-      "loss": 2.7868,
+      "loss": 4.9015,
       "step": 263
     },
     {
       "epoch": 0.25984251968503935,
-      "grad_norm": 11.448728561401367,
+      "grad_norm": 13.894003868103027,
       "learning_rate": 2.5745526838966207e-07,
-      "loss": 2.5113,
+      "loss": 2.7866,
       "step": 264
     },
     {
       "epoch": 0.2608267716535433,
-      "grad_norm": 31.349775314331055,
+      "grad_norm": 12.785796165466309,
       "learning_rate": 2.5844930417495037e-07,
-      "loss": 5.5063,
+      "loss": 2.4588,
       "step": 265
     },
     {
       "epoch": 0.2618110236220472,
-      "grad_norm": 10.772862434387207,
+      "grad_norm": 13.115778923034668,
       "learning_rate": 2.5944333996023857e-07,
-      "loss": 2.5678,
+      "loss": 2.3909,
       "step": 266
     },
     {
       "epoch": 0.26279527559055116,
-      "grad_norm": 39.487667083740234,
+      "grad_norm": 38.795326232910156,
       "learning_rate": 2.604373757455269e-07,
-      "loss": 5.6321,
+      "loss": 4.7394,
       "step": 267
     },
     {
       "epoch": 0.2637795275590551,
-      "grad_norm": 12.496540069580078,
+      "grad_norm": 14.078845977783203,
       "learning_rate": 2.614314115308152e-07,
-      "loss": 2.3219,
+      "loss": 3.1581,
       "step": 268
     },
     {
       "epoch": 0.26476377952755903,
-      "grad_norm": 9.853267669677734,
+      "grad_norm": 36.33890151977539,
       "learning_rate": 2.6242544731610343e-07,
-      "loss": 2.0576,
+      "loss": 3.973,
       "step": 269
     },
     {
       "epoch": 0.265748031496063,
-      "grad_norm": 11.759198188781738,
+      "grad_norm": 32.12986755371094,
       "learning_rate": 2.634194831013917e-07,
-      "loss": 2.7055,
+      "loss": 4.1565,
       "step": 270
     },
     {
       "epoch": 0.26673228346456695,
-      "grad_norm": 11.300050735473633,
+      "grad_norm": 13.353660583496094,
       "learning_rate": 2.6441351888667994e-07,
-      "loss": 2.2286,
+      "loss": 2.5183,
       "step": 271
     },
     {
       "epoch": 0.2677165354330709,
-      "grad_norm": 13.588562965393066,
+      "grad_norm": 24.31439208984375,
       "learning_rate": 2.6540755467196824e-07,
-      "loss": 2.6628,
+      "loss": 3.614,
       "step": 272
     },
     {
       "epoch": 0.2687007874015748,
-      "grad_norm": 37.17454528808594,
+      "grad_norm": 11.95603084564209,
       "learning_rate": 2.664015904572565e-07,
-      "loss": 5.2455,
+      "loss": 2.6858,
       "step": 273
     },
     {
       "epoch": 0.26968503937007876,
-      "grad_norm": 12.299790382385254,
+      "grad_norm": 14.08092975616455,
       "learning_rate": 2.6739562624254475e-07,
-      "loss": 2.6966,
+      "loss": 3.1182,
       "step": 274
     },
     {
       "epoch": 0.2706692913385827,
-      "grad_norm": 12.163739204406738,
+      "grad_norm": 11.759414672851562,
       "learning_rate": 2.6838966202783305e-07,
-      "loss": 2.65,
+      "loss": 2.9628,
       "step": 275
     },
     {
       "epoch": 0.27165354330708663,
-      "grad_norm": 12.674065589904785,
+      "grad_norm": 13.239761352539062,
       "learning_rate": 2.693836978131213e-07,
-      "loss": 2.6302,
+      "loss": 2.8376,
       "step": 276
     },
     {
       "epoch": 0.27263779527559057,
-      "grad_norm": 13.80542278289795,
+      "grad_norm": 13.551993370056152,
       "learning_rate": 2.703777335984096e-07,
-      "loss": 2.6564,
+      "loss": 2.7858,
       "step": 277
     },
     {
       "epoch": 0.2736220472440945,
-      "grad_norm": 11.924317359924316,
+      "grad_norm": 12.110513687133789,
       "learning_rate": 2.7137176938369786e-07,
-      "loss": 2.0339,
+      "loss": 2.1037,
       "step": 278
     },
     {
       "epoch": 0.27460629921259844,
-      "grad_norm": 15.281108856201172,
+      "grad_norm": 14.467287063598633,
       "learning_rate": 2.723658051689861e-07,
-      "loss": 2.8719,
+      "loss": 3.0436,
       "step": 279
     },
     {
       "epoch": 0.2755905511811024,
-      "grad_norm": 14.091357231140137,
+      "grad_norm": 17.148061752319336,
       "learning_rate": 2.7335984095427437e-07,
-      "loss": 2.5937,
+      "loss": 3.4125,
       "step": 280
     },
     {
       "epoch": 0.2765748031496063,
-      "grad_norm": 37.14817428588867,
+      "grad_norm": 12.052011489868164,
       "learning_rate": 2.743538767395627e-07,
-      "loss": 5.97,
+      "loss": 2.5027,
       "step": 281
     },
     {
       "epoch": 0.27755905511811024,
-      "grad_norm": 11.206151962280273,
+      "grad_norm": 11.502729415893555,
       "learning_rate": 2.75347912524851e-07,
-      "loss": 2.1888,
+      "loss": 2.7922,
       "step": 282
     },
     {
       "epoch": 0.2785433070866142,
-      "grad_norm": 11.060794830322266,
+      "grad_norm": 15.440553665161133,
       "learning_rate": 2.763419483101392e-07,
-      "loss": 2.4111,
+      "loss": 2.9762,
       "step": 283
     },
     {
       "epoch": 0.2795275590551181,
-      "grad_norm": 11.33033561706543,
+      "grad_norm": 12.604522705078125,
       "learning_rate": 2.773359840954275e-07,
-      "loss": 2.2334,
+      "loss": 2.6458,
       "step": 284
     },
     {
       "epoch": 0.28051181102362205,
-      "grad_norm": 46.83280563354492,
+      "grad_norm": 12.404953002929688,
       "learning_rate": 2.7833001988071574e-07,
-      "loss": 6.7992,
+      "loss": 2.962,
       "step": 285
     },
     {
       "epoch": 0.281496062992126,
-      "grad_norm": 12.653908729553223,
+      "grad_norm": 12.527341842651367,
       "learning_rate": 2.7932405566600404e-07,
-      "loss": 1.8102,
+      "loss": 2.5439,
       "step": 286
     },
     {
       "epoch": 0.2824803149606299,
-      "grad_norm": 12.369451522827148,
+      "grad_norm": 12.331761360168457,
       "learning_rate": 2.803180914512923e-07,
-      "loss": 2.3559,
+      "loss": 2.8437,
       "step": 287
     },
     {
       "epoch": 0.28346456692913385,
-      "grad_norm": 47.14701461791992,
+      "grad_norm": 17.64342498779297,
       "learning_rate": 2.8131212723658055e-07,
-      "loss": 6.0575,
+      "loss": 3.2134,
       "step": 288
     },
     {
       "epoch": 0.2844488188976378,
-      "grad_norm": 11.92288875579834,
+      "grad_norm": 12.74220085144043,
       "learning_rate": 2.8230616302186885e-07,
-      "loss": 2.3846,
+      "loss": 2.5655,
       "step": 289
     },
     {
       "epoch": 0.2854330708661417,
-      "grad_norm": 12.775434494018555,
+      "grad_norm": 13.071759223937988,
       "learning_rate": 2.833001988071571e-07,
-      "loss": 2.7311,
+      "loss": 2.9465,
       "step": 290
     },
     {
       "epoch": 0.28641732283464566,
-      "grad_norm": 66.64407348632812,
+      "grad_norm": 12.093460083007812,
       "learning_rate": 2.842942345924454e-07,
-      "loss": 7.2351,
+      "loss": 2.4653,
       "step": 291
     },
     {
       "epoch": 0.2874015748031496,
-      "grad_norm": 12.071573257446289,
+      "grad_norm": 18.303346633911133,
       "learning_rate": 2.852882703777336e-07,
-      "loss": 2.3992,
+      "loss": 3.1467,
       "step": 292
     },
     {
       "epoch": 0.28838582677165353,
-      "grad_norm": 9.866507530212402,
+      "grad_norm": 11.613899230957031,
       "learning_rate": 2.862823061630219e-07,
-      "loss": 2.1329,
+      "loss": 2.6551,
       "step": 293
     },
     {
       "epoch": 0.28937007874015747,
-      "grad_norm": 10.664796829223633,
+      "grad_norm": 12.590253829956055,
       "learning_rate": 2.8727634194831017e-07,
-      "loss": 2.0372,
+      "loss": 2.5098,
       "step": 294
     },
     {
       "epoch": 0.2903543307086614,
-      "grad_norm": 11.258280754089355,
+      "grad_norm": 13.879076957702637,
       "learning_rate": 2.8827037773359847e-07,
-      "loss": 1.9959,
+      "loss": 2.5988,
       "step": 295
     },
     {
       "epoch": 0.29133858267716534,
-      "grad_norm": 11.857529640197754,
+      "grad_norm": 23.504112243652344,
       "learning_rate": 2.892644135188867e-07,
-      "loss": 2.5762,
+      "loss": 3.778,
       "step": 296
     },
     {
       "epoch": 0.29232283464566927,
-      "grad_norm": 12.668633460998535,
+      "grad_norm": 11.794881820678711,
       "learning_rate": 2.90258449304175e-07,
-      "loss": 2.454,
+      "loss": 2.6257,
       "step": 297
     },
     {
       "epoch": 0.2933070866141732,
-      "grad_norm": 11.908174514770508,
+      "grad_norm": 11.491183280944824,
       "learning_rate": 2.912524850894633e-07,
-      "loss": 2.4293,
+      "loss": 2.5142,
       "step": 298
     },
     {
       "epoch": 0.29429133858267714,
-      "grad_norm": 10.250532150268555,
+      "grad_norm": 11.801483154296875,
       "learning_rate": 2.9224652087475153e-07,
-      "loss": 1.6916,
+      "loss": 2.3182,
       "step": 299
     },
     {
       "epoch": 0.2952755905511811,
-      "grad_norm": 11.219808578491211,
+      "grad_norm": 21.28721046447754,
       "learning_rate": 2.9324055666003984e-07,
-      "loss": 2.0638,
+      "loss": 3.3505,
       "step": 300
     },
     {
       "epoch": 0.296259842519685,
-      "grad_norm": 10.419336318969727,
+      "grad_norm": 14.353005409240723,
       "learning_rate": 2.9423459244532804e-07,
-      "loss": 1.967,
+      "loss": 2.9615,
       "step": 301
     },
     {
       "epoch": 0.297244094488189,
-      "grad_norm": 12.338981628417969,
+      "grad_norm": 17.378894805908203,
       "learning_rate": 2.9522862823061634e-07,
-      "loss": 2.3572,
+      "loss": 2.9136,
       "step": 302
     },
     {
       "epoch": 0.29822834645669294,
-      "grad_norm": 11.970561027526855,
+      "grad_norm": 12.216327667236328,
       "learning_rate": 2.9622266401590465e-07,
-      "loss": 2.1485,
+      "loss": 2.6192,
       "step": 303
     },
     {
       "epoch": 0.2992125984251969,
-      "grad_norm": 52.54031753540039,
+      "grad_norm": 12.561356544494629,
       "learning_rate": 2.972166998011929e-07,
-      "loss": 7.7883,
+      "loss": 2.3255,
       "step": 304
     },
     {
       "epoch": 0.3001968503937008,
-      "grad_norm": 11.679478645324707,
+      "grad_norm": 15.897427558898926,
       "learning_rate": 2.9821073558648115e-07,
-      "loss": 2.1426,
+      "loss": 2.7168,
       "step": 305
     },
     {
       "epoch": 0.30118110236220474,
-      "grad_norm": 10.657000541687012,
+      "grad_norm": 20.39179039001465,
       "learning_rate": 2.992047713717694e-07,
-      "loss": 1.7839,
+      "loss": 2.9137,
       "step": 306
     },
     {
       "epoch": 0.30118110236220474,
-      "eval_Qnli-dev_cosine_accuracy": 0.603515625,
-      "eval_Qnli-dev_cosine_accuracy_threshold": 0.90336012840271,
-      "eval_Qnli-dev_cosine_ap": 0.5993966680292693,
-      "eval_Qnli-dev_cosine_f1": 0.637883008356546,
-      "eval_Qnli-dev_cosine_f1_threshold": 0.7992255687713623,
-      "eval_Qnli-dev_cosine_precision": 0.475103734439834,
-      "eval_Qnli-dev_cosine_recall": 0.9703389830508474,
-      "eval_Qnli-dev_dot_accuracy": 0.603515625,
-      "eval_Qnli-dev_dot_accuracy_threshold": 693.1090087890625,
-      "eval_Qnli-dev_dot_ap": 0.5990996370946063,
-      "eval_Qnli-dev_dot_f1": 0.637883008356546,
-      "eval_Qnli-dev_dot_f1_threshold": 613.3412475585938,
-      "eval_Qnli-dev_dot_precision": 0.475103734439834,
-      "eval_Qnli-dev_dot_recall": 0.9703389830508474,
-      "eval_Qnli-dev_euclidean_accuracy": 0.603515625,
-      "eval_Qnli-dev_euclidean_accuracy_threshold": 12.177652359008789,
-      "eval_Qnli-dev_euclidean_ap": 0.5994114375803649,
-      "eval_Qnli-dev_euclidean_f1": 0.637883008356546,
-      "eval_Qnli-dev_euclidean_f1_threshold": 17.55432891845703,
-      "eval_Qnli-dev_euclidean_precision": 0.475103734439834,
-      "eval_Qnli-dev_euclidean_recall": 0.9703389830508474,
-      "eval_Qnli-dev_manhattan_accuracy": 0.623046875,
-      "eval_Qnli-dev_manhattan_accuracy_threshold": 253.44888305664062,
-      "eval_Qnli-dev_manhattan_ap": 0.6114921536490292,
-      "eval_Qnli-dev_manhattan_f1": 0.6529411764705882,
-      "eval_Qnli-dev_manhattan_f1_threshold": 327.55078125,
-      "eval_Qnli-dev_manhattan_precision": 0.5,
-      "eval_Qnli-dev_manhattan_recall": 0.940677966101695,
-      "eval_Qnli-dev_max_accuracy": 0.623046875,
-      "eval_Qnli-dev_max_accuracy_threshold": 693.1090087890625,
-      "eval_Qnli-dev_max_ap": 0.6114921536490292,
-      "eval_Qnli-dev_max_f1": 0.6529411764705882,
-      "eval_Qnli-dev_max_f1_threshold": 613.3412475585938,
-      "eval_Qnli-dev_max_precision": 0.5,
-      "eval_Qnli-dev_max_recall": 0.9703389830508474,
-      "eval_allNLI-dev_cosine_accuracy": 0.669921875,
-      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9743889570236206,
-      "eval_allNLI-dev_cosine_ap": 0.41419182158586465,
-      "eval_allNLI-dev_cosine_f1": 0.5081669691470054,
-      "eval_allNLI-dev_cosine_f1_threshold": 0.8874048590660095,
-      "eval_allNLI-dev_cosine_precision": 0.37037037037037035,
-      "eval_allNLI-dev_cosine_recall": 0.8092485549132948,
+      "eval_Qnli-dev_cosine_accuracy": 0.59765625,
+      "eval_Qnli-dev_cosine_accuracy_threshold": 0.9195176362991333,
+      "eval_Qnli-dev_cosine_ap": 0.5806713513254954,
+      "eval_Qnli-dev_cosine_f1": 0.6419400855920115,
+      "eval_Qnli-dev_cosine_f1_threshold": 0.8419712781906128,
+      "eval_Qnli-dev_cosine_precision": 0.4838709677419355,
+      "eval_Qnli-dev_cosine_recall": 0.9533898305084746,
+      "eval_Qnli-dev_dot_accuracy": 0.59765625,
+      "eval_Qnli-dev_dot_accuracy_threshold": 706.0679931640625,
+      "eval_Qnli-dev_dot_ap": 0.5806120989705038,
+      "eval_Qnli-dev_dot_f1": 0.641025641025641,
+      "eval_Qnli-dev_dot_f1_threshold": 646.50732421875,
+      "eval_Qnli-dev_dot_precision": 0.48283261802575106,
+      "eval_Qnli-dev_dot_recall": 0.9533898305084746,
+      "eval_Qnli-dev_euclidean_accuracy": 0.59765625,
+      "eval_Qnli-dev_euclidean_accuracy_threshold": 11.117534637451172,
+      "eval_Qnli-dev_euclidean_ap": 0.5806941879083722,
+      "eval_Qnli-dev_euclidean_f1": 0.6419400855920115,
+      "eval_Qnli-dev_euclidean_f1_threshold": 15.578910827636719,
+      "eval_Qnli-dev_euclidean_precision": 0.4838709677419355,
+      "eval_Qnli-dev_euclidean_recall": 0.9533898305084746,
+      "eval_Qnli-dev_manhattan_accuracy": 0.607421875,
+      "eval_Qnli-dev_manhattan_accuracy_threshold": 233.11520385742188,
+      "eval_Qnli-dev_manhattan_ap": 0.5947537164269667,
+      "eval_Qnli-dev_manhattan_f1": 0.6514459665144596,
+      "eval_Qnli-dev_manhattan_f1_threshold": 283.4456481933594,
+      "eval_Qnli-dev_manhattan_precision": 0.5083135391923991,
+      "eval_Qnli-dev_manhattan_recall": 0.9067796610169492,
+      "eval_Qnli-dev_max_accuracy": 0.607421875,
+      "eval_Qnli-dev_max_accuracy_threshold": 706.0679931640625,
+      "eval_Qnli-dev_max_ap": 0.5947537164269667,
+      "eval_Qnli-dev_max_f1": 0.6514459665144596,
+      "eval_Qnli-dev_max_f1_threshold": 646.50732421875,
+      "eval_Qnli-dev_max_precision": 0.5083135391923991,
+      "eval_Qnli-dev_max_recall": 0.9533898305084746,
+      "eval_allNLI-dev_cosine_accuracy": 0.671875,
+      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9753081798553467,
+      "eval_allNLI-dev_cosine_ap": 0.41030395214559057,
+      "eval_allNLI-dev_cosine_f1": 0.5126475548060708,
+      "eval_allNLI-dev_cosine_f1_threshold": 0.8886709213256836,
+      "eval_allNLI-dev_cosine_precision": 0.3619047619047619,
+      "eval_allNLI-dev_cosine_recall": 0.8786127167630058,
       "eval_allNLI-dev_dot_accuracy": 0.669921875,
-      "eval_allNLI-dev_dot_accuracy_threshold": 750.77294921875,
-      "eval_allNLI-dev_dot_ap": 0.41489360676575493,
-      "eval_allNLI-dev_dot_f1": 0.5078864353312303,
-      "eval_allNLI-dev_dot_f1_threshold": 647.291259765625,
-      "eval_allNLI-dev_dot_precision": 0.3492407809110629,
-      "eval_allNLI-dev_dot_recall": 0.930635838150289,
-      "eval_allNLI-dev_euclidean_accuracy": 0.669921875,
-      "eval_allNLI-dev_euclidean_accuracy_threshold": 6.270920753479004,
-      "eval_allNLI-dev_euclidean_ap": 0.4142019252328734,
-      "eval_allNLI-dev_euclidean_f1": 0.5081669691470054,
-      "eval_allNLI-dev_euclidean_f1_threshold": 13.145881652832031,
-      "eval_allNLI-dev_euclidean_precision": 0.37037037037037035,
-      "eval_allNLI-dev_euclidean_recall": 0.8092485549132948,
+      "eval_allNLI-dev_dot_accuracy_threshold": 751.201904296875,
+      "eval_allNLI-dev_dot_ap": 0.40946066450522606,
+      "eval_allNLI-dev_dot_f1": 0.5126475548060708,
+      "eval_allNLI-dev_dot_f1_threshold": 682.495849609375,
+      "eval_allNLI-dev_dot_precision": 0.3619047619047619,
+      "eval_allNLI-dev_dot_recall": 0.8786127167630058,
+      "eval_allNLI-dev_euclidean_accuracy": 0.671875,
+      "eval_allNLI-dev_euclidean_accuracy_threshold": 6.158434867858887,
+      "eval_allNLI-dev_euclidean_ap": 0.4103105551716732,
+      "eval_allNLI-dev_euclidean_f1": 0.5126475548060708,
+      "eval_allNLI-dev_euclidean_f1_threshold": 13.076848983764648,
+      "eval_allNLI-dev_euclidean_precision": 0.3619047619047619,
+      "eval_allNLI-dev_euclidean_recall": 0.8786127167630058,
       "eval_allNLI-dev_manhattan_accuracy": 0.66796875,
-      "eval_allNLI-dev_manhattan_accuracy_threshold": 108.75921630859375,
-      "eval_allNLI-dev_manhattan_ap": 0.41121266093840314,
-      "eval_allNLI-dev_manhattan_f1": 0.5156794425087108,
-      "eval_allNLI-dev_manhattan_f1_threshold": 267.4344787597656,
-      "eval_allNLI-dev_manhattan_precision": 0.3690773067331671,
-      "eval_allNLI-dev_manhattan_recall": 0.8554913294797688,
-      "eval_allNLI-dev_max_accuracy": 0.669921875,
-      "eval_allNLI-dev_max_accuracy_threshold": 750.77294921875,
-      "eval_allNLI-dev_max_ap": 0.41489360676575493,
-      "eval_allNLI-dev_max_f1": 0.5156794425087108,
-      "eval_allNLI-dev_max_f1_threshold": 647.291259765625,
-      "eval_allNLI-dev_max_precision": 0.37037037037037035,
-      "eval_allNLI-dev_max_recall": 0.930635838150289,
-      "eval_loss": 2.382014036178589,
-      "eval_runtime": 54.4017,
-      "eval_samples_per_second": 30.587,
-      "eval_sequential_score": 0.6114921536490292,
-      "eval_steps_per_second": 0.129,
-      "eval_sts-test_pearson_cosine": 0.21416737125119828,
-      "eval_sts-test_pearson_dot": 0.2130784047531441,
-      "eval_sts-test_pearson_euclidean": 0.24175068780760803,
-      "eval_sts-test_pearson_manhattan": 0.2519623781736086,
-      "eval_sts-test_pearson_max": 0.2519623781736086,
-      "eval_sts-test_spearman_cosine": 0.2708540559881937,
-      "eval_sts-test_spearman_dot": 0.26943982662570265,
-      "eval_sts-test_spearman_euclidean": 0.2709061912077147,
-      "eval_sts-test_spearman_manhattan": 0.27978654201570125,
-      "eval_sts-test_spearman_max": 0.27978654201570125,
+      "eval_allNLI-dev_manhattan_accuracy_threshold": 125.19190979003906,
+      "eval_allNLI-dev_manhattan_ap": 0.405662874748899,
+      "eval_allNLI-dev_manhattan_f1": 0.5164644714038128,
+      "eval_allNLI-dev_manhattan_f1_threshold": 245.46649169921875,
+      "eval_allNLI-dev_manhattan_precision": 0.3688118811881188,
+      "eval_allNLI-dev_manhattan_recall": 0.861271676300578,
+      "eval_allNLI-dev_max_accuracy": 0.671875,
+      "eval_allNLI-dev_max_accuracy_threshold": 751.201904296875,
+      "eval_allNLI-dev_max_ap": 0.4103105551716732,
+      "eval_allNLI-dev_max_f1": 0.5164644714038128,
+      "eval_allNLI-dev_max_f1_threshold": 682.495849609375,
+      "eval_allNLI-dev_max_precision": 0.3688118811881188,
+      "eval_allNLI-dev_max_recall": 0.8786127167630058,
+      "eval_loss": 2.4280221462249756,
+      "eval_runtime": 52.6639,
+      "eval_samples_per_second": 31.597,
+      "eval_sequential_score": 0.5947537164269667,
+      "eval_steps_per_second": 0.133,
+      "eval_sts-test_pearson_cosine": 0.20581042639062724,
+      "eval_sts-test_pearson_dot": 0.2051481762928456,
+      "eval_sts-test_pearson_euclidean": 0.22966459127158345,
+      "eval_sts-test_pearson_manhattan": 0.23576329100758983,
+      "eval_sts-test_pearson_max": 0.23576329100758983,
+      "eval_sts-test_spearman_cosine": 0.25074163083484124,
+      "eval_sts-test_spearman_dot": 0.250431638847183,
+      "eval_sts-test_spearman_euclidean": 0.2507327700850857,
+      "eval_sts-test_spearman_manhattan": 0.256412465927098,
+      "eval_sts-test_spearman_max": 0.256412465927098,
       "step": 306
     },
     {
       "epoch": 0.3021653543307087,
-      "grad_norm": 10.734902381896973,
+      "grad_norm": 16.60762596130371,
       "learning_rate": 3.001988071570577e-07,
-      "loss": 1.7923,
+      "loss": 2.6681,
       "step": 307
     },
     {
       "epoch": 0.3031496062992126,
-      "grad_norm": 10.291141510009766,
+      "grad_norm": 15.935681343078613,
       "learning_rate": 3.01192842942346e-07,
-      "loss": 1.8325,
+      "loss": 2.7219,
       "step": 308
     },
     {
       "epoch": 0.30413385826771655,
-      "grad_norm": 10.940272331237793,
+      "grad_norm": 12.386747360229492,
       "learning_rate": 3.021868787276342e-07,
-      "loss": 1.8853,
+      "loss": 2.4057,
       "step": 309
     },
     {
       "epoch": 0.3051181102362205,
-      "grad_norm": 79.87450408935547,
+      "grad_norm": 11.695796966552734,
       "learning_rate": 3.031809145129225e-07,
-      "loss": 9.9739,
+      "loss": 2.7402,
       "step": 310
     },
     {
       "epoch": 0.3061023622047244,
-      "grad_norm": 58.47280502319336,
+      "grad_norm": 14.046793937683105,
       "learning_rate": 3.0417495029821077e-07,
-      "loss": 7.9147,
+      "loss": 2.5512,
       "step": 311
     },
     {
       "epoch": 0.30708661417322836,
-      "grad_norm": 11.367313385009766,
+      "grad_norm": 11.52070426940918,
       "learning_rate": 3.051689860834991e-07,
-      "loss": 2.2901,
+      "loss": 2.8553,
       "step": 312
     },
     {
       "epoch": 0.3080708661417323,
-      "grad_norm": 11.76492691040039,
+      "grad_norm": 12.387203216552734,
       "learning_rate": 3.0616302186878733e-07,
-      "loss": 2.2322,
+      "loss": 2.598,
       "step": 313
     },
     {
       "epoch": 0.3090551181102362,
-      "grad_norm": 10.923507690429688,
+      "grad_norm": 14.028305053710938,
       "learning_rate": 3.071570576540756e-07,
-      "loss": 1.7877,
+      "loss": 2.6186,
       "step": 314
     },
     {
       "epoch": 0.31003937007874016,
-      "grad_norm": 9.680418014526367,
+      "grad_norm": 11.698431968688965,
       "learning_rate": 3.0815109343936384e-07,
-      "loss": 1.7058,
+      "loss": 2.3678,
       "step": 315
     },
     {
       "epoch": 0.3110236220472441,
-      "grad_norm": 52.868309020996094,
+      "grad_norm": 12.938248634338379,
       "learning_rate": 3.0914512922465214e-07,
-      "loss": 7.9885,
+      "loss": 2.886,
       "step": 316
     },
     {
       "epoch": 0.31200787401574803,
-      "grad_norm": 11.308423042297363,
+      "grad_norm": 12.183907508850098,
       "learning_rate": 3.1013916500994045e-07,
-      "loss": 2.0856,
+      "loss": 2.1738,
       "step": 317
     },
     {
       "epoch": 0.31299212598425197,
-      "grad_norm": 13.043806076049805,
+      "grad_norm": 15.912786483764648,
       "learning_rate": 3.1113320079522865e-07,
-      "loss": 2.0736,
+      "loss": 2.6619,
       "step": 318
     },
     {
       "epoch": 0.3139763779527559,
-      "grad_norm": 13.645163536071777,
+      "grad_norm": 10.824213981628418,
       "learning_rate": 3.1212723658051695e-07,
-      "loss": 2.4477,
+      "loss": 2.1818,
       "step": 319
     },
     {
       "epoch": 0.31496062992125984,
-      "grad_norm": 11.428119659423828,
+      "grad_norm": 13.062846183776855,
       "learning_rate": 3.131212723658052e-07,
-      "loss": 2.3418,
+      "loss": 3.0407,
       "step": 320
     },
     {
       "epoch": 0.3159448818897638,
-      "grad_norm": 12.059881210327148,
+      "grad_norm": 11.457890510559082,
       "learning_rate": 3.141153081510935e-07,
-      "loss": 2.113,
+      "loss": 2.464,
       "step": 321
     },
     {
       "epoch": 0.3169291338582677,
-      "grad_norm": 62.59613800048828,
+      "grad_norm": 12.722960472106934,
       "learning_rate": 3.151093439363818e-07,
-      "loss": 9.2682,
+      "loss": 2.7415,
       "step": 322
     },
     {
       "epoch": 0.31791338582677164,
-      "grad_norm": 61.35698699951172,
+      "grad_norm": 13.098994255065918,
       "learning_rate": 3.1610337972167e-07,
-      "loss": 8.0044,
+      "loss": 2.7455,
       "step": 323
     },
     {
       "epoch": 0.3188976377952756,
-      "grad_norm": 45.25782775878906,
+      "grad_norm": 12.980748176574707,
       "learning_rate": 3.170974155069583e-07,
-      "loss": 6.6853,
+      "loss": 2.4061,
       "step": 324
     },
     {
       "epoch": 0.3198818897637795,
-      "grad_norm": 10.384099006652832,
+      "grad_norm": 11.015922546386719,
       "learning_rate": 3.1809145129224657e-07,
-      "loss": 1.9131,
+      "loss": 2.0491,
       "step": 325
     },
     {
       "epoch": 0.32086614173228345,
-      "grad_norm": 10.898626327514648,
+      "grad_norm": 23.543014526367188,
       "learning_rate": 3.190854870775349e-07,
-      "loss": 1.9983,
+      "loss": 3.3097,
       "step": 326
     },
     {
       "epoch": 0.3218503937007874,
-      "grad_norm": 11.038727760314941,
+      "grad_norm": 13.360864639282227,
       "learning_rate": 3.200795228628231e-07,
-      "loss": 2.1731,
+      "loss": 2.3587,
       "step": 327
     },
     {
       "epoch": 0.3228346456692913,
-      "grad_norm": 11.289586067199707,
+      "grad_norm": 11.620161056518555,
       "learning_rate": 3.210735586481114e-07,
-      "loss": 1.6724,
+      "loss": 1.9493,
       "step": 328
     },
     {
       "epoch": 0.32381889763779526,
-      "grad_norm": 10.893710136413574,
+      "grad_norm": 16.858875274658203,
       "learning_rate": 3.220675944333997e-07,
-      "loss": 1.9086,
+      "loss": 2.5399,
       "step": 329
     },
     {
       "epoch": 0.3248031496062992,
-      "grad_norm": 9.912519454956055,
+      "grad_norm": 16.233020782470703,
       "learning_rate": 3.2306163021868794e-07,
-      "loss": 1.9539,
+      "loss": 2.3569,
       "step": 330
     },
     {
       "epoch": 0.3257874015748031,
-      "grad_norm": 47.91545104980469,
+      "grad_norm": 10.670516014099121,
       "learning_rate": 3.240556660039762e-07,
-      "loss": 5.7975,
+      "loss": 1.9024,
       "step": 331
     },
     {
       "epoch": 0.32677165354330706,
-      "grad_norm": 10.154540061950684,
+      "grad_norm": 13.305569648742676,
       "learning_rate": 3.2504970178926444e-07,
-      "loss": 1.8287,
+      "loss": 2.3513,
       "step": 332
     },
     {
       "epoch": 0.327755905511811,
-      "grad_norm": 9.187210083007812,
+      "grad_norm": 16.92148208618164,
       "learning_rate": 3.2604373757455275e-07,
-      "loss": 1.8369,
+      "loss": 2.2488,
       "step": 333
     },
     {
       "epoch": 0.328740157480315,
-      "grad_norm": 9.3264799118042,
+      "grad_norm": 13.141341209411621,
       "learning_rate": 3.27037773359841e-07,
-      "loss": 1.4491,
+      "loss": 1.9141,
       "step": 334
     },
     {
       "epoch": 0.3297244094488189,
-      "grad_norm": 11.264083862304688,
+      "grad_norm": 12.373048782348633,
       "learning_rate": 3.280318091451293e-07,
-      "loss": 1.9481,
+      "loss": 2.7065,
       "step": 335
     },
     {
       "epoch": 0.33070866141732286,
-      "grad_norm": 56.167755126953125,
+      "grad_norm": 11.926918029785156,
       "learning_rate": 3.2902584493041756e-07,
-      "loss": 6.3526,
+      "loss": 2.139,
       "step": 336
     },
     {
       "epoch": 0.3316929133858268,
-      "grad_norm": 11.542598724365234,
+      "grad_norm": 14.438637733459473,
       "learning_rate": 3.300198807157058e-07,
-      "loss": 2.2364,
+      "loss": 2.2345,
       "step": 337
     },
     {
       "epoch": 0.33267716535433073,
-      "grad_norm": 9.792985916137695,
+      "grad_norm": 11.567721366882324,
       "learning_rate": 3.310139165009941e-07,
-      "loss": 1.7361,
+      "loss": 2.3612,
       "step": 338
     },
     {
       "epoch": 0.33366141732283466,
-      "grad_norm": 11.404271125793457,
+      "grad_norm": 11.467390060424805,
       "learning_rate": 3.3200795228628237e-07,
-      "loss": 2.0384,
+      "loss": 2.1413,
       "step": 339
     },
     {
       "epoch": 0.3346456692913386,
-      "grad_norm": 10.791166305541992,
+      "grad_norm": 13.05380630493164,
       "learning_rate": 3.330019880715706e-07,
-      "loss": 1.7119,
+      "loss": 2.2214,
       "step": 340
     },
     {
       "epoch": 0.33562992125984253,
-      "grad_norm": 11.242233276367188,
+      "grad_norm": 13.208922386169434,
       "learning_rate": 3.3399602385685887e-07,
-      "loss": 2.0796,
+      "loss": 2.9006,
       "step": 341
     },
     {
       "epoch": 0.33661417322834647,
-      "grad_norm": 8.66325569152832,
+      "grad_norm": 12.578964233398438,
       "learning_rate": 3.349900596421472e-07,
-      "loss": 1.423,
+      "loss": 2.417,
       "step": 342
     },
     {
       "epoch": 0.3375984251968504,
-      "grad_norm": 49.26806640625,
+      "grad_norm": 11.79666519165039,
       "learning_rate": 3.359840954274355e-07,
-      "loss": 7.3127,
+      "loss": 2.2348,
       "step": 343
     },
     {
       "epoch": 0.33858267716535434,
-      "grad_norm": 12.85938549041748,
+      "grad_norm": 12.465178489685059,
       "learning_rate": 3.369781312127237e-07,
-      "loss": 1.7333,
+      "loss": 2.4369,
       "step": 344
     },
     {
       "epoch": 0.3395669291338583,
-      "grad_norm": 11.912487030029297,
+      "grad_norm": 16.569061279296875,
       "learning_rate": 3.37972166998012e-07,
-      "loss": 2.0256,
+      "loss": 2.7623,
       "step": 345
     },
     {
       "epoch": 0.3405511811023622,
-      "grad_norm": 13.381354331970215,
+      "grad_norm": 18.68236541748047,
       "learning_rate": 3.3896620278330024e-07,
-      "loss": 2.281,
+      "loss": 2.6741,
       "step": 346
     },
     {
       "epoch": 0.34153543307086615,
-      "grad_norm": 11.3228178024292,
+      "grad_norm": 22.597843170166016,
       "learning_rate": 3.3996023856858855e-07,
-      "loss": 2.6135,
+      "loss": 3.0515,
       "step": 347
     },
     {
       "epoch": 0.3425196850393701,
-      "grad_norm": 13.237906455993652,
+      "grad_norm": 13.142956733703613,
       "learning_rate": 3.409542743538768e-07,
-      "loss": 1.7188,
+      "loss": 2.4952,
       "step": 348
     },
     {
       "epoch": 0.343503937007874,
-      "grad_norm": 51.69926071166992,
+      "grad_norm": 10.252715110778809,
       "learning_rate": 3.4194831013916505e-07,
-      "loss": 7.2075,
+      "loss": 2.1265,
       "step": 349
     },
     {
       "epoch": 0.34448818897637795,
-      "grad_norm": 14.641998291015625,
+      "grad_norm": 10.546085357666016,
       "learning_rate": 3.4294234592445336e-07,
-      "loss": 2.2859,
+      "loss": 2.0359,
       "step": 350
     },
     {
       "epoch": 0.3454724409448819,
-      "grad_norm": 12.565581321716309,
+      "grad_norm": 15.560839653015137,
       "learning_rate": 3.439363817097416e-07,
-      "loss": 2.3042,
+      "loss": 3.107,
       "step": 351
     },
     {
       "epoch": 0.3464566929133858,
-      "grad_norm": 11.288939476013184,
+      "grad_norm": 10.87291145324707,
       "learning_rate": 3.449304174950299e-07,
-      "loss": 1.8944,
+      "loss": 2.116,
       "step": 352
     },
     {
       "epoch": 0.34744094488188976,
-      "grad_norm": 35.340904235839844,
+      "grad_norm": 11.75866413116455,
       "learning_rate": 3.459244532803181e-07,
-      "loss": 3.2227,
+      "loss": 2.1996,
       "step": 353
     },
     {
       "epoch": 0.3484251968503937,
-      "grad_norm": 12.464967727661133,
+      "grad_norm": 14.872318267822266,
       "learning_rate": 3.469184890656064e-07,
-      "loss": 1.9706,
+      "loss": 2.9312,
       "step": 354
     },
     {
       "epoch": 0.3494094488188976,
-      "grad_norm": 10.959451675415039,
+      "grad_norm": 14.699666023254395,
       "learning_rate": 3.4791252485089467e-07,
-      "loss": 1.4694,
+      "loss": 2.2885,
       "step": 355
     },
     {
       "epoch": 0.35039370078740156,
-      "grad_norm": 13.121545791625977,
+      "grad_norm": 17.621458053588867,
       "learning_rate": 3.48906560636183e-07,
-      "loss": 2.6948,
+      "loss": 3.0302,
       "step": 356
     },
     {
       "epoch": 0.3513779527559055,
-      "grad_norm": 52.15006637573242,
+      "grad_norm": 12.756147384643555,
       "learning_rate": 3.499005964214713e-07,
-      "loss": 7.2386,
+      "loss": 2.2163,
       "step": 357
     },
     {
       "epoch": 0.35236220472440943,
-      "grad_norm": 53.013832092285156,
+      "grad_norm": 14.842194557189941,
       "learning_rate": 3.508946322067595e-07,
-      "loss": 6.7146,
+      "loss": 2.8304,
       "step": 358
     },
     {
       "epoch": 0.35334645669291337,
-      "grad_norm": 56.041343688964844,
+      "grad_norm": 11.481346130371094,
       "learning_rate": 3.518886679920478e-07,
-      "loss": 7.1446,
+      "loss": 2.2715,
       "step": 359
     },
     {
       "epoch": 0.3543307086614173,
-      "grad_norm": 13.036025047302246,
+      "grad_norm": 11.839008331298828,
       "learning_rate": 3.5288270377733604e-07,
-      "loss": 2.1246,
+      "loss": 2.3388,
       "step": 360
     },
     {
       "epoch": 0.35531496062992124,
-      "grad_norm": 12.321011543273926,
+      "grad_norm": 12.406773567199707,
       "learning_rate": 3.5387673956262434e-07,
-      "loss": 2.2676,
+      "loss": 2.2098,
       "step": 361
     },
     {
       "epoch": 0.3562992125984252,
-      "grad_norm": 12.051976203918457,
+      "grad_norm": 11.005632400512695,
       "learning_rate": 3.5487077534791254e-07,
-      "loss": 1.8415,
+      "loss": 2.0911,
       "step": 362
     },
     {
       "epoch": 0.3572834645669291,
-      "grad_norm": 40.2060661315918,
+      "grad_norm": 11.084342002868652,
       "learning_rate": 3.5586481113320085e-07,
-      "loss": 5.9398,
+      "loss": 2.3582,
       "step": 363
     },
     {
       "epoch": 0.35826771653543305,
-      "grad_norm": 11.134186744689941,
+      "grad_norm": 10.120821952819824,
       "learning_rate": 3.5685884691848915e-07,
-      "loss": 2.022,
+      "loss": 1.8605,
       "step": 364
     },
     {
       "epoch": 0.359251968503937,
-      "grad_norm": 13.448582649230957,
+      "grad_norm": 13.282901763916016,
       "learning_rate": 3.578528827037774e-07,
-      "loss": 2.4997,
+      "loss": 2.2252,
       "step": 365
     },
     {
       "epoch": 0.36023622047244097,
-      "grad_norm": 9.906368255615234,
+      "grad_norm": 13.67626667022705,
       "learning_rate": 3.5884691848906566e-07,
-      "loss": 1.8627,
+      "loss": 2.2018,
       "step": 366
     },
     {
       "epoch": 0.3612204724409449,
-      "grad_norm": 12.35049819946289,
+      "grad_norm": 13.320021629333496,
       "learning_rate": 3.598409542743539e-07,
-      "loss": 2.0428,
+      "loss": 2.1099,
       "step": 367
     },
     {
       "epoch": 0.36220472440944884,
-      "grad_norm": 11.315701484680176,
+      "grad_norm": 11.790545463562012,
       "learning_rate": 3.608349900596422e-07,
-      "loss": 1.9337,
+      "loss": 2.1323,
       "step": 368
     },
     {
       "epoch": 0.3631889763779528,
-      "grad_norm": 12.36326789855957,
+      "grad_norm": 13.921866416931152,
       "learning_rate": 3.6182902584493047e-07,
-      "loss": 2.2796,
+      "loss": 2.4203,
       "step": 369
     },
     {
       "epoch": 0.3641732283464567,
-      "grad_norm": 12.894477844238281,
+      "grad_norm": 23.001510620117188,
       "learning_rate": 3.6282306163021877e-07,
-      "loss": 2.1442,
+      "loss": 2.7768,
       "step": 370
     },
     {
       "epoch": 0.36515748031496065,
-      "grad_norm": 11.680022239685059,
+      "grad_norm": 14.317642211914062,
       "learning_rate": 3.63817097415507e-07,
-      "loss": 1.856,
+      "loss": 2.3359,
       "step": 371
     },
     {
       "epoch": 0.3661417322834646,
-      "grad_norm": 35.735713958740234,
+      "grad_norm": 16.946765899658203,
       "learning_rate": 3.648111332007953e-07,
-      "loss": 5.1339,
+      "loss": 2.3773,
       "step": 372
     },
     {
       "epoch": 0.3671259842519685,
-      "grad_norm": 14.422959327697754,
+      "grad_norm": 12.836204528808594,
       "learning_rate": 3.658051689860836e-07,
-      "loss": 2.5143,
+      "loss": 2.4424,
       "step": 373
     },
     {
       "epoch": 0.36811023622047245,
-      "grad_norm": 10.153562545776367,
+      "grad_norm": 13.058905601501465,
       "learning_rate": 3.6679920477137184e-07,
-      "loss": 1.3353,
+      "loss": 1.9478,
       "step": 374
     },
     {
       "epoch": 0.3690944881889764,
-      "grad_norm": 9.560186386108398,
+      "grad_norm": 15.509428977966309,
       "learning_rate": 3.677932405566601e-07,
-      "loss": 1.1054,
+      "loss": 1.6047,
       "step": 375
     },
     {
       "epoch": 0.3700787401574803,
-      "grad_norm": 52.342227935791016,
+      "grad_norm": 11.629678726196289,
       "learning_rate": 3.6878727634194834e-07,
-      "loss": 5.1893,
+      "loss": 1.7384,
       "step": 376
     },
     {
       "epoch": 0.37106299212598426,
-      "grad_norm": 9.996565818786621,
+      "grad_norm": 13.820512771606445,
       "learning_rate": 3.6978131212723665e-07,
-      "loss": 1.5849,
+      "loss": 2.1147,
       "step": 377
     },
     {
       "epoch": 0.3720472440944882,
-      "grad_norm": 10.249340057373047,
+      "grad_norm": 11.474569320678711,
       "learning_rate": 3.7077534791252495e-07,
-      "loss": 1.7363,
+      "loss": 1.8449,
       "step": 378
     },
     {
       "epoch": 0.37303149606299213,
-      "grad_norm": 54.82731628417969,
+      "grad_norm": 15.554925918579102,
       "learning_rate": 3.7176938369781315e-07,
-      "loss": 6.3385,
+      "loss": 2.6009,
       "step": 379
     },
     {
       "epoch": 0.37401574803149606,
-      "grad_norm": 11.697839736938477,
+      "grad_norm": 13.037839889526367,
       "learning_rate": 3.7276341948310145e-07,
-      "loss": 1.7896,
+      "loss": 2.4051,
       "step": 380
     },
     {
       "epoch": 0.375,
-      "grad_norm": 10.580652236938477,
+      "grad_norm": 12.058863639831543,
       "learning_rate": 3.737574552683897e-07,
-      "loss": 1.8278,
+      "loss": 2.3035,
       "step": 381
     },
     {
       "epoch": 0.37598425196850394,
-      "grad_norm": 9.671079635620117,
+      "grad_norm": 11.572028160095215,
       "learning_rate": 3.74751491053678e-07,
-      "loss": 1.3783,
+      "loss": 1.8955,
       "step": 382
     },
     {
       "epoch": 0.37696850393700787,
-      "grad_norm": 10.716227531433105,
+      "grad_norm": 16.923925399780273,
       "learning_rate": 3.7574552683896626e-07,
-      "loss": 1.5611,
+      "loss": 2.287,
       "step": 383
     },
     {
       "epoch": 0.3779527559055118,
-      "grad_norm": 50.10944366455078,
+      "grad_norm": 10.901126861572266,
       "learning_rate": 3.767395626242545e-07,
-      "loss": 6.6285,
+      "loss": 1.9123,
       "step": 384
     },
     {
       "epoch": 0.37893700787401574,
-      "grad_norm": 11.59723949432373,
+      "grad_norm": 10.82137680053711,
       "learning_rate": 3.777335984095428e-07,
-      "loss": 1.54,
+      "loss": 1.9369,
       "step": 385
     },
     {
       "epoch": 0.3799212598425197,
-      "grad_norm": 11.77585220336914,
+      "grad_norm": 11.884252548217773,
       "learning_rate": 3.787276341948311e-07,
-      "loss": 1.4216,
+      "loss": 2.1367,
       "step": 386
     },
     {
       "epoch": 0.3809055118110236,
-      "grad_norm": 45.65110397338867,
+      "grad_norm": 12.42955207824707,
       "learning_rate": 3.797216699801194e-07,
-      "loss": 5.1756,
+      "loss": 1.9437,
       "step": 387
     },
     {
       "epoch": 0.38188976377952755,
-      "grad_norm": 13.375621795654297,
+      "grad_norm": 13.655105590820312,
       "learning_rate": 3.807157057654076e-07,
-      "loss": 2.2009,
+      "loss": 2.3873,
       "step": 388
     },
     {
       "epoch": 0.3828740157480315,
-      "grad_norm": 10.92442512512207,
+      "grad_norm": 11.53514575958252,
       "learning_rate": 3.817097415506959e-07,
-      "loss": 1.8205,
+      "loss": 1.7463,
       "step": 389
     },
     {
       "epoch": 0.3838582677165354,
-      "grad_norm": 14.326327323913574,
+      "grad_norm": 22.04655647277832,
       "learning_rate": 3.8270377733598414e-07,
-      "loss": 2.437,
+      "loss": 2.8438,
       "step": 390
     },
     {
       "epoch": 0.38484251968503935,
-      "grad_norm": 12.294797897338867,
+      "grad_norm": 17.399572372436523,
       "learning_rate": 3.8369781312127244e-07,
-      "loss": 1.9934,
+      "loss": 2.4875,
       "step": 391
     },
     {
       "epoch": 0.3858267716535433,
-      "grad_norm": 11.739425659179688,
+      "grad_norm": 13.103324890136719,
       "learning_rate": 3.8469184890656075e-07,
-      "loss": 1.9723,
+      "loss": 2.0798,
       "step": 392
     },
     {
       "epoch": 0.3868110236220472,
-      "grad_norm": 12.739344596862793,
+      "grad_norm": 13.370649337768555,
       "learning_rate": 3.8568588469184895e-07,
-      "loss": 1.7051,
+      "loss": 2.2242,
       "step": 393
     },
     {
       "epoch": 0.38779527559055116,
-      "grad_norm": 11.261693954467773,
+      "grad_norm": 12.390323638916016,
       "learning_rate": 3.8667992047713725e-07,
-      "loss": 1.7474,
+      "loss": 1.8714,
       "step": 394
     },
     {
       "epoch": 0.3887795275590551,
-      "grad_norm": 39.55622100830078,
+      "grad_norm": 13.5247802734375,
       "learning_rate": 3.876739562624255e-07,
-      "loss": 5.1513,
+      "loss": 1.5893,
       "step": 395
     },
     {
       "epoch": 0.38976377952755903,
-      "grad_norm": 12.222193717956543,
+      "grad_norm": 10.50505542755127,
       "learning_rate": 3.886679920477138e-07,
-      "loss": 1.6779,
+      "loss": 1.5633,
       "step": 396
     },
     {
       "epoch": 0.390748031496063,
-      "grad_norm": 9.99728012084961,
+      "grad_norm": 11.619284629821777,
       "learning_rate": 3.89662027833002e-07,
-      "loss": 1.4055,
+      "loss": 1.8645,
       "step": 397
     },
     {
       "epoch": 0.39173228346456695,
-      "grad_norm": 11.975881576538086,
+      "grad_norm": 11.723767280578613,
       "learning_rate": 3.906560636182903e-07,
-      "loss": 1.6386,
+      "loss": 1.8928,
       "step": 398
     },
     {
       "epoch": 0.3927165354330709,
-      "grad_norm": 10.65119457244873,
+      "grad_norm": 9.949848175048828,
       "learning_rate": 3.916500994035786e-07,
-      "loss": 1.2141,
+      "loss": 1.3352,
       "step": 399
     },
     {
       "epoch": 0.3937007874015748,
-      "grad_norm": 45.11775588989258,
+      "grad_norm": 30.210622787475586,
       "learning_rate": 3.9264413518886687e-07,
-      "loss": 6.2437,
+      "loss": 3.3052,
       "step": 400
     },
     {
       "epoch": 0.39468503937007876,
-      "grad_norm": 11.461906433105469,
+      "grad_norm": 12.481863975524902,
       "learning_rate": 3.936381709741551e-07,
-      "loss": 1.3342,
+      "loss": 1.6256,
       "step": 401
     },
     {
       "epoch": 0.3956692913385827,
-      "grad_norm": 11.590656280517578,
+      "grad_norm": 12.854496955871582,
       "learning_rate": 3.946322067594434e-07,
-      "loss": 1.5153,
+      "loss": 1.8856,
       "step": 402
     },
     {
       "epoch": 0.39665354330708663,
-      "grad_norm": 11.251949310302734,
+      "grad_norm": 11.837615013122559,
       "learning_rate": 3.956262425447317e-07,
-      "loss": 1.4861,
+      "loss": 1.8355,
       "step": 403
     },
     {
       "epoch": 0.39763779527559057,
-      "grad_norm": 11.35289478302002,
+      "grad_norm": 14.314859390258789,
       "learning_rate": 3.9662027833001993e-07,
-      "loss": 1.6596,
+      "loss": 1.8944,
       "step": 404
     },
     {
       "epoch": 0.3986220472440945,
-      "grad_norm": 57.254600524902344,
+      "grad_norm": 11.3548002243042,
       "learning_rate": 3.9761431411530824e-07,
-      "loss": 6.3475,
+      "loss": 1.7636,
       "step": 405
     },
     {
       "epoch": 0.39960629921259844,
-      "grad_norm": 13.950235366821289,
+      "grad_norm": 25.222524642944336,
       "learning_rate": 3.986083499005965e-07,
-      "loss": 1.7669,
+      "loss": 2.8097,
       "step": 406
     },
     {
       "epoch": 0.4005905511811024,
-      "grad_norm": 12.42131519317627,
+      "grad_norm": 13.39458179473877,
       "learning_rate": 3.9960238568588474e-07,
-      "loss": 1.5388,
+      "loss": 1.9121,
       "step": 407
     },
     {
       "epoch": 0.4015748031496063,
-      "grad_norm": 11.60079288482666,
+      "grad_norm": 12.579276084899902,
       "learning_rate": 4.0059642147117305e-07,
-      "loss": 1.6276,
+      "loss": 1.9233,
       "step": 408
     },
     {
       "epoch": 0.40255905511811024,
-      "grad_norm": 12.122014999389648,
+      "grad_norm": 12.75617504119873,
       "learning_rate": 4.015904572564613e-07,
-      "loss": 1.4932,
+      "loss": 1.543,
       "step": 409
     },
     {
       "epoch": 0.4035433070866142,
-      "grad_norm": 63.547462463378906,
+      "grad_norm": 13.689290046691895,
       "learning_rate": 4.0258449304174955e-07,
-      "loss": 7.097,
+      "loss": 1.7207,
       "step": 410
     },
     {
       "epoch": 0.4045275590551181,
-      "grad_norm": 11.58280086517334,
+      "grad_norm": 11.717874526977539,
       "learning_rate": 4.035785288270378e-07,
-      "loss": 1.4621,
+      "loss": 1.6344,
       "step": 411
     },
     {
       "epoch": 0.40551181102362205,
-      "grad_norm": 44.8221549987793,
+      "grad_norm": 22.84028434753418,
       "learning_rate": 4.045725646123261e-07,
-      "loss": 5.6547,
+      "loss": 2.4177,
       "step": 412
     },
     {
       "epoch": 0.406496062992126,
-      "grad_norm": 44.49589157104492,
+      "grad_norm": 15.062310218811035,
       "learning_rate": 4.055666003976144e-07,
-      "loss": 5.7338,
+      "loss": 2.2995,
       "step": 413
     },
     {
       "epoch": 0.4074803149606299,
-      "grad_norm": 11.479046821594238,
+      "grad_norm": 13.126559257507324,
       "learning_rate": 4.065606361829026e-07,
-      "loss": 1.2834,
+      "loss": 1.7681,
       "step": 414
     },
     {
       "epoch": 0.40846456692913385,
-      "grad_norm": 12.855639457702637,
+      "grad_norm": 11.371601104736328,
       "learning_rate": 4.075546719681909e-07,
-      "loss": 1.8426,
+      "loss": 1.6562,
       "step": 415
     },
     {
       "epoch": 0.4094488188976378,
-      "grad_norm": 14.07402229309082,
+      "grad_norm": 13.057047843933105,
       "learning_rate": 4.085487077534792e-07,
-      "loss": 1.6923,
+      "loss": 1.8896,
       "step": 416
     },
     {
       "epoch": 0.4104330708661417,
-      "grad_norm": 11.612604141235352,
+      "grad_norm": 14.517732620239258,
       "learning_rate": 4.095427435387675e-07,
-      "loss": 1.7412,
+      "loss": 2.0671,
       "step": 417
     },
     {
       "epoch": 0.41141732283464566,
-      "grad_norm": 13.344563484191895,
+      "grad_norm": 15.08285903930664,
       "learning_rate": 4.1053677932405573e-07,
-      "loss": 1.5589,
+      "loss": 1.6097,
       "step": 418
     },
     {
       "epoch": 0.4124015748031496,
-      "grad_norm": 11.442570686340332,
+      "grad_norm": 23.114534378051758,
       "learning_rate": 4.11530815109344e-07,
-      "loss": 1.2435,
+      "loss": 2.8126,
       "step": 419
     },
     {
       "epoch": 0.41338582677165353,
-      "grad_norm": 12.59207820892334,
+      "grad_norm": 12.036443710327148,
       "learning_rate": 4.125248508946323e-07,
-      "loss": 1.9214,
+      "loss": 1.7028,
       "step": 420
     },
     {
       "epoch": 0.41437007874015747,
-      "grad_norm": 10.514676094055176,
+      "grad_norm": 11.2406644821167,
       "learning_rate": 4.1351888667992054e-07,
-      "loss": 1.432,
+      "loss": 1.526,
       "step": 421
     },
     {
       "epoch": 0.4153543307086614,
-      "grad_norm": 11.39521598815918,
+      "grad_norm": 23.773788452148438,
       "learning_rate": 4.1451292246520885e-07,
-      "loss": 1.643,
+      "loss": 2.5029,
       "step": 422
     },
     {
       "epoch": 0.41633858267716534,
-      "grad_norm": 12.807600021362305,
+      "grad_norm": 13.637556076049805,
       "learning_rate": 4.1550695825049705e-07,
-      "loss": 1.7221,
+      "loss": 1.7668,
       "step": 423
     },
     {
       "epoch": 0.41732283464566927,
-      "grad_norm": 11.254819869995117,
+      "grad_norm": 17.130098342895508,
       "learning_rate": 4.1650099403578535e-07,
-      "loss": 1.6566,
+      "loss": 1.9065,
       "step": 424
     },
     {
       "epoch": 0.4183070866141732,
-      "grad_norm": 47.7083625793457,
+      "grad_norm": 13.371904373168945,
       "learning_rate": 4.174950298210736e-07,
-      "loss": 5.7164,
+      "loss": 1.6645,
       "step": 425
     },
     {
       "epoch": 0.41929133858267714,
-      "grad_norm": 10.80250072479248,
+      "grad_norm": 12.900718688964844,
       "learning_rate": 4.184890656063619e-07,
-      "loss": 1.3455,
+      "loss": 1.8075,
       "step": 426
     },
     {
       "epoch": 0.4202755905511811,
-      "grad_norm": 10.916194915771484,
+      "grad_norm": 12.189275741577148,
       "learning_rate": 4.194831013916502e-07,
-      "loss": 1.2572,
+      "loss": 1.872,
       "step": 427
     },
     {
       "epoch": 0.421259842519685,
-      "grad_norm": 11.888566017150879,
+      "grad_norm": 15.679601669311523,
       "learning_rate": 4.204771371769384e-07,
-      "loss": 1.3673,
+      "loss": 2.0487,
       "step": 428
     },
     {
       "epoch": 0.422244094488189,
-      "grad_norm": 37.930355072021484,
+      "grad_norm": 11.605814933776855,
       "learning_rate": 4.214711729622267e-07,
-      "loss": 4.6922,
+      "loss": 1.535,
       "step": 429
     },
     {
       "epoch": 0.42322834645669294,
-      "grad_norm": 13.136495590209961,
+      "grad_norm": 11.776004791259766,
       "learning_rate": 4.2246520874751497e-07,
-      "loss": 1.5377,
+      "loss": 1.8046,
       "step": 430
     },
     {
       "epoch": 0.4242125984251969,
-      "grad_norm": 50.90897750854492,
+      "grad_norm": 14.196041107177734,
       "learning_rate": 4.234592445328033e-07,
-      "loss": 6.0487,
+      "loss": 2.2561,
       "step": 431
     },
     {
       "epoch": 0.4251968503937008,
-      "grad_norm": 57.34842300415039,
+      "grad_norm": 13.536731719970703,
       "learning_rate": 4.244532803180915e-07,
-      "loss": 6.3802,
+      "loss": 2.0306,
       "step": 432
     },
     {
       "epoch": 0.42618110236220474,
-      "grad_norm": 44.35916519165039,
+      "grad_norm": 10.69434642791748,
       "learning_rate": 4.254473161033798e-07,
-      "loss": 5.7241,
+      "loss": 2.1311,
       "step": 433
     },
     {
       "epoch": 0.4271653543307087,
-      "grad_norm": 39.626949310302734,
+      "grad_norm": 14.568777084350586,
       "learning_rate": 4.264413518886681e-07,
-      "loss": 4.4259,
+      "loss": 2.3013,
       "step": 434
     },
     {
       "epoch": 0.4281496062992126,
-      "grad_norm": 11.860918045043945,
+      "grad_norm": 11.407812118530273,
       "learning_rate": 4.2743538767395634e-07,
-      "loss": 1.4789,
+      "loss": 1.6402,
       "step": 435
     },
     {
       "epoch": 0.42913385826771655,
-      "grad_norm": 13.54674243927002,
+      "grad_norm": 13.241829872131348,
       "learning_rate": 4.284294234592446e-07,
-      "loss": 2.1121,
+      "loss": 1.9572,
       "step": 436
     },
     {
       "epoch": 0.4301181102362205,
-      "grad_norm": 10.985451698303223,
+      "grad_norm": 11.766160011291504,
       "learning_rate": 4.2942345924453284e-07,
-      "loss": 1.5906,
+      "loss": 1.6364,
       "step": 437
     },
     {
       "epoch": 0.4311023622047244,
-      "grad_norm": 12.662684440612793,
+      "grad_norm": 12.111750602722168,
       "learning_rate": 4.3041749502982115e-07,
-      "loss": 1.6671,
+      "loss": 1.446,
       "step": 438
     },
     {
       "epoch": 0.43208661417322836,
-      "grad_norm": 19.172922134399414,
+      "grad_norm": 12.901723861694336,
       "learning_rate": 4.3141153081510945e-07,
-      "loss": 1.5505,
+      "loss": 1.6009,
       "step": 439
     },
     {
       "epoch": 0.4330708661417323,
-      "grad_norm": 28.78730583190918,
+      "grad_norm": 13.12215518951416,
       "learning_rate": 4.324055666003977e-07,
-      "loss": 3.6253,
+      "loss": 1.9469,
       "step": 440
     },
     {
       "epoch": 0.4340551181102362,
-      "grad_norm": 12.55512523651123,
+      "grad_norm": 14.054530143737793,
       "learning_rate": 4.3339960238568596e-07,
-      "loss": 2.1088,
+      "loss": 2.1951,
       "step": 441
     },
     {
       "epoch": 0.43503937007874016,
-      "grad_norm": 10.92598819732666,
+      "grad_norm": 13.3843412399292,
       "learning_rate": 4.343936381709742e-07,
-      "loss": 1.8861,
+      "loss": 1.675,
       "step": 442
     },
     {
       "epoch": 0.4360236220472441,
-      "grad_norm": 10.909286499023438,
+      "grad_norm": 13.083832740783691,
       "learning_rate": 4.353876739562625e-07,
-      "loss": 1.377,
+      "loss": 1.4182,
       "step": 443
     },
     {
       "epoch": 0.43700787401574803,
-      "grad_norm": 11.649571418762207,
+      "grad_norm": 13.504453659057617,
       "learning_rate": 4.3638170974155077e-07,
-      "loss": 1.7832,
+      "loss": 2.2317,
       "step": 444
     },
     {
       "epoch": 0.43799212598425197,
-      "grad_norm": 12.003958702087402,
+      "grad_norm": 13.614442825317383,
       "learning_rate": 4.37375745526839e-07,
-      "loss": 2.1039,
+      "loss": 2.1076,
       "step": 445
     },
     {
       "epoch": 0.4389763779527559,
-      "grad_norm": 11.032736778259277,
+      "grad_norm": 12.694330215454102,
       "learning_rate": 4.383697813121273e-07,
-      "loss": 1.5845,
+      "loss": 1.6691,
       "step": 446
     },
     {
       "epoch": 0.43996062992125984,
-      "grad_norm": 10.186923027038574,
+      "grad_norm": 12.593806266784668,
       "learning_rate": 4.393638170974156e-07,
-      "loss": 1.5359,
+      "loss": 1.6909,
       "step": 447
     },
     {
       "epoch": 0.4409448818897638,
-      "grad_norm": 12.455483436584473,
+      "grad_norm": 37.6284065246582,
       "learning_rate": 4.403578528827039e-07,
-      "loss": 2.0012,
+      "loss": 3.1056,
       "step": 448
     },
     {
       "epoch": 0.4419291338582677,
-      "grad_norm": 29.561321258544922,
+      "grad_norm": 12.051369667053223,
       "learning_rate": 4.413518886679921e-07,
-      "loss": 3.5535,
+      "loss": 1.4069,
       "step": 449
     },
     {
       "epoch": 0.44291338582677164,
-      "grad_norm": 13.571199417114258,
+      "grad_norm": 15.523516654968262,
       "learning_rate": 4.423459244532804e-07,
-      "loss": 1.8301,
+      "loss": 2.1639,
       "step": 450
     },
     {
       "epoch": 0.4438976377952756,
-      "grad_norm": 12.335179328918457,
+      "grad_norm": 11.322219848632812,
       "learning_rate": 4.4333996023856864e-07,
-      "loss": 1.5625,
+      "loss": 1.5531,
       "step": 451
     },
     {
       "epoch": 0.4448818897637795,
-      "grad_norm": 14.14587116241455,
+      "grad_norm": 12.900222778320312,
       "learning_rate": 4.4433399602385695e-07,
-      "loss": 1.9825,
+      "loss": 2.1895,
       "step": 452
     },
     {
       "epoch": 0.44586614173228345,
-      "grad_norm": 11.227643013000488,
+      "grad_norm": 14.88873291015625,
       "learning_rate": 4.4532803180914525e-07,
-      "loss": 1.404,
+      "loss": 1.9384,
       "step": 453
     },
     {
       "epoch": 0.4468503937007874,
-      "grad_norm": 11.070554733276367,
+      "grad_norm": 13.40294075012207,
       "learning_rate": 4.4632206759443345e-07,
-      "loss": 1.1998,
+      "loss": 1.7761,
       "step": 454
     },
     {
       "epoch": 0.4478346456692913,
-      "grad_norm": 35.71782302856445,
+      "grad_norm": 27.584877014160156,
       "learning_rate": 4.4731610337972176e-07,
-      "loss": 3.9972,
+      "loss": 2.8286,
       "step": 455
     },
     {
       "epoch": 0.44881889763779526,
-      "grad_norm": 13.148333549499512,
+      "grad_norm": 25.157712936401367,
       "learning_rate": 4.4831013916501e-07,
-      "loss": 2.01,
+      "loss": 2.4877,
       "step": 456
     },
     {
       "epoch": 0.4498031496062992,
-      "grad_norm": 11.549042701721191,
+      "grad_norm": 22.14501953125,
       "learning_rate": 4.493041749502983e-07,
-      "loss": 1.3631,
+      "loss": 1.7636,
       "step": 457
     },
     {
       "epoch": 0.4507874015748031,
-      "grad_norm": 12.581208229064941,
+      "grad_norm": 10.37475872039795,
       "learning_rate": 4.502982107355865e-07,
-      "loss": 1.3897,
+      "loss": 1.1849,
       "step": 458
     },
     {
       "epoch": 0.45177165354330706,
-      "grad_norm": 34.60251235961914,
+      "grad_norm": 14.996706008911133,
       "learning_rate": 4.512922465208748e-07,
-      "loss": 3.3806,
+      "loss": 1.8331,
       "step": 459
     },
     {
       "epoch": 0.45177165354330706,
-      "eval_Qnli-dev_cosine_accuracy": 0.623046875,
-      "eval_Qnli-dev_cosine_accuracy_threshold": 0.9098526239395142,
-      "eval_Qnli-dev_cosine_ap": 0.6152473349893568,
-      "eval_Qnli-dev_cosine_f1": 0.6391184573002756,
-      "eval_Qnli-dev_cosine_f1_threshold": 0.7495293617248535,
-      "eval_Qnli-dev_cosine_precision": 0.47346938775510206,
-      "eval_Qnli-dev_cosine_recall": 0.9830508474576272,
-      "eval_Qnli-dev_dot_accuracy": 0.625,
-      "eval_Qnli-dev_dot_accuracy_threshold": 698.5355224609375,
-      "eval_Qnli-dev_dot_ap": 0.6174855915379028,
-      "eval_Qnli-dev_dot_f1": 0.6392318244170097,
-      "eval_Qnli-dev_dot_f1_threshold": 571.7469482421875,
-      "eval_Qnli-dev_dot_precision": 0.4726166328600406,
-      "eval_Qnli-dev_dot_recall": 0.9872881355932204,
-      "eval_Qnli-dev_euclidean_accuracy": 0.623046875,
-      "eval_Qnli-dev_euclidean_accuracy_threshold": 11.752104759216309,
-      "eval_Qnli-dev_euclidean_ap": 0.6152099088218845,
-      "eval_Qnli-dev_euclidean_f1": 0.6392318244170097,
-      "eval_Qnli-dev_euclidean_f1_threshold": 19.713125228881836,
-      "eval_Qnli-dev_euclidean_precision": 0.4726166328600406,
-      "eval_Qnli-dev_euclidean_recall": 0.9872881355932204,
-      "eval_Qnli-dev_manhattan_accuracy": 0.62890625,
-      "eval_Qnli-dev_manhattan_accuracy_threshold": 269.57421875,
-      "eval_Qnli-dev_manhattan_ap": 0.6311673051130458,
-      "eval_Qnli-dev_manhattan_f1": 0.6490683229813664,
-      "eval_Qnli-dev_manhattan_f1_threshold": 338.50018310546875,
-      "eval_Qnli-dev_manhattan_precision": 0.5122549019607843,
-      "eval_Qnli-dev_manhattan_recall": 0.885593220338983,
-      "eval_Qnli-dev_max_accuracy": 0.62890625,
-      "eval_Qnli-dev_max_accuracy_threshold": 698.5355224609375,
-      "eval_Qnli-dev_max_ap": 0.6311673051130458,
-      "eval_Qnli-dev_max_f1": 0.6490683229813664,
-      "eval_Qnli-dev_max_f1_threshold": 571.7469482421875,
-      "eval_Qnli-dev_max_precision": 0.5122549019607843,
-      "eval_Qnli-dev_max_recall": 0.9872881355932204,
+      "eval_Qnli-dev_cosine_accuracy": 0.634765625,
+      "eval_Qnli-dev_cosine_accuracy_threshold": 0.9121971130371094,
+      "eval_Qnli-dev_cosine_ap": 0.6377161139177543,
+      "eval_Qnli-dev_cosine_f1": 0.6430868167202571,
+      "eval_Qnli-dev_cosine_f1_threshold": 0.8449763059616089,
+      "eval_Qnli-dev_cosine_precision": 0.5181347150259067,
+      "eval_Qnli-dev_cosine_recall": 0.847457627118644,
+      "eval_Qnli-dev_dot_accuracy": 0.63671875,
+      "eval_Qnli-dev_dot_accuracy_threshold": 699.1280517578125,
+      "eval_Qnli-dev_dot_ap": 0.6388138195772171,
+      "eval_Qnli-dev_dot_f1": 0.6430868167202571,
+      "eval_Qnli-dev_dot_f1_threshold": 647.91845703125,
+      "eval_Qnli-dev_dot_precision": 0.5181347150259067,
+      "eval_Qnli-dev_dot_recall": 0.847457627118644,
+      "eval_Qnli-dev_euclidean_accuracy": 0.634765625,
+      "eval_Qnli-dev_euclidean_accuracy_threshold": 11.602351188659668,
+      "eval_Qnli-dev_euclidean_ap": 0.6377918821678507,
+      "eval_Qnli-dev_euclidean_f1": 0.6430868167202571,
+      "eval_Qnli-dev_euclidean_f1_threshold": 15.418830871582031,
+      "eval_Qnli-dev_euclidean_precision": 0.5181347150259067,
+      "eval_Qnli-dev_euclidean_recall": 0.847457627118644,
+      "eval_Qnli-dev_manhattan_accuracy": 0.642578125,
+      "eval_Qnli-dev_manhattan_accuracy_threshold": 233.09597778320312,
+      "eval_Qnli-dev_manhattan_ap": 0.6510660300493925,
+      "eval_Qnli-dev_manhattan_f1": 0.6605783866057838,
+      "eval_Qnli-dev_manhattan_f1_threshold": 315.7362976074219,
+      "eval_Qnli-dev_manhattan_precision": 0.5154394299287411,
+      "eval_Qnli-dev_manhattan_recall": 0.9194915254237288,
+      "eval_Qnli-dev_max_accuracy": 0.642578125,
+      "eval_Qnli-dev_max_accuracy_threshold": 699.1280517578125,
+      "eval_Qnli-dev_max_ap": 0.6510660300493925,
+      "eval_Qnli-dev_max_f1": 0.6605783866057838,
+      "eval_Qnli-dev_max_f1_threshold": 647.91845703125,
+      "eval_Qnli-dev_max_precision": 0.5181347150259067,
+      "eval_Qnli-dev_max_recall": 0.9194915254237288,
       "eval_allNLI-dev_cosine_accuracy": 0.673828125,
-      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9571062922477722,
-      "eval_allNLI-dev_cosine_ap": 0.43412785256494396,
-      "eval_allNLI-dev_cosine_f1": 0.5173116089613035,
-      "eval_allNLI-dev_cosine_f1_threshold": 0.8916290998458862,
-      "eval_allNLI-dev_cosine_precision": 0.39937106918238996,
-      "eval_allNLI-dev_cosine_recall": 0.7341040462427746,
-      "eval_allNLI-dev_dot_accuracy": 0.671875,
-      "eval_allNLI-dev_dot_accuracy_threshold": 734.5493774414062,
-      "eval_allNLI-dev_dot_ap": 0.43404768564943036,
-      "eval_allNLI-dev_dot_f1": 0.515748031496063,
-      "eval_allNLI-dev_dot_f1_threshold": 681.0838012695312,
-      "eval_allNLI-dev_dot_precision": 0.39104477611940297,
-      "eval_allNLI-dev_dot_recall": 0.7572254335260116,
+      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9788761138916016,
+      "eval_allNLI-dev_cosine_ap": 0.4327025118722887,
+      "eval_allNLI-dev_cosine_f1": 0.5157894736842105,
+      "eval_allNLI-dev_cosine_f1_threshold": 0.8889895081520081,
+      "eval_allNLI-dev_cosine_precision": 0.3702770780856423,
+      "eval_allNLI-dev_cosine_recall": 0.8497109826589595,
+      "eval_allNLI-dev_dot_accuracy": 0.673828125,
+      "eval_allNLI-dev_dot_accuracy_threshold": 751.3026733398438,
+      "eval_allNLI-dev_dot_ap": 0.43253813511417205,
+      "eval_allNLI-dev_dot_f1": 0.5166959578207382,
+      "eval_allNLI-dev_dot_f1_threshold": 682.115234375,
+      "eval_allNLI-dev_dot_precision": 0.3712121212121212,
+      "eval_allNLI-dev_dot_recall": 0.8497109826589595,
       "eval_allNLI-dev_euclidean_accuracy": 0.673828125,
-      "eval_allNLI-dev_euclidean_accuracy_threshold": 8.112812042236328,
-      "eval_allNLI-dev_euclidean_ap": 0.4341037986713034,
-      "eval_allNLI-dev_euclidean_f1": 0.515212981744422,
-      "eval_allNLI-dev_euclidean_f1_threshold": 12.893339157104492,
-      "eval_allNLI-dev_euclidean_precision": 0.396875,
-      "eval_allNLI-dev_euclidean_recall": 0.7341040462427746,
+      "eval_allNLI-dev_euclidean_accuracy_threshold": 5.694375038146973,
+      "eval_allNLI-dev_euclidean_ap": 0.4325438108928368,
+      "eval_allNLI-dev_euclidean_f1": 0.5157894736842105,
+      "eval_allNLI-dev_euclidean_f1_threshold": 13.050301551818848,
+      "eval_allNLI-dev_euclidean_precision": 0.3702770780856423,
+      "eval_allNLI-dev_euclidean_recall": 0.8497109826589595,
       "eval_allNLI-dev_manhattan_accuracy": 0.671875,
-      "eval_allNLI-dev_manhattan_accuracy_threshold": 172.47096252441406,
-      "eval_allNLI-dev_manhattan_ap": 0.4307335026703495,
-      "eval_allNLI-dev_manhattan_f1": 0.5261044176706827,
-      "eval_allNLI-dev_manhattan_f1_threshold": 258.1781005859375,
-      "eval_allNLI-dev_manhattan_precision": 0.40307692307692305,
-      "eval_allNLI-dev_manhattan_recall": 0.7572254335260116,
+      "eval_allNLI-dev_manhattan_accuracy_threshold": 105.85862731933594,
+      "eval_allNLI-dev_manhattan_ap": 0.4278948508381489,
+      "eval_allNLI-dev_manhattan_f1": 0.5217391304347826,
+      "eval_allNLI-dev_manhattan_f1_threshold": 241.55101013183594,
+      "eval_allNLI-dev_manhattan_precision": 0.38764044943820225,
+      "eval_allNLI-dev_manhattan_recall": 0.7976878612716763,
       "eval_allNLI-dev_max_accuracy": 0.673828125,
-      "eval_allNLI-dev_max_accuracy_threshold": 734.5493774414062,
-      "eval_allNLI-dev_max_ap": 0.43412785256494396,
-      "eval_allNLI-dev_max_f1": 0.5261044176706827,
-      "eval_allNLI-dev_max_f1_threshold": 681.0838012695312,
-      "eval_allNLI-dev_max_precision": 0.40307692307692305,
-      "eval_allNLI-dev_max_recall": 0.7572254335260116,
-      "eval_loss": 1.9933717250823975,
-      "eval_runtime": 54.2062,
-      "eval_samples_per_second": 30.698,
-      "eval_sequential_score": 0.6311673051130458,
-      "eval_steps_per_second": 0.129,
-      "eval_sts-test_pearson_cosine": 0.2805005187117122,
-      "eval_sts-test_pearson_dot": 0.2779761844381874,
-      "eval_sts-test_pearson_euclidean": 0.3103382429908865,
-      "eval_sts-test_pearson_manhattan": 0.3211130291412454,
-      "eval_sts-test_pearson_max": 0.3211130291412454,
-      "eval_sts-test_spearman_cosine": 0.3370077871450756,
-      "eval_sts-test_spearman_dot": 0.3341743610300792,
-      "eval_sts-test_spearman_euclidean": 0.3370845355583619,
-      "eval_sts-test_spearman_manhattan": 0.3502758201274433,
-      "eval_sts-test_spearman_max": 0.3502758201274433,
+      "eval_allNLI-dev_max_accuracy_threshold": 751.3026733398438,
+      "eval_allNLI-dev_max_ap": 0.4327025118722887,
+      "eval_allNLI-dev_max_f1": 0.5217391304347826,
+      "eval_allNLI-dev_max_f1_threshold": 682.115234375,
+      "eval_allNLI-dev_max_precision": 0.38764044943820225,
+      "eval_allNLI-dev_max_recall": 0.8497109826589595,
+      "eval_loss": 1.9853787422180176,
+      "eval_runtime": 52.6119,
+      "eval_samples_per_second": 31.628,
+      "eval_sequential_score": 0.6510660300493925,
+      "eval_steps_per_second": 0.133,
+      "eval_sts-test_pearson_cosine": 0.27561429926791464,
+      "eval_sts-test_pearson_dot": 0.27336219058729005,
+      "eval_sts-test_pearson_euclidean": 0.29994864031298485,
+      "eval_sts-test_pearson_manhattan": 0.3112396414398868,
+      "eval_sts-test_pearson_max": 0.3112396414398868,
+      "eval_sts-test_spearman_cosine": 0.32606859471811517,
+      "eval_sts-test_spearman_dot": 0.3235796341494495,
+      "eval_sts-test_spearman_euclidean": 0.3260361203897462,
+      "eval_sts-test_spearman_manhattan": 0.3379918226318111,
+      "eval_sts-test_spearman_max": 0.3379918226318111,
       "step": 459
     },
     {
       "epoch": 0.452755905511811,
-      "grad_norm": 13.650534629821777,
+      "grad_norm": 15.168042182922363,
       "learning_rate": 4.522862823061631e-07,
-      "loss": 1.9031,
+      "loss": 2.0416,
       "step": 460
     },
     {
       "epoch": 0.453740157480315,
-      "grad_norm": 15.621660232543945,
+      "grad_norm": 12.258943557739258,
       "learning_rate": 4.532803180914514e-07,
-      "loss": 2.1993,
+      "loss": 2.1907,
       "step": 461
     },
     {
       "epoch": 0.4547244094488189,
-      "grad_norm": 12.056524276733398,
+      "grad_norm": 16.1510009765625,
       "learning_rate": 4.542743538767397e-07,
-      "loss": 1.4149,
+      "loss": 1.7478,
       "step": 462
     },
     {
       "epoch": 0.45570866141732286,
-      "grad_norm": 12.095454216003418,
+      "grad_norm": 13.124279022216797,
       "learning_rate": 4.552683896620279e-07,
-      "loss": 1.2881,
+      "loss": 1.9,
       "step": 463
     },
     {
       "epoch": 0.4566929133858268,
-      "grad_norm": 42.66931915283203,
+      "grad_norm": 11.794757843017578,
       "learning_rate": 4.562624254473162e-07,
-      "loss": 5.0921,
+      "loss": 1.6876,
       "step": 464
     },
     {
       "epoch": 0.45767716535433073,
-      "grad_norm": 38.241668701171875,
+      "grad_norm": 13.155580520629883,
       "learning_rate": 4.5725646123260444e-07,
-      "loss": 3.8827,
+      "loss": 2.0035,
       "step": 465
     },
     {
       "epoch": 0.45866141732283466,
-      "grad_norm": 14.144487380981445,
+      "grad_norm": 11.5040922164917,
       "learning_rate": 4.5825049701789274e-07,
-      "loss": 1.5278,
+      "loss": 1.4127,
       "step": 466
     },
     {
       "epoch": 0.4596456692913386,
-      "grad_norm": 13.078521728515625,
+      "grad_norm": 11.074214935302734,
       "learning_rate": 4.59244532803181e-07,
-      "loss": 1.5084,
+      "loss": 1.5593,
       "step": 467
     },
     {
       "epoch": 0.46062992125984253,
-      "grad_norm": 12.216127395629883,
+      "grad_norm": 13.357093811035156,
       "learning_rate": 4.6023856858846925e-07,
-      "loss": 1.5728,
+      "loss": 1.7,
       "step": 468
     },
     {
       "epoch": 0.46161417322834647,
-      "grad_norm": 12.086899757385254,
+      "grad_norm": 14.933716773986816,
       "learning_rate": 4.6123260437375755e-07,
-      "loss": 1.3861,
+      "loss": 1.5157,
       "step": 469
     },
     {
       "epoch": 0.4625984251968504,
-      "grad_norm": 12.2554349899292,
+      "grad_norm": 12.29159164428711,
       "learning_rate": 4.622266401590458e-07,
-      "loss": 1.2923,
+      "loss": 1.6554,
       "step": 470
     },
     {
       "epoch": 0.46358267716535434,
-      "grad_norm": 14.57932186126709,
+      "grad_norm": 12.008153915405273,
       "learning_rate": 4.6322067594433406e-07,
-      "loss": 1.8902,
+      "loss": 1.7404,
       "step": 471
     },
     {
       "epoch": 0.4645669291338583,
-      "grad_norm": 13.89941692352295,
+      "grad_norm": 13.209589958190918,
       "learning_rate": 4.642147117296223e-07,
-      "loss": 1.7272,
+      "loss": 2.1432,
       "step": 472
     },
     {
       "epoch": 0.4655511811023622,
-      "grad_norm": 13.036828994750977,
+      "grad_norm": 13.247814178466797,
       "learning_rate": 4.652087475149106e-07,
-      "loss": 1.292,
+      "loss": 1.7322,
       "step": 473
     },
     {
       "epoch": 0.46653543307086615,
-      "grad_norm": 10.900352478027344,
+      "grad_norm": 12.133121490478516,
       "learning_rate": 4.662027833001989e-07,
-      "loss": 1.2791,
+      "loss": 1.7281,
       "step": 474
     },
     {
       "epoch": 0.4675196850393701,
-      "grad_norm": 11.765789031982422,
+      "grad_norm": 12.511950492858887,
       "learning_rate": 4.671968190854872e-07,
-      "loss": 1.8328,
+      "loss": 1.5107,
       "step": 475
     },
     {
       "epoch": 0.468503937007874,
-      "grad_norm": 15.065398216247559,
+      "grad_norm": 13.9373779296875,
       "learning_rate": 4.681908548707754e-07,
-      "loss": 1.5654,
+      "loss": 1.779,
       "step": 476
     },
     {
       "epoch": 0.46948818897637795,
-      "grad_norm": 11.131512641906738,
+      "grad_norm": 10.878262519836426,
       "learning_rate": 4.691848906560637e-07,
-      "loss": 0.865,
+      "loss": 1.325,
       "step": 477
     },
     {
       "epoch": 0.4704724409448819,
-      "grad_norm": 11.469169616699219,
+      "grad_norm": 11.128897666931152,
       "learning_rate": 4.70178926441352e-07,
-      "loss": 1.47,
+      "loss": 1.073,
       "step": 478
     },
     {
       "epoch": 0.4714566929133858,
-      "grad_norm": 13.804464340209961,
+      "grad_norm": 14.144214630126953,
       "learning_rate": 4.7117296222664024e-07,
-      "loss": 1.7305,
+      "loss": 1.864,
       "step": 479
     },
     {
       "epoch": 0.47244094488188976,
-      "grad_norm": 26.671695709228516,
+      "grad_norm": 15.306927680969238,
       "learning_rate": 4.721669980119285e-07,
-      "loss": 2.009,
+      "loss": 2.3645,
       "step": 480
     },
     {
       "epoch": 0.4734251968503937,
-      "grad_norm": 11.814390182495117,
+      "grad_norm": 10.306663513183594,
       "learning_rate": 4.731610337972168e-07,
-      "loss": 1.4719,
+      "loss": 1.181,
       "step": 481
     },
     {
       "epoch": 0.4744094488188976,
-      "grad_norm": 13.533790588378906,
+      "grad_norm": 12.236723899841309,
       "learning_rate": 4.7415506958250505e-07,
-      "loss": 1.6774,
+      "loss": 1.4562,
       "step": 482
     },
     {
       "epoch": 0.47539370078740156,
-      "grad_norm": 37.1170768737793,
+      "grad_norm": 12.010734558105469,
       "learning_rate": 4.7514910536779335e-07,
-      "loss": 3.9534,
+      "loss": 1.3105,
       "step": 483
     },
     {
       "epoch": 0.4763779527559055,
-      "grad_norm": 13.464319229125977,
+      "grad_norm": 24.507387161254883,
       "learning_rate": 4.7614314115308155e-07,
-      "loss": 1.7503,
+      "loss": 2.8012,
       "step": 484
     },
     {
       "epoch": 0.47736220472440943,
-      "grad_norm": 12.962063789367676,
+      "grad_norm": 15.473782539367676,
       "learning_rate": 4.771371769383698e-07,
-      "loss": 1.4652,
+      "loss": 2.0114,
       "step": 485
     },
     {
       "epoch": 0.47834645669291337,
-      "grad_norm": 11.213044166564941,
+      "grad_norm": 13.31457805633545,
       "learning_rate": 4.781312127236581e-07,
-      "loss": 1.4057,
+      "loss": 1.6307,
       "step": 486
     },
     {
       "epoch": 0.4793307086614173,
-      "grad_norm": 12.803650856018066,
+      "grad_norm": 26.823562622070312,
       "learning_rate": 4.791252485089464e-07,
-      "loss": 1.7381,
+      "loss": 2.7733,
       "step": 487
     },
     {
       "epoch": 0.48031496062992124,
-      "grad_norm": 42.93519592285156,
+      "grad_norm": 13.975008010864258,
       "learning_rate": 4.801192842942347e-07,
-      "loss": 3.1813,
+      "loss": 1.8211,
       "step": 488
     },
     {
       "epoch": 0.4812992125984252,
-      "grad_norm": 14.303778648376465,
+      "grad_norm": 12.852688789367676,
       "learning_rate": 4.811133200795229e-07,
-      "loss": 1.6575,
+      "loss": 1.574,
       "step": 489
     },
     {
       "epoch": 0.4822834645669291,
-      "grad_norm": 11.379716873168945,
+      "grad_norm": 18.221059799194336,
       "learning_rate": 4.821073558648112e-07,
-      "loss": 1.2314,
+      "loss": 1.9713,
       "step": 490
     },
     {
       "epoch": 0.48326771653543305,
-      "grad_norm": 63.755924224853516,
+      "grad_norm": 12.657098770141602,
       "learning_rate": 4.831013916500995e-07,
-      "loss": 5.9239,
+      "loss": 1.2774,
       "step": 491
     },
     {
       "epoch": 0.484251968503937,
-      "grad_norm": 15.35229778289795,
+      "grad_norm": 16.00031280517578,
       "learning_rate": 4.840954274353877e-07,
-      "loss": 2.1689,
+      "loss": 2.58,
       "step": 492
     },
     {
       "epoch": 0.48523622047244097,
-      "grad_norm": 11.58899974822998,
+      "grad_norm": 12.299332618713379,
       "learning_rate": 4.85089463220676e-07,
-      "loss": 1.5536,
+      "loss": 2.0594,
       "step": 493
     },
     {
       "epoch": 0.4862204724409449,
-      "grad_norm": 42.49075698852539,
+      "grad_norm": 16.219257354736328,
       "learning_rate": 4.860834990059642e-07,
-      "loss": 4.2167,
+      "loss": 1.5857,
       "step": 494
     },
     {
       "epoch": 0.48720472440944884,
-      "grad_norm": 11.30247974395752,
+      "grad_norm": 12.892383575439453,
       "learning_rate": 4.870775347912525e-07,
-      "loss": 1.3489,
+      "loss": 2.0028,
       "step": 495
     },
     {
       "epoch": 0.4881889763779528,
-      "grad_norm": 12.173792839050293,
+      "grad_norm": 16.72523307800293,
       "learning_rate": 4.880715705765408e-07,
-      "loss": 1.6137,
+      "loss": 1.8863,
       "step": 496
     },
     {
       "epoch": 0.4891732283464567,
-      "grad_norm": 10.37173080444336,
+      "grad_norm": 14.14151382446289,
       "learning_rate": 4.890656063618291e-07,
-      "loss": 1.3183,
+      "loss": 1.5171,
       "step": 497
     },
     {
       "epoch": 0.49015748031496065,
-      "grad_norm": 35.126182556152344,
+      "grad_norm": 16.287216186523438,
       "learning_rate": 4.900596421471173e-07,
-      "loss": 3.9189,
+      "loss": 1.9355,
       "step": 498
     },
     {
       "epoch": 0.4911417322834646,
-      "grad_norm": 30.792539596557617,
+      "grad_norm": 16.423860549926758,
       "learning_rate": 4.910536779324057e-07,
-      "loss": 3.003,
+      "loss": 2.0675,
       "step": 499
     },
     {
       "epoch": 0.4921259842519685,
-      "grad_norm": 9.988119125366211,
+      "grad_norm": 12.384491920471191,
       "learning_rate": 4.92047713717694e-07,
-      "loss": 1.3985,
+      "loss": 1.6017,
       "step": 500
     },
     {
       "epoch": 0.49311023622047245,
-      "grad_norm": 10.691671371459961,
+      "grad_norm": 10.480212211608887,
       "learning_rate": 4.930417495029823e-07,
-      "loss": 1.1173,
+      "loss": 1.4089,
       "step": 501
     },
     {
       "epoch": 0.4940944881889764,
-      "grad_norm": 9.386334419250488,
+      "grad_norm": 9.902453422546387,
       "learning_rate": 4.940357852882705e-07,
-      "loss": 0.9892,
+      "loss": 1.3836,
       "step": 502
     },
     {
       "epoch": 0.4950787401574803,
-      "grad_norm": 13.273354530334473,
+      "grad_norm": 12.404056549072266,
       "learning_rate": 4.950298210735588e-07,
-      "loss": 1.6817,
+      "loss": 1.6033,
       "step": 503
     },
     {
       "epoch": 0.49606299212598426,
-      "grad_norm": 9.470836639404297,
+      "grad_norm": 9.605865478515625,
       "learning_rate": 4.96023856858847e-07,
-      "loss": 1.1026,
+      "loss": 1.0891,
       "step": 504
     },
     {
       "epoch": 0.4970472440944882,
-      "grad_norm": 27.343177795410156,
+      "grad_norm": 13.224746704101562,
       "learning_rate": 4.970178926441353e-07,
-      "loss": 3.0063,
+      "loss": 1.7119,
       "step": 505
     },
     {
       "epoch": 0.49803149606299213,
-      "grad_norm": 12.136895179748535,
+      "grad_norm": 13.01671314239502,
       "learning_rate": 4.980119284294235e-07,
-      "loss": 1.4292,
+      "loss": 1.3685,
       "step": 506
     },
     {
       "epoch": 0.49901574803149606,
-      "grad_norm": 13.389548301696777,
+      "grad_norm": 11.965983390808105,
       "learning_rate": 4.990059642147118e-07,
-      "loss": 1.8554,
+      "loss": 1.4252,
       "step": 507
     },
     {
       "epoch": 0.5,
-      "grad_norm": 10.676069259643555,
+      "grad_norm": 13.703736305236816,
       "learning_rate": 5.000000000000001e-07,
-      "loss": 1.419,
+      "loss": 1.5538,
       "step": 508
     },
     {
       "epoch": 0.5009842519685039,
-      "grad_norm": 11.071086883544922,
+      "grad_norm": 16.96326446533203,
       "learning_rate": 5.009940357852883e-07,
-      "loss": 1.2399,
+      "loss": 1.7513,
       "step": 509
     },
     {
       "epoch": 0.5019685039370079,
-      "grad_norm": 10.330881118774414,
+      "grad_norm": 9.837270736694336,
       "learning_rate": 5.019880715705767e-07,
-      "loss": 1.2939,
+      "loss": 1.1831,
       "step": 510
     },
     {
       "epoch": 0.5029527559055118,
-      "grad_norm": 10.792057037353516,
+      "grad_norm": 13.675102233886719,
       "learning_rate": 5.029821073558649e-07,
-      "loss": 1.4505,
+      "loss": 1.7767,
       "step": 511
     },
     {
       "epoch": 0.5039370078740157,
-      "grad_norm": 11.177830696105957,
+      "grad_norm": 12.791515350341797,
       "learning_rate": 5.039761431411532e-07,
-      "loss": 1.0514,
+      "loss": 1.4324,
       "step": 512
     },
     {
       "epoch": 0.5049212598425197,
-      "grad_norm": 10.643841743469238,
+      "grad_norm": 18.81296157836914,
       "learning_rate": 5.049701789264414e-07,
-      "loss": 1.2327,
+      "loss": 2.1672,
       "step": 513
     },
     {
       "epoch": 0.5059055118110236,
-      "grad_norm": 11.909382820129395,
+      "grad_norm": 13.022212028503418,
       "learning_rate": 5.059642147117297e-07,
-      "loss": 1.3414,
+      "loss": 1.6348,
       "step": 514
     },
     {
       "epoch": 0.5068897637795275,
-      "grad_norm": 10.281572341918945,
+      "grad_norm": 16.152873992919922,
       "learning_rate": 5.06958250497018e-07,
-      "loss": 1.3721,
+      "loss": 1.7285,
       "step": 515
     },
     {
       "epoch": 0.5078740157480315,
-      "grad_norm": 53.167823791503906,
+      "grad_norm": 19.10329818725586,
       "learning_rate": 5.079522862823062e-07,
-      "loss": 5.4099,
+      "loss": 2.0186,
       "step": 516
     },
     {
       "epoch": 0.5088582677165354,
-      "grad_norm": 10.074481010437012,
+      "grad_norm": 12.92306137084961,
       "learning_rate": 5.089463220675945e-07,
-      "loss": 1.0425,
+      "loss": 1.382,
       "step": 517
     },
     {
       "epoch": 0.5098425196850394,
-      "grad_norm": 47.982383728027344,
+      "grad_norm": 12.875118255615234,
       "learning_rate": 5.099403578528828e-07,
-      "loss": 4.6786,
+      "loss": 1.4509,
       "step": 518
     },
     {
       "epoch": 0.5108267716535433,
-      "grad_norm": 10.051337242126465,
+      "grad_norm": 10.416156768798828,
       "learning_rate": 5.10934393638171e-07,
-      "loss": 0.9301,
+      "loss": 1.1043,
       "step": 519
     },
     {
       "epoch": 0.5118110236220472,
-      "grad_norm": 11.556605339050293,
+      "grad_norm": 12.095355987548828,
       "learning_rate": 5.119284294234593e-07,
-      "loss": 1.3494,
+      "loss": 1.3322,
       "step": 520
     },
     {
       "epoch": 0.5127952755905512,
-      "grad_norm": 11.6151762008667,
+      "grad_norm": 10.407857894897461,
       "learning_rate": 5.129224652087476e-07,
-      "loss": 1.158,
+      "loss": 1.3267,
       "step": 521
     },
     {
       "epoch": 0.5137795275590551,
-      "grad_norm": 9.960102081298828,
+      "grad_norm": 12.354635238647461,
       "learning_rate": 5.139165009940358e-07,
-      "loss": 1.0346,
+      "loss": 1.3639,
       "step": 522
     },
     {
       "epoch": 0.514763779527559,
-      "grad_norm": 10.822325706481934,
+      "grad_norm": 11.841660499572754,
       "learning_rate": 5.149105367793241e-07,
-      "loss": 1.3054,
+      "loss": 1.203,
       "step": 523
     },
     {
       "epoch": 0.515748031496063,
-      "grad_norm": 11.666457176208496,
+      "grad_norm": 15.488195419311523,
       "learning_rate": 5.159045725646124e-07,
-      "loss": 1.1146,
+      "loss": 1.8583,
       "step": 524
     },
     {
       "epoch": 0.5167322834645669,
-      "grad_norm": 12.345129013061523,
+      "grad_norm": 18.47950553894043,
       "learning_rate": 5.168986083499007e-07,
-      "loss": 1.1232,
+      "loss": 2.267,
       "step": 525
     },
     {
       "epoch": 0.5177165354330708,
-      "grad_norm": 15.775688171386719,
+      "grad_norm": 11.875829696655273,
       "learning_rate": 5.178926441351889e-07,
-      "loss": 1.9019,
+      "loss": 1.2935,
       "step": 526
     },
     {
       "epoch": 0.5187007874015748,
-      "grad_norm": 10.290173530578613,
+      "grad_norm": 13.655136108398438,
       "learning_rate": 5.188866799204771e-07,
-      "loss": 0.7564,
+      "loss": 1.7431,
       "step": 527
     },
     {
       "epoch": 0.5196850393700787,
-      "grad_norm": 12.738898277282715,
+      "grad_norm": 16.041044235229492,
       "learning_rate": 5.198807157057656e-07,
-      "loss": 1.2245,
+      "loss": 1.8484,
       "step": 528
     },
     {
       "epoch": 0.5206692913385826,
-      "grad_norm": 14.781623840332031,
+      "grad_norm": 15.453495979309082,
       "learning_rate": 5.208747514910538e-07,
-      "loss": 1.2729,
+      "loss": 1.5626,
       "step": 529
     },
     {
       "epoch": 0.5216535433070866,
-      "grad_norm": 12.573616981506348,
+      "grad_norm": 18.945079803466797,
       "learning_rate": 5.21868787276342e-07,
-      "loss": 1.3485,
+      "loss": 2.2645,
       "step": 530
     },
     {
       "epoch": 0.5226377952755905,
-      "grad_norm": 14.437003135681152,
+      "grad_norm": 12.458165168762207,
       "learning_rate": 5.228628230616304e-07,
-      "loss": 1.3507,
+      "loss": 1.4313,
       "step": 531
     },
     {
       "epoch": 0.5236220472440944,
-      "grad_norm": 13.429966926574707,
+      "grad_norm": 13.550048828125,
       "learning_rate": 5.238568588469186e-07,
-      "loss": 1.1656,
+      "loss": 1.8204,
       "step": 532
     },
     {
       "epoch": 0.5246062992125984,
-      "grad_norm": 56.66192626953125,
+      "grad_norm": 14.208359718322754,
       "learning_rate": 5.248508946322069e-07,
-      "loss": 5.7052,
+      "loss": 1.5659,
       "step": 533
     },
     {
       "epoch": 0.5255905511811023,
-      "grad_norm": 13.437881469726562,
+      "grad_norm": 11.332735061645508,
       "learning_rate": 5.258449304174951e-07,
-      "loss": 1.2702,
+      "loss": 1.2689,
       "step": 534
     },
     {
       "epoch": 0.5265748031496063,
-      "grad_norm": 14.13820743560791,
+      "grad_norm": 12.082714080810547,
       "learning_rate": 5.268389662027834e-07,
-      "loss": 1.5765,
+      "loss": 1.8193,
       "step": 535
     },
     {
       "epoch": 0.5275590551181102,
-      "grad_norm": 12.186126708984375,
+      "grad_norm": 15.39877700805664,
       "learning_rate": 5.278330019880717e-07,
-      "loss": 1.1904,
+      "loss": 2.2902,
       "step": 536
     },
     {
       "epoch": 0.5285433070866141,
-      "grad_norm": 12.874736785888672,
+      "grad_norm": 11.852607727050781,
       "learning_rate": 5.288270377733599e-07,
-      "loss": 1.2439,
+      "loss": 1.6936,
       "step": 537
     },
     {
       "epoch": 0.5295275590551181,
-      "grad_norm": 13.560784339904785,
+      "grad_norm": 12.435094833374023,
       "learning_rate": 5.298210735586483e-07,
-      "loss": 1.3851,
+      "loss": 1.7305,
       "step": 538
     },
     {
       "epoch": 0.530511811023622,
-      "grad_norm": 48.91597366333008,
+      "grad_norm": 12.685184478759766,
       "learning_rate": 5.308151093439365e-07,
-      "loss": 5.4258,
+      "loss": 1.4449,
       "step": 539
     },
     {
       "epoch": 0.531496062992126,
-      "grad_norm": 63.82156753540039,
+      "grad_norm": 12.26480484008789,
       "learning_rate": 5.318091451292247e-07,
-      "loss": 6.0733,
+      "loss": 1.5594,
       "step": 540
     },
     {
       "epoch": 0.53248031496063,
-      "grad_norm": 11.866876602172852,
+      "grad_norm": 14.554289817810059,
       "learning_rate": 5.32803180914513e-07,
-      "loss": 1.1416,
+      "loss": 1.9678,
       "step": 541
     },
     {
       "epoch": 0.5334645669291339,
-      "grad_norm": 13.9098482131958,
+      "grad_norm": 13.191105842590332,
       "learning_rate": 5.337972166998013e-07,
-      "loss": 1.6498,
+      "loss": 2.0327,
       "step": 542
     },
     {
       "epoch": 0.5344488188976378,
-      "grad_norm": 15.363849639892578,
+      "grad_norm": 14.845519065856934,
       "learning_rate": 5.347912524850895e-07,
-      "loss": 1.9752,
+      "loss": 2.0456,
       "step": 543
     },
     {
       "epoch": 0.5354330708661418,
-      "grad_norm": 53.16739273071289,
+      "grad_norm": 14.311629295349121,
       "learning_rate": 5.357852882703778e-07,
-      "loss": 5.0479,
+      "loss": 2.0452,
       "step": 544
     },
     {
       "epoch": 0.5364173228346457,
-      "grad_norm": 11.984163284301758,
+      "grad_norm": 14.074420928955078,
       "learning_rate": 5.367793240556661e-07,
-      "loss": 1.4322,
+      "loss": 1.9435,
       "step": 545
     },
     {
       "epoch": 0.5374015748031497,
-      "grad_norm": 9.779897689819336,
+      "grad_norm": 16.501874923706055,
       "learning_rate": 5.377733598409544e-07,
-      "loss": 1.1549,
+      "loss": 1.8963,
       "step": 546
     },
     {
       "epoch": 0.5383858267716536,
-      "grad_norm": 12.604474067687988,
+      "grad_norm": 16.289310455322266,
       "learning_rate": 5.387673956262426e-07,
-      "loss": 1.623,
+      "loss": 1.9536,
       "step": 547
     },
     {
       "epoch": 0.5393700787401575,
-      "grad_norm": 10.861937522888184,
+      "grad_norm": 11.657831192016602,
       "learning_rate": 5.397614314115308e-07,
-      "loss": 1.2541,
+      "loss": 1.0665,
       "step": 548
     },
     {
       "epoch": 0.5403543307086615,
-      "grad_norm": 11.303938865661621,
+      "grad_norm": 17.12558364868164,
       "learning_rate": 5.407554671968192e-07,
-      "loss": 1.3253,
+      "loss": 1.8067,
       "step": 549
     },
     {
       "epoch": 0.5413385826771654,
-      "grad_norm": 11.52147388458252,
+      "grad_norm": 12.209671974182129,
       "learning_rate": 5.417495029821074e-07,
-      "loss": 1.4802,
+      "loss": 1.6227,
       "step": 550
     },
     {
       "epoch": 0.5423228346456693,
-      "grad_norm": 9.711042404174805,
+      "grad_norm": 13.168261528015137,
       "learning_rate": 5.427435387673957e-07,
-      "loss": 0.993,
+      "loss": 1.687,
       "step": 551
     },
     {
       "epoch": 0.5433070866141733,
-      "grad_norm": 10.261675834655762,
+      "grad_norm": 11.820549011230469,
       "learning_rate": 5.43737574552684e-07,
-      "loss": 1.2354,
+      "loss": 1.5937,
       "step": 552
     },
     {
       "epoch": 0.5442913385826772,
-      "grad_norm": 9.668618202209473,
+      "grad_norm": 12.037482261657715,
       "learning_rate": 5.447316103379722e-07,
-      "loss": 0.7297,
+      "loss": 0.9216,
       "step": 553
     },
     {
       "epoch": 0.5452755905511811,
-      "grad_norm": 59.58056640625,
+      "grad_norm": 15.135348320007324,
       "learning_rate": 5.457256461232605e-07,
-      "loss": 5.1613,
+      "loss": 1.3895,
       "step": 554
     },
     {
       "epoch": 0.5462598425196851,
-      "grad_norm": 10.98855209350586,
+      "grad_norm": 16.537757873535156,
       "learning_rate": 5.467196819085487e-07,
-      "loss": 1.0977,
+      "loss": 1.7863,
       "step": 555
     },
     {
       "epoch": 0.547244094488189,
-      "grad_norm": 10.048385620117188,
+      "grad_norm": 10.809349060058594,
       "learning_rate": 5.477137176938371e-07,
-      "loss": 1.0681,
+      "loss": 1.2574,
       "step": 556
     },
     {
       "epoch": 0.5482283464566929,
-      "grad_norm": 59.76496124267578,
+      "grad_norm": 17.314552307128906,
       "learning_rate": 5.487077534791253e-07,
-      "loss": 5.1421,
+      "loss": 2.108,
       "step": 557
     },
     {
       "epoch": 0.5492125984251969,
-      "grad_norm": 12.414549827575684,
+      "grad_norm": 11.581759452819824,
       "learning_rate": 5.497017892644135e-07,
-      "loss": 1.1879,
+      "loss": 1.2782,
       "step": 558
     },
     {
       "epoch": 0.5501968503937008,
-      "grad_norm": 10.891206741333008,
+      "grad_norm": 11.438880920410156,
       "learning_rate": 5.50695825049702e-07,
-      "loss": 1.106,
+      "loss": 1.4959,
       "step": 559
     },
     {
       "epoch": 0.5511811023622047,
-      "grad_norm": 46.950321197509766,
+      "grad_norm": 12.535311698913574,
       "learning_rate": 5.516898608349902e-07,
-      "loss": 4.4247,
+      "loss": 1.9191,
       "step": 560
     },
     {
       "epoch": 0.5521653543307087,
-      "grad_norm": 10.903446197509766,
+      "grad_norm": 17.22417449951172,
       "learning_rate": 5.526838966202784e-07,
-      "loss": 1.2965,
+      "loss": 2.0049,
       "step": 561
     },
     {
       "epoch": 0.5531496062992126,
-      "grad_norm": 10.382317543029785,
+      "grad_norm": 11.439227104187012,
       "learning_rate": 5.536779324055667e-07,
-      "loss": 0.878,
+      "loss": 1.2511,
       "step": 562
     },
     {
       "epoch": 0.5541338582677166,
-      "grad_norm": 11.724767684936523,
+      "grad_norm": 13.34407901763916,
       "learning_rate": 5.54671968190855e-07,
-      "loss": 1.3601,
+      "loss": 1.3912,
       "step": 563
     },
     {
       "epoch": 0.5551181102362205,
-      "grad_norm": 12.926590919494629,
+      "grad_norm": 13.90283203125,
       "learning_rate": 5.556660039761433e-07,
-      "loss": 1.4492,
+      "loss": 1.371,
       "step": 564
     },
     {
       "epoch": 0.5561023622047244,
-      "grad_norm": 10.146951675415039,
+      "grad_norm": 18.280315399169922,
       "learning_rate": 5.566600397614315e-07,
-      "loss": 0.9265,
+      "loss": 1.6155,
       "step": 565
     },
     {
       "epoch": 0.5570866141732284,
-      "grad_norm": 11.620803833007812,
+      "grad_norm": 10.993673324584961,
       "learning_rate": 5.576540755467198e-07,
-      "loss": 1.2781,
+      "loss": 1.4625,
       "step": 566
     },
     {
       "epoch": 0.5580708661417323,
-      "grad_norm": 9.574142456054688,
+      "grad_norm": 9.647687911987305,
       "learning_rate": 5.586481113320081e-07,
-      "loss": 0.8514,
+      "loss": 0.86,
       "step": 567
     },
     {
       "epoch": 0.5590551181102362,
-      "grad_norm": 13.647089004516602,
+      "grad_norm": 14.242208480834961,
       "learning_rate": 5.596421471172963e-07,
-      "loss": 1.3312,
+      "loss": 1.5753,
       "step": 568
     },
     {
       "epoch": 0.5600393700787402,
-      "grad_norm": 28.685733795166016,
+      "grad_norm": 11.613588333129883,
       "learning_rate": 5.606361829025846e-07,
-      "loss": 1.9533,
+      "loss": 1.6126,
       "step": 569
     },
     {
       "epoch": 0.5610236220472441,
-      "grad_norm": 11.902270317077637,
+      "grad_norm": 11.57154369354248,
       "learning_rate": 5.616302186878729e-07,
-      "loss": 0.9773,
+      "loss": 1.3171,
       "step": 570
     },
     {
       "epoch": 0.562007874015748,
-      "grad_norm": 12.151090621948242,
+      "grad_norm": 15.930388450622559,
       "learning_rate": 5.626242544731611e-07,
-      "loss": 1.5187,
+      "loss": 1.9378,
       "step": 571
     },
     {
       "epoch": 0.562992125984252,
-      "grad_norm": 11.929316520690918,
+      "grad_norm": 12.384969711303711,
       "learning_rate": 5.636182902584494e-07,
-      "loss": 1.1786,
+      "loss": 1.2736,
       "step": 572
     },
     {
       "epoch": 0.5639763779527559,
-      "grad_norm": 11.160627365112305,
+      "grad_norm": 10.599091529846191,
       "learning_rate": 5.646123260437377e-07,
-      "loss": 1.1154,
+      "loss": 1.2368,
       "step": 573
     },
     {
       "epoch": 0.5649606299212598,
-      "grad_norm": 10.790230751037598,
+      "grad_norm": 10.539170265197754,
       "learning_rate": 5.656063618290259e-07,
-      "loss": 0.9937,
+      "loss": 1.1005,
       "step": 574
     },
     {
       "epoch": 0.5659448818897638,
-      "grad_norm": 46.701942443847656,
+      "grad_norm": 11.336695671081543,
       "learning_rate": 5.666003976143142e-07,
-      "loss": 3.9975,
+      "loss": 1.1765,
       "step": 575
     },
     {
       "epoch": 0.5669291338582677,
-      "grad_norm": 10.748030662536621,
+      "grad_norm": 12.062975883483887,
       "learning_rate": 5.675944333996024e-07,
-      "loss": 0.8966,
+      "loss": 1.3557,
       "step": 576
     },
     {
       "epoch": 0.5679133858267716,
-      "grad_norm": 43.31929397583008,
+      "grad_norm": 12.290358543395996,
       "learning_rate": 5.685884691848908e-07,
-      "loss": 3.9168,
+      "loss": 1.3224,
       "step": 577
     },
     {
       "epoch": 0.5688976377952756,
-      "grad_norm": 11.500582695007324,
+      "grad_norm": 17.76947593688965,
       "learning_rate": 5.69582504970179e-07,
-      "loss": 1.0795,
+      "loss": 1.7914,
       "step": 578
     },
     {
       "epoch": 0.5698818897637795,
-      "grad_norm": 11.253996849060059,
+      "grad_norm": 10.218934059143066,
       "learning_rate": 5.705765407554672e-07,
-      "loss": 0.8258,
+      "loss": 1.0633,
       "step": 579
     },
     {
       "epoch": 0.5708661417322834,
-      "grad_norm": 34.77997970581055,
+      "grad_norm": 12.503771781921387,
       "learning_rate": 5.715705765407556e-07,
-      "loss": 2.9027,
+      "loss": 1.3624,
       "step": 580
     },
     {
       "epoch": 0.5718503937007874,
-      "grad_norm": 43.59735107421875,
+      "grad_norm": 11.047056198120117,
       "learning_rate": 5.725646123260438e-07,
-      "loss": 3.9928,
+      "loss": 0.9804,
       "step": 581
     },
     {
       "epoch": 0.5728346456692913,
-      "grad_norm": 10.751973152160645,
+      "grad_norm": 17.329021453857422,
       "learning_rate": 5.735586481113321e-07,
-      "loss": 1.0328,
+      "loss": 1.8246,
       "step": 582
     },
     {
       "epoch": 0.5738188976377953,
-      "grad_norm": 37.12349319458008,
+      "grad_norm": 13.590007781982422,
       "learning_rate": 5.745526838966203e-07,
-      "loss": 3.5106,
+      "loss": 1.1806,
       "step": 583
     },
     {
       "epoch": 0.5748031496062992,
-      "grad_norm": 10.426263809204102,
+      "grad_norm": 12.608687400817871,
       "learning_rate": 5.755467196819086e-07,
-      "loss": 1.0963,
+      "loss": 1.6243,
       "step": 584
     },
     {
       "epoch": 0.5757874015748031,
-      "grad_norm": 35.70732116699219,
+      "grad_norm": 17.16189956665039,
       "learning_rate": 5.765407554671969e-07,
-      "loss": 3.2798,
+      "loss": 1.739,
       "step": 585
     },
     {
       "epoch": 0.5767716535433071,
-      "grad_norm": 11.900673866271973,
+      "grad_norm": 11.89455509185791,
       "learning_rate": 5.775347912524851e-07,
-      "loss": 1.3846,
+      "loss": 1.2502,
       "step": 586
     },
     {
       "epoch": 0.577755905511811,
-      "grad_norm": 12.225541114807129,
+      "grad_norm": 17.690635681152344,
       "learning_rate": 5.785288270377734e-07,
-      "loss": 1.4765,
+      "loss": 1.6328,
       "step": 587
     },
     {
       "epoch": 0.5787401574803149,
-      "grad_norm": 12.415202140808105,
+      "grad_norm": 12.663801193237305,
       "learning_rate": 5.795228628230618e-07,
-      "loss": 1.5319,
+      "loss": 1.3618,
       "step": 588
     },
     {
       "epoch": 0.5797244094488189,
-      "grad_norm": 8.8843412399292,
+      "grad_norm": 14.101471900939941,
       "learning_rate": 5.8051689860835e-07,
-      "loss": 0.8532,
+      "loss": 1.1535,
       "step": 589
     },
     {
       "epoch": 0.5807086614173228,
-      "grad_norm": 10.474663734436035,
+      "grad_norm": 13.188057899475098,
       "learning_rate": 5.815109343936383e-07,
-      "loss": 0.9573,
+      "loss": 1.2214,
       "step": 590
     },
     {
       "epoch": 0.5816929133858267,
-      "grad_norm": 11.181775093078613,
+      "grad_norm": 14.179088592529297,
       "learning_rate": 5.825049701789266e-07,
-      "loss": 1.2687,
+      "loss": 1.4884,
       "step": 591
     },
     {
       "epoch": 0.5826771653543307,
-      "grad_norm": 23.9904842376709,
+      "grad_norm": 10.61578369140625,
       "learning_rate": 5.834990059642148e-07,
-      "loss": 2.5192,
+      "loss": 1.4029,
       "step": 592
     },
     {
       "epoch": 0.5836614173228346,
-      "grad_norm": 13.039846420288086,
+      "grad_norm": 10.220196723937988,
       "learning_rate": 5.844930417495031e-07,
-      "loss": 1.3108,
+      "loss": 1.0542,
       "step": 593
     },
     {
       "epoch": 0.5846456692913385,
-      "grad_norm": 13.52120590209961,
+      "grad_norm": 12.367060661315918,
       "learning_rate": 5.854870775347914e-07,
-      "loss": 1.7258,
+      "loss": 1.5848,
       "step": 594
     },
     {
       "epoch": 0.5856299212598425,
-      "grad_norm": 10.26899528503418,
+      "grad_norm": 11.365569114685059,
       "learning_rate": 5.864811133200797e-07,
-      "loss": 0.9112,
+      "loss": 1.405,
       "step": 595
     },
     {
       "epoch": 0.5866141732283464,
-      "grad_norm": 11.116101264953613,
+      "grad_norm": 12.305469512939453,
       "learning_rate": 5.874751491053679e-07,
-      "loss": 1.2594,
+      "loss": 1.6281,
       "step": 596
     },
     {
       "epoch": 0.5875984251968503,
-      "grad_norm": 32.65583038330078,
+      "grad_norm": 12.690118789672852,
       "learning_rate": 5.884691848906561e-07,
-      "loss": 2.8427,
+      "loss": 1.5228,
       "step": 597
     },
     {
       "epoch": 0.5885826771653543,
-      "grad_norm": 8.176895141601562,
+      "grad_norm": 14.444221496582031,
       "learning_rate": 5.894632206759445e-07,
-      "loss": 0.7377,
+      "loss": 1.8192,
       "step": 598
     },
     {
       "epoch": 0.5895669291338582,
-      "grad_norm": 12.208337783813477,
+      "grad_norm": 10.509269714355469,
       "learning_rate": 5.904572564612327e-07,
-      "loss": 1.0449,
+      "loss": 1.2403,
       "step": 599
     },
     {
       "epoch": 0.5905511811023622,
-      "grad_norm": 35.78986740112305,
+      "grad_norm": 16.332834243774414,
       "learning_rate": 5.914512922465209e-07,
-      "loss": 3.2569,
+      "loss": 1.9368,
       "step": 600
     },
     {
       "epoch": 0.5915354330708661,
-      "grad_norm": 10.76382827758789,
+      "grad_norm": 14.727821350097656,
       "learning_rate": 5.924453280318093e-07,
-      "loss": 1.0936,
+      "loss": 1.6623,
       "step": 601
     },
     {
       "epoch": 0.59251968503937,
-      "grad_norm": 11.242823600769043,
+      "grad_norm": 14.168722152709961,
       "learning_rate": 5.934393638170975e-07,
-      "loss": 1.0935,
+      "loss": 1.495,
       "step": 602
     },
     {
       "epoch": 0.593503937007874,
-      "grad_norm": 15.131819725036621,
+      "grad_norm": 12.486218452453613,
       "learning_rate": 5.944333996023858e-07,
-      "loss": 1.5585,
+      "loss": 1.7079,
       "step": 603
     },
     {
       "epoch": 0.594488188976378,
-      "grad_norm": 10.615683555603027,
+      "grad_norm": 11.449934959411621,
       "learning_rate": 5.95427435387674e-07,
-      "loss": 0.8383,
+      "loss": 1.0651,
       "step": 604
     },
     {
       "epoch": 0.5954724409448819,
-      "grad_norm": 11.556757926940918,
+      "grad_norm": 11.699078559875488,
       "learning_rate": 5.964214711729623e-07,
-      "loss": 0.9299,
+      "loss": 1.2121,
       "step": 605
     },
     {
       "epoch": 0.5964566929133859,
-      "grad_norm": 12.370502471923828,
+      "grad_norm": 12.642477989196777,
       "learning_rate": 5.974155069582506e-07,
-      "loss": 1.2652,
+      "loss": 1.5385,
       "step": 606
     },
     {
       "epoch": 0.5974409448818898,
-      "grad_norm": 10.400749206542969,
+      "grad_norm": 13.445345878601074,
       "learning_rate": 5.984095427435388e-07,
-      "loss": 0.9225,
+      "loss": 1.1015,
       "step": 607
     },
     {
       "epoch": 0.5984251968503937,
-      "grad_norm": 11.763941764831543,
+      "grad_norm": 15.42099380493164,
       "learning_rate": 5.994035785288272e-07,
-      "loss": 1.2258,
+      "loss": 1.7596,
       "step": 608
     },
     {
       "epoch": 0.5994094488188977,
-      "grad_norm": 11.520954132080078,
+      "grad_norm": 13.161843299865723,
       "learning_rate": 6.003976143141154e-07,
-      "loss": 1.0937,
+      "loss": 1.5597,
       "step": 609
     },
     {
       "epoch": 0.6003937007874016,
-      "grad_norm": 12.111637115478516,
+      "grad_norm": 12.604528427124023,
       "learning_rate": 6.013916500994036e-07,
-      "loss": 1.0317,
+      "loss": 1.3254,
       "step": 610
     },
     {
       "epoch": 0.6013779527559056,
-      "grad_norm": 12.023083686828613,
+      "grad_norm": 12.85647964477539,
       "learning_rate": 6.02385685884692e-07,
-      "loss": 0.9925,
+      "loss": 1.3595,
       "step": 611
     },
     {
       "epoch": 0.6023622047244095,
-      "grad_norm": 10.958356857299805,
+      "grad_norm": 11.856555938720703,
       "learning_rate": 6.033797216699802e-07,
-      "loss": 1.0347,
+      "loss": 1.0326,
       "step": 612
     },
     {
       "epoch": 0.6023622047244095,
-      "eval_Qnli-dev_cosine_accuracy": 0.6328125,
-      "eval_Qnli-dev_cosine_accuracy_threshold": 0.8777655363082886,
-      "eval_Qnli-dev_cosine_ap": 0.636082184555122,
-      "eval_Qnli-dev_cosine_f1": 0.6467817896389324,
-      "eval_Qnli-dev_cosine_f1_threshold": 0.78821861743927,
-      "eval_Qnli-dev_cosine_precision": 0.513715710723192,
-      "eval_Qnli-dev_cosine_recall": 0.8728813559322034,
-      "eval_Qnli-dev_dot_accuracy": 0.6328125,
-      "eval_Qnli-dev_dot_accuracy_threshold": 671.5428466796875,
-      "eval_Qnli-dev_dot_ap": 0.6345552658179916,
-      "eval_Qnli-dev_dot_f1": 0.6466876971608833,
-      "eval_Qnli-dev_dot_f1_threshold": 603.1448974609375,
-      "eval_Qnli-dev_dot_precision": 0.5150753768844221,
-      "eval_Qnli-dev_dot_recall": 0.8686440677966102,
-      "eval_Qnli-dev_euclidean_accuracy": 0.6328125,
-      "eval_Qnli-dev_euclidean_accuracy_threshold": 13.673242568969727,
-      "eval_Qnli-dev_euclidean_ap": 0.6364530484256814,
-      "eval_Qnli-dev_euclidean_f1": 0.6467817896389324,
-      "eval_Qnli-dev_euclidean_f1_threshold": 17.990753173828125,
-      "eval_Qnli-dev_euclidean_precision": 0.513715710723192,
-      "eval_Qnli-dev_euclidean_recall": 0.8728813559322034,
-      "eval_Qnli-dev_manhattan_accuracy": 0.646484375,
-      "eval_Qnli-dev_manhattan_accuracy_threshold": 294.42578125,
-      "eval_Qnli-dev_manhattan_ap": 0.6537409929922839,
-      "eval_Qnli-dev_manhattan_f1": 0.6527570789865872,
-      "eval_Qnli-dev_manhattan_f1_threshold": 385.14154052734375,
-      "eval_Qnli-dev_manhattan_precision": 0.503448275862069,
-      "eval_Qnli-dev_manhattan_recall": 0.9279661016949152,
-      "eval_Qnli-dev_max_accuracy": 0.646484375,
-      "eval_Qnli-dev_max_accuracy_threshold": 671.5428466796875,
-      "eval_Qnli-dev_max_ap": 0.6537409929922839,
-      "eval_Qnli-dev_max_f1": 0.6527570789865872,
-      "eval_Qnli-dev_max_f1_threshold": 603.1448974609375,
-      "eval_Qnli-dev_max_precision": 0.5150753768844221,
-      "eval_Qnli-dev_max_recall": 0.9279661016949152,
-      "eval_allNLI-dev_cosine_accuracy": 0.6796875,
-      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9677231311798096,
-      "eval_allNLI-dev_cosine_ap": 0.46405508143112484,
-      "eval_allNLI-dev_cosine_f1": 0.5277246653919695,
-      "eval_allNLI-dev_cosine_f1_threshold": 0.8723927140235901,
-      "eval_allNLI-dev_cosine_precision": 0.3942857142857143,
-      "eval_allNLI-dev_cosine_recall": 0.7976878612716763,
-      "eval_allNLI-dev_dot_accuracy": 0.681640625,
-      "eval_allNLI-dev_dot_accuracy_threshold": 728.7240600585938,
-      "eval_allNLI-dev_dot_ap": 0.46535056618997517,
-      "eval_allNLI-dev_dot_f1": 0.5288640595903166,
-      "eval_allNLI-dev_dot_f1_threshold": 664.1909790039062,
-      "eval_allNLI-dev_dot_precision": 0.3901098901098901,
-      "eval_allNLI-dev_dot_recall": 0.8208092485549133,
-      "eval_allNLI-dev_euclidean_accuracy": 0.6796875,
-      "eval_allNLI-dev_euclidean_accuracy_threshold": 7.031080722808838,
-      "eval_allNLI-dev_euclidean_ap": 0.46418150524434637,
-      "eval_allNLI-dev_euclidean_f1": 0.5287356321839081,
-      "eval_allNLI-dev_euclidean_f1_threshold": 13.958231925964355,
-      "eval_allNLI-dev_euclidean_precision": 0.3954154727793696,
-      "eval_allNLI-dev_euclidean_recall": 0.7976878612716763,
-      "eval_allNLI-dev_manhattan_accuracy": 0.67578125,
-      "eval_allNLI-dev_manhattan_accuracy_threshold": 178.7028045654297,
-      "eval_allNLI-dev_manhattan_ap": 0.45843625593780585,
-      "eval_allNLI-dev_manhattan_f1": 0.5275142314990512,
-      "eval_allNLI-dev_manhattan_f1_threshold": 285.9658203125,
-      "eval_allNLI-dev_manhattan_precision": 0.3926553672316384,
-      "eval_allNLI-dev_manhattan_recall": 0.8034682080924855,
-      "eval_allNLI-dev_max_accuracy": 0.681640625,
-      "eval_allNLI-dev_max_accuracy_threshold": 728.7240600585938,
-      "eval_allNLI-dev_max_ap": 0.46535056618997517,
-      "eval_allNLI-dev_max_f1": 0.5288640595903166,
-      "eval_allNLI-dev_max_f1_threshold": 664.1909790039062,
-      "eval_allNLI-dev_max_precision": 0.3954154727793696,
-      "eval_allNLI-dev_max_recall": 0.8208092485549133,
-      "eval_loss": 1.6341288089752197,
-      "eval_runtime": 54.4154,
-      "eval_samples_per_second": 30.58,
-      "eval_sequential_score": 0.6537409929922839,
-      "eval_steps_per_second": 0.129,
-      "eval_sts-test_pearson_cosine": 0.37733602550373435,
-      "eval_sts-test_pearson_dot": 0.37306741740811944,
-      "eval_sts-test_pearson_euclidean": 0.39932799761186,
-      "eval_sts-test_pearson_manhattan": 0.4113008973824271,
-      "eval_sts-test_pearson_max": 0.4113008973824271,
-      "eval_sts-test_spearman_cosine": 0.4216275445487205,
-      "eval_sts-test_spearman_dot": 0.41610367259887304,
-      "eval_sts-test_spearman_euclidean": 0.4220386743871232,
-      "eval_sts-test_spearman_manhattan": 0.4336883675037937,
-      "eval_sts-test_spearman_max": 0.4336883675037937,
+      "eval_Qnli-dev_cosine_accuracy": 0.646484375,
+      "eval_Qnli-dev_cosine_accuracy_threshold": 0.8883535861968994,
+      "eval_Qnli-dev_cosine_ap": 0.6529248124763861,
+      "eval_Qnli-dev_cosine_f1": 0.656195462478185,
+      "eval_Qnli-dev_cosine_f1_threshold": 0.8338419795036316,
+      "eval_Qnli-dev_cosine_precision": 0.5578635014836796,
+      "eval_Qnli-dev_cosine_recall": 0.7966101694915254,
+      "eval_Qnli-dev_dot_accuracy": 0.64453125,
+      "eval_Qnli-dev_dot_accuracy_threshold": 681.0067138671875,
+      "eval_Qnli-dev_dot_ap": 0.6519832432307548,
+      "eval_Qnli-dev_dot_f1": 0.657293497363796,
+      "eval_Qnli-dev_dot_f1_threshold": 639.3973999023438,
+      "eval_Qnli-dev_dot_precision": 0.5615615615615616,
+      "eval_Qnli-dev_dot_recall": 0.7923728813559322,
+      "eval_Qnli-dev_euclidean_accuracy": 0.646484375,
+      "eval_Qnli-dev_euclidean_accuracy_threshold": 13.06938362121582,
+      "eval_Qnli-dev_euclidean_ap": 0.6529216999296377,
+      "eval_Qnli-dev_euclidean_f1": 0.656195462478185,
+      "eval_Qnli-dev_euclidean_f1_threshold": 15.943912506103516,
+      "eval_Qnli-dev_euclidean_precision": 0.5578635014836796,
+      "eval_Qnli-dev_euclidean_recall": 0.7966101694915254,
+      "eval_Qnli-dev_manhattan_accuracy": 0.65234375,
+      "eval_Qnli-dev_manhattan_accuracy_threshold": 260.0885314941406,
+      "eval_Qnli-dev_manhattan_ap": 0.6597671356125537,
+      "eval_Qnli-dev_manhattan_f1": 0.6655948553054662,
+      "eval_Qnli-dev_manhattan_f1_threshold": 340.8906555175781,
+      "eval_Qnli-dev_manhattan_precision": 0.5362694300518135,
+      "eval_Qnli-dev_manhattan_recall": 0.8771186440677966,
+      "eval_Qnli-dev_max_accuracy": 0.65234375,
+      "eval_Qnli-dev_max_accuracy_threshold": 681.0067138671875,
+      "eval_Qnli-dev_max_ap": 0.6597671356125537,
+      "eval_Qnli-dev_max_f1": 0.6655948553054662,
+      "eval_Qnli-dev_max_f1_threshold": 639.3973999023438,
+      "eval_Qnli-dev_max_precision": 0.5615615615615616,
+      "eval_Qnli-dev_max_recall": 0.8771186440677966,
+      "eval_allNLI-dev_cosine_accuracy": 0.671875,
+      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9779053926467896,
+      "eval_allNLI-dev_cosine_ap": 0.447770779436667,
+      "eval_allNLI-dev_cosine_f1": 0.5192629815745394,
+      "eval_allNLI-dev_cosine_f1_threshold": 0.8607640862464905,
+      "eval_allNLI-dev_cosine_precision": 0.36556603773584906,
+      "eval_allNLI-dev_cosine_recall": 0.8959537572254336,
+      "eval_allNLI-dev_dot_accuracy": 0.671875,
+      "eval_allNLI-dev_dot_accuracy_threshold": 749.451904296875,
+      "eval_allNLI-dev_dot_ap": 0.4471854335007342,
+      "eval_allNLI-dev_dot_f1": 0.5192629815745394,
+      "eval_allNLI-dev_dot_f1_threshold": 659.1720581054688,
+      "eval_allNLI-dev_dot_precision": 0.36556603773584906,
+      "eval_allNLI-dev_dot_recall": 0.8959537572254336,
+      "eval_allNLI-dev_euclidean_accuracy": 0.671875,
+      "eval_allNLI-dev_euclidean_accuracy_threshold": 5.819397449493408,
+      "eval_allNLI-dev_euclidean_ap": 0.44774506125516483,
+      "eval_allNLI-dev_euclidean_f1": 0.5192629815745394,
+      "eval_allNLI-dev_euclidean_f1_threshold": 14.603199005126953,
+      "eval_allNLI-dev_euclidean_precision": 0.36556603773584906,
+      "eval_allNLI-dev_euclidean_recall": 0.8959537572254336,
+      "eval_allNLI-dev_manhattan_accuracy": 0.671875,
+      "eval_allNLI-dev_manhattan_accuracy_threshold": 119.47159576416016,
+      "eval_allNLI-dev_manhattan_ap": 0.4448387653657968,
+      "eval_allNLI-dev_manhattan_f1": 0.5315904139433552,
+      "eval_allNLI-dev_manhattan_f1_threshold": 245.42477416992188,
+      "eval_allNLI-dev_manhattan_precision": 0.42657342657342656,
+      "eval_allNLI-dev_manhattan_recall": 0.7052023121387283,
+      "eval_allNLI-dev_max_accuracy": 0.671875,
+      "eval_allNLI-dev_max_accuracy_threshold": 749.451904296875,
+      "eval_allNLI-dev_max_ap": 0.447770779436667,
+      "eval_allNLI-dev_max_f1": 0.5315904139433552,
+      "eval_allNLI-dev_max_f1_threshold": 659.1720581054688,
+      "eval_allNLI-dev_max_precision": 0.42657342657342656,
+      "eval_allNLI-dev_max_recall": 0.8959537572254336,
+      "eval_loss": 1.6315439939498901,
+      "eval_runtime": 52.9207,
+      "eval_samples_per_second": 31.443,
+      "eval_sequential_score": 0.6597671356125537,
+      "eval_steps_per_second": 0.132,
+      "eval_sts-test_pearson_cosine": 0.31607787444026963,
+      "eval_sts-test_pearson_dot": 0.31281708435791855,
+      "eval_sts-test_pearson_euclidean": 0.33857077884567166,
+      "eval_sts-test_pearson_manhattan": 0.3555732019703781,
+      "eval_sts-test_pearson_max": 0.3555732019703781,
+      "eval_sts-test_spearman_cosine": 0.35971175784560405,
+      "eval_sts-test_spearman_dot": 0.3566100031673048,
+      "eval_sts-test_spearman_euclidean": 0.35988199164393697,
+      "eval_sts-test_spearman_manhattan": 0.376050443379604,
+      "eval_sts-test_spearman_max": 0.376050443379604,
       "step": 612
     },
     {
       "epoch": 0.6033464566929134,
-      "grad_norm": 10.701803207397461,
+      "grad_norm": 15.954797744750977,
       "learning_rate": 6.043737574552684e-07,
-      "loss": 0.7597,
+      "loss": 1.4822,
       "step": 613
     },
     {
       "epoch": 0.6043307086614174,
-      "grad_norm": 44.23878860473633,
+      "grad_norm": 11.097407341003418,
       "learning_rate": 6.053677932405567e-07,
-      "loss": 4.1533,
+      "loss": 0.9997,
       "step": 614
     },
     {
       "epoch": 0.6053149606299213,
-      "grad_norm": 11.628232955932617,
+      "grad_norm": 14.609465599060059,
       "learning_rate": 6.06361829025845e-07,
-      "loss": 0.9556,
+      "loss": 1.4946,
       "step": 615
     },
     {
       "epoch": 0.6062992125984252,
-      "grad_norm": 15.238367080688477,
+      "grad_norm": 11.441244125366211,
       "learning_rate": 6.073558648111333e-07,
-      "loss": 1.3079,
+      "loss": 1.3491,
       "step": 616
     },
     {
       "epoch": 0.6072834645669292,
-      "grad_norm": 14.660409927368164,
+      "grad_norm": 12.638875961303711,
       "learning_rate": 6.083499005964215e-07,
-      "loss": 1.5096,
+      "loss": 1.2118,
       "step": 617
     },
     {
       "epoch": 0.6082677165354331,
-      "grad_norm": 40.2839469909668,
+      "grad_norm": 15.746068000793457,
       "learning_rate": 6.093439363817097e-07,
-      "loss": 3.8311,
+      "loss": 1.6948,
       "step": 618
     },
     {
       "epoch": 0.609251968503937,
-      "grad_norm": 13.805837631225586,
+      "grad_norm": 20.515592575073242,
       "learning_rate": 6.103379721669982e-07,
-      "loss": 1.379,
+      "loss": 2.0457,
       "step": 619
     },
     {
       "epoch": 0.610236220472441,
-      "grad_norm": 46.00523376464844,
+      "grad_norm": 13.20988655090332,
       "learning_rate": 6.113320079522864e-07,
-      "loss": 3.8273,
+      "loss": 1.2724,
       "step": 620
     },
     {
       "epoch": 0.6112204724409449,
-      "grad_norm": 13.986534118652344,
+      "grad_norm": 11.521592140197754,
       "learning_rate": 6.123260437375747e-07,
-      "loss": 1.296,
+      "loss": 1.2677,
       "step": 621
     },
     {
       "epoch": 0.6122047244094488,
-      "grad_norm": 15.609278678894043,
+      "grad_norm": 14.984801292419434,
       "learning_rate": 6.13320079522863e-07,
-      "loss": 1.6233,
+      "loss": 1.6639,
       "step": 622
     },
     {
       "epoch": 0.6131889763779528,
-      "grad_norm": 11.434459686279297,
+      "grad_norm": 13.088191032409668,
       "learning_rate": 6.143141153081512e-07,
-      "loss": 0.7542,
+      "loss": 1.139,
       "step": 623
     },
     {
       "epoch": 0.6141732283464567,
-      "grad_norm": 13.063506126403809,
+      "grad_norm": 14.182477951049805,
       "learning_rate": 6.153081510934395e-07,
-      "loss": 1.0578,
+      "loss": 1.2424,
       "step": 624
     },
     {
       "epoch": 0.6151574803149606,
-      "grad_norm": 30.77376365661621,
+      "grad_norm": 10.056435585021973,
       "learning_rate": 6.163021868787277e-07,
-      "loss": 2.4721,
+      "loss": 0.8214,
       "step": 625
     },
     {
       "epoch": 0.6161417322834646,
-      "grad_norm": 30.752870559692383,
+      "grad_norm": 11.403252601623535,
       "learning_rate": 6.172962226640161e-07,
-      "loss": 2.6996,
+      "loss": 1.4398,
       "step": 626
     },
     {
       "epoch": 0.6171259842519685,
-      "grad_norm": 13.579484939575195,
+      "grad_norm": 12.417586326599121,
       "learning_rate": 6.182902584493043e-07,
-      "loss": 1.6649,
+      "loss": 1.6008,
       "step": 627
     },
     {
       "epoch": 0.6181102362204725,
-      "grad_norm": 30.221582412719727,
+      "grad_norm": 14.504150390625,
       "learning_rate": 6.192842942345925e-07,
-      "loss": 2.5207,
+      "loss": 1.3225,
       "step": 628
     },
     {
       "epoch": 0.6190944881889764,
-      "grad_norm": 12.64453411102295,
+      "grad_norm": 15.959015846252441,
       "learning_rate": 6.202783300198809e-07,
-      "loss": 1.2837,
+      "loss": 1.4452,
       "step": 629
     },
     {
       "epoch": 0.6200787401574803,
-      "grad_norm": 11.140278816223145,
+      "grad_norm": 17.7681827545166,
       "learning_rate": 6.212723658051691e-07,
-      "loss": 1.1731,
+      "loss": 1.7723,
       "step": 630
     },
     {
       "epoch": 0.6210629921259843,
-      "grad_norm": 11.152597427368164,
+      "grad_norm": 10.61798095703125,
       "learning_rate": 6.222664015904573e-07,
-      "loss": 0.9778,
+      "loss": 1.035,
       "step": 631
     },
     {
       "epoch": 0.6220472440944882,
-      "grad_norm": 12.832036972045898,
+      "grad_norm": 13.359830856323242,
       "learning_rate": 6.232604373757457e-07,
-      "loss": 1.6909,
+      "loss": 1.6132,
       "step": 632
     },
     {
       "epoch": 0.6230314960629921,
-      "grad_norm": 12.344291687011719,
+      "grad_norm": 11.638774871826172,
       "learning_rate": 6.242544731610339e-07,
-      "loss": 1.5441,
+      "loss": 1.4167,
       "step": 633
     },
     {
       "epoch": 0.6240157480314961,
-      "grad_norm": 11.180489540100098,
+      "grad_norm": 11.718339920043945,
       "learning_rate": 6.252485089463222e-07,
-      "loss": 0.9197,
+      "loss": 1.1764,
       "step": 634
     },
     {
       "epoch": 0.625,
-      "grad_norm": 23.448375701904297,
+      "grad_norm": 9.881278991699219,
       "learning_rate": 6.262425447316104e-07,
-      "loss": 1.781,
+      "loss": 1.0702,
       "step": 635
     },
     {
       "epoch": 0.6259842519685039,
-      "grad_norm": 10.159034729003906,
+      "grad_norm": 15.579713821411133,
       "learning_rate": 6.272365805168987e-07,
-      "loss": 1.0285,
+      "loss": 1.3096,
       "step": 636
     },
     {
       "epoch": 0.6269685039370079,
-      "grad_norm": 13.20181941986084,
+      "grad_norm": 12.148575782775879,
       "learning_rate": 6.28230616302187e-07,
-      "loss": 1.213,
+      "loss": 1.4174,
       "step": 637
     },
     {
       "epoch": 0.6279527559055118,
-      "grad_norm": 11.607129096984863,
+      "grad_norm": 13.778498649597168,
       "learning_rate": 6.292246520874752e-07,
-      "loss": 1.0478,
+      "loss": 1.8742,
       "step": 638
     },
     {
       "epoch": 0.6289370078740157,
-      "grad_norm": 16.6951961517334,
+      "grad_norm": 11.97413158416748,
       "learning_rate": 6.302186878727636e-07,
-      "loss": 1.6243,
+      "loss": 1.3372,
       "step": 639
     },
     {
       "epoch": 0.6299212598425197,
-      "grad_norm": 13.98421573638916,
+      "grad_norm": 11.662862777709961,
       "learning_rate": 6.312127236580518e-07,
-      "loss": 1.4093,
+      "loss": 1.4453,
       "step": 640
     },
     {
       "epoch": 0.6309055118110236,
-      "grad_norm": 26.42371368408203,
+      "grad_norm": 10.738568305969238,
       "learning_rate": 6.3220675944334e-07,
-      "loss": 1.8847,
+      "loss": 0.6792,
       "step": 641
     },
     {
       "epoch": 0.6318897637795275,
-      "grad_norm": 10.854850769042969,
+      "grad_norm": 11.657475471496582,
       "learning_rate": 6.332007952286283e-07,
-      "loss": 1.0713,
+      "loss": 1.2363,
       "step": 642
     },
     {
       "epoch": 0.6328740157480315,
-      "grad_norm": 11.060506820678711,
+      "grad_norm": 12.533220291137695,
       "learning_rate": 6.341948310139166e-07,
-      "loss": 0.9998,
+      "loss": 1.2037,
       "step": 643
     },
     {
       "epoch": 0.6338582677165354,
-      "grad_norm": 10.658958435058594,
+      "grad_norm": 10.946784019470215,
       "learning_rate": 6.351888667992048e-07,
-      "loss": 0.9684,
+      "loss": 1.105,
       "step": 644
     },
     {
       "epoch": 0.6348425196850394,
-      "grad_norm": 14.677898406982422,
+      "grad_norm": 17.624608993530273,
       "learning_rate": 6.361829025844931e-07,
-      "loss": 2.1272,
+      "loss": 1.8319,
       "step": 645
     },
     {
       "epoch": 0.6358267716535433,
-      "grad_norm": 12.596633911132812,
+      "grad_norm": 13.263932228088379,
       "learning_rate": 6.371769383697813e-07,
-      "loss": 1.416,
+      "loss": 1.4261,
       "step": 646
     },
     {
       "epoch": 0.6368110236220472,
-      "grad_norm": 10.448723793029785,
+      "grad_norm": 11.63705062866211,
       "learning_rate": 6.381709741550698e-07,
-      "loss": 1.0463,
+      "loss": 1.0431,
       "step": 647
     },
     {
       "epoch": 0.6377952755905512,
-      "grad_norm": 11.114981651306152,
+      "grad_norm": 13.618841171264648,
       "learning_rate": 6.39165009940358e-07,
-      "loss": 0.9779,
+      "loss": 1.2132,
       "step": 648
     },
     {
       "epoch": 0.6387795275590551,
-      "grad_norm": 8.395352363586426,
+      "grad_norm": 12.975966453552246,
       "learning_rate": 6.401590457256462e-07,
-      "loss": 0.5093,
+      "loss": 1.1464,
       "step": 649
     },
     {
       "epoch": 0.639763779527559,
-      "grad_norm": 12.399072647094727,
+      "grad_norm": 12.637139320373535,
       "learning_rate": 6.411530815109346e-07,
-      "loss": 1.1007,
+      "loss": 1.3076,
       "step": 650
     },
     {
       "epoch": 0.640748031496063,
-      "grad_norm": 10.01904582977295,
+      "grad_norm": 12.68572998046875,
       "learning_rate": 6.421471172962228e-07,
-      "loss": 0.9436,
+      "loss": 0.9162,
       "step": 651
     },
     {
       "epoch": 0.6417322834645669,
-      "grad_norm": 14.65745735168457,
+      "grad_norm": 15.307023048400879,
       "learning_rate": 6.431411530815111e-07,
-      "loss": 1.4446,
+      "loss": 1.6739,
       "step": 652
     },
     {
       "epoch": 0.6427165354330708,
-      "grad_norm": 15.491090774536133,
+      "grad_norm": 16.65964698791504,
       "learning_rate": 6.441351888667994e-07,
-      "loss": 1.7327,
+      "loss": 1.3627,
       "step": 653
     },
     {
       "epoch": 0.6437007874015748,
-      "grad_norm": 10.455428123474121,
+      "grad_norm": 10.542638778686523,
       "learning_rate": 6.451292246520876e-07,
-      "loss": 0.8326,
+      "loss": 1.084,
       "step": 654
     },
     {
       "epoch": 0.6446850393700787,
-      "grad_norm": 9.583117485046387,
+      "grad_norm": 11.227409362792969,
       "learning_rate": 6.461232604373759e-07,
-      "loss": 0.6514,
+      "loss": 0.8349,
       "step": 655
     },
     {
       "epoch": 0.6456692913385826,
-      "grad_norm": 9.990079879760742,
+      "grad_norm": 14.71562671661377,
       "learning_rate": 6.471172962226641e-07,
-      "loss": 0.6155,
+      "loss": 1.5575,
       "step": 656
     },
     {
       "epoch": 0.6466535433070866,
-      "grad_norm": 10.108931541442871,
+      "grad_norm": 10.660284042358398,
       "learning_rate": 6.481113320079524e-07,
-      "loss": 0.7906,
+      "loss": 0.7503,
       "step": 657
     },
     {
       "epoch": 0.6476377952755905,
-      "grad_norm": 13.665596008300781,
+      "grad_norm": 12.722469329833984,
       "learning_rate": 6.491053677932407e-07,
-      "loss": 1.4234,
+      "loss": 1.5593,
       "step": 658
     },
     {
       "epoch": 0.6486220472440944,
-      "grad_norm": 11.515149116516113,
+      "grad_norm": 13.289987564086914,
       "learning_rate": 6.500994035785289e-07,
-      "loss": 0.9298,
+      "loss": 1.1788,
       "step": 659
     },
     {
       "epoch": 0.6496062992125984,
-      "grad_norm": 17.00029945373535,
+      "grad_norm": 13.905250549316406,
       "learning_rate": 6.510934393638173e-07,
-      "loss": 1.7384,
+      "loss": 1.5119,
       "step": 660
     },
     {
       "epoch": 0.6505905511811023,
-      "grad_norm": 15.062308311462402,
+      "grad_norm": 14.397658348083496,
       "learning_rate": 6.520874751491055e-07,
-      "loss": 1.7171,
+      "loss": 1.1793,
       "step": 661
     },
     {
       "epoch": 0.6515748031496063,
-      "grad_norm": 15.243254661560059,
+      "grad_norm": 13.769511222839355,
       "learning_rate": 6.530815109343937e-07,
-      "loss": 1.5247,
+      "loss": 1.1577,
       "step": 662
     },
     {
       "epoch": 0.6525590551181102,
-      "grad_norm": 12.88428783416748,
+      "grad_norm": 17.989856719970703,
       "learning_rate": 6.54075546719682e-07,
-      "loss": 1.2509,
+      "loss": 1.4391,
       "step": 663
     },
     {
       "epoch": 0.6535433070866141,
-      "grad_norm": 34.313804626464844,
+      "grad_norm": 13.776409149169922,
       "learning_rate": 6.550695825049703e-07,
-      "loss": 3.0989,
+      "loss": 1.2291,
       "step": 664
     },
     {
       "epoch": 0.6545275590551181,
-      "grad_norm": 10.890783309936523,
+      "grad_norm": 11.71041202545166,
       "learning_rate": 6.560636182902586e-07,
-      "loss": 0.8984,
+      "loss": 1.3118,
       "step": 665
     },
     {
       "epoch": 0.655511811023622,
-      "grad_norm": 12.740053176879883,
+      "grad_norm": 12.15799331665039,
       "learning_rate": 6.570576540755468e-07,
-      "loss": 1.0749,
+      "loss": 1.1358,
       "step": 666
     },
     {
       "epoch": 0.656496062992126,
-      "grad_norm": 65.30494689941406,
+      "grad_norm": 15.679516792297363,
       "learning_rate": 6.580516898608351e-07,
-      "loss": 5.5445,
+      "loss": 1.814,
       "step": 667
     },
     {
       "epoch": 0.65748031496063,
-      "grad_norm": 12.40191650390625,
+      "grad_norm": 13.018841743469238,
       "learning_rate": 6.590457256461234e-07,
-      "loss": 1.15,
+      "loss": 1.038,
       "step": 668
     },
     {
       "epoch": 0.6584645669291339,
-      "grad_norm": 13.165613174438477,
+      "grad_norm": 12.75416088104248,
       "learning_rate": 6.600397614314116e-07,
-      "loss": 1.0755,
+      "loss": 1.2344,
       "step": 669
     },
     {
       "epoch": 0.6594488188976378,
-      "grad_norm": 11.294352531433105,
+      "grad_norm": 12.071971893310547,
       "learning_rate": 6.610337972166998e-07,
-      "loss": 0.93,
+      "loss": 1.1064,
       "step": 670
     },
     {
       "epoch": 0.6604330708661418,
-      "grad_norm": 14.429823875427246,
+      "grad_norm": 13.168946266174316,
       "learning_rate": 6.620278330019882e-07,
-      "loss": 1.6097,
+      "loss": 1.8745,
       "step": 671
     },
     {
       "epoch": 0.6614173228346457,
-      "grad_norm": 13.069869995117188,
+      "grad_norm": 14.136277198791504,
       "learning_rate": 6.630218687872764e-07,
-      "loss": 1.0799,
+      "loss": 1.8775,
       "step": 672
     },
     {
       "epoch": 0.6624015748031497,
-      "grad_norm": 12.076016426086426,
+      "grad_norm": 11.940804481506348,
       "learning_rate": 6.640159045725647e-07,
-      "loss": 1.1492,
+      "loss": 1.2582,
       "step": 673
     },
     {
       "epoch": 0.6633858267716536,
-      "grad_norm": 13.563544273376465,
+      "grad_norm": 11.855066299438477,
       "learning_rate": 6.65009940357853e-07,
-      "loss": 1.0605,
+      "loss": 1.3454,
       "step": 674
     },
     {
       "epoch": 0.6643700787401575,
-      "grad_norm": 12.923056602478027,
+      "grad_norm": 15.009392738342285,
       "learning_rate": 6.660039761431412e-07,
-      "loss": 0.9923,
+      "loss": 1.2374,
       "step": 675
     },
     {
       "epoch": 0.6653543307086615,
-      "grad_norm": 8.722559928894043,
+      "grad_norm": 9.056105613708496,
       "learning_rate": 6.669980119284295e-07,
-      "loss": 0.5501,
+      "loss": 0.7227,
       "step": 676
     },
     {
       "epoch": 0.6663385826771654,
-      "grad_norm": 15.11403751373291,
+      "grad_norm": 11.413203239440918,
       "learning_rate": 6.679920477137177e-07,
-      "loss": 1.1595,
+      "loss": 1.2036,
       "step": 677
     },
     {
       "epoch": 0.6673228346456693,
-      "grad_norm": 12.444266319274902,
+      "grad_norm": 12.716734886169434,
       "learning_rate": 6.689860834990062e-07,
-      "loss": 0.968,
+      "loss": 1.5248,
       "step": 678
     },
     {
       "epoch": 0.6683070866141733,
-      "grad_norm": 39.157779693603516,
+      "grad_norm": 11.400150299072266,
       "learning_rate": 6.699801192842944e-07,
-      "loss": 3.3578,
+      "loss": 1.0584,
       "step": 679
     },
     {
       "epoch": 0.6692913385826772,
-      "grad_norm": 47.95648193359375,
+      "grad_norm": 10.649139404296875,
       "learning_rate": 6.709741550695826e-07,
-      "loss": 4.1605,
+      "loss": 0.7455,
       "step": 680
     },
     {
       "epoch": 0.6702755905511811,
-      "grad_norm": 11.361122131347656,
+      "grad_norm": 11.595582962036133,
       "learning_rate": 6.71968190854871e-07,
-      "loss": 0.7277,
+      "loss": 1.0616,
       "step": 681
     },
     {
       "epoch": 0.6712598425196851,
-      "grad_norm": 8.56065559387207,
+      "grad_norm": 13.45544147491455,
       "learning_rate": 6.729622266401592e-07,
-      "loss": 0.7908,
+      "loss": 1.1377,
       "step": 682
     },
     {
       "epoch": 0.672244094488189,
-      "grad_norm": 9.499028205871582,
+      "grad_norm": 13.17196273803711,
       "learning_rate": 6.739562624254474e-07,
-      "loss": 0.898,
+      "loss": 1.7022,
       "step": 683
     },
     {
       "epoch": 0.6732283464566929,
-      "grad_norm": 11.447650909423828,
+      "grad_norm": 14.776825904846191,
       "learning_rate": 6.749502982107357e-07,
-      "loss": 1.0208,
+      "loss": 1.4008,
       "step": 684
     },
     {
       "epoch": 0.6742125984251969,
-      "grad_norm": 14.240975379943848,
+      "grad_norm": 15.56824016571045,
       "learning_rate": 6.75944333996024e-07,
-      "loss": 1.652,
+      "loss": 1.4894,
       "step": 685
     },
     {
       "epoch": 0.6751968503937008,
-      "grad_norm": 13.617683410644531,
+      "grad_norm": 14.228174209594727,
       "learning_rate": 6.769383697813123e-07,
-      "loss": 1.4247,
+      "loss": 1.5464,
       "step": 686
     },
     {
       "epoch": 0.6761811023622047,
-      "grad_norm": 36.25943374633789,
+      "grad_norm": 15.94443130493164,
       "learning_rate": 6.779324055666005e-07,
-      "loss": 3.3876,
+      "loss": 1.5018,
       "step": 687
     },
     {
       "epoch": 0.6771653543307087,
-      "grad_norm": 12.248785018920898,
+      "grad_norm": 20.589385986328125,
       "learning_rate": 6.789264413518888e-07,
-      "loss": 1.5638,
+      "loss": 1.5803,
       "step": 688
     },
     {
       "epoch": 0.6781496062992126,
-      "grad_norm": 9.890121459960938,
+      "grad_norm": 18.15032386779785,
       "learning_rate": 6.799204771371771e-07,
-      "loss": 1.009,
+      "loss": 1.5654,
       "step": 689
     },
     {
       "epoch": 0.6791338582677166,
-      "grad_norm": 11.641477584838867,
+      "grad_norm": 13.775176048278809,
       "learning_rate": 6.809145129224653e-07,
-      "loss": 1.1282,
+      "loss": 1.2537,
       "step": 690
     },
     {
       "epoch": 0.6801181102362205,
-      "grad_norm": 11.62278938293457,
+      "grad_norm": 15.423773765563965,
       "learning_rate": 6.819085487077536e-07,
-      "loss": 1.2746,
+      "loss": 1.4901,
       "step": 691
     },
     {
       "epoch": 0.6811023622047244,
-      "grad_norm": 11.104626655578613,
+      "grad_norm": 11.028498649597168,
       "learning_rate": 6.829025844930419e-07,
-      "loss": 1.1079,
+      "loss": 0.9597,
       "step": 692
     },
     {
       "epoch": 0.6820866141732284,
-      "grad_norm": 10.597916603088379,
+      "grad_norm": 14.687771797180176,
       "learning_rate": 6.838966202783301e-07,
-      "loss": 1.1276,
+      "loss": 1.5515,
       "step": 693
     },
     {
       "epoch": 0.6830708661417323,
-      "grad_norm": 28.668432235717773,
+      "grad_norm": 11.73508071899414,
       "learning_rate": 6.848906560636184e-07,
-      "loss": 2.5246,
+      "loss": 1.1314,
       "step": 694
     },
     {
       "epoch": 0.6840551181102362,
-      "grad_norm": 9.761340141296387,
+      "grad_norm": 11.076573371887207,
       "learning_rate": 6.858846918489067e-07,
-      "loss": 0.807,
+      "loss": 1.2442,
       "step": 695
     },
     {
       "epoch": 0.6850393700787402,
-      "grad_norm": 11.57912826538086,
+      "grad_norm": 14.134485244750977,
       "learning_rate": 6.86878727634195e-07,
-      "loss": 1.0713,
+      "loss": 1.5542,
       "step": 696
     },
     {
       "epoch": 0.6860236220472441,
-      "grad_norm": 9.34850788116455,
+      "grad_norm": 9.774703979492188,
       "learning_rate": 6.878727634194832e-07,
-      "loss": 0.8295,
+      "loss": 0.7617,
       "step": 697
     },
     {
       "epoch": 0.687007874015748,
-      "grad_norm": 9.771157264709473,
+      "grad_norm": 14.868870735168457,
       "learning_rate": 6.888667992047714e-07,
-      "loss": 1.1652,
+      "loss": 1.4626,
       "step": 698
     },
     {
       "epoch": 0.687992125984252,
-      "grad_norm": 43.168426513671875,
+      "grad_norm": 11.818559646606445,
       "learning_rate": 6.898608349900598e-07,
-      "loss": 2.9347,
+      "loss": 0.6546,
       "step": 699
     },
     {
       "epoch": 0.6889763779527559,
-      "grad_norm": 9.494790077209473,
+      "grad_norm": 13.716598510742188,
       "learning_rate": 6.90854870775348e-07,
-      "loss": 0.7807,
+      "loss": 1.1726,
       "step": 700
     },
     {
       "epoch": 0.6899606299212598,
-      "grad_norm": 23.544239044189453,
+      "grad_norm": 10.127545356750488,
       "learning_rate": 6.918489065606362e-07,
-      "loss": 1.3768,
+      "loss": 0.7522,
       "step": 701
     },
     {
       "epoch": 0.6909448818897638,
-      "grad_norm": 11.414657592773438,
+      "grad_norm": 14.03452205657959,
       "learning_rate": 6.928429423459246e-07,
-      "loss": 1.4596,
+      "loss": 1.6203,
       "step": 702
     },
     {
       "epoch": 0.6919291338582677,
-      "grad_norm": 13.431307792663574,
+      "grad_norm": 11.538923263549805,
       "learning_rate": 6.938369781312128e-07,
-      "loss": 1.5273,
+      "loss": 1.0822,
       "step": 703
     },
     {
       "epoch": 0.6929133858267716,
-      "grad_norm": 10.998498916625977,
+      "grad_norm": 11.976446151733398,
       "learning_rate": 6.948310139165011e-07,
-      "loss": 0.8857,
+      "loss": 0.9652,
       "step": 704
     },
     {
       "epoch": 0.6938976377952756,
-      "grad_norm": 12.236613273620605,
+      "grad_norm": 11.482376098632812,
       "learning_rate": 6.958250497017893e-07,
-      "loss": 1.0254,
+      "loss": 0.9884,
       "step": 705
     },
     {
       "epoch": 0.6948818897637795,
-      "grad_norm": 10.905499458312988,
+      "grad_norm": 12.661773681640625,
       "learning_rate": 6.968190854870776e-07,
-      "loss": 0.6571,
+      "loss": 1.3876,
       "step": 706
     },
     {
       "epoch": 0.6958661417322834,
-      "grad_norm": 35.93000411987305,
+      "grad_norm": 11.961230278015137,
       "learning_rate": 6.97813121272366e-07,
-      "loss": 2.6553,
+      "loss": 1.1726,
       "step": 707
     },
     {
       "epoch": 0.6968503937007874,
-      "grad_norm": 12.845637321472168,
+      "grad_norm": 11.375978469848633,
       "learning_rate": 6.988071570576542e-07,
-      "loss": 1.3157,
+      "loss": 1.1426,
       "step": 708
     },
     {
       "epoch": 0.6978346456692913,
-      "grad_norm": 14.18862247467041,
+      "grad_norm": 13.53009033203125,
       "learning_rate": 6.998011928429426e-07,
-      "loss": 1.3675,
+      "loss": 1.2303,
       "step": 709
     },
     {
       "epoch": 0.6988188976377953,
-      "grad_norm": 10.507378578186035,
+      "grad_norm": 12.14291763305664,
       "learning_rate": 7.007952286282308e-07,
-      "loss": 0.8492,
+      "loss": 1.3438,
       "step": 710
     },
     {
       "epoch": 0.6998031496062992,
-      "grad_norm": 12.715144157409668,
+      "grad_norm": 13.431678771972656,
       "learning_rate": 7.01789264413519e-07,
-      "loss": 1.0831,
+      "loss": 1.3823,
       "step": 711
     },
     {
       "epoch": 0.7007874015748031,
-      "grad_norm": 10.003392219543457,
+      "grad_norm": 14.398938179016113,
       "learning_rate": 7.027833001988073e-07,
-      "loss": 0.7888,
+      "loss": 1.5166,
       "step": 712
     },
     {
       "epoch": 0.7017716535433071,
-      "grad_norm": 13.558688163757324,
+      "grad_norm": 12.663697242736816,
       "learning_rate": 7.037773359840956e-07,
-      "loss": 1.2725,
+      "loss": 1.4148,
       "step": 713
     },
     {
       "epoch": 0.702755905511811,
-      "grad_norm": 12.053690910339355,
+      "grad_norm": 11.511486053466797,
       "learning_rate": 7.047713717693838e-07,
-      "loss": 1.1746,
+      "loss": 0.9564,
       "step": 714
     },
     {
       "epoch": 0.7037401574803149,
-      "grad_norm": 8.193392753601074,
+      "grad_norm": 14.274373054504395,
       "learning_rate": 7.057654075546721e-07,
-      "loss": 0.6175,
+      "loss": 1.1031,
       "step": 715
     },
     {
       "epoch": 0.7047244094488189,
-      "grad_norm": 8.654151916503906,
+      "grad_norm": 22.680253982543945,
       "learning_rate": 7.067594433399604e-07,
-      "loss": 0.5935,
+      "loss": 1.6628,
       "step": 716
     },
     {
       "epoch": 0.7057086614173228,
-      "grad_norm": 13.256050109863281,
+      "grad_norm": 14.918832778930664,
       "learning_rate": 7.077534791252487e-07,
-      "loss": 1.1907,
+      "loss": 1.3693,
       "step": 717
     },
     {
       "epoch": 0.7066929133858267,
-      "grad_norm": 13.537592887878418,
+      "grad_norm": 9.6978120803833,
       "learning_rate": 7.087475149105369e-07,
-      "loss": 1.4747,
+      "loss": 1.0247,
       "step": 718
     },
     {
       "epoch": 0.7076771653543307,
-      "grad_norm": 11.818737030029297,
+      "grad_norm": 10.600605010986328,
       "learning_rate": 7.097415506958251e-07,
-      "loss": 0.7373,
+      "loss": 0.7778,
       "step": 719
     },
     {
       "epoch": 0.7086614173228346,
-      "grad_norm": 38.487003326416016,
+      "grad_norm": 13.40506649017334,
       "learning_rate": 7.107355864811135e-07,
-      "loss": 3.7169,
+      "loss": 1.3158,
       "step": 720
     },
     {
       "epoch": 0.7096456692913385,
-      "grad_norm": 13.312034606933594,
+      "grad_norm": 14.183016777038574,
       "learning_rate": 7.117296222664017e-07,
-      "loss": 1.0055,
+      "loss": 1.5329,
       "step": 721
     },
     {
       "epoch": 0.7106299212598425,
-      "grad_norm": 12.317737579345703,
+      "grad_norm": 12.718838691711426,
       "learning_rate": 7.1272365805169e-07,
-      "loss": 1.0311,
+      "loss": 1.2845,
       "step": 722
     },
     {
       "epoch": 0.7116141732283464,
-      "grad_norm": 8.979562759399414,
+      "grad_norm": 10.078266143798828,
       "learning_rate": 7.137176938369783e-07,
-      "loss": 0.6121,
+      "loss": 0.7895,
       "step": 723
     },
     {
       "epoch": 0.7125984251968503,
-      "grad_norm": 11.30450439453125,
+      "grad_norm": 9.505541801452637,
       "learning_rate": 7.147117296222665e-07,
-      "loss": 0.8375,
+      "loss": 0.9098,
       "step": 724
     },
     {
       "epoch": 0.7135826771653543,
-      "grad_norm": 10.606871604919434,
+      "grad_norm": 9.941719055175781,
       "learning_rate": 7.157057654075548e-07,
-      "loss": 0.6943,
+      "loss": 0.7648,
       "step": 725
     },
     {
       "epoch": 0.7145669291338582,
-      "grad_norm": 12.569619178771973,
+      "grad_norm": 13.63866138458252,
       "learning_rate": 7.16699801192843e-07,
-      "loss": 1.0684,
+      "loss": 1.0951,
       "step": 726
     },
     {
       "epoch": 0.7155511811023622,
-      "grad_norm": 9.287858009338379,
+      "grad_norm": 11.125609397888184,
       "learning_rate": 7.176938369781313e-07,
-      "loss": 0.7662,
+      "loss": 1.1351,
       "step": 727
     },
     {
       "epoch": 0.7165354330708661,
-      "grad_norm": 11.010154724121094,
+      "grad_norm": 13.634588241577148,
       "learning_rate": 7.186878727634196e-07,
-      "loss": 0.8516,
+      "loss": 1.0514,
       "step": 728
     },
     {
       "epoch": 0.71751968503937,
-      "grad_norm": 12.050772666931152,
+      "grad_norm": 12.125421524047852,
       "learning_rate": 7.196819085487078e-07,
-      "loss": 1.2911,
+      "loss": 1.2328,
       "step": 729
     },
     {
       "epoch": 0.718503937007874,
-      "grad_norm": 12.157442092895508,
+      "grad_norm": 10.89931583404541,
       "learning_rate": 7.206759443339962e-07,
-      "loss": 1.0036,
+      "loss": 1.0664,
       "step": 730
     },
     {
       "epoch": 0.719488188976378,
-      "grad_norm": 38.21113586425781,
+      "grad_norm": 12.2229585647583,
       "learning_rate": 7.216699801192844e-07,
-      "loss": 3.1986,
+      "loss": 1.1058,
       "step": 731
     },
     {
       "epoch": 0.7204724409448819,
-      "grad_norm": 13.931038856506348,
+      "grad_norm": 7.885047912597656,
       "learning_rate": 7.226640159045726e-07,
-      "loss": 0.9704,
+      "loss": 0.6698,
       "step": 732
     },
     {
       "epoch": 0.7214566929133859,
-      "grad_norm": 8.16901683807373,
+      "grad_norm": 13.67959976196289,
       "learning_rate": 7.236580516898609e-07,
-      "loss": 0.637,
+      "loss": 1.1438,
       "step": 733
     },
     {
       "epoch": 0.7224409448818898,
-      "grad_norm": 9.032388687133789,
+      "grad_norm": 13.96220874786377,
       "learning_rate": 7.246520874751492e-07,
-      "loss": 0.5723,
+      "loss": 1.0473,
       "step": 734
     },
     {
       "epoch": 0.7234251968503937,
-      "grad_norm": 10.507246971130371,
+      "grad_norm": 17.09735679626465,
       "learning_rate": 7.256461232604375e-07,
-      "loss": 1.0665,
+      "loss": 1.5307,
       "step": 735
     },
     {
       "epoch": 0.7244094488188977,
-      "grad_norm": 12.778374671936035,
+      "grad_norm": 13.294286727905273,
       "learning_rate": 7.266401590457257e-07,
-      "loss": 1.0845,
+      "loss": 1.2537,
       "step": 736
     },
     {
       "epoch": 0.7253937007874016,
-      "grad_norm": 12.195361137390137,
+      "grad_norm": 9.892070770263672,
       "learning_rate": 7.27634194831014e-07,
-      "loss": 1.0019,
+      "loss": 0.7006,
       "step": 737
     },
     {
       "epoch": 0.7263779527559056,
-      "grad_norm": 8.443058013916016,
+      "grad_norm": 11.488076210021973,
       "learning_rate": 7.286282306163024e-07,
-      "loss": 0.616,
+      "loss": 0.8487,
       "step": 738
     },
     {
       "epoch": 0.7273622047244095,
-      "grad_norm": 37.86750030517578,
+      "grad_norm": 12.460606575012207,
       "learning_rate": 7.296222664015906e-07,
-      "loss": 2.9126,
+      "loss": 1.193,
       "step": 739
     },
     {
       "epoch": 0.7283464566929134,
-      "grad_norm": 9.833470344543457,
+      "grad_norm": 12.237475395202637,
       "learning_rate": 7.306163021868788e-07,
-      "loss": 0.7203,
+      "loss": 0.9711,
       "step": 740
     },
     {
       "epoch": 0.7293307086614174,
-      "grad_norm": 10.621623992919922,
+      "grad_norm": 8.606354713439941,
       "learning_rate": 7.316103379721672e-07,
-      "loss": 0.617,
+      "loss": 0.5139,
       "step": 741
     },
     {
       "epoch": 0.7303149606299213,
-      "grad_norm": 54.747642517089844,
+      "grad_norm": 14.769782066345215,
       "learning_rate": 7.326043737574554e-07,
-      "loss": 4.7813,
+      "loss": 1.1185,
       "step": 742
     },
     {
       "epoch": 0.7312992125984252,
-      "grad_norm": 11.494114875793457,
+      "grad_norm": 11.785113334655762,
       "learning_rate": 7.335984095427437e-07,
-      "loss": 0.8951,
+      "loss": 0.9982,
       "step": 743
     },
     {
       "epoch": 0.7322834645669292,
-      "grad_norm": 9.549132347106934,
+      "grad_norm": 10.399678230285645,
       "learning_rate": 7.34592445328032e-07,
-      "loss": 0.6171,
+      "loss": 0.7987,
       "step": 744
     },
     {
       "epoch": 0.7332677165354331,
-      "grad_norm": 11.160932540893555,
+      "grad_norm": 14.253144264221191,
       "learning_rate": 7.355864811133202e-07,
-      "loss": 0.8032,
+      "loss": 1.161,
       "step": 745
     },
     {
       "epoch": 0.734251968503937,
-      "grad_norm": 13.318976402282715,
+      "grad_norm": 12.234946250915527,
       "learning_rate": 7.365805168986085e-07,
-      "loss": 1.1307,
+      "loss": 0.9832,
       "step": 746
     },
     {
       "epoch": 0.735236220472441,
-      "grad_norm": 38.743255615234375,
+      "grad_norm": 17.171052932739258,
       "learning_rate": 7.375745526838967e-07,
-      "loss": 3.4791,
+      "loss": 1.6041,
       "step": 747
     },
     {
       "epoch": 0.7362204724409449,
-      "grad_norm": 28.96615982055664,
+      "grad_norm": 12.20317268371582,
       "learning_rate": 7.385685884691851e-07,
-      "loss": 1.6899,
+      "loss": 1.1084,
       "step": 748
     },
     {
       "epoch": 0.7372047244094488,
-      "grad_norm": 9.659252166748047,
+      "grad_norm": 13.474609375,
       "learning_rate": 7.395626242544733e-07,
-      "loss": 0.7096,
+      "loss": 1.1264,
       "step": 749
     },
     {
       "epoch": 0.7381889763779528,
-      "grad_norm": 11.343924522399902,
+      "grad_norm": 10.15322494506836,
       "learning_rate": 7.405566600397615e-07,
-      "loss": 0.7839,
+      "loss": 0.6759,
       "step": 750
     },
     {
       "epoch": 0.7391732283464567,
-      "grad_norm": 13.733336448669434,
+      "grad_norm": 13.045278549194336,
       "learning_rate": 7.415506958250499e-07,
-      "loss": 1.3269,
+      "loss": 1.5241,
       "step": 751
     },
     {
       "epoch": 0.7401574803149606,
-      "grad_norm": 12.469435691833496,
+      "grad_norm": 14.404647827148438,
       "learning_rate": 7.425447316103381e-07,
-      "loss": 1.4379,
+      "loss": 1.6181,
       "step": 752
     },
     {
       "epoch": 0.7411417322834646,
-      "grad_norm": 13.951380729675293,
+      "grad_norm": 10.859138488769531,
       "learning_rate": 7.435387673956263e-07,
-      "loss": 1.055,
+      "loss": 1.1862,
       "step": 753
     },
     {
       "epoch": 0.7421259842519685,
-      "grad_norm": 12.946548461914062,
+      "grad_norm": 12.503321647644043,
       "learning_rate": 7.445328031809146e-07,
-      "loss": 0.9262,
+      "loss": 1.0147,
       "step": 754
     },
     {
       "epoch": 0.7431102362204725,
-      "grad_norm": 9.735945701599121,
+      "grad_norm": 12.434426307678223,
       "learning_rate": 7.455268389662029e-07,
-      "loss": 0.7023,
+      "loss": 1.0435,
       "step": 755
     },
     {
       "epoch": 0.7440944881889764,
-      "grad_norm": 32.48512649536133,
+      "grad_norm": 12.646941184997559,
       "learning_rate": 7.465208747514912e-07,
-      "loss": 2.6348,
+      "loss": 0.9175,
       "step": 756
     },
     {
       "epoch": 0.7450787401574803,
-      "grad_norm": 12.768858909606934,
+      "grad_norm": 14.548798561096191,
       "learning_rate": 7.475149105367794e-07,
-      "loss": 1.3171,
+      "loss": 1.5608,
       "step": 757
     },
     {
       "epoch": 0.7460629921259843,
-      "grad_norm": 26.709489822387695,
+      "grad_norm": 15.520113945007324,
       "learning_rate": 7.485089463220677e-07,
-      "loss": 2.5155,
+      "loss": 1.7027,
       "step": 758
     },
     {
       "epoch": 0.7470472440944882,
-      "grad_norm": 12.552984237670898,
+      "grad_norm": 11.138642311096191,
       "learning_rate": 7.49502982107356e-07,
-      "loss": 1.0247,
+      "loss": 0.9282,
       "step": 759
     },
     {
       "epoch": 0.7480314960629921,
-      "grad_norm": 13.25454044342041,
+      "grad_norm": 10.51353931427002,
       "learning_rate": 7.504970178926442e-07,
-      "loss": 0.9433,
+      "loss": 0.8072,
       "step": 760
     },
     {
       "epoch": 0.7490157480314961,
-      "grad_norm": 11.653966903686523,
+      "grad_norm": 15.635726928710938,
       "learning_rate": 7.514910536779325e-07,
-      "loss": 1.1192,
+      "loss": 1.13,
       "step": 761
     },
     {
       "epoch": 0.75,
-      "grad_norm": 11.929251670837402,
+      "grad_norm": 14.228267669677734,
       "learning_rate": 7.524850894632208e-07,
-      "loss": 1.0187,
+      "loss": 1.153,
       "step": 762
     },
     {
       "epoch": 0.7509842519685039,
-      "grad_norm": 11.60064697265625,
+      "grad_norm": 18.568565368652344,
       "learning_rate": 7.53479125248509e-07,
-      "loss": 0.8483,
+      "loss": 1.7723,
       "step": 763
     },
     {
       "epoch": 0.7519685039370079,
-      "grad_norm": 10.26369571685791,
+      "grad_norm": 11.066574096679688,
       "learning_rate": 7.544731610337973e-07,
-      "loss": 0.7776,
+      "loss": 0.9833,
       "step": 764
     },
     {
       "epoch": 0.7529527559055118,
-      "grad_norm": 32.42780303955078,
+      "grad_norm": 15.546330451965332,
       "learning_rate": 7.554671968190856e-07,
-      "loss": 2.2929,
+      "loss": 1.1766,
       "step": 765
     },
     {
       "epoch": 0.7529527559055118,
-      "eval_Qnli-dev_cosine_accuracy": 0.654296875,
-      "eval_Qnli-dev_cosine_accuracy_threshold": 0.8493565320968628,
-      "eval_Qnli-dev_cosine_ap": 0.6651485883101813,
-      "eval_Qnli-dev_cosine_f1": 0.6678023850085179,
-      "eval_Qnli-dev_cosine_f1_threshold": 0.7991494536399841,
-      "eval_Qnli-dev_cosine_precision": 0.5584045584045584,
-      "eval_Qnli-dev_cosine_recall": 0.8305084745762712,
-      "eval_Qnli-dev_dot_accuracy": 0.654296875,
-      "eval_Qnli-dev_dot_accuracy_threshold": 649.9716186523438,
-      "eval_Qnli-dev_dot_ap": 0.6631292557807136,
-      "eval_Qnli-dev_dot_f1": 0.6700507614213197,
-      "eval_Qnli-dev_dot_f1_threshold": 609.7298583984375,
-      "eval_Qnli-dev_dot_precision": 0.5577464788732395,
-      "eval_Qnli-dev_dot_recall": 0.8389830508474576,
-      "eval_Qnli-dev_euclidean_accuracy": 0.654296875,
-      "eval_Qnli-dev_euclidean_accuracy_threshold": 14.864194869995117,
-      "eval_Qnli-dev_euclidean_ap": 0.6652217784881351,
-      "eval_Qnli-dev_euclidean_f1": 0.6678023850085179,
-      "eval_Qnli-dev_euclidean_f1_threshold": 17.510093688964844,
-      "eval_Qnli-dev_euclidean_precision": 0.5584045584045584,
-      "eval_Qnli-dev_euclidean_recall": 0.8305084745762712,
-      "eval_Qnli-dev_manhattan_accuracy": 0.662109375,
-      "eval_Qnli-dev_manhattan_accuracy_threshold": 301.39422607421875,
-      "eval_Qnli-dev_manhattan_ap": 0.6763570733667617,
-      "eval_Qnli-dev_manhattan_f1": 0.6762820512820512,
-      "eval_Qnli-dev_manhattan_f1_threshold": 376.59600830078125,
-      "eval_Qnli-dev_manhattan_precision": 0.5438144329896907,
-      "eval_Qnli-dev_manhattan_recall": 0.8940677966101694,
-      "eval_Qnli-dev_max_accuracy": 0.662109375,
-      "eval_Qnli-dev_max_accuracy_threshold": 649.9716186523438,
-      "eval_Qnli-dev_max_ap": 0.6763570733667617,
-      "eval_Qnli-dev_max_f1": 0.6762820512820512,
-      "eval_Qnli-dev_max_f1_threshold": 609.7298583984375,
-      "eval_Qnli-dev_max_precision": 0.5584045584045584,
-      "eval_Qnli-dev_max_recall": 0.8940677966101694,
-      "eval_allNLI-dev_cosine_accuracy": 0.6875,
-      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9424331188201904,
-      "eval_allNLI-dev_cosine_ap": 0.4692955852848747,
-      "eval_allNLI-dev_cosine_f1": 0.5341365461847389,
-      "eval_allNLI-dev_cosine_f1_threshold": 0.8611772060394287,
-      "eval_allNLI-dev_cosine_precision": 0.40923076923076923,
-      "eval_allNLI-dev_cosine_recall": 0.7687861271676301,
-      "eval_allNLI-dev_dot_accuracy": 0.6875,
-      "eval_allNLI-dev_dot_accuracy_threshold": 722.18798828125,
-      "eval_allNLI-dev_dot_ap": 0.4694961600900589,
-      "eval_allNLI-dev_dot_f1": 0.5357873210633947,
-      "eval_allNLI-dev_dot_f1_threshold": 661.792724609375,
-      "eval_allNLI-dev_dot_precision": 0.41455696202531644,
-      "eval_allNLI-dev_dot_recall": 0.7572254335260116,
-      "eval_allNLI-dev_euclidean_accuracy": 0.6875,
-      "eval_allNLI-dev_euclidean_accuracy_threshold": 9.392923355102539,
-      "eval_allNLI-dev_euclidean_ap": 0.4693675979912956,
-      "eval_allNLI-dev_euclidean_f1": 0.5341365461847389,
-      "eval_allNLI-dev_euclidean_f1_threshold": 14.569525718688965,
-      "eval_allNLI-dev_euclidean_precision": 0.40923076923076923,
-      "eval_allNLI-dev_euclidean_recall": 0.7687861271676301,
+      "eval_Qnli-dev_cosine_accuracy": 0.650390625,
+      "eval_Qnli-dev_cosine_accuracy_threshold": 0.8711802363395691,
+      "eval_Qnli-dev_cosine_ap": 0.6636184479177414,
+      "eval_Qnli-dev_cosine_f1": 0.6573208722741433,
+      "eval_Qnli-dev_cosine_f1_threshold": 0.779100775718689,
+      "eval_Qnli-dev_cosine_precision": 0.5197044334975369,
+      "eval_Qnli-dev_cosine_recall": 0.8940677966101694,
+      "eval_Qnli-dev_dot_accuracy": 0.650390625,
+      "eval_Qnli-dev_dot_accuracy_threshold": 664.5684814453125,
+      "eval_Qnli-dev_dot_ap": 0.6615631968512774,
+      "eval_Qnli-dev_dot_f1": 0.6568047337278107,
+      "eval_Qnli-dev_dot_f1_threshold": 582.806396484375,
+      "eval_Qnli-dev_dot_precision": 0.5045454545454545,
+      "eval_Qnli-dev_dot_recall": 0.940677966101695,
+      "eval_Qnli-dev_euclidean_accuracy": 0.650390625,
+      "eval_Qnli-dev_euclidean_accuracy_threshold": 14.019062995910645,
+      "eval_Qnli-dev_euclidean_ap": 0.663466792452131,
+      "eval_Qnli-dev_euclidean_f1": 0.6573208722741433,
+      "eval_Qnli-dev_euclidean_f1_threshold": 18.369121551513672,
+      "eval_Qnli-dev_euclidean_precision": 0.5197044334975369,
+      "eval_Qnli-dev_euclidean_recall": 0.8940677966101694,
+      "eval_Qnli-dev_manhattan_accuracy": 0.65625,
+      "eval_Qnli-dev_manhattan_accuracy_threshold": 303.57794189453125,
+      "eval_Qnli-dev_manhattan_ap": 0.6740758222672537,
+      "eval_Qnli-dev_manhattan_f1": 0.6689655172413793,
+      "eval_Qnli-dev_manhattan_f1_threshold": 349.1427001953125,
+      "eval_Qnli-dev_manhattan_precision": 0.563953488372093,
+      "eval_Qnli-dev_manhattan_recall": 0.8220338983050848,
+      "eval_Qnli-dev_max_accuracy": 0.65625,
+      "eval_Qnli-dev_max_accuracy_threshold": 664.5684814453125,
+      "eval_Qnli-dev_max_ap": 0.6740758222672537,
+      "eval_Qnli-dev_max_f1": 0.6689655172413793,
+      "eval_Qnli-dev_max_f1_threshold": 582.806396484375,
+      "eval_Qnli-dev_max_precision": 0.563953488372093,
+      "eval_Qnli-dev_max_recall": 0.940677966101695,
+      "eval_allNLI-dev_cosine_accuracy": 0.677734375,
+      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9697319269180298,
+      "eval_allNLI-dev_cosine_ap": 0.47097839371304523,
+      "eval_allNLI-dev_cosine_f1": 0.5189003436426117,
+      "eval_allNLI-dev_cosine_f1_threshold": 0.8454164862632751,
+      "eval_allNLI-dev_cosine_precision": 0.3691931540342298,
+      "eval_allNLI-dev_cosine_recall": 0.8728323699421965,
+      "eval_allNLI-dev_dot_accuracy": 0.677734375,
+      "eval_allNLI-dev_dot_accuracy_threshold": 737.8129272460938,
+      "eval_allNLI-dev_dot_ap": 0.4710099600965258,
+      "eval_allNLI-dev_dot_f1": 0.5189003436426117,
+      "eval_allNLI-dev_dot_f1_threshold": 647.1653442382812,
+      "eval_allNLI-dev_dot_precision": 0.3691931540342298,
+      "eval_allNLI-dev_dot_recall": 0.8728323699421965,
+      "eval_allNLI-dev_euclidean_accuracy": 0.677734375,
+      "eval_allNLI-dev_euclidean_accuracy_threshold": 6.808244228363037,
+      "eval_allNLI-dev_euclidean_ap": 0.47125375440506867,
+      "eval_allNLI-dev_euclidean_f1": 0.5189003436426117,
+      "eval_allNLI-dev_euclidean_f1_threshold": 15.383974075317383,
+      "eval_allNLI-dev_euclidean_precision": 0.3691931540342298,
+      "eval_allNLI-dev_euclidean_recall": 0.8728323699421965,
       "eval_allNLI-dev_manhattan_accuracy": 0.677734375,
-      "eval_allNLI-dev_manhattan_accuracy_threshold": 195.32534790039062,
-      "eval_allNLI-dev_manhattan_ap": 0.4693162755377951,
-      "eval_allNLI-dev_manhattan_f1": 0.5338645418326693,
-      "eval_allNLI-dev_manhattan_f1_threshold": 301.98321533203125,
-      "eval_allNLI-dev_manhattan_precision": 0.4072948328267477,
-      "eval_allNLI-dev_manhattan_recall": 0.7745664739884393,
-      "eval_allNLI-dev_max_accuracy": 0.6875,
-      "eval_allNLI-dev_max_accuracy_threshold": 722.18798828125,
-      "eval_allNLI-dev_max_ap": 0.4694961600900589,
-      "eval_allNLI-dev_max_f1": 0.5357873210633947,
-      "eval_allNLI-dev_max_f1_threshold": 661.792724609375,
-      "eval_allNLI-dev_max_precision": 0.41455696202531644,
-      "eval_allNLI-dev_max_recall": 0.7745664739884393,
-      "eval_loss": 1.460240364074707,
-      "eval_runtime": 54.1723,
-      "eval_samples_per_second": 30.717,
-      "eval_sequential_score": 0.6763570733667617,
-      "eval_steps_per_second": 0.129,
-      "eval_sts-test_pearson_cosine": 0.39163897925771674,
-      "eval_sts-test_pearson_dot": 0.3859127227394329,
-      "eval_sts-test_pearson_euclidean": 0.416510609899249,
-      "eval_sts-test_pearson_manhattan": 0.42444677513396134,
-      "eval_sts-test_pearson_max": 0.42444677513396134,
-      "eval_sts-test_spearman_cosine": 0.43941218808953797,
-      "eval_sts-test_spearman_dot": 0.4330934889861178,
-      "eval_sts-test_spearman_euclidean": 0.4398173660101757,
-      "eval_sts-test_spearman_manhattan": 0.4483853082623888,
-      "eval_sts-test_spearman_max": 0.4483853082623888,
+      "eval_allNLI-dev_manhattan_accuracy_threshold": 182.8551788330078,
+      "eval_allNLI-dev_manhattan_ap": 0.4685839203620036,
+      "eval_allNLI-dev_manhattan_f1": 0.5330296127562643,
+      "eval_allNLI-dev_manhattan_f1_threshold": 261.1466979980469,
+      "eval_allNLI-dev_manhattan_precision": 0.4398496240601504,
+      "eval_allNLI-dev_manhattan_recall": 0.6763005780346821,
+      "eval_allNLI-dev_max_accuracy": 0.677734375,
+      "eval_allNLI-dev_max_accuracy_threshold": 737.8129272460938,
+      "eval_allNLI-dev_max_ap": 0.47125375440506867,
+      "eval_allNLI-dev_max_f1": 0.5330296127562643,
+      "eval_allNLI-dev_max_f1_threshold": 647.1653442382812,
+      "eval_allNLI-dev_max_precision": 0.4398496240601504,
+      "eval_allNLI-dev_max_recall": 0.8728323699421965,
+      "eval_loss": 1.3821483850479126,
+      "eval_runtime": 52.6629,
+      "eval_samples_per_second": 31.597,
+      "eval_sequential_score": 0.6740758222672537,
+      "eval_steps_per_second": 0.133,
+      "eval_sts-test_pearson_cosine": 0.41074334497722165,
+      "eval_sts-test_pearson_dot": 0.407552380755139,
+      "eval_sts-test_pearson_euclidean": 0.4333967868726292,
+      "eval_sts-test_pearson_manhattan": 0.44332181906543383,
+      "eval_sts-test_pearson_max": 0.44332181906543383,
+      "eval_sts-test_spearman_cosine": 0.4583378890486231,
+      "eval_sts-test_spearman_dot": 0.4553192375302001,
+      "eval_sts-test_spearman_euclidean": 0.458448737920612,
+      "eval_sts-test_spearman_manhattan": 0.4653892556459323,
+      "eval_sts-test_spearman_max": 0.4653892556459323,
       "step": 765
     },
     {
       "epoch": 0.7539370078740157,
-      "grad_norm": 14.009566307067871,
+      "grad_norm": 16.530473709106445,
       "learning_rate": 7.56461232604374e-07,
-      "loss": 1.1385,
+      "loss": 1.2351,
       "step": 766
     },
     {
       "epoch": 0.7549212598425197,
-      "grad_norm": 9.024477005004883,
+      "grad_norm": 14.440428733825684,
       "learning_rate": 7.574552683896621e-07,
-      "loss": 0.5783,
+      "loss": 0.9521,
       "step": 767
     },
     {
       "epoch": 0.7559055118110236,
-      "grad_norm": 14.691167831420898,
+      "grad_norm": 12.20114517211914,
       "learning_rate": 7.584493041749503e-07,
-      "loss": 1.2594,
+      "loss": 1.1981,
       "step": 768
     },
     {
       "epoch": 0.7568897637795275,
-      "grad_norm": 30.12328338623047,
+      "grad_norm": 11.787071228027344,
       "learning_rate": 7.594433399602388e-07,
-      "loss": 2.1814,
+      "loss": 0.8031,
       "step": 769
     },
     {
       "epoch": 0.7578740157480315,
-      "grad_norm": 11.479731559753418,
+      "grad_norm": 12.98501205444336,
       "learning_rate": 7.60437375745527e-07,
-      "loss": 0.9557,
+      "loss": 1.2502,
       "step": 770
     },
     {
       "epoch": 0.7588582677165354,
-      "grad_norm": 9.11027717590332,
+      "grad_norm": 10.086615562438965,
       "learning_rate": 7.614314115308152e-07,
-      "loss": 0.6649,
+      "loss": 0.9313,
       "step": 771
     },
     {
       "epoch": 0.7598425196850394,
-      "grad_norm": 11.371647834777832,
+      "grad_norm": 9.566267013549805,
       "learning_rate": 7.624254473161036e-07,
-      "loss": 0.8908,
+      "loss": 0.8738,
       "step": 772
     },
     {
       "epoch": 0.7608267716535433,
-      "grad_norm": 10.068663597106934,
+      "grad_norm": 11.557473182678223,
       "learning_rate": 7.634194831013918e-07,
-      "loss": 0.8234,
+      "loss": 1.0316,
       "step": 773
     },
     {
       "epoch": 0.7618110236220472,
-      "grad_norm": 12.031545639038086,
+      "grad_norm": 10.591485023498535,
       "learning_rate": 7.644135188866801e-07,
-      "loss": 1.0889,
+      "loss": 1.0679,
       "step": 774
     },
     {
       "epoch": 0.7627952755905512,
-      "grad_norm": 11.676977157592773,
+      "grad_norm": 12.123056411743164,
       "learning_rate": 7.654075546719683e-07,
-      "loss": 1.0294,
+      "loss": 1.0308,
       "step": 775
     },
     {
       "epoch": 0.7637795275590551,
-      "grad_norm": 12.664962768554688,
+      "grad_norm": 11.250408172607422,
       "learning_rate": 7.664015904572566e-07,
-      "loss": 1.207,
+      "loss": 1.0663,
       "step": 776
     },
     {
       "epoch": 0.764763779527559,
-      "grad_norm": 11.427294731140137,
+      "grad_norm": 13.228967666625977,
       "learning_rate": 7.673956262425449e-07,
-      "loss": 0.9177,
+      "loss": 1.3565,
       "step": 777
     },
     {
       "epoch": 0.765748031496063,
-      "grad_norm": 11.521842002868652,
+      "grad_norm": 9.849665641784668,
       "learning_rate": 7.683896620278331e-07,
-      "loss": 1.118,
+      "loss": 0.9103,
       "step": 778
     },
     {
       "epoch": 0.7667322834645669,
-      "grad_norm": 12.860966682434082,
+      "grad_norm": 14.536798477172852,
       "learning_rate": 7.693836978131215e-07,
-      "loss": 1.2775,
+      "loss": 1.477,
       "step": 779
     },
     {
       "epoch": 0.7677165354330708,
-      "grad_norm": 34.428462982177734,
+      "grad_norm": 10.974257469177246,
       "learning_rate": 7.703777335984097e-07,
-      "loss": 3.6265,
+      "loss": 1.2526,
       "step": 780
     },
     {
       "epoch": 0.7687007874015748,
-      "grad_norm": 11.847333908081055,
+      "grad_norm": 9.089542388916016,
       "learning_rate": 7.713717693836979e-07,
-      "loss": 1.0477,
+      "loss": 0.8647,
       "step": 781
     },
     {
       "epoch": 0.7696850393700787,
-      "grad_norm": 11.537668228149414,
+      "grad_norm": 11.030006408691406,
       "learning_rate": 7.723658051689862e-07,
-      "loss": 1.0475,
+      "loss": 0.8498,
       "step": 782
     },
     {
       "epoch": 0.7706692913385826,
-      "grad_norm": 10.223224639892578,
+      "grad_norm": 12.539122581481934,
       "learning_rate": 7.733598409542745e-07,
-      "loss": 0.8965,
+      "loss": 1.5257,
       "step": 783
     },
     {
       "epoch": 0.7716535433070866,
-      "grad_norm": 9.149170875549316,
+      "grad_norm": 8.903121948242188,
       "learning_rate": 7.743538767395627e-07,
-      "loss": 0.6398,
+      "loss": 0.5493,
       "step": 784
     },
     {
       "epoch": 0.7726377952755905,
-      "grad_norm": 9.898460388183594,
+      "grad_norm": 12.023555755615234,
       "learning_rate": 7.75347912524851e-07,
-      "loss": 0.8387,
+      "loss": 1.2321,
       "step": 785
     },
     {
       "epoch": 0.7736220472440944,
-      "grad_norm": 14.476304054260254,
+      "grad_norm": 14.207155227661133,
       "learning_rate": 7.763419483101393e-07,
-      "loss": 1.2206,
+      "loss": 1.4306,
       "step": 786
     },
     {
       "epoch": 0.7746062992125984,
-      "grad_norm": 10.1232328414917,
+      "grad_norm": 12.97458553314209,
       "learning_rate": 7.773359840954276e-07,
-      "loss": 0.7254,
+      "loss": 1.0428,
       "step": 787
     },
     {
       "epoch": 0.7755905511811023,
-      "grad_norm": 29.228551864624023,
+      "grad_norm": 14.624722480773926,
       "learning_rate": 7.783300198807158e-07,
-      "loss": 2.674,
+      "loss": 1.3553,
       "step": 788
     },
     {
       "epoch": 0.7765748031496063,
-      "grad_norm": 11.327478408813477,
+      "grad_norm": 11.624364852905273,
       "learning_rate": 7.79324055666004e-07,
-      "loss": 0.9569,
+      "loss": 1.1102,
       "step": 789
     },
     {
       "epoch": 0.7775590551181102,
-      "grad_norm": 12.248971939086914,
+      "grad_norm": 11.759018898010254,
       "learning_rate": 7.803180914512924e-07,
-      "loss": 1.1421,
+      "loss": 0.9225,
       "step": 790
     },
     {
       "epoch": 0.7785433070866141,
-      "grad_norm": 11.644203186035156,
+      "grad_norm": 11.817508697509766,
       "learning_rate": 7.813121272365806e-07,
-      "loss": 1.0097,
+      "loss": 1.2585,
       "step": 791
     },
     {
       "epoch": 0.7795275590551181,
-      "grad_norm": 9.664244651794434,
+      "grad_norm": 15.326850891113281,
       "learning_rate": 7.823061630218689e-07,
-      "loss": 0.7003,
+      "loss": 1.1335,
       "step": 792
     },
     {
       "epoch": 0.780511811023622,
-      "grad_norm": 10.34313678741455,
+      "grad_norm": 14.029367446899414,
       "learning_rate": 7.833001988071572e-07,
-      "loss": 0.9617,
+      "loss": 1.2781,
       "step": 793
     },
     {
       "epoch": 0.781496062992126,
-      "grad_norm": 10.9442720413208,
+      "grad_norm": 15.042057037353516,
       "learning_rate": 7.842942345924454e-07,
-      "loss": 0.7657,
+      "loss": 1.0265,
       "step": 794
     },
     {
       "epoch": 0.78248031496063,
-      "grad_norm": 13.110860824584961,
+      "grad_norm": 13.87203311920166,
       "learning_rate": 7.852882703777337e-07,
-      "loss": 1.0451,
+      "loss": 1.1846,
       "step": 795
     },
     {
       "epoch": 0.7834645669291339,
-      "grad_norm": 32.66413879394531,
+      "grad_norm": 16.487451553344727,
       "learning_rate": 7.862823061630219e-07,
-      "loss": 2.6532,
+      "loss": 1.2123,
       "step": 796
     },
     {
       "epoch": 0.7844488188976378,
-      "grad_norm": 31.113216400146484,
+      "grad_norm": 11.526494026184082,
       "learning_rate": 7.872763419483102e-07,
-      "loss": 2.2881,
+      "loss": 1.0861,
       "step": 797
     },
     {
       "epoch": 0.7854330708661418,
-      "grad_norm": 8.918703079223633,
+      "grad_norm": 12.820661544799805,
       "learning_rate": 7.882703777335986e-07,
-      "loss": 0.6881,
+      "loss": 1.1929,
       "step": 798
     },
     {
       "epoch": 0.7864173228346457,
-      "grad_norm": 13.535049438476562,
+      "grad_norm": 8.670584678649902,
       "learning_rate": 7.892644135188868e-07,
-      "loss": 0.8976,
+      "loss": 0.8202,
       "step": 799
     },
     {
       "epoch": 0.7874015748031497,
-      "grad_norm": 26.764535903930664,
+      "grad_norm": 14.31804084777832,
       "learning_rate": 7.902584493041752e-07,
-      "loss": 2.1921,
+      "loss": 1.0745,
       "step": 800
     },
     {
       "epoch": 0.7883858267716536,
-      "grad_norm": 9.47220516204834,
+      "grad_norm": 15.200295448303223,
       "learning_rate": 7.912524850894634e-07,
-      "loss": 0.5112,
+      "loss": 0.9162,
       "step": 801
     },
     {
       "epoch": 0.7893700787401575,
-      "grad_norm": 10.911540985107422,
+      "grad_norm": 14.89583969116211,
       "learning_rate": 7.922465208747516e-07,
-      "loss": 0.775,
+      "loss": 1.2077,
       "step": 802
     },
     {
       "epoch": 0.7903543307086615,
-      "grad_norm": 28.311187744140625,
+      "grad_norm": 17.461294174194336,
       "learning_rate": 7.932405566600399e-07,
-      "loss": 1.9774,
+      "loss": 1.432,
       "step": 803
     },
     {
       "epoch": 0.7913385826771654,
-      "grad_norm": 28.16878890991211,
+      "grad_norm": 13.266355514526367,
       "learning_rate": 7.942345924453282e-07,
-      "loss": 1.8192,
+      "loss": 0.9384,
       "step": 804
     },
     {
       "epoch": 0.7923228346456693,
-      "grad_norm": 9.268046379089355,
+      "grad_norm": 13.99533748626709,
       "learning_rate": 7.952286282306165e-07,
-      "loss": 0.6617,
+      "loss": 1.4686,
       "step": 805
     },
     {
       "epoch": 0.7933070866141733,
-      "grad_norm": 12.68924331665039,
+      "grad_norm": 11.035348892211914,
       "learning_rate": 7.962226640159047e-07,
-      "loss": 1.1582,
+      "loss": 0.8565,
       "step": 806
     },
     {
       "epoch": 0.7942913385826772,
-      "grad_norm": 21.48848533630371,
+      "grad_norm": 14.682604789733887,
       "learning_rate": 7.97216699801193e-07,
-      "loss": 1.7469,
+      "loss": 1.1208,
       "step": 807
     },
     {
       "epoch": 0.7952755905511811,
-      "grad_norm": 10.589978218078613,
+      "grad_norm": 11.655988693237305,
       "learning_rate": 7.982107355864813e-07,
-      "loss": 1.0597,
+      "loss": 1.006,
       "step": 808
     },
     {
       "epoch": 0.7962598425196851,
-      "grad_norm": 9.825285911560059,
+      "grad_norm": 10.584676742553711,
       "learning_rate": 7.992047713717695e-07,
-      "loss": 0.4811,
+      "loss": 0.71,
       "step": 809
     },
     {
       "epoch": 0.797244094488189,
-      "grad_norm": 14.238350868225098,
+      "grad_norm": 11.906569480895996,
       "learning_rate": 8.001988071570577e-07,
-      "loss": 1.1675,
+      "loss": 1.0219,
       "step": 810
     },
     {
       "epoch": 0.7982283464566929,
-      "grad_norm": 11.2032470703125,
+      "grad_norm": 12.940180778503418,
       "learning_rate": 8.011928429423461e-07,
-      "loss": 0.9254,
+      "loss": 1.0836,
       "step": 811
     },
     {
       "epoch": 0.7992125984251969,
-      "grad_norm": 12.099225997924805,
+      "grad_norm": 14.408658027648926,
       "learning_rate": 8.021868787276343e-07,
-      "loss": 0.9448,
+      "loss": 1.1318,
       "step": 812
     },
     {
       "epoch": 0.8001968503937008,
-      "grad_norm": 14.134735107421875,
+      "grad_norm": 15.621986389160156,
       "learning_rate": 8.031809145129226e-07,
-      "loss": 1.2215,
+      "loss": 1.5411,
       "step": 813
     },
     {
       "epoch": 0.8011811023622047,
-      "grad_norm": 13.462716102600098,
+      "grad_norm": 11.737788200378418,
       "learning_rate": 8.041749502982109e-07,
-      "loss": 1.1021,
+      "loss": 1.0063,
       "step": 814
     },
     {
       "epoch": 0.8021653543307087,
-      "grad_norm": 15.155725479125977,
+      "grad_norm": 13.536646842956543,
       "learning_rate": 8.051689860834991e-07,
-      "loss": 1.6263,
+      "loss": 1.277,
       "step": 815
     },
     {
       "epoch": 0.8031496062992126,
-      "grad_norm": 9.317359924316406,
+      "grad_norm": 12.623762130737305,
       "learning_rate": 8.061630218687874e-07,
-      "loss": 0.6675,
+      "loss": 1.0547,
       "step": 816
     },
     {
       "epoch": 0.8041338582677166,
-      "grad_norm": 11.237923622131348,
+      "grad_norm": 10.43212890625,
       "learning_rate": 8.071570576540756e-07,
-      "loss": 0.8904,
+      "loss": 0.7084,
       "step": 817
     },
     {
       "epoch": 0.8051181102362205,
-      "grad_norm": 11.468343734741211,
+      "grad_norm": 12.9672269821167,
       "learning_rate": 8.08151093439364e-07,
-      "loss": 0.8919,
+      "loss": 1.2581,
       "step": 818
     },
     {
       "epoch": 0.8061023622047244,
-      "grad_norm": 10.934569358825684,
+      "grad_norm": 11.470852851867676,
       "learning_rate": 8.091451292246522e-07,
-      "loss": 0.6128,
+      "loss": 0.7857,
       "step": 819
     },
     {
       "epoch": 0.8070866141732284,
-      "grad_norm": 10.651111602783203,
+      "grad_norm": 10.311899185180664,
       "learning_rate": 8.101391650099404e-07,
-      "loss": 0.7364,
+      "loss": 0.7219,
       "step": 820
     },
     {
       "epoch": 0.8080708661417323,
-      "grad_norm": 28.13587760925293,
+      "grad_norm": 14.127314567565918,
       "learning_rate": 8.111332007952288e-07,
-      "loss": 1.8094,
+      "loss": 1.2658,
       "step": 821
     },
     {
       "epoch": 0.8090551181102362,
-      "grad_norm": 11.294393539428711,
+      "grad_norm": 17.850603103637695,
       "learning_rate": 8.12127236580517e-07,
-      "loss": 0.7423,
+      "loss": 1.4563,
       "step": 822
     },
     {
       "epoch": 0.8100393700787402,
-      "grad_norm": 8.074543952941895,
+      "grad_norm": 11.794126510620117,
       "learning_rate": 8.131212723658052e-07,
-      "loss": 0.4273,
+      "loss": 0.9098,
       "step": 823
     },
     {
       "epoch": 0.8110236220472441,
-      "grad_norm": 10.790502548217773,
+      "grad_norm": 11.173970222473145,
       "learning_rate": 8.141153081510935e-07,
-      "loss": 0.7915,
+      "loss": 0.8606,
       "step": 824
     },
     {
       "epoch": 0.812007874015748,
-      "grad_norm": 11.241899490356445,
+      "grad_norm": 13.826883316040039,
       "learning_rate": 8.151093439363818e-07,
-      "loss": 0.6667,
+      "loss": 1.1116,
       "step": 825
     },
     {
       "epoch": 0.812992125984252,
-      "grad_norm": 14.739508628845215,
+      "grad_norm": 10.755367279052734,
       "learning_rate": 8.161033797216701e-07,
-      "loss": 1.0842,
+      "loss": 0.952,
       "step": 826
     },
     {
       "epoch": 0.8139763779527559,
-      "grad_norm": 10.02929973602295,
+      "grad_norm": 11.222610473632812,
       "learning_rate": 8.170974155069583e-07,
-      "loss": 0.6472,
+      "loss": 0.6831,
       "step": 827
     },
     {
       "epoch": 0.8149606299212598,
-      "grad_norm": 11.892731666564941,
+      "grad_norm": 10.882960319519043,
       "learning_rate": 8.180914512922467e-07,
-      "loss": 0.6185,
+      "loss": 0.8267,
       "step": 828
     },
     {
       "epoch": 0.8159448818897638,
-      "grad_norm": 15.663006782531738,
+      "grad_norm": 12.082534790039062,
       "learning_rate": 8.19085487077535e-07,
-      "loss": 1.1912,
+      "loss": 0.934,
       "step": 829
     },
     {
       "epoch": 0.8169291338582677,
-      "grad_norm": 15.349569320678711,
+      "grad_norm": 16.402433395385742,
       "learning_rate": 8.200795228628232e-07,
-      "loss": 1.3546,
+      "loss": 1.7555,
       "step": 830
     },
     {
       "epoch": 0.8179133858267716,
-      "grad_norm": 11.670661926269531,
+      "grad_norm": 12.048799514770508,
       "learning_rate": 8.210735586481115e-07,
-      "loss": 0.8114,
+      "loss": 1.0109,
       "step": 831
     },
     {
       "epoch": 0.8188976377952756,
-      "grad_norm": 27.554731369018555,
+      "grad_norm": 15.013277053833008,
       "learning_rate": 8.220675944333998e-07,
-      "loss": 1.5822,
+      "loss": 1.4509,
       "step": 832
     },
     {
       "epoch": 0.8198818897637795,
-      "grad_norm": 12.875972747802734,
+      "grad_norm": 10.118528366088867,
       "learning_rate": 8.23061630218688e-07,
-      "loss": 0.7137,
+      "loss": 0.6548,
       "step": 833
     },
     {
       "epoch": 0.8208661417322834,
-      "grad_norm": 6.835382461547852,
+      "grad_norm": 10.921168327331543,
       "learning_rate": 8.240556660039763e-07,
-      "loss": 0.2886,
+      "loss": 0.7777,
       "step": 834
     },
     {
       "epoch": 0.8218503937007874,
-      "grad_norm": 15.828883171081543,
+      "grad_norm": 12.746010780334473,
       "learning_rate": 8.250497017892646e-07,
-      "loss": 1.2391,
+      "loss": 0.8929,
       "step": 835
     },
     {
       "epoch": 0.8228346456692913,
-      "grad_norm": 15.309415817260742,
+      "grad_norm": 15.578157424926758,
       "learning_rate": 8.260437375745529e-07,
-      "loss": 1.1536,
+      "loss": 1.282,
       "step": 836
     },
     {
       "epoch": 0.8238188976377953,
-      "grad_norm": 16.423490524291992,
+      "grad_norm": 11.15747356414795,
       "learning_rate": 8.270377733598411e-07,
-      "loss": 1.1422,
+      "loss": 0.6726,
       "step": 837
     },
     {
       "epoch": 0.8248031496062992,
-      "grad_norm": 35.645301818847656,
+      "grad_norm": 14.893630027770996,
       "learning_rate": 8.280318091451293e-07,
-      "loss": 2.7215,
+      "loss": 1.0643,
       "step": 838
     },
     {
       "epoch": 0.8257874015748031,
-      "grad_norm": 30.38851547241211,
+      "grad_norm": 11.360194206237793,
       "learning_rate": 8.290258449304177e-07,
-      "loss": 1.9721,
+      "loss": 1.01,
       "step": 839
     },
     {
       "epoch": 0.8267716535433071,
-      "grad_norm": 9.624545097351074,
+      "grad_norm": 12.738455772399902,
       "learning_rate": 8.300198807157059e-07,
-      "loss": 0.5596,
+      "loss": 1.0631,
       "step": 840
     },
     {
       "epoch": 0.827755905511811,
-      "grad_norm": 24.00203514099121,
+      "grad_norm": 13.322991371154785,
       "learning_rate": 8.310139165009941e-07,
-      "loss": 1.1538,
+      "loss": 1.092,
       "step": 841
     },
     {
       "epoch": 0.8287401574803149,
-      "grad_norm": 15.759881973266602,
+      "grad_norm": 11.534586906433105,
       "learning_rate": 8.320079522862825e-07,
-      "loss": 1.1463,
+      "loss": 1.0219,
       "step": 842
     },
     {
       "epoch": 0.8297244094488189,
-      "grad_norm": 15.030658721923828,
+      "grad_norm": 10.090067863464355,
       "learning_rate": 8.330019880715707e-07,
-      "loss": 1.2118,
+      "loss": 1.0437,
       "step": 843
     },
     {
       "epoch": 0.8307086614173228,
-      "grad_norm": 17.78954315185547,
+      "grad_norm": 11.919770240783691,
       "learning_rate": 8.33996023856859e-07,
-      "loss": 1.0761,
+      "loss": 0.8789,
       "step": 844
     },
     {
       "epoch": 0.8316929133858267,
-      "grad_norm": 28.6446533203125,
+      "grad_norm": 13.919114112854004,
       "learning_rate": 8.349900596421472e-07,
-      "loss": 2.2337,
+      "loss": 1.5409,
       "step": 845
     },
     {
       "epoch": 0.8326771653543307,
-      "grad_norm": 15.061113357543945,
+      "grad_norm": 12.6904296875,
       "learning_rate": 8.359840954274355e-07,
-      "loss": 1.4387,
+      "loss": 1.365,
       "step": 846
     },
     {
       "epoch": 0.8336614173228346,
-      "grad_norm": 16.369281768798828,
+      "grad_norm": 13.091736793518066,
       "learning_rate": 8.369781312127238e-07,
-      "loss": 1.3689,
+      "loss": 1.0567,
       "step": 847
     },
     {
       "epoch": 0.8346456692913385,
-      "grad_norm": 11.550053596496582,
+      "grad_norm": 9.207596778869629,
       "learning_rate": 8.37972166998012e-07,
-      "loss": 0.6953,
+      "loss": 0.59,
       "step": 848
     },
     {
       "epoch": 0.8356299212598425,
-      "grad_norm": 11.115499496459961,
+      "grad_norm": 9.879922866821289,
       "learning_rate": 8.389662027833004e-07,
-      "loss": 0.7809,
+      "loss": 0.839,
       "step": 849
     },
     {
       "epoch": 0.8366141732283464,
-      "grad_norm": 11.123430252075195,
+      "grad_norm": 9.073246955871582,
       "learning_rate": 8.399602385685886e-07,
-      "loss": 0.8823,
+      "loss": 0.727,
       "step": 850
     },
     {
       "epoch": 0.8375984251968503,
-      "grad_norm": 14.044934272766113,
+      "grad_norm": 10.825157165527344,
       "learning_rate": 8.409542743538768e-07,
-      "loss": 1.2892,
+      "loss": 0.8681,
       "step": 851
     },
     {
       "epoch": 0.8385826771653543,
-      "grad_norm": 11.716177940368652,
+      "grad_norm": 14.727147102355957,
       "learning_rate": 8.419483101391651e-07,
-      "loss": 1.0262,
+      "loss": 1.5186,
       "step": 852
     },
     {
       "epoch": 0.8395669291338582,
-      "grad_norm": 14.854663848876953,
+      "grad_norm": 11.10617446899414,
       "learning_rate": 8.429423459244534e-07,
-      "loss": 1.053,
+      "loss": 0.8923,
       "step": 853
     },
     {
       "epoch": 0.8405511811023622,
-      "grad_norm": 12.035706520080566,
+      "grad_norm": 10.533689498901367,
       "learning_rate": 8.439363817097416e-07,
-      "loss": 0.8695,
+      "loss": 0.8248,
       "step": 854
     },
     {
       "epoch": 0.8415354330708661,
-      "grad_norm": 11.932565689086914,
+      "grad_norm": 13.720536231994629,
       "learning_rate": 8.449304174950299e-07,
-      "loss": 0.9463,
+      "loss": 0.9884,
       "step": 855
     },
     {
       "epoch": 0.84251968503937,
-      "grad_norm": 30.260635375976562,
+      "grad_norm": 12.623149871826172,
       "learning_rate": 8.459244532803182e-07,
-      "loss": 2.3975,
+      "loss": 1.257,
       "step": 856
     },
     {
       "epoch": 0.843503937007874,
-      "grad_norm": 13.055898666381836,
+      "grad_norm": 11.030608177185059,
       "learning_rate": 8.469184890656066e-07,
-      "loss": 1.032,
+      "loss": 0.8841,
       "step": 857
     },
     {
       "epoch": 0.844488188976378,
-      "grad_norm": 29.741281509399414,
+      "grad_norm": 14.188507080078125,
       "learning_rate": 8.479125248508948e-07,
-      "loss": 2.1536,
+      "loss": 1.232,
       "step": 858
     },
     {
       "epoch": 0.8454724409448819,
-      "grad_norm": 24.180944442749023,
+      "grad_norm": 15.049591064453125,
       "learning_rate": 8.48906560636183e-07,
-      "loss": 1.5867,
+      "loss": 1.2214,
       "step": 859
     },
     {
       "epoch": 0.8464566929133859,
-      "grad_norm": 10.7355318069458,
+      "grad_norm": 12.906851768493652,
       "learning_rate": 8.499005964214714e-07,
-      "loss": 0.8153,
+      "loss": 1.3332,
       "step": 860
     },
     {
       "epoch": 0.8474409448818898,
-      "grad_norm": 10.552571296691895,
+      "grad_norm": 12.399016380310059,
       "learning_rate": 8.508946322067596e-07,
-      "loss": 0.7053,
+      "loss": 0.8272,
       "step": 861
     },
     {
       "epoch": 0.8484251968503937,
-      "grad_norm": 11.440624237060547,
+      "grad_norm": 12.69371223449707,
       "learning_rate": 8.518886679920479e-07,
-      "loss": 1.0604,
+      "loss": 1.0901,
       "step": 862
     },
     {
       "epoch": 0.8494094488188977,
-      "grad_norm": 24.281822204589844,
+      "grad_norm": 14.04008960723877,
       "learning_rate": 8.528827037773362e-07,
-      "loss": 2.0091,
+      "loss": 1.2892,
       "step": 863
     },
     {
       "epoch": 0.8503937007874016,
-      "grad_norm": 13.694790840148926,
+      "grad_norm": 11.152538299560547,
       "learning_rate": 8.538767395626244e-07,
-      "loss": 0.9021,
+      "loss": 0.8817,
       "step": 864
     },
     {
       "epoch": 0.8513779527559056,
-      "grad_norm": 23.159679412841797,
+      "grad_norm": 13.230735778808594,
       "learning_rate": 8.548707753479127e-07,
-      "loss": 1.6572,
+      "loss": 0.944,
       "step": 865
     },
     {
       "epoch": 0.8523622047244095,
-      "grad_norm": 9.769048690795898,
+      "grad_norm": 10.027914047241211,
       "learning_rate": 8.558648111332009e-07,
-      "loss": 0.7634,
+      "loss": 0.6212,
       "step": 866
     },
     {
       "epoch": 0.8533464566929134,
-      "grad_norm": 15.126567840576172,
+      "grad_norm": 14.65113639831543,
       "learning_rate": 8.568588469184892e-07,
-      "loss": 1.3602,
+      "loss": 1.4346,
       "step": 867
     },
     {
       "epoch": 0.8543307086614174,
-      "grad_norm": 18.267560958862305,
+      "grad_norm": 9.519758224487305,
       "learning_rate": 8.578528827037775e-07,
-      "loss": 1.8969,
+      "loss": 0.7204,
       "step": 868
     },
     {
       "epoch": 0.8553149606299213,
-      "grad_norm": 19.578649520874023,
+      "grad_norm": 17.111120223999023,
       "learning_rate": 8.588469184890657e-07,
-      "loss": 1.8319,
+      "loss": 1.4009,
       "step": 869
     },
     {
       "epoch": 0.8562992125984252,
-      "grad_norm": 19.867412567138672,
+      "grad_norm": 13.902044296264648,
       "learning_rate": 8.598409542743541e-07,
-      "loss": 1.8445,
+      "loss": 1.1522,
       "step": 870
     },
     {
       "epoch": 0.8572834645669292,
-      "grad_norm": 15.06701374053955,
+      "grad_norm": 11.630669593811035,
       "learning_rate": 8.608349900596423e-07,
-      "loss": 1.3896,
+      "loss": 0.9558,
       "step": 871
     },
     {
       "epoch": 0.8582677165354331,
-      "grad_norm": 9.85556697845459,
+      "grad_norm": 10.162226676940918,
       "learning_rate": 8.618290258449305e-07,
-      "loss": 0.5278,
+      "loss": 0.8871,
       "step": 872
     },
     {
       "epoch": 0.859251968503937,
-      "grad_norm": 10.843356132507324,
+      "grad_norm": 12.421677589416504,
       "learning_rate": 8.628230616302189e-07,
-      "loss": 0.9049,
+      "loss": 1.0025,
       "step": 873
     },
     {
       "epoch": 0.860236220472441,
-      "grad_norm": 11.80095386505127,
+      "grad_norm": 11.895072937011719,
       "learning_rate": 8.638170974155071e-07,
-      "loss": 0.8758,
+      "loss": 0.8617,
       "step": 874
     },
     {
       "epoch": 0.8612204724409449,
-      "grad_norm": 13.865303993225098,
+      "grad_norm": 12.787313461303711,
       "learning_rate": 8.648111332007954e-07,
-      "loss": 1.2579,
+      "loss": 1.1982,
       "step": 875
     },
     {
       "epoch": 0.8622047244094488,
-      "grad_norm": 10.507223129272461,
+      "grad_norm": 18.052860260009766,
       "learning_rate": 8.658051689860836e-07,
-      "loss": 0.7051,
+      "loss": 1.2484,
       "step": 876
     },
     {
       "epoch": 0.8631889763779528,
-      "grad_norm": 12.494022369384766,
+      "grad_norm": 15.959498405456543,
       "learning_rate": 8.667992047713719e-07,
-      "loss": 0.9826,
+      "loss": 1.0181,
       "step": 877
     },
     {
       "epoch": 0.8641732283464567,
-      "grad_norm": 13.588288307189941,
+      "grad_norm": 15.16512680053711,
       "learning_rate": 8.677932405566602e-07,
-      "loss": 0.7903,
+      "loss": 0.8535,
       "step": 878
     },
     {
       "epoch": 0.8651574803149606,
-      "grad_norm": 13.005169868469238,
+      "grad_norm": 15.01723575592041,
       "learning_rate": 8.687872763419484e-07,
-      "loss": 1.0113,
+      "loss": 1.144,
       "step": 879
     },
     {
       "epoch": 0.8661417322834646,
-      "grad_norm": 14.407569885253906,
+      "grad_norm": 13.688557624816895,
       "learning_rate": 8.697813121272366e-07,
-      "loss": 1.0594,
+      "loss": 1.1927,
       "step": 880
     },
     {
       "epoch": 0.8671259842519685,
-      "grad_norm": 12.332335472106934,
+      "grad_norm": 10.62550163269043,
       "learning_rate": 8.70775347912525e-07,
-      "loss": 0.9289,
+      "loss": 0.7245,
       "step": 881
     },
     {
       "epoch": 0.8681102362204725,
-      "grad_norm": 30.00078010559082,
+      "grad_norm": 11.776147842407227,
       "learning_rate": 8.717693836978132e-07,
-      "loss": 2.2593,
+      "loss": 1.0118,
       "step": 882
     },
     {
       "epoch": 0.8690944881889764,
-      "grad_norm": 12.043009757995605,
+      "grad_norm": 15.631773948669434,
       "learning_rate": 8.727634194831015e-07,
-      "loss": 0.8689,
+      "loss": 1.4399,
       "step": 883
     },
     {
       "epoch": 0.8700787401574803,
-      "grad_norm": 11.589649200439453,
+      "grad_norm": 13.422005653381348,
       "learning_rate": 8.737574552683898e-07,
-      "loss": 0.7923,
+      "loss": 1.0666,
       "step": 884
     },
     {
       "epoch": 0.8710629921259843,
-      "grad_norm": 15.759047508239746,
+      "grad_norm": 17.7437686920166,
       "learning_rate": 8.74751491053678e-07,
-      "loss": 0.9832,
+      "loss": 1.1947,
       "step": 885
     },
     {
       "epoch": 0.8720472440944882,
-      "grad_norm": 40.35478591918945,
+      "grad_norm": 11.687212944030762,
       "learning_rate": 8.757455268389663e-07,
-      "loss": 2.681,
+      "loss": 0.8908,
       "step": 886
     },
     {
       "epoch": 0.8730314960629921,
-      "grad_norm": 14.283058166503906,
+      "grad_norm": 15.046998977661133,
       "learning_rate": 8.767395626242545e-07,
-      "loss": 1.0878,
+      "loss": 1.394,
       "step": 887
     },
     {
       "epoch": 0.8740157480314961,
-      "grad_norm": 13.471712112426758,
+      "grad_norm": 11.96061897277832,
       "learning_rate": 8.77733598409543e-07,
-      "loss": 0.6764,
+      "loss": 0.802,
       "step": 888
     },
     {
       "epoch": 0.875,
-      "grad_norm": 24.483654022216797,
+      "grad_norm": 12.169413566589355,
       "learning_rate": 8.787276341948312e-07,
-      "loss": 1.7032,
+      "loss": 0.9879,
       "step": 889
     },
     {
       "epoch": 0.8759842519685039,
-      "grad_norm": 16.756513595581055,
+      "grad_norm": 11.521828651428223,
       "learning_rate": 8.797216699801194e-07,
-      "loss": 1.4734,
+      "loss": 0.9895,
       "step": 890
     },
     {
       "epoch": 0.8769685039370079,
-      "grad_norm": 12.61288833618164,
+      "grad_norm": 12.837910652160645,
       "learning_rate": 8.807157057654078e-07,
-      "loss": 0.8396,
+      "loss": 1.0927,
       "step": 891
     },
     {
       "epoch": 0.8779527559055118,
-      "grad_norm": 13.15617561340332,
+      "grad_norm": 14.709928512573242,
       "learning_rate": 8.81709741550696e-07,
-      "loss": 1.0006,
+      "loss": 1.4376,
       "step": 892
     },
     {
       "epoch": 0.8789370078740157,
-      "grad_norm": 14.844758987426758,
+      "grad_norm": 13.70672607421875,
       "learning_rate": 8.827037773359842e-07,
-      "loss": 1.2926,
+      "loss": 1.1509,
       "step": 893
     },
     {
       "epoch": 0.8799212598425197,
-      "grad_norm": 11.292862892150879,
+      "grad_norm": 12.354025840759277,
       "learning_rate": 8.836978131212726e-07,
-      "loss": 0.7681,
+      "loss": 0.7944,
       "step": 894
     },
     {
       "epoch": 0.8809055118110236,
-      "grad_norm": 11.68862247467041,
+      "grad_norm": 14.6171236038208,
       "learning_rate": 8.846918489065608e-07,
-      "loss": 0.8424,
+      "loss": 1.4574,
       "step": 895
     },
     {
       "epoch": 0.8818897637795275,
-      "grad_norm": 8.260144233703613,
+      "grad_norm": 12.5920991897583,
       "learning_rate": 8.856858846918491e-07,
-      "loss": 0.4961,
+      "loss": 1.2278,
       "step": 896
     },
     {
       "epoch": 0.8828740157480315,
-      "grad_norm": 21.9805965423584,
+      "grad_norm": 13.8173189163208,
       "learning_rate": 8.866799204771373e-07,
-      "loss": 1.8518,
+      "loss": 1.1974,
       "step": 897
     },
     {
       "epoch": 0.8838582677165354,
-      "grad_norm": 12.657113075256348,
+      "grad_norm": 12.641080856323242,
       "learning_rate": 8.876739562624256e-07,
-      "loss": 0.7808,
+      "loss": 1.0711,
       "step": 898
     },
     {
       "epoch": 0.8848425196850394,
-      "grad_norm": 10.799315452575684,
+      "grad_norm": 14.381937026977539,
       "learning_rate": 8.886679920477139e-07,
-      "loss": 0.7176,
+      "loss": 1.3942,
       "step": 899
     },
     {
       "epoch": 0.8858267716535433,
-      "grad_norm": 12.964784622192383,
+      "grad_norm": 13.973237037658691,
       "learning_rate": 8.896620278330021e-07,
-      "loss": 0.8946,
+      "loss": 1.1718,
       "step": 900
     },
     {
       "epoch": 0.8868110236220472,
-      "grad_norm": 23.136194229125977,
+      "grad_norm": 12.490571022033691,
       "learning_rate": 8.906560636182905e-07,
-      "loss": 1.5311,
+      "loss": 1.1274,
       "step": 901
     },
     {
       "epoch": 0.8877952755905512,
-      "grad_norm": 12.264001846313477,
+      "grad_norm": 10.961933135986328,
       "learning_rate": 8.916500994035787e-07,
-      "loss": 1.0758,
+      "loss": 0.9954,
       "step": 902
     },
     {
       "epoch": 0.8887795275590551,
-      "grad_norm": 11.262816429138184,
+      "grad_norm": 10.65230655670166,
       "learning_rate": 8.926441351888669e-07,
-      "loss": 0.843,
+      "loss": 1.0155,
       "step": 903
     },
     {
       "epoch": 0.889763779527559,
-      "grad_norm": 14.829652786254883,
+      "grad_norm": 14.470539093017578,
       "learning_rate": 8.936381709741552e-07,
-      "loss": 1.3121,
+      "loss": 1.4032,
       "step": 904
     },
     {
       "epoch": 0.890748031496063,
-      "grad_norm": 12.111909866333008,
+      "grad_norm": 15.179549217224121,
       "learning_rate": 8.946322067594435e-07,
-      "loss": 0.9709,
+      "loss": 1.4135,
       "step": 905
     },
     {
       "epoch": 0.8917322834645669,
-      "grad_norm": 13.451350212097168,
+      "grad_norm": 11.116039276123047,
       "learning_rate": 8.956262425447318e-07,
-      "loss": 0.9791,
+      "loss": 1.1893,
       "step": 906
     },
     {
       "epoch": 0.8927165354330708,
-      "grad_norm": 12.7373628616333,
+      "grad_norm": 10.903943061828613,
       "learning_rate": 8.9662027833002e-07,
-      "loss": 1.0457,
+      "loss": 0.7243,
       "step": 907
     },
     {
       "epoch": 0.8937007874015748,
-      "grad_norm": 12.571308135986328,
+      "grad_norm": 12.204026222229004,
       "learning_rate": 8.976143141153082e-07,
-      "loss": 0.8604,
+      "loss": 1.1674,
       "step": 908
     },
     {
       "epoch": 0.8946850393700787,
-      "grad_norm": 15.486084938049316,
+      "grad_norm": 12.262521743774414,
       "learning_rate": 8.986083499005966e-07,
-      "loss": 1.1543,
+      "loss": 1.3777,
       "step": 909
     },
     {
       "epoch": 0.8956692913385826,
-      "grad_norm": 12.579278945922852,
+      "grad_norm": 12.773459434509277,
       "learning_rate": 8.996023856858848e-07,
-      "loss": 0.642,
+      "loss": 1.0281,
       "step": 910
     },
     {
       "epoch": 0.8966535433070866,
-      "grad_norm": 20.628713607788086,
+      "grad_norm": 16.213043212890625,
       "learning_rate": 9.00596421471173e-07,
-      "loss": 1.0866,
+      "loss": 1.2869,
       "step": 911
     },
     {
       "epoch": 0.8976377952755905,
-      "grad_norm": 24.661611557006836,
+      "grad_norm": 10.814340591430664,
       "learning_rate": 9.015904572564614e-07,
-      "loss": 1.5848,
+      "loss": 1.0125,
       "step": 912
     },
     {
       "epoch": 0.8986220472440944,
-      "grad_norm": 31.02180290222168,
+      "grad_norm": 11.734081268310547,
       "learning_rate": 9.025844930417496e-07,
-      "loss": 1.7155,
+      "loss": 1.0047,
       "step": 913
     },
     {
       "epoch": 0.8996062992125984,
-      "grad_norm": 13.382023811340332,
+      "grad_norm": 10.440587997436523,
       "learning_rate": 9.035785288270379e-07,
-      "loss": 0.8018,
+      "loss": 0.8839,
       "step": 914
     },
     {
       "epoch": 0.9005905511811023,
-      "grad_norm": 12.55356216430664,
+      "grad_norm": 13.46262264251709,
       "learning_rate": 9.045725646123262e-07,
-      "loss": 0.879,
+      "loss": 1.4279,
       "step": 915
     },
     {
       "epoch": 0.9015748031496063,
-      "grad_norm": 12.50846004486084,
+      "grad_norm": 13.135570526123047,
       "learning_rate": 9.055666003976144e-07,
-      "loss": 0.7473,
+      "loss": 1.053,
       "step": 916
     },
     {
       "epoch": 0.9025590551181102,
-      "grad_norm": 8.537793159484863,
+      "grad_norm": 10.498846054077148,
       "learning_rate": 9.065606361829028e-07,
-      "loss": 0.4899,
+      "loss": 0.7841,
       "step": 917
     },
     {
       "epoch": 0.9035433070866141,
-      "grad_norm": 9.605969429016113,
+      "grad_norm": 8.559603691101074,
       "learning_rate": 9.07554671968191e-07,
-      "loss": 0.6282,
+      "loss": 0.7738,
       "step": 918
     },
     {
       "epoch": 0.9035433070866141,
-      "eval_Qnli-dev_cosine_accuracy": 0.662109375,
-      "eval_Qnli-dev_cosine_accuracy_threshold": 0.8222019672393799,
-      "eval_Qnli-dev_cosine_ap": 0.6690525867530521,
-      "eval_Qnli-dev_cosine_f1": 0.6626506024096385,
-      "eval_Qnli-dev_cosine_f1_threshold": 0.736522376537323,
-      "eval_Qnli-dev_cosine_precision": 0.514018691588785,
-      "eval_Qnli-dev_cosine_recall": 0.9322033898305084,
-      "eval_Qnli-dev_dot_accuracy": 0.662109375,
-      "eval_Qnli-dev_dot_accuracy_threshold": 630.4925537109375,
-      "eval_Qnli-dev_dot_ap": 0.6666029674148094,
-      "eval_Qnli-dev_dot_f1": 0.6635802469135803,
-      "eval_Qnli-dev_dot_f1_threshold": 569.1029052734375,
-      "eval_Qnli-dev_dot_precision": 0.5218446601941747,
-      "eval_Qnli-dev_dot_recall": 0.9110169491525424,
-      "eval_Qnli-dev_euclidean_accuracy": 0.666015625,
-      "eval_Qnli-dev_euclidean_accuracy_threshold": 16.523191452026367,
-      "eval_Qnli-dev_euclidean_ap": 0.6688456278122232,
-      "eval_Qnli-dev_euclidean_f1": 0.6626506024096385,
-      "eval_Qnli-dev_euclidean_f1_threshold": 20.06503677368164,
-      "eval_Qnli-dev_euclidean_precision": 0.514018691588785,
-      "eval_Qnli-dev_euclidean_recall": 0.9322033898305084,
-      "eval_Qnli-dev_manhattan_accuracy": 0.666015625,
-      "eval_Qnli-dev_manhattan_accuracy_threshold": 349.4272766113281,
-      "eval_Qnli-dev_manhattan_ap": 0.6815804028415055,
-      "eval_Qnli-dev_manhattan_f1": 0.6720257234726689,
-      "eval_Qnli-dev_manhattan_f1_threshold": 396.77142333984375,
-      "eval_Qnli-dev_manhattan_precision": 0.5414507772020726,
-      "eval_Qnli-dev_manhattan_recall": 0.885593220338983,
-      "eval_Qnli-dev_max_accuracy": 0.666015625,
-      "eval_Qnli-dev_max_accuracy_threshold": 630.4925537109375,
-      "eval_Qnli-dev_max_ap": 0.6815804028415055,
-      "eval_Qnli-dev_max_f1": 0.6720257234726689,
-      "eval_Qnli-dev_max_f1_threshold": 569.1029052734375,
-      "eval_Qnli-dev_max_precision": 0.5414507772020726,
-      "eval_Qnli-dev_max_recall": 0.9322033898305084,
-      "eval_allNLI-dev_cosine_accuracy": 0.685546875,
-      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9381226897239685,
-      "eval_allNLI-dev_cosine_ap": 0.49098585826236835,
-      "eval_allNLI-dev_cosine_f1": 0.5296610169491525,
-      "eval_allNLI-dev_cosine_f1_threshold": 0.8554738163948059,
-      "eval_allNLI-dev_cosine_precision": 0.4180602006688963,
-      "eval_allNLI-dev_cosine_recall": 0.7225433526011561,
-      "eval_allNLI-dev_dot_accuracy": 0.685546875,
-      "eval_allNLI-dev_dot_accuracy_threshold": 715.43310546875,
-      "eval_allNLI-dev_dot_ap": 0.49162275304477426,
-      "eval_allNLI-dev_dot_f1": 0.5311871227364185,
-      "eval_allNLI-dev_dot_f1_threshold": 646.8074951171875,
-      "eval_allNLI-dev_dot_precision": 0.4074074074074074,
-      "eval_allNLI-dev_dot_recall": 0.7630057803468208,
-      "eval_allNLI-dev_euclidean_accuracy": 0.685546875,
-      "eval_allNLI-dev_euclidean_accuracy_threshold": 9.73391342163086,
-      "eval_allNLI-dev_euclidean_ap": 0.4908290509025364,
-      "eval_allNLI-dev_euclidean_f1": 0.5296610169491525,
-      "eval_allNLI-dev_euclidean_f1_threshold": 14.872737884521484,
-      "eval_allNLI-dev_euclidean_precision": 0.4180602006688963,
-      "eval_allNLI-dev_euclidean_recall": 0.7225433526011561,
-      "eval_allNLI-dev_manhattan_accuracy": 0.6875,
-      "eval_allNLI-dev_manhattan_accuracy_threshold": 210.50613403320312,
-      "eval_allNLI-dev_manhattan_ap": 0.49318808387723423,
-      "eval_allNLI-dev_manhattan_f1": 0.5353319057815846,
-      "eval_allNLI-dev_manhattan_f1_threshold": 307.93359375,
-      "eval_allNLI-dev_manhattan_precision": 0.42517006802721086,
-      "eval_allNLI-dev_manhattan_recall": 0.7225433526011561,
-      "eval_allNLI-dev_max_accuracy": 0.6875,
-      "eval_allNLI-dev_max_accuracy_threshold": 715.43310546875,
-      "eval_allNLI-dev_max_ap": 0.49318808387723423,
-      "eval_allNLI-dev_max_f1": 0.5353319057815846,
-      "eval_allNLI-dev_max_f1_threshold": 646.8074951171875,
-      "eval_allNLI-dev_max_precision": 0.42517006802721086,
-      "eval_allNLI-dev_max_recall": 0.7630057803468208,
-      "eval_loss": 1.3715771436691284,
-      "eval_runtime": 54.2633,
-      "eval_samples_per_second": 30.665,
-      "eval_sequential_score": 0.6815804028415055,
-      "eval_steps_per_second": 0.129,
-      "eval_sts-test_pearson_cosine": 0.4591231950836093,
-      "eval_sts-test_pearson_dot": 0.4551866277132004,
-      "eval_sts-test_pearson_euclidean": 0.48026487148147445,
-      "eval_sts-test_pearson_manhattan": 0.49118985842833635,
-      "eval_sts-test_pearson_max": 0.49118985842833635,
-      "eval_sts-test_spearman_cosine": 0.4969829405703791,
-      "eval_sts-test_spearman_dot": 0.491899063022532,
-      "eval_sts-test_spearman_euclidean": 0.49736032795769275,
-      "eval_sts-test_spearman_manhattan": 0.5076723611201129,
-      "eval_sts-test_spearman_max": 0.5076723611201129,
+      "eval_Qnli-dev_cosine_accuracy": 0.6484375,
+      "eval_Qnli-dev_cosine_accuracy_threshold": 0.8556903004646301,
+      "eval_Qnli-dev_cosine_ap": 0.6712050759826647,
+      "eval_Qnli-dev_cosine_f1": 0.6836734693877551,
+      "eval_Qnli-dev_cosine_f1_threshold": 0.817956805229187,
+      "eval_Qnli-dev_cosine_precision": 0.5710227272727273,
+      "eval_Qnli-dev_cosine_recall": 0.8516949152542372,
+      "eval_Qnli-dev_dot_accuracy": 0.646484375,
+      "eval_Qnli-dev_dot_accuracy_threshold": 663.1381225585938,
+      "eval_Qnli-dev_dot_ap": 0.6701392631858906,
+      "eval_Qnli-dev_dot_f1": 0.6835871404399323,
+      "eval_Qnli-dev_dot_f1_threshold": 623.9208984375,
+      "eval_Qnli-dev_dot_precision": 0.5690140845070423,
+      "eval_Qnli-dev_dot_recall": 0.8559322033898306,
+      "eval_Qnli-dev_euclidean_accuracy": 0.650390625,
+      "eval_Qnli-dev_euclidean_accuracy_threshold": 14.574091911315918,
+      "eval_Qnli-dev_euclidean_ap": 0.6711308988804017,
+      "eval_Qnli-dev_euclidean_f1": 0.6844741235392321,
+      "eval_Qnli-dev_euclidean_f1_threshold": 16.858455657958984,
+      "eval_Qnli-dev_euclidean_precision": 0.5647382920110193,
+      "eval_Qnli-dev_euclidean_recall": 0.8686440677966102,
+      "eval_Qnli-dev_manhattan_accuracy": 0.65625,
+      "eval_Qnli-dev_manhattan_accuracy_threshold": 308.8890380859375,
+      "eval_Qnli-dev_manhattan_ap": 0.6758066941744509,
+      "eval_Qnli-dev_manhattan_f1": 0.6814580031695722,
+      "eval_Qnli-dev_manhattan_f1_threshold": 359.8348083496094,
+      "eval_Qnli-dev_manhattan_precision": 0.5443037974683544,
+      "eval_Qnli-dev_manhattan_recall": 0.9110169491525424,
+      "eval_Qnli-dev_max_accuracy": 0.65625,
+      "eval_Qnli-dev_max_accuracy_threshold": 663.1381225585938,
+      "eval_Qnli-dev_max_ap": 0.6758066941744509,
+      "eval_Qnli-dev_max_f1": 0.6844741235392321,
+      "eval_Qnli-dev_max_f1_threshold": 623.9208984375,
+      "eval_Qnli-dev_max_precision": 0.5710227272727273,
+      "eval_Qnli-dev_max_recall": 0.9110169491525424,
+      "eval_allNLI-dev_cosine_accuracy": 0.681640625,
+      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9492451548576355,
+      "eval_allNLI-dev_cosine_ap": 0.4860204244756696,
+      "eval_allNLI-dev_cosine_f1": 0.5408805031446541,
+      "eval_allNLI-dev_cosine_f1_threshold": 0.8793689012527466,
+      "eval_allNLI-dev_cosine_precision": 0.4243421052631579,
+      "eval_allNLI-dev_cosine_recall": 0.7456647398843931,
+      "eval_allNLI-dev_dot_accuracy": 0.681640625,
+      "eval_allNLI-dev_dot_accuracy_threshold": 719.29296875,
+      "eval_allNLI-dev_dot_ap": 0.4850643671480781,
+      "eval_allNLI-dev_dot_f1": 0.5408805031446541,
+      "eval_allNLI-dev_dot_f1_threshold": 671.677001953125,
+      "eval_allNLI-dev_dot_precision": 0.4243421052631579,
+      "eval_allNLI-dev_dot_recall": 0.7456647398843931,
+      "eval_allNLI-dev_euclidean_accuracy": 0.681640625,
+      "eval_allNLI-dev_euclidean_accuracy_threshold": 8.806386947631836,
+      "eval_allNLI-dev_euclidean_ap": 0.4857904504087363,
+      "eval_allNLI-dev_euclidean_f1": 0.5408805031446541,
+      "eval_allNLI-dev_euclidean_f1_threshold": 13.574195861816406,
+      "eval_allNLI-dev_euclidean_precision": 0.4243421052631579,
+      "eval_allNLI-dev_euclidean_recall": 0.7456647398843931,
+      "eval_allNLI-dev_manhattan_accuracy": 0.6796875,
+      "eval_allNLI-dev_manhattan_accuracy_threshold": 179.25799560546875,
+      "eval_allNLI-dev_manhattan_ap": 0.48263260956262277,
+      "eval_allNLI-dev_manhattan_f1": 0.5512820512820513,
+      "eval_allNLI-dev_manhattan_f1_threshold": 281.2195739746094,
+      "eval_allNLI-dev_manhattan_precision": 0.43728813559322033,
+      "eval_allNLI-dev_manhattan_recall": 0.7456647398843931,
+      "eval_allNLI-dev_max_accuracy": 0.681640625,
+      "eval_allNLI-dev_max_accuracy_threshold": 719.29296875,
+      "eval_allNLI-dev_max_ap": 0.4860204244756696,
+      "eval_allNLI-dev_max_f1": 0.5512820512820513,
+      "eval_allNLI-dev_max_f1_threshold": 671.677001953125,
+      "eval_allNLI-dev_max_precision": 0.43728813559322033,
+      "eval_allNLI-dev_max_recall": 0.7456647398843931,
+      "eval_loss": 1.2464040517807007,
+      "eval_runtime": 52.8966,
+      "eval_samples_per_second": 31.458,
+      "eval_sequential_score": 0.6758066941744509,
+      "eval_steps_per_second": 0.132,
+      "eval_sts-test_pearson_cosine": 0.49084594999467995,
+      "eval_sts-test_pearson_dot": 0.48706933209590814,
+      "eval_sts-test_pearson_euclidean": 0.5093765998387638,
+      "eval_sts-test_pearson_manhattan": 0.5202808585227008,
+      "eval_sts-test_pearson_max": 0.5202808585227008,
+      "eval_sts-test_spearman_cosine": 0.5248642742038014,
+      "eval_sts-test_spearman_dot": 0.5211632670331621,
+      "eval_sts-test_spearman_euclidean": 0.5250007655510451,
+      "eval_sts-test_spearman_manhattan": 0.5343063428447727,
+      "eval_sts-test_spearman_max": 0.5343063428447727,
       "step": 918
     },
     {
       "epoch": 0.9045275590551181,
-      "grad_norm": 15.066818237304688,
+      "grad_norm": 10.508577346801758,
       "learning_rate": 9.085487077534794e-07,
-      "loss": 1.2808,
+      "loss": 0.8998,
       "step": 919
     },
     {
       "epoch": 0.905511811023622,
-      "grad_norm": 15.124238014221191,
+      "grad_norm": 15.358297348022461,
       "learning_rate": 9.095427435387676e-07,
-      "loss": 1.2134,
+      "loss": 1.7733,
       "step": 920
     },
     {
       "epoch": 0.906496062992126,
-      "grad_norm": 10.836934089660645,
+      "grad_norm": 11.635780334472656,
       "learning_rate": 9.105367793240558e-07,
-      "loss": 0.6815,
+      "loss": 0.8032,
       "step": 921
     },
     {
       "epoch": 0.90748031496063,
-      "grad_norm": 12.821209907531738,
+      "grad_norm": 9.231951713562012,
       "learning_rate": 9.115308151093442e-07,
-      "loss": 1.0554,
+      "loss": 0.6311,
       "step": 922
     },
     {
       "epoch": 0.9084645669291339,
-      "grad_norm": 23.893203735351562,
+      "grad_norm": 10.531167030334473,
       "learning_rate": 9.125248508946324e-07,
-      "loss": 1.8122,
+      "loss": 0.9534,
       "step": 923
     },
     {
       "epoch": 0.9094488188976378,
-      "grad_norm": 25.48162841796875,
+      "grad_norm": 12.956085205078125,
       "learning_rate": 9.135188866799206e-07,
-      "loss": 1.041,
+      "loss": 1.2538,
       "step": 924
     },
     {
       "epoch": 0.9104330708661418,
-      "grad_norm": 10.63532829284668,
+      "grad_norm": 13.205366134643555,
       "learning_rate": 9.145129224652089e-07,
-      "loss": 0.4742,
+      "loss": 0.5411,
       "step": 925
     },
     {
       "epoch": 0.9114173228346457,
-      "grad_norm": 13.105849266052246,
+      "grad_norm": 11.612566947937012,
       "learning_rate": 9.155069582504972e-07,
-      "loss": 0.986,
+      "loss": 0.797,
       "step": 926
     },
     {
       "epoch": 0.9124015748031497,
-      "grad_norm": 16.00253677368164,
+      "grad_norm": 15.050639152526855,
       "learning_rate": 9.165009940357855e-07,
-      "loss": 1.1247,
+      "loss": 1.2115,
       "step": 927
     },
     {
       "epoch": 0.9133858267716536,
-      "grad_norm": 13.528996467590332,
+      "grad_norm": 11.717726707458496,
       "learning_rate": 9.174950298210737e-07,
-      "loss": 0.8918,
+      "loss": 0.8412,
       "step": 928
     },
     {
       "epoch": 0.9143700787401575,
-      "grad_norm": 12.672636985778809,
+      "grad_norm": 11.777231216430664,
       "learning_rate": 9.18489065606362e-07,
-      "loss": 0.9735,
+      "loss": 0.7325,
       "step": 929
     },
     {
       "epoch": 0.9153543307086615,
-      "grad_norm": 15.981063842773438,
+      "grad_norm": 22.675230026245117,
       "learning_rate": 9.194831013916503e-07,
-      "loss": 1.4467,
+      "loss": 1.6631,
       "step": 930
     },
     {
       "epoch": 0.9163385826771654,
-      "grad_norm": 10.165881156921387,
+      "grad_norm": 12.434297561645508,
       "learning_rate": 9.204771371769385e-07,
-      "loss": 0.6623,
+      "loss": 0.8573,
       "step": 931
     },
     {
       "epoch": 0.9173228346456693,
-      "grad_norm": 35.951602935791016,
+      "grad_norm": 14.405913352966309,
       "learning_rate": 9.214711729622268e-07,
-      "loss": 2.3602,
+      "loss": 0.8194,
       "step": 932
     },
     {
       "epoch": 0.9183070866141733,
-      "grad_norm": 37.820743560791016,
+      "grad_norm": 14.679107666015625,
       "learning_rate": 9.224652087475151e-07,
-      "loss": 2.5324,
+      "loss": 1.1627,
       "step": 933
     },
     {
       "epoch": 0.9192913385826772,
-      "grad_norm": 11.071194648742676,
+      "grad_norm": 18.158557891845703,
       "learning_rate": 9.234592445328033e-07,
-      "loss": 0.6558,
+      "loss": 1.3729,
       "step": 934
     },
     {
       "epoch": 0.9202755905511811,
-      "grad_norm": 14.736655235290527,
+      "grad_norm": 16.845603942871094,
       "learning_rate": 9.244532803180916e-07,
-      "loss": 0.856,
+      "loss": 1.0679,
       "step": 935
     },
     {
       "epoch": 0.9212598425196851,
-      "grad_norm": 23.33149528503418,
+      "grad_norm": 14.517767906188965,
       "learning_rate": 9.254473161033799e-07,
-      "loss": 1.8413,
+      "loss": 1.2123,
       "step": 936
     },
     {
       "epoch": 0.922244094488189,
-      "grad_norm": 29.5730037689209,
+      "grad_norm": 13.135607719421387,
       "learning_rate": 9.264413518886681e-07,
-      "loss": 2.0715,
+      "loss": 0.9217,
       "step": 937
     },
     {
       "epoch": 0.9232283464566929,
-      "grad_norm": 17.301631927490234,
+      "grad_norm": 11.955747604370117,
       "learning_rate": 9.274353876739564e-07,
-      "loss": 0.9652,
+      "loss": 0.6959,
       "step": 938
     },
     {
       "epoch": 0.9242125984251969,
-      "grad_norm": 14.46304702758789,
+      "grad_norm": 12.874065399169922,
       "learning_rate": 9.284294234592446e-07,
-      "loss": 0.7404,
+      "loss": 0.9306,
       "step": 939
     },
     {
       "epoch": 0.9251968503937008,
-      "grad_norm": 15.705856323242188,
+      "grad_norm": 15.19837760925293,
       "learning_rate": 9.29423459244533e-07,
-      "loss": 1.6298,
+      "loss": 1.2815,
       "step": 940
     },
     {
       "epoch": 0.9261811023622047,
-      "grad_norm": 10.920563697814941,
+      "grad_norm": 11.53364372253418,
       "learning_rate": 9.304174950298212e-07,
-      "loss": 0.7701,
+      "loss": 1.1188,
       "step": 941
     },
     {
       "epoch": 0.9271653543307087,
-      "grad_norm": 11.190475463867188,
+      "grad_norm": 10.0062837600708,
       "learning_rate": 9.314115308151094e-07,
-      "loss": 0.967,
+      "loss": 0.8454,
       "step": 942
     },
     {
       "epoch": 0.9281496062992126,
-      "grad_norm": 10.520164489746094,
+      "grad_norm": 10.954655647277832,
       "learning_rate": 9.324055666003978e-07,
-      "loss": 0.934,
+      "loss": 0.7369,
       "step": 943
     },
     {
       "epoch": 0.9291338582677166,
-      "grad_norm": 26.160831451416016,
+      "grad_norm": 12.967487335205078,
       "learning_rate": 9.33399602385686e-07,
-      "loss": 1.7068,
+      "loss": 1.2244,
       "step": 944
     },
     {
       "epoch": 0.9301181102362205,
-      "grad_norm": 13.173468589782715,
+      "grad_norm": 12.993568420410156,
       "learning_rate": 9.343936381709743e-07,
-      "loss": 1.1774,
+      "loss": 1.1402,
       "step": 945
     },
     {
       "epoch": 0.9311023622047244,
-      "grad_norm": 8.563469886779785,
+      "grad_norm": 7.9335832595825195,
       "learning_rate": 9.353876739562625e-07,
-      "loss": 0.6848,
+      "loss": 0.4213,
       "step": 946
     },
     {
       "epoch": 0.9320866141732284,
-      "grad_norm": 9.22998046875,
+      "grad_norm": 14.420933723449707,
       "learning_rate": 9.363817097415509e-07,
-      "loss": 0.6364,
+      "loss": 1.0047,
       "step": 947
     },
     {
       "epoch": 0.9330708661417323,
-      "grad_norm": 7.827816963195801,
+      "grad_norm": 13.709823608398438,
       "learning_rate": 9.373757455268392e-07,
-      "loss": 0.6477,
+      "loss": 1.1353,
       "step": 948
     },
     {
       "epoch": 0.9340551181102362,
-      "grad_norm": 27.504056930541992,
+      "grad_norm": 16.349287033081055,
       "learning_rate": 9.383697813121274e-07,
-      "loss": 1.5631,
+      "loss": 1.2862,
       "step": 949
     },
     {
       "epoch": 0.9350393700787402,
-      "grad_norm": 9.704866409301758,
+      "grad_norm": 11.642127990722656,
       "learning_rate": 9.393638170974157e-07,
-      "loss": 0.7204,
+      "loss": 0.6358,
       "step": 950
     },
     {
       "epoch": 0.9360236220472441,
-      "grad_norm": 9.747822761535645,
+      "grad_norm": 11.004044532775879,
       "learning_rate": 9.40357852882704e-07,
-      "loss": 0.7169,
+      "loss": 0.676,
       "step": 951
     },
     {
       "epoch": 0.937007874015748,
-      "grad_norm": 10.555800437927246,
+      "grad_norm": 13.824701309204102,
       "learning_rate": 9.413518886679922e-07,
-      "loss": 0.7335,
+      "loss": 1.1628,
       "step": 952
     },
     {
       "epoch": 0.937992125984252,
-      "grad_norm": 7.41068172454834,
+      "grad_norm": 12.448197364807129,
       "learning_rate": 9.423459244532805e-07,
-      "loss": 0.5237,
+      "loss": 0.861,
       "step": 953
     },
     {
       "epoch": 0.9389763779527559,
-      "grad_norm": 11.667238235473633,
+      "grad_norm": 11.722825050354004,
       "learning_rate": 9.433399602385688e-07,
-      "loss": 0.6509,
+      "loss": 0.7121,
       "step": 954
     },
     {
       "epoch": 0.9399606299212598,
-      "grad_norm": 13.451628684997559,
+      "grad_norm": 11.611307144165039,
       "learning_rate": 9.44333996023857e-07,
-      "loss": 0.8209,
+      "loss": 0.6625,
       "step": 955
     },
     {
       "epoch": 0.9409448818897638,
-      "grad_norm": 9.289586067199707,
+      "grad_norm": 11.159773826599121,
       "learning_rate": 9.453280318091453e-07,
-      "loss": 0.509,
+      "loss": 0.9343,
       "step": 956
     },
     {
       "epoch": 0.9419291338582677,
-      "grad_norm": 12.067706108093262,
+      "grad_norm": 10.660136222839355,
       "learning_rate": 9.463220675944336e-07,
-      "loss": 0.8048,
+      "loss": 0.8103,
       "step": 957
     },
     {
       "epoch": 0.9429133858267716,
-      "grad_norm": 10.956072807312012,
+      "grad_norm": 11.581384658813477,
       "learning_rate": 9.473161033797219e-07,
-      "loss": 0.663,
+      "loss": 1.047,
       "step": 958
     },
     {
       "epoch": 0.9438976377952756,
-      "grad_norm": 10.91523551940918,
+      "grad_norm": 10.32679557800293,
       "learning_rate": 9.483101391650101e-07,
-      "loss": 0.714,
+      "loss": 0.6455,
       "step": 959
     },
     {
       "epoch": 0.9448818897637795,
-      "grad_norm": 29.80571174621582,
+      "grad_norm": 12.044934272766113,
       "learning_rate": 9.493041749502983e-07,
-      "loss": 2.1961,
+      "loss": 1.0978,
       "step": 960
     },
     {
       "epoch": 0.9458661417322834,
-      "grad_norm": 12.550529479980469,
+      "grad_norm": 11.067245483398438,
       "learning_rate": 9.502982107355867e-07,
-      "loss": 0.8293,
+      "loss": 0.9522,
       "step": 961
     },
     {
       "epoch": 0.9468503937007874,
-      "grad_norm": 22.91265296936035,
+      "grad_norm": 11.00229549407959,
       "learning_rate": 9.512922465208749e-07,
-      "loss": 1.4108,
+      "loss": 0.9151,
       "step": 962
     },
     {
       "epoch": 0.9478346456692913,
-      "grad_norm": 15.225966453552246,
+      "grad_norm": 13.090099334716797,
       "learning_rate": 9.522862823061631e-07,
-      "loss": 1.2291,
+      "loss": 1.2515,
       "step": 963
     },
     {
       "epoch": 0.9488188976377953,
-      "grad_norm": 26.223369598388672,
+      "grad_norm": 13.134993553161621,
       "learning_rate": 9.532803180914515e-07,
-      "loss": 1.7917,
+      "loss": 1.0809,
       "step": 964
     },
     {
       "epoch": 0.9498031496062992,
-      "grad_norm": 25.121417999267578,
+      "grad_norm": 10.051626205444336,
       "learning_rate": 9.542743538767396e-07,
-      "loss": 1.5266,
+      "loss": 0.6971,
       "step": 965
     },
     {
       "epoch": 0.9507874015748031,
-      "grad_norm": 17.08681869506836,
+      "grad_norm": 15.127157211303711,
       "learning_rate": 9.55268389662028e-07,
-      "loss": 1.4995,
+      "loss": 1.4086,
       "step": 966
     },
     {
       "epoch": 0.9517716535433071,
-      "grad_norm": 14.251176834106445,
+      "grad_norm": 13.528825759887695,
       "learning_rate": 9.562624254473162e-07,
-      "loss": 1.1383,
+      "loss": 1.3102,
       "step": 967
     },
     {
       "epoch": 0.952755905511811,
-      "grad_norm": 9.796082496643066,
+      "grad_norm": 15.755900382995605,
       "learning_rate": 9.572564612326044e-07,
-      "loss": 0.7347,
+      "loss": 1.5008,
       "step": 968
     },
     {
       "epoch": 0.9537401574803149,
-      "grad_norm": 16.32306480407715,
+      "grad_norm": 11.590021133422852,
       "learning_rate": 9.582504970178928e-07,
-      "loss": 1.2644,
+      "loss": 0.9449,
       "step": 969
     },
     {
       "epoch": 0.9547244094488189,
-      "grad_norm": 13.050923347473145,
+      "grad_norm": 12.689349174499512,
       "learning_rate": 9.59244532803181e-07,
-      "loss": 0.8973,
+      "loss": 1.0705,
       "step": 970
     },
     {
       "epoch": 0.9557086614173228,
-      "grad_norm": 11.90676498413086,
+      "grad_norm": 12.062538146972656,
       "learning_rate": 9.602385685884694e-07,
-      "loss": 0.8478,
+      "loss": 0.7788,
       "step": 971
     },
     {
       "epoch": 0.9566929133858267,
-      "grad_norm": 9.020946502685547,
+      "grad_norm": 11.848228454589844,
       "learning_rate": 9.612326043737576e-07,
-      "loss": 0.5924,
+      "loss": 1.0007,
       "step": 972
     },
     {
       "epoch": 0.9576771653543307,
-      "grad_norm": 12.121260643005371,
+      "grad_norm": 12.216765403747559,
       "learning_rate": 9.622266401590458e-07,
-      "loss": 1.0661,
+      "loss": 0.9345,
       "step": 973
     },
     {
       "epoch": 0.9586614173228346,
-      "grad_norm": 13.3746337890625,
+      "grad_norm": 13.60435962677002,
       "learning_rate": 9.632206759443342e-07,
-      "loss": 1.3449,
+      "loss": 1.4525,
       "step": 974
     },
     {
       "epoch": 0.9596456692913385,
-      "grad_norm": 11.143105506896973,
+      "grad_norm": 11.18852710723877,
       "learning_rate": 9.642147117296224e-07,
-      "loss": 0.8305,
+      "loss": 0.5892,
       "step": 975
     },
     {
       "epoch": 0.9606299212598425,
-      "grad_norm": 14.122886657714844,
+      "grad_norm": 12.86732006072998,
       "learning_rate": 9.652087475149109e-07,
-      "loss": 1.3399,
+      "loss": 0.927,
       "step": 976
     },
     {
       "epoch": 0.9616141732283464,
-      "grad_norm": 26.04822540283203,
+      "grad_norm": 11.5518798828125,
       "learning_rate": 9.66202783300199e-07,
-      "loss": 2.1596,
+      "loss": 0.7473,
       "step": 977
     },
     {
       "epoch": 0.9625984251968503,
-      "grad_norm": 11.19150447845459,
+      "grad_norm": 10.896742820739746,
       "learning_rate": 9.671968190854873e-07,
-      "loss": 0.6784,
+      "loss": 0.8173,
       "step": 978
     },
     {
       "epoch": 0.9635826771653543,
-      "grad_norm": 18.581256866455078,
+      "grad_norm": 9.958429336547852,
       "learning_rate": 9.681908548707755e-07,
-      "loss": 1.3365,
+      "loss": 0.6888,
       "step": 979
     },
     {
       "epoch": 0.9645669291338582,
-      "grad_norm": 12.220732688903809,
+      "grad_norm": 11.125531196594238,
       "learning_rate": 9.691848906560639e-07,
-      "loss": 0.8564,
+      "loss": 0.8059,
       "step": 980
     },
     {
       "epoch": 0.9655511811023622,
-      "grad_norm": 12.121989250183105,
+      "grad_norm": 10.909327507019043,
       "learning_rate": 9.70178926441352e-07,
-      "loss": 0.9305,
+      "loss": 0.9087,
       "step": 981
     },
     {
       "epoch": 0.9665354330708661,
-      "grad_norm": 12.424619674682617,
+      "grad_norm": 16.282583236694336,
       "learning_rate": 9.711729622266403e-07,
-      "loss": 1.0597,
+      "loss": 1.5197,
       "step": 982
     },
     {
       "epoch": 0.96751968503937,
-      "grad_norm": 25.869260787963867,
+      "grad_norm": 11.268091201782227,
       "learning_rate": 9.721669980119285e-07,
-      "loss": 2.0984,
+      "loss": 0.6625,
       "step": 983
     },
     {
       "epoch": 0.968503937007874,
-      "grad_norm": 12.994894027709961,
+      "grad_norm": 12.072832107543945,
       "learning_rate": 9.731610337972169e-07,
-      "loss": 1.2005,
+      "loss": 1.0372,
       "step": 984
     },
     {
       "epoch": 0.969488188976378,
-      "grad_norm": 11.917289733886719,
+      "grad_norm": 13.086760520935059,
       "learning_rate": 9.74155069582505e-07,
-      "loss": 0.9083,
+      "loss": 0.91,
       "step": 985
     },
     {
       "epoch": 0.9704724409448819,
-      "grad_norm": 24.574920654296875,
+      "grad_norm": 11.189858436584473,
       "learning_rate": 9.751491053677933e-07,
-      "loss": 1.8997,
+      "loss": 0.7797,
       "step": 986
     },
     {
       "epoch": 0.9714566929133859,
-      "grad_norm": 9.57914924621582,
+      "grad_norm": 13.82830810546875,
       "learning_rate": 9.761431411530817e-07,
-      "loss": 0.7606,
+      "loss": 0.838,
       "step": 987
     },
     {
       "epoch": 0.9724409448818898,
-      "grad_norm": 11.350131034851074,
+      "grad_norm": 14.06183910369873,
       "learning_rate": 9.771371769383699e-07,
-      "loss": 0.6725,
+      "loss": 0.9568,
       "step": 988
     },
     {
       "epoch": 0.9734251968503937,
-      "grad_norm": 11.479193687438965,
+      "grad_norm": 18.976341247558594,
       "learning_rate": 9.781312127236583e-07,
-      "loss": 0.8839,
+      "loss": 1.4704,
       "step": 989
     },
     {
       "epoch": 0.9744094488188977,
-      "grad_norm": 8.321087837219238,
+      "grad_norm": 12.61073112487793,
       "learning_rate": 9.791252485089465e-07,
-      "loss": 0.4199,
+      "loss": 0.8209,
       "step": 990
     },
     {
       "epoch": 0.9753937007874016,
-      "grad_norm": 10.547780990600586,
+      "grad_norm": 14.609846115112305,
       "learning_rate": 9.801192842942347e-07,
-      "loss": 0.6008,
+      "loss": 0.86,
       "step": 991
     },
     {
       "epoch": 0.9763779527559056,
-      "grad_norm": 6.902090549468994,
+      "grad_norm": 11.316509246826172,
       "learning_rate": 9.811133200795231e-07,
-      "loss": 0.3017,
+      "loss": 0.7749,
       "step": 992
     },
     {
       "epoch": 0.9773622047244095,
-      "grad_norm": 9.781502723693848,
+      "grad_norm": 11.958404541015625,
       "learning_rate": 9.821073558648113e-07,
-      "loss": 0.5473,
+      "loss": 0.7225,
       "step": 993
     },
     {
       "epoch": 0.9783464566929134,
-      "grad_norm": 12.86219310760498,
+      "grad_norm": 14.443065643310547,
       "learning_rate": 9.831013916500995e-07,
-      "loss": 0.854,
+      "loss": 0.8488,
       "step": 994
     },
     {
       "epoch": 0.9793307086614174,
-      "grad_norm": 12.871740341186523,
+      "grad_norm": 19.713075637817383,
       "learning_rate": 9.84095427435388e-07,
-      "loss": 0.961,
+      "loss": 1.4838,
       "step": 995
     },
     {
       "epoch": 0.9803149606299213,
-      "grad_norm": 12.065014839172363,
+      "grad_norm": 15.081480979919434,
       "learning_rate": 9.850894632206761e-07,
-      "loss": 0.8797,
+      "loss": 1.3282,
       "step": 996
     },
     {
       "epoch": 0.9812992125984252,
-      "grad_norm": 15.222761154174805,
+      "grad_norm": 14.910409927368164,
       "learning_rate": 9.860834990059645e-07,
-      "loss": 0.9166,
+      "loss": 0.9905,
       "step": 997
     },
     {
       "epoch": 0.9822834645669292,
-      "grad_norm": 13.594207763671875,
+      "grad_norm": 21.365158081054688,
       "learning_rate": 9.870775347912527e-07,
-      "loss": 0.8646,
+      "loss": 1.7902,
       "step": 998
     },
     {
       "epoch": 0.9832677165354331,
-      "grad_norm": 9.33657455444336,
+      "grad_norm": 14.85123062133789,
       "learning_rate": 9.88071570576541e-07,
-      "loss": 0.4182,
+      "loss": 0.7411,
       "step": 999
     },
     {
       "epoch": 0.984251968503937,
-      "grad_norm": 13.540481567382812,
+      "grad_norm": 12.506738662719727,
       "learning_rate": 9.890656063618291e-07,
-      "loss": 1.1083,
+      "loss": 1.2985,
       "step": 1000
     },
     {
       "epoch": 0.985236220472441,
-      "grad_norm": 27.116676330566406,
+      "grad_norm": 12.294382095336914,
       "learning_rate": 9.900596421471175e-07,
-      "loss": 1.6306,
+      "loss": 0.9858,
       "step": 1001
     },
     {
       "epoch": 0.9862204724409449,
-      "grad_norm": 40.88335037231445,
+      "grad_norm": 10.234291076660156,
       "learning_rate": 9.910536779324057e-07,
-      "loss": 3.1844,
+      "loss": 0.8951,
       "step": 1002
     },
     {
       "epoch": 0.9872047244094488,
-      "grad_norm": 11.378805160522461,
+      "grad_norm": 8.601326942443848,
       "learning_rate": 9.92047713717694e-07,
-      "loss": 0.7701,
+      "loss": 0.5526,
       "step": 1003
     },
     {
       "epoch": 0.9881889763779528,
-      "grad_norm": 13.258501052856445,
+      "grad_norm": 15.147130966186523,
       "learning_rate": 9.930417495029821e-07,
-      "loss": 0.8628,
+      "loss": 1.466,
       "step": 1004
     },
     {
       "epoch": 0.9891732283464567,
-      "grad_norm": 11.201334953308105,
+      "grad_norm": 8.791632652282715,
       "learning_rate": 9.940357852882705e-07,
-      "loss": 0.6919,
+      "loss": 0.5767,
       "step": 1005
     },
     {
       "epoch": 0.9901574803149606,
-      "grad_norm": 10.30268669128418,
+      "grad_norm": 13.056974411010742,
       "learning_rate": 9.950298210735587e-07,
-      "loss": 0.5191,
+      "loss": 0.8946,
       "step": 1006
     },
     {
       "epoch": 0.9911417322834646,
-      "grad_norm": 16.8773136138916,
+      "grad_norm": 14.539746284484863,
       "learning_rate": 9.96023856858847e-07,
-      "loss": 0.8675,
+      "loss": 1.0825,
       "step": 1007
     },
     {
       "epoch": 0.9921259842519685,
-      "grad_norm": 21.878103256225586,
+      "grad_norm": 12.327513694763184,
       "learning_rate": 9.970178926441354e-07,
-      "loss": 1.4837,
+      "loss": 1.0152,
       "step": 1008
     },
     {
       "epoch": 0.9931102362204725,
-      "grad_norm": 25.748367309570312,
+      "grad_norm": 10.933964729309082,
       "learning_rate": 9.980119284294236e-07,
-      "loss": 1.5177,
+      "loss": 0.7586,
       "step": 1009
     },
     {
       "epoch": 0.9940944881889764,
-      "grad_norm": 16.585498809814453,
+      "grad_norm": 12.646042823791504,
       "learning_rate": 9.99005964214712e-07,
-      "loss": 1.0422,
+      "loss": 1.3353,
       "step": 1010
     },
     {
       "epoch": 0.9950787401574803,
-      "grad_norm": 20.055673599243164,
+      "grad_norm": 11.157730102539062,
       "learning_rate": 1.0000000000000002e-06,
-      "loss": 1.4038,
+      "loss": 0.9555,
       "step": 1011
     },
     {
       "epoch": 0.9960629921259843,
-      "grad_norm": 10.144420623779297,
+      "grad_norm": 13.540596961975098,
       "learning_rate": 9.999994102361802e-07,
-      "loss": 0.6402,
+      "loss": 1.0585,
       "step": 1012
     },
     {
       "epoch": 0.9970472440944882,
-      "grad_norm": 17.399948120117188,
+      "grad_norm": 9.059813499450684,
       "learning_rate": 9.999976409461164e-07,
-      "loss": 1.0376,
+      "loss": 0.6774,
       "step": 1013
     },
     {
       "epoch": 0.9980314960629921,
-      "grad_norm": 16.318910598754883,
+      "grad_norm": 14.586808204650879,
       "learning_rate": 9.999946921339963e-07,
-      "loss": 1.5319,
+      "loss": 1.5326,
       "step": 1014
     },
     {
       "epoch": 0.9990157480314961,
-      "grad_norm": 12.954554557800293,
+      "grad_norm": 13.352030754089355,
       "learning_rate": 9.999905638068e-07,
-      "loss": 1.0956,
+      "loss": 1.0034,
       "step": 1015
     },
     {
       "epoch": 1.0,
-      "grad_norm": 14.94405460357666,
+      "grad_norm": 17.30084228515625,
       "learning_rate": 9.999852559742985e-07,
-      "loss": 0.6806,
+      "loss": 0.8925,
       "step": 1016
     },
     {
       "epoch": 1.000984251968504,
-      "grad_norm": 0.0,
+      "grad_norm": 13.423869132995605,
       "learning_rate": 9.999787686490555e-07,
-      "loss": 0.0,
+      "loss": 1.1286,
       "step": 1017
     },
     {
-      "epoch": 1.000984251968504,
-      "grad_norm": 13.080833435058594,
+      "epoch": 1.0019685039370079,
+      "grad_norm": 13.345012664794922,
       "learning_rate": 9.99971101846426e-07,
-      "loss": 1.0824,
+      "loss": 0.9836,
       "step": 1018
     },
     {
-      "epoch": 1.0019685039370079,
-      "grad_norm": 15.298651695251465,
+      "epoch": 1.0029527559055118,
+      "grad_norm": 10.137012481689453,
       "learning_rate": 9.999622555845573e-07,
-      "loss": 1.0969,
+      "loss": 0.6727,
       "step": 1019
     },
     {
-      "epoch": 1.0029527559055118,
-      "grad_norm": 11.248276710510254,
+      "epoch": 1.0039370078740157,
+      "grad_norm": 12.473076820373535,
       "learning_rate": 9.999522298843875e-07,
-      "loss": 0.8334,
+      "loss": 0.8101,
       "step": 1020
     },
     {
-      "epoch": 1.0039370078740157,
-      "grad_norm": 10.074872016906738,
+      "epoch": 1.0049212598425197,
+      "grad_norm": 12.631579399108887,
       "learning_rate": 9.99941024769647e-07,
-      "loss": 0.7022,
+      "loss": 0.7747,
       "step": 1021
     },
     {
-      "epoch": 1.0049212598425197,
-      "grad_norm": 8.650634765625,
+      "epoch": 1.0059055118110236,
+      "grad_norm": 12.579475402832031,
       "learning_rate": 9.999286402668578e-07,
-      "loss": 0.502,
+      "loss": 0.6931,
       "step": 1022
     },
     {
-      "epoch": 1.0059055118110236,
-      "grad_norm": 8.788339614868164,
+      "epoch": 1.0068897637795275,
+      "grad_norm": 17.56043815612793,
       "learning_rate": 9.999150764053334e-07,
-      "loss": 0.5146,
+      "loss": 1.2261,
       "step": 1023
     },
     {
-      "epoch": 1.0068897637795275,
-      "grad_norm": 12.089606285095215,
+      "epoch": 1.0078740157480315,
+      "grad_norm": 13.309534072875977,
       "learning_rate": 9.999003332171784e-07,
-      "loss": 0.9768,
+      "loss": 0.7315,
       "step": 1024
     },
     {
-      "epoch": 1.0078740157480315,
-      "grad_norm": 11.496919631958008,
+      "epoch": 1.0088582677165354,
+      "grad_norm": 12.705358505249023,
       "learning_rate": 9.998844107372894e-07,
-      "loss": 0.7295,
+      "loss": 0.61,
       "step": 1025
     },
     {
-      "epoch": 1.0088582677165354,
-      "grad_norm": 51.444732666015625,
+      "epoch": 1.0098425196850394,
+      "grad_norm": 10.458514213562012,
       "learning_rate": 9.99867309003354e-07,
-      "loss": 2.0448,
+      "loss": 0.5402,
       "step": 1026
     },
     {
-      "epoch": 1.0098425196850394,
-      "grad_norm": 11.470032691955566,
+      "epoch": 1.0108267716535433,
+      "grad_norm": 17.223058700561523,
       "learning_rate": 9.998490280558511e-07,
-      "loss": 0.7609,
+      "loss": 1.3353,
       "step": 1027
     },
     {
-      "epoch": 1.0108267716535433,
-      "grad_norm": 14.040739059448242,
+      "epoch": 1.0118110236220472,
+      "grad_norm": 15.71506118774414,
       "learning_rate": 9.998295679380506e-07,
-      "loss": 1.3264,
+      "loss": 0.9483,
       "step": 1028
     },
     {
-      "epoch": 1.0118110236220472,
-      "grad_norm": 16.276897430419922,
+      "epoch": 1.0127952755905512,
+      "grad_norm": 16.565704345703125,
       "learning_rate": 9.998089286960135e-07,
-      "loss": 1.1084,
+      "loss": 0.9226,
       "step": 1029
     },
     {
-      "epoch": 1.0127952755905512,
-      "grad_norm": 13.747846603393555,
+      "epoch": 1.013779527559055,
+      "grad_norm": 8.595297813415527,
       "learning_rate": 9.99787110378592e-07,
-      "loss": 1.0857,
+      "loss": 0.6536,
       "step": 1030
     },
     {
-      "epoch": 1.013779527559055,
-      "grad_norm": 34.06081008911133,
+      "epoch": 1.014763779527559,
+      "grad_norm": 13.655412673950195,
       "learning_rate": 9.997641130374284e-07,
-      "loss": 1.7838,
+      "loss": 0.7343,
       "step": 1031
     },
     {
-      "epoch": 1.014763779527559,
-      "grad_norm": 23.071144104003906,
+      "epoch": 1.015748031496063,
+      "grad_norm": 13.065082550048828,
       "learning_rate": 9.997399367269565e-07,
-      "loss": 1.2398,
+      "loss": 0.7647,
       "step": 1032
     },
     {
-      "epoch": 1.015748031496063,
-      "grad_norm": 11.253214836120605,
+      "epoch": 1.016732283464567,
+      "grad_norm": 11.816627502441406,
       "learning_rate": 9.997145815044003e-07,
-      "loss": 0.7096,
+      "loss": 0.7541,
       "step": 1033
     },
     {
-      "epoch": 1.016732283464567,
-      "grad_norm": 10.326289176940918,
+      "epoch": 1.0177165354330708,
+      "grad_norm": 15.631732940673828,
       "learning_rate": 9.99688047429774e-07,
-      "loss": 0.7082,
+      "loss": 1.1746,
       "step": 1034
     },
     {
-      "epoch": 1.0177165354330708,
-      "grad_norm": 10.67347526550293,
+      "epoch": 1.0187007874015748,
+      "grad_norm": 17.95444679260254,
       "learning_rate": 9.996603345658824e-07,
-      "loss": 0.6649,
+      "loss": 1.3935,
       "step": 1035
     },
     {
-      "epoch": 1.0187007874015748,
-      "grad_norm": 20.553695678710938,
+      "epoch": 1.0196850393700787,
+      "grad_norm": 15.705811500549316,
       "learning_rate": 9.996314429783204e-07,
-      "loss": 1.6809,
+      "loss": 1.5224,
       "step": 1036
     },
     {
-      "epoch": 1.0196850393700787,
-      "grad_norm": 14.0806303024292,
+      "epoch": 1.0206692913385826,
+      "grad_norm": 13.482087135314941,
       "learning_rate": 9.99601372735473e-07,
-      "loss": 1.0725,
+      "loss": 0.9067,
       "step": 1037
     },
     {
-      "epoch": 1.0206692913385826,
-      "grad_norm": 12.312373161315918,
+      "epoch": 1.0216535433070866,
+      "grad_norm": 10.162127494812012,
       "learning_rate": 9.995701239085143e-07,
-      "loss": 0.7607,
+      "loss": 0.7184,
       "step": 1038
     },
     {
-      "epoch": 1.0216535433070866,
-      "grad_norm": 13.746301651000977,
+      "epoch": 1.0226377952755905,
+      "grad_norm": 10.461275100708008,
       "learning_rate": 9.99537696571409e-07,
-      "loss": 0.8318,
+      "loss": 0.5991,
       "step": 1039
     },
     {
-      "epoch": 1.0226377952755905,
-      "grad_norm": 27.883398056030273,
+      "epoch": 1.0236220472440944,
+      "grad_norm": 11.486288070678711,
       "learning_rate": 9.995040908009107e-07,
-      "loss": 1.2795,
+      "loss": 0.8841,
       "step": 1040
     },
     {
-      "epoch": 1.0236220472440944,
-      "grad_norm": 12.371521949768066,
+      "epoch": 1.0246062992125984,
+      "grad_norm": 10.825572967529297,
       "learning_rate": 9.994693066765622e-07,
-      "loss": 0.5544,
+      "loss": 0.7474,
       "step": 1041
     },
     {
-      "epoch": 1.0246062992125984,
-      "grad_norm": 10.142796516418457,
+      "epoch": 1.0255905511811023,
+      "grad_norm": 12.2081298828125,
       "learning_rate": 9.994333442806959e-07,
-      "loss": 0.6573,
+      "loss": 0.7444,
       "step": 1042
     },
     {
-      "epoch": 1.0255905511811023,
-      "grad_norm": 13.032113075256348,
+      "epoch": 1.0265748031496063,
+      "grad_norm": 15.263197898864746,
       "learning_rate": 9.993962036984325e-07,
-      "loss": 0.868,
+      "loss": 1.4098,
       "step": 1043
     },
     {
-      "epoch": 1.0265748031496063,
-      "grad_norm": 15.365907669067383,
+      "epoch": 1.0275590551181102,
+      "grad_norm": 9.019082069396973,
       "learning_rate": 9.99357885017682e-07,
-      "loss": 1.1831,
+      "loss": 0.6033,
       "step": 1044
     },
     {
-      "epoch": 1.0275590551181102,
-      "grad_norm": 14.052511215209961,
+      "epoch": 1.0285433070866141,
+      "grad_norm": 14.380988121032715,
       "learning_rate": 9.993183883291425e-07,
-      "loss": 0.9724,
+      "loss": 0.93,
       "step": 1045
     },
     {
-      "epoch": 1.0285433070866141,
-      "grad_norm": 16.52604866027832,
+      "epoch": 1.029527559055118,
+      "grad_norm": 12.470325469970703,
       "learning_rate": 9.992777137263007e-07,
-      "loss": 1.2018,
+      "loss": 0.7308,
       "step": 1046
     },
     {
-      "epoch": 1.029527559055118,
-      "grad_norm": 14.135658264160156,
+      "epoch": 1.030511811023622,
+      "grad_norm": 11.483294486999512,
       "learning_rate": 9.99235861305431e-07,
-      "loss": 0.9539,
+      "loss": 0.7339,
       "step": 1047
     },
     {
-      "epoch": 1.030511811023622,
-      "grad_norm": 9.573715209960938,
+      "epoch": 1.031496062992126,
+      "grad_norm": 17.470186233520508,
       "learning_rate": 9.991928311655956e-07,
-      "loss": 0.5348,
+      "loss": 1.3459,
       "step": 1048
     },
     {
-      "epoch": 1.031496062992126,
-      "grad_norm": 11.995317459106445,
+      "epoch": 1.0324803149606299,
+      "grad_norm": 12.41200065612793,
       "learning_rate": 9.991486234086447e-07,
-      "loss": 0.7531,
+      "loss": 0.6866,
       "step": 1049
     },
     {
-      "epoch": 1.0324803149606299,
-      "grad_norm": 14.20138168334961,
+      "epoch": 1.0334645669291338,
+      "grad_norm": 11.915864944458008,
       "learning_rate": 9.991032381392158e-07,
-      "loss": 1.1062,
+      "loss": 0.8323,
       "step": 1050
     },
     {
-      "epoch": 1.0334645669291338,
-      "grad_norm": 13.821763038635254,
+      "epoch": 1.0344488188976377,
+      "grad_norm": 10.167206764221191,
       "learning_rate": 9.990566754647328e-07,
-      "loss": 1.067,
+      "loss": 0.5301,
       "step": 1051
     },
     {
-      "epoch": 1.0344488188976377,
-      "grad_norm": 13.316102027893066,
+      "epoch": 1.0354330708661417,
+      "grad_norm": 15.526559829711914,
       "learning_rate": 9.990089354954073e-07,
-      "loss": 0.866,
+      "loss": 1.1091,
       "step": 1052
     },
     {
-      "epoch": 1.0354330708661417,
-      "grad_norm": 15.538060188293457,
+      "epoch": 1.0364173228346456,
+      "grad_norm": 11.07171630859375,
       "learning_rate": 9.989600183442373e-07,
-      "loss": 1.2976,
+      "loss": 0.7301,
       "step": 1053
     },
     {
-      "epoch": 1.0364173228346456,
-      "grad_norm": 25.953405380249023,
+      "epoch": 1.0374015748031495,
+      "grad_norm": 13.301454544067383,
       "learning_rate": 9.98909924127007e-07,
-      "loss": 2.035,
+      "loss": 1.1041,
       "step": 1054
     },
     {
-      "epoch": 1.0374015748031495,
-      "grad_norm": 14.323229789733887,
+      "epoch": 1.0383858267716535,
+      "grad_norm": 7.886115074157715,
       "learning_rate": 9.988586529622864e-07,
-      "loss": 1.1908,
+      "loss": 0.433,
       "step": 1055
     },
     {
-      "epoch": 1.0383858267716535,
-      "grad_norm": 13.404814720153809,
+      "epoch": 1.0393700787401574,
+      "grad_norm": 13.891656875610352,
       "learning_rate": 9.98806204971432e-07,
-      "loss": 0.5581,
+      "loss": 0.8808,
       "step": 1056
     },
     {
-      "epoch": 1.0393700787401574,
-      "grad_norm": 14.532567977905273,
+      "epoch": 1.0403543307086613,
+      "grad_norm": 10.688889503479004,
       "learning_rate": 9.987525802785847e-07,
-      "loss": 1.0315,
+      "loss": 0.5836,
       "step": 1057
     },
     {
-      "epoch": 1.0403543307086613,
-      "grad_norm": 13.722807884216309,
+      "epoch": 1.0413385826771653,
+      "grad_norm": 12.516069412231445,
       "learning_rate": 9.986977790106715e-07,
-      "loss": 1.1023,
+      "loss": 0.6459,
       "step": 1058
     },
     {
-      "epoch": 1.0413385826771653,
-      "grad_norm": 10.097068786621094,
+      "epoch": 1.0423228346456692,
+      "grad_norm": 11.766775131225586,
       "learning_rate": 9.986418012974042e-07,
-      "loss": 0.4518,
+      "loss": 0.4647,
       "step": 1059
     },
     {
-      "epoch": 1.0423228346456692,
-      "grad_norm": 13.188447952270508,
+      "epoch": 1.0433070866141732,
+      "grad_norm": 18.46944236755371,
       "learning_rate": 9.985846472712786e-07,
-      "loss": 0.8027,
+      "loss": 1.4988,
       "step": 1060
     },
     {
-      "epoch": 1.0433070866141732,
-      "grad_norm": 27.605440139770508,
+      "epoch": 1.044291338582677,
+      "grad_norm": 14.120805740356445,
       "learning_rate": 9.985263170675756e-07,
-      "loss": 1.9934,
+      "loss": 1.2506,
       "step": 1061
     },
     {
-      "epoch": 1.044291338582677,
-      "grad_norm": 12.822807312011719,
+      "epoch": 1.045275590551181,
+      "grad_norm": 11.723356246948242,
       "learning_rate": 9.98466810824359e-07,
-      "loss": 0.9329,
+      "loss": 0.5674,
       "step": 1062
     },
     {
-      "epoch": 1.045275590551181,
-      "grad_norm": 14.287766456604004,
+      "epoch": 1.046259842519685,
+      "grad_norm": 11.574830055236816,
       "learning_rate": 9.984061286824774e-07,
-      "loss": 0.9931,
+      "loss": 0.8192,
       "step": 1063
     },
     {
-      "epoch": 1.046259842519685,
-      "grad_norm": 14.493118286132812,
+      "epoch": 1.047244094488189,
+      "grad_norm": 19.33531379699707,
       "learning_rate": 9.983442707855616e-07,
-      "loss": 0.7169,
+      "loss": 1.443,
       "step": 1064
     },
     {
-      "epoch": 1.047244094488189,
-      "grad_norm": 20.291234970092773,
+      "epoch": 1.0482283464566928,
+      "grad_norm": 9.310123443603516,
       "learning_rate": 9.982812372800262e-07,
-      "loss": 1.0229,
+      "loss": 0.4771,
       "step": 1065
     },
     {
-      "epoch": 1.0482283464566928,
-      "grad_norm": 10.355449676513672,
+      "epoch": 1.0492125984251968,
+      "grad_norm": 11.541061401367188,
       "learning_rate": 9.982170283150678e-07,
-      "loss": 0.6757,
+      "loss": 0.5886,
       "step": 1066
     },
     {
-      "epoch": 1.0492125984251968,
-      "grad_norm": 13.256071090698242,
+      "epoch": 1.0501968503937007,
+      "grad_norm": 11.022961616516113,
       "learning_rate": 9.981516440426659e-07,
-      "loss": 0.7128,
+      "loss": 0.9119,
       "step": 1067
     },
     {
-      "epoch": 1.0501968503937007,
-      "grad_norm": 8.21869945526123,
+      "epoch": 1.0511811023622046,
+      "grad_norm": 12.574899673461914,
       "learning_rate": 9.98085084617581e-07,
-      "loss": 0.3775,
+      "loss": 0.7875,
       "step": 1068
     },
     {
-      "epoch": 1.0511811023622046,
-      "grad_norm": 14.035961151123047,
+      "epoch": 1.0521653543307086,
+      "grad_norm": 10.970486640930176,
       "learning_rate": 9.980173501973558e-07,
-      "loss": 1.1158,
+      "loss": 0.7041,
       "step": 1069
     },
     {
-      "epoch": 1.0521653543307086,
-      "grad_norm": 14.340359687805176,
+      "epoch": 1.0531496062992125,
+      "grad_norm": 10.811155319213867,
       "learning_rate": 9.97948440942314e-07,
-      "loss": 1.084,
+      "loss": 0.5232,
       "step": 1070
     },
     {
-      "epoch": 1.0531496062992125,
-      "grad_norm": 16.62276840209961,
+      "epoch": 1.0541338582677164,
+      "grad_norm": 10.792261123657227,
       "learning_rate": 9.978783570155597e-07,
-      "loss": 1.0829,
+      "loss": 0.57,
       "step": 1071
     },
     {
-      "epoch": 1.0531496062992125,
-      "eval_Qnli-dev_cosine_accuracy": 0.654296875,
-      "eval_Qnli-dev_cosine_accuracy_threshold": 0.842217206954956,
-      "eval_Qnli-dev_cosine_ap": 0.6736642973454704,
-      "eval_Qnli-dev_cosine_f1": 0.6656534954407294,
-      "eval_Qnli-dev_cosine_f1_threshold": 0.7517247200012207,
-      "eval_Qnli-dev_cosine_precision": 0.518957345971564,
-      "eval_Qnli-dev_cosine_recall": 0.9279661016949152,
-      "eval_Qnli-dev_dot_accuracy": 0.658203125,
-      "eval_Qnli-dev_dot_accuracy_threshold": 660.471923828125,
-      "eval_Qnli-dev_dot_ap": 0.6716435318899119,
-      "eval_Qnli-dev_dot_f1": 0.6666666666666667,
-      "eval_Qnli-dev_dot_f1_threshold": 575.9989013671875,
-      "eval_Qnli-dev_dot_precision": 0.5228915662650603,
-      "eval_Qnli-dev_dot_recall": 0.9194915254237288,
-      "eval_Qnli-dev_euclidean_accuracy": 0.65625,
-      "eval_Qnli-dev_euclidean_accuracy_threshold": 15.113401412963867,
-      "eval_Qnli-dev_euclidean_ap": 0.673907896308458,
-      "eval_Qnli-dev_euclidean_f1": 0.6666666666666667,
-      "eval_Qnli-dev_euclidean_f1_threshold": 19.382671356201172,
-      "eval_Qnli-dev_euclidean_precision": 0.5228915662650603,
-      "eval_Qnli-dev_euclidean_recall": 0.9194915254237288,
-      "eval_Qnli-dev_manhattan_accuracy": 0.66015625,
-      "eval_Qnli-dev_manhattan_accuracy_threshold": 342.40484619140625,
-      "eval_Qnli-dev_manhattan_ap": 0.6866744707607304,
-      "eval_Qnli-dev_manhattan_f1": 0.6666666666666667,
-      "eval_Qnli-dev_manhattan_f1_threshold": 357.13262939453125,
-      "eval_Qnli-dev_manhattan_precision": 0.5846645367412141,
-      "eval_Qnli-dev_manhattan_recall": 0.7754237288135594,
-      "eval_Qnli-dev_max_accuracy": 0.66015625,
-      "eval_Qnli-dev_max_accuracy_threshold": 660.471923828125,
-      "eval_Qnli-dev_max_ap": 0.6866744707607304,
-      "eval_Qnli-dev_max_f1": 0.6666666666666667,
-      "eval_Qnli-dev_max_f1_threshold": 575.9989013671875,
-      "eval_Qnli-dev_max_precision": 0.5846645367412141,
-      "eval_Qnli-dev_max_recall": 0.9279661016949152,
-      "eval_allNLI-dev_cosine_accuracy": 0.689453125,
-      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9341496229171753,
-      "eval_allNLI-dev_cosine_ap": 0.489327543271253,
-      "eval_allNLI-dev_cosine_f1": 0.5300668151447662,
-      "eval_allNLI-dev_cosine_f1_threshold": 0.865491509437561,
-      "eval_allNLI-dev_cosine_precision": 0.4311594202898551,
-      "eval_allNLI-dev_cosine_recall": 0.6878612716763006,
-      "eval_allNLI-dev_dot_accuracy": 0.689453125,
-      "eval_allNLI-dev_dot_accuracy_threshold": 715.212890625,
-      "eval_allNLI-dev_dot_ap": 0.4881798241956048,
-      "eval_allNLI-dev_dot_f1": 0.5309734513274337,
-      "eval_allNLI-dev_dot_f1_threshold": 661.7647094726562,
-      "eval_allNLI-dev_dot_precision": 0.43010752688172044,
-      "eval_allNLI-dev_dot_recall": 0.6936416184971098,
-      "eval_allNLI-dev_euclidean_accuracy": 0.689453125,
-      "eval_allNLI-dev_euclidean_accuracy_threshold": 10.04066276550293,
-      "eval_allNLI-dev_euclidean_ap": 0.4892035422795702,
-      "eval_allNLI-dev_euclidean_f1": 0.5330296127562643,
-      "eval_allNLI-dev_euclidean_f1_threshold": 14.194154739379883,
-      "eval_allNLI-dev_euclidean_precision": 0.4398496240601504,
-      "eval_allNLI-dev_euclidean_recall": 0.6763005780346821,
-      "eval_allNLI-dev_manhattan_accuracy": 0.6875,
-      "eval_allNLI-dev_manhattan_accuracy_threshold": 212.68006896972656,
-      "eval_allNLI-dev_manhattan_ap": 0.4893894820749276,
-      "eval_allNLI-dev_manhattan_f1": 0.5400457665903889,
-      "eval_allNLI-dev_manhattan_f1_threshold": 294.4931640625,
-      "eval_allNLI-dev_manhattan_precision": 0.44696969696969696,
-      "eval_allNLI-dev_manhattan_recall": 0.6820809248554913,
-      "eval_allNLI-dev_max_accuracy": 0.689453125,
-      "eval_allNLI-dev_max_accuracy_threshold": 715.212890625,
-      "eval_allNLI-dev_max_ap": 0.4893894820749276,
-      "eval_allNLI-dev_max_f1": 0.5400457665903889,
-      "eval_allNLI-dev_max_f1_threshold": 661.7647094726562,
-      "eval_allNLI-dev_max_precision": 0.44696969696969696,
-      "eval_allNLI-dev_max_recall": 0.6936416184971098,
-      "eval_loss": 1.2497835159301758,
-      "eval_runtime": 54.1817,
-      "eval_samples_per_second": 30.711,
-      "eval_sequential_score": 0.6866744707607304,
-      "eval_steps_per_second": 0.129,
-      "eval_sts-test_pearson_cosine": 0.5218244250314455,
-      "eval_sts-test_pearson_dot": 0.5186838879324001,
-      "eval_sts-test_pearson_euclidean": 0.5454337101963522,
-      "eval_sts-test_pearson_manhattan": 0.554711626366034,
-      "eval_sts-test_pearson_max": 0.554711626366034,
-      "eval_sts-test_spearman_cosine": 0.5611271163344802,
-      "eval_sts-test_spearman_dot": 0.5566891784935565,
-      "eval_sts-test_spearman_euclidean": 0.5611494472139146,
-      "eval_sts-test_spearman_manhattan": 0.5698060417149441,
-      "eval_sts-test_spearman_max": 0.5698060417149441,
+      "epoch": 1.0541338582677164,
+      "eval_Qnli-dev_cosine_accuracy": 0.6640625,
+      "eval_Qnli-dev_cosine_accuracy_threshold": 0.8653759956359863,
+      "eval_Qnli-dev_cosine_ap": 0.6846632153971953,
+      "eval_Qnli-dev_cosine_f1": 0.6816608996539792,
+      "eval_Qnli-dev_cosine_f1_threshold": 0.7897687554359436,
+      "eval_Qnli-dev_cosine_precision": 0.5760233918128655,
+      "eval_Qnli-dev_cosine_recall": 0.8347457627118644,
+      "eval_Qnli-dev_dot_accuracy": 0.66015625,
+      "eval_Qnli-dev_dot_accuracy_threshold": 657.5789794921875,
+      "eval_Qnli-dev_dot_ap": 0.6832739733257174,
+      "eval_Qnli-dev_dot_f1": 0.6814310051107325,
+      "eval_Qnli-dev_dot_f1_threshold": 598.9862670898438,
+      "eval_Qnli-dev_dot_precision": 0.5698005698005698,
+      "eval_Qnli-dev_dot_recall": 0.847457627118644,
+      "eval_Qnli-dev_euclidean_accuracy": 0.6640625,
+      "eval_Qnli-dev_euclidean_accuracy_threshold": 14.331533432006836,
+      "eval_Qnli-dev_euclidean_ap": 0.6849916947033118,
+      "eval_Qnli-dev_euclidean_f1": 0.6804123711340205,
+      "eval_Qnli-dev_euclidean_f1_threshold": 17.989498138427734,
+      "eval_Qnli-dev_euclidean_precision": 0.5722543352601156,
+      "eval_Qnli-dev_euclidean_recall": 0.8389830508474576,
+      "eval_Qnli-dev_manhattan_accuracy": 0.6796875,
+      "eval_Qnli-dev_manhattan_accuracy_threshold": 294.7474670410156,
+      "eval_Qnli-dev_manhattan_ap": 0.6961691800504137,
+      "eval_Qnli-dev_manhattan_f1": 0.6860068259385665,
+      "eval_Qnli-dev_manhattan_f1_threshold": 376.3381652832031,
+      "eval_Qnli-dev_manhattan_precision": 0.5742857142857143,
+      "eval_Qnli-dev_manhattan_recall": 0.8516949152542372,
+      "eval_Qnli-dev_max_accuracy": 0.6796875,
+      "eval_Qnli-dev_max_accuracy_threshold": 657.5789794921875,
+      "eval_Qnli-dev_max_ap": 0.6961691800504137,
+      "eval_Qnli-dev_max_f1": 0.6860068259385665,
+      "eval_Qnli-dev_max_f1_threshold": 598.9862670898438,
+      "eval_Qnli-dev_max_precision": 0.5760233918128655,
+      "eval_Qnli-dev_max_recall": 0.8516949152542372,
+      "eval_allNLI-dev_cosine_accuracy": 0.69140625,
+      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9270304441452026,
+      "eval_allNLI-dev_cosine_ap": 0.49228397783239236,
+      "eval_allNLI-dev_cosine_f1": 0.5363984674329503,
+      "eval_allNLI-dev_cosine_f1_threshold": 0.8337951898574829,
+      "eval_allNLI-dev_cosine_precision": 0.40114613180515757,
+      "eval_allNLI-dev_cosine_recall": 0.8092485549132948,
+      "eval_allNLI-dev_dot_accuracy": 0.69140625,
+      "eval_allNLI-dev_dot_accuracy_threshold": 709.9082641601562,
+      "eval_allNLI-dev_dot_ap": 0.4925267288107693,
+      "eval_allNLI-dev_dot_f1": 0.5384615384615384,
+      "eval_allNLI-dev_dot_f1_threshold": 638.5916748046875,
+      "eval_allNLI-dev_dot_precision": 0.4034582132564842,
+      "eval_allNLI-dev_dot_recall": 0.8092485549132948,
+      "eval_allNLI-dev_euclidean_accuracy": 0.69140625,
+      "eval_allNLI-dev_euclidean_accuracy_threshold": 10.571220397949219,
+      "eval_allNLI-dev_euclidean_ap": 0.4923798851447917,
+      "eval_allNLI-dev_euclidean_f1": 0.5363984674329503,
+      "eval_allNLI-dev_euclidean_f1_threshold": 15.952140808105469,
+      "eval_allNLI-dev_euclidean_precision": 0.40114613180515757,
+      "eval_allNLI-dev_euclidean_recall": 0.8092485549132948,
+      "eval_allNLI-dev_manhattan_accuracy": 0.689453125,
+      "eval_allNLI-dev_manhattan_accuracy_threshold": 214.40182495117188,
+      "eval_allNLI-dev_manhattan_ap": 0.49086550398875856,
+      "eval_allNLI-dev_manhattan_f1": 0.5391621129326047,
+      "eval_allNLI-dev_manhattan_f1_threshold": 346.060302734375,
+      "eval_allNLI-dev_manhattan_precision": 0.39361702127659576,
+      "eval_allNLI-dev_manhattan_recall": 0.8554913294797688,
+      "eval_allNLI-dev_max_accuracy": 0.69140625,
+      "eval_allNLI-dev_max_accuracy_threshold": 709.9082641601562,
+      "eval_allNLI-dev_max_ap": 0.4925267288107693,
+      "eval_allNLI-dev_max_f1": 0.5391621129326047,
+      "eval_allNLI-dev_max_f1_threshold": 638.5916748046875,
+      "eval_allNLI-dev_max_precision": 0.4034582132564842,
+      "eval_allNLI-dev_max_recall": 0.8554913294797688,
+      "eval_loss": 1.1570727825164795,
+      "eval_runtime": 52.6333,
+      "eval_samples_per_second": 31.615,
+      "eval_sequential_score": 0.6961691800504137,
+      "eval_steps_per_second": 0.133,
+      "eval_sts-test_pearson_cosine": 0.509950648911252,
+      "eval_sts-test_pearson_dot": 0.5045563506807493,
+      "eval_sts-test_pearson_euclidean": 0.5284216887479147,
+      "eval_sts-test_pearson_manhattan": 0.5423369169040173,
+      "eval_sts-test_pearson_max": 0.5423369169040173,
+      "eval_sts-test_spearman_cosine": 0.5372664597807683,
+      "eval_sts-test_spearman_dot": 0.5314762283231191,
+      "eval_sts-test_spearman_euclidean": 0.5375220789858359,
+      "eval_sts-test_spearman_manhattan": 0.5495945820201323,
+      "eval_sts-test_spearman_max": 0.5495945820201323,
       "step": 1071
     },
     {
-      "epoch": 1.0541338582677164,
-      "grad_norm": 28.540851593017578,
+      "epoch": 1.0551181102362204,
+      "grad_norm": 11.420138359069824,
       "learning_rate": 9.978070985829782e-07,
-      "loss": 1.4476,
+      "loss": 0.8948,
       "step": 1072
     },
     {
-      "epoch": 1.0551181102362204,
-      "grad_norm": 17.484149932861328,
+      "epoch": 1.0561023622047243,
+      "grad_norm": 14.187203407287598,
       "learning_rate": 9.97734665813234e-07,
-      "loss": 1.008,
+      "loss": 1.0059,
       "step": 1073
     },
     {
-      "epoch": 1.0561023622047243,
-      "grad_norm": 13.71534538269043,
+      "epoch": 1.0570866141732282,
+      "grad_norm": 11.898282051086426,
       "learning_rate": 9.976610588777717e-07,
-      "loss": 0.8832,
+      "loss": 0.9796,
       "step": 1074
     },
     {
-      "epoch": 1.0570866141732282,
-      "grad_norm": 9.226608276367188,
+      "epoch": 1.0580708661417322,
+      "grad_norm": 12.518900871276855,
       "learning_rate": 9.975862779508149e-07,
-      "loss": 0.5766,
+      "loss": 0.9632,
       "step": 1075
     },
     {
-      "epoch": 1.0580708661417322,
-      "grad_norm": 10.914620399475098,
+      "epoch": 1.0590551181102361,
+      "grad_norm": 15.27397632598877,
       "learning_rate": 9.975103232093656e-07,
-      "loss": 0.8722,
+      "loss": 1.1254,
       "step": 1076
     },
     {
-      "epoch": 1.0590551181102361,
-      "grad_norm": 10.53770923614502,
+      "epoch": 1.06003937007874,
+      "grad_norm": 12.165975570678711,
       "learning_rate": 9.974331948332048e-07,
-      "loss": 0.6285,
+      "loss": 0.8915,
       "step": 1077
     },
     {
-      "epoch": 1.06003937007874,
-      "grad_norm": 11.654394149780273,
+      "epoch": 1.061023622047244,
+      "grad_norm": 11.72553825378418,
       "learning_rate": 9.97354893004891e-07,
-      "loss": 0.8074,
+      "loss": 0.5213,
       "step": 1078
     },
     {
-      "epoch": 1.061023622047244,
-      "grad_norm": 11.405344009399414,
+      "epoch": 1.0620078740157481,
+      "grad_norm": 10.605430603027344,
       "learning_rate": 9.972754179097602e-07,
-      "loss": 0.7646,
+      "loss": 0.6755,
       "step": 1079
     },
     {
-      "epoch": 1.0620078740157481,
-      "grad_norm": 8.713008880615234,
+      "epoch": 1.0629921259842519,
+      "grad_norm": 11.675688743591309,
       "learning_rate": 9.971947697359257e-07,
-      "loss": 0.3947,
+      "loss": 0.6665,
       "step": 1080
     },
     {
-      "epoch": 1.0629921259842519,
-      "grad_norm": 14.822614669799805,
+      "epoch": 1.063976377952756,
+      "grad_norm": 11.114697456359863,
       "learning_rate": 9.971129486742777e-07,
-      "loss": 1.1946,
+      "loss": 0.729,
       "step": 1081
     },
     {
-      "epoch": 1.063976377952756,
-      "grad_norm": 13.049259185791016,
+      "epoch": 1.0649606299212597,
+      "grad_norm": 13.48147201538086,
       "learning_rate": 9.970299549184814e-07,
-      "loss": 1.3833,
+      "loss": 0.9047,
       "step": 1082
     },
     {
-      "epoch": 1.0649606299212597,
-      "grad_norm": 24.78580665588379,
+      "epoch": 1.0659448818897639,
+      "grad_norm": 13.161925315856934,
       "learning_rate": 9.969457886649791e-07,
-      "loss": 1.5897,
+      "loss": 1.0514,
       "step": 1083
     },
     {
-      "epoch": 1.0659448818897639,
-      "grad_norm": 24.568845748901367,
+      "epoch": 1.0669291338582678,
+      "grad_norm": 18.14025115966797,
       "learning_rate": 9.968604501129877e-07,
-      "loss": 1.7621,
+      "loss": 1.6729,
       "step": 1084
     },
     {
-      "epoch": 1.0669291338582678,
-      "grad_norm": 15.05627155303955,
+      "epoch": 1.0679133858267718,
+      "grad_norm": 15.771471977233887,
       "learning_rate": 9.967739394644982e-07,
-      "loss": 1.4584,
+      "loss": 0.8698,
       "step": 1085
     },
     {
-      "epoch": 1.0679133858267718,
-      "grad_norm": 11.705534934997559,
+      "epoch": 1.0688976377952757,
+      "grad_norm": 13.432958602905273,
       "learning_rate": 9.966862569242773e-07,
-      "loss": 0.715,
+      "loss": 1.1209,
       "step": 1086
     },
     {
-      "epoch": 1.0688976377952757,
-      "grad_norm": 11.141972541809082,
+      "epoch": 1.0698818897637796,
+      "grad_norm": 13.478239059448242,
       "learning_rate": 9.965974026998642e-07,
-      "loss": 0.7493,
+      "loss": 0.9425,
       "step": 1087
     },
     {
-      "epoch": 1.0698818897637796,
-      "grad_norm": 11.882009506225586,
+      "epoch": 1.0708661417322836,
+      "grad_norm": 12.611392974853516,
       "learning_rate": 9.965073770015722e-07,
-      "loss": 0.7076,
+      "loss": 0.7151,
       "step": 1088
     },
     {
-      "epoch": 1.0708661417322836,
-      "grad_norm": 16.43975067138672,
+      "epoch": 1.0718503937007875,
+      "grad_norm": 12.270779609680176,
       "learning_rate": 9.964161800424873e-07,
-      "loss": 0.7919,
+      "loss": 0.7336,
       "step": 1089
     },
     {
-      "epoch": 1.0718503937007875,
-      "grad_norm": 11.75568675994873,
+      "epoch": 1.0728346456692914,
+      "grad_norm": 18.48605728149414,
       "learning_rate": 9.963238120384677e-07,
-      "loss": 0.9057,
+      "loss": 1.0971,
       "step": 1090
     },
     {
-      "epoch": 1.0728346456692914,
-      "grad_norm": 9.294759750366211,
+      "epoch": 1.0738188976377954,
+      "grad_norm": 11.107337951660156,
       "learning_rate": 9.962302732081432e-07,
-      "loss": 0.5515,
+      "loss": 0.5334,
       "step": 1091
     },
     {
-      "epoch": 1.0738188976377954,
-      "grad_norm": 9.674452781677246,
+      "epoch": 1.0748031496062993,
+      "grad_norm": 13.366519927978516,
       "learning_rate": 9.961355637729153e-07,
-      "loss": 0.5275,
+      "loss": 1.0171,
       "step": 1092
     },
     {
-      "epoch": 1.0748031496062993,
-      "grad_norm": 10.010574340820312,
+      "epoch": 1.0757874015748032,
+      "grad_norm": 12.98841381072998,
       "learning_rate": 9.960396839569557e-07,
-      "loss": 0.4794,
+      "loss": 1.0153,
       "step": 1093
     },
     {
-      "epoch": 1.0757874015748032,
-      "grad_norm": 23.5357608795166,
+      "epoch": 1.0767716535433072,
+      "grad_norm": 12.056806564331055,
       "learning_rate": 9.95942633987207e-07,
-      "loss": 1.4877,
+      "loss": 0.6662,
       "step": 1094
     },
     {
-      "epoch": 1.0767716535433072,
-      "grad_norm": 18.584951400756836,
+      "epoch": 1.077755905511811,
+      "grad_norm": 11.31603717803955,
       "learning_rate": 9.95844414093381e-07,
-      "loss": 1.1599,
+      "loss": 0.5451,
       "step": 1095
     },
     {
-      "epoch": 1.077755905511811,
-      "grad_norm": 10.287535667419434,
+      "epoch": 1.078740157480315,
+      "grad_norm": 11.599329948425293,
       "learning_rate": 9.957450245079588e-07,
-      "loss": 0.5241,
+      "loss": 0.7743,
       "step": 1096
     },
     {
-      "epoch": 1.078740157480315,
-      "grad_norm": 21.15087890625,
+      "epoch": 1.079724409448819,
+      "grad_norm": 17.17649269104004,
       "learning_rate": 9.9564446546619e-07,
-      "loss": 1.7957,
+      "loss": 0.9187,
       "step": 1097
     },
     {
-      "epoch": 1.079724409448819,
-      "grad_norm": 13.045212745666504,
+      "epoch": 1.080708661417323,
+      "grad_norm": 14.856340408325195,
       "learning_rate": 9.955427372060928e-07,
-      "loss": 1.1295,
+      "loss": 1.1919,
       "step": 1098
     },
     {
-      "epoch": 1.080708661417323,
-      "grad_norm": 19.12482452392578,
+      "epoch": 1.0816929133858268,
+      "grad_norm": 12.51814079284668,
       "learning_rate": 9.954398399684515e-07,
-      "loss": 1.5688,
+      "loss": 0.9405,
       "step": 1099
     },
     {
-      "epoch": 1.0816929133858268,
-      "grad_norm": 9.251370429992676,
+      "epoch": 1.0826771653543308,
+      "grad_norm": 11.004323959350586,
       "learning_rate": 9.953357739968192e-07,
-      "loss": 0.7658,
+      "loss": 0.5361,
       "step": 1100
     },
     {
-      "epoch": 1.0826771653543308,
-      "grad_norm": 9.279908180236816,
+      "epoch": 1.0836614173228347,
+      "grad_norm": 12.910338401794434,
       "learning_rate": 9.952305395375137e-07,
-      "loss": 0.5828,
+      "loss": 1.233,
       "step": 1101
     },
     {
-      "epoch": 1.0836614173228347,
-      "grad_norm": 10.163260459899902,
+      "epoch": 1.0846456692913387,
+      "grad_norm": 13.418724060058594,
       "learning_rate": 9.951241368396194e-07,
-      "loss": 0.8962,
+      "loss": 0.8222,
       "step": 1102
     },
     {
-      "epoch": 1.0846456692913387,
-      "grad_norm": 10.750070571899414,
+      "epoch": 1.0856299212598426,
+      "grad_norm": 11.835745811462402,
       "learning_rate": 9.950165661549856e-07,
-      "loss": 0.9649,
+      "loss": 0.906,
       "step": 1103
     },
     {
-      "epoch": 1.0856299212598426,
-      "grad_norm": 10.144333839416504,
+      "epoch": 1.0866141732283465,
+      "grad_norm": 10.636372566223145,
       "learning_rate": 9.949078277382262e-07,
-      "loss": 0.8328,
+      "loss": 0.62,
       "step": 1104
     },
     {
-      "epoch": 1.0866141732283465,
-      "grad_norm": 19.090774536132812,
+      "epoch": 1.0875984251968505,
+      "grad_norm": 13.1072416305542,
       "learning_rate": 9.947979218467193e-07,
-      "loss": 1.2495,
+      "loss": 0.887,
       "step": 1105
     },
     {
-      "epoch": 1.0875984251968505,
-      "grad_norm": 12.030488967895508,
+      "epoch": 1.0885826771653544,
+      "grad_norm": 9.505353927612305,
       "learning_rate": 9.946868487406057e-07,
-      "loss": 1.1155,
+      "loss": 0.6135,
       "step": 1106
     },
     {
-      "epoch": 1.0885826771653544,
-      "grad_norm": 10.546430587768555,
+      "epoch": 1.0895669291338583,
+      "grad_norm": 8.03498649597168,
       "learning_rate": 9.945746086827896e-07,
-      "loss": 0.6959,
+      "loss": 0.2279,
       "step": 1107
     },
     {
-      "epoch": 1.0895669291338583,
-      "grad_norm": 7.1781415939331055,
+      "epoch": 1.0905511811023623,
+      "grad_norm": 12.267210006713867,
       "learning_rate": 9.94461201938937e-07,
-      "loss": 0.3202,
+      "loss": 1.0091,
       "step": 1108
     },
     {
-      "epoch": 1.0905511811023623,
-      "grad_norm": 12.0957670211792,
+      "epoch": 1.0915354330708662,
+      "grad_norm": 13.677517890930176,
       "learning_rate": 9.943466287774756e-07,
-      "loss": 0.7917,
+      "loss": 1.2457,
       "step": 1109
     },
     {
-      "epoch": 1.0915354330708662,
-      "grad_norm": 21.6634578704834,
+      "epoch": 1.0925196850393701,
+      "grad_norm": 10.236763954162598,
       "learning_rate": 9.942308894695932e-07,
-      "loss": 1.6036,
+      "loss": 0.6233,
       "step": 1110
     },
     {
-      "epoch": 1.0925196850393701,
-      "grad_norm": 19.509628295898438,
+      "epoch": 1.093503937007874,
+      "grad_norm": 8.855351448059082,
       "learning_rate": 9.941139842892391e-07,
-      "loss": 1.2433,
+      "loss": 0.6004,
       "step": 1111
     },
     {
-      "epoch": 1.093503937007874,
-      "grad_norm": 10.067360877990723,
+      "epoch": 1.094488188976378,
+      "grad_norm": 13.564655303955078,
       "learning_rate": 9.939959135131214e-07,
-      "loss": 0.5881,
+      "loss": 1.0716,
       "step": 1112
     },
     {
-      "epoch": 1.094488188976378,
-      "grad_norm": 15.549373626708984,
+      "epoch": 1.095472440944882,
+      "grad_norm": 10.523872375488281,
       "learning_rate": 9.938766774207064e-07,
-      "loss": 1.2943,
+      "loss": 0.5854,
       "step": 1113
     },
     {
-      "epoch": 1.095472440944882,
-      "grad_norm": 8.6616792678833,
+      "epoch": 1.0964566929133859,
+      "grad_norm": 12.336915016174316,
       "learning_rate": 9.937562762942202e-07,
-      "loss": 0.4795,
+      "loss": 0.5708,
       "step": 1114
     },
     {
-      "epoch": 1.0964566929133859,
-      "grad_norm": 17.8710994720459,
+      "epoch": 1.0974409448818898,
+      "grad_norm": 15.987162590026855,
       "learning_rate": 9.936347104186452e-07,
-      "loss": 1.2879,
+      "loss": 1.0223,
       "step": 1115
     },
     {
-      "epoch": 1.0974409448818898,
-      "grad_norm": 11.639845848083496,
+      "epoch": 1.0984251968503937,
+      "grad_norm": 11.978333473205566,
       "learning_rate": 9.935119800817215e-07,
-      "loss": 0.7128,
+      "loss": 0.6434,
       "step": 1116
     },
     {
-      "epoch": 1.0984251968503937,
-      "grad_norm": 11.503822326660156,
+      "epoch": 1.0994094488188977,
+      "grad_norm": 8.372782707214355,
       "learning_rate": 9.933880855739446e-07,
-      "loss": 0.579,
+      "loss": 0.363,
       "step": 1117
     },
     {
-      "epoch": 1.0994094488188977,
-      "grad_norm": 8.639110565185547,
+      "epoch": 1.1003937007874016,
+      "grad_norm": 17.084253311157227,
       "learning_rate": 9.932630271885662e-07,
-      "loss": 0.4919,
+      "loss": 1.2779,
       "step": 1118
     },
     {
-      "epoch": 1.1003937007874016,
-      "grad_norm": 11.358433723449707,
+      "epoch": 1.1013779527559056,
+      "grad_norm": 13.697333335876465,
       "learning_rate": 9.93136805221593e-07,
-      "loss": 0.8719,
+      "loss": 0.8207,
       "step": 1119
     },
     {
-      "epoch": 1.1013779527559056,
-      "grad_norm": 12.555047988891602,
+      "epoch": 1.1023622047244095,
+      "grad_norm": 12.337996482849121,
       "learning_rate": 9.930094199717848e-07,
-      "loss": 1.1696,
+      "loss": 0.6738,
       "step": 1120
     },
     {
-      "epoch": 1.1023622047244095,
-      "grad_norm": 16.483301162719727,
+      "epoch": 1.1033464566929134,
+      "grad_norm": 15.996990203857422,
       "learning_rate": 9.92880871740656e-07,
-      "loss": 1.0572,
+      "loss": 1.0475,
       "step": 1121
     },
     {
-      "epoch": 1.1033464566929134,
-      "grad_norm": 10.812005043029785,
+      "epoch": 1.1043307086614174,
+      "grad_norm": 12.84057331085205,
       "learning_rate": 9.927511608324732e-07,
-      "loss": 0.746,
+      "loss": 0.9189,
       "step": 1122
     },
     {
-      "epoch": 1.1043307086614174,
-      "grad_norm": 7.33451509475708,
+      "epoch": 1.1053149606299213,
+      "grad_norm": 18.542404174804688,
       "learning_rate": 9.926202875542546e-07,
-      "loss": 0.4024,
+      "loss": 1.0991,
       "step": 1123
     },
     {
-      "epoch": 1.1053149606299213,
-      "grad_norm": 22.47677230834961,
+      "epoch": 1.1062992125984252,
+      "grad_norm": 11.175984382629395,
       "learning_rate": 9.924882522157706e-07,
-      "loss": 1.506,
+      "loss": 0.7348,
       "step": 1124
     },
     {
-      "epoch": 1.1062992125984252,
-      "grad_norm": 12.610832214355469,
+      "epoch": 1.1072834645669292,
+      "grad_norm": 11.271822929382324,
       "learning_rate": 9.923550551295411e-07,
-      "loss": 0.8658,
+      "loss": 0.7178,
       "step": 1125
     },
     {
-      "epoch": 1.1072834645669292,
-      "grad_norm": 11.658797264099121,
+      "epoch": 1.108267716535433,
+      "grad_norm": 15.300597190856934,
       "learning_rate": 9.922206966108367e-07,
-      "loss": 0.8811,
+      "loss": 0.893,
       "step": 1126
     },
     {
-      "epoch": 1.108267716535433,
-      "grad_norm": 10.466711044311523,
+      "epoch": 1.109251968503937,
+      "grad_norm": 12.767745018005371,
       "learning_rate": 9.920851769776761e-07,
-      "loss": 0.5166,
+      "loss": 0.7549,
       "step": 1127
     },
     {
-      "epoch": 1.109251968503937,
-      "grad_norm": 12.000123023986816,
+      "epoch": 1.110236220472441,
+      "grad_norm": 15.920074462890625,
       "learning_rate": 9.919484965508276e-07,
-      "loss": 0.6676,
+      "loss": 1.4074,
       "step": 1128
     },
     {
-      "epoch": 1.110236220472441,
-      "grad_norm": 10.173340797424316,
+      "epoch": 1.111220472440945,
+      "grad_norm": 15.358880996704102,
       "learning_rate": 9.918106556538057e-07,
-      "loss": 0.6104,
+      "loss": 0.8321,
       "step": 1129
     },
     {
-      "epoch": 1.111220472440945,
-      "grad_norm": 12.212109565734863,
+      "epoch": 1.1122047244094488,
+      "grad_norm": 15.989846229553223,
       "learning_rate": 9.91671654612872e-07,
-      "loss": 0.9263,
+      "loss": 1.0289,
       "step": 1130
     },
     {
-      "epoch": 1.1122047244094488,
-      "grad_norm": 15.39807415008545,
+      "epoch": 1.1131889763779528,
+      "grad_norm": 15.226240158081055,
       "learning_rate": 9.91531493757035e-07,
-      "loss": 1.27,
+      "loss": 0.8296,
       "step": 1131
     },
     {
-      "epoch": 1.1131889763779528,
-      "grad_norm": 11.779406547546387,
+      "epoch": 1.1141732283464567,
+      "grad_norm": 12.149371147155762,
       "learning_rate": 9.913901734180476e-07,
-      "loss": 0.7009,
+      "loss": 0.7587,
       "step": 1132
     },
     {
-      "epoch": 1.1141732283464567,
-      "grad_norm": 22.418588638305664,
+      "epoch": 1.1151574803149606,
+      "grad_norm": 9.797683715820312,
       "learning_rate": 9.912476939304069e-07,
-      "loss": 1.7367,
+      "loss": 0.6198,
       "step": 1133
     },
     {
-      "epoch": 1.1151574803149606,
-      "grad_norm": 9.897189140319824,
+      "epoch": 1.1161417322834646,
+      "grad_norm": 12.730241775512695,
       "learning_rate": 9.911040556313543e-07,
-      "loss": 0.5924,
+      "loss": 0.7474,
       "step": 1134
     },
     {
-      "epoch": 1.1161417322834646,
-      "grad_norm": 25.989843368530273,
+      "epoch": 1.1171259842519685,
+      "grad_norm": 11.825194358825684,
       "learning_rate": 9.909592588608738e-07,
-      "loss": 1.4601,
+      "loss": 0.719,
       "step": 1135
     },
     {
-      "epoch": 1.1171259842519685,
-      "grad_norm": 24.011293411254883,
+      "epoch": 1.1181102362204725,
+      "grad_norm": 10.120687484741211,
       "learning_rate": 9.908133039616911e-07,
-      "loss": 1.4386,
+      "loss": 0.517,
       "step": 1136
     },
     {
-      "epoch": 1.1181102362204725,
-      "grad_norm": 11.448659896850586,
+      "epoch": 1.1190944881889764,
+      "grad_norm": 9.076939582824707,
       "learning_rate": 9.90666191279274e-07,
-      "loss": 0.7663,
+      "loss": 0.4452,
       "step": 1137
     },
     {
-      "epoch": 1.1190944881889764,
-      "grad_norm": 13.132770538330078,
+      "epoch": 1.1200787401574803,
+      "grad_norm": 17.70829200744629,
       "learning_rate": 9.905179211618296e-07,
-      "loss": 1.4043,
+      "loss": 1.1683,
       "step": 1138
     },
     {
-      "epoch": 1.1200787401574803,
-      "grad_norm": 12.378631591796875,
+      "epoch": 1.1210629921259843,
+      "grad_norm": 11.209772109985352,
       "learning_rate": 9.903684939603054e-07,
-      "loss": 1.0486,
+      "loss": 0.7705,
       "step": 1139
     },
     {
-      "epoch": 1.1210629921259843,
-      "grad_norm": 10.946438789367676,
+      "epoch": 1.1220472440944882,
+      "grad_norm": 12.0265531539917,
       "learning_rate": 9.902179100283875e-07,
-      "loss": 0.4541,
+      "loss": 0.7873,
       "step": 1140
     },
     {
-      "epoch": 1.1220472440944882,
-      "grad_norm": 9.304766654968262,
+      "epoch": 1.1230314960629921,
+      "grad_norm": 12.67635440826416,
       "learning_rate": 9.900661697224995e-07,
-      "loss": 0.6372,
+      "loss": 0.7939,
       "step": 1141
     },
     {
-      "epoch": 1.1230314960629921,
-      "grad_norm": 9.965361595153809,
+      "epoch": 1.124015748031496,
+      "grad_norm": 15.988715171813965,
       "learning_rate": 9.899132734018024e-07,
-      "loss": 0.6563,
+      "loss": 0.6513,
       "step": 1142
     },
     {
-      "epoch": 1.124015748031496,
-      "grad_norm": 9.754569053649902,
+      "epoch": 1.125,
+      "grad_norm": 16.87392234802246,
       "learning_rate": 9.897592214281936e-07,
-      "loss": 0.4799,
+      "loss": 1.4787,
       "step": 1143
     },
     {
-      "epoch": 1.125,
-      "grad_norm": 17.194753646850586,
+      "epoch": 1.125984251968504,
+      "grad_norm": 10.136147499084473,
       "learning_rate": 9.896040141663054e-07,
-      "loss": 1.1878,
+      "loss": 0.7314,
       "step": 1144
     },
     {
-      "epoch": 1.125984251968504,
-      "grad_norm": 10.091248512268066,
+      "epoch": 1.1269685039370079,
+      "grad_norm": 12.934366226196289,
       "learning_rate": 9.894476519835052e-07,
-      "loss": 0.6507,
+      "loss": 0.4927,
       "step": 1145
     },
     {
-      "epoch": 1.1269685039370079,
-      "grad_norm": 16.77588653564453,
+      "epoch": 1.1279527559055118,
+      "grad_norm": 12.322945594787598,
       "learning_rate": 9.892901352498932e-07,
-      "loss": 0.8475,
+      "loss": 0.9718,
       "step": 1146
     },
     {
-      "epoch": 1.1279527559055118,
-      "grad_norm": 12.568476676940918,
+      "epoch": 1.1289370078740157,
+      "grad_norm": 12.120319366455078,
       "learning_rate": 9.891314643383037e-07,
-      "loss": 1.107,
+      "loss": 0.6738,
       "step": 1147
     },
     {
-      "epoch": 1.1289370078740157,
-      "grad_norm": 9.016369819641113,
+      "epoch": 1.1299212598425197,
+      "grad_norm": 13.021424293518066,
       "learning_rate": 9.88971639624301e-07,
-      "loss": 0.3035,
+      "loss": 1.2356,
       "step": 1148
     },
     {
-      "epoch": 1.1299212598425197,
-      "grad_norm": 9.463494300842285,
+      "epoch": 1.1309055118110236,
+      "grad_norm": 12.181797981262207,
       "learning_rate": 9.888106614861823e-07,
-      "loss": 0.6317,
+      "loss": 0.902,
       "step": 1149
     },
     {
-      "epoch": 1.1309055118110236,
-      "grad_norm": 16.098876953125,
+      "epoch": 1.1318897637795275,
+      "grad_norm": 11.161940574645996,
       "learning_rate": 9.886485303049734e-07,
-      "loss": 1.0571,
+      "loss": 0.5957,
       "step": 1150
     },
     {
-      "epoch": 1.1318897637795275,
-      "grad_norm": 11.583295822143555,
+      "epoch": 1.1328740157480315,
+      "grad_norm": 12.929612159729004,
       "learning_rate": 9.884852464644305e-07,
-      "loss": 0.6284,
+      "loss": 1.1595,
       "step": 1151
     },
     {
-      "epoch": 1.1328740157480315,
-      "grad_norm": 11.977601051330566,
+      "epoch": 1.1338582677165354,
+      "grad_norm": 12.247835159301758,
       "learning_rate": 9.883208103510367e-07,
-      "loss": 0.9719,
+      "loss": 0.9146,
       "step": 1152
     },
     {
-      "epoch": 1.1338582677165354,
-      "grad_norm": 12.4282808303833,
+      "epoch": 1.1348425196850394,
+      "grad_norm": 10.77420425415039,
       "learning_rate": 9.88155222354004e-07,
-      "loss": 0.9043,
+      "loss": 0.5402,
       "step": 1153
     },
     {
-      "epoch": 1.1348425196850394,
-      "grad_norm": 11.01272201538086,
+      "epoch": 1.1358267716535433,
+      "grad_norm": 10.38335132598877,
       "learning_rate": 9.879884828652698e-07,
-      "loss": 0.6841,
+      "loss": 0.6177,
       "step": 1154
     },
     {
-      "epoch": 1.1358267716535433,
-      "grad_norm": 12.006009101867676,
+      "epoch": 1.1368110236220472,
+      "grad_norm": 11.626049995422363,
       "learning_rate": 9.878205922794972e-07,
-      "loss": 0.608,
+      "loss": 0.6705,
       "step": 1155
     },
     {
-      "epoch": 1.1368110236220472,
-      "grad_norm": 9.32274055480957,
+      "epoch": 1.1377952755905512,
+      "grad_norm": 10.968165397644043,
       "learning_rate": 9.876515509940742e-07,
-      "loss": 0.5592,
+      "loss": 0.603,
       "step": 1156
     },
     {
-      "epoch": 1.1377952755905512,
-      "grad_norm": 22.739818572998047,
+      "epoch": 1.138779527559055,
+      "grad_norm": 13.380332946777344,
       "learning_rate": 9.874813594091122e-07,
-      "loss": 1.4329,
+      "loss": 0.733,
       "step": 1157
     },
     {
-      "epoch": 1.138779527559055,
-      "grad_norm": 26.053585052490234,
+      "epoch": 1.139763779527559,
+      "grad_norm": 15.77918815612793,
       "learning_rate": 9.873100179274453e-07,
-      "loss": 1.6512,
+      "loss": 1.1947,
       "step": 1158
     },
     {
-      "epoch": 1.139763779527559,
-      "grad_norm": 23.698720932006836,
+      "epoch": 1.140748031496063,
+      "grad_norm": 13.412628173828125,
       "learning_rate": 9.871375269546292e-07,
-      "loss": 1.9971,
+      "loss": 0.6454,
       "step": 1159
     },
     {
-      "epoch": 1.140748031496063,
-      "grad_norm": 16.412256240844727,
+      "epoch": 1.141732283464567,
+      "grad_norm": 18.430992126464844,
       "learning_rate": 9.869638868989407e-07,
-      "loss": 0.9536,
+      "loss": 1.0318,
       "step": 1160
     },
     {
-      "epoch": 1.141732283464567,
-      "grad_norm": 14.230514526367188,
+      "epoch": 1.1427165354330708,
+      "grad_norm": 15.241898536682129,
       "learning_rate": 9.867890981713764e-07,
-      "loss": 0.7213,
+      "loss": 1.0791,
       "step": 1161
     },
     {
-      "epoch": 1.1427165354330708,
-      "grad_norm": 10.371112823486328,
+      "epoch": 1.1437007874015748,
+      "grad_norm": 13.973746299743652,
       "learning_rate": 9.866131611856514e-07,
-      "loss": 0.4714,
+      "loss": 0.928,
       "step": 1162
     },
     {
-      "epoch": 1.1437007874015748,
-      "grad_norm": 9.045125961303711,
+      "epoch": 1.1446850393700787,
+      "grad_norm": 8.853302955627441,
       "learning_rate": 9.864360763581994e-07,
-      "loss": 0.5469,
+      "loss": 0.5488,
       "step": 1163
     },
     {
-      "epoch": 1.1446850393700787,
-      "grad_norm": 13.141439437866211,
+      "epoch": 1.1456692913385826,
+      "grad_norm": 11.555764198303223,
       "learning_rate": 9.862578441081699e-07,
-      "loss": 0.7808,
+      "loss": 0.7937,
       "step": 1164
     },
     {
-      "epoch": 1.1456692913385826,
-      "grad_norm": 20.23772430419922,
+      "epoch": 1.1466535433070866,
+      "grad_norm": 14.18860149383545,
       "learning_rate": 9.860784648574288e-07,
-      "loss": 2.1225,
+      "loss": 0.6262,
       "step": 1165
     },
     {
-      "epoch": 1.1466535433070866,
-      "grad_norm": 12.796348571777344,
+      "epoch": 1.1476377952755905,
+      "grad_norm": 11.897910118103027,
       "learning_rate": 9.858979390305573e-07,
-      "loss": 1.0035,
+      "loss": 0.6925,
       "step": 1166
     },
     {
-      "epoch": 1.1476377952755905,
-      "grad_norm": 11.537859916687012,
+      "epoch": 1.1486220472440944,
+      "grad_norm": 10.954864501953125,
       "learning_rate": 9.857162670548502e-07,
-      "loss": 0.8881,
+      "loss": 0.6949,
       "step": 1167
     },
     {
-      "epoch": 1.1486220472440944,
-      "grad_norm": 11.287174224853516,
+      "epoch": 1.1496062992125984,
+      "grad_norm": 8.025981903076172,
       "learning_rate": 9.855334493603147e-07,
-      "loss": 1.0899,
+      "loss": 0.2552,
       "step": 1168
     },
     {
-      "epoch": 1.1496062992125984,
-      "grad_norm": 10.592592239379883,
+      "epoch": 1.1505905511811023,
+      "grad_norm": 13.133017539978027,
       "learning_rate": 9.853494863796705e-07,
-      "loss": 0.814,
+      "loss": 0.7877,
       "step": 1169
     },
     {
-      "epoch": 1.1505905511811023,
-      "grad_norm": 13.20379638671875,
+      "epoch": 1.1515748031496063,
+      "grad_norm": 13.997974395751953,
       "learning_rate": 9.851643785483478e-07,
-      "loss": 1.1564,
+      "loss": 0.7581,
       "step": 1170
     },
     {
-      "epoch": 1.1515748031496063,
-      "grad_norm": 12.289298057556152,
+      "epoch": 1.1525590551181102,
+      "grad_norm": 10.509552001953125,
       "learning_rate": 9.849781263044867e-07,
-      "loss": 1.1262,
+      "loss": 0.4293,
       "step": 1171
     },
     {
-      "epoch": 1.1525590551181102,
-      "grad_norm": 10.34423828125,
+      "epoch": 1.1535433070866141,
+      "grad_norm": 8.699508666992188,
       "learning_rate": 9.847907300889358e-07,
-      "loss": 0.4826,
+      "loss": 0.3826,
       "step": 1172
     },
     {
-      "epoch": 1.1535433070866141,
-      "grad_norm": 13.520644187927246,
+      "epoch": 1.154527559055118,
+      "grad_norm": 14.012443542480469,
       "learning_rate": 9.846021903452519e-07,
-      "loss": 1.072,
+      "loss": 0.9302,
       "step": 1173
     },
     {
-      "epoch": 1.154527559055118,
-      "grad_norm": 10.882362365722656,
+      "epoch": 1.155511811023622,
+      "grad_norm": 13.481894493103027,
       "learning_rate": 9.844125075196983e-07,
-      "loss": 0.7588,
+      "loss": 0.9708,
       "step": 1174
     },
     {
-      "epoch": 1.155511811023622,
-      "grad_norm": 12.282682418823242,
+      "epoch": 1.156496062992126,
+      "grad_norm": 13.585330963134766,
       "learning_rate": 9.842216820612438e-07,
-      "loss": 0.9597,
+      "loss": 0.4477,
       "step": 1175
     },
     {
-      "epoch": 1.156496062992126,
-      "grad_norm": 14.361637115478516,
+      "epoch": 1.1574803149606299,
+      "grad_norm": 13.615195274353027,
       "learning_rate": 9.840297144215615e-07,
-      "loss": 0.9954,
+      "loss": 0.7183,
       "step": 1176
     },
     {
-      "epoch": 1.1574803149606299,
-      "grad_norm": 10.931534767150879,
+      "epoch": 1.1584645669291338,
+      "grad_norm": 11.44372272491455,
       "learning_rate": 9.838366050550287e-07,
-      "loss": 0.7169,
+      "loss": 0.6087,
       "step": 1177
     },
     {
-      "epoch": 1.1584645669291338,
-      "grad_norm": 8.973247528076172,
+      "epoch": 1.1594488188976377,
+      "grad_norm": 15.111401557922363,
       "learning_rate": 9.836423544187245e-07,
-      "loss": 0.6022,
+      "loss": 0.8324,
       "step": 1178
     },
     {
-      "epoch": 1.1594488188976377,
-      "grad_norm": 13.785079002380371,
+      "epoch": 1.1604330708661417,
+      "grad_norm": 10.387323379516602,
       "learning_rate": 9.834469629724293e-07,
-      "loss": 0.9108,
+      "loss": 0.3906,
       "step": 1179
     },
     {
-      "epoch": 1.1604330708661417,
-      "grad_norm": 13.577901840209961,
+      "epoch": 1.1614173228346456,
+      "grad_norm": 19.592727661132812,
       "learning_rate": 9.832504311786238e-07,
-      "loss": 0.7757,
+      "loss": 1.5957,
       "step": 1180
     },
     {
-      "epoch": 1.1614173228346456,
-      "grad_norm": 15.905202865600586,
+      "epoch": 1.1624015748031495,
+      "grad_norm": 16.09340476989746,
       "learning_rate": 9.830527595024885e-07,
-      "loss": 1.3626,
+      "loss": 1.2033,
       "step": 1181
     },
     {
-      "epoch": 1.1624015748031495,
-      "grad_norm": 10.73210334777832,
+      "epoch": 1.1633858267716535,
+      "grad_norm": 7.257636070251465,
       "learning_rate": 9.82853948411901e-07,
-      "loss": 0.7546,
+      "loss": 0.473,
       "step": 1182
     },
     {
-      "epoch": 1.1633858267716535,
-      "grad_norm": 7.652894020080566,
+      "epoch": 1.1643700787401574,
+      "grad_norm": 9.731563568115234,
       "learning_rate": 9.826539983774365e-07,
-      "loss": 0.4991,
+      "loss": 0.6048,
       "step": 1183
     },
     {
-      "epoch": 1.1643700787401574,
-      "grad_norm": 7.244395732879639,
+      "epoch": 1.1653543307086613,
+      "grad_norm": 13.61319637298584,
       "learning_rate": 9.824529098723657e-07,
-      "loss": 0.359,
+      "loss": 0.861,
       "step": 1184
     },
     {
-      "epoch": 1.1653543307086613,
-      "grad_norm": 12.004920959472656,
+      "epoch": 1.1663385826771653,
+      "grad_norm": 10.9859619140625,
       "learning_rate": 9.822506833726537e-07,
-      "loss": 0.8804,
+      "loss": 0.9065,
       "step": 1185
     },
     {
-      "epoch": 1.1663385826771653,
-      "grad_norm": 12.034069061279297,
+      "epoch": 1.1673228346456692,
+      "grad_norm": 13.370589256286621,
       "learning_rate": 9.8204731935696e-07,
-      "loss": 0.7941,
+      "loss": 0.6253,
       "step": 1186
     },
     {
-      "epoch": 1.1673228346456692,
-      "grad_norm": 22.406587600708008,
+      "epoch": 1.1683070866141732,
+      "grad_norm": 16.47172737121582,
       "learning_rate": 9.818428183066356e-07,
-      "loss": 1.6374,
+      "loss": 1.227,
       "step": 1187
     },
     {
-      "epoch": 1.1683070866141732,
-      "grad_norm": 13.935935020446777,
+      "epoch": 1.169291338582677,
+      "grad_norm": 12.20095443725586,
       "learning_rate": 9.816371807057237e-07,
-      "loss": 0.8655,
+      "loss": 0.7943,
       "step": 1188
     },
     {
-      "epoch": 1.169291338582677,
-      "grad_norm": 12.156387329101562,
+      "epoch": 1.170275590551181,
+      "grad_norm": 15.804914474487305,
       "learning_rate": 9.81430407040957e-07,
-      "loss": 0.7996,
+      "loss": 0.6892,
       "step": 1189
     },
     {
-      "epoch": 1.170275590551181,
-      "grad_norm": 10.461490631103516,
+      "epoch": 1.171259842519685,
+      "grad_norm": 15.404899597167969,
       "learning_rate": 9.812224978017573e-07,
-      "loss": 0.77,
+      "loss": 1.0856,
       "step": 1190
     },
     {
-      "epoch": 1.171259842519685,
-      "grad_norm": 9.46623706817627,
+      "epoch": 1.172244094488189,
+      "grad_norm": 14.905608177185059,
       "learning_rate": 9.810134534802346e-07,
-      "loss": 0.7023,
+      "loss": 0.7547,
       "step": 1191
     },
     {
-      "epoch": 1.172244094488189,
-      "grad_norm": 10.952614784240723,
+      "epoch": 1.1732283464566928,
+      "grad_norm": 8.873392105102539,
       "learning_rate": 9.80803274571185e-07,
-      "loss": 0.8983,
+      "loss": 0.4079,
       "step": 1192
     },
     {
-      "epoch": 1.1732283464566928,
-      "grad_norm": 10.143698692321777,
+      "epoch": 1.1742125984251968,
+      "grad_norm": 15.257758140563965,
       "learning_rate": 9.80591961572091e-07,
-      "loss": 0.7334,
+      "loss": 0.6927,
       "step": 1193
     },
     {
-      "epoch": 1.1742125984251968,
-      "grad_norm": 10.362936019897461,
+      "epoch": 1.1751968503937007,
+      "grad_norm": 16.975690841674805,
       "learning_rate": 9.803795149831183e-07,
-      "loss": 0.588,
+      "loss": 1.0321,
       "step": 1194
     },
     {
-      "epoch": 1.1751968503937007,
-      "grad_norm": 7.492079257965088,
+      "epoch": 1.1761811023622046,
+      "grad_norm": 18.011857986450195,
       "learning_rate": 9.801659353071165e-07,
-      "loss": 0.3758,
+      "loss": 1.1502,
       "step": 1195
     },
     {
-      "epoch": 1.1761811023622046,
-      "grad_norm": 11.673321723937988,
+      "epoch": 1.1771653543307086,
+      "grad_norm": 15.919508934020996,
       "learning_rate": 9.799512230496172e-07,
-      "loss": 0.8057,
+      "loss": 0.9344,
       "step": 1196
     },
     {
-      "epoch": 1.1771653543307086,
-      "grad_norm": 28.830455780029297,
+      "epoch": 1.1781496062992125,
+      "grad_norm": 16.511510848999023,
       "learning_rate": 9.79735378718832e-07,
-      "loss": 1.6137,
+      "loss": 1.0466,
       "step": 1197
     },
     {
-      "epoch": 1.1781496062992125,
-      "grad_norm": 12.126893043518066,
+      "epoch": 1.1791338582677164,
+      "grad_norm": 13.429428100585938,
       "learning_rate": 9.795184028256534e-07,
-      "loss": 0.6031,
+      "loss": 0.6004,
       "step": 1198
     },
     {
-      "epoch": 1.1791338582677164,
-      "grad_norm": 13.487756729125977,
+      "epoch": 1.1801181102362204,
+      "grad_norm": 15.42149543762207,
       "learning_rate": 9.793002958836506e-07,
-      "loss": 0.9049,
+      "loss": 0.7064,
       "step": 1199
     },
     {
-      "epoch": 1.1801181102362204,
-      "grad_norm": 21.260787963867188,
+      "epoch": 1.1811023622047245,
+      "grad_norm": 16.78218650817871,
       "learning_rate": 9.790810584090712e-07,
-      "loss": 0.9688,
+      "loss": 0.659,
       "step": 1200
     },
     {
-      "epoch": 1.1811023622047245,
-      "grad_norm": 10.346322059631348,
+      "epoch": 1.1820866141732282,
+      "grad_norm": 12.032279968261719,
       "learning_rate": 9.78860690920838e-07,
-      "loss": 0.4923,
+      "loss": 0.6445,
       "step": 1201
     },
     {
-      "epoch": 1.1820866141732282,
-      "grad_norm": 12.38548755645752,
+      "epoch": 1.1830708661417324,
+      "grad_norm": 17.824844360351562,
       "learning_rate": 9.78639193940549e-07,
-      "loss": 0.6468,
+      "loss": 1.1254,
       "step": 1202
     },
     {
-      "epoch": 1.1830708661417324,
-      "grad_norm": 26.022510528564453,
+      "epoch": 1.1840551181102361,
+      "grad_norm": 11.511914253234863,
       "learning_rate": 9.784165679924752e-07,
-      "loss": 1.907,
+      "loss": 0.7727,
       "step": 1203
     },
     {
-      "epoch": 1.1840551181102361,
-      "grad_norm": 12.170500755310059,
+      "epoch": 1.1850393700787403,
+      "grad_norm": 14.361802101135254,
       "learning_rate": 9.781928136035604e-07,
-      "loss": 0.81,
+      "loss": 0.8912,
       "step": 1204
     },
     {
-      "epoch": 1.1850393700787403,
-      "grad_norm": 19.44597625732422,
+      "epoch": 1.186023622047244,
+      "grad_norm": 15.696497917175293,
       "learning_rate": 9.779679313034182e-07,
-      "loss": 1.1321,
+      "loss": 1.3996,
       "step": 1205
     },
     {
-      "epoch": 1.186023622047244,
-      "grad_norm": 24.822078704833984,
+      "epoch": 1.1870078740157481,
+      "grad_norm": 12.164986610412598,
       "learning_rate": 9.777419216243333e-07,
-      "loss": 1.901,
+      "loss": 0.6232,
       "step": 1206
     },
     {
-      "epoch": 1.1870078740157481,
-      "grad_norm": 12.606575965881348,
+      "epoch": 1.1879921259842519,
+      "grad_norm": 18.16672134399414,
       "learning_rate": 9.77514785101258e-07,
-      "loss": 0.9074,
+      "loss": 1.2074,
       "step": 1207
     },
     {
-      "epoch": 1.1879921259842519,
-      "grad_norm": 12.32088851928711,
+      "epoch": 1.188976377952756,
+      "grad_norm": 10.229595184326172,
       "learning_rate": 9.77286522271812e-07,
-      "loss": 0.7982,
+      "loss": 0.7199,
       "step": 1208
     },
     {
-      "epoch": 1.188976377952756,
-      "grad_norm": 11.44482707977295,
+      "epoch": 1.1899606299212597,
+      "grad_norm": 11.71091079711914,
       "learning_rate": 9.77057133676281e-07,
-      "loss": 0.6944,
+      "loss": 0.5826,
       "step": 1209
     },
     {
-      "epoch": 1.1899606299212597,
-      "grad_norm": 8.395225524902344,
+      "epoch": 1.1909448818897639,
+      "grad_norm": 11.842780113220215,
       "learning_rate": 9.768266198576147e-07,
-      "loss": 0.3568,
+      "loss": 0.7618,
       "step": 1210
     },
     {
-      "epoch": 1.1909448818897639,
-      "grad_norm": 10.126123428344727,
+      "epoch": 1.1919291338582676,
+      "grad_norm": 18.636600494384766,
       "learning_rate": 9.765949813614273e-07,
-      "loss": 0.605,
+      "loss": 1.1995,
       "step": 1211
     },
     {
-      "epoch": 1.1919291338582676,
-      "grad_norm": 9.255550384521484,
+      "epoch": 1.1929133858267718,
+      "grad_norm": 11.783096313476562,
       "learning_rate": 9.763622187359942e-07,
-      "loss": 0.515,
+      "loss": 0.6375,
       "step": 1212
     },
     {
-      "epoch": 1.1929133858267718,
-      "grad_norm": 18.876310348510742,
+      "epoch": 1.1938976377952755,
+      "grad_norm": 13.159656524658203,
       "learning_rate": 9.761283325322515e-07,
-      "loss": 1.2768,
+      "loss": 0.5623,
       "step": 1213
     },
     {
-      "epoch": 1.1938976377952755,
-      "grad_norm": 10.210433959960938,
+      "epoch": 1.1948818897637796,
+      "grad_norm": 13.676416397094727,
       "learning_rate": 9.758933233037952e-07,
-      "loss": 0.5625,
+      "loss": 0.7756,
       "step": 1214
     },
     {
-      "epoch": 1.1948818897637796,
-      "grad_norm": 8.423827171325684,
+      "epoch": 1.1958661417322836,
+      "grad_norm": 12.268596649169922,
       "learning_rate": 9.756571916068793e-07,
-      "loss": 0.3923,
+      "loss": 0.8011,
       "step": 1215
     },
     {
-      "epoch": 1.1958661417322836,
-      "grad_norm": 11.02546215057373,
+      "epoch": 1.1968503937007875,
+      "grad_norm": 9.940037727355957,
       "learning_rate": 9.754199380004145e-07,
-      "loss": 0.5358,
+      "loss": 0.4404,
       "step": 1216
     },
     {
-      "epoch": 1.1968503937007875,
-      "grad_norm": 20.572431564331055,
+      "epoch": 1.1978346456692914,
+      "grad_norm": 11.429154396057129,
       "learning_rate": 9.75181563045967e-07,
-      "loss": 0.9994,
+      "loss": 0.8401,
       "step": 1217
     },
     {
-      "epoch": 1.1978346456692914,
-      "grad_norm": 13.444281578063965,
+      "epoch": 1.1988188976377954,
+      "grad_norm": 11.402060508728027,
       "learning_rate": 9.749420673077574e-07,
-      "loss": 0.5955,
+      "loss": 0.8971,
       "step": 1218
     },
     {
-      "epoch": 1.1988188976377954,
-      "grad_norm": 11.807435035705566,
+      "epoch": 1.1998031496062993,
+      "grad_norm": 12.799181938171387,
       "learning_rate": 9.747014513526588e-07,
-      "loss": 0.6657,
+      "loss": 1.0111,
       "step": 1219
     },
     {
-      "epoch": 1.1998031496062993,
-      "grad_norm": 11.162291526794434,
+      "epoch": 1.2007874015748032,
+      "grad_norm": 10.88442611694336,
       "learning_rate": 9.744597157501957e-07,
-      "loss": 0.7751,
+      "loss": 0.8536,
       "step": 1220
     },
     {
-      "epoch": 1.2007874015748032,
-      "grad_norm": 12.134571075439453,
+      "epoch": 1.2017716535433072,
+      "grad_norm": 14.517230987548828,
       "learning_rate": 9.742168610725437e-07,
-      "loss": 0.8443,
+      "loss": 1.1102,
       "step": 1221
     },
     {
-      "epoch": 1.2017716535433072,
-      "grad_norm": 12.475569725036621,
+      "epoch": 1.202755905511811,
+      "grad_norm": 9.811517715454102,
       "learning_rate": 9.73972887894526e-07,
-      "loss": 0.6662,
+      "loss": 0.581,
       "step": 1222
     },
     {
-      "epoch": 1.202755905511811,
-      "grad_norm": 10.001572608947754,
+      "epoch": 1.203740157480315,
+      "grad_norm": 13.741737365722656,
       "learning_rate": 9.737277967936137e-07,
-      "loss": 0.7022,
+      "loss": 0.8276,
       "step": 1223
     },
     {
-      "epoch": 1.203740157480315,
-      "grad_norm": 13.214725494384766,
+      "epoch": 1.204724409448819,
+      "grad_norm": 13.864313125610352,
       "learning_rate": 9.734815883499241e-07,
-      "loss": 0.8183,
+      "loss": 1.1131,
       "step": 1224
     },
     {
-      "epoch": 1.203740157480315,
-      "eval_Qnli-dev_cosine_accuracy": 0.681640625,
-      "eval_Qnli-dev_cosine_accuracy_threshold": 0.8366125226020813,
-      "eval_Qnli-dev_cosine_ap": 0.6914476487679233,
-      "eval_Qnli-dev_cosine_f1": 0.6784,
-      "eval_Qnli-dev_cosine_f1_threshold": 0.7529718279838562,
-      "eval_Qnli-dev_cosine_precision": 0.5449871465295629,
-      "eval_Qnli-dev_cosine_recall": 0.8983050847457628,
-      "eval_Qnli-dev_dot_accuracy": 0.67578125,
-      "eval_Qnli-dev_dot_accuracy_threshold": 639.7266845703125,
-      "eval_Qnli-dev_dot_ap": 0.6893503306461788,
-      "eval_Qnli-dev_dot_f1": 0.6772655007949125,
-      "eval_Qnli-dev_dot_f1_threshold": 571.5009155273438,
-      "eval_Qnli-dev_dot_precision": 0.5419847328244275,
-      "eval_Qnli-dev_dot_recall": 0.902542372881356,
-      "eval_Qnli-dev_euclidean_accuracy": 0.681640625,
-      "eval_Qnli-dev_euclidean_accuracy_threshold": 15.806737899780273,
-      "eval_Qnli-dev_euclidean_ap": 0.6915811548453872,
-      "eval_Qnli-dev_euclidean_f1": 0.6784,
-      "eval_Qnli-dev_euclidean_f1_threshold": 19.451509475708008,
-      "eval_Qnli-dev_euclidean_precision": 0.5449871465295629,
-      "eval_Qnli-dev_euclidean_recall": 0.8983050847457628,
-      "eval_Qnli-dev_manhattan_accuracy": 0.68359375,
-      "eval_Qnli-dev_manhattan_accuracy_threshold": 344.76611328125,
-      "eval_Qnli-dev_manhattan_ap": 0.6984341364216773,
-      "eval_Qnli-dev_manhattan_f1": 0.6807817589576547,
-      "eval_Qnli-dev_manhattan_f1_threshold": 400.9693603515625,
-      "eval_Qnli-dev_manhattan_precision": 0.5529100529100529,
-      "eval_Qnli-dev_manhattan_recall": 0.885593220338983,
-      "eval_Qnli-dev_max_accuracy": 0.68359375,
-      "eval_Qnli-dev_max_accuracy_threshold": 639.7266845703125,
-      "eval_Qnli-dev_max_ap": 0.6984341364216773,
-      "eval_Qnli-dev_max_f1": 0.6807817589576547,
-      "eval_Qnli-dev_max_f1_threshold": 571.5009155273438,
-      "eval_Qnli-dev_max_precision": 0.5529100529100529,
-      "eval_Qnli-dev_max_recall": 0.902542372881356,
-      "eval_allNLI-dev_cosine_accuracy": 0.68359375,
-      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9374961853027344,
-      "eval_allNLI-dev_cosine_ap": 0.4873775690193095,
-      "eval_allNLI-dev_cosine_f1": 0.533066132264529,
-      "eval_allNLI-dev_cosine_f1_threshold": 0.8453455567359924,
-      "eval_allNLI-dev_cosine_precision": 0.40797546012269936,
-      "eval_allNLI-dev_cosine_recall": 0.7687861271676301,
-      "eval_allNLI-dev_dot_accuracy": 0.68359375,
-      "eval_allNLI-dev_dot_accuracy_threshold": 717.646484375,
-      "eval_allNLI-dev_dot_ap": 0.48976942081457003,
-      "eval_allNLI-dev_dot_f1": 0.5338645418326693,
-      "eval_allNLI-dev_dot_f1_threshold": 646.2739868164062,
-      "eval_allNLI-dev_dot_precision": 0.4072948328267477,
-      "eval_allNLI-dev_dot_recall": 0.7745664739884393,
-      "eval_allNLI-dev_euclidean_accuracy": 0.68359375,
-      "eval_allNLI-dev_euclidean_accuracy_threshold": 9.785663604736328,
-      "eval_allNLI-dev_euclidean_ap": 0.4873970729071831,
-      "eval_allNLI-dev_euclidean_f1": 0.5319999999999999,
-      "eval_allNLI-dev_euclidean_f1_threshold": 15.387359619140625,
-      "eval_allNLI-dev_euclidean_precision": 0.40672782874617736,
-      "eval_allNLI-dev_euclidean_recall": 0.7687861271676301,
-      "eval_allNLI-dev_manhattan_accuracy": 0.681640625,
-      "eval_allNLI-dev_manhattan_accuracy_threshold": 205.621337890625,
-      "eval_allNLI-dev_manhattan_ap": 0.48655776096738934,
-      "eval_allNLI-dev_manhattan_f1": 0.5379876796714579,
-      "eval_allNLI-dev_manhattan_f1_threshold": 313.74407958984375,
-      "eval_allNLI-dev_manhattan_precision": 0.4171974522292994,
-      "eval_allNLI-dev_manhattan_recall": 0.7572254335260116,
-      "eval_allNLI-dev_max_accuracy": 0.68359375,
-      "eval_allNLI-dev_max_accuracy_threshold": 717.646484375,
-      "eval_allNLI-dev_max_ap": 0.48976942081457003,
-      "eval_allNLI-dev_max_f1": 0.5379876796714579,
-      "eval_allNLI-dev_max_f1_threshold": 646.2739868164062,
-      "eval_allNLI-dev_max_precision": 0.4171974522292994,
-      "eval_allNLI-dev_max_recall": 0.7745664739884393,
-      "eval_loss": 1.1757967472076416,
-      "eval_runtime": 54.4087,
-      "eval_samples_per_second": 30.583,
-      "eval_sequential_score": 0.6984341364216773,
-      "eval_steps_per_second": 0.129,
-      "eval_sts-test_pearson_cosine": 0.5452067506531562,
-      "eval_sts-test_pearson_dot": 0.5408381648828177,
-      "eval_sts-test_pearson_euclidean": 0.5684167455418432,
-      "eval_sts-test_pearson_manhattan": 0.5766689893476588,
-      "eval_sts-test_pearson_max": 0.5766689893476588,
-      "eval_sts-test_spearman_cosine": 0.5782599130018705,
-      "eval_sts-test_spearman_dot": 0.5738021503521504,
-      "eval_sts-test_spearman_euclidean": 0.5785303001231974,
-      "eval_sts-test_spearman_manhattan": 0.5863883978672202,
-      "eval_sts-test_spearman_max": 0.5863883978672202,
+      "epoch": 1.204724409448819,
+      "eval_Qnli-dev_cosine_accuracy": 0.662109375,
+      "eval_Qnli-dev_cosine_accuracy_threshold": 0.8402308225631714,
+      "eval_Qnli-dev_cosine_ap": 0.6870663974200225,
+      "eval_Qnli-dev_cosine_f1": 0.679646017699115,
+      "eval_Qnli-dev_cosine_f1_threshold": 0.7844527959823608,
+      "eval_Qnli-dev_cosine_precision": 0.5835866261398176,
+      "eval_Qnli-dev_cosine_recall": 0.8135593220338984,
+      "eval_Qnli-dev_dot_accuracy": 0.662109375,
+      "eval_Qnli-dev_dot_accuracy_threshold": 642.9093017578125,
+      "eval_Qnli-dev_dot_ap": 0.6854636952685231,
+      "eval_Qnli-dev_dot_f1": 0.6830985915492958,
+      "eval_Qnli-dev_dot_f1_threshold": 598.628662109375,
+      "eval_Qnli-dev_dot_precision": 0.5843373493975904,
+      "eval_Qnli-dev_dot_recall": 0.8220338983050848,
+      "eval_Qnli-dev_euclidean_accuracy": 0.6640625,
+      "eval_Qnli-dev_euclidean_accuracy_threshold": 15.632356643676758,
+      "eval_Qnli-dev_euclidean_ap": 0.6875052693474604,
+      "eval_Qnli-dev_euclidean_f1": 0.679646017699115,
+      "eval_Qnli-dev_euclidean_f1_threshold": 18.142810821533203,
+      "eval_Qnli-dev_euclidean_precision": 0.5835866261398176,
+      "eval_Qnli-dev_euclidean_recall": 0.8135593220338984,
+      "eval_Qnli-dev_manhattan_accuracy": 0.66796875,
+      "eval_Qnli-dev_manhattan_accuracy_threshold": 313.628662109375,
+      "eval_Qnli-dev_manhattan_ap": 0.6998084195621053,
+      "eval_Qnli-dev_manhattan_f1": 0.6832740213523131,
+      "eval_Qnli-dev_manhattan_f1_threshold": 375.14727783203125,
+      "eval_Qnli-dev_manhattan_precision": 0.588957055214724,
+      "eval_Qnli-dev_manhattan_recall": 0.8135593220338984,
+      "eval_Qnli-dev_max_accuracy": 0.66796875,
+      "eval_Qnli-dev_max_accuracy_threshold": 642.9093017578125,
+      "eval_Qnli-dev_max_ap": 0.6998084195621053,
+      "eval_Qnli-dev_max_f1": 0.6832740213523131,
+      "eval_Qnli-dev_max_f1_threshold": 598.628662109375,
+      "eval_Qnli-dev_max_precision": 0.588957055214724,
+      "eval_Qnli-dev_max_recall": 0.8220338983050848,
+      "eval_allNLI-dev_cosine_accuracy": 0.6953125,
+      "eval_allNLI-dev_cosine_accuracy_threshold": 0.924639105796814,
+      "eval_allNLI-dev_cosine_ap": 0.4972415283301538,
+      "eval_allNLI-dev_cosine_f1": 0.5409836065573771,
+      "eval_allNLI-dev_cosine_f1_threshold": 0.8456767797470093,
+      "eval_allNLI-dev_cosine_precision": 0.41904761904761906,
+      "eval_allNLI-dev_cosine_recall": 0.7630057803468208,
+      "eval_allNLI-dev_dot_accuracy": 0.693359375,
+      "eval_allNLI-dev_dot_accuracy_threshold": 709.039794921875,
+      "eval_allNLI-dev_dot_ap": 0.4974522853968933,
+      "eval_allNLI-dev_dot_f1": 0.539877300613497,
+      "eval_allNLI-dev_dot_f1_threshold": 646.7660522460938,
+      "eval_allNLI-dev_dot_precision": 0.4177215189873418,
+      "eval_allNLI-dev_dot_recall": 0.7630057803468208,
+      "eval_allNLI-dev_euclidean_accuracy": 0.6953125,
+      "eval_allNLI-dev_euclidean_accuracy_threshold": 10.734097480773926,
+      "eval_allNLI-dev_euclidean_ap": 0.4970191291040448,
+      "eval_allNLI-dev_euclidean_f1": 0.5409836065573771,
+      "eval_allNLI-dev_euclidean_f1_threshold": 15.374061584472656,
+      "eval_allNLI-dev_euclidean_precision": 0.41904761904761906,
+      "eval_allNLI-dev_euclidean_recall": 0.7630057803468208,
+      "eval_allNLI-dev_manhattan_accuracy": 0.69140625,
+      "eval_allNLI-dev_manhattan_accuracy_threshold": 223.44715881347656,
+      "eval_allNLI-dev_manhattan_ap": 0.4947474533883158,
+      "eval_allNLI-dev_manhattan_f1": 0.5476190476190477,
+      "eval_allNLI-dev_manhattan_f1_threshold": 330.39208984375,
+      "eval_allNLI-dev_manhattan_precision": 0.4169184290030212,
+      "eval_allNLI-dev_manhattan_recall": 0.7976878612716763,
+      "eval_allNLI-dev_max_accuracy": 0.6953125,
+      "eval_allNLI-dev_max_accuracy_threshold": 709.039794921875,
+      "eval_allNLI-dev_max_ap": 0.4974522853968933,
+      "eval_allNLI-dev_max_f1": 0.5476190476190477,
+      "eval_allNLI-dev_max_f1_threshold": 646.7660522460938,
+      "eval_allNLI-dev_max_precision": 0.41904761904761906,
+      "eval_allNLI-dev_max_recall": 0.7976878612716763,
+      "eval_loss": 1.0811175107955933,
+      "eval_runtime": 53.0985,
+      "eval_samples_per_second": 31.338,
+      "eval_sequential_score": 0.6998084195621053,
+      "eval_steps_per_second": 0.132,
+      "eval_sts-test_pearson_cosine": 0.570631571629065,
+      "eval_sts-test_pearson_dot": 0.5666437665381776,
+      "eval_sts-test_pearson_euclidean": 0.5864775199138799,
+      "eval_sts-test_pearson_manhattan": 0.5931981726951323,
+      "eval_sts-test_pearson_max": 0.5931981726951323,
+      "eval_sts-test_spearman_cosine": 0.5896615946279099,
+      "eval_sts-test_spearman_dot": 0.5849432006318465,
+      "eval_sts-test_spearman_euclidean": 0.5902238942272928,
+      "eval_sts-test_spearman_manhattan": 0.5950810171336699,
+      "eval_sts-test_spearman_max": 0.5950810171336699,
       "step": 1224
     },
     {
-      "epoch": 1.204724409448819,
-      "grad_norm": 27.127208709716797,
+      "epoch": 1.205708661417323,
+      "grad_norm": 14.709962844848633,
       "learning_rate": 9.73234263146219e-07,
-      "loss": 1.7184,
+      "loss": 0.7577,
       "step": 1225
     },
     {
-      "epoch": 1.205708661417323,
-      "grad_norm": 27.30764389038086,
+      "epoch": 1.2066929133858268,
+      "grad_norm": 12.769095420837402,
       "learning_rate": 9.72985821767904e-07,
-      "loss": 2.0612,
+      "loss": 0.7671,
       "step": 1226
     },
     {
-      "epoch": 1.2066929133858268,
-      "grad_norm": 12.28317928314209,
+      "epoch": 1.2076771653543308,
+      "grad_norm": 13.248802185058594,
       "learning_rate": 9.727362648030254e-07,
-      "loss": 0.5645,
+      "loss": 1.0486,
       "step": 1227
     },
     {
-      "epoch": 1.2076771653543308,
-      "grad_norm": 11.45736312866211,
+      "epoch": 1.2086614173228347,
+      "grad_norm": 11.779559135437012,
       "learning_rate": 9.724855928422714e-07,
-      "loss": 1.0122,
+      "loss": 0.5076,
       "step": 1228
     },
     {
-      "epoch": 1.2086614173228347,
-      "grad_norm": 7.623450756072998,
+      "epoch": 1.2096456692913387,
+      "grad_norm": 10.296235084533691,
       "learning_rate": 9.722338064789688e-07,
-      "loss": 0.4022,
+      "loss": 0.7368,
       "step": 1229
     },
     {
-      "epoch": 1.2096456692913387,
-      "grad_norm": 18.705665588378906,
+      "epoch": 1.2106299212598426,
+      "grad_norm": 11.898944854736328,
       "learning_rate": 9.719809063090818e-07,
-      "loss": 1.2996,
+      "loss": 0.8573,
       "step": 1230
     },
     {
-      "epoch": 1.2106299212598426,
-      "grad_norm": 10.509492874145508,
+      "epoch": 1.2116141732283465,
+      "grad_norm": 12.604121208190918,
       "learning_rate": 9.717268929312112e-07,
-      "loss": 0.5619,
+      "loss": 0.4998,
       "step": 1231
     },
     {
-      "epoch": 1.2116141732283465,
-      "grad_norm": 7.465246200561523,
+      "epoch": 1.2125984251968505,
+      "grad_norm": 16.115983963012695,
       "learning_rate": 9.714717669465932e-07,
-      "loss": 0.2787,
+      "loss": 0.9818,
       "step": 1232
     },
     {
-      "epoch": 1.2125984251968505,
-      "grad_norm": 10.551802635192871,
+      "epoch": 1.2135826771653544,
+      "grad_norm": 12.059069633483887,
       "learning_rate": 9.712155289590964e-07,
-      "loss": 0.423,
+      "loss": 0.5621,
       "step": 1233
     },
     {
-      "epoch": 1.2135826771653544,
-      "grad_norm": 14.872739791870117,
+      "epoch": 1.2145669291338583,
+      "grad_norm": 12.484044075012207,
       "learning_rate": 9.709581795752225e-07,
-      "loss": 0.7389,
+      "loss": 0.8134,
       "step": 1234
     },
     {
-      "epoch": 1.2145669291338583,
-      "grad_norm": 14.757356643676758,
+      "epoch": 1.2155511811023623,
+      "grad_norm": 13.233085632324219,
       "learning_rate": 9.70699719404103e-07,
-      "loss": 1.0838,
+      "loss": 0.7607,
       "step": 1235
     },
     {
-      "epoch": 1.2155511811023623,
-      "grad_norm": 10.091594696044922,
+      "epoch": 1.2165354330708662,
+      "grad_norm": 15.704331398010254,
       "learning_rate": 9.704401490574993e-07,
-      "loss": 0.5171,
+      "loss": 1.0499,
       "step": 1236
     },
     {
-      "epoch": 1.2165354330708662,
-      "grad_norm": 11.12486743927002,
+      "epoch": 1.2175196850393701,
+      "grad_norm": 11.812478065490723,
       "learning_rate": 9.701794691497997e-07,
-      "loss": 0.5735,
+      "loss": 0.594,
       "step": 1237
     },
     {
-      "epoch": 1.2175196850393701,
-      "grad_norm": 10.067895889282227,
+      "epoch": 1.218503937007874,
+      "grad_norm": 12.935066223144531,
       "learning_rate": 9.699176802980199e-07,
-      "loss": 0.716,
+      "loss": 0.9004,
       "step": 1238
     },
     {
-      "epoch": 1.218503937007874,
-      "grad_norm": 14.799269676208496,
+      "epoch": 1.219488188976378,
+      "grad_norm": 15.198156356811523,
       "learning_rate": 9.69654783121799e-07,
-      "loss": 1.0944,
+      "loss": 0.9055,
       "step": 1239
     },
     {
-      "epoch": 1.219488188976378,
-      "grad_norm": 8.924025535583496,
+      "epoch": 1.220472440944882,
+      "grad_norm": 15.90765380859375,
       "learning_rate": 9.693907782434008e-07,
-      "loss": 0.409,
+      "loss": 1.2924,
       "step": 1240
     },
     {
-      "epoch": 1.220472440944882,
-      "grad_norm": 23.94953155517578,
+      "epoch": 1.2214566929133859,
+      "grad_norm": 14.856472969055176,
       "learning_rate": 9.691256662877099e-07,
-      "loss": 1.5226,
+      "loss": 0.9998,
       "step": 1241
     },
     {
-      "epoch": 1.2214566929133859,
-      "grad_norm": 11.415987014770508,
+      "epoch": 1.2224409448818898,
+      "grad_norm": 13.175382614135742,
       "learning_rate": 9.688594478822317e-07,
-      "loss": 0.6814,
+      "loss": 0.8414,
       "step": 1242
     },
     {
-      "epoch": 1.2224409448818898,
-      "grad_norm": 13.597917556762695,
+      "epoch": 1.2234251968503937,
+      "grad_norm": 12.428009986877441,
       "learning_rate": 9.685921236570914e-07,
-      "loss": 0.9776,
+      "loss": 0.8105,
       "step": 1243
     },
     {
-      "epoch": 1.2234251968503937,
-      "grad_norm": 12.646177291870117,
+      "epoch": 1.2244094488188977,
+      "grad_norm": 11.19924545288086,
       "learning_rate": 9.683236942450294e-07,
-      "loss": 0.6928,
+      "loss": 0.862,
       "step": 1244
     },
     {
-      "epoch": 1.2244094488188977,
-      "grad_norm": 24.075674057006836,
+      "epoch": 1.2253937007874016,
+      "grad_norm": 16.816848754882812,
       "learning_rate": 9.680541602814047e-07,
-      "loss": 1.1014,
+      "loss": 1.1903,
       "step": 1245
     },
     {
-      "epoch": 1.2253937007874016,
-      "grad_norm": 13.862969398498535,
+      "epoch": 1.2263779527559056,
+      "grad_norm": 9.508321762084961,
       "learning_rate": 9.677835224041885e-07,
-      "loss": 1.0794,
+      "loss": 0.5475,
       "step": 1246
     },
     {
-      "epoch": 1.2263779527559056,
-      "grad_norm": 10.0800199508667,
+      "epoch": 1.2273622047244095,
+      "grad_norm": 8.490301132202148,
       "learning_rate": 9.67511781253966e-07,
-      "loss": 0.4316,
+      "loss": 0.4714,
       "step": 1247
     },
     {
-      "epoch": 1.2273622047244095,
-      "grad_norm": 16.01119613647461,
+      "epoch": 1.2283464566929134,
+      "grad_norm": 10.48558235168457,
       "learning_rate": 9.67238937473934e-07,
-      "loss": 1.0016,
+      "loss": 0.582,
       "step": 1248
     },
     {
-      "epoch": 1.2283464566929134,
-      "grad_norm": 9.573379516601562,
+      "epoch": 1.2293307086614174,
+      "grad_norm": 11.935273170471191,
       "learning_rate": 9.669649917098983e-07,
-      "loss": 0.4825,
+      "loss": 0.8019,
       "step": 1249
     },
     {
-      "epoch": 1.2293307086614174,
-      "grad_norm": 19.94251823425293,
+      "epoch": 1.2303149606299213,
+      "grad_norm": 8.721261978149414,
       "learning_rate": 9.666899446102736e-07,
-      "loss": 1.2514,
+      "loss": 0.46,
       "step": 1250
     },
     {
-      "epoch": 1.2303149606299213,
-      "grad_norm": 9.705304145812988,
+      "epoch": 1.2312992125984252,
+      "grad_norm": 9.019754409790039,
       "learning_rate": 9.664137968260814e-07,
-      "loss": 0.462,
+      "loss": 0.4769,
       "step": 1251
     },
     {
-      "epoch": 1.2312992125984252,
-      "grad_norm": 13.630583763122559,
+      "epoch": 1.2322834645669292,
+      "grad_norm": 10.739852905273438,
       "learning_rate": 9.661365490109482e-07,
-      "loss": 0.9723,
+      "loss": 0.727,
       "step": 1252
     },
     {
-      "epoch": 1.2322834645669292,
-      "grad_norm": 10.716950416564941,
+      "epoch": 1.233267716535433,
+      "grad_norm": 9.422754287719727,
       "learning_rate": 9.658582018211045e-07,
-      "loss": 0.6441,
+      "loss": 0.3343,
       "step": 1253
     },
     {
-      "epoch": 1.233267716535433,
-      "grad_norm": 6.912999153137207,
+      "epoch": 1.234251968503937,
+      "grad_norm": 15.779256820678711,
       "learning_rate": 9.655787559153824e-07,
-      "loss": 0.3009,
+      "loss": 1.0573,
       "step": 1254
     },
     {
-      "epoch": 1.234251968503937,
-      "grad_norm": 15.139958381652832,
+      "epoch": 1.235236220472441,
+      "grad_norm": 11.698262214660645,
       "learning_rate": 9.652982119552154e-07,
-      "loss": 1.0618,
+      "loss": 0.7796,
       "step": 1255
     },
     {
-      "epoch": 1.235236220472441,
-      "grad_norm": 29.90904998779297,
+      "epoch": 1.236220472440945,
+      "grad_norm": 11.095013618469238,
       "learning_rate": 9.650165706046359e-07,
-      "loss": 1.9636,
+      "loss": 0.7146,
       "step": 1256
     },
     {
-      "epoch": 1.236220472440945,
-      "grad_norm": 22.840852737426758,
+      "epoch": 1.2372047244094488,
+      "grad_norm": 11.641757011413574,
       "learning_rate": 9.64733832530273e-07,
-      "loss": 1.7589,
+      "loss": 0.6668,
       "step": 1257
     },
     {
-      "epoch": 1.2372047244094488,
-      "grad_norm": 12.522883415222168,
+      "epoch": 1.2381889763779528,
+      "grad_norm": 12.829182624816895,
       "learning_rate": 9.644499984013525e-07,
-      "loss": 0.6347,
+      "loss": 0.7252,
       "step": 1258
     },
     {
-      "epoch": 1.2381889763779528,
-      "grad_norm": 14.57872200012207,
+      "epoch": 1.2391732283464567,
+      "grad_norm": 7.879256725311279,
       "learning_rate": 9.64165068889694e-07,
-      "loss": 0.6183,
+      "loss": 0.4865,
       "step": 1259
     },
     {
-      "epoch": 1.2391732283464567,
-      "grad_norm": 11.782417297363281,
+      "epoch": 1.2401574803149606,
+      "grad_norm": 8.990915298461914,
       "learning_rate": 9.6387904466971e-07,
-      "loss": 0.6868,
+      "loss": 0.479,
       "step": 1260
     },
     {
-      "epoch": 1.2401574803149606,
-      "grad_norm": 10.705400466918945,
+      "epoch": 1.2411417322834646,
+      "grad_norm": 16.715126037597656,
       "learning_rate": 9.635919264184043e-07,
-      "loss": 0.6504,
+      "loss": 1.0787,
       "step": 1261
     },
     {
-      "epoch": 1.2411417322834646,
-      "grad_norm": 9.323182106018066,
+      "epoch": 1.2421259842519685,
+      "grad_norm": 11.397772789001465,
       "learning_rate": 9.6330371481537e-07,
-      "loss": 0.4346,
+      "loss": 0.619,
       "step": 1262
     },
     {
-      "epoch": 1.2421259842519685,
-      "grad_norm": 11.712157249450684,
+      "epoch": 1.2431102362204725,
+      "grad_norm": 9.551841735839844,
       "learning_rate": 9.630144105427883e-07,
-      "loss": 0.6985,
+      "loss": 0.6382,
       "step": 1263
     },
     {
-      "epoch": 1.2431102362204725,
-      "grad_norm": 11.031139373779297,
+      "epoch": 1.2440944881889764,
+      "grad_norm": 9.50159740447998,
       "learning_rate": 9.627240142854262e-07,
-      "loss": 0.5558,
+      "loss": 0.333,
       "step": 1264
     },
     {
-      "epoch": 1.2440944881889764,
-      "grad_norm": 15.171723365783691,
+      "epoch": 1.2450787401574803,
+      "grad_norm": 10.793553352355957,
       "learning_rate": 9.624325267306357e-07,
-      "loss": 0.9465,
+      "loss": 0.6462,
       "step": 1265
     },
     {
-      "epoch": 1.2450787401574803,
-      "grad_norm": 16.123682022094727,
+      "epoch": 1.2460629921259843,
+      "grad_norm": 13.310639381408691,
       "learning_rate": 9.621399485683523e-07,
-      "loss": 1.0173,
+      "loss": 0.8646,
       "step": 1266
     },
     {
-      "epoch": 1.2460629921259843,
-      "grad_norm": 11.502947807312012,
+      "epoch": 1.2470472440944882,
+      "grad_norm": 15.79319953918457,
       "learning_rate": 9.618462804910917e-07,
-      "loss": 0.6847,
+      "loss": 1.1484,
       "step": 1267
     },
     {
-      "epoch": 1.2470472440944882,
-      "grad_norm": 13.352543830871582,
+      "epoch": 1.2480314960629921,
+      "grad_norm": 12.593992233276367,
       "learning_rate": 9.61551523193951e-07,
-      "loss": 1.0888,
+      "loss": 0.8422,
       "step": 1268
     },
     {
-      "epoch": 1.2480314960629921,
-      "grad_norm": 26.212068557739258,
+      "epoch": 1.249015748031496,
+      "grad_norm": 10.541960716247559,
       "learning_rate": 9.612556773746038e-07,
-      "loss": 1.5393,
+      "loss": 0.7198,
       "step": 1269
     },
     {
-      "epoch": 1.249015748031496,
-      "grad_norm": 7.848837852478027,
+      "epoch": 1.25,
+      "grad_norm": 9.414874076843262,
       "learning_rate": 9.609587437333011e-07,
-      "loss": 0.466,
+      "loss": 0.524,
       "step": 1270
     },
     {
-      "epoch": 1.25,
-      "grad_norm": 9.938773155212402,
+      "epoch": 1.250984251968504,
+      "grad_norm": 11.995010375976562,
       "learning_rate": 9.606607229728688e-07,
-      "loss": 0.4438,
+      "loss": 0.7523,
       "step": 1271
     },
     {
-      "epoch": 1.250984251968504,
-      "grad_norm": 11.60131549835205,
+      "epoch": 1.2519685039370079,
+      "grad_norm": 11.432456016540527,
       "learning_rate": 9.603616157987054e-07,
-      "loss": 0.6365,
+      "loss": 0.7618,
       "step": 1272
     },
     {
-      "epoch": 1.2519685039370079,
-      "grad_norm": 20.468914031982422,
+      "epoch": 1.2529527559055118,
+      "grad_norm": 14.102582931518555,
       "learning_rate": 9.600614229187812e-07,
-      "loss": 1.244,
+      "loss": 0.9805,
       "step": 1273
     },
     {
-      "epoch": 1.2529527559055118,
-      "grad_norm": 14.907753944396973,
+      "epoch": 1.2539370078740157,
+      "grad_norm": 14.343584060668945,
       "learning_rate": 9.597601450436364e-07,
-      "loss": 1.0352,
+      "loss": 0.8194,
       "step": 1274
     },
     {
-      "epoch": 1.2539370078740157,
-      "grad_norm": 13.460160255432129,
+      "epoch": 1.2549212598425197,
+      "grad_norm": 12.313678741455078,
       "learning_rate": 9.594577828863794e-07,
-      "loss": 0.8419,
+      "loss": 0.3379,
       "step": 1275
     },
     {
-      "epoch": 1.2549212598425197,
-      "grad_norm": 22.66904067993164,
+      "epoch": 1.2559055118110236,
+      "grad_norm": 14.750896453857422,
       "learning_rate": 9.591543371626842e-07,
-      "loss": 1.6384,
+      "loss": 0.925,
       "step": 1276
     },
     {
-      "epoch": 1.2559055118110236,
-      "grad_norm": 19.56178855895996,
+      "epoch": 1.2568897637795275,
+      "grad_norm": 7.83335018157959,
       "learning_rate": 9.588498085907905e-07,
-      "loss": 1.3742,
+      "loss": 0.2099,
       "step": 1277
     },
     {
-      "epoch": 1.2568897637795275,
-      "grad_norm": 7.275899410247803,
+      "epoch": 1.2578740157480315,
+      "grad_norm": 10.596028327941895,
       "learning_rate": 9.585441978915008e-07,
-      "loss": 0.2408,
+      "loss": 0.4021,
       "step": 1278
     },
     {
-      "epoch": 1.2578740157480315,
-      "grad_norm": 12.173895835876465,
+      "epoch": 1.2588582677165354,
+      "grad_norm": 15.246892929077148,
       "learning_rate": 9.582375057881785e-07,
-      "loss": 0.4633,
+      "loss": 1.002,
       "step": 1279
     },
     {
-      "epoch": 1.2588582677165354,
-      "grad_norm": 12.677711486816406,
+      "epoch": 1.2598425196850394,
+      "grad_norm": 11.348398208618164,
       "learning_rate": 9.579297330067476e-07,
-      "loss": 0.9162,
+      "loss": 0.5184,
       "step": 1280
     },
     {
-      "epoch": 1.2598425196850394,
-      "grad_norm": 17.700607299804688,
+      "epoch": 1.2608267716535433,
+      "grad_norm": 15.725797653198242,
       "learning_rate": 9.576208802756888e-07,
-      "loss": 1.0861,
+      "loss": 0.9787,
       "step": 1281
     },
     {
-      "epoch": 1.2608267716535433,
-      "grad_norm": 10.266142845153809,
+      "epoch": 1.2618110236220472,
+      "grad_norm": 17.523651123046875,
       "learning_rate": 9.5731094832604e-07,
-      "loss": 0.5915,
+      "loss": 1.0499,
       "step": 1282
     },
     {
-      "epoch": 1.2618110236220472,
-      "grad_norm": 20.697246551513672,
+      "epoch": 1.2627952755905512,
+      "grad_norm": 13.885104179382324,
       "learning_rate": 9.569999378913925e-07,
-      "loss": 0.9804,
+      "loss": 0.7918,
       "step": 1283
     },
     {
-      "epoch": 1.2627952755905512,
-      "grad_norm": 12.583255767822266,
+      "epoch": 1.263779527559055,
+      "grad_norm": 16.724008560180664,
       "learning_rate": 9.566878497078914e-07,
-      "loss": 0.7697,
+      "loss": 0.8564,
       "step": 1284
     },
     {
-      "epoch": 1.263779527559055,
-      "grad_norm": 12.262706756591797,
+      "epoch": 1.264763779527559,
+      "grad_norm": 11.258442878723145,
       "learning_rate": 9.563746845142326e-07,
-      "loss": 0.6647,
+      "loss": 0.7226,
       "step": 1285
     },
     {
-      "epoch": 1.264763779527559,
-      "grad_norm": 10.353370666503906,
+      "epoch": 1.265748031496063,
+      "grad_norm": 15.972453117370605,
       "learning_rate": 9.5606044305166e-07,
-      "loss": 0.5544,
+      "loss": 0.7506,
       "step": 1286
     },
     {
-      "epoch": 1.265748031496063,
-      "grad_norm": 9.361234664916992,
+      "epoch": 1.266732283464567,
+      "grad_norm": 14.38155460357666,
       "learning_rate": 9.557451260639665e-07,
-      "loss": 0.6577,
+      "loss": 0.7241,
       "step": 1287
     },
     {
-      "epoch": 1.266732283464567,
-      "grad_norm": 9.571710586547852,
+      "epoch": 1.2677165354330708,
+      "grad_norm": 8.162917137145996,
       "learning_rate": 9.5542873429749e-07,
-      "loss": 0.4959,
+      "loss": 0.3387,
       "step": 1288
     },
     {
-      "epoch": 1.2677165354330708,
-      "grad_norm": 18.31670570373535,
+      "epoch": 1.2687007874015748,
+      "grad_norm": 15.059714317321777,
       "learning_rate": 9.551112685011127e-07,
-      "loss": 0.8166,
+      "loss": 1.0883,
       "step": 1289
     },
     {
-      "epoch": 1.2687007874015748,
-      "grad_norm": 10.438338279724121,
+      "epoch": 1.2696850393700787,
+      "grad_norm": 15.431548118591309,
       "learning_rate": 9.547927294262583e-07,
-      "loss": 0.7087,
+      "loss": 1.0991,
       "step": 1290
     },
     {
-      "epoch": 1.2696850393700787,
-      "grad_norm": 9.65834903717041,
+      "epoch": 1.2706692913385826,
+      "grad_norm": 13.569391250610352,
       "learning_rate": 9.544731178268913e-07,
-      "loss": 0.5905,
+      "loss": 0.7014,
       "step": 1291
     },
     {
-      "epoch": 1.2706692913385826,
-      "grad_norm": 23.963003158569336,
+      "epoch": 1.2716535433070866,
+      "grad_norm": 10.554574966430664,
       "learning_rate": 9.54152434459515e-07,
-      "loss": 1.406,
+      "loss": 0.4188,
       "step": 1292
     },
     {
-      "epoch": 1.2716535433070866,
-      "grad_norm": 23.614809036254883,
+      "epoch": 1.2726377952755905,
+      "grad_norm": 11.29153823852539,
       "learning_rate": 9.538306800831695e-07,
-      "loss": 1.1811,
+      "loss": 0.5075,
       "step": 1293
     },
     {
-      "epoch": 1.2726377952755905,
-      "grad_norm": 10.113353729248047,
+      "epoch": 1.2736220472440944,
+      "grad_norm": 15.637104034423828,
       "learning_rate": 9.535078554594293e-07,
-      "loss": 0.6425,
+      "loss": 0.9734,
       "step": 1294
     },
     {
-      "epoch": 1.2736220472440944,
-      "grad_norm": 16.463342666625977,
+      "epoch": 1.2746062992125984,
+      "grad_norm": 14.084452629089355,
       "learning_rate": 9.531839613524029e-07,
-      "loss": 0.9818,
+      "loss": 0.9324,
       "step": 1295
     },
     {
-      "epoch": 1.2746062992125984,
-      "grad_norm": 13.193273544311523,
+      "epoch": 1.2755905511811023,
+      "grad_norm": 12.533185005187988,
       "learning_rate": 9.528589985287297e-07,
-      "loss": 0.8521,
+      "loss": 0.6765,
       "step": 1296
     },
     {
-      "epoch": 1.2755905511811023,
-      "grad_norm": 11.638998031616211,
+      "epoch": 1.2765748031496063,
+      "grad_norm": 11.510200500488281,
       "learning_rate": 9.525329677575787e-07,
-      "loss": 0.6998,
+      "loss": 0.4597,
       "step": 1297
     },
     {
-      "epoch": 1.2765748031496063,
-      "grad_norm": 15.33539867401123,
+      "epoch": 1.2775590551181102,
+      "grad_norm": 16.040021896362305,
       "learning_rate": 9.522058698106475e-07,
-      "loss": 0.8087,
+      "loss": 0.8729,
       "step": 1298
     },
     {
-      "epoch": 1.2775590551181102,
-      "grad_norm": 11.569411277770996,
+      "epoch": 1.2785433070866141,
+      "grad_norm": 8.879847526550293,
       "learning_rate": 9.518777054621582e-07,
-      "loss": 0.6473,
+      "loss": 0.4972,
       "step": 1299
     },
     {
-      "epoch": 1.2785433070866141,
-      "grad_norm": 13.36752700805664,
+      "epoch": 1.279527559055118,
+      "grad_norm": 9.760492324829102,
       "learning_rate": 9.515484754888582e-07,
-      "loss": 1.2283,
+      "loss": 0.6675,
       "step": 1300
     },
     {
-      "epoch": 1.279527559055118,
-      "grad_norm": 11.46446418762207,
+      "epoch": 1.280511811023622,
+      "grad_norm": 13.019458770751953,
       "learning_rate": 9.512181806700167e-07,
-      "loss": 0.7829,
+      "loss": 0.7899,
       "step": 1301
     },
     {
-      "epoch": 1.280511811023622,
-      "grad_norm": 12.435479164123535,
+      "epoch": 1.281496062992126,
+      "grad_norm": 10.449447631835938,
       "learning_rate": 9.508868217874233e-07,
-      "loss": 0.7921,
+      "loss": 0.4958,
       "step": 1302
     },
     {
-      "epoch": 1.281496062992126,
-      "grad_norm": 8.405648231506348,
+      "epoch": 1.2824803149606299,
+      "grad_norm": 9.376012802124023,
       "learning_rate": 9.505543996253865e-07,
-      "loss": 0.3911,
+      "loss": 0.6902,
       "step": 1303
     },
     {
-      "epoch": 1.2824803149606299,
-      "grad_norm": 9.165940284729004,
+      "epoch": 1.2834645669291338,
+      "grad_norm": 14.257211685180664,
       "learning_rate": 9.502209149707312e-07,
-      "loss": 0.4446,
+      "loss": 0.9887,
       "step": 1304
     },
     {
-      "epoch": 1.2834645669291338,
-      "grad_norm": 9.730753898620605,
+      "epoch": 1.2844488188976377,
+      "grad_norm": 12.741817474365234,
       "learning_rate": 9.498863686127971e-07,
-      "loss": 0.53,
+      "loss": 0.9335,
       "step": 1305
     },
     {
-      "epoch": 1.2844488188976377,
-      "grad_norm": 13.173008918762207,
+      "epoch": 1.2854330708661417,
+      "grad_norm": 16.605112075805664,
       "learning_rate": 9.495507613434374e-07,
-      "loss": 0.9651,
+      "loss": 1.4299,
       "step": 1306
     },
     {
-      "epoch": 1.2854330708661417,
-      "grad_norm": 11.323593139648438,
+      "epoch": 1.2864173228346456,
+      "grad_norm": 12.681513786315918,
       "learning_rate": 9.492140939570159e-07,
-      "loss": 0.8219,
+      "loss": 0.7355,
       "step": 1307
     },
     {
-      "epoch": 1.2864173228346456,
-      "grad_norm": 12.519793510437012,
+      "epoch": 1.2874015748031495,
+      "grad_norm": 12.397991180419922,
       "learning_rate": 9.488763672504058e-07,
-      "loss": 0.7061,
+      "loss": 0.6879,
       "step": 1308
     },
     {
-      "epoch": 1.2874015748031495,
-      "grad_norm": 8.543204307556152,
+      "epoch": 1.2883858267716535,
+      "grad_norm": 14.25597095489502,
       "learning_rate": 9.485375820229876e-07,
-      "loss": 0.493,
+      "loss": 0.8728,
       "step": 1309
     },
     {
-      "epoch": 1.2883858267716535,
-      "grad_norm": 17.205244064331055,
+      "epoch": 1.2893700787401574,
+      "grad_norm": 17.068870544433594,
       "learning_rate": 9.481977390766475e-07,
-      "loss": 1.315,
+      "loss": 1.3413,
       "step": 1310
     },
     {
-      "epoch": 1.2893700787401574,
-      "grad_norm": 18.341398239135742,
+      "epoch": 1.2903543307086613,
+      "grad_norm": 8.730855941772461,
       "learning_rate": 9.47856839215775e-07,
-      "loss": 1.4424,
+      "loss": 0.4279,
       "step": 1311
     },
     {
-      "epoch": 1.2903543307086613,
-      "grad_norm": 13.18166446685791,
+      "epoch": 1.2913385826771653,
+      "grad_norm": 15.96691608428955,
       "learning_rate": 9.475148832472613e-07,
-      "loss": 0.6492,
+      "loss": 1.1984,
       "step": 1312
     },
     {
-      "epoch": 1.2913385826771653,
-      "grad_norm": 20.222919464111328,
+      "epoch": 1.2923228346456692,
+      "grad_norm": 10.096443176269531,
       "learning_rate": 9.471718719804975e-07,
-      "loss": 1.3142,
+      "loss": 0.5541,
       "step": 1313
     },
     {
-      "epoch": 1.2923228346456692,
-      "grad_norm": 13.445477485656738,
+      "epoch": 1.2933070866141732,
+      "grad_norm": 10.184749603271484,
       "learning_rate": 9.468278062273726e-07,
-      "loss": 0.8536,
+      "loss": 0.6735,
       "step": 1314
     },
     {
-      "epoch": 1.2933070866141732,
-      "grad_norm": 11.16597843170166,
+      "epoch": 1.294291338582677,
+      "grad_norm": 11.718093872070312,
       "learning_rate": 9.464826868022709e-07,
-      "loss": 0.6758,
+      "loss": 0.6738,
       "step": 1315
     },
     {
-      "epoch": 1.294291338582677,
-      "grad_norm": 16.122098922729492,
+      "epoch": 1.295275590551181,
+      "grad_norm": 12.755319595336914,
       "learning_rate": 9.461365145220715e-07,
-      "loss": 1.0877,
+      "loss": 0.9851,
       "step": 1316
     },
     {
-      "epoch": 1.295275590551181,
-      "grad_norm": 19.477783203125,
+      "epoch": 1.296259842519685,
+      "grad_norm": 13.120318412780762,
       "learning_rate": 9.457892902061452e-07,
-      "loss": 1.2856,
+      "loss": 0.8041,
       "step": 1317
     },
     {
-      "epoch": 1.296259842519685,
-      "grad_norm": 16.64370346069336,
+      "epoch": 1.297244094488189,
+      "grad_norm": 12.807595252990723,
       "learning_rate": 9.454410146763525e-07,
-      "loss": 1.1316,
+      "loss": 0.9031,
       "step": 1318
     },
     {
-      "epoch": 1.297244094488189,
-      "grad_norm": 16.68716049194336,
+      "epoch": 1.298228346456693,
+      "grad_norm": 10.72183895111084,
       "learning_rate": 9.450916887570429e-07,
-      "loss": 1.137,
+      "loss": 0.7003,
       "step": 1319
     },
     {
-      "epoch": 1.298228346456693,
-      "grad_norm": 10.482457160949707,
+      "epoch": 1.2992125984251968,
+      "grad_norm": 8.036816596984863,
       "learning_rate": 9.447413132750515e-07,
-      "loss": 0.6438,
+      "loss": 0.4731,
       "step": 1320
     },
     {
-      "epoch": 1.2992125984251968,
-      "grad_norm": 10.563558578491211,
+      "epoch": 1.300196850393701,
+      "grad_norm": 13.362663269042969,
       "learning_rate": 9.443898890596977e-07,
-      "loss": 0.5434,
+      "loss": 0.863,
       "step": 1321
     },
     {
-      "epoch": 1.300196850393701,
-      "grad_norm": 20.146427154541016,
+      "epoch": 1.3011811023622046,
+      "grad_norm": 16.159162521362305,
       "learning_rate": 9.440374169427834e-07,
-      "loss": 1.117,
+      "loss": 1.4337,
       "step": 1322
     },
     {
-      "epoch": 1.3011811023622046,
-      "grad_norm": 17.821557998657227,
+      "epoch": 1.3021653543307088,
+      "grad_norm": 10.059019088745117,
       "learning_rate": 9.43683897758591e-07,
-      "loss": 1.0887,
+      "loss": 0.7687,
       "step": 1323
     },
     {
-      "epoch": 1.3021653543307088,
-      "grad_norm": 17.06334686279297,
+      "epoch": 1.3031496062992125,
+      "grad_norm": 10.327445030212402,
       "learning_rate": 9.433293323438807e-07,
-      "loss": 0.9753,
+      "loss": 0.5768,
       "step": 1324
     },
     {
-      "epoch": 1.3031496062992125,
-      "grad_norm": 14.24149227142334,
+      "epoch": 1.3041338582677167,
+      "grad_norm": 10.336861610412598,
       "learning_rate": 9.429737215378894e-07,
-      "loss": 0.7109,
+      "loss": 0.6752,
       "step": 1325
     },
     {
-      "epoch": 1.3041338582677167,
-      "grad_norm": 16.65839958190918,
+      "epoch": 1.3051181102362204,
+      "grad_norm": 8.284307479858398,
       "learning_rate": 9.426170661823285e-07,
-      "loss": 1.1552,
+      "loss": 0.4609,
       "step": 1326
     },
     {
-      "epoch": 1.3051181102362204,
-      "grad_norm": 13.389101028442383,
+      "epoch": 1.3061023622047245,
+      "grad_norm": 8.756792068481445,
       "learning_rate": 9.422593671213816e-07,
-      "loss": 0.8534,
+      "loss": 0.5695,
       "step": 1327
     },
     {
-      "epoch": 1.3061023622047245,
-      "grad_norm": 10.902589797973633,
+      "epoch": 1.3070866141732282,
+      "grad_norm": 13.238386154174805,
       "learning_rate": 9.419006252017027e-07,
-      "loss": 0.7198,
+      "loss": 1.0559,
       "step": 1328
     },
     {
-      "epoch": 1.3070866141732282,
-      "grad_norm": 13.474040985107422,
+      "epoch": 1.3080708661417324,
+      "grad_norm": 11.581672668457031,
       "learning_rate": 9.415408412724141e-07,
-      "loss": 1.2167,
+      "loss": 0.5529,
       "step": 1329
     },
     {
-      "epoch": 1.3080708661417324,
-      "grad_norm": 11.535694122314453,
+      "epoch": 1.3090551181102361,
+      "grad_norm": 14.326388359069824,
       "learning_rate": 9.411800161851048e-07,
-      "loss": 0.979,
+      "loss": 0.8975,
       "step": 1330
     },
     {
-      "epoch": 1.3090551181102361,
-      "grad_norm": 15.960291862487793,
+      "epoch": 1.3100393700787403,
+      "grad_norm": 10.52392578125,
       "learning_rate": 9.408181507938279e-07,
-      "loss": 1.1667,
+      "loss": 0.4789,
       "step": 1331
     },
     {
-      "epoch": 1.3100393700787403,
-      "grad_norm": 10.989014625549316,
+      "epoch": 1.311023622047244,
+      "grad_norm": 16.630456924438477,
       "learning_rate": 9.404552459550988e-07,
-      "loss": 0.5707,
+      "loss": 1.0715,
       "step": 1332
     },
     {
-      "epoch": 1.311023622047244,
-      "grad_norm": 12.71398639678955,
+      "epoch": 1.3120078740157481,
+      "grad_norm": 13.835437774658203,
       "learning_rate": 9.400913025278937e-07,
-      "loss": 0.8196,
+      "loss": 0.6386,
       "step": 1333
     },
     {
-      "epoch": 1.3120078740157481,
-      "grad_norm": 15.065571784973145,
+      "epoch": 1.3129921259842519,
+      "grad_norm": 12.128693580627441,
       "learning_rate": 9.397263213736464e-07,
-      "loss": 1.1255,
+      "loss": 0.7176,
       "step": 1334
     },
     {
-      "epoch": 1.3129921259842519,
-      "grad_norm": 9.93442440032959,
+      "epoch": 1.313976377952756,
+      "grad_norm": 11.691065788269043,
       "learning_rate": 9.393603033562471e-07,
-      "loss": 0.7614,
+      "loss": 0.4927,
       "step": 1335
     },
     {
-      "epoch": 1.313976377952756,
-      "grad_norm": 10.066417694091797,
+      "epoch": 1.3149606299212597,
+      "grad_norm": 13.579904556274414,
       "learning_rate": 9.389932493420407e-07,
-      "loss": 0.616,
+      "loss": 1.1553,
       "step": 1336
     },
     {
-      "epoch": 1.3149606299212597,
-      "grad_norm": 11.152246475219727,
+      "epoch": 1.3159448818897639,
+      "grad_norm": 16.762948989868164,
       "learning_rate": 9.386251601998239e-07,
-      "loss": 0.8282,
+      "loss": 1.0901,
       "step": 1337
     },
     {
-      "epoch": 1.3159448818897639,
-      "grad_norm": 13.376919746398926,
+      "epoch": 1.3169291338582676,
+      "grad_norm": 15.133347511291504,
       "learning_rate": 9.38256036800843e-07,
-      "loss": 0.9118,
+      "loss": 0.7623,
       "step": 1338
     },
     {
-      "epoch": 1.3169291338582676,
-      "grad_norm": 8.617725372314453,
+      "epoch": 1.3179133858267718,
+      "grad_norm": 11.118714332580566,
       "learning_rate": 9.378858800187935e-07,
-      "loss": 0.578,
+      "loss": 0.5397,
       "step": 1339
     },
     {
-      "epoch": 1.3179133858267718,
-      "grad_norm": 9.212502479553223,
+      "epoch": 1.3188976377952755,
+      "grad_norm": 11.781917572021484,
       "learning_rate": 9.375146907298157e-07,
-      "loss": 0.5914,
+      "loss": 0.6808,
       "step": 1340
     },
     {
-      "epoch": 1.3188976377952755,
-      "grad_norm": 9.490756034851074,
+      "epoch": 1.3198818897637796,
+      "grad_norm": 10.816554069519043,
       "learning_rate": 9.371424698124945e-07,
-      "loss": 0.612,
+      "loss": 0.4346,
       "step": 1341
     },
     {
-      "epoch": 1.3198818897637796,
-      "grad_norm": 10.681628227233887,
+      "epoch": 1.3208661417322833,
+      "grad_norm": 12.161731719970703,
       "learning_rate": 9.367692181478564e-07,
-      "loss": 0.5543,
+      "loss": 0.8217,
       "step": 1342
     },
     {
-      "epoch": 1.3208661417322833,
-      "grad_norm": 9.452579498291016,
+      "epoch": 1.3218503937007875,
+      "grad_norm": 14.671911239624023,
       "learning_rate": 9.363949366193673e-07,
-      "loss": 0.4898,
+      "loss": 0.9094,
       "step": 1343
     },
     {
-      "epoch": 1.3218503937007875,
-      "grad_norm": 11.30627727508545,
+      "epoch": 1.3228346456692912,
+      "grad_norm": 8.16436767578125,
       "learning_rate": 9.360196261129313e-07,
-      "loss": 0.7363,
+      "loss": 0.4829,
       "step": 1344
     },
     {
-      "epoch": 1.3228346456692912,
-      "grad_norm": 8.573918342590332,
+      "epoch": 1.3238188976377954,
+      "grad_norm": 11.594481468200684,
       "learning_rate": 9.356432875168876e-07,
-      "loss": 0.3767,
+      "loss": 0.706,
       "step": 1345
     },
     {
-      "epoch": 1.3238188976377954,
-      "grad_norm": 13.486555099487305,
+      "epoch": 1.324803149606299,
+      "grad_norm": 11.576638221740723,
       "learning_rate": 9.352659217220091e-07,
-      "loss": 0.4314,
+      "loss": 0.7423,
       "step": 1346
     },
     {
-      "epoch": 1.324803149606299,
-      "grad_norm": 10.113042831420898,
+      "epoch": 1.3257874015748032,
+      "grad_norm": 14.552374839782715,
       "learning_rate": 9.348875296215001e-07,
-      "loss": 0.4874,
+      "loss": 0.864,
       "step": 1347
     },
     {
-      "epoch": 1.3257874015748032,
-      "grad_norm": 9.970690727233887,
+      "epoch": 1.326771653543307,
+      "grad_norm": 9.43635082244873,
       "learning_rate": 9.345081121109935e-07,
-      "loss": 0.467,
+      "loss": 0.3376,
       "step": 1348
     },
     {
-      "epoch": 1.326771653543307,
-      "grad_norm": 10.185608863830566,
+      "epoch": 1.327755905511811,
+      "grad_norm": 11.919918060302734,
       "learning_rate": 9.341276700885501e-07,
-      "loss": 0.4353,
+      "loss": 0.6119,
       "step": 1349
     },
     {
-      "epoch": 1.327755905511811,
-      "grad_norm": 9.48005199432373,
+      "epoch": 1.328740157480315,
+      "grad_norm": 10.393912315368652,
       "learning_rate": 9.337462044546548e-07,
-      "loss": 0.6328,
+      "loss": 0.5668,
       "step": 1350
     },
     {
-      "epoch": 1.328740157480315,
-      "grad_norm": 19.29398536682129,
+      "epoch": 1.329724409448819,
+      "grad_norm": 15.983757019042969,
       "learning_rate": 9.333637161122166e-07,
-      "loss": 0.9637,
+      "loss": 0.702,
       "step": 1351
     },
     {
-      "epoch": 1.329724409448819,
-      "grad_norm": 12.811509132385254,
+      "epoch": 1.330708661417323,
+      "grad_norm": 11.626129150390625,
       "learning_rate": 9.329802059665637e-07,
-      "loss": 0.9188,
+      "loss": 0.5933,
       "step": 1352
     },
     {
-      "epoch": 1.330708661417323,
-      "grad_norm": 11.238883018493652,
+      "epoch": 1.3316929133858268,
+      "grad_norm": 9.455058097839355,
       "learning_rate": 9.325956749254437e-07,
-      "loss": 0.5365,
+      "loss": 0.4192,
       "step": 1353
     },
     {
-      "epoch": 1.3316929133858268,
-      "grad_norm": 9.164770126342773,
+      "epoch": 1.3326771653543308,
+      "grad_norm": 14.05883502960205,
       "learning_rate": 9.322101238990206e-07,
-      "loss": 0.2543,
+      "loss": 1.047,
       "step": 1354
     },
     {
-      "epoch": 1.3326771653543308,
-      "grad_norm": 13.3213472366333,
+      "epoch": 1.3336614173228347,
+      "grad_norm": 11.624855995178223,
       "learning_rate": 9.318235537998726e-07,
-      "loss": 0.8462,
+      "loss": 0.7922,
       "step": 1355
     },
     {
-      "epoch": 1.3336614173228347,
-      "grad_norm": 12.852092742919922,
+      "epoch": 1.3346456692913387,
+      "grad_norm": 12.157319068908691,
       "learning_rate": 9.314359655429897e-07,
-      "loss": 0.824,
+      "loss": 1.0565,
       "step": 1356
     },
     {
-      "epoch": 1.3346456692913387,
-      "grad_norm": 10.420133590698242,
+      "epoch": 1.3356299212598426,
+      "grad_norm": 14.961847305297852,
       "learning_rate": 9.310473600457721e-07,
-      "loss": 0.6009,
+      "loss": 0.6672,
       "step": 1357
     },
     {
-      "epoch": 1.3356299212598426,
-      "grad_norm": 13.898282051086426,
+      "epoch": 1.3366141732283465,
+      "grad_norm": 14.245471000671387,
       "learning_rate": 9.306577382280277e-07,
-      "loss": 0.8251,
+      "loss": 0.993,
       "step": 1358
     },
     {
-      "epoch": 1.3366141732283465,
-      "grad_norm": 8.93106746673584,
+      "epoch": 1.3375984251968505,
+      "grad_norm": 10.663938522338867,
       "learning_rate": 9.3026710101197e-07,
-      "loss": 0.5137,
+      "loss": 0.6533,
       "step": 1359
     },
     {
-      "epoch": 1.3375984251968505,
-      "grad_norm": 20.12264633178711,
+      "epoch": 1.3385826771653544,
+      "grad_norm": 13.007502555847168,
       "learning_rate": 9.298754493222155e-07,
-      "loss": 1.1384,
+      "loss": 1.1159,
       "step": 1360
     },
     {
-      "epoch": 1.3385826771653544,
-      "grad_norm": 6.826462268829346,
+      "epoch": 1.3395669291338583,
+      "grad_norm": 10.378064155578613,
       "learning_rate": 9.294827840857827e-07,
-      "loss": 0.3927,
+      "loss": 0.6079,
       "step": 1361
     },
     {
-      "epoch": 1.3395669291338583,
-      "grad_norm": 21.808618545532227,
+      "epoch": 1.3405511811023623,
+      "grad_norm": 10.60908031463623,
       "learning_rate": 9.290891062320884e-07,
-      "loss": 1.2509,
+      "loss": 0.6253,
       "step": 1362
     },
     {
-      "epoch": 1.3405511811023623,
-      "grad_norm": 17.233945846557617,
+      "epoch": 1.3415354330708662,
+      "grad_norm": 14.619140625,
       "learning_rate": 9.286944166929466e-07,
-      "loss": 1.0992,
+      "loss": 0.9925,
       "step": 1363
     },
     {
-      "epoch": 1.3415354330708662,
-      "grad_norm": 16.560678482055664,
+      "epoch": 1.3425196850393701,
+      "grad_norm": 9.355093002319336,
       "learning_rate": 9.282987164025658e-07,
-      "loss": 1.0847,
+      "loss": 0.4941,
       "step": 1364
     },
     {
-      "epoch": 1.3425196850393701,
-      "grad_norm": 13.578652381896973,
+      "epoch": 1.343503937007874,
+      "grad_norm": 5.725484848022461,
       "learning_rate": 9.279020062975465e-07,
-      "loss": 1.1022,
+      "loss": 0.2237,
       "step": 1365
     },
     {
-      "epoch": 1.343503937007874,
-      "grad_norm": 16.567989349365234,
+      "epoch": 1.344488188976378,
+      "grad_norm": 10.741724014282227,
       "learning_rate": 9.275042873168802e-07,
-      "loss": 0.8804,
+      "loss": 0.5003,
       "step": 1366
     },
     {
-      "epoch": 1.344488188976378,
-      "grad_norm": 13.353636741638184,
+      "epoch": 1.345472440944882,
+      "grad_norm": 15.184252738952637,
       "learning_rate": 9.271055604019458e-07,
-      "loss": 0.8076,
+      "loss": 0.8567,
       "step": 1367
     },
     {
-      "epoch": 1.345472440944882,
-      "grad_norm": 11.190862655639648,
+      "epoch": 1.3464566929133859,
+      "grad_norm": 11.08669662475586,
       "learning_rate": 9.267058264965076e-07,
-      "loss": 0.7171,
+      "loss": 0.5952,
       "step": 1368
     },
     {
-      "epoch": 1.3464566929133859,
-      "grad_norm": 11.54306411743164,
+      "epoch": 1.3474409448818898,
+      "grad_norm": 15.194894790649414,
       "learning_rate": 9.263050865467143e-07,
-      "loss": 0.7283,
+      "loss": 0.785,
       "step": 1369
     },
     {
-      "epoch": 1.3474409448818898,
-      "grad_norm": 12.018163681030273,
+      "epoch": 1.3484251968503937,
+      "grad_norm": 12.53934383392334,
       "learning_rate": 9.259033415010952e-07,
-      "loss": 0.7313,
+      "loss": 1.0094,
       "step": 1370
     },
     {
-      "epoch": 1.3484251968503937,
-      "grad_norm": 14.016282081604004,
+      "epoch": 1.3494094488188977,
+      "grad_norm": 10.500348091125488,
       "learning_rate": 9.255005923105587e-07,
-      "loss": 0.9396,
+      "loss": 0.8874,
       "step": 1371
     },
     {
-      "epoch": 1.3494094488188977,
-      "grad_norm": 12.047124862670898,
+      "epoch": 1.3503937007874016,
+      "grad_norm": 12.779960632324219,
       "learning_rate": 9.250968399283902e-07,
-      "loss": 0.661,
+      "loss": 0.7712,
       "step": 1372
     },
     {
-      "epoch": 1.3503937007874016,
-      "grad_norm": 21.686264038085938,
+      "epoch": 1.3513779527559056,
+      "grad_norm": 13.599752426147461,
       "learning_rate": 9.24692085310249e-07,
-      "loss": 1.8785,
+      "loss": 0.8822,
       "step": 1373
     },
     {
-      "epoch": 1.3513779527559056,
-      "grad_norm": 13.831476211547852,
+      "epoch": 1.3523622047244095,
+      "grad_norm": 15.221623420715332,
       "learning_rate": 9.242863294141675e-07,
-      "loss": 0.4109,
+      "loss": 1.0364,
       "step": 1374
     },
     {
-      "epoch": 1.3523622047244095,
-      "grad_norm": 10.79392147064209,
+      "epoch": 1.3533464566929134,
+      "grad_norm": 11.4656400680542,
       "learning_rate": 9.238795732005475e-07,
-      "loss": 0.7084,
+      "loss": 0.5006,
       "step": 1375
     },
     {
-      "epoch": 1.3533464566929134,
-      "grad_norm": 9.851486206054688,
+      "epoch": 1.3543307086614174,
+      "grad_norm": 14.487397193908691,
       "learning_rate": 9.234718176321586e-07,
-      "loss": 0.4369,
+      "loss": 1.1828,
       "step": 1376
     },
     {
-      "epoch": 1.3543307086614174,
-      "grad_norm": 11.24117660522461,
+      "epoch": 1.3553149606299213,
+      "grad_norm": 12.548595428466797,
       "learning_rate": 9.230630636741358e-07,
-      "loss": 0.7483,
+      "loss": 0.4488,
       "step": 1377
     },
     {
-      "epoch": 1.3543307086614174,
-      "eval_Qnli-dev_cosine_accuracy": 0.671875,
-      "eval_Qnli-dev_cosine_accuracy_threshold": 0.8206880688667297,
-      "eval_Qnli-dev_cosine_ap": 0.7020016169571128,
-      "eval_Qnli-dev_cosine_f1": 0.6984667802385007,
-      "eval_Qnli-dev_cosine_f1_threshold": 0.7731959819793701,
-      "eval_Qnli-dev_cosine_precision": 0.584045584045584,
-      "eval_Qnli-dev_cosine_recall": 0.8686440677966102,
-      "eval_Qnli-dev_dot_accuracy": 0.669921875,
-      "eval_Qnli-dev_dot_accuracy_threshold": 637.1887817382812,
-      "eval_Qnli-dev_dot_ap": 0.70064192309746,
-      "eval_Qnli-dev_dot_f1": 0.696095076400679,
-      "eval_Qnli-dev_dot_f1_threshold": 588.2445068359375,
-      "eval_Qnli-dev_dot_precision": 0.5807365439093485,
-      "eval_Qnli-dev_dot_recall": 0.8686440677966102,
-      "eval_Qnli-dev_euclidean_accuracy": 0.671875,
-      "eval_Qnli-dev_euclidean_accuracy_threshold": 16.55626106262207,
-      "eval_Qnli-dev_euclidean_ap": 0.7017081005305794,
-      "eval_Qnli-dev_euclidean_f1": 0.6958041958041957,
-      "eval_Qnli-dev_euclidean_f1_threshold": 18.35430335998535,
-      "eval_Qnli-dev_euclidean_precision": 0.5922619047619048,
-      "eval_Qnli-dev_euclidean_recall": 0.8432203389830508,
-      "eval_Qnli-dev_manhattan_accuracy": 0.693359375,
-      "eval_Qnli-dev_manhattan_accuracy_threshold": 337.3518981933594,
-      "eval_Qnli-dev_manhattan_ap": 0.7143798529397983,
-      "eval_Qnli-dev_manhattan_f1": 0.6998284734133791,
-      "eval_Qnli-dev_manhattan_f1_threshold": 390.21343994140625,
-      "eval_Qnli-dev_manhattan_precision": 0.5878962536023055,
-      "eval_Qnli-dev_manhattan_recall": 0.864406779661017,
-      "eval_Qnli-dev_max_accuracy": 0.693359375,
-      "eval_Qnli-dev_max_accuracy_threshold": 637.1887817382812,
-      "eval_Qnli-dev_max_ap": 0.7143798529397983,
-      "eval_Qnli-dev_max_f1": 0.6998284734133791,
-      "eval_Qnli-dev_max_f1_threshold": 588.2445068359375,
-      "eval_Qnli-dev_max_precision": 0.5922619047619048,
-      "eval_Qnli-dev_max_recall": 0.8686440677966102,
-      "eval_allNLI-dev_cosine_accuracy": 0.689453125,
-      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9173126816749573,
-      "eval_allNLI-dev_cosine_ap": 0.5006976649965474,
-      "eval_allNLI-dev_cosine_f1": 0.5413223140495869,
-      "eval_allNLI-dev_cosine_f1_threshold": 0.848089337348938,
-      "eval_allNLI-dev_cosine_precision": 0.4212218649517685,
-      "eval_allNLI-dev_cosine_recall": 0.7572254335260116,
-      "eval_allNLI-dev_dot_accuracy": 0.69140625,
-      "eval_allNLI-dev_dot_accuracy_threshold": 701.9290771484375,
-      "eval_allNLI-dev_dot_ap": 0.5021573167737639,
-      "eval_allNLI-dev_dot_f1": 0.5424430641821947,
-      "eval_allNLI-dev_dot_f1_threshold": 649.3282470703125,
-      "eval_allNLI-dev_dot_precision": 0.42258064516129035,
-      "eval_allNLI-dev_dot_recall": 0.7572254335260116,
-      "eval_allNLI-dev_euclidean_accuracy": 0.6875,
-      "eval_allNLI-dev_euclidean_accuracy_threshold": 10.926703453063965,
-      "eval_allNLI-dev_euclidean_ap": 0.5003891082478613,
-      "eval_allNLI-dev_euclidean_f1": 0.5413223140495869,
-      "eval_allNLI-dev_euclidean_f1_threshold": 15.256593704223633,
-      "eval_allNLI-dev_euclidean_precision": 0.4212218649517685,
-      "eval_allNLI-dev_euclidean_recall": 0.7572254335260116,
-      "eval_allNLI-dev_manhattan_accuracy": 0.6875,
-      "eval_allNLI-dev_manhattan_accuracy_threshold": 229.8043212890625,
-      "eval_allNLI-dev_manhattan_ap": 0.4991680725185854,
-      "eval_allNLI-dev_manhattan_f1": 0.5409090909090909,
-      "eval_allNLI-dev_manhattan_f1_threshold": 306.140380859375,
-      "eval_allNLI-dev_manhattan_precision": 0.44569288389513106,
-      "eval_allNLI-dev_manhattan_recall": 0.6878612716763006,
-      "eval_allNLI-dev_max_accuracy": 0.69140625,
-      "eval_allNLI-dev_max_accuracy_threshold": 701.9290771484375,
-      "eval_allNLI-dev_max_ap": 0.5021573167737639,
-      "eval_allNLI-dev_max_f1": 0.5424430641821947,
-      "eval_allNLI-dev_max_f1_threshold": 649.3282470703125,
-      "eval_allNLI-dev_max_precision": 0.44569288389513106,
-      "eval_allNLI-dev_max_recall": 0.7572254335260116,
-      "eval_loss": 1.1422334909439087,
-      "eval_runtime": 54.1068,
-      "eval_samples_per_second": 30.754,
-      "eval_sequential_score": 0.7143798529397983,
-      "eval_steps_per_second": 0.129,
-      "eval_sts-test_pearson_cosine": 0.559231077840739,
-      "eval_sts-test_pearson_dot": 0.5569194332314316,
-      "eval_sts-test_pearson_euclidean": 0.5735407858028017,
-      "eval_sts-test_pearson_manhattan": 0.5839195979621228,
-      "eval_sts-test_pearson_max": 0.5839195979621228,
-      "eval_sts-test_spearman_cosine": 0.5765562772319613,
-      "eval_sts-test_spearman_dot": 0.575029274690764,
-      "eval_sts-test_spearman_euclidean": 0.5768639421540271,
-      "eval_sts-test_spearman_manhattan": 0.5881996156695143,
-      "eval_sts-test_spearman_max": 0.5881996156695143,
+      "epoch": 1.3553149606299213,
+      "eval_Qnli-dev_cosine_accuracy": 0.662109375,
+      "eval_Qnli-dev_cosine_accuracy_threshold": 0.8539660573005676,
+      "eval_Qnli-dev_cosine_ap": 0.6829334293773432,
+      "eval_Qnli-dev_cosine_f1": 0.6643478260869564,
+      "eval_Qnli-dev_cosine_f1_threshold": 0.7722476124763489,
+      "eval_Qnli-dev_cosine_precision": 0.5634218289085545,
+      "eval_Qnli-dev_cosine_recall": 0.809322033898305,
+      "eval_Qnli-dev_dot_accuracy": 0.6640625,
+      "eval_Qnli-dev_dot_accuracy_threshold": 650.67578125,
+      "eval_Qnli-dev_dot_ap": 0.6802062887370495,
+      "eval_Qnli-dev_dot_f1": 0.6654929577464789,
+      "eval_Qnli-dev_dot_f1_threshold": 590.8773193359375,
+      "eval_Qnli-dev_dot_precision": 0.5692771084337349,
+      "eval_Qnli-dev_dot_recall": 0.8008474576271186,
+      "eval_Qnli-dev_euclidean_accuracy": 0.662109375,
+      "eval_Qnli-dev_euclidean_accuracy_threshold": 14.918805122375488,
+      "eval_Qnli-dev_euclidean_ap": 0.683062530657187,
+      "eval_Qnli-dev_euclidean_f1": 0.6655052264808362,
+      "eval_Qnli-dev_euclidean_f1_threshold": 18.635833740234375,
+      "eval_Qnli-dev_euclidean_precision": 0.5650887573964497,
+      "eval_Qnli-dev_euclidean_recall": 0.809322033898305,
+      "eval_Qnli-dev_manhattan_accuracy": 0.662109375,
+      "eval_Qnli-dev_manhattan_accuracy_threshold": 313.5548095703125,
+      "eval_Qnli-dev_manhattan_ap": 0.6859473493159097,
+      "eval_Qnli-dev_manhattan_f1": 0.6719999999999999,
+      "eval_Qnli-dev_manhattan_f1_threshold": 409.43829345703125,
+      "eval_Qnli-dev_manhattan_precision": 0.5398457583547558,
+      "eval_Qnli-dev_manhattan_recall": 0.8898305084745762,
+      "eval_Qnli-dev_max_accuracy": 0.6640625,
+      "eval_Qnli-dev_max_accuracy_threshold": 650.67578125,
+      "eval_Qnli-dev_max_ap": 0.6859473493159097,
+      "eval_Qnli-dev_max_f1": 0.6719999999999999,
+      "eval_Qnli-dev_max_f1_threshold": 590.8773193359375,
+      "eval_Qnli-dev_max_precision": 0.5692771084337349,
+      "eval_Qnli-dev_max_recall": 0.8898305084745762,
+      "eval_allNLI-dev_cosine_accuracy": 0.6953125,
+      "eval_allNLI-dev_cosine_accuracy_threshold": 0.922626793384552,
+      "eval_allNLI-dev_cosine_ap": 0.4975335085770213,
+      "eval_allNLI-dev_cosine_f1": 0.5322896281800392,
+      "eval_allNLI-dev_cosine_f1_threshold": 0.8138325214385986,
+      "eval_allNLI-dev_cosine_precision": 0.40236686390532544,
+      "eval_allNLI-dev_cosine_recall": 0.7861271676300579,
+      "eval_allNLI-dev_dot_accuracy": 0.6953125,
+      "eval_allNLI-dev_dot_accuracy_threshold": 705.760009765625,
+      "eval_allNLI-dev_dot_ap": 0.4999639133660367,
+      "eval_allNLI-dev_dot_f1": 0.5322896281800392,
+      "eval_allNLI-dev_dot_f1_threshold": 621.754150390625,
+      "eval_allNLI-dev_dot_precision": 0.40236686390532544,
+      "eval_allNLI-dev_dot_recall": 0.7861271676300579,
+      "eval_allNLI-dev_euclidean_accuracy": 0.6953125,
+      "eval_allNLI-dev_euclidean_accuracy_threshold": 10.87257194519043,
+      "eval_allNLI-dev_euclidean_ap": 0.4976068306794622,
+      "eval_allNLI-dev_euclidean_f1": 0.5322896281800392,
+      "eval_allNLI-dev_euclidean_f1_threshold": 16.863723754882812,
+      "eval_allNLI-dev_euclidean_precision": 0.40236686390532544,
+      "eval_allNLI-dev_euclidean_recall": 0.7861271676300579,
+      "eval_allNLI-dev_manhattan_accuracy": 0.689453125,
+      "eval_allNLI-dev_manhattan_accuracy_threshold": 211.14002990722656,
+      "eval_allNLI-dev_manhattan_ap": 0.49973125387204415,
+      "eval_allNLI-dev_manhattan_f1": 0.5386313465783665,
+      "eval_allNLI-dev_manhattan_f1_threshold": 331.33514404296875,
+      "eval_allNLI-dev_manhattan_precision": 0.4357142857142857,
+      "eval_allNLI-dev_manhattan_recall": 0.7052023121387283,
+      "eval_allNLI-dev_max_accuracy": 0.6953125,
+      "eval_allNLI-dev_max_accuracy_threshold": 705.760009765625,
+      "eval_allNLI-dev_max_ap": 0.4999639133660367,
+      "eval_allNLI-dev_max_f1": 0.5386313465783665,
+      "eval_allNLI-dev_max_f1_threshold": 621.754150390625,
+      "eval_allNLI-dev_max_precision": 0.4357142857142857,
+      "eval_allNLI-dev_max_recall": 0.7861271676300579,
+      "eval_loss": 1.0092445611953735,
+      "eval_runtime": 52.2693,
+      "eval_samples_per_second": 31.835,
+      "eval_sequential_score": 0.6859473493159097,
+      "eval_steps_per_second": 0.134,
+      "eval_sts-test_pearson_cosine": 0.6105508995363205,
+      "eval_sts-test_pearson_dot": 0.6085951424647134,
+      "eval_sts-test_pearson_euclidean": 0.625154216426298,
+      "eval_sts-test_pearson_manhattan": 0.6346928228602939,
+      "eval_sts-test_pearson_max": 0.6346928228602939,
+      "eval_sts-test_spearman_cosine": 0.6250538904707419,
+      "eval_sts-test_spearman_dot": 0.6216908331771782,
+      "eval_sts-test_spearman_euclidean": 0.625315864354927,
+      "eval_sts-test_spearman_manhattan": 0.6320356242629369,
+      "eval_sts-test_spearman_max": 0.6320356242629369,
       "step": 1377
     },
     {
-      "epoch": 1.3553149606299213,
-      "grad_norm": 14.724981307983398,
+      "epoch": 1.3562992125984252,
+      "grad_norm": 9.63126277923584,
       "learning_rate": 9.226533122939772e-07,
-      "loss": 1.2169,
+      "loss": 0.4031,
       "step": 1378
     },
     {
-      "epoch": 1.3562992125984252,
-      "grad_norm": 20.81302261352539,
+      "epoch": 1.3572834645669292,
+      "grad_norm": 17.997404098510742,
       "learning_rate": 9.222425644615421e-07,
-      "loss": 1.2488,
+      "loss": 1.1215,
       "step": 1379
     },
     {
-      "epoch": 1.3572834645669292,
-      "grad_norm": 21.09238052368164,
+      "epoch": 1.358267716535433,
+      "grad_norm": 16.002639770507812,
       "learning_rate": 9.218308211490479e-07,
-      "loss": 1.641,
+      "loss": 0.685,
       "step": 1380
     },
     {
-      "epoch": 1.358267716535433,
-      "grad_norm": 11.956694602966309,
+      "epoch": 1.359251968503937,
+      "grad_norm": 13.864032745361328,
       "learning_rate": 9.214180833310684e-07,
-      "loss": 0.7456,
+      "loss": 0.9461,
       "step": 1381
     },
     {
-      "epoch": 1.359251968503937,
-      "grad_norm": 16.93556022644043,
+      "epoch": 1.360236220472441,
+      "grad_norm": 11.994694709777832,
       "learning_rate": 9.210043519845313e-07,
-      "loss": 1.1028,
+      "loss": 0.6278,
       "step": 1382
     },
     {
-      "epoch": 1.360236220472441,
-      "grad_norm": 16.941574096679688,
+      "epoch": 1.361220472440945,
+      "grad_norm": 11.811822891235352,
       "learning_rate": 9.205896280887163e-07,
-      "loss": 1.0972,
+      "loss": 0.6777,
       "step": 1383
     },
     {
-      "epoch": 1.361220472440945,
-      "grad_norm": 11.039828300476074,
+      "epoch": 1.3622047244094488,
+      "grad_norm": 11.404610633850098,
       "learning_rate": 9.201739126252518e-07,
-      "loss": 0.6608,
+      "loss": 0.9092,
       "step": 1384
     },
     {
-      "epoch": 1.3622047244094488,
-      "grad_norm": 10.035400390625,
+      "epoch": 1.3631889763779528,
+      "grad_norm": 16.141279220581055,
       "learning_rate": 9.197572065781133e-07,
-      "loss": 0.4861,
+      "loss": 1.107,
       "step": 1385
     },
     {
-      "epoch": 1.3631889763779528,
-      "grad_norm": 14.977723121643066,
+      "epoch": 1.3641732283464567,
+      "grad_norm": 13.142462730407715,
       "learning_rate": 9.193395109336216e-07,
-      "loss": 0.9011,
+      "loss": 0.8068,
       "step": 1386
     },
     {
-      "epoch": 1.3641732283464567,
-      "grad_norm": 16.243133544921875,
+      "epoch": 1.3651574803149606,
+      "grad_norm": 11.724642753601074,
       "learning_rate": 9.189208266804389e-07,
-      "loss": 1.1415,
+      "loss": 0.7613,
       "step": 1387
     },
     {
-      "epoch": 1.3651574803149606,
-      "grad_norm": 11.963306427001953,
+      "epoch": 1.3661417322834646,
+      "grad_norm": 10.146293640136719,
       "learning_rate": 9.185011548095682e-07,
-      "loss": 0.5773,
+      "loss": 0.6967,
       "step": 1388
     },
     {
-      "epoch": 1.3661417322834646,
-      "grad_norm": 11.360391616821289,
+      "epoch": 1.3671259842519685,
+      "grad_norm": 11.84849739074707,
       "learning_rate": 9.180804963143496e-07,
-      "loss": 0.6052,
+      "loss": 0.9097,
       "step": 1389
     },
     {
-      "epoch": 1.3671259842519685,
-      "grad_norm": 12.168638229370117,
+      "epoch": 1.3681102362204725,
+      "grad_norm": 11.217120170593262,
       "learning_rate": 9.176588521904588e-07,
-      "loss": 0.7691,
+      "loss": 0.8659,
       "step": 1390
     },
     {
-      "epoch": 1.3681102362204725,
-      "grad_norm": 7.507850170135498,
+      "epoch": 1.3690944881889764,
+      "grad_norm": 8.344671249389648,
       "learning_rate": 9.172362234359041e-07,
-      "loss": 0.3499,
+      "loss": 0.5029,
       "step": 1391
     },
     {
-      "epoch": 1.3690944881889764,
-      "grad_norm": 17.015483856201172,
+      "epoch": 1.3700787401574803,
+      "grad_norm": 7.371416091918945,
       "learning_rate": 9.168126110510246e-07,
-      "loss": 1.0002,
+      "loss": 0.3766,
       "step": 1392
     },
     {
-      "epoch": 1.3700787401574803,
-      "grad_norm": 11.774580001831055,
+      "epoch": 1.3710629921259843,
+      "grad_norm": 10.682507514953613,
       "learning_rate": 9.163880160384877e-07,
-      "loss": 0.8713,
+      "loss": 0.7413,
       "step": 1393
     },
     {
-      "epoch": 1.3710629921259843,
-      "grad_norm": 10.965306282043457,
+      "epoch": 1.3720472440944882,
+      "grad_norm": 9.902941703796387,
       "learning_rate": 9.159624394032864e-07,
-      "loss": 0.7215,
+      "loss": 0.6002,
       "step": 1394
     },
     {
-      "epoch": 1.3720472440944882,
-      "grad_norm": 17.416032791137695,
+      "epoch": 1.3730314960629921,
+      "grad_norm": 14.252848625183105,
       "learning_rate": 9.155358821527371e-07,
-      "loss": 0.954,
+      "loss": 1.002,
       "step": 1395
     },
     {
-      "epoch": 1.3730314960629921,
-      "grad_norm": 15.789299011230469,
+      "epoch": 1.374015748031496,
+      "grad_norm": 12.179991722106934,
       "learning_rate": 9.151083452964775e-07,
-      "loss": 1.0503,
+      "loss": 0.8959,
       "step": 1396
     },
     {
-      "epoch": 1.374015748031496,
-      "grad_norm": 12.647138595581055,
+      "epoch": 1.375,
+      "grad_norm": 13.375877380371094,
       "learning_rate": 9.146798298464642e-07,
-      "loss": 1.0006,
+      "loss": 1.0575,
       "step": 1397
     },
     {
-      "epoch": 1.375,
-      "grad_norm": 11.720510482788086,
+      "epoch": 1.375984251968504,
+      "grad_norm": 12.646140098571777,
       "learning_rate": 9.14250336816969e-07,
-      "loss": 0.8779,
+      "loss": 0.746,
       "step": 1398
     },
     {
-      "epoch": 1.375984251968504,
-      "grad_norm": 9.829345703125,
+      "epoch": 1.3769685039370079,
+      "grad_norm": 13.08265495300293,
       "learning_rate": 9.13819867224579e-07,
-      "loss": 0.3058,
+      "loss": 0.9076,
       "step": 1399
     },
     {
-      "epoch": 1.3769685039370079,
-      "grad_norm": 9.576454162597656,
+      "epoch": 1.3779527559055118,
+      "grad_norm": 13.691642761230469,
       "learning_rate": 9.133884220881917e-07,
-      "loss": 0.4527,
+      "loss": 0.8383,
       "step": 1400
     },
     {
-      "epoch": 1.3779527559055118,
-      "grad_norm": 14.037468910217285,
+      "epoch": 1.3789370078740157,
+      "grad_norm": 13.106060028076172,
       "learning_rate": 9.129560024290145e-07,
-      "loss": 1.17,
+      "loss": 0.6814,
       "step": 1401
     },
     {
-      "epoch": 1.3789370078740157,
-      "grad_norm": 8.34862232208252,
+      "epoch": 1.3799212598425197,
+      "grad_norm": 11.385546684265137,
       "learning_rate": 9.125226092705603e-07,
-      "loss": 0.4068,
+      "loss": 0.5709,
       "step": 1402
     },
     {
-      "epoch": 1.3799212598425197,
-      "grad_norm": 10.172402381896973,
+      "epoch": 1.3809055118110236,
+      "grad_norm": 12.104999542236328,
       "learning_rate": 9.120882436386474e-07,
-      "loss": 0.5578,
+      "loss": 0.5641,
       "step": 1403
     },
     {
-      "epoch": 1.3809055118110236,
-      "grad_norm": 10.482789993286133,
+      "epoch": 1.3818897637795275,
+      "grad_norm": 16.76909828186035,
       "learning_rate": 9.116529065613954e-07,
-      "loss": 0.5727,
+      "loss": 1.0729,
       "step": 1404
     },
     {
-      "epoch": 1.3818897637795275,
-      "grad_norm": 12.011210441589355,
+      "epoch": 1.3828740157480315,
+      "grad_norm": 8.080307960510254,
       "learning_rate": 9.112165990692231e-07,
-      "loss": 1.0053,
+      "loss": 0.3779,
       "step": 1405
     },
     {
-      "epoch": 1.3828740157480315,
-      "grad_norm": 10.86913013458252,
+      "epoch": 1.3838582677165354,
+      "grad_norm": 14.46143913269043,
       "learning_rate": 9.107793221948464e-07,
-      "loss": 0.7875,
+      "loss": 1.0705,
       "step": 1406
     },
     {
-      "epoch": 1.3838582677165354,
-      "grad_norm": 17.681474685668945,
+      "epoch": 1.3848425196850394,
+      "grad_norm": 13.886582374572754,
       "learning_rate": 9.103410769732755e-07,
-      "loss": 1.2545,
+      "loss": 0.8394,
       "step": 1407
     },
     {
-      "epoch": 1.3848425196850394,
-      "grad_norm": 13.744503021240234,
+      "epoch": 1.3858267716535433,
+      "grad_norm": 15.248581886291504,
       "learning_rate": 9.099018644418133e-07,
-      "loss": 1.0035,
+      "loss": 1.0838,
       "step": 1408
     },
     {
-      "epoch": 1.3858267716535433,
-      "grad_norm": 12.040654182434082,
+      "epoch": 1.3868110236220472,
+      "grad_norm": 18.300304412841797,
       "learning_rate": 9.094616856400512e-07,
-      "loss": 0.7365,
+      "loss": 1.4442,
       "step": 1409
     },
     {
-      "epoch": 1.3868110236220472,
-      "grad_norm": 7.282107353210449,
+      "epoch": 1.3877952755905512,
+      "grad_norm": 13.699790954589844,
       "learning_rate": 9.090205416098685e-07,
-      "loss": 0.3283,
+      "loss": 0.8921,
       "step": 1410
     },
     {
-      "epoch": 1.3877952755905512,
-      "grad_norm": 12.776232719421387,
+      "epoch": 1.388779527559055,
+      "grad_norm": 7.783717155456543,
       "learning_rate": 9.08578433395429e-07,
-      "loss": 0.839,
+      "loss": 0.3109,
       "step": 1411
     },
     {
-      "epoch": 1.388779527559055,
-      "grad_norm": 14.230220794677734,
+      "epoch": 1.389763779527559,
+      "grad_norm": 12.392820358276367,
       "learning_rate": 9.081353620431783e-07,
-      "loss": 0.5744,
+      "loss": 0.6138,
       "step": 1412
     },
     {
-      "epoch": 1.389763779527559,
-      "grad_norm": 13.209874153137207,
+      "epoch": 1.390748031496063,
+      "grad_norm": 11.631244659423828,
       "learning_rate": 9.07691328601842e-07,
-      "loss": 0.6241,
+      "loss": 0.7296,
       "step": 1413
     },
     {
-      "epoch": 1.390748031496063,
-      "grad_norm": 7.803886413574219,
+      "epoch": 1.391732283464567,
+      "grad_norm": 11.222329139709473,
       "learning_rate": 9.072463341224232e-07,
-      "loss": 0.3192,
+      "loss": 0.6515,
       "step": 1414
     },
     {
-      "epoch": 1.391732283464567,
-      "grad_norm": 19.139183044433594,
+      "epoch": 1.3927165354330708,
+      "grad_norm": 8.38576889038086,
       "learning_rate": 9.068003796581991e-07,
-      "loss": 1.1741,
+      "loss": 0.323,
       "step": 1415
     },
     {
-      "epoch": 1.3927165354330708,
-      "grad_norm": 8.626065254211426,
+      "epoch": 1.3937007874015748,
+      "grad_norm": 14.723484992980957,
       "learning_rate": 9.063534662647195e-07,
-      "loss": 0.4036,
+      "loss": 1.2172,
       "step": 1416
     },
     {
-      "epoch": 1.3937007874015748,
-      "grad_norm": 13.161855697631836,
+      "epoch": 1.3946850393700787,
+      "grad_norm": 10.360933303833008,
       "learning_rate": 9.059055949998039e-07,
-      "loss": 1.107,
+      "loss": 0.6128,
       "step": 1417
     },
     {
-      "epoch": 1.3946850393700787,
-      "grad_norm": 10.349092483520508,
+      "epoch": 1.3956692913385826,
+      "grad_norm": 13.539901733398438,
       "learning_rate": 9.05456766923539e-07,
-      "loss": 0.5198,
+      "loss": 0.7743,
       "step": 1418
     },
     {
-      "epoch": 1.3956692913385826,
-      "grad_norm": 11.868388175964355,
+      "epoch": 1.3966535433070866,
+      "grad_norm": 11.068268775939941,
       "learning_rate": 9.05006983098276e-07,
-      "loss": 0.5921,
+      "loss": 0.5889,
       "step": 1419
     },
     {
-      "epoch": 1.3966535433070866,
-      "grad_norm": 8.070558547973633,
+      "epoch": 1.3976377952755905,
+      "grad_norm": 12.779909133911133,
       "learning_rate": 9.045562445886289e-07,
-      "loss": 0.3738,
+      "loss": 0.9509,
       "step": 1420
     },
     {
-      "epoch": 1.3976377952755905,
-      "grad_norm": 16.018455505371094,
+      "epoch": 1.3986220472440944,
+      "grad_norm": 9.246415138244629,
       "learning_rate": 9.041045524614707e-07,
-      "loss": 0.8777,
+      "loss": 0.358,
       "step": 1421
     },
     {
-      "epoch": 1.3986220472440944,
-      "grad_norm": 9.561680793762207,
+      "epoch": 1.3996062992125984,
+      "grad_norm": 12.120144844055176,
       "learning_rate": 9.03651907785932e-07,
-      "loss": 0.4179,
+      "loss": 0.8089,
       "step": 1422
     },
     {
-      "epoch": 1.3996062992125984,
-      "grad_norm": 13.034375190734863,
+      "epoch": 1.4005905511811023,
+      "grad_norm": 9.861230850219727,
       "learning_rate": 9.03198311633398e-07,
-      "loss": 1.0642,
+      "loss": 0.4282,
       "step": 1423
     },
     {
-      "epoch": 1.4005905511811023,
-      "grad_norm": 9.478670120239258,
+      "epoch": 1.4015748031496063,
+      "grad_norm": 10.857460975646973,
       "learning_rate": 9.027437650775055e-07,
-      "loss": 0.4586,
+      "loss": 0.7527,
       "step": 1424
     },
     {
-      "epoch": 1.4015748031496063,
-      "grad_norm": 8.93643569946289,
+      "epoch": 1.4025590551181102,
+      "grad_norm": 7.980360507965088,
       "learning_rate": 9.022882691941417e-07,
-      "loss": 0.5648,
+      "loss": 0.3626,
       "step": 1425
     },
     {
-      "epoch": 1.4025590551181102,
-      "grad_norm": 9.910337448120117,
+      "epoch": 1.4035433070866141,
+      "grad_norm": 12.22629451751709,
       "learning_rate": 9.018318250614402e-07,
-      "loss": 0.551,
+      "loss": 0.7605,
       "step": 1426
     },
     {
-      "epoch": 1.4035433070866141,
-      "grad_norm": 14.57641887664795,
+      "epoch": 1.404527559055118,
+      "grad_norm": 15.700676918029785,
       "learning_rate": 9.01374433759779e-07,
-      "loss": 0.9472,
+      "loss": 0.8407,
       "step": 1427
     },
     {
-      "epoch": 1.404527559055118,
-      "grad_norm": 9.881360054016113,
+      "epoch": 1.405511811023622,
+      "grad_norm": 10.976335525512695,
       "learning_rate": 9.009160963717784e-07,
-      "loss": 0.3759,
+      "loss": 0.6551,
       "step": 1428
     },
     {
-      "epoch": 1.405511811023622,
-      "grad_norm": 12.745662689208984,
+      "epoch": 1.406496062992126,
+      "grad_norm": 11.496158599853516,
       "learning_rate": 9.004568139822978e-07,
-      "loss": 0.8034,
+      "loss": 0.6433,
       "step": 1429
     },
     {
-      "epoch": 1.406496062992126,
-      "grad_norm": 23.145109176635742,
+      "epoch": 1.4074803149606299,
+      "grad_norm": 11.197528839111328,
       "learning_rate": 8.999965876784333e-07,
-      "loss": 1.5744,
+      "loss": 0.6133,
       "step": 1430
     },
     {
-      "epoch": 1.4074803149606299,
-      "grad_norm": 8.001358032226562,
+      "epoch": 1.4084645669291338,
+      "grad_norm": 10.293011665344238,
       "learning_rate": 8.995354185495155e-07,
-      "loss": 0.2946,
+      "loss": 0.4582,
       "step": 1431
     },
     {
-      "epoch": 1.4084645669291338,
-      "grad_norm": 12.089577674865723,
+      "epoch": 1.4094488188976377,
+      "grad_norm": 17.310544967651367,
       "learning_rate": 8.990733076871063e-07,
-      "loss": 0.6559,
+      "loss": 0.8696,
       "step": 1432
     },
     {
-      "epoch": 1.4094488188976377,
-      "grad_norm": 11.743132591247559,
+      "epoch": 1.4104330708661417,
+      "grad_norm": 12.043225288391113,
       "learning_rate": 8.986102561849967e-07,
-      "loss": 0.5264,
+      "loss": 0.6652,
       "step": 1433
     },
     {
-      "epoch": 1.4104330708661417,
-      "grad_norm": 10.895527839660645,
+      "epoch": 1.4114173228346456,
+      "grad_norm": 9.704212188720703,
       "learning_rate": 8.981462651392041e-07,
-      "loss": 0.6314,
+      "loss": 0.5286,
       "step": 1434
     },
     {
-      "epoch": 1.4114173228346456,
-      "grad_norm": 17.545785903930664,
+      "epoch": 1.4124015748031495,
+      "grad_norm": 11.730301856994629,
       "learning_rate": 8.9768133564797e-07,
-      "loss": 1.0635,
+      "loss": 0.5371,
       "step": 1435
     },
     {
-      "epoch": 1.4124015748031495,
-      "grad_norm": 8.993785858154297,
+      "epoch": 1.4133858267716535,
+      "grad_norm": 9.802748680114746,
       "learning_rate": 8.972154688117568e-07,
-      "loss": 0.373,
+      "loss": 0.5219,
       "step": 1436
     },
     {
-      "epoch": 1.4133858267716535,
-      "grad_norm": 20.556406021118164,
+      "epoch": 1.4143700787401574,
+      "grad_norm": 10.800559043884277,
       "learning_rate": 8.967486657332462e-07,
-      "loss": 1.1461,
+      "loss": 0.593,
       "step": 1437
     },
     {
-      "epoch": 1.4143700787401574,
-      "grad_norm": 18.048484802246094,
+      "epoch": 1.4153543307086613,
+      "grad_norm": 15.391337394714355,
       "learning_rate": 8.962809275173347e-07,
-      "loss": 0.8015,
+      "loss": 0.8033,
       "step": 1438
     },
     {
-      "epoch": 1.4153543307086613,
-      "grad_norm": 8.574625015258789,
+      "epoch": 1.4163385826771653,
+      "grad_norm": 11.981468200683594,
       "learning_rate": 8.958122552711336e-07,
-      "loss": 0.4479,
+      "loss": 0.7256,
       "step": 1439
     },
     {
-      "epoch": 1.4163385826771653,
-      "grad_norm": 9.441163063049316,
+      "epoch": 1.4173228346456692,
+      "grad_norm": 8.102356910705566,
       "learning_rate": 8.953426501039641e-07,
-      "loss": 0.3387,
+      "loss": 0.5703,
       "step": 1440
     },
     {
-      "epoch": 1.4173228346456692,
-      "grad_norm": 21.563499450683594,
+      "epoch": 1.4183070866141732,
+      "grad_norm": 12.768531799316406,
       "learning_rate": 8.94872113127356e-07,
-      "loss": 0.7423,
+      "loss": 0.8677,
       "step": 1441
     },
     {
-      "epoch": 1.4183070866141732,
-      "grad_norm": 14.528879165649414,
+      "epoch": 1.419291338582677,
+      "grad_norm": 12.328608512878418,
       "learning_rate": 8.944006454550443e-07,
-      "loss": 0.9856,
+      "loss": 0.6869,
       "step": 1442
     },
     {
-      "epoch": 1.419291338582677,
-      "grad_norm": 9.45664119720459,
+      "epoch": 1.420275590551181,
+      "grad_norm": 8.419022560119629,
       "learning_rate": 8.939282482029671e-07,
-      "loss": 0.449,
+      "loss": 0.4755,
       "step": 1443
     },
     {
-      "epoch": 1.420275590551181,
-      "grad_norm": 14.329258918762207,
+      "epoch": 1.421259842519685,
+      "grad_norm": 12.00171947479248,
       "learning_rate": 8.934549224892629e-07,
-      "loss": 0.6372,
+      "loss": 0.7355,
       "step": 1444
     },
     {
-      "epoch": 1.421259842519685,
-      "grad_norm": 8.746792793273926,
+      "epoch": 1.422244094488189,
+      "grad_norm": 6.688138008117676,
       "learning_rate": 8.929806694342677e-07,
-      "loss": 0.3884,
+      "loss": 0.2593,
       "step": 1445
     },
     {
-      "epoch": 1.422244094488189,
-      "grad_norm": 10.391413688659668,
+      "epoch": 1.423228346456693,
+      "grad_norm": 10.913597106933594,
       "learning_rate": 8.925054901605124e-07,
-      "loss": 0.5394,
+      "loss": 0.6159,
       "step": 1446
     },
     {
-      "epoch": 1.423228346456693,
-      "grad_norm": 15.459757804870605,
+      "epoch": 1.4242125984251968,
+      "grad_norm": 12.266181945800781,
       "learning_rate": 8.9202938579272e-07,
-      "loss": 0.8704,
+      "loss": 0.6839,
       "step": 1447
     },
     {
-      "epoch": 1.4242125984251968,
-      "grad_norm": 9.648519515991211,
+      "epoch": 1.425196850393701,
+      "grad_norm": 15.073808670043945,
       "learning_rate": 8.915523574578038e-07,
-      "loss": 0.6108,
+      "loss": 0.9492,
       "step": 1448
     },
     {
-      "epoch": 1.425196850393701,
-      "grad_norm": 19.38627052307129,
+      "epoch": 1.4261811023622046,
+      "grad_norm": 12.433513641357422,
       "learning_rate": 8.910744062848633e-07,
-      "loss": 1.1582,
+      "loss": 1.066,
       "step": 1449
     },
     {
-      "epoch": 1.4261811023622046,
-      "grad_norm": 21.816274642944336,
+      "epoch": 1.4271653543307088,
+      "grad_norm": 10.259384155273438,
       "learning_rate": 8.90595533405183e-07,
-      "loss": 1.5682,
+      "loss": 0.7793,
       "step": 1450
     },
     {
-      "epoch": 1.4271653543307088,
-      "grad_norm": 10.449719429016113,
+      "epoch": 1.4281496062992125,
+      "grad_norm": 9.378206253051758,
       "learning_rate": 8.901157399522286e-07,
-      "loss": 0.4988,
+      "loss": 0.4892,
       "step": 1451
     },
     {
-      "epoch": 1.4281496062992125,
-      "grad_norm": 11.805849075317383,
+      "epoch": 1.4291338582677167,
+      "grad_norm": 14.53237247467041,
       "learning_rate": 8.896350270616449e-07,
-      "loss": 0.8312,
+      "loss": 1.1506,
       "step": 1452
     },
     {
-      "epoch": 1.4291338582677167,
-      "grad_norm": 20.020519256591797,
+      "epoch": 1.4301181102362204,
+      "grad_norm": 9.962061882019043,
       "learning_rate": 8.891533958712529e-07,
-      "loss": 0.9784,
+      "loss": 0.6908,
       "step": 1453
     },
     {
-      "epoch": 1.4301181102362204,
-      "grad_norm": 11.389593124389648,
+      "epoch": 1.4311023622047245,
+      "grad_norm": 8.699627876281738,
       "learning_rate": 8.88670847521047e-07,
-      "loss": 0.6923,
+      "loss": 0.4107,
       "step": 1454
     },
     {
-      "epoch": 1.4311023622047245,
-      "grad_norm": 10.024092674255371,
+      "epoch": 1.4320866141732282,
+      "grad_norm": 13.175297737121582,
       "learning_rate": 8.881873831531928e-07,
-      "loss": 0.4661,
+      "loss": 0.7537,
       "step": 1455
     },
     {
-      "epoch": 1.4320866141732282,
-      "grad_norm": 6.586660385131836,
+      "epoch": 1.4330708661417324,
+      "grad_norm": 13.997385025024414,
       "learning_rate": 8.877030039120242e-07,
-      "loss": 0.229,
+      "loss": 1.0748,
       "step": 1456
     },
     {
-      "epoch": 1.4330708661417324,
-      "grad_norm": 11.63663387298584,
+      "epoch": 1.4340551181102361,
+      "grad_norm": 12.485882759094238,
       "learning_rate": 8.872177109440398e-07,
-      "loss": 0.6479,
+      "loss": 0.8493,
       "step": 1457
     },
     {
-      "epoch": 1.4340551181102361,
-      "grad_norm": 11.15836238861084,
+      "epoch": 1.4350393700787403,
+      "grad_norm": 10.378874778747559,
       "learning_rate": 8.867315053979016e-07,
-      "loss": 0.6628,
+      "loss": 0.5319,
       "step": 1458
     },
     {
-      "epoch": 1.4350393700787403,
-      "grad_norm": 13.676897048950195,
+      "epoch": 1.436023622047244,
+      "grad_norm": 7.443114280700684,
       "learning_rate": 8.862443884244314e-07,
-      "loss": 0.7929,
+      "loss": 0.3999,
       "step": 1459
     },
     {
-      "epoch": 1.436023622047244,
-      "grad_norm": 6.395401477813721,
+      "epoch": 1.4370078740157481,
+      "grad_norm": 12.757030487060547,
       "learning_rate": 8.857563611766085e-07,
-      "loss": 0.2601,
+      "loss": 0.679,
       "step": 1460
     },
     {
-      "epoch": 1.4370078740157481,
-      "grad_norm": 11.027069091796875,
+      "epoch": 1.4379921259842519,
+      "grad_norm": 12.555755615234375,
       "learning_rate": 8.852674248095663e-07,
-      "loss": 0.5944,
+      "loss": 1.1676,
       "step": 1461
     },
     {
-      "epoch": 1.4379921259842519,
-      "grad_norm": 14.573845863342285,
+      "epoch": 1.438976377952756,
+      "grad_norm": 11.089254379272461,
       "learning_rate": 8.847775804805907e-07,
-      "loss": 0.9913,
+      "loss": 0.7234,
       "step": 1462
     },
     {
-      "epoch": 1.438976377952756,
-      "grad_norm": 9.79361629486084,
+      "epoch": 1.4399606299212597,
+      "grad_norm": 10.351755142211914,
       "learning_rate": 8.84286829349116e-07,
-      "loss": 0.4411,
+      "loss": 0.5784,
       "step": 1463
     },
     {
-      "epoch": 1.4399606299212597,
-      "grad_norm": 7.446544647216797,
+      "epoch": 1.4409448818897639,
+      "grad_norm": 11.980927467346191,
       "learning_rate": 8.837951725767234e-07,
-      "loss": 0.3742,
+      "loss": 0.8432,
       "step": 1464
     },
     {
-      "epoch": 1.4409448818897639,
-      "grad_norm": 10.105347633361816,
+      "epoch": 1.4419291338582676,
+      "grad_norm": 11.256375312805176,
       "learning_rate": 8.833026113271374e-07,
-      "loss": 0.398,
+      "loss": 0.5748,
       "step": 1465
     },
     {
-      "epoch": 1.4419291338582676,
-      "grad_norm": 12.443421363830566,
+      "epoch": 1.4429133858267718,
+      "grad_norm": 12.114046096801758,
       "learning_rate": 8.828091467662234e-07,
-      "loss": 0.579,
+      "loss": 0.7716,
       "step": 1466
     },
     {
-      "epoch": 1.4429133858267718,
-      "grad_norm": 14.672910690307617,
+      "epoch": 1.4438976377952755,
+      "grad_norm": 9.06035327911377,
       "learning_rate": 8.82314780061985e-07,
-      "loss": 0.9689,
+      "loss": 0.6102,
       "step": 1467
     },
     {
-      "epoch": 1.4438976377952755,
-      "grad_norm": 11.923147201538086,
+      "epoch": 1.4448818897637796,
+      "grad_norm": 15.105749130249023,
       "learning_rate": 8.81819512384561e-07,
-      "loss": 0.5551,
+      "loss": 1.3959,
       "step": 1468
     },
     {
-      "epoch": 1.4448818897637796,
-      "grad_norm": 11.983245849609375,
+      "epoch": 1.4458661417322833,
+      "grad_norm": 15.754951477050781,
       "learning_rate": 8.813233449062229e-07,
-      "loss": 0.4934,
+      "loss": 1.1255,
       "step": 1469
     },
     {
-      "epoch": 1.4458661417322833,
-      "grad_norm": 16.73973274230957,
+      "epoch": 1.4468503937007875,
+      "grad_norm": 11.91762924194336,
       "learning_rate": 8.808262788013718e-07,
-      "loss": 0.9175,
+      "loss": 0.5782,
       "step": 1470
     },
     {
-      "epoch": 1.4468503937007875,
-      "grad_norm": 23.527509689331055,
+      "epoch": 1.4478346456692912,
+      "grad_norm": 13.89993667602539,
       "learning_rate": 8.803283152465357e-07,
-      "loss": 1.5047,
+      "loss": 0.9801,
       "step": 1471
     },
     {
-      "epoch": 1.4478346456692912,
-      "grad_norm": 10.171259880065918,
+      "epoch": 1.4488188976377954,
+      "grad_norm": 9.969817161560059,
       "learning_rate": 8.798294554203674e-07,
-      "loss": 0.5172,
+      "loss": 0.446,
       "step": 1472
     },
     {
-      "epoch": 1.4488188976377954,
-      "grad_norm": 22.97465705871582,
+      "epoch": 1.449803149606299,
+      "grad_norm": 10.130709648132324,
       "learning_rate": 8.793297005036402e-07,
-      "loss": 1.2806,
+      "loss": 0.4698,
       "step": 1473
     },
     {
-      "epoch": 1.449803149606299,
-      "grad_norm": 11.81129264831543,
+      "epoch": 1.4507874015748032,
+      "grad_norm": 5.649473667144775,
       "learning_rate": 8.788290516792471e-07,
-      "loss": 0.8314,
+      "loss": 0.1574,
       "step": 1474
     },
     {
-      "epoch": 1.4507874015748032,
-      "grad_norm": 14.46216869354248,
+      "epoch": 1.451771653543307,
+      "grad_norm": 10.669340133666992,
       "learning_rate": 8.783275101321962e-07,
-      "loss": 0.5415,
+      "loss": 0.5371,
       "step": 1475
     },
     {
-      "epoch": 1.451771653543307,
-      "grad_norm": 13.006589889526367,
+      "epoch": 1.452755905511811,
+      "grad_norm": 14.848026275634766,
       "learning_rate": 8.778250770496084e-07,
-      "loss": 0.5999,
+      "loss": 0.8717,
       "step": 1476
     },
     {
-      "epoch": 1.452755905511811,
-      "grad_norm": 12.510464668273926,
+      "epoch": 1.453740157480315,
+      "grad_norm": 11.245269775390625,
       "learning_rate": 8.773217536207158e-07,
-      "loss": 0.7689,
+      "loss": 0.6271,
       "step": 1477
     },
     {
-      "epoch": 1.453740157480315,
-      "grad_norm": 19.87265968322754,
+      "epoch": 1.454724409448819,
+      "grad_norm": 10.957762718200684,
       "learning_rate": 8.76817541036857e-07,
-      "loss": 1.1297,
+      "loss": 0.4344,
       "step": 1478
     },
     {
-      "epoch": 1.454724409448819,
-      "grad_norm": 10.278497695922852,
+      "epoch": 1.455708661417323,
+      "grad_norm": 11.351705551147461,
       "learning_rate": 8.763124404914754e-07,
-      "loss": 0.3801,
+      "loss": 0.3845,
       "step": 1479
     },
     {
-      "epoch": 1.455708661417323,
-      "grad_norm": 13.605413436889648,
+      "epoch": 1.4566929133858268,
+      "grad_norm": 10.221748352050781,
       "learning_rate": 8.758064531801166e-07,
-      "loss": 0.3896,
+      "loss": 0.7941,
       "step": 1480
     },
     {
-      "epoch": 1.4566929133858268,
-      "grad_norm": 10.33932876586914,
+      "epoch": 1.4576771653543308,
+      "grad_norm": 11.94582462310791,
       "learning_rate": 8.752995803004247e-07,
-      "loss": 0.4557,
+      "loss": 0.824,
       "step": 1481
     },
     {
-      "epoch": 1.4576771653543308,
-      "grad_norm": 10.69277286529541,
+      "epoch": 1.4586614173228347,
+      "grad_norm": 10.737421989440918,
       "learning_rate": 8.747918230521398e-07,
-      "loss": 0.5844,
+      "loss": 0.439,
       "step": 1482
     },
     {
-      "epoch": 1.4586614173228347,
-      "grad_norm": 15.579158782958984,
+      "epoch": 1.4596456692913387,
+      "grad_norm": 6.845982074737549,
       "learning_rate": 8.742831826370956e-07,
-      "loss": 0.8164,
+      "loss": 0.29,
       "step": 1483
     },
     {
-      "epoch": 1.4596456692913387,
-      "grad_norm": 9.021865844726562,
+      "epoch": 1.4606299212598426,
+      "grad_norm": 10.5374116897583,
       "learning_rate": 8.737736602592158e-07,
-      "loss": 0.4885,
+      "loss": 0.7063,
       "step": 1484
     },
     {
-      "epoch": 1.4606299212598426,
-      "grad_norm": 11.235892295837402,
+      "epoch": 1.4616141732283465,
+      "grad_norm": 9.620530128479004,
       "learning_rate": 8.732632571245121e-07,
-      "loss": 0.9719,
+      "loss": 0.5149,
       "step": 1485
     },
     {
-      "epoch": 1.4616141732283465,
-      "grad_norm": 13.27570629119873,
+      "epoch": 1.4625984251968505,
+      "grad_norm": 11.207847595214844,
       "learning_rate": 8.727519744410806e-07,
-      "loss": 0.6612,
+      "loss": 0.6015,
       "step": 1486
     },
     {
-      "epoch": 1.4625984251968505,
-      "grad_norm": 11.570671081542969,
+      "epoch": 1.4635826771653544,
+      "grad_norm": 9.3619966506958,
       "learning_rate": 8.722398134190995e-07,
-      "loss": 0.4546,
+      "loss": 0.4672,
       "step": 1487
     },
     {
-      "epoch": 1.4635826771653544,
-      "grad_norm": 17.243309020996094,
+      "epoch": 1.4645669291338583,
+      "grad_norm": 10.878146171569824,
       "learning_rate": 8.717267752708257e-07,
-      "loss": 1.1661,
+      "loss": 0.5126,
       "step": 1488
     },
     {
-      "epoch": 1.4645669291338583,
-      "grad_norm": 14.900501251220703,
+      "epoch": 1.4655511811023623,
+      "grad_norm": 10.388224601745605,
       "learning_rate": 8.712128612105924e-07,
-      "loss": 0.6853,
+      "loss": 0.5502,
       "step": 1489
     },
     {
-      "epoch": 1.4655511811023623,
-      "grad_norm": 9.174758911132812,
+      "epoch": 1.4665354330708662,
+      "grad_norm": 8.63163948059082,
       "learning_rate": 8.706980724548058e-07,
-      "loss": 0.3457,
+      "loss": 0.368,
       "step": 1490
     },
     {
-      "epoch": 1.4665354330708662,
-      "grad_norm": 10.846331596374512,
+      "epoch": 1.4675196850393701,
+      "grad_norm": 14.595239639282227,
       "learning_rate": 8.701824102219428e-07,
-      "loss": 0.5259,
+      "loss": 0.8213,
       "step": 1491
     },
     {
-      "epoch": 1.4675196850393701,
-      "grad_norm": 12.600656509399414,
+      "epoch": 1.468503937007874,
+      "grad_norm": 12.493700981140137,
       "learning_rate": 8.696658757325474e-07,
-      "loss": 0.8553,
+      "loss": 0.7298,
       "step": 1492
     },
     {
-      "epoch": 1.468503937007874,
-      "grad_norm": 10.123764991760254,
+      "epoch": 1.469488188976378,
+      "grad_norm": 11.516324996948242,
       "learning_rate": 8.691484702092286e-07,
-      "loss": 0.3868,
+      "loss": 0.4096,
       "step": 1493
     },
     {
-      "epoch": 1.469488188976378,
-      "grad_norm": 8.318851470947266,
+      "epoch": 1.470472440944882,
+      "grad_norm": 9.997244834899902,
       "learning_rate": 8.686301948766568e-07,
-      "loss": 0.2448,
+      "loss": 0.483,
       "step": 1494
     },
     {
-      "epoch": 1.470472440944882,
-      "grad_norm": 8.971074104309082,
+      "epoch": 1.4714566929133859,
+      "grad_norm": 14.764153480529785,
       "learning_rate": 8.681110509615612e-07,
-      "loss": 0.435,
+      "loss": 0.6826,
       "step": 1495
     },
     {
-      "epoch": 1.4714566929133859,
-      "grad_norm": 13.754014015197754,
+      "epoch": 1.4724409448818898,
+      "grad_norm": 14.205328941345215,
       "learning_rate": 8.675910396927269e-07,
-      "loss": 0.6314,
+      "loss": 0.7304,
       "step": 1496
     },
     {
-      "epoch": 1.4724409448818898,
-      "grad_norm": 8.66455364227295,
+      "epoch": 1.4734251968503937,
+      "grad_norm": 8.074531555175781,
       "learning_rate": 8.670701623009921e-07,
-      "loss": 0.5653,
+      "loss": 0.3569,
       "step": 1497
     },
     {
-      "epoch": 1.4734251968503937,
-      "grad_norm": 15.178778648376465,
+      "epoch": 1.4744094488188977,
+      "grad_norm": 10.682283401489258,
       "learning_rate": 8.66548420019245e-07,
-      "loss": 0.8,
+      "loss": 0.5318,
       "step": 1498
     },
     {
-      "epoch": 1.4744094488188977,
-      "grad_norm": 10.51429557800293,
+      "epoch": 1.4753937007874016,
+      "grad_norm": 8.804259300231934,
       "learning_rate": 8.66025814082421e-07,
-      "loss": 0.5797,
+      "loss": 0.2428,
       "step": 1499
     },
     {
-      "epoch": 1.4753937007874016,
-      "grad_norm": 16.99217414855957,
+      "epoch": 1.4763779527559056,
+      "grad_norm": 15.6033353805542,
       "learning_rate": 8.655023457274994e-07,
-      "loss": 0.827,
+      "loss": 1.1592,
       "step": 1500
     },
     {
-      "epoch": 1.4763779527559056,
-      "grad_norm": 14.463443756103516,
+      "epoch": 1.4773622047244095,
+      "grad_norm": 15.73362922668457,
       "learning_rate": 8.649780161935012e-07,
-      "loss": 1.0582,
+      "loss": 0.957,
       "step": 1501
     },
     {
-      "epoch": 1.4773622047244095,
-      "grad_norm": 9.892260551452637,
+      "epoch": 1.4783464566929134,
+      "grad_norm": 15.021441459655762,
       "learning_rate": 8.644528267214859e-07,
-      "loss": 0.2981,
+      "loss": 0.7483,
       "step": 1502
     },
     {
-      "epoch": 1.4783464566929134,
-      "grad_norm": 10.980031967163086,
+      "epoch": 1.4793307086614174,
+      "grad_norm": 10.988287925720215,
       "learning_rate": 8.639267785545478e-07,
-      "loss": 0.8084,
+      "loss": 0.739,
       "step": 1503
     },
     {
-      "epoch": 1.4793307086614174,
-      "grad_norm": 11.92680835723877,
+      "epoch": 1.4803149606299213,
+      "grad_norm": 16.389732360839844,
       "learning_rate": 8.63399872937814e-07,
-      "loss": 1.1578,
+      "loss": 0.6278,
       "step": 1504
     },
     {
-      "epoch": 1.4803149606299213,
-      "grad_norm": 14.131094932556152,
+      "epoch": 1.4812992125984252,
+      "grad_norm": 13.051485061645508,
       "learning_rate": 8.628721111184414e-07,
-      "loss": 0.7456,
+      "loss": 0.752,
       "step": 1505
     },
     {
-      "epoch": 1.4812992125984252,
-      "grad_norm": 13.81746768951416,
+      "epoch": 1.4822834645669292,
+      "grad_norm": 11.54430866241455,
       "learning_rate": 8.623434943456132e-07,
-      "loss": 0.8266,
+      "loss": 0.6608,
       "step": 1506
     },
     {
-      "epoch": 1.4822834645669292,
-      "grad_norm": 9.297046661376953,
+      "epoch": 1.483267716535433,
+      "grad_norm": 8.369393348693848,
       "learning_rate": 8.61814023870536e-07,
-      "loss": 0.3407,
+      "loss": 0.2598,
       "step": 1507
     },
     {
-      "epoch": 1.483267716535433,
-      "grad_norm": 13.54496955871582,
+      "epoch": 1.484251968503937,
+      "grad_norm": 18.80166244506836,
       "learning_rate": 8.612837009464377e-07,
-      "loss": 0.7785,
+      "loss": 1.664,
       "step": 1508
     },
     {
-      "epoch": 1.484251968503937,
-      "grad_norm": 16.752914428710938,
+      "epoch": 1.485236220472441,
+      "grad_norm": 13.08558177947998,
       "learning_rate": 8.607525268285634e-07,
-      "loss": 1.2844,
+      "loss": 0.9461,
       "step": 1509
     },
     {
-      "epoch": 1.485236220472441,
-      "grad_norm": 14.00351333618164,
+      "epoch": 1.486220472440945,
+      "grad_norm": 9.269903182983398,
       "learning_rate": 8.602205027741731e-07,
-      "loss": 0.6928,
+      "loss": 0.3373,
       "step": 1510
     },
     {
-      "epoch": 1.486220472440945,
-      "grad_norm": 12.860536575317383,
+      "epoch": 1.4872047244094488,
+      "grad_norm": 11.707659721374512,
       "learning_rate": 8.596876300425384e-07,
-      "loss": 0.3945,
+      "loss": 0.9237,
       "step": 1511
     },
     {
-      "epoch": 1.4872047244094488,
-      "grad_norm": 10.744647026062012,
+      "epoch": 1.4881889763779528,
+      "grad_norm": 7.032736301422119,
       "learning_rate": 8.591539098949401e-07,
-      "loss": 0.5651,
+      "loss": 0.359,
       "step": 1512
     },
     {
-      "epoch": 1.4881889763779528,
-      "grad_norm": 9.299052238464355,
+      "epoch": 1.4891732283464567,
+      "grad_norm": 12.72712516784668,
       "learning_rate": 8.586193435946641e-07,
-      "loss": 0.4731,
+      "loss": 0.7701,
       "step": 1513
     },
     {
-      "epoch": 1.4891732283464567,
-      "grad_norm": 7.104952812194824,
+      "epoch": 1.4901574803149606,
+      "grad_norm": 11.758654594421387,
       "learning_rate": 8.580839324069998e-07,
-      "loss": 0.249,
+      "loss": 0.5773,
       "step": 1514
     },
     {
-      "epoch": 1.4901574803149606,
-      "grad_norm": 11.674582481384277,
+      "epoch": 1.4911417322834646,
+      "grad_norm": 13.19546127319336,
       "learning_rate": 8.57547677599236e-07,
-      "loss": 0.4961,
+      "loss": 0.8257,
       "step": 1515
     },
     {
-      "epoch": 1.4911417322834646,
-      "grad_norm": 11.296091079711914,
+      "epoch": 1.4921259842519685,
+      "grad_norm": 10.869878768920898,
       "learning_rate": 8.570105804406583e-07,
-      "loss": 0.6058,
+      "loss": 0.7083,
       "step": 1516
     },
     {
-      "epoch": 1.4921259842519685,
-      "grad_norm": 15.509420394897461,
+      "epoch": 1.4931102362204725,
+      "grad_norm": 9.778740882873535,
       "learning_rate": 8.564726422025462e-07,
-      "loss": 1.0121,
+      "loss": 0.6284,
       "step": 1517
     },
     {
-      "epoch": 1.4931102362204725,
-      "grad_norm": 9.842504501342773,
+      "epoch": 1.4940944881889764,
+      "grad_norm": 8.870782852172852,
       "learning_rate": 8.5593386415817e-07,
-      "loss": 0.4994,
+      "loss": 0.6834,
       "step": 1518
     },
     {
-      "epoch": 1.4940944881889764,
-      "grad_norm": 11.032633781433105,
+      "epoch": 1.4950787401574803,
+      "grad_norm": 9.989852905273438,
       "learning_rate": 8.553942475827879e-07,
-      "loss": 0.3958,
+      "loss": 0.574,
       "step": 1519
     },
     {
-      "epoch": 1.4950787401574803,
-      "grad_norm": 13.140924453735352,
+      "epoch": 1.4960629921259843,
+      "grad_norm": 6.89227819442749,
       "learning_rate": 8.548537937536425e-07,
-      "loss": 0.5271,
+      "loss": 0.3719,
       "step": 1520
     },
     {
-      "epoch": 1.4960629921259843,
-      "grad_norm": 7.572324275970459,
+      "epoch": 1.4970472440944882,
+      "grad_norm": 11.572386741638184,
       "learning_rate": 8.543125039499582e-07,
-      "loss": 0.3546,
+      "loss": 0.7524,
       "step": 1521
     },
     {
-      "epoch": 1.4970472440944882,
-      "grad_norm": 11.842230796813965,
+      "epoch": 1.4980314960629921,
+      "grad_norm": 8.91762924194336,
       "learning_rate": 8.537703794529387e-07,
-      "loss": 0.5617,
+      "loss": 0.374,
       "step": 1522
     },
     {
-      "epoch": 1.4980314960629921,
-      "grad_norm": 9.245935440063477,
+      "epoch": 1.499015748031496,
+      "grad_norm": 10.259950637817383,
       "learning_rate": 8.532274215457625e-07,
-      "loss": 0.4117,
+      "loss": 0.6744,
       "step": 1523
     },
     {
-      "epoch": 1.499015748031496,
-      "grad_norm": 19.322467803955078,
+      "epoch": 1.5,
+      "grad_norm": 9.1063871383667,
       "learning_rate": 8.526836315135815e-07,
-      "loss": 1.4661,
+      "loss": 0.4346,
       "step": 1524
     },
     {
-      "epoch": 1.5,
-      "grad_norm": 14.873554229736328,
+      "epoch": 1.500984251968504,
+      "grad_norm": 9.23418140411377,
       "learning_rate": 8.521390106435168e-07,
-      "loss": 0.8395,
+      "loss": 0.4382,
       "step": 1525
     },
     {
-      "epoch": 1.500984251968504,
-      "grad_norm": 9.65881633758545,
+      "epoch": 1.5019685039370079,
+      "grad_norm": 8.439826965332031,
       "learning_rate": 8.515935602246558e-07,
-      "loss": 0.4478,
+      "loss": 0.5753,
       "step": 1526
     },
     {
-      "epoch": 1.5019685039370079,
-      "grad_norm": 10.6986665725708,
+      "epoch": 1.5029527559055118,
+      "grad_norm": 12.370939254760742,
       "learning_rate": 8.510472815480498e-07,
-      "loss": 0.5156,
+      "loss": 0.981,
       "step": 1527
     },
     {
-      "epoch": 1.5029527559055118,
-      "grad_norm": 10.98793888092041,
+      "epoch": 1.5039370078740157,
+      "grad_norm": 12.242222785949707,
       "learning_rate": 8.505001759067106e-07,
-      "loss": 0.6582,
+      "loss": 0.7227,
       "step": 1528
     },
     {
-      "epoch": 1.5039370078740157,
-      "grad_norm": 7.674596786499023,
+      "epoch": 1.5049212598425197,
+      "grad_norm": 10.096325874328613,
       "learning_rate": 8.499522445956071e-07,
-      "loss": 0.3326,
+      "loss": 0.8546,
       "step": 1529
     },
     {
-      "epoch": 1.5049212598425197,
-      "grad_norm": 9.579343795776367,
+      "epoch": 1.5059055118110236,
+      "grad_norm": 9.368844985961914,
       "learning_rate": 8.494034889116626e-07,
-      "loss": 0.6277,
+      "loss": 0.577,
       "step": 1530
     },
     {
-      "epoch": 1.5049212598425197,
-      "eval_Qnli-dev_cosine_accuracy": 0.66796875,
-      "eval_Qnli-dev_cosine_accuracy_threshold": 0.7962973117828369,
-      "eval_Qnli-dev_cosine_ap": 0.691759185255588,
-      "eval_Qnli-dev_cosine_f1": 0.6839546191247974,
-      "eval_Qnli-dev_cosine_f1_threshold": 0.7521535158157349,
-      "eval_Qnli-dev_cosine_precision": 0.5538057742782152,
-      "eval_Qnli-dev_cosine_recall": 0.8940677966101694,
-      "eval_Qnli-dev_dot_accuracy": 0.662109375,
-      "eval_Qnli-dev_dot_accuracy_threshold": 606.8631591796875,
-      "eval_Qnli-dev_dot_ap": 0.6893728512808852,
-      "eval_Qnli-dev_dot_f1": 0.6848816029143898,
-      "eval_Qnli-dev_dot_f1_threshold": 601.61083984375,
-      "eval_Qnli-dev_dot_precision": 0.6006389776357828,
-      "eval_Qnli-dev_dot_recall": 0.7966101694915254,
-      "eval_Qnli-dev_euclidean_accuracy": 0.671875,
-      "eval_Qnli-dev_euclidean_accuracy_threshold": 17.649578094482422,
-      "eval_Qnli-dev_euclidean_ap": 0.6921233720273546,
-      "eval_Qnli-dev_euclidean_f1": 0.6839546191247974,
-      "eval_Qnli-dev_euclidean_f1_threshold": 19.418928146362305,
-      "eval_Qnli-dev_euclidean_precision": 0.5538057742782152,
-      "eval_Qnli-dev_euclidean_recall": 0.8940677966101694,
-      "eval_Qnli-dev_manhattan_accuracy": 0.669921875,
-      "eval_Qnli-dev_manhattan_accuracy_threshold": 355.29461669921875,
-      "eval_Qnli-dev_manhattan_ap": 0.7063039458792412,
-      "eval_Qnli-dev_manhattan_f1": 0.6957983193277311,
-      "eval_Qnli-dev_manhattan_f1_threshold": 397.1031799316406,
-      "eval_Qnli-dev_manhattan_precision": 0.5766016713091922,
-      "eval_Qnli-dev_manhattan_recall": 0.8771186440677966,
-      "eval_Qnli-dev_max_accuracy": 0.671875,
-      "eval_Qnli-dev_max_accuracy_threshold": 606.8631591796875,
-      "eval_Qnli-dev_max_ap": 0.7063039458792412,
-      "eval_Qnli-dev_max_f1": 0.6957983193277311,
-      "eval_Qnli-dev_max_f1_threshold": 601.61083984375,
-      "eval_Qnli-dev_max_precision": 0.6006389776357828,
-      "eval_Qnli-dev_max_recall": 0.8940677966101694,
+      "epoch": 1.5059055118110236,
+      "eval_Qnli-dev_cosine_accuracy": 0.654296875,
+      "eval_Qnli-dev_cosine_accuracy_threshold": 0.8250534534454346,
+      "eval_Qnli-dev_cosine_ap": 0.68496915260985,
+      "eval_Qnli-dev_cosine_f1": 0.6811352253756261,
+      "eval_Qnli-dev_cosine_f1_threshold": 0.7778436541557312,
+      "eval_Qnli-dev_cosine_precision": 0.5619834710743802,
+      "eval_Qnli-dev_cosine_recall": 0.864406779661017,
+      "eval_Qnli-dev_dot_accuracy": 0.65625,
+      "eval_Qnli-dev_dot_accuracy_threshold": 661.041259765625,
+      "eval_Qnli-dev_dot_ap": 0.6829464453911642,
+      "eval_Qnli-dev_dot_f1": 0.6787479406919276,
+      "eval_Qnli-dev_dot_f1_threshold": 588.8831176757812,
+      "eval_Qnli-dev_dot_precision": 0.555256064690027,
+      "eval_Qnli-dev_dot_recall": 0.8728813559322034,
+      "eval_Qnli-dev_euclidean_accuracy": 0.65625,
+      "eval_Qnli-dev_euclidean_accuracy_threshold": 16.30521011352539,
+      "eval_Qnli-dev_euclidean_ap": 0.6850774562069436,
+      "eval_Qnli-dev_euclidean_f1": 0.6811352253756261,
+      "eval_Qnli-dev_euclidean_f1_threshold": 18.404125213623047,
+      "eval_Qnli-dev_euclidean_precision": 0.5619834710743802,
+      "eval_Qnli-dev_euclidean_recall": 0.864406779661017,
+      "eval_Qnli-dev_manhattan_accuracy": 0.66796875,
+      "eval_Qnli-dev_manhattan_accuracy_threshold": 349.23028564453125,
+      "eval_Qnli-dev_manhattan_ap": 0.695617919723666,
+      "eval_Qnli-dev_manhattan_f1": 0.6791171477079796,
+      "eval_Qnli-dev_manhattan_f1_threshold": 380.68841552734375,
+      "eval_Qnli-dev_manhattan_precision": 0.56657223796034,
+      "eval_Qnli-dev_manhattan_recall": 0.847457627118644,
+      "eval_Qnli-dev_max_accuracy": 0.66796875,
+      "eval_Qnli-dev_max_accuracy_threshold": 661.041259765625,
+      "eval_Qnli-dev_max_ap": 0.695617919723666,
+      "eval_Qnli-dev_max_f1": 0.6811352253756261,
+      "eval_Qnli-dev_max_f1_threshold": 588.8831176757812,
+      "eval_Qnli-dev_max_precision": 0.56657223796034,
+      "eval_Qnli-dev_max_recall": 0.8728813559322034,
       "eval_allNLI-dev_cosine_accuracy": 0.689453125,
-      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9331935048103333,
-      "eval_allNLI-dev_cosine_ap": 0.5030724645677387,
-      "eval_allNLI-dev_cosine_f1": 0.5440613026819924,
-      "eval_allNLI-dev_cosine_f1_threshold": 0.819710373878479,
-      "eval_allNLI-dev_cosine_precision": 0.4068767908309456,
-      "eval_allNLI-dev_cosine_recall": 0.8208092485549133,
-      "eval_allNLI-dev_dot_accuracy": 0.689453125,
-      "eval_allNLI-dev_dot_accuracy_threshold": 707.578857421875,
-      "eval_allNLI-dev_dot_ap": 0.5031528402578558,
-      "eval_allNLI-dev_dot_f1": 0.543726235741445,
-      "eval_allNLI-dev_dot_f1_threshold": 624.6593017578125,
-      "eval_allNLI-dev_dot_precision": 0.40509915014164305,
-      "eval_allNLI-dev_dot_recall": 0.8265895953757225,
+      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9387403726577759,
+      "eval_allNLI-dev_cosine_ap": 0.5065804330173296,
+      "eval_allNLI-dev_cosine_f1": 0.5330739299610895,
+      "eval_allNLI-dev_cosine_f1_threshold": 0.8264200687408447,
+      "eval_allNLI-dev_cosine_precision": 0.40175953079178883,
+      "eval_allNLI-dev_cosine_recall": 0.791907514450867,
+      "eval_allNLI-dev_dot_accuracy": 0.6875,
+      "eval_allNLI-dev_dot_accuracy_threshold": 716.1957397460938,
+      "eval_allNLI-dev_dot_ap": 0.5073912490291572,
+      "eval_allNLI-dev_dot_f1": 0.5291828793774318,
+      "eval_allNLI-dev_dot_f1_threshold": 630.6777954101562,
+      "eval_allNLI-dev_dot_precision": 0.39882697947214074,
+      "eval_allNLI-dev_dot_recall": 0.7861271676300579,
       "eval_allNLI-dev_euclidean_accuracy": 0.689453125,
-      "eval_allNLI-dev_euclidean_accuracy_threshold": 10.111281394958496,
-      "eval_allNLI-dev_euclidean_ap": 0.5029907933861366,
-      "eval_allNLI-dev_euclidean_f1": 0.5461538461538462,
-      "eval_allNLI-dev_euclidean_f1_threshold": 16.565919876098633,
-      "eval_allNLI-dev_euclidean_precision": 0.4092219020172911,
-      "eval_allNLI-dev_euclidean_recall": 0.8208092485549133,
-      "eval_allNLI-dev_manhattan_accuracy": 0.685546875,
-      "eval_allNLI-dev_manhattan_accuracy_threshold": 224.79913330078125,
-      "eval_allNLI-dev_manhattan_ap": 0.5004588997249526,
-      "eval_allNLI-dev_manhattan_f1": 0.5419103313840157,
-      "eval_allNLI-dev_manhattan_f1_threshold": 342.75079345703125,
-      "eval_allNLI-dev_manhattan_precision": 0.4088235294117647,
-      "eval_allNLI-dev_manhattan_recall": 0.8034682080924855,
-      "eval_allNLI-dev_max_accuracy": 0.689453125,
-      "eval_allNLI-dev_max_accuracy_threshold": 707.578857421875,
-      "eval_allNLI-dev_max_ap": 0.5031528402578558,
-      "eval_allNLI-dev_max_f1": 0.5461538461538462,
-      "eval_allNLI-dev_max_f1_threshold": 624.6593017578125,
-      "eval_allNLI-dev_max_precision": 0.4092219020172911,
-      "eval_allNLI-dev_max_recall": 0.8265895953757225,
-      "eval_loss": 1.0736732482910156,
-      "eval_runtime": 54.2804,
-      "eval_samples_per_second": 30.656,
-      "eval_sequential_score": 0.7063039458792412,
-      "eval_steps_per_second": 0.129,
-      "eval_sts-test_pearson_cosine": 0.6063324046452339,
-      "eval_sts-test_pearson_dot": 0.6046695712154431,
-      "eval_sts-test_pearson_euclidean": 0.6274252948479226,
-      "eval_sts-test_pearson_manhattan": 0.6331295272217956,
-      "eval_sts-test_pearson_max": 0.6331295272217956,
-      "eval_sts-test_spearman_cosine": 0.6293434778801531,
-      "eval_sts-test_spearman_dot": 0.6275092131782424,
-      "eval_sts-test_spearman_euclidean": 0.6293739534891606,
-      "eval_sts-test_spearman_manhattan": 0.6344549222050173,
-      "eval_sts-test_spearman_max": 0.6344549222050173,
+      "eval_allNLI-dev_euclidean_accuracy_threshold": 9.675241470336914,
+      "eval_allNLI-dev_euclidean_ap": 0.506529663287556,
+      "eval_allNLI-dev_euclidean_f1": 0.5330739299610895,
+      "eval_allNLI-dev_euclidean_f1_threshold": 16.28316879272461,
+      "eval_allNLI-dev_euclidean_precision": 0.40175953079178883,
+      "eval_allNLI-dev_euclidean_recall": 0.791907514450867,
+      "eval_allNLI-dev_manhattan_accuracy": 0.69140625,
+      "eval_allNLI-dev_manhattan_accuracy_threshold": 241.33181762695312,
+      "eval_allNLI-dev_manhattan_ap": 0.5061775704986504,
+      "eval_allNLI-dev_manhattan_f1": 0.5314960629921259,
+      "eval_allNLI-dev_manhattan_f1_threshold": 340.0284118652344,
+      "eval_allNLI-dev_manhattan_precision": 0.40298507462686567,
+      "eval_allNLI-dev_manhattan_recall": 0.7803468208092486,
+      "eval_allNLI-dev_max_accuracy": 0.69140625,
+      "eval_allNLI-dev_max_accuracy_threshold": 716.1957397460938,
+      "eval_allNLI-dev_max_ap": 0.5073912490291572,
+      "eval_allNLI-dev_max_f1": 0.5330739299610895,
+      "eval_allNLI-dev_max_f1_threshold": 630.6777954101562,
+      "eval_allNLI-dev_max_precision": 0.40298507462686567,
+      "eval_allNLI-dev_max_recall": 0.791907514450867,
+      "eval_loss": 0.9738673567771912,
+      "eval_runtime": 53.2399,
+      "eval_samples_per_second": 31.255,
+      "eval_sequential_score": 0.695617919723666,
+      "eval_steps_per_second": 0.131,
+      "eval_sts-test_pearson_cosine": 0.6234338499114922,
+      "eval_sts-test_pearson_dot": 0.620329131617912,
+      "eval_sts-test_pearson_euclidean": 0.6421665682768289,
+      "eval_sts-test_pearson_manhattan": 0.6520687044199087,
+      "eval_sts-test_pearson_max": 0.6520687044199087,
+      "eval_sts-test_spearman_cosine": 0.6420121124674444,
+      "eval_sts-test_spearman_dot": 0.6384889351606223,
+      "eval_sts-test_spearman_euclidean": 0.6422998183269296,
+      "eval_sts-test_spearman_manhattan": 0.6504231567972193,
+      "eval_sts-test_spearman_max": 0.6504231567972193,
       "step": 1530
     },
     {
-      "epoch": 1.5059055118110236,
-      "grad_norm": 12.969947814941406,
+      "epoch": 1.5068897637795275,
+      "grad_norm": 9.270878791809082,
       "learning_rate": 8.488539101537518e-07,
-      "loss": 0.6134,
+      "loss": 0.6163,
       "step": 1531
     },
     {
-      "epoch": 1.5068897637795275,
-      "grad_norm": 16.989253997802734,
+      "epoch": 1.5078740157480315,
+      "grad_norm": 13.084099769592285,
       "learning_rate": 8.483035096226975e-07,
-      "loss": 0.8329,
+      "loss": 0.852,
       "step": 1532
     },
     {
-      "epoch": 1.5078740157480315,
-      "grad_norm": 9.056514739990234,
+      "epoch": 1.5088582677165354,
+      "grad_norm": 7.97643518447876,
       "learning_rate": 8.477522886212673e-07,
-      "loss": 0.458,
+      "loss": 0.3445,
       "step": 1533
     },
     {
-      "epoch": 1.5088582677165354,
-      "grad_norm": 18.400096893310547,
+      "epoch": 1.5098425196850394,
+      "grad_norm": 7.906027793884277,
       "learning_rate": 8.472002484541711e-07,
-      "loss": 0.7997,
+      "loss": 0.437,
       "step": 1534
     },
     {
-      "epoch": 1.5098425196850394,
-      "grad_norm": 14.392437934875488,
+      "epoch": 1.5108267716535433,
+      "grad_norm": 8.713579177856445,
       "learning_rate": 8.466473904280579e-07,
-      "loss": 0.9097,
+      "loss": 0.4558,
       "step": 1535
     },
     {
-      "epoch": 1.5108267716535433,
-      "grad_norm": 9.243882179260254,
+      "epoch": 1.5118110236220472,
+      "grad_norm": 9.118990898132324,
       "learning_rate": 8.46093715851512e-07,
-      "loss": 0.4139,
+      "loss": 0.3667,
       "step": 1536
     },
     {
-      "epoch": 1.5118110236220472,
-      "grad_norm": 21.380508422851562,
+      "epoch": 1.5127952755905512,
+      "grad_norm": 5.652214050292969,
       "learning_rate": 8.455392260350507e-07,
-      "loss": 1.1263,
+      "loss": 0.1558,
       "step": 1537
     },
     {
-      "epoch": 1.5127952755905512,
-      "grad_norm": 13.089934349060059,
+      "epoch": 1.513779527559055,
+      "grad_norm": 8.721686363220215,
       "learning_rate": 8.449839222911214e-07,
-      "loss": 0.4691,
+      "loss": 0.4603,
       "step": 1538
     },
     {
-      "epoch": 1.513779527559055,
-      "grad_norm": 8.922409057617188,
+      "epoch": 1.514763779527559,
+      "grad_norm": 9.682229995727539,
       "learning_rate": 8.444278059340971e-07,
-      "loss": 0.4162,
+      "loss": 0.416,
       "step": 1539
     },
     {
-      "epoch": 1.514763779527559,
-      "grad_norm": 10.464616775512695,
+      "epoch": 1.515748031496063,
+      "grad_norm": 12.936091423034668,
       "learning_rate": 8.438708782802748e-07,
-      "loss": 0.4039,
+      "loss": 0.8853,
       "step": 1540
     },
     {
-      "epoch": 1.515748031496063,
-      "grad_norm": 11.656524658203125,
+      "epoch": 1.516732283464567,
+      "grad_norm": 13.46939754486084,
       "learning_rate": 8.43313140647872e-07,
-      "loss": 0.6962,
+      "loss": 0.7362,
       "step": 1541
     },
     {
-      "epoch": 1.516732283464567,
-      "grad_norm": 9.229491233825684,
+      "epoch": 1.5177165354330708,
+      "grad_norm": 11.548934936523438,
       "learning_rate": 8.427545943570226e-07,
-      "loss": 0.4833,
+      "loss": 0.569,
       "step": 1542
     },
     {
-      "epoch": 1.5177165354330708,
-      "grad_norm": 13.765993118286133,
+      "epoch": 1.5187007874015748,
+      "grad_norm": 9.297774314880371,
       "learning_rate": 8.421952407297752e-07,
-      "loss": 0.8962,
+      "loss": 0.5174,
       "step": 1543
     },
     {
-      "epoch": 1.5187007874015748,
-      "grad_norm": 5.461718559265137,
+      "epoch": 1.5196850393700787,
+      "grad_norm": 12.480260848999023,
       "learning_rate": 8.416350810900892e-07,
-      "loss": 0.1782,
+      "loss": 0.797,
       "step": 1544
     },
     {
-      "epoch": 1.5196850393700787,
-      "grad_norm": 11.364346504211426,
+      "epoch": 1.5206692913385826,
+      "grad_norm": 12.73460865020752,
       "learning_rate": 8.410741167638317e-07,
-      "loss": 0.7229,
+      "loss": 0.7091,
       "step": 1545
     },
     {
-      "epoch": 1.5206692913385826,
-      "grad_norm": 8.603964805603027,
+      "epoch": 1.5216535433070866,
+      "grad_norm": 10.627336502075195,
       "learning_rate": 8.405123490787741e-07,
-      "loss": 0.3197,
+      "loss": 0.512,
       "step": 1546
     },
     {
-      "epoch": 1.5216535433070866,
-      "grad_norm": 12.854615211486816,
+      "epoch": 1.5226377952755905,
+      "grad_norm": 10.014727592468262,
       "learning_rate": 8.3994977936459e-07,
-      "loss": 0.7508,
+      "loss": 0.5039,
       "step": 1547
     },
     {
-      "epoch": 1.5226377952755905,
-      "grad_norm": 12.71799087524414,
+      "epoch": 1.5236220472440944,
+      "grad_norm": 14.352371215820312,
       "learning_rate": 8.393864089528514e-07,
-      "loss": 0.6946,
+      "loss": 0.9192,
       "step": 1548
     },
     {
-      "epoch": 1.5236220472440944,
-      "grad_norm": 7.034970760345459,
+      "epoch": 1.5246062992125984,
+      "grad_norm": 14.823103904724121,
       "learning_rate": 8.388222391770245e-07,
-      "loss": 0.3118,
+      "loss": 1.0385,
       "step": 1549
     },
     {
-      "epoch": 1.5246062992125984,
-      "grad_norm": 16.605329513549805,
+      "epoch": 1.5255905511811023,
+      "grad_norm": 10.62708854675293,
       "learning_rate": 8.382572713724683e-07,
-      "loss": 0.8608,
+      "loss": 0.4519,
       "step": 1550
     },
     {
-      "epoch": 1.5255905511811023,
-      "grad_norm": 15.885974884033203,
+      "epoch": 1.5265748031496063,
+      "grad_norm": 13.908456802368164,
       "learning_rate": 8.376915068764308e-07,
-      "loss": 0.7907,
+      "loss": 0.8472,
       "step": 1551
     },
     {
-      "epoch": 1.5265748031496063,
-      "grad_norm": 22.729400634765625,
+      "epoch": 1.5275590551181102,
+      "grad_norm": 13.704379081726074,
       "learning_rate": 8.371249470280453e-07,
-      "loss": 1.2424,
+      "loss": 0.8557,
       "step": 1552
     },
     {
-      "epoch": 1.5275590551181102,
-      "grad_norm": 10.601911544799805,
+      "epoch": 1.5285433070866141,
+      "grad_norm": 12.586782455444336,
       "learning_rate": 8.36557593168328e-07,
-      "loss": 0.5366,
+      "loss": 0.652,
       "step": 1553
     },
     {
-      "epoch": 1.5285433070866141,
-      "grad_norm": 10.668947219848633,
+      "epoch": 1.529527559055118,
+      "grad_norm": 13.087150573730469,
       "learning_rate": 8.359894466401743e-07,
-      "loss": 0.5525,
+      "loss": 0.8557,
       "step": 1554
     },
     {
-      "epoch": 1.529527559055118,
-      "grad_norm": 7.3187150955200195,
+      "epoch": 1.530511811023622,
+      "grad_norm": 11.236658096313477,
       "learning_rate": 8.354205087883556e-07,
-      "loss": 0.2904,
+      "loss": 0.5835,
       "step": 1555
     },
     {
-      "epoch": 1.530511811023622,
-      "grad_norm": 13.080991744995117,
+      "epoch": 1.531496062992126,
+      "grad_norm": 9.262578010559082,
       "learning_rate": 8.348507809595169e-07,
-      "loss": 0.7339,
+      "loss": 0.5346,
       "step": 1556
     },
     {
-      "epoch": 1.531496062992126,
-      "grad_norm": 7.837886333465576,
+      "epoch": 1.53248031496063,
+      "grad_norm": 11.856017112731934,
       "learning_rate": 8.342802645021727e-07,
-      "loss": 0.4061,
+      "loss": 0.5621,
       "step": 1557
     },
     {
-      "epoch": 1.53248031496063,
-      "grad_norm": 9.239282608032227,
+      "epoch": 1.5334645669291338,
+      "grad_norm": 10.604304313659668,
       "learning_rate": 8.337089607667039e-07,
-      "loss": 0.3988,
+      "loss": 0.6299,
       "step": 1558
     },
     {
-      "epoch": 1.5334645669291338,
-      "grad_norm": 14.300249099731445,
+      "epoch": 1.534448818897638,
+      "grad_norm": 13.338227272033691,
       "learning_rate": 8.331368711053552e-07,
-      "loss": 0.6063,
+      "loss": 1.0536,
       "step": 1559
     },
     {
-      "epoch": 1.534448818897638,
-      "grad_norm": 13.190995216369629,
+      "epoch": 1.5354330708661417,
+      "grad_norm": 12.48371696472168,
       "learning_rate": 8.325639968722313e-07,
-      "loss": 0.7394,
+      "loss": 0.676,
       "step": 1560
     },
     {
-      "epoch": 1.5354330708661417,
-      "grad_norm": 22.914588928222656,
+      "epoch": 1.5364173228346458,
+      "grad_norm": 14.458796501159668,
       "learning_rate": 8.319903394232942e-07,
-      "loss": 1.2118,
+      "loss": 1.15,
       "step": 1561
     },
     {
-      "epoch": 1.5364173228346458,
-      "grad_norm": 12.508071899414062,
+      "epoch": 1.5374015748031495,
+      "grad_norm": 9.366189002990723,
       "learning_rate": 8.314159001163596e-07,
-      "loss": 0.5494,
+      "loss": 0.7588,
       "step": 1562
     },
     {
-      "epoch": 1.5374015748031495,
-      "grad_norm": 10.416210174560547,
+      "epoch": 1.5383858267716537,
+      "grad_norm": 12.09666633605957,
       "learning_rate": 8.308406803110936e-07,
-      "loss": 0.4957,
+      "loss": 0.8398,
       "step": 1563
     },
     {
-      "epoch": 1.5383858267716537,
-      "grad_norm": 12.602110862731934,
+      "epoch": 1.5393700787401574,
+      "grad_norm": 10.621899604797363,
       "learning_rate": 8.302646813690101e-07,
-      "loss": 0.628,
+      "loss": 0.5209,
       "step": 1564
     },
     {
-      "epoch": 1.5393700787401574,
-      "grad_norm": 10.616961479187012,
+      "epoch": 1.5403543307086616,
+      "grad_norm": 10.583375930786133,
       "learning_rate": 8.296879046534668e-07,
-      "loss": 0.454,
+      "loss": 0.8099,
       "step": 1565
     },
     {
-      "epoch": 1.5403543307086616,
-      "grad_norm": 10.201837539672852,
+      "epoch": 1.5413385826771653,
+      "grad_norm": 9.15184497833252,
       "learning_rate": 8.291103515296628e-07,
-      "loss": 0.4168,
+      "loss": 0.6272,
       "step": 1566
     },
     {
-      "epoch": 1.5413385826771653,
-      "grad_norm": 10.088976860046387,
+      "epoch": 1.5423228346456694,
+      "grad_norm": 10.965386390686035,
       "learning_rate": 8.285320233646345e-07,
-      "loss": 0.4472,
+      "loss": 0.8252,
       "step": 1567
     },
     {
-      "epoch": 1.5423228346456694,
-      "grad_norm": 11.572919845581055,
+      "epoch": 1.5433070866141732,
+      "grad_norm": 12.962262153625488,
       "learning_rate": 8.279529215272526e-07,
-      "loss": 0.4394,
+      "loss": 0.9155,
       "step": 1568
     },
     {
-      "epoch": 1.5433070866141732,
-      "grad_norm": 24.046588897705078,
+      "epoch": 1.5442913385826773,
+      "grad_norm": 5.538671016693115,
       "learning_rate": 8.273730473882198e-07,
-      "loss": 1.1084,
+      "loss": 0.2462,
       "step": 1569
     },
     {
-      "epoch": 1.5442913385826773,
-      "grad_norm": 10.44725513458252,
+      "epoch": 1.545275590551181,
+      "grad_norm": 7.206732749938965,
       "learning_rate": 8.267924023200662e-07,
-      "loss": 0.3262,
+      "loss": 0.3308,
       "step": 1570
     },
     {
-      "epoch": 1.545275590551181,
-      "grad_norm": 8.531761169433594,
+      "epoch": 1.5462598425196852,
+      "grad_norm": 10.74162483215332,
       "learning_rate": 8.262109876971466e-07,
-      "loss": 0.3507,
+      "loss": 0.3919,
       "step": 1571
     },
     {
-      "epoch": 1.5462598425196852,
-      "grad_norm": 16.767059326171875,
+      "epoch": 1.547244094488189,
+      "grad_norm": 9.28885269165039,
       "learning_rate": 8.256288048956379e-07,
-      "loss": 1.0028,
+      "loss": 0.5583,
       "step": 1572
     },
     {
-      "epoch": 1.547244094488189,
-      "grad_norm": 9.508438110351562,
+      "epoch": 1.548228346456693,
+      "grad_norm": 12.702869415283203,
       "learning_rate": 8.250458552935343e-07,
-      "loss": 0.5218,
+      "loss": 0.6686,
       "step": 1573
     },
     {
-      "epoch": 1.548228346456693,
-      "grad_norm": 13.773395538330078,
+      "epoch": 1.5492125984251968,
+      "grad_norm": 10.253847122192383,
       "learning_rate": 8.244621402706462e-07,
-      "loss": 0.9723,
+      "loss": 0.4003,
       "step": 1574
     },
     {
-      "epoch": 1.5492125984251968,
-      "grad_norm": 14.080069541931152,
+      "epoch": 1.550196850393701,
+      "grad_norm": 11.96845531463623,
       "learning_rate": 8.238776612085943e-07,
-      "loss": 0.5616,
+      "loss": 0.7001,
       "step": 1575
     },
     {
-      "epoch": 1.550196850393701,
-      "grad_norm": 11.193357467651367,
+      "epoch": 1.5511811023622046,
+      "grad_norm": 11.859036445617676,
       "learning_rate": 8.232924194908088e-07,
-      "loss": 0.4887,
+      "loss": 0.9504,
       "step": 1576
     },
     {
-      "epoch": 1.5511811023622046,
-      "grad_norm": 9.106607437133789,
+      "epoch": 1.5521653543307088,
+      "grad_norm": 10.334456443786621,
       "learning_rate": 8.227064165025247e-07,
-      "loss": 0.3899,
+      "loss": 0.453,
       "step": 1577
     },
     {
-      "epoch": 1.5521653543307088,
-      "grad_norm": 8.196263313293457,
+      "epoch": 1.5531496062992125,
+      "grad_norm": 10.861915588378906,
       "learning_rate": 8.221196536307788e-07,
-      "loss": 0.3545,
+      "loss": 0.5156,
       "step": 1578
     },
     {
-      "epoch": 1.5531496062992125,
-      "grad_norm": 9.231218338012695,
+      "epoch": 1.5541338582677167,
+      "grad_norm": 8.276926040649414,
       "learning_rate": 8.215321322644067e-07,
-      "loss": 0.3574,
+      "loss": 0.363,
       "step": 1579
     },
     {
-      "epoch": 1.5541338582677167,
-      "grad_norm": 22.20115852355957,
+      "epoch": 1.5551181102362204,
+      "grad_norm": 13.232145309448242,
       "learning_rate": 8.209438537940393e-07,
-      "loss": 1.3911,
+      "loss": 0.9124,
       "step": 1580
     },
     {
-      "epoch": 1.5551181102362204,
-      "grad_norm": 11.824051856994629,
+      "epoch": 1.5561023622047245,
+      "grad_norm": 12.50605583190918,
       "learning_rate": 8.203548196120994e-07,
-      "loss": 0.7395,
+      "loss": 0.5979,
       "step": 1581
     },
     {
-      "epoch": 1.5561023622047245,
-      "grad_norm": 9.031845092773438,
+      "epoch": 1.5570866141732282,
+      "grad_norm": 10.92392635345459,
       "learning_rate": 8.197650311127981e-07,
-      "loss": 0.2491,
+      "loss": 0.7076,
       "step": 1582
     },
     {
-      "epoch": 1.5570866141732282,
-      "grad_norm": 10.274392127990723,
+      "epoch": 1.5580708661417324,
+      "grad_norm": 6.655268669128418,
       "learning_rate": 8.191744896921331e-07,
-      "loss": 0.6825,
+      "loss": 0.272,
       "step": 1583
     },
     {
-      "epoch": 1.5580708661417324,
-      "grad_norm": 5.4450154304504395,
+      "epoch": 1.5590551181102361,
+      "grad_norm": 13.023505210876465,
       "learning_rate": 8.185831967478831e-07,
-      "loss": 0.1434,
+      "loss": 0.7369,
       "step": 1584
     },
     {
-      "epoch": 1.5590551181102361,
-      "grad_norm": 11.918195724487305,
+      "epoch": 1.5600393700787403,
+      "grad_norm": 12.396964073181152,
       "learning_rate": 8.179911536796062e-07,
-      "loss": 0.3764,
+      "loss": 0.66,
       "step": 1585
     },
     {
-      "epoch": 1.5600393700787403,
-      "grad_norm": 11.091309547424316,
+      "epoch": 1.561023622047244,
+      "grad_norm": 10.249530792236328,
       "learning_rate": 8.173983618886358e-07,
-      "loss": 0.5378,
+      "loss": 0.4354,
       "step": 1586
     },
     {
-      "epoch": 1.561023622047244,
-      "grad_norm": 8.367671966552734,
+      "epoch": 1.5620078740157481,
+      "grad_norm": 11.370365142822266,
       "learning_rate": 8.168048227780775e-07,
-      "loss": 0.2975,
+      "loss": 0.8265,
       "step": 1587
     },
     {
-      "epoch": 1.5620078740157481,
-      "grad_norm": 15.36435604095459,
+      "epoch": 1.5629921259842519,
+      "grad_norm": 10.647195816040039,
       "learning_rate": 8.162105377528057e-07,
-      "loss": 1.0245,
+      "loss": 0.4936,
       "step": 1588
     },
     {
-      "epoch": 1.5629921259842519,
-      "grad_norm": 14.115655899047852,
+      "epoch": 1.563976377952756,
+      "grad_norm": 9.339645385742188,
       "learning_rate": 8.156155082194607e-07,
-      "loss": 0.6128,
+      "loss": 0.4236,
       "step": 1589
     },
     {
-      "epoch": 1.563976377952756,
-      "grad_norm": 17.05628776550293,
+      "epoch": 1.5649606299212597,
+      "grad_norm": 8.706368446350098,
       "learning_rate": 8.150197355864446e-07,
-      "loss": 0.89,
+      "loss": 0.5496,
       "step": 1590
     },
     {
-      "epoch": 1.5649606299212597,
-      "grad_norm": 14.739027976989746,
+      "epoch": 1.5659448818897639,
+      "grad_norm": 9.876303672790527,
       "learning_rate": 8.144232212639184e-07,
-      "loss": 0.8377,
+      "loss": 0.545,
       "step": 1591
     },
     {
-      "epoch": 1.5659448818897639,
-      "grad_norm": 21.410964965820312,
+      "epoch": 1.5669291338582676,
+      "grad_norm": 14.000279426574707,
       "learning_rate": 8.138259666637987e-07,
-      "loss": 0.7711,
+      "loss": 0.804,
       "step": 1592
     },
     {
-      "epoch": 1.5669291338582676,
-      "grad_norm": 14.888060569763184,
+      "epoch": 1.5679133858267718,
+      "grad_norm": 11.417315483093262,
       "learning_rate": 8.132279731997544e-07,
-      "loss": 0.8045,
+      "loss": 0.6017,
       "step": 1593
     },
     {
-      "epoch": 1.5679133858267718,
-      "grad_norm": 11.131561279296875,
+      "epoch": 1.5688976377952755,
+      "grad_norm": 8.553269386291504,
       "learning_rate": 8.126292422872034e-07,
-      "loss": 0.5162,
+      "loss": 0.5229,
       "step": 1594
     },
     {
-      "epoch": 1.5688976377952755,
-      "grad_norm": 12.947890281677246,
+      "epoch": 1.5698818897637796,
+      "grad_norm": 7.077866554260254,
       "learning_rate": 8.120297753433087e-07,
-      "loss": 0.5009,
+      "loss": 0.3998,
       "step": 1595
     },
     {
-      "epoch": 1.5698818897637796,
-      "grad_norm": 16.48666763305664,
+      "epoch": 1.5708661417322833,
+      "grad_norm": 10.789830207824707,
       "learning_rate": 8.114295737869759e-07,
-      "loss": 0.6454,
+      "loss": 0.6741,
       "step": 1596
     },
     {
-      "epoch": 1.5708661417322833,
-      "grad_norm": 13.440455436706543,
+      "epoch": 1.5718503937007875,
+      "grad_norm": 9.581600189208984,
       "learning_rate": 8.108286390388487e-07,
-      "loss": 0.4929,
+      "loss": 0.4781,
       "step": 1597
     },
     {
-      "epoch": 1.5718503937007875,
-      "grad_norm": 9.928903579711914,
+      "epoch": 1.5728346456692912,
+      "grad_norm": 10.937962532043457,
       "learning_rate": 8.102269725213071e-07,
-      "loss": 0.4363,
+      "loss": 0.4971,
       "step": 1598
     },
     {
-      "epoch": 1.5728346456692912,
-      "grad_norm": 13.543412208557129,
+      "epoch": 1.5738188976377954,
+      "grad_norm": 10.09300708770752,
       "learning_rate": 8.096245756584625e-07,
-      "loss": 0.4686,
+      "loss": 0.6007,
       "step": 1599
     },
     {
-      "epoch": 1.5738188976377954,
-      "grad_norm": 9.794742584228516,
+      "epoch": 1.574803149606299,
+      "grad_norm": 12.481731414794922,
       "learning_rate": 8.090214498761551e-07,
-      "loss": 0.3621,
+      "loss": 0.8647,
       "step": 1600
     },
     {
-      "epoch": 1.574803149606299,
-      "grad_norm": 10.056382179260254,
+      "epoch": 1.5757874015748032,
+      "grad_norm": 13.888327598571777,
       "learning_rate": 8.084175966019507e-07,
-      "loss": 0.5088,
+      "loss": 1.0286,
       "step": 1601
     },
     {
-      "epoch": 1.5757874015748032,
-      "grad_norm": 12.101970672607422,
+      "epoch": 1.576771653543307,
+      "grad_norm": 9.011124610900879,
       "learning_rate": 8.078130172651368e-07,
-      "loss": 0.5757,
+      "loss": 0.4067,
       "step": 1602
     },
     {
-      "epoch": 1.576771653543307,
-      "grad_norm": 11.040324211120605,
+      "epoch": 1.577755905511811,
+      "grad_norm": 13.957579612731934,
       "learning_rate": 8.072077132967193e-07,
-      "loss": 0.5269,
+      "loss": 0.731,
       "step": 1603
     },
     {
-      "epoch": 1.577755905511811,
-      "grad_norm": 14.213335037231445,
+      "epoch": 1.5787401574803148,
+      "grad_norm": 8.332479476928711,
       "learning_rate": 8.066016861294198e-07,
-      "loss": 0.7907,
+      "loss": 0.2758,
       "step": 1604
     },
     {
-      "epoch": 1.5787401574803148,
-      "grad_norm": 12.529051780700684,
+      "epoch": 1.579724409448819,
+      "grad_norm": 11.11207389831543,
       "learning_rate": 8.059949371976711e-07,
-      "loss": 0.5693,
+      "loss": 0.3977,
       "step": 1605
     },
     {
-      "epoch": 1.579724409448819,
-      "grad_norm": 8.595606803894043,
+      "epoch": 1.5807086614173227,
+      "grad_norm": 11.47100830078125,
       "learning_rate": 8.053874679376147e-07,
-      "loss": 0.4048,
+      "loss": 0.6459,
       "step": 1606
     },
     {
-      "epoch": 1.5807086614173227,
-      "grad_norm": 9.17170524597168,
+      "epoch": 1.5816929133858268,
+      "grad_norm": 13.429177284240723,
       "learning_rate": 8.047792797870968e-07,
-      "loss": 0.4727,
+      "loss": 0.9274,
       "step": 1607
     },
     {
-      "epoch": 1.5816929133858268,
-      "grad_norm": 10.802497863769531,
+      "epoch": 1.5826771653543306,
+      "grad_norm": 15.43608570098877,
       "learning_rate": 8.041703741856657e-07,
-      "loss": 0.5753,
+      "loss": 0.74,
       "step": 1608
     },
     {
-      "epoch": 1.5826771653543306,
-      "grad_norm": 9.498733520507812,
+      "epoch": 1.5836614173228347,
+      "grad_norm": 12.196651458740234,
       "learning_rate": 8.035607525745672e-07,
-      "loss": 0.554,
+      "loss": 0.3715,
       "step": 1609
     },
     {
-      "epoch": 1.5836614173228347,
-      "grad_norm": 10.088098526000977,
+      "epoch": 1.5846456692913384,
+      "grad_norm": 11.486495018005371,
       "learning_rate": 8.029504163967423e-07,
-      "loss": 0.5813,
+      "loss": 0.6587,
       "step": 1610
     },
     {
-      "epoch": 1.5846456692913384,
-      "grad_norm": 31.178342819213867,
+      "epoch": 1.5856299212598426,
+      "grad_norm": 11.401627540588379,
       "learning_rate": 8.023393670968235e-07,
-      "loss": 2.0048,
+      "loss": 0.5264,
       "step": 1611
     },
     {
-      "epoch": 1.5856299212598426,
-      "grad_norm": 20.6037540435791,
+      "epoch": 1.5866141732283463,
+      "grad_norm": 11.985355377197266,
       "learning_rate": 8.017276061211303e-07,
-      "loss": 0.9611,
+      "loss": 0.8698,
       "step": 1612
     },
     {
-      "epoch": 1.5866141732283463,
-      "grad_norm": 11.645776748657227,
+      "epoch": 1.5875984251968505,
+      "grad_norm": 12.371281623840332,
       "learning_rate": 8.011151349176681e-07,
-      "loss": 0.7653,
+      "loss": 0.6965,
       "step": 1613
     },
     {
-      "epoch": 1.5875984251968505,
-      "grad_norm": 14.203062057495117,
+      "epoch": 1.5885826771653542,
+      "grad_norm": 12.63413143157959,
       "learning_rate": 8.005019549361222e-07,
-      "loss": 0.8667,
+      "loss": 0.8431,
       "step": 1614
     },
     {
-      "epoch": 1.5885826771653542,
-      "grad_norm": 12.750677108764648,
+      "epoch": 1.5895669291338583,
+      "grad_norm": 11.72999382019043,
       "learning_rate": 7.998880676278558e-07,
-      "loss": 0.7168,
+      "loss": 0.5191,
       "step": 1615
     },
     {
-      "epoch": 1.5895669291338583,
-      "grad_norm": 9.95725154876709,
+      "epoch": 1.590551181102362,
+      "grad_norm": 16.434574127197266,
       "learning_rate": 7.992734744459068e-07,
-      "loss": 0.6818,
+      "loss": 1.1508,
       "step": 1616
     },
     {
-      "epoch": 1.590551181102362,
-      "grad_norm": 12.161149024963379,
+      "epoch": 1.5915354330708662,
+      "grad_norm": 16.757034301757812,
       "learning_rate": 7.986581768449833e-07,
-      "loss": 0.8315,
+      "loss": 1.2072,
       "step": 1617
     },
     {
-      "epoch": 1.5915354330708662,
-      "grad_norm": 17.93672752380371,
+      "epoch": 1.59251968503937,
+      "grad_norm": 12.101422309875488,
       "learning_rate": 7.98042176281461e-07,
-      "loss": 0.9919,
+      "loss": 0.5955,
       "step": 1618
     },
     {
-      "epoch": 1.59251968503937,
-      "grad_norm": 17.684009552001953,
+      "epoch": 1.593503937007874,
+      "grad_norm": 12.935314178466797,
       "learning_rate": 7.974254742133795e-07,
-      "loss": 0.7954,
+      "loss": 0.9112,
       "step": 1619
     },
     {
-      "epoch": 1.593503937007874,
-      "grad_norm": 19.807653427124023,
+      "epoch": 1.594488188976378,
+      "grad_norm": 9.056806564331055,
       "learning_rate": 7.968080721004386e-07,
-      "loss": 0.9298,
+      "loss": 0.2759,
       "step": 1620
     },
     {
-      "epoch": 1.594488188976378,
-      "grad_norm": 9.452728271484375,
+      "epoch": 1.595472440944882,
+      "grad_norm": 8.368066787719727,
       "learning_rate": 7.961899714039954e-07,
-      "loss": 0.3301,
+      "loss": 0.3118,
       "step": 1621
     },
     {
-      "epoch": 1.595472440944882,
-      "grad_norm": 11.764351844787598,
+      "epoch": 1.5964566929133859,
+      "grad_norm": 14.533076286315918,
       "learning_rate": 7.955711735870601e-07,
-      "loss": 0.468,
+      "loss": 1.0573,
       "step": 1622
     },
     {
-      "epoch": 1.5964566929133859,
-      "grad_norm": 18.4998836517334,
+      "epoch": 1.5974409448818898,
+      "grad_norm": 13.674250602722168,
       "learning_rate": 7.949516801142932e-07,
-      "loss": 1.0635,
+      "loss": 0.9392,
       "step": 1623
     },
     {
-      "epoch": 1.5974409448818898,
-      "grad_norm": 11.700940132141113,
+      "epoch": 1.5984251968503937,
+      "grad_norm": 9.712000846862793,
       "learning_rate": 7.943314924520017e-07,
-      "loss": 0.6376,
+      "loss": 0.6772,
       "step": 1624
     },
     {
-      "epoch": 1.5984251968503937,
-      "grad_norm": 12.373494148254395,
+      "epoch": 1.5994094488188977,
+      "grad_norm": 9.423108100891113,
       "learning_rate": 7.93710612068136e-07,
-      "loss": 0.8462,
+      "loss": 0.6695,
       "step": 1625
     },
     {
-      "epoch": 1.5994094488188977,
-      "grad_norm": 11.433577537536621,
+      "epoch": 1.6003937007874016,
+      "grad_norm": 11.593213081359863,
       "learning_rate": 7.930890404322857e-07,
-      "loss": 0.6499,
+      "loss": 0.6276,
       "step": 1626
     },
     {
-      "epoch": 1.6003937007874016,
-      "grad_norm": 11.250558853149414,
+      "epoch": 1.6013779527559056,
+      "grad_norm": 11.381474494934082,
       "learning_rate": 7.924667790156766e-07,
-      "loss": 0.5233,
+      "loss": 0.527,
       "step": 1627
     },
     {
-      "epoch": 1.6013779527559056,
-      "grad_norm": 13.41759204864502,
+      "epoch": 1.6023622047244095,
+      "grad_norm": 8.85727310180664,
       "learning_rate": 7.918438292911674e-07,
-      "loss": 0.5169,
+      "loss": 0.4637,
       "step": 1628
     },
     {
-      "epoch": 1.6023622047244095,
-      "grad_norm": 17.72492218017578,
+      "epoch": 1.6033464566929134,
+      "grad_norm": 8.845136642456055,
       "learning_rate": 7.912201927332461e-07,
-      "loss": 0.8332,
+      "loss": 0.4799,
       "step": 1629
     },
     {
-      "epoch": 1.6033464566929134,
-      "grad_norm": 6.448797225952148,
+      "epoch": 1.6043307086614174,
+      "grad_norm": 11.491124153137207,
       "learning_rate": 7.905958708180258e-07,
-      "loss": 0.1964,
+      "loss": 0.5972,
       "step": 1630
     },
     {
-      "epoch": 1.6043307086614174,
-      "grad_norm": 17.034406661987305,
+      "epoch": 1.6053149606299213,
+      "grad_norm": 14.663311958312988,
       "learning_rate": 7.899708650232426e-07,
-      "loss": 0.914,
+      "loss": 0.667,
       "step": 1631
     },
     {
-      "epoch": 1.6053149606299213,
-      "grad_norm": 11.307120323181152,
+      "epoch": 1.6062992125984252,
+      "grad_norm": 8.870426177978516,
       "learning_rate": 7.893451768282508e-07,
-      "loss": 0.4956,
+      "loss": 0.5926,
       "step": 1632
     },
     {
-      "epoch": 1.6062992125984252,
-      "grad_norm": 9.852806091308594,
+      "epoch": 1.6072834645669292,
+      "grad_norm": 11.694674491882324,
       "learning_rate": 7.8871880771402e-07,
-      "loss": 0.5823,
+      "loss": 0.6911,
       "step": 1633
     },
     {
-      "epoch": 1.6072834645669292,
-      "grad_norm": 13.381124496459961,
+      "epoch": 1.608267716535433,
+      "grad_norm": 14.012720108032227,
       "learning_rate": 7.880917591631312e-07,
-      "loss": 0.9373,
+      "loss": 0.8098,
       "step": 1634
     },
     {
-      "epoch": 1.608267716535433,
-      "grad_norm": 14.188089370727539,
+      "epoch": 1.609251968503937,
+      "grad_norm": 14.760422706604004,
       "learning_rate": 7.874640326597741e-07,
-      "loss": 0.879,
+      "loss": 0.8524,
       "step": 1635
     },
     {
-      "epoch": 1.609251968503937,
-      "grad_norm": 11.136621475219727,
+      "epoch": 1.610236220472441,
+      "grad_norm": 9.076541900634766,
       "learning_rate": 7.868356296897431e-07,
-      "loss": 0.8167,
+      "loss": 0.3731,
       "step": 1636
     },
     {
-      "epoch": 1.610236220472441,
-      "grad_norm": 19.828542709350586,
+      "epoch": 1.611220472440945,
+      "grad_norm": 11.972025871276855,
       "learning_rate": 7.862065517404334e-07,
-      "loss": 0.8329,
+      "loss": 0.6751,
       "step": 1637
     },
     {
-      "epoch": 1.611220472440945,
-      "grad_norm": 10.303707122802734,
+      "epoch": 1.6122047244094488,
+      "grad_norm": 14.013944625854492,
       "learning_rate": 7.855768003008376e-07,
-      "loss": 0.5512,
+      "loss": 0.9858,
       "step": 1638
     },
     {
-      "epoch": 1.6122047244094488,
-      "grad_norm": 11.766749382019043,
+      "epoch": 1.6131889763779528,
+      "grad_norm": 10.247302055358887,
       "learning_rate": 7.849463768615432e-07,
-      "loss": 0.6024,
+      "loss": 0.439,
       "step": 1639
     },
     {
-      "epoch": 1.6131889763779528,
-      "grad_norm": 15.314605712890625,
+      "epoch": 1.6141732283464567,
+      "grad_norm": 7.529063701629639,
       "learning_rate": 7.843152829147278e-07,
-      "loss": 0.966,
+      "loss": 0.2996,
       "step": 1640
     },
     {
-      "epoch": 1.6141732283464567,
-      "grad_norm": 11.736525535583496,
+      "epoch": 1.6151574803149606,
+      "grad_norm": 5.548278331756592,
       "learning_rate": 7.836835199541562e-07,
-      "loss": 0.6611,
+      "loss": 0.2333,
       "step": 1641
     },
     {
-      "epoch": 1.6151574803149606,
-      "grad_norm": 4.6410369873046875,
+      "epoch": 1.6161417322834646,
+      "grad_norm": 9.582456588745117,
       "learning_rate": 7.830510894751764e-07,
-      "loss": 0.1169,
+      "loss": 0.4415,
       "step": 1642
     },
     {
-      "epoch": 1.6161417322834646,
-      "grad_norm": 9.750079154968262,
+      "epoch": 1.6171259842519685,
+      "grad_norm": 11.477483749389648,
       "learning_rate": 7.824179929747168e-07,
-      "loss": 0.4613,
+      "loss": 0.7798,
       "step": 1643
     },
     {
-      "epoch": 1.6171259842519685,
-      "grad_norm": 17.277965545654297,
+      "epoch": 1.6181102362204725,
+      "grad_norm": 11.246156692504883,
       "learning_rate": 7.817842319512818e-07,
-      "loss": 1.3485,
+      "loss": 0.4089,
       "step": 1644
     },
     {
-      "epoch": 1.6181102362204725,
-      "grad_norm": 10.674585342407227,
+      "epoch": 1.6190944881889764,
+      "grad_norm": 13.863037109375,
       "learning_rate": 7.811498079049492e-07,
-      "loss": 0.4776,
+      "loss": 0.9624,
       "step": 1645
     },
     {
-      "epoch": 1.6190944881889764,
-      "grad_norm": 14.96063232421875,
+      "epoch": 1.6200787401574803,
+      "grad_norm": 13.357006072998047,
       "learning_rate": 7.805147223373658e-07,
-      "loss": 1.061,
+      "loss": 0.6725,
       "step": 1646
     },
     {
-      "epoch": 1.6200787401574803,
-      "grad_norm": 8.888649940490723,
+      "epoch": 1.6210629921259843,
+      "grad_norm": 8.112847328186035,
       "learning_rate": 7.798789767517444e-07,
-      "loss": 0.4102,
+      "loss": 0.4314,
       "step": 1647
     },
     {
-      "epoch": 1.6210629921259843,
-      "grad_norm": 11.162688255310059,
+      "epoch": 1.6220472440944882,
+      "grad_norm": 10.525714874267578,
       "learning_rate": 7.792425726528596e-07,
-      "loss": 0.5991,
+      "loss": 0.9129,
       "step": 1648
     },
     {
-      "epoch": 1.6220472440944882,
-      "grad_norm": 12.890787124633789,
+      "epoch": 1.6230314960629921,
+      "grad_norm": 12.334314346313477,
       "learning_rate": 7.786055115470454e-07,
-      "loss": 0.9039,
+      "loss": 0.8468,
       "step": 1649
     },
     {
-      "epoch": 1.6230314960629921,
-      "grad_norm": 11.972373962402344,
+      "epoch": 1.624015748031496,
+      "grad_norm": 10.935612678527832,
       "learning_rate": 7.779677949421902e-07,
-      "loss": 0.7915,
+      "loss": 0.6209,
       "step": 1650
     },
     {
-      "epoch": 1.624015748031496,
-      "grad_norm": 9.724061965942383,
+      "epoch": 1.625,
+      "grad_norm": 8.299671173095703,
       "learning_rate": 7.77329424347734e-07,
-      "loss": 0.5604,
+      "loss": 0.5043,
       "step": 1651
     },
     {
-      "epoch": 1.625,
-      "grad_norm": 8.689451217651367,
+      "epoch": 1.625984251968504,
+      "grad_norm": 9.101208686828613,
       "learning_rate": 7.766904012746653e-07,
-      "loss": 0.2933,
+      "loss": 0.4109,
       "step": 1652
     },
     {
-      "epoch": 1.625984251968504,
-      "grad_norm": 10.402259826660156,
+      "epoch": 1.6269685039370079,
+      "grad_norm": 12.939144134521484,
       "learning_rate": 7.760507272355165e-07,
-      "loss": 0.6606,
+      "loss": 0.7351,
       "step": 1653
     },
     {
-      "epoch": 1.6269685039370079,
-      "grad_norm": 11.47604751586914,
+      "epoch": 1.6279527559055118,
+      "grad_norm": 13.521783828735352,
       "learning_rate": 7.754104037443609e-07,
-      "loss": 0.6412,
+      "loss": 0.8728,
       "step": 1654
     },
     {
-      "epoch": 1.6279527559055118,
-      "grad_norm": 9.512560844421387,
+      "epoch": 1.6289370078740157,
+      "grad_norm": 8.386841773986816,
       "learning_rate": 7.747694323168091e-07,
-      "loss": 0.5141,
+      "loss": 0.4225,
       "step": 1655
     },
     {
-      "epoch": 1.6289370078740157,
-      "grad_norm": 14.507397651672363,
+      "epoch": 1.6299212598425197,
+      "grad_norm": 13.039405822753906,
       "learning_rate": 7.741278144700052e-07,
-      "loss": 1.0262,
+      "loss": 0.8833,
       "step": 1656
     },
     {
-      "epoch": 1.6299212598425197,
-      "grad_norm": 10.463685989379883,
+      "epoch": 1.6309055118110236,
+      "grad_norm": 7.570183277130127,
       "learning_rate": 7.734855517226234e-07,
-      "loss": 0.535,
+      "loss": 0.2562,
       "step": 1657
     },
     {
-      "epoch": 1.6309055118110236,
-      "grad_norm": 13.42270565032959,
+      "epoch": 1.6318897637795275,
+      "grad_norm": 12.169554710388184,
       "learning_rate": 7.728426455948644e-07,
-      "loss": 0.6027,
+      "loss": 0.6762,
       "step": 1658
     },
     {
-      "epoch": 1.6318897637795275,
-      "grad_norm": 9.59408950805664,
+      "epoch": 1.6328740157480315,
+      "grad_norm": 7.591965675354004,
       "learning_rate": 7.721990976084517e-07,
-      "loss": 0.6127,
+      "loss": 0.2752,
       "step": 1659
     },
     {
-      "epoch": 1.6328740157480315,
-      "grad_norm": 17.450523376464844,
+      "epoch": 1.6338582677165354,
+      "grad_norm": 10.623973846435547,
       "learning_rate": 7.715549092866282e-07,
-      "loss": 1.2897,
+      "loss": 0.426,
       "step": 1660
     },
     {
-      "epoch": 1.6338582677165354,
-      "grad_norm": 8.003664016723633,
+      "epoch": 1.6348425196850394,
+      "grad_norm": 12.923746109008789,
       "learning_rate": 7.70910082154152e-07,
-      "loss": 0.3319,
+      "loss": 0.95,
       "step": 1661
     },
     {
-      "epoch": 1.6348425196850394,
-      "grad_norm": 11.838500022888184,
+      "epoch": 1.6358267716535433,
+      "grad_norm": 13.924927711486816,
       "learning_rate": 7.702646177372938e-07,
-      "loss": 1.0125,
+      "loss": 0.9488,
       "step": 1662
     },
     {
-      "epoch": 1.6358267716535433,
-      "grad_norm": 19.441804885864258,
+      "epoch": 1.6368110236220472,
+      "grad_norm": 8.399730682373047,
       "learning_rate": 7.696185175638325e-07,
-      "loss": 1.0042,
+      "loss": 0.3755,
       "step": 1663
     },
     {
-      "epoch": 1.6368110236220472,
-      "grad_norm": 13.763493537902832,
+      "epoch": 1.6377952755905512,
+      "grad_norm": 8.661364555358887,
       "learning_rate": 7.689717831630513e-07,
-      "loss": 0.6483,
+      "loss": 0.3523,
       "step": 1664
     },
     {
-      "epoch": 1.6377952755905512,
-      "grad_norm": 14.2445068359375,
+      "epoch": 1.638779527559055,
+      "grad_norm": 12.495870590209961,
       "learning_rate": 7.683244160657353e-07,
-      "loss": 0.6933,
+      "loss": 0.6591,
       "step": 1665
     },
     {
-      "epoch": 1.638779527559055,
-      "grad_norm": 5.94786262512207,
+      "epoch": 1.639763779527559,
+      "grad_norm": 12.014862060546875,
       "learning_rate": 7.676764178041668e-07,
-      "loss": 0.2073,
+      "loss": 0.5441,
       "step": 1666
     },
     {
-      "epoch": 1.639763779527559,
-      "grad_norm": 9.012726783752441,
+      "epoch": 1.640748031496063,
+      "grad_norm": 14.289363861083984,
       "learning_rate": 7.670277899121223e-07,
-      "loss": 0.3838,
+      "loss": 0.7484,
       "step": 1667
     },
     {
-      "epoch": 1.640748031496063,
-      "grad_norm": 11.752018928527832,
+      "epoch": 1.641732283464567,
+      "grad_norm": 12.573966026306152,
       "learning_rate": 7.663785339248683e-07,
-      "loss": 0.6676,
+      "loss": 0.8561,
       "step": 1668
     },
     {
-      "epoch": 1.641732283464567,
-      "grad_norm": 11.327059745788574,
+      "epoch": 1.6427165354330708,
+      "grad_norm": 8.506410598754883,
       "learning_rate": 7.657286513791578e-07,
-      "loss": 0.5122,
+      "loss": 0.4679,
       "step": 1669
     },
     {
-      "epoch": 1.6427165354330708,
-      "grad_norm": 12.427508354187012,
+      "epoch": 1.6437007874015748,
+      "grad_norm": 11.476618766784668,
       "learning_rate": 7.650781438132276e-07,
-      "loss": 0.8016,
+      "loss": 0.5952,
       "step": 1670
     },
     {
-      "epoch": 1.6437007874015748,
-      "grad_norm": 8.388118743896484,
+      "epoch": 1.6446850393700787,
+      "grad_norm": 10.105586051940918,
       "learning_rate": 7.644270127667932e-07,
-      "loss": 0.3227,
+      "loss": 0.4511,
       "step": 1671
     },
     {
-      "epoch": 1.6446850393700787,
-      "grad_norm": 3.641969919204712,
+      "epoch": 1.6456692913385826,
+      "grad_norm": 11.244239807128906,
       "learning_rate": 7.637752597810456e-07,
-      "loss": 0.1016,
+      "loss": 0.7256,
       "step": 1672
     },
     {
-      "epoch": 1.6456692913385826,
-      "grad_norm": 7.827102184295654,
+      "epoch": 1.6466535433070866,
+      "grad_norm": 9.892069816589355,
       "learning_rate": 7.631228863986491e-07,
-      "loss": 0.1843,
+      "loss": 0.4359,
       "step": 1673
     },
     {
-      "epoch": 1.6466535433070866,
-      "grad_norm": 11.204242706298828,
+      "epoch": 1.6476377952755905,
+      "grad_norm": 11.57856273651123,
       "learning_rate": 7.624698941637352e-07,
-      "loss": 0.4915,
+      "loss": 0.71,
       "step": 1674
     },
     {
-      "epoch": 1.6476377952755905,
-      "grad_norm": 12.071660995483398,
+      "epoch": 1.6486220472440944,
+      "grad_norm": 9.806062698364258,
       "learning_rate": 7.618162846219006e-07,
-      "loss": 0.8414,
+      "loss": 0.5335,
       "step": 1675
     },
     {
-      "epoch": 1.6486220472440944,
-      "grad_norm": 10.194080352783203,
+      "epoch": 1.6496062992125984,
+      "grad_norm": 11.548669815063477,
       "learning_rate": 7.611620593202037e-07,
-      "loss": 0.493,
+      "loss": 0.6162,
       "step": 1676
     },
     {
-      "epoch": 1.6496062992125984,
-      "grad_norm": 12.145285606384277,
+      "epoch": 1.6505905511811023,
+      "grad_norm": 9.03017520904541,
       "learning_rate": 7.605072198071592e-07,
-      "loss": 0.9144,
+      "loss": 0.4306,
       "step": 1677
     },
     {
-      "epoch": 1.6505905511811023,
-      "grad_norm": 24.41216468811035,
+      "epoch": 1.6515748031496063,
+      "grad_norm": 8.461210250854492,
       "learning_rate": 7.598517676327368e-07,
-      "loss": 1.6354,
+      "loss": 0.3156,
       "step": 1678
     },
     {
-      "epoch": 1.6515748031496063,
-      "grad_norm": 23.160465240478516,
+      "epoch": 1.6525590551181102,
+      "grad_norm": 10.077200889587402,
       "learning_rate": 7.591957043483555e-07,
-      "loss": 2.1757,
+      "loss": 0.5425,
       "step": 1679
     },
     {
-      "epoch": 1.6525590551181102,
-      "grad_norm": 10.578500747680664,
+      "epoch": 1.6535433070866141,
+      "grad_norm": 10.720576286315918,
       "learning_rate": 7.585390315068812e-07,
-      "loss": 0.6866,
+      "loss": 0.5371,
       "step": 1680
     },
     {
-      "epoch": 1.6535433070866141,
-      "grad_norm": 10.035861015319824,
+      "epoch": 1.654527559055118,
+      "grad_norm": 10.244686126708984,
       "learning_rate": 7.578817506626224e-07,
-      "loss": 0.4759,
+      "loss": 0.5377,
       "step": 1681
     },
     {
-      "epoch": 1.654527559055118,
-      "grad_norm": 9.562938690185547,
+      "epoch": 1.655511811023622,
+      "grad_norm": 6.805355072021484,
       "learning_rate": 7.572238633713268e-07,
-      "loss": 0.4671,
+      "loss": 0.3161,
       "step": 1682
     },
     {
-      "epoch": 1.655511811023622,
-      "grad_norm": 9.952853202819824,
+      "epoch": 1.656496062992126,
+      "grad_norm": 21.622095108032227,
       "learning_rate": 7.565653711901778e-07,
-      "loss": 0.474,
+      "loss": 1.1868,
       "step": 1683
     },
     {
-      "epoch": 1.655511811023622,
-      "eval_Qnli-dev_cosine_accuracy": 0.662109375,
-      "eval_Qnli-dev_cosine_accuracy_threshold": 0.8541384935379028,
-      "eval_Qnli-dev_cosine_ap": 0.6939321999394883,
-      "eval_Qnli-dev_cosine_f1": 0.6912,
-      "eval_Qnli-dev_cosine_f1_threshold": 0.748054027557373,
-      "eval_Qnli-dev_cosine_precision": 0.5552699228791774,
-      "eval_Qnli-dev_cosine_recall": 0.9152542372881356,
-      "eval_Qnli-dev_dot_accuracy": 0.662109375,
-      "eval_Qnli-dev_dot_accuracy_threshold": 652.1033935546875,
-      "eval_Qnli-dev_dot_ap": 0.6900299415325823,
-      "eval_Qnli-dev_dot_f1": 0.6900958466453675,
-      "eval_Qnli-dev_dot_f1_threshold": 568.281982421875,
-      "eval_Qnli-dev_dot_precision": 0.5538461538461539,
-      "eval_Qnli-dev_dot_recall": 0.9152542372881356,
+      "epoch": 1.656496062992126,
+      "eval_Qnli-dev_cosine_accuracy": 0.66015625,
+      "eval_Qnli-dev_cosine_accuracy_threshold": 0.8450385332107544,
+      "eval_Qnli-dev_cosine_ap": 0.6877296306406283,
+      "eval_Qnli-dev_cosine_f1": 0.6910299003322258,
+      "eval_Qnli-dev_cosine_f1_threshold": 0.7468088269233704,
+      "eval_Qnli-dev_cosine_precision": 0.5683060109289617,
+      "eval_Qnli-dev_cosine_recall": 0.8813559322033898,
+      "eval_Qnli-dev_dot_accuracy": 0.654296875,
+      "eval_Qnli-dev_dot_accuracy_threshold": 639.7537231445312,
+      "eval_Qnli-dev_dot_ap": 0.6842454894163914,
+      "eval_Qnli-dev_dot_f1": 0.6900000000000001,
+      "eval_Qnli-dev_dot_f1_threshold": 570.31103515625,
+      "eval_Qnli-dev_dot_precision": 0.5686813186813187,
+      "eval_Qnli-dev_dot_recall": 0.8771186440677966,
       "eval_Qnli-dev_euclidean_accuracy": 0.66015625,
-      "eval_Qnli-dev_euclidean_accuracy_threshold": 14.833984375,
-      "eval_Qnli-dev_euclidean_ap": 0.6946097416292264,
-      "eval_Qnli-dev_euclidean_f1": 0.6921850079744816,
-      "eval_Qnli-dev_euclidean_f1_threshold": 19.65032196044922,
-      "eval_Qnli-dev_euclidean_precision": 0.5549872122762148,
-      "eval_Qnli-dev_euclidean_recall": 0.9194915254237288,
-      "eval_Qnli-dev_manhattan_accuracy": 0.673828125,
-      "eval_Qnli-dev_manhattan_accuracy_threshold": 335.0054016113281,
-      "eval_Qnli-dev_manhattan_ap": 0.7052862991616473,
-      "eval_Qnli-dev_manhattan_f1": 0.6913183279742765,
-      "eval_Qnli-dev_manhattan_f1_threshold": 413.606201171875,
-      "eval_Qnli-dev_manhattan_precision": 0.5569948186528497,
-      "eval_Qnli-dev_manhattan_recall": 0.9110169491525424,
-      "eval_Qnli-dev_max_accuracy": 0.673828125,
-      "eval_Qnli-dev_max_accuracy_threshold": 652.1033935546875,
-      "eval_Qnli-dev_max_ap": 0.7052862991616473,
-      "eval_Qnli-dev_max_f1": 0.6921850079744816,
-      "eval_Qnli-dev_max_f1_threshold": 568.281982421875,
-      "eval_Qnli-dev_max_precision": 0.5569948186528497,
-      "eval_Qnli-dev_max_recall": 0.9194915254237288,
-      "eval_allNLI-dev_cosine_accuracy": 0.693359375,
-      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9193407297134399,
-      "eval_allNLI-dev_cosine_ap": 0.5107511056200436,
-      "eval_allNLI-dev_cosine_f1": 0.5541666666666667,
-      "eval_allNLI-dev_cosine_f1_threshold": 0.8247472047805786,
-      "eval_allNLI-dev_cosine_precision": 0.43322475570032576,
-      "eval_allNLI-dev_cosine_recall": 0.7687861271676301,
-      "eval_allNLI-dev_dot_accuracy": 0.6953125,
-      "eval_allNLI-dev_dot_accuracy_threshold": 701.783447265625,
-      "eval_allNLI-dev_dot_ap": 0.5115874668538737,
-      "eval_allNLI-dev_dot_f1": 0.551440329218107,
-      "eval_allNLI-dev_dot_f1_threshold": 627.7216186523438,
-      "eval_allNLI-dev_dot_precision": 0.4281150159744409,
-      "eval_allNLI-dev_dot_recall": 0.7745664739884393,
-      "eval_allNLI-dev_euclidean_accuracy": 0.693359375,
-      "eval_allNLI-dev_euclidean_accuracy_threshold": 11.089791297912598,
-      "eval_allNLI-dev_euclidean_ap": 0.5111656459885842,
-      "eval_allNLI-dev_euclidean_f1": 0.5541666666666667,
-      "eval_allNLI-dev_euclidean_f1_threshold": 16.34630584716797,
-      "eval_allNLI-dev_euclidean_precision": 0.43322475570032576,
-      "eval_allNLI-dev_euclidean_recall": 0.7687861271676301,
-      "eval_allNLI-dev_manhattan_accuracy": 0.69140625,
-      "eval_allNLI-dev_manhattan_accuracy_threshold": 234.15560913085938,
-      "eval_allNLI-dev_manhattan_ap": 0.5103208400600096,
-      "eval_allNLI-dev_manhattan_f1": 0.5557809330628803,
-      "eval_allNLI-dev_manhattan_f1_threshold": 347.42694091796875,
-      "eval_allNLI-dev_manhattan_precision": 0.428125,
-      "eval_allNLI-dev_manhattan_recall": 0.791907514450867,
-      "eval_allNLI-dev_max_accuracy": 0.6953125,
-      "eval_allNLI-dev_max_accuracy_threshold": 701.783447265625,
-      "eval_allNLI-dev_max_ap": 0.5115874668538737,
-      "eval_allNLI-dev_max_f1": 0.5557809330628803,
-      "eval_allNLI-dev_max_f1_threshold": 627.7216186523438,
-      "eval_allNLI-dev_max_precision": 0.43322475570032576,
-      "eval_allNLI-dev_max_recall": 0.791907514450867,
-      "eval_loss": 1.046770691871643,
-      "eval_runtime": 54.2587,
-      "eval_samples_per_second": 30.668,
-      "eval_sequential_score": 0.7052862991616473,
-      "eval_steps_per_second": 0.129,
-      "eval_sts-test_pearson_cosine": 0.6387787653991626,
-      "eval_sts-test_pearson_dot": 0.6375419066754788,
-      "eval_sts-test_pearson_euclidean": 0.6534286404135119,
-      "eval_sts-test_pearson_manhattan": 0.6562338816714135,
-      "eval_sts-test_pearson_max": 0.6562338816714135,
-      "eval_sts-test_spearman_cosine": 0.6501734447586552,
-      "eval_sts-test_spearman_dot": 0.6498578588633231,
-      "eval_sts-test_spearman_euclidean": 0.650259948946925,
-      "eval_sts-test_spearman_manhattan": 0.6551447281328417,
-      "eval_sts-test_spearman_max": 0.6551447281328417,
+      "eval_Qnli-dev_euclidean_accuracy_threshold": 15.3107328414917,
+      "eval_Qnli-dev_euclidean_ap": 0.6880049661294799,
+      "eval_Qnli-dev_euclidean_f1": 0.6909090909090909,
+      "eval_Qnli-dev_euclidean_f1_threshold": 19.696701049804688,
+      "eval_Qnli-dev_euclidean_precision": 0.5663956639566395,
+      "eval_Qnli-dev_euclidean_recall": 0.885593220338983,
+      "eval_Qnli-dev_manhattan_accuracy": 0.6640625,
+      "eval_Qnli-dev_manhattan_accuracy_threshold": 329.4808349609375,
+      "eval_Qnli-dev_manhattan_ap": 0.6966692812556274,
+      "eval_Qnli-dev_manhattan_f1": 0.6902356902356903,
+      "eval_Qnli-dev_manhattan_f1_threshold": 411.3583679199219,
+      "eval_Qnli-dev_manhattan_precision": 0.5726256983240223,
+      "eval_Qnli-dev_manhattan_recall": 0.8686440677966102,
+      "eval_Qnli-dev_max_accuracy": 0.6640625,
+      "eval_Qnli-dev_max_accuracy_threshold": 639.7537231445312,
+      "eval_Qnli-dev_max_ap": 0.6966692812556274,
+      "eval_Qnli-dev_max_f1": 0.6910299003322258,
+      "eval_Qnli-dev_max_f1_threshold": 570.31103515625,
+      "eval_Qnli-dev_max_precision": 0.5726256983240223,
+      "eval_Qnli-dev_max_recall": 0.885593220338983,
+      "eval_allNLI-dev_cosine_accuracy": 0.701171875,
+      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9206767082214355,
+      "eval_allNLI-dev_cosine_ap": 0.513006237897071,
+      "eval_allNLI-dev_cosine_f1": 0.5411255411255411,
+      "eval_allNLI-dev_cosine_f1_threshold": 0.8377725481987,
+      "eval_allNLI-dev_cosine_precision": 0.43252595155709345,
+      "eval_allNLI-dev_cosine_recall": 0.7225433526011561,
+      "eval_allNLI-dev_dot_accuracy": 0.69921875,
+      "eval_allNLI-dev_dot_accuracy_threshold": 699.6907348632812,
+      "eval_allNLI-dev_dot_ap": 0.5131686680131453,
+      "eval_allNLI-dev_dot_f1": 0.5414847161572053,
+      "eval_allNLI-dev_dot_f1_threshold": 641.1671752929688,
+      "eval_allNLI-dev_dot_precision": 0.43508771929824563,
+      "eval_allNLI-dev_dot_recall": 0.7167630057803468,
+      "eval_allNLI-dev_euclidean_accuracy": 0.701171875,
+      "eval_allNLI-dev_euclidean_accuracy_threshold": 11.018213272094727,
+      "eval_allNLI-dev_euclidean_ap": 0.5129168021855273,
+      "eval_allNLI-dev_euclidean_f1": 0.5411255411255411,
+      "eval_allNLI-dev_euclidean_f1_threshold": 15.73725700378418,
+      "eval_allNLI-dev_euclidean_precision": 0.43252595155709345,
+      "eval_allNLI-dev_euclidean_recall": 0.7225433526011561,
+      "eval_allNLI-dev_manhattan_accuracy": 0.69921875,
+      "eval_allNLI-dev_manhattan_accuracy_threshold": 240.844970703125,
+      "eval_allNLI-dev_manhattan_ap": 0.5089409558106986,
+      "eval_allNLI-dev_manhattan_f1": 0.545054945054945,
+      "eval_allNLI-dev_manhattan_f1_threshold": 328.6717529296875,
+      "eval_allNLI-dev_manhattan_precision": 0.4397163120567376,
+      "eval_allNLI-dev_manhattan_recall": 0.7167630057803468,
+      "eval_allNLI-dev_max_accuracy": 0.701171875,
+      "eval_allNLI-dev_max_accuracy_threshold": 699.6907348632812,
+      "eval_allNLI-dev_max_ap": 0.5131686680131453,
+      "eval_allNLI-dev_max_f1": 0.545054945054945,
+      "eval_allNLI-dev_max_f1_threshold": 641.1671752929688,
+      "eval_allNLI-dev_max_precision": 0.4397163120567376,
+      "eval_allNLI-dev_max_recall": 0.7225433526011561,
+      "eval_loss": 0.9177284240722656,
+      "eval_runtime": 52.8643,
+      "eval_samples_per_second": 31.477,
+      "eval_sequential_score": 0.6966692812556274,
+      "eval_steps_per_second": 0.132,
+      "eval_sts-test_pearson_cosine": 0.6276409873243156,
+      "eval_sts-test_pearson_dot": 0.6255027331931474,
+      "eval_sts-test_pearson_euclidean": 0.6433595257550158,
+      "eval_sts-test_pearson_manhattan": 0.6522532291708293,
+      "eval_sts-test_pearson_max": 0.6522532291708293,
+      "eval_sts-test_spearman_cosine": 0.6429558638145366,
+      "eval_sts-test_spearman_dot": 0.6409545506582428,
+      "eval_sts-test_spearman_euclidean": 0.6431744817308246,
+      "eval_sts-test_spearman_manhattan": 0.6508304827784037,
+      "eval_sts-test_spearman_max": 0.6508304827784037,
       "step": 1683
     },
     {
-      "epoch": 1.656496062992126,
-      "grad_norm": 7.453148365020752,
+      "epoch": 1.65748031496063,
+      "grad_norm": 8.047771453857422,
       "learning_rate": 7.559062756777898e-07,
-      "loss": 0.3317,
+      "loss": 0.2412,
       "step": 1684
     },
     {
-      "epoch": 1.65748031496063,
-      "grad_norm": 14.089736938476562,
+      "epoch": 1.6584645669291338,
+      "grad_norm": 9.497953414916992,
       "learning_rate": 7.552465783942058e-07,
-      "loss": 0.6944,
+      "loss": 0.3353,
       "step": 1685
     },
     {
-      "epoch": 1.6584645669291338,
-      "grad_norm": 6.532533168792725,
+      "epoch": 1.659448818897638,
+      "grad_norm": 11.23277759552002,
       "learning_rate": 7.545862809008931e-07,
-      "loss": 0.2652,
+      "loss": 0.5455,
       "step": 1686
     },
     {
-      "epoch": 1.659448818897638,
-      "grad_norm": 9.175405502319336,
+      "epoch": 1.6604330708661417,
+      "grad_norm": 12.001265525817871,
       "learning_rate": 7.539253847607397e-07,
-      "loss": 0.4397,
+      "loss": 0.7836,
       "step": 1687
     },
     {
-      "epoch": 1.6604330708661417,
-      "grad_norm": 13.5653076171875,
+      "epoch": 1.6614173228346458,
+      "grad_norm": 15.05136775970459,
       "learning_rate": 7.532638915380503e-07,
-      "loss": 1.0538,
+      "loss": 0.8547,
       "step": 1688
     },
     {
-      "epoch": 1.6614173228346458,
-      "grad_norm": 7.647918224334717,
+      "epoch": 1.6624015748031495,
+      "grad_norm": 8.400712966918945,
       "learning_rate": 7.526018027985434e-07,
-      "loss": 0.3413,
+      "loss": 0.3242,
       "step": 1689
     },
     {
-      "epoch": 1.6624015748031495,
-      "grad_norm": 9.469629287719727,
+      "epoch": 1.6633858267716537,
+      "grad_norm": 10.37802791595459,
       "learning_rate": 7.519391201093462e-07,
-      "loss": 0.4697,
+      "loss": 0.5512,
       "step": 1690
     },
     {
-      "epoch": 1.6633858267716537,
-      "grad_norm": 9.352253913879395,
+      "epoch": 1.6643700787401574,
+      "grad_norm": 9.163196563720703,
       "learning_rate": 7.512758450389925e-07,
-      "loss": 0.4027,
+      "loss": 0.3552,
       "step": 1691
     },
     {
-      "epoch": 1.6643700787401574,
-      "grad_norm": 10.696208000183105,
+      "epoch": 1.6653543307086616,
+      "grad_norm": 6.402004718780518,
       "learning_rate": 7.50611979157418e-07,
-      "loss": 0.5004,
+      "loss": 0.1621,
       "step": 1692
     },
     {
-      "epoch": 1.6653543307086616,
-      "grad_norm": 8.486995697021484,
+      "epoch": 1.6663385826771653,
+      "grad_norm": 11.6072359085083,
       "learning_rate": 7.499475240359567e-07,
-      "loss": 0.402,
+      "loss": 0.6054,
       "step": 1693
     },
     {
-      "epoch": 1.6663385826771653,
-      "grad_norm": 8.808696746826172,
+      "epoch": 1.6673228346456694,
+      "grad_norm": 13.645760536193848,
       "learning_rate": 7.492824812473376e-07,
-      "loss": 0.336,
+      "loss": 0.8546,
       "step": 1694
     },
     {
-      "epoch": 1.6673228346456694,
-      "grad_norm": 10.125554084777832,
+      "epoch": 1.6683070866141732,
+      "grad_norm": 10.80756950378418,
       "learning_rate": 7.4861685236568e-07,
-      "loss": 0.4602,
+      "loss": 0.5073,
       "step": 1695
     },
     {
-      "epoch": 1.6683070866141732,
-      "grad_norm": 13.361590385437012,
+      "epoch": 1.6692913385826773,
+      "grad_norm": 11.263961791992188,
       "learning_rate": 7.479506389664914e-07,
-      "loss": 0.8332,
+      "loss": 0.4197,
       "step": 1696
     },
     {
-      "epoch": 1.6692913385826773,
-      "grad_norm": 17.709083557128906,
+      "epoch": 1.670275590551181,
+      "grad_norm": 11.742910385131836,
       "learning_rate": 7.47283842626662e-07,
-      "loss": 0.701,
+      "loss": 0.5708,
       "step": 1697
     },
     {
-      "epoch": 1.670275590551181,
-      "grad_norm": 8.722396850585938,
+      "epoch": 1.6712598425196852,
+      "grad_norm": 13.023048400878906,
       "learning_rate": 7.466164649244625e-07,
-      "loss": 0.2903,
+      "loss": 0.6482,
       "step": 1698
     },
     {
-      "epoch": 1.6712598425196852,
-      "grad_norm": 9.08720588684082,
+      "epoch": 1.672244094488189,
+      "grad_norm": 18.381389617919922,
       "learning_rate": 7.45948507439539e-07,
-      "loss": 0.3562,
+      "loss": 1.2711,
       "step": 1699
     },
     {
-      "epoch": 1.672244094488189,
-      "grad_norm": 8.78695011138916,
+      "epoch": 1.673228346456693,
+      "grad_norm": 13.040267944335938,
       "learning_rate": 7.452799717529102e-07,
-      "loss": 0.4957,
+      "loss": 0.6077,
       "step": 1700
     },
     {
-      "epoch": 1.673228346456693,
-      "grad_norm": 11.732802391052246,
+      "epoch": 1.6742125984251968,
+      "grad_norm": 13.312483787536621,
       "learning_rate": 7.446108594469634e-07,
-      "loss": 0.5276,
+      "loss": 0.9495,
       "step": 1701
     },
     {
-      "epoch": 1.6742125984251968,
-      "grad_norm": 14.943826675415039,
+      "epoch": 1.675196850393701,
+      "grad_norm": 13.009159088134766,
       "learning_rate": 7.439411721054508e-07,
-      "loss": 1.0879,
+      "loss": 1.2484,
       "step": 1702
     },
     {
-      "epoch": 1.675196850393701,
-      "grad_norm": 10.825159072875977,
+      "epoch": 1.6761811023622046,
+      "grad_norm": 12.714216232299805,
       "learning_rate": 7.432709113134856e-07,
-      "loss": 0.4225,
+      "loss": 0.9573,
       "step": 1703
     },
     {
-      "epoch": 1.6761811023622046,
-      "grad_norm": 14.165266990661621,
+      "epoch": 1.6771653543307088,
+      "grad_norm": 10.643980026245117,
       "learning_rate": 7.42600078657538e-07,
-      "loss": 0.7649,
+      "loss": 0.4614,
       "step": 1704
     },
     {
-      "epoch": 1.6771653543307088,
-      "grad_norm": 13.455033302307129,
+      "epoch": 1.6781496062992125,
+      "grad_norm": 10.612221717834473,
       "learning_rate": 7.419286757254324e-07,
-      "loss": 0.8458,
+      "loss": 0.7239,
       "step": 1705
     },
     {
-      "epoch": 1.6781496062992125,
-      "grad_norm": 8.739083290100098,
+      "epoch": 1.6791338582677167,
+      "grad_norm": 13.159558296203613,
       "learning_rate": 7.412567041063424e-07,
-      "loss": 0.377,
+      "loss": 0.5873,
       "step": 1706
     },
     {
-      "epoch": 1.6791338582677167,
-      "grad_norm": 11.673897743225098,
+      "epoch": 1.6801181102362204,
+      "grad_norm": 12.812065124511719,
       "learning_rate": 7.40584165390788e-07,
-      "loss": 0.4533,
+      "loss": 0.6671,
       "step": 1707
     },
     {
-      "epoch": 1.6801181102362204,
-      "grad_norm": 12.348942756652832,
+      "epoch": 1.6811023622047245,
+      "grad_norm": 7.931811332702637,
       "learning_rate": 7.399110611706316e-07,
-      "loss": 0.6464,
+      "loss": 0.3904,
       "step": 1708
     },
     {
-      "epoch": 1.6811023622047245,
-      "grad_norm": 8.441234588623047,
+      "epoch": 1.6820866141732282,
+      "grad_norm": 13.664514541625977,
       "learning_rate": 7.392373930390738e-07,
-      "loss": 0.3329,
+      "loss": 1.0458,
       "step": 1709
     },
     {
-      "epoch": 1.6820866141732282,
-      "grad_norm": 9.699177742004395,
+      "epoch": 1.6830708661417324,
+      "grad_norm": 8.342589378356934,
       "learning_rate": 7.3856316259065e-07,
-      "loss": 0.5933,
+      "loss": 0.4296,
       "step": 1710
     },
     {
-      "epoch": 1.6830708661417324,
-      "grad_norm": 9.690885543823242,
+      "epoch": 1.6840551181102361,
+      "grad_norm": 12.8305025100708,
       "learning_rate": 7.378883714212268e-07,
-      "loss": 0.2693,
+      "loss": 0.847,
       "step": 1711
     },
     {
-      "epoch": 1.6840551181102361,
-      "grad_norm": 9.51298713684082,
+      "epoch": 1.6850393700787403,
+      "grad_norm": 10.930636405944824,
       "learning_rate": 7.372130211279977e-07,
-      "loss": 0.3935,
+      "loss": 0.6004,
       "step": 1712
     },
     {
-      "epoch": 1.6850393700787403,
-      "grad_norm": 21.50560760498047,
+      "epoch": 1.686023622047244,
+      "grad_norm": 7.919064998626709,
       "learning_rate": 7.365371133094799e-07,
-      "loss": 0.8808,
+      "loss": 0.2714,
       "step": 1713
     },
     {
-      "epoch": 1.686023622047244,
-      "grad_norm": 8.320510864257812,
+      "epoch": 1.6870078740157481,
+      "grad_norm": 10.656820297241211,
       "learning_rate": 7.3586064956551e-07,
-      "loss": 0.3107,
+      "loss": 0.5642,
       "step": 1714
     },
     {
-      "epoch": 1.6870078740157481,
-      "grad_norm": 13.256457328796387,
+      "epoch": 1.6879921259842519,
+      "grad_norm": 6.634957790374756,
       "learning_rate": 7.351836314972408e-07,
-      "loss": 0.7207,
+      "loss": 0.1839,
       "step": 1715
     },
     {
-      "epoch": 1.6879921259842519,
-      "grad_norm": 9.8543119430542,
+      "epoch": 1.688976377952756,
+      "grad_norm": 10.987984657287598,
       "learning_rate": 7.345060607071365e-07,
-      "loss": 0.2448,
+      "loss": 0.5378,
       "step": 1716
     },
     {
-      "epoch": 1.688976377952756,
-      "grad_norm": 16.452669143676758,
+      "epoch": 1.6899606299212597,
+      "grad_norm": 11.241539001464844,
       "learning_rate": 7.338279387989703e-07,
-      "loss": 0.5973,
+      "loss": 0.4204,
       "step": 1717
     },
     {
-      "epoch": 1.6899606299212597,
-      "grad_norm": 12.07260513305664,
+      "epoch": 1.6909448818897639,
+      "grad_norm": 13.65767765045166,
       "learning_rate": 7.331492673778193e-07,
-      "loss": 0.7099,
+      "loss": 0.7513,
       "step": 1718
     },
     {
-      "epoch": 1.6909448818897639,
-      "grad_norm": 12.174388885498047,
+      "epoch": 1.6919291338582676,
+      "grad_norm": 8.150182723999023,
       "learning_rate": 7.324700480500615e-07,
-      "loss": 0.7829,
+      "loss": 0.3122,
       "step": 1719
     },
     {
-      "epoch": 1.6919291338582676,
-      "grad_norm": 17.407093048095703,
+      "epoch": 1.6929133858267718,
+      "grad_norm": 8.780075073242188,
       "learning_rate": 7.317902824233719e-07,
-      "loss": 1.0199,
+      "loss": 0.4128,
       "step": 1720
     },
     {
-      "epoch": 1.6929133858267718,
-      "grad_norm": 11.410693168640137,
+      "epoch": 1.6938976377952755,
+      "grad_norm": 9.636549949645996,
       "learning_rate": 7.311099721067183e-07,
-      "loss": 0.5131,
+      "loss": 0.3942,
       "step": 1721
     },
     {
-      "epoch": 1.6938976377952755,
-      "grad_norm": 10.559515953063965,
+      "epoch": 1.6948818897637796,
+      "grad_norm": 14.47591781616211,
       "learning_rate": 7.304291187103578e-07,
-      "loss": 0.3552,
+      "loss": 1.0144,
       "step": 1722
     },
     {
-      "epoch": 1.6948818897637796,
-      "grad_norm": 7.444983005523682,
+      "epoch": 1.6958661417322833,
+      "grad_norm": 8.88793659210205,
       "learning_rate": 7.297477238458333e-07,
-      "loss": 0.2427,
+      "loss": 0.3428,
       "step": 1723
     },
     {
-      "epoch": 1.6958661417322833,
-      "grad_norm": 10.84617805480957,
+      "epoch": 1.6968503937007875,
+      "grad_norm": 9.890840530395508,
       "learning_rate": 7.290657891259688e-07,
-      "loss": 0.5793,
+      "loss": 0.535,
       "step": 1724
     },
     {
-      "epoch": 1.6968503937007875,
-      "grad_norm": 11.741364479064941,
+      "epoch": 1.6978346456692912,
+      "grad_norm": 9.66598892211914,
       "learning_rate": 7.283833161648664e-07,
-      "loss": 0.6017,
+      "loss": 0.4357,
       "step": 1725
     },
     {
-      "epoch": 1.6978346456692912,
-      "grad_norm": 12.906566619873047,
+      "epoch": 1.6988188976377954,
+      "grad_norm": 11.194934844970703,
       "learning_rate": 7.277003065779022e-07,
-      "loss": 0.868,
+      "loss": 0.6145,
       "step": 1726
     },
     {
-      "epoch": 1.6988188976377954,
-      "grad_norm": 9.38211441040039,
+      "epoch": 1.699803149606299,
+      "grad_norm": 9.703887939453125,
       "learning_rate": 7.270167619817222e-07,
-      "loss": 0.437,
+      "loss": 0.5191,
       "step": 1727
     },
     {
-      "epoch": 1.699803149606299,
-      "grad_norm": 12.161310195922852,
+      "epoch": 1.7007874015748032,
+      "grad_norm": 14.260202407836914,
       "learning_rate": 7.26332683994239e-07,
-      "loss": 0.6072,
+      "loss": 0.9029,
       "step": 1728
     },
     {
-      "epoch": 1.7007874015748032,
-      "grad_norm": 9.744953155517578,
+      "epoch": 1.701771653543307,
+      "grad_norm": 12.470251083374023,
       "learning_rate": 7.256480742346278e-07,
-      "loss": 0.4196,
+      "loss": 0.8404,
       "step": 1729
     },
     {
-      "epoch": 1.701771653543307,
-      "grad_norm": 11.653627395629883,
+      "epoch": 1.702755905511811,
+      "grad_norm": 10.260608673095703,
       "learning_rate": 7.249629343233221e-07,
-      "loss": 0.5175,
+      "loss": 0.4985,
       "step": 1730
     },
     {
-      "epoch": 1.702755905511811,
-      "grad_norm": 7.9525628089904785,
+      "epoch": 1.7037401574803148,
+      "grad_norm": 13.649051666259766,
       "learning_rate": 7.242772658820105e-07,
-      "loss": 0.3439,
+      "loss": 0.5642,
       "step": 1731
     },
     {
-      "epoch": 1.7037401574803148,
-      "grad_norm": 8.413471221923828,
+      "epoch": 1.704724409448819,
+      "grad_norm": 9.734668731689453,
       "learning_rate": 7.235910705336327e-07,
-      "loss": 0.3967,
+      "loss": 0.4307,
       "step": 1732
     },
     {
-      "epoch": 1.704724409448819,
-      "grad_norm": 15.535460472106934,
+      "epoch": 1.7057086614173227,
+      "grad_norm": 10.934605598449707,
       "learning_rate": 7.229043499023753e-07,
-      "loss": 0.6726,
+      "loss": 0.5458,
       "step": 1733
     },
     {
-      "epoch": 1.7057086614173227,
-      "grad_norm": 23.099462509155273,
+      "epoch": 1.7066929133858268,
+      "grad_norm": 10.280840873718262,
       "learning_rate": 7.222171056136683e-07,
-      "loss": 1.0673,
+      "loss": 0.6761,
       "step": 1734
     },
     {
-      "epoch": 1.7066929133858268,
-      "grad_norm": 20.861698150634766,
+      "epoch": 1.7076771653543306,
+      "grad_norm": 10.9468412399292,
       "learning_rate": 7.215293392941814e-07,
-      "loss": 1.5876,
+      "loss": 0.4366,
       "step": 1735
     },
     {
-      "epoch": 1.7076771653543306,
-      "grad_norm": 6.357061386108398,
+      "epoch": 1.7086614173228347,
+      "grad_norm": 9.874605178833008,
       "learning_rate": 7.208410525718195e-07,
-      "loss": 0.3725,
+      "loss": 0.7759,
       "step": 1736
     },
     {
-      "epoch": 1.7086614173228347,
-      "grad_norm": 9.408873558044434,
+      "epoch": 1.7096456692913384,
+      "grad_norm": 10.595252990722656,
       "learning_rate": 7.201522470757196e-07,
-      "loss": 0.644,
+      "loss": 0.6839,
       "step": 1737
     },
     {
-      "epoch": 1.7096456692913384,
-      "grad_norm": 13.452515602111816,
+      "epoch": 1.7106299212598426,
+      "grad_norm": 11.22783088684082,
       "learning_rate": 7.194629244362462e-07,
-      "loss": 0.7511,
+      "loss": 0.5315,
       "step": 1738
     },
     {
-      "epoch": 1.7106299212598426,
-      "grad_norm": 11.06453800201416,
+      "epoch": 1.7116141732283463,
+      "grad_norm": 8.873421669006348,
       "learning_rate": 7.187730862849886e-07,
-      "loss": 0.4075,
+      "loss": 0.3093,
       "step": 1739
     },
     {
-      "epoch": 1.7116141732283463,
-      "grad_norm": 7.75278377532959,
+      "epoch": 1.7125984251968505,
+      "grad_norm": 8.114155769348145,
       "learning_rate": 7.180827342547556e-07,
-      "loss": 0.325,
+      "loss": 0.3635,
       "step": 1740
     },
     {
-      "epoch": 1.7125984251968505,
-      "grad_norm": 18.88231086730957,
+      "epoch": 1.7135826771653542,
+      "grad_norm": 5.080150604248047,
       "learning_rate": 7.173918699795726e-07,
-      "loss": 0.6408,
+      "loss": 0.1818,
       "step": 1741
     },
     {
-      "epoch": 1.7135826771653542,
-      "grad_norm": 11.604063987731934,
+      "epoch": 1.7145669291338583,
+      "grad_norm": 13.469075202941895,
       "learning_rate": 7.167004950946774e-07,
-      "loss": 0.5693,
+      "loss": 0.6372,
       "step": 1742
     },
     {
-      "epoch": 1.7145669291338583,
-      "grad_norm": 12.89757251739502,
+      "epoch": 1.715551181102362,
+      "grad_norm": 10.697860717773438,
       "learning_rate": 7.160086112365163e-07,
-      "loss": 0.755,
+      "loss": 0.5375,
       "step": 1743
     },
     {
-      "epoch": 1.715551181102362,
-      "grad_norm": 9.268168449401855,
+      "epoch": 1.7165354330708662,
+      "grad_norm": 9.85250473022461,
       "learning_rate": 7.153162200427404e-07,
-      "loss": 0.3953,
+      "loss": 0.5094,
       "step": 1744
     },
     {
-      "epoch": 1.7165354330708662,
-      "grad_norm": 7.196349620819092,
+      "epoch": 1.71751968503937,
+      "grad_norm": 15.063056945800781,
       "learning_rate": 7.146233231522018e-07,
-      "loss": 0.2483,
+      "loss": 0.8326,
       "step": 1745
     },
     {
-      "epoch": 1.71751968503937,
-      "grad_norm": 18.42107582092285,
+      "epoch": 1.718503937007874,
+      "grad_norm": 6.529819965362549,
       "learning_rate": 7.139299222049493e-07,
-      "loss": 0.8284,
+      "loss": 0.2582,
       "step": 1746
     },
     {
-      "epoch": 1.718503937007874,
-      "grad_norm": 16.944561004638672,
+      "epoch": 1.719488188976378,
+      "grad_norm": 8.93335247039795,
       "learning_rate": 7.132360188422249e-07,
-      "loss": 0.8637,
+      "loss": 0.3566,
       "step": 1747
     },
     {
-      "epoch": 1.719488188976378,
-      "grad_norm": 15.024608612060547,
+      "epoch": 1.720472440944882,
+      "grad_norm": 9.361804008483887,
       "learning_rate": 7.125416147064597e-07,
-      "loss": 0.809,
+      "loss": 0.4642,
       "step": 1748
     },
     {
-      "epoch": 1.720472440944882,
-      "grad_norm": 10.315958976745605,
+      "epoch": 1.7214566929133859,
+      "grad_norm": 11.460365295410156,
       "learning_rate": 7.118467114412702e-07,
-      "loss": 0.4039,
+      "loss": 0.5632,
       "step": 1749
     },
     {
-      "epoch": 1.7214566929133859,
-      "grad_norm": 14.782455444335938,
+      "epoch": 1.7224409448818898,
+      "grad_norm": 11.371411323547363,
       "learning_rate": 7.111513106914544e-07,
-      "loss": 0.7835,
+      "loss": 0.4932,
       "step": 1750
     },
     {
-      "epoch": 1.7224409448818898,
-      "grad_norm": 7.03788423538208,
+      "epoch": 1.7234251968503937,
+      "grad_norm": 8.059605598449707,
       "learning_rate": 7.104554141029874e-07,
-      "loss": 0.2731,
+      "loss": 0.2973,
       "step": 1751
     },
     {
-      "epoch": 1.7234251968503937,
-      "grad_norm": 17.06793785095215,
+      "epoch": 1.7244094488188977,
+      "grad_norm": 9.684464454650879,
       "learning_rate": 7.097590233230186e-07,
-      "loss": 1.1285,
+      "loss": 0.4152,
       "step": 1752
     },
     {
-      "epoch": 1.7244094488188977,
-      "grad_norm": 12.640857696533203,
+      "epoch": 1.7253937007874016,
+      "grad_norm": 9.201460838317871,
       "learning_rate": 7.090621399998665e-07,
-      "loss": 0.8982,
+      "loss": 0.3536,
       "step": 1753
     },
     {
-      "epoch": 1.7253937007874016,
-      "grad_norm": 10.769369125366211,
+      "epoch": 1.7263779527559056,
+      "grad_norm": 12.385725975036621,
       "learning_rate": 7.083647657830155e-07,
-      "loss": 0.5333,
+      "loss": 0.4387,
       "step": 1754
     },
     {
-      "epoch": 1.7263779527559056,
-      "grad_norm": 4.927238941192627,
+      "epoch": 1.7273622047244095,
+      "grad_norm": 13.970943450927734,
       "learning_rate": 7.076669023231126e-07,
-      "loss": 0.1284,
+      "loss": 0.8308,
       "step": 1755
     },
     {
-      "epoch": 1.7273622047244095,
-      "grad_norm": 12.758343696594238,
+      "epoch": 1.7283464566929134,
+      "grad_norm": 11.431367874145508,
       "learning_rate": 7.069685512719619e-07,
-      "loss": 0.6596,
+      "loss": 0.6075,
       "step": 1756
     },
     {
-      "epoch": 1.7283464566929134,
-      "grad_norm": 9.049027442932129,
+      "epoch": 1.7293307086614174,
+      "grad_norm": 8.143542289733887,
       "learning_rate": 7.06269714282522e-07,
-      "loss": 0.3537,
+      "loss": 0.323,
       "step": 1757
     },
     {
-      "epoch": 1.7293307086614174,
-      "grad_norm": 8.433937072753906,
+      "epoch": 1.7303149606299213,
+      "grad_norm": 14.46448802947998,
       "learning_rate": 7.055703930089018e-07,
-      "loss": 0.4507,
+      "loss": 0.6818,
       "step": 1758
     },
     {
-      "epoch": 1.7303149606299213,
-      "grad_norm": 16.079524993896484,
+      "epoch": 1.7312992125984252,
+      "grad_norm": 9.4772310256958,
       "learning_rate": 7.048705891063562e-07,
-      "loss": 0.7807,
+      "loss": 0.3929,
       "step": 1759
     },
     {
-      "epoch": 1.7312992125984252,
-      "grad_norm": 9.565560340881348,
+      "epoch": 1.7322834645669292,
+      "grad_norm": 9.145750999450684,
       "learning_rate": 7.04170304231283e-07,
-      "loss": 0.3956,
+      "loss": 0.4531,
       "step": 1760
     },
     {
-      "epoch": 1.7322834645669292,
-      "grad_norm": 5.272612571716309,
+      "epoch": 1.733267716535433,
+      "grad_norm": 9.935554504394531,
       "learning_rate": 7.034695400412177e-07,
-      "loss": 0.1616,
+      "loss": 0.4499,
       "step": 1761
     },
     {
-      "epoch": 1.733267716535433,
-      "grad_norm": 9.687300682067871,
+      "epoch": 1.734251968503937,
+      "grad_norm": 9.503782272338867,
       "learning_rate": 7.027682981948306e-07,
-      "loss": 0.4198,
+      "loss": 0.5292,
       "step": 1762
     },
     {
-      "epoch": 1.734251968503937,
-      "grad_norm": 11.882638931274414,
+      "epoch": 1.735236220472441,
+      "grad_norm": 10.536495208740234,
       "learning_rate": 7.020665803519229e-07,
-      "loss": 0.5712,
+      "loss": 0.57,
       "step": 1763
     },
     {
-      "epoch": 1.735236220472441,
-      "grad_norm": 12.776277542114258,
+      "epoch": 1.736220472440945,
+      "grad_norm": 10.385976791381836,
       "learning_rate": 7.013643881734222e-07,
-      "loss": 0.8004,
+      "loss": 0.3814,
       "step": 1764
     },
     {
-      "epoch": 1.736220472440945,
-      "grad_norm": 10.282646179199219,
+      "epoch": 1.7372047244094488,
+      "grad_norm": 7.928142547607422,
       "learning_rate": 7.006617233213787e-07,
-      "loss": 0.493,
+      "loss": 0.5125,
       "step": 1765
     },
     {
-      "epoch": 1.7372047244094488,
-      "grad_norm": 9.987165451049805,
+      "epoch": 1.7381889763779528,
+      "grad_norm": 11.099848747253418,
       "learning_rate": 6.999585874589616e-07,
-      "loss": 0.3876,
+      "loss": 0.4302,
       "step": 1766
     },
     {
-      "epoch": 1.7381889763779528,
-      "grad_norm": 8.786449432373047,
+      "epoch": 1.7391732283464567,
+      "grad_norm": 11.634474754333496,
       "learning_rate": 6.992549822504548e-07,
-      "loss": 0.2701,
+      "loss": 0.7763,
       "step": 1767
     },
     {
-      "epoch": 1.7391732283464567,
-      "grad_norm": 17.34662437438965,
+      "epoch": 1.7401574803149606,
+      "grad_norm": 10.054056167602539,
       "learning_rate": 6.985509093612534e-07,
-      "loss": 1.1244,
+      "loss": 0.6339,
       "step": 1768
     },
     {
-      "epoch": 1.7401574803149606,
-      "grad_norm": 13.91609001159668,
+      "epoch": 1.7411417322834646,
+      "grad_norm": 9.210363388061523,
       "learning_rate": 6.978463704578589e-07,
-      "loss": 0.9328,
+      "loss": 0.5771,
       "step": 1769
     },
     {
-      "epoch": 1.7411417322834646,
-      "grad_norm": 12.426129341125488,
+      "epoch": 1.7421259842519685,
+      "grad_norm": 9.674139022827148,
       "learning_rate": 6.971413672078766e-07,
-      "loss": 0.6236,
+      "loss": 0.4903,
       "step": 1770
     },
     {
-      "epoch": 1.7421259842519685,
-      "grad_norm": 16.320554733276367,
+      "epoch": 1.7431102362204725,
+      "grad_norm": 8.702906608581543,
       "learning_rate": 6.9643590128001e-07,
-      "loss": 0.5158,
+      "loss": 0.3864,
       "step": 1771
     },
     {
-      "epoch": 1.7431102362204725,
-      "grad_norm": 7.571167469024658,
+      "epoch": 1.7440944881889764,
+      "grad_norm": 9.84926986694336,
       "learning_rate": 6.957299743440589e-07,
-      "loss": 0.2337,
+      "loss": 0.4726,
       "step": 1772
     },
     {
-      "epoch": 1.7440944881889764,
-      "grad_norm": 10.513776779174805,
+      "epoch": 1.7450787401574803,
+      "grad_norm": 14.524797439575195,
       "learning_rate": 6.950235880709131e-07,
-      "loss": 0.5936,
+      "loss": 0.9534,
       "step": 1773
     },
     {
-      "epoch": 1.7450787401574803,
-      "grad_norm": 12.005305290222168,
+      "epoch": 1.7460629921259843,
+      "grad_norm": 14.925498008728027,
       "learning_rate": 6.943167441325504e-07,
-      "loss": 0.6437,
+      "loss": 0.886,
       "step": 1774
     },
     {
-      "epoch": 1.7460629921259843,
-      "grad_norm": 20.60370445251465,
+      "epoch": 1.7470472440944882,
+      "grad_norm": 12.312790870666504,
       "learning_rate": 6.936094442020314e-07,
-      "loss": 0.9568,
+      "loss": 0.7257,
       "step": 1775
     },
     {
-      "epoch": 1.7470472440944882,
-      "grad_norm": 12.10549545288086,
+      "epoch": 1.7480314960629921,
+      "grad_norm": 9.439497947692871,
       "learning_rate": 6.929016899534963e-07,
-      "loss": 0.5351,
+      "loss": 0.4611,
       "step": 1776
     },
     {
-      "epoch": 1.7480314960629921,
-      "grad_norm": 12.539772987365723,
+      "epoch": 1.749015748031496,
+      "grad_norm": 8.903450012207031,
       "learning_rate": 6.921934830621603e-07,
-      "loss": 0.6413,
+      "loss": 0.422,
       "step": 1777
     },
     {
-      "epoch": 1.749015748031496,
-      "grad_norm": 12.27780818939209,
+      "epoch": 1.75,
+      "grad_norm": 10.335160255432129,
       "learning_rate": 6.914848252043106e-07,
-      "loss": 0.7472,
+      "loss": 0.5538,
       "step": 1778
     },
     {
-      "epoch": 1.75,
-      "grad_norm": 14.630345344543457,
+      "epoch": 1.750984251968504,
+      "grad_norm": 12.994903564453125,
       "learning_rate": 6.907757180573013e-07,
-      "loss": 0.6276,
+      "loss": 0.6973,
       "step": 1779
     },
     {
-      "epoch": 1.750984251968504,
-      "grad_norm": 11.79141616821289,
+      "epoch": 1.7519685039370079,
+      "grad_norm": 10.423318862915039,
       "learning_rate": 6.9006616329955e-07,
-      "loss": 0.4015,
+      "loss": 0.5833,
       "step": 1780
     },
     {
-      "epoch": 1.7519685039370079,
-      "grad_norm": 9.032299041748047,
+      "epoch": 1.7529527559055118,
+      "grad_norm": 13.148845672607422,
       "learning_rate": 6.893561626105338e-07,
-      "loss": 0.3602,
+      "loss": 0.7693,
       "step": 1781
     },
     {
-      "epoch": 1.7529527559055118,
-      "grad_norm": 22.263681411743164,
+      "epoch": 1.7539370078740157,
+      "grad_norm": 8.706790924072266,
       "learning_rate": 6.886457176707855e-07,
-      "loss": 1.1816,
+      "loss": 0.4957,
       "step": 1782
     },
     {
-      "epoch": 1.7539370078740157,
-      "grad_norm": 11.743998527526855,
+      "epoch": 1.7549212598425197,
+      "grad_norm": 6.322276592254639,
       "learning_rate": 6.879348301618894e-07,
-      "loss": 0.4165,
+      "loss": 0.3354,
       "step": 1783
     },
     {
-      "epoch": 1.7549212598425197,
-      "grad_norm": 19.353025436401367,
+      "epoch": 1.7559055118110236,
+      "grad_norm": 11.415865898132324,
       "learning_rate": 6.872235017664769e-07,
-      "loss": 0.6603,
+      "loss": 0.6255,
       "step": 1784
     },
     {
-      "epoch": 1.7559055118110236,
-      "grad_norm": 13.100909233093262,
+      "epoch": 1.7568897637795275,
+      "grad_norm": 8.03885555267334,
       "learning_rate": 6.865117341682234e-07,
-      "loss": 0.628,
+      "loss": 0.2834,
       "step": 1785
     },
     {
-      "epoch": 1.7568897637795275,
-      "grad_norm": 20.460329055786133,
+      "epoch": 1.7578740157480315,
+      "grad_norm": 8.716883659362793,
       "learning_rate": 6.857995290518439e-07,
-      "loss": 0.8104,
+      "loss": 0.4637,
       "step": 1786
     },
     {
-      "epoch": 1.7578740157480315,
-      "grad_norm": 19.350191116333008,
+      "epoch": 1.7588582677165354,
+      "grad_norm": 12.705853462219238,
       "learning_rate": 6.850868881030883e-07,
-      "loss": 0.7918,
+      "loss": 0.8133,
       "step": 1787
     },
     {
-      "epoch": 1.7588582677165354,
-      "grad_norm": 11.969427108764648,
+      "epoch": 1.7598425196850394,
+      "grad_norm": 6.630486488342285,
       "learning_rate": 6.843738130087393e-07,
-      "loss": 0.6674,
+      "loss": 0.2364,
       "step": 1788
     },
     {
-      "epoch": 1.7598425196850394,
-      "grad_norm": 13.915922164916992,
+      "epoch": 1.7608267716535433,
+      "grad_norm": 14.637483596801758,
       "learning_rate": 6.836603054566059e-07,
-      "loss": 0.696,
+      "loss": 0.8109,
       "step": 1789
     },
     {
-      "epoch": 1.7608267716535433,
-      "grad_norm": 16.64994239807129,
+      "epoch": 1.7618110236220472,
+      "grad_norm": 9.314077377319336,
       "learning_rate": 6.829463671355216e-07,
-      "loss": 0.6102,
+      "loss": 0.4795,
       "step": 1790
     },
     {
-      "epoch": 1.7618110236220472,
-      "grad_norm": 12.086501121520996,
+      "epoch": 1.7627952755905512,
+      "grad_norm": 10.287254333496094,
       "learning_rate": 6.822319997353391e-07,
-      "loss": 0.6323,
+      "loss": 0.3508,
       "step": 1791
     },
     {
-      "epoch": 1.7627952755905512,
-      "grad_norm": 10.079103469848633,
+      "epoch": 1.763779527559055,
+      "grad_norm": 11.251386642456055,
       "learning_rate": 6.815172049469268e-07,
-      "loss": 0.447,
+      "loss": 0.5447,
       "step": 1792
     },
     {
-      "epoch": 1.763779527559055,
-      "grad_norm": 13.176106452941895,
+      "epoch": 1.764763779527559,
+      "grad_norm": 14.856246948242188,
       "learning_rate": 6.808019844621649e-07,
-      "loss": 0.5301,
+      "loss": 0.7391,
       "step": 1793
     },
     {
-      "epoch": 1.764763779527559,
-      "grad_norm": 10.617973327636719,
+      "epoch": 1.765748031496063,
+      "grad_norm": 9.681744575500488,
       "learning_rate": 6.800863399739409e-07,
-      "loss": 0.4242,
+      "loss": 0.3975,
       "step": 1794
     },
     {
-      "epoch": 1.765748031496063,
-      "grad_norm": 16.50306510925293,
+      "epoch": 1.766732283464567,
+      "grad_norm": 12.79214096069336,
       "learning_rate": 6.793702731761459e-07,
-      "loss": 1.0912,
+      "loss": 0.7205,
       "step": 1795
     },
     {
-      "epoch": 1.766732283464567,
-      "grad_norm": 10.760600090026855,
+      "epoch": 1.7677165354330708,
+      "grad_norm": 11.755746841430664,
       "learning_rate": 6.786537857636707e-07,
-      "loss": 0.5947,
+      "loss": 0.7735,
       "step": 1796
     },
     {
-      "epoch": 1.7677165354330708,
-      "grad_norm": 17.52591323852539,
+      "epoch": 1.7687007874015748,
+      "grad_norm": 9.981927871704102,
       "learning_rate": 6.77936879432402e-07,
-      "loss": 0.9659,
+      "loss": 0.4532,
       "step": 1797
     },
     {
-      "epoch": 1.7687007874015748,
-      "grad_norm": 10.315180778503418,
+      "epoch": 1.7696850393700787,
+      "grad_norm": 7.891877174377441,
       "learning_rate": 6.772195558792175e-07,
-      "loss": 0.3915,
+      "loss": 0.331,
       "step": 1798
     },
     {
-      "epoch": 1.7696850393700787,
-      "grad_norm": 16.027626037597656,
+      "epoch": 1.7706692913385826,
+      "grad_norm": 13.395078659057617,
       "learning_rate": 6.765018168019825e-07,
-      "loss": 1.0406,
+      "loss": 0.8354,
       "step": 1799
     },
     {
-      "epoch": 1.7706692913385826,
-      "grad_norm": 11.071070671081543,
+      "epoch": 1.7716535433070866,
+      "grad_norm": 3.719597578048706,
       "learning_rate": 6.757836638995461e-07,
-      "loss": 0.601,
+      "loss": 0.0976,
       "step": 1800
     },
     {
-      "epoch": 1.7716535433070866,
-      "grad_norm": 7.411619663238525,
+      "epoch": 1.7726377952755905,
+      "grad_norm": 10.299521446228027,
       "learning_rate": 6.75065098871737e-07,
-      "loss": 0.2213,
+      "loss": 0.6051,
       "step": 1801
     },
     {
-      "epoch": 1.7726377952755905,
-      "grad_norm": 13.08354663848877,
+      "epoch": 1.7736220472440944,
+      "grad_norm": 10.495471000671387,
       "learning_rate": 6.74346123419359e-07,
-      "loss": 0.6773,
+      "loss": 0.707,
       "step": 1802
     },
     {
-      "epoch": 1.7736220472440944,
-      "grad_norm": 14.372445106506348,
+      "epoch": 1.7746062992125984,
+      "grad_norm": 9.168440818786621,
       "learning_rate": 6.736267392441874e-07,
-      "loss": 0.5883,
+      "loss": 0.419,
       "step": 1803
     },
     {
-      "epoch": 1.7746062992125984,
-      "grad_norm": 10.676488876342773,
+      "epoch": 1.7755905511811023,
+      "grad_norm": 13.20716667175293,
       "learning_rate": 6.729069480489652e-07,
-      "loss": 0.4667,
+      "loss": 0.9268,
       "step": 1804
     },
     {
-      "epoch": 1.7755905511811023,
-      "grad_norm": 12.284120559692383,
+      "epoch": 1.7765748031496063,
+      "grad_norm": 9.664036750793457,
       "learning_rate": 6.721867515373987e-07,
-      "loss": 0.6663,
+      "loss": 0.639,
       "step": 1805
     },
     {
-      "epoch": 1.7765748031496063,
-      "grad_norm": 10.81378173828125,
+      "epoch": 1.7775590551181102,
+      "grad_norm": 8.314957618713379,
       "learning_rate": 6.714661514141535e-07,
-      "loss": 0.6057,
+      "loss": 0.427,
       "step": 1806
     },
     {
-      "epoch": 1.7775590551181102,
-      "grad_norm": 12.88852310180664,
+      "epoch": 1.7785433070866141,
+      "grad_norm": 8.465643882751465,
       "learning_rate": 6.707451493848502e-07,
-      "loss": 0.8443,
+      "loss": 0.6253,
       "step": 1807
     },
     {
-      "epoch": 1.7785433070866141,
-      "grad_norm": 11.20924186706543,
+      "epoch": 1.779527559055118,
+      "grad_norm": 11.997491836547852,
       "learning_rate": 6.700237471560614e-07,
-      "loss": 0.6473,
+      "loss": 0.6166,
       "step": 1808
     },
     {
-      "epoch": 1.779527559055118,
-      "grad_norm": 6.5453691482543945,
+      "epoch": 1.780511811023622,
+      "grad_norm": 11.27194881439209,
       "learning_rate": 6.693019464353065e-07,
-      "loss": 0.2648,
+      "loss": 0.6593,
       "step": 1809
     },
     {
-      "epoch": 1.780511811023622,
-      "grad_norm": 11.30556583404541,
+      "epoch": 1.781496062992126,
+      "grad_norm": 8.32484245300293,
       "learning_rate": 6.685797489310481e-07,
-      "loss": 0.5918,
+      "loss": 0.3922,
       "step": 1810
     },
     {
-      "epoch": 1.781496062992126,
-      "grad_norm": 15.30104923248291,
+      "epoch": 1.78248031496063,
+      "grad_norm": 10.298493385314941,
       "learning_rate": 6.67857156352688e-07,
-      "loss": 0.6399,
+      "loss": 0.7626,
       "step": 1811
     },
     {
-      "epoch": 1.78248031496063,
-      "grad_norm": 15.234845161437988,
+      "epoch": 1.7834645669291338,
+      "grad_norm": 13.35094928741455,
       "learning_rate": 6.671341704105632e-07,
-      "loss": 0.5961,
+      "loss": 0.6866,
       "step": 1812
     },
     {
-      "epoch": 1.7834645669291338,
-      "grad_norm": 13.376496315002441,
+      "epoch": 1.784448818897638,
+      "grad_norm": 9.887539863586426,
       "learning_rate": 6.664107928159419e-07,
-      "loss": 0.2637,
+      "loss": 0.4619,
       "step": 1813
     },
     {
-      "epoch": 1.784448818897638,
-      "grad_norm": 18.91242790222168,
+      "epoch": 1.7854330708661417,
+      "grad_norm": 11.572975158691406,
       "learning_rate": 6.656870252810187e-07,
-      "loss": 0.8332,
+      "loss": 0.8688,
       "step": 1814
     },
     {
-      "epoch": 1.7854330708661417,
-      "grad_norm": 6.986865043640137,
+      "epoch": 1.7864173228346458,
+      "grad_norm": 9.701433181762695,
       "learning_rate": 6.649628695189123e-07,
-      "loss": 0.3121,
+      "loss": 0.5084,
       "step": 1815
     },
     {
-      "epoch": 1.7864173228346458,
-      "grad_norm": 17.72435188293457,
+      "epoch": 1.7874015748031495,
+      "grad_norm": 12.317909240722656,
       "learning_rate": 6.642383272436589e-07,
-      "loss": 0.9966,
+      "loss": 0.7005,
       "step": 1816
     },
     {
-      "epoch": 1.7874015748031495,
-      "grad_norm": 13.330727577209473,
+      "epoch": 1.7883858267716537,
+      "grad_norm": 9.139718055725098,
       "learning_rate": 6.635134001702107e-07,
-      "loss": 0.9226,
+      "loss": 0.2802,
       "step": 1817
     },
     {
-      "epoch": 1.7883858267716537,
-      "grad_norm": 3.9042019844055176,
+      "epoch": 1.7893700787401574,
+      "grad_norm": 11.69360065460205,
       "learning_rate": 6.627880900144303e-07,
-      "loss": 0.1044,
+      "loss": 0.6338,
       "step": 1818
     },
     {
-      "epoch": 1.7893700787401574,
-      "grad_norm": 12.28075885772705,
+      "epoch": 1.7903543307086616,
+      "grad_norm": 11.106903076171875,
       "learning_rate": 6.620623984930869e-07,
-      "loss": 0.67,
+      "loss": 0.4616,
       "step": 1819
     },
     {
-      "epoch": 1.7903543307086616,
-      "grad_norm": 9.03951358795166,
+      "epoch": 1.7913385826771653,
+      "grad_norm": 7.193187236785889,
       "learning_rate": 6.613363273238527e-07,
-      "loss": 0.4401,
+      "loss": 0.1921,
       "step": 1820
     },
     {
-      "epoch": 1.7913385826771653,
-      "grad_norm": 9.525883674621582,
+      "epoch": 1.7923228346456694,
+      "grad_norm": 13.572063446044922,
       "learning_rate": 6.606098782252978e-07,
-      "loss": 0.4222,
+      "loss": 0.9873,
       "step": 1821
     },
     {
-      "epoch": 1.7923228346456694,
-      "grad_norm": 10.233338356018066,
+      "epoch": 1.7933070866141732,
+      "grad_norm": 6.819611549377441,
       "learning_rate": 6.598830529168877e-07,
-      "loss": 0.5033,
+      "loss": 0.2347,
       "step": 1822
     },
     {
-      "epoch": 1.7933070866141732,
-      "grad_norm": 13.113160133361816,
+      "epoch": 1.7942913385826773,
+      "grad_norm": 10.862784385681152,
       "learning_rate": 6.591558531189781e-07,
-      "loss": 0.7639,
+      "loss": 0.4585,
       "step": 1823
     },
     {
-      "epoch": 1.7942913385826773,
-      "grad_norm": 10.392827987670898,
+      "epoch": 1.795275590551181,
+      "grad_norm": 13.653882026672363,
       "learning_rate": 6.584282805528106e-07,
-      "loss": 0.5791,
+      "loss": 0.7981,
       "step": 1824
     },
     {
-      "epoch": 1.795275590551181,
-      "grad_norm": 13.654716491699219,
+      "epoch": 1.7962598425196852,
+      "grad_norm": 9.333664894104004,
       "learning_rate": 6.577003369405099e-07,
-      "loss": 0.7615,
+      "loss": 0.3834,
       "step": 1825
     },
     {
-      "epoch": 1.7962598425196852,
-      "grad_norm": 9.4904146194458,
+      "epoch": 1.797244094488189,
+      "grad_norm": 8.496042251586914,
       "learning_rate": 6.569720240050784e-07,
-      "loss": 0.26,
+      "loss": 0.3476,
       "step": 1826
     },
     {
-      "epoch": 1.797244094488189,
-      "grad_norm": 7.8181071281433105,
+      "epoch": 1.798228346456693,
+      "grad_norm": 9.476622581481934,
       "learning_rate": 6.562433434703928e-07,
-      "loss": 0.2372,
+      "loss": 0.6374,
       "step": 1827
     },
     {
-      "epoch": 1.798228346456693,
-      "grad_norm": 11.520416259765625,
+      "epoch": 1.7992125984251968,
+      "grad_norm": 7.275120258331299,
       "learning_rate": 6.555142970612001e-07,
-      "loss": 0.3546,
+      "loss": 0.2443,
       "step": 1828
     },
     {
-      "epoch": 1.7992125984251968,
-      "grad_norm": 10.961482048034668,
+      "epoch": 1.800196850393701,
+      "grad_norm": 12.427238464355469,
       "learning_rate": 6.54784886503113e-07,
-      "loss": 0.4989,
+      "loss": 0.8733,
       "step": 1829
     },
     {
-      "epoch": 1.800196850393701,
-      "grad_norm": 19.619661331176758,
+      "epoch": 1.8011811023622046,
+      "grad_norm": 12.010316848754883,
       "learning_rate": 6.540551135226061e-07,
-      "loss": 1.1705,
+      "loss": 0.6572,
       "step": 1830
     },
     {
-      "epoch": 1.8011811023622046,
-      "grad_norm": 20.04231071472168,
+      "epoch": 1.8021653543307088,
+      "grad_norm": 8.92840576171875,
       "learning_rate": 6.533249798470123e-07,
-      "loss": 1.1289,
+      "loss": 0.5134,
       "step": 1831
     },
     {
-      "epoch": 1.8021653543307088,
-      "grad_norm": 14.368165969848633,
+      "epoch": 1.8031496062992125,
+      "grad_norm": 8.452524185180664,
       "learning_rate": 6.525944872045175e-07,
-      "loss": 0.8638,
+      "loss": 0.3629,
       "step": 1832
     },
     {
-      "epoch": 1.8031496062992125,
-      "grad_norm": 8.188169479370117,
+      "epoch": 1.8041338582677167,
+      "grad_norm": 7.315922260284424,
       "learning_rate": 6.51863637324158e-07,
-      "loss": 0.2951,
+      "loss": 0.2229,
       "step": 1833
     },
     {
-      "epoch": 1.8041338582677167,
-      "grad_norm": 7.987644672393799,
+      "epoch": 1.8051181102362204,
+      "grad_norm": 10.379257202148438,
       "learning_rate": 6.511324319358154e-07,
-      "loss": 0.3667,
+      "loss": 0.4706,
       "step": 1834
     },
     {
-      "epoch": 1.8051181102362204,
-      "grad_norm": 11.989052772521973,
+      "epoch": 1.8061023622047245,
+      "grad_norm": 9.111647605895996,
       "learning_rate": 6.504008727702123e-07,
-      "loss": 0.5948,
+      "loss": 0.4832,
       "step": 1835
     },
     {
-      "epoch": 1.8061023622047245,
-      "grad_norm": 6.342445373535156,
+      "epoch": 1.8070866141732282,
+      "grad_norm": 8.726858139038086,
       "learning_rate": 6.496689615589096e-07,
-      "loss": 0.1921,
+      "loss": 0.3765,
       "step": 1836
     },
     {
-      "epoch": 1.8061023622047245,
-      "eval_Qnli-dev_cosine_accuracy": 0.6640625,
-      "eval_Qnli-dev_cosine_accuracy_threshold": 0.8278301954269409,
-      "eval_Qnli-dev_cosine_ap": 0.6921415813364429,
-      "eval_Qnli-dev_cosine_f1": 0.687813021702838,
-      "eval_Qnli-dev_cosine_f1_threshold": 0.7467654943466187,
-      "eval_Qnli-dev_cosine_precision": 0.5674931129476584,
-      "eval_Qnli-dev_cosine_recall": 0.8728813559322034,
-      "eval_Qnli-dev_dot_accuracy": 0.658203125,
-      "eval_Qnli-dev_dot_accuracy_threshold": 638.0369262695312,
-      "eval_Qnli-dev_dot_ap": 0.6907455906812887,
-      "eval_Qnli-dev_dot_f1": 0.6901172529313233,
-      "eval_Qnli-dev_dot_f1_threshold": 568.925537109375,
-      "eval_Qnli-dev_dot_precision": 0.5706371191135734,
-      "eval_Qnli-dev_dot_recall": 0.8728813559322034,
-      "eval_Qnli-dev_euclidean_accuracy": 0.6640625,
-      "eval_Qnli-dev_euclidean_accuracy_threshold": 16.232084274291992,
-      "eval_Qnli-dev_euclidean_ap": 0.692218626949551,
-      "eval_Qnli-dev_euclidean_f1": 0.6877076411960132,
-      "eval_Qnli-dev_euclidean_f1_threshold": 19.771434783935547,
-      "eval_Qnli-dev_euclidean_precision": 0.5655737704918032,
-      "eval_Qnli-dev_euclidean_recall": 0.8771186440677966,
-      "eval_Qnli-dev_manhattan_accuracy": 0.6796875,
-      "eval_Qnli-dev_manhattan_accuracy_threshold": 370.8492431640625,
-      "eval_Qnli-dev_manhattan_ap": 0.7016599415151726,
-      "eval_Qnli-dev_manhattan_f1": 0.6907216494845361,
-      "eval_Qnli-dev_manhattan_f1_threshold": 403.894775390625,
-      "eval_Qnli-dev_manhattan_precision": 0.5809248554913294,
-      "eval_Qnli-dev_manhattan_recall": 0.8516949152542372,
-      "eval_Qnli-dev_max_accuracy": 0.6796875,
-      "eval_Qnli-dev_max_accuracy_threshold": 638.0369262695312,
-      "eval_Qnli-dev_max_ap": 0.7016599415151726,
-      "eval_Qnli-dev_max_f1": 0.6907216494845361,
-      "eval_Qnli-dev_max_f1_threshold": 568.925537109375,
-      "eval_Qnli-dev_max_precision": 0.5809248554913294,
-      "eval_Qnli-dev_max_recall": 0.8771186440677966,
+      "epoch": 1.8070866141732282,
+      "eval_Qnli-dev_cosine_accuracy": 0.65234375,
+      "eval_Qnli-dev_cosine_accuracy_threshold": 0.8536216020584106,
+      "eval_Qnli-dev_cosine_ap": 0.68444801736511,
+      "eval_Qnli-dev_cosine_f1": 0.6926829268292684,
+      "eval_Qnli-dev_cosine_f1_threshold": 0.7339719533920288,
+      "eval_Qnli-dev_cosine_precision": 0.5620052770448549,
+      "eval_Qnli-dev_cosine_recall": 0.902542372881356,
+      "eval_Qnli-dev_dot_accuracy": 0.650390625,
+      "eval_Qnli-dev_dot_accuracy_threshold": 652.3082885742188,
+      "eval_Qnli-dev_dot_ap": 0.6819551428799111,
+      "eval_Qnli-dev_dot_f1": 0.6949429037520392,
+      "eval_Qnli-dev_dot_f1_threshold": 561.5039672851562,
+      "eval_Qnli-dev_dot_precision": 0.5649867374005305,
+      "eval_Qnli-dev_dot_recall": 0.902542372881356,
+      "eval_Qnli-dev_euclidean_accuracy": 0.654296875,
+      "eval_Qnli-dev_euclidean_accuracy_threshold": 14.972758293151855,
+      "eval_Qnli-dev_euclidean_ap": 0.6845367142236655,
+      "eval_Qnli-dev_euclidean_f1": 0.6916802610114192,
+      "eval_Qnli-dev_euclidean_f1_threshold": 20.030675888061523,
+      "eval_Qnli-dev_euclidean_precision": 0.5623342175066313,
+      "eval_Qnli-dev_euclidean_recall": 0.8983050847457628,
+      "eval_Qnli-dev_manhattan_accuracy": 0.65625,
+      "eval_Qnli-dev_manhattan_accuracy_threshold": 343.36370849609375,
+      "eval_Qnli-dev_manhattan_ap": 0.6940033401652869,
+      "eval_Qnli-dev_manhattan_f1": 0.6960784313725491,
+      "eval_Qnli-dev_manhattan_f1_threshold": 421.79571533203125,
+      "eval_Qnli-dev_manhattan_precision": 0.5664893617021277,
+      "eval_Qnli-dev_manhattan_recall": 0.902542372881356,
+      "eval_Qnli-dev_max_accuracy": 0.65625,
+      "eval_Qnli-dev_max_accuracy_threshold": 652.3082885742188,
+      "eval_Qnli-dev_max_ap": 0.6940033401652869,
+      "eval_Qnli-dev_max_f1": 0.6960784313725491,
+      "eval_Qnli-dev_max_f1_threshold": 561.5039672851562,
+      "eval_Qnli-dev_max_precision": 0.5664893617021277,
+      "eval_Qnli-dev_max_recall": 0.902542372881356,
       "eval_allNLI-dev_cosine_accuracy": 0.6953125,
-      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9156147241592407,
-      "eval_allNLI-dev_cosine_ap": 0.5190278899962391,
-      "eval_allNLI-dev_cosine_f1": 0.5562632696390658,
-      "eval_allNLI-dev_cosine_f1_threshold": 0.83030104637146,
-      "eval_allNLI-dev_cosine_precision": 0.4395973154362416,
-      "eval_allNLI-dev_cosine_recall": 0.7572254335260116,
-      "eval_allNLI-dev_dot_accuracy": 0.6953125,
-      "eval_allNLI-dev_dot_accuracy_threshold": 709.8850708007812,
-      "eval_allNLI-dev_dot_ap": 0.5167210503115699,
-      "eval_allNLI-dev_dot_f1": 0.5520169851380042,
-      "eval_allNLI-dev_dot_f1_threshold": 634.1051635742188,
-      "eval_allNLI-dev_dot_precision": 0.436241610738255,
-      "eval_allNLI-dev_dot_recall": 0.7514450867052023,
+      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9244678616523743,
+      "eval_allNLI-dev_cosine_ap": 0.5193634578230323,
+      "eval_allNLI-dev_cosine_f1": 0.5596707818930042,
+      "eval_allNLI-dev_cosine_f1_threshold": 0.8289010524749756,
+      "eval_allNLI-dev_cosine_precision": 0.43450479233226835,
+      "eval_allNLI-dev_cosine_recall": 0.7861271676300579,
+      "eval_allNLI-dev_dot_accuracy": 0.693359375,
+      "eval_allNLI-dev_dot_accuracy_threshold": 708.5109252929688,
+      "eval_allNLI-dev_dot_ap": 0.5179064331299997,
+      "eval_allNLI-dev_dot_f1": 0.5591836734693878,
+      "eval_allNLI-dev_dot_f1_threshold": 631.8807373046875,
+      "eval_allNLI-dev_dot_precision": 0.43217665615141954,
+      "eval_allNLI-dev_dot_recall": 0.791907514450867,
       "eval_allNLI-dev_euclidean_accuracy": 0.6953125,
-      "eval_allNLI-dev_euclidean_accuracy_threshold": 10.66873550415039,
-      "eval_allNLI-dev_euclidean_ap": 0.5195587301363868,
-      "eval_allNLI-dev_euclidean_f1": 0.5562632696390658,
-      "eval_allNLI-dev_euclidean_f1_threshold": 16.106643676757812,
-      "eval_allNLI-dev_euclidean_precision": 0.4395973154362416,
-      "eval_allNLI-dev_euclidean_recall": 0.7572254335260116,
-      "eval_allNLI-dev_manhattan_accuracy": 0.69140625,
-      "eval_allNLI-dev_manhattan_accuracy_threshold": 259.05010986328125,
-      "eval_allNLI-dev_manhattan_ap": 0.515787932229368,
-      "eval_allNLI-dev_manhattan_f1": 0.5493562231759657,
-      "eval_allNLI-dev_manhattan_f1_threshold": 338.73870849609375,
-      "eval_allNLI-dev_manhattan_precision": 0.43686006825938567,
-      "eval_allNLI-dev_manhattan_recall": 0.7398843930635838,
+      "eval_allNLI-dev_euclidean_accuracy_threshold": 10.730307579040527,
+      "eval_allNLI-dev_euclidean_ap": 0.5194794509210287,
+      "eval_allNLI-dev_euclidean_f1": 0.5585215605749487,
+      "eval_allNLI-dev_euclidean_f1_threshold": 16.171306610107422,
+      "eval_allNLI-dev_euclidean_precision": 0.43312101910828027,
+      "eval_allNLI-dev_euclidean_recall": 0.7861271676300579,
+      "eval_allNLI-dev_manhattan_accuracy": 0.6953125,
+      "eval_allNLI-dev_manhattan_accuracy_threshold": 231.04771423339844,
+      "eval_allNLI-dev_manhattan_ap": 0.5177496703367335,
+      "eval_allNLI-dev_manhattan_f1": 0.5542168674698795,
+      "eval_allNLI-dev_manhattan_f1_threshold": 344.5424499511719,
+      "eval_allNLI-dev_manhattan_precision": 0.4246153846153846,
+      "eval_allNLI-dev_manhattan_recall": 0.7976878612716763,
       "eval_allNLI-dev_max_accuracy": 0.6953125,
-      "eval_allNLI-dev_max_accuracy_threshold": 709.8850708007812,
-      "eval_allNLI-dev_max_ap": 0.5195587301363868,
-      "eval_allNLI-dev_max_f1": 0.5562632696390658,
-      "eval_allNLI-dev_max_f1_threshold": 634.1051635742188,
-      "eval_allNLI-dev_max_precision": 0.4395973154362416,
-      "eval_allNLI-dev_max_recall": 0.7572254335260116,
-      "eval_loss": 0.9842851758003235,
-      "eval_runtime": 54.3651,
-      "eval_samples_per_second": 30.608,
-      "eval_sequential_score": 0.7016599415151726,
-      "eval_steps_per_second": 0.129,
-      "eval_sts-test_pearson_cosine": 0.6309265304897379,
-      "eval_sts-test_pearson_dot": 0.6292250868831678,
-      "eval_sts-test_pearson_euclidean": 0.6445377889980463,
-      "eval_sts-test_pearson_manhattan": 0.6504294362294498,
-      "eval_sts-test_pearson_max": 0.6504294362294498,
-      "eval_sts-test_spearman_cosine": 0.642238464347562,
-      "eval_sts-test_spearman_dot": 0.6404100620608937,
-      "eval_sts-test_spearman_euclidean": 0.6423653341736068,
-      "eval_sts-test_spearman_manhattan": 0.6474620105822118,
-      "eval_sts-test_spearman_max": 0.6474620105822118,
+      "eval_allNLI-dev_max_accuracy_threshold": 708.5109252929688,
+      "eval_allNLI-dev_max_ap": 0.5194794509210287,
+      "eval_allNLI-dev_max_f1": 0.5596707818930042,
+      "eval_allNLI-dev_max_f1_threshold": 631.8807373046875,
+      "eval_allNLI-dev_max_precision": 0.43450479233226835,
+      "eval_allNLI-dev_max_recall": 0.7976878612716763,
+      "eval_loss": 0.9066572785377502,
+      "eval_runtime": 53.0735,
+      "eval_samples_per_second": 31.353,
+      "eval_sequential_score": 0.6940033401652869,
+      "eval_steps_per_second": 0.132,
+      "eval_sts-test_pearson_cosine": 0.6409463231775903,
+      "eval_sts-test_pearson_dot": 0.6394445667117477,
+      "eval_sts-test_pearson_euclidean": 0.6536559030100324,
+      "eval_sts-test_pearson_manhattan": 0.6598899192675511,
+      "eval_sts-test_pearson_max": 0.6598899192675511,
+      "eval_sts-test_spearman_cosine": 0.65107415339668,
+      "eval_sts-test_spearman_dot": 0.6495059349437412,
+      "eval_sts-test_spearman_euclidean": 0.6512988047288648,
+      "eval_sts-test_spearman_manhattan": 0.6569007676298402,
+      "eval_sts-test_spearman_max": 0.6569007676298402,
       "step": 1836
     },
     {
-      "epoch": 1.8070866141732282,
-      "grad_norm": 8.733999252319336,
+      "epoch": 1.8080708661417324,
+      "grad_norm": 11.966014862060547,
       "learning_rate": 6.489367000343005e-07,
-      "loss": 0.3578,
+      "loss": 0.6207,
       "step": 1837
     },
     {
-      "epoch": 1.8080708661417324,
-      "grad_norm": 13.05897331237793,
+      "epoch": 1.8090551181102361,
+      "grad_norm": 13.144506454467773,
       "learning_rate": 6.482040899296081e-07,
-      "loss": 0.5384,
+      "loss": 0.6119,
       "step": 1838
     },
     {
-      "epoch": 1.8090551181102361,
-      "grad_norm": 8.08520793914795,
+      "epoch": 1.8100393700787403,
+      "grad_norm": 13.123624801635742,
       "learning_rate": 6.474711329788802e-07,
-      "loss": 0.3188,
+      "loss": 0.6193,
       "step": 1839
     },
     {
-      "epoch": 1.8100393700787403,
-      "grad_norm": 4.597148895263672,
+      "epoch": 1.811023622047244,
+      "grad_norm": 9.731718063354492,
       "learning_rate": 6.467378309169858e-07,
-      "loss": 0.1175,
+      "loss": 0.5251,
       "step": 1840
     },
     {
-      "epoch": 1.811023622047244,
-      "grad_norm": 9.91828727722168,
+      "epoch": 1.8120078740157481,
+      "grad_norm": 11.679033279418945,
       "learning_rate": 6.460041854796105e-07,
-      "loss": 0.403,
+      "loss": 0.5778,
       "step": 1841
     },
     {
-      "epoch": 1.8120078740157481,
-      "grad_norm": 6.756091594696045,
+      "epoch": 1.8129921259842519,
+      "grad_norm": 9.962485313415527,
       "learning_rate": 6.452701984032526e-07,
-      "loss": 0.2802,
+      "loss": 0.457,
       "step": 1842
     },
     {
-      "epoch": 1.8129921259842519,
-      "grad_norm": 17.230257034301758,
+      "epoch": 1.813976377952756,
+      "grad_norm": 8.286011695861816,
       "learning_rate": 6.445358714252195e-07,
-      "loss": 0.8946,
+      "loss": 0.2371,
       "step": 1843
     },
     {
-      "epoch": 1.813976377952756,
-      "grad_norm": 14.641213417053223,
+      "epoch": 1.8149606299212597,
+      "grad_norm": 11.306199073791504,
       "learning_rate": 6.438012062836226e-07,
-      "loss": 0.678,
+      "loss": 0.4343,
       "step": 1844
     },
     {
-      "epoch": 1.8149606299212597,
-      "grad_norm": 10.255255699157715,
+      "epoch": 1.8159448818897639,
+      "grad_norm": 7.489519119262695,
       "learning_rate": 6.430662047173742e-07,
-      "loss": 0.4228,
+      "loss": 0.3065,
       "step": 1845
     },
     {
-      "epoch": 1.8159448818897639,
-      "grad_norm": 13.459108352661133,
+      "epoch": 1.8169291338582676,
+      "grad_norm": 15.40380859375,
       "learning_rate": 6.423308684661825e-07,
-      "loss": 0.848,
+      "loss": 1.0271,
       "step": 1846
     },
     {
-      "epoch": 1.8169291338582676,
-      "grad_norm": 13.858319282531738,
+      "epoch": 1.8179133858267718,
+      "grad_norm": 10.585190773010254,
       "learning_rate": 6.415951992705482e-07,
-      "loss": 0.9801,
+      "loss": 0.618,
       "step": 1847
     },
     {
-      "epoch": 1.8179133858267718,
-      "grad_norm": 9.967562675476074,
+      "epoch": 1.8188976377952755,
+      "grad_norm": 12.50681209564209,
       "learning_rate": 6.408591988717594e-07,
-      "loss": 0.4032,
+      "loss": 0.7867,
       "step": 1848
     },
     {
-      "epoch": 1.8188976377952755,
-      "grad_norm": 11.916775703430176,
+      "epoch": 1.8198818897637796,
+      "grad_norm": 9.978963851928711,
       "learning_rate": 6.401228690118893e-07,
-      "loss": 0.6389,
+      "loss": 0.3885,
       "step": 1849
     },
     {
-      "epoch": 1.8198818897637796,
-      "grad_norm": 13.487015724182129,
+      "epoch": 1.8208661417322833,
+      "grad_norm": 6.844351768493652,
       "learning_rate": 6.3938621143379e-07,
-      "loss": 0.4744,
+      "loss": 0.3757,
       "step": 1850
     },
     {
-      "epoch": 1.8208661417322833,
-      "grad_norm": 6.505801200866699,
+      "epoch": 1.8218503937007875,
+      "grad_norm": 12.80474853515625,
       "learning_rate": 6.386492278810894e-07,
-      "loss": 0.2109,
+      "loss": 0.7492,
       "step": 1851
     },
     {
-      "epoch": 1.8218503937007875,
-      "grad_norm": 13.570316314697266,
+      "epoch": 1.8228346456692912,
+      "grad_norm": 13.616663932800293,
       "learning_rate": 6.37911920098187e-07,
-      "loss": 0.7483,
+      "loss": 0.7607,
       "step": 1852
     },
     {
-      "epoch": 1.8228346456692912,
-      "grad_norm": 13.152098655700684,
+      "epoch": 1.8238188976377954,
+      "grad_norm": 9.825035095214844,
       "learning_rate": 6.371742898302503e-07,
-      "loss": 0.7438,
+      "loss": 0.287,
       "step": 1853
     },
     {
-      "epoch": 1.8238188976377954,
-      "grad_norm": 13.68462085723877,
+      "epoch": 1.824803149606299,
+      "grad_norm": 9.413983345031738,
       "learning_rate": 6.364363388232091e-07,
-      "loss": 0.6909,
+      "loss": 0.4212,
       "step": 1854
     },
     {
-      "epoch": 1.824803149606299,
-      "grad_norm": 10.51680850982666,
+      "epoch": 1.8257874015748032,
+      "grad_norm": 8.40971851348877,
       "learning_rate": 6.356980688237535e-07,
-      "loss": 0.5785,
+      "loss": 0.5182,
       "step": 1855
     },
     {
-      "epoch": 1.8257874015748032,
-      "grad_norm": 12.95541000366211,
+      "epoch": 1.826771653543307,
+      "grad_norm": 8.446026802062988,
       "learning_rate": 6.349594815793275e-07,
-      "loss": 0.6868,
+      "loss": 0.3489,
       "step": 1856
     },
     {
-      "epoch": 1.826771653543307,
-      "grad_norm": 10.486471176147461,
+      "epoch": 1.827755905511811,
+      "grad_norm": 11.696182250976562,
       "learning_rate": 6.34220578838127e-07,
-      "loss": 0.3436,
+      "loss": 0.535,
       "step": 1857
     },
     {
-      "epoch": 1.827755905511811,
-      "grad_norm": 14.526586532592773,
+      "epoch": 1.8287401574803148,
+      "grad_norm": 10.527113914489746,
       "learning_rate": 6.334813623490939e-07,
-      "loss": 1.1297,
+      "loss": 0.5276,
       "step": 1858
     },
     {
-      "epoch": 1.8287401574803148,
-      "grad_norm": 13.752555847167969,
+      "epoch": 1.829724409448819,
+      "grad_norm": 13.748953819274902,
       "learning_rate": 6.327418338619132e-07,
-      "loss": 0.8586,
+      "loss": 0.9417,
       "step": 1859
     },
     {
-      "epoch": 1.829724409448819,
-      "grad_norm": 24.5888729095459,
+      "epoch": 1.8307086614173227,
+      "grad_norm": 9.856019973754883,
       "learning_rate": 6.320019951270082e-07,
-      "loss": 1.3802,
+      "loss": 0.4258,
       "step": 1860
     },
     {
-      "epoch": 1.8307086614173227,
-      "grad_norm": 12.717546463012695,
+      "epoch": 1.8316929133858268,
+      "grad_norm": 9.23482608795166,
       "learning_rate": 6.312618478955368e-07,
-      "loss": 0.4151,
+      "loss": 0.3864,
       "step": 1861
     },
     {
-      "epoch": 1.8316929133858268,
-      "grad_norm": 9.724681854248047,
+      "epoch": 1.8326771653543306,
+      "grad_norm": 10.299756050109863,
       "learning_rate": 6.305213939193864e-07,
-      "loss": 0.3279,
+      "loss": 0.6552,
       "step": 1862
     },
     {
-      "epoch": 1.8326771653543306,
-      "grad_norm": 9.508359909057617,
+      "epoch": 1.8336614173228347,
+      "grad_norm": 9.539176940917969,
       "learning_rate": 6.297806349511711e-07,
-      "loss": 0.6177,
+      "loss": 0.4263,
       "step": 1863
     },
     {
-      "epoch": 1.8336614173228347,
-      "grad_norm": 10.602931022644043,
+      "epoch": 1.8346456692913384,
+      "grad_norm": 8.211603164672852,
       "learning_rate": 6.290395727442269e-07,
-      "loss": 0.4888,
+      "loss": 0.2489,
       "step": 1864
     },
     {
-      "epoch": 1.8346456692913384,
-      "grad_norm": 12.10499095916748,
+      "epoch": 1.8356299212598426,
+      "grad_norm": 10.142160415649414,
       "learning_rate": 6.282982090526071e-07,
-      "loss": 0.3599,
+      "loss": 0.5805,
       "step": 1865
     },
     {
-      "epoch": 1.8356299212598426,
-      "grad_norm": 7.4049272537231445,
+      "epoch": 1.8366141732283463,
+      "grad_norm": 9.113184928894043,
       "learning_rate": 6.27556545631079e-07,
-      "loss": 0.2112,
+      "loss": 0.3668,
       "step": 1866
     },
     {
-      "epoch": 1.8366141732283463,
-      "grad_norm": 9.28160285949707,
+      "epoch": 1.8375984251968505,
+      "grad_norm": 9.238096237182617,
       "learning_rate": 6.268145842351191e-07,
-      "loss": 0.2462,
+      "loss": 0.3874,
       "step": 1867
     },
     {
-      "epoch": 1.8375984251968505,
-      "grad_norm": 14.189945220947266,
+      "epoch": 1.8385826771653542,
+      "grad_norm": 15.01919174194336,
       "learning_rate": 6.260723266209094e-07,
-      "loss": 0.9246,
+      "loss": 1.1184,
       "step": 1868
     },
     {
-      "epoch": 1.8385826771653542,
-      "grad_norm": 10.58701229095459,
+      "epoch": 1.8395669291338583,
+      "grad_norm": 7.7761149406433105,
       "learning_rate": 6.253297745453329e-07,
-      "loss": 0.5992,
+      "loss": 0.2684,
       "step": 1869
     },
     {
-      "epoch": 1.8395669291338583,
-      "grad_norm": 16.576274871826172,
+      "epoch": 1.840551181102362,
+      "grad_norm": 7.935420513153076,
       "learning_rate": 6.245869297659694e-07,
-      "loss": 0.7455,
+      "loss": 0.3389,
       "step": 1870
     },
     {
-      "epoch": 1.840551181102362,
-      "grad_norm": 14.601115226745605,
+      "epoch": 1.8415354330708662,
+      "grad_norm": 8.66396713256836,
       "learning_rate": 6.238437940410921e-07,
-      "loss": 0.4792,
+      "loss": 0.3668,
       "step": 1871
     },
     {
-      "epoch": 1.8415354330708662,
-      "grad_norm": 13.090888023376465,
+      "epoch": 1.84251968503937,
+      "grad_norm": 13.415223121643066,
       "learning_rate": 6.231003691296622e-07,
-      "loss": 0.6048,
+      "loss": 0.9874,
       "step": 1872
     },
     {
-      "epoch": 1.84251968503937,
-      "grad_norm": 16.715045928955078,
+      "epoch": 1.843503937007874,
+      "grad_norm": 7.629619121551514,
       "learning_rate": 6.223566567913256e-07,
-      "loss": 1.1413,
+      "loss": 0.4333,
       "step": 1873
     },
     {
-      "epoch": 1.843503937007874,
-      "grad_norm": 10.588872909545898,
+      "epoch": 1.844488188976378,
+      "grad_norm": 11.345224380493164,
       "learning_rate": 6.216126587864084e-07,
-      "loss": 0.5035,
+      "loss": 0.6744,
       "step": 1874
     },
     {
-      "epoch": 1.844488188976378,
-      "grad_norm": 18.341047286987305,
+      "epoch": 1.845472440944882,
+      "grad_norm": 10.896759986877441,
       "learning_rate": 6.208683768759136e-07,
-      "loss": 0.8205,
+      "loss": 0.5133,
       "step": 1875
     },
     {
-      "epoch": 1.845472440944882,
-      "grad_norm": 10.638150215148926,
+      "epoch": 1.8464566929133859,
+      "grad_norm": 11.083782196044922,
       "learning_rate": 6.201238128215152e-07,
-      "loss": 0.3922,
+      "loss": 0.7328,
       "step": 1876
     },
     {
-      "epoch": 1.8464566929133859,
-      "grad_norm": 10.543109893798828,
+      "epoch": 1.8474409448818898,
+      "grad_norm": 8.74697494506836,
       "learning_rate": 6.193789683855555e-07,
-      "loss": 0.4792,
+      "loss": 0.2997,
       "step": 1877
     },
     {
-      "epoch": 1.8474409448818898,
-      "grad_norm": 7.625102996826172,
+      "epoch": 1.8484251968503937,
+      "grad_norm": 11.063111305236816,
       "learning_rate": 6.186338453310404e-07,
-      "loss": 0.2323,
+      "loss": 0.6519,
       "step": 1878
     },
     {
-      "epoch": 1.8484251968503937,
-      "grad_norm": 21.77037811279297,
+      "epoch": 1.8494094488188977,
+      "grad_norm": 12.17970085144043,
       "learning_rate": 6.178884454216355e-07,
-      "loss": 1.0537,
+      "loss": 0.5825,
       "step": 1879
     },
     {
-      "epoch": 1.8494094488188977,
-      "grad_norm": 11.93779182434082,
+      "epoch": 1.8503937007874016,
+      "grad_norm": 11.113762855529785,
       "learning_rate": 6.171427704216609e-07,
-      "loss": 0.7624,
+      "loss": 0.7681,
       "step": 1880
     },
     {
-      "epoch": 1.8503937007874016,
-      "grad_norm": 13.156431198120117,
+      "epoch": 1.8513779527559056,
+      "grad_norm": 8.454630851745605,
       "learning_rate": 6.163968220960891e-07,
-      "loss": 0.6861,
+      "loss": 0.3848,
       "step": 1881
     },
     {
-      "epoch": 1.8513779527559056,
-      "grad_norm": 10.549013137817383,
+      "epoch": 1.8523622047244095,
+      "grad_norm": 7.186916351318359,
       "learning_rate": 6.156506022105382e-07,
-      "loss": 0.4958,
+      "loss": 0.2117,
       "step": 1882
     },
     {
-      "epoch": 1.8523622047244095,
-      "grad_norm": 9.736396789550781,
+      "epoch": 1.8533464566929134,
+      "grad_norm": 11.546930313110352,
       "learning_rate": 6.1490411253127e-07,
-      "loss": 0.4112,
+      "loss": 1.0945,
       "step": 1883
     },
     {
-      "epoch": 1.8533464566929134,
-      "grad_norm": 11.155174255371094,
+      "epoch": 1.8543307086614174,
+      "grad_norm": 9.67233943939209,
       "learning_rate": 6.141573548251847e-07,
-      "loss": 0.6358,
+      "loss": 0.5184,
       "step": 1884
     },
     {
-      "epoch": 1.8543307086614174,
-      "grad_norm": 13.777209281921387,
+      "epoch": 1.8553149606299213,
+      "grad_norm": 11.523322105407715,
       "learning_rate": 6.134103308598166e-07,
-      "loss": 0.7354,
+      "loss": 0.4794,
       "step": 1885
     },
     {
-      "epoch": 1.8553149606299213,
-      "grad_norm": 13.100604057312012,
+      "epoch": 1.8562992125984252,
+      "grad_norm": 16.110071182250977,
       "learning_rate": 6.126630424033303e-07,
-      "loss": 0.7199,
+      "loss": 0.8211,
       "step": 1886
     },
     {
-      "epoch": 1.8562992125984252,
-      "grad_norm": 12.631733894348145,
+      "epoch": 1.8572834645669292,
+      "grad_norm": 10.432527542114258,
       "learning_rate": 6.119154912245167e-07,
-      "loss": 0.8309,
+      "loss": 0.495,
       "step": 1887
     },
     {
-      "epoch": 1.8572834645669292,
-      "grad_norm": 11.478351593017578,
+      "epoch": 1.858267716535433,
+      "grad_norm": 9.711058616638184,
       "learning_rate": 6.111676790927884e-07,
-      "loss": 0.6902,
+      "loss": 0.4292,
       "step": 1888
     },
     {
-      "epoch": 1.858267716535433,
-      "grad_norm": 5.461014270782471,
+      "epoch": 1.859251968503937,
+      "grad_norm": 12.808670997619629,
       "learning_rate": 6.104196077781757e-07,
-      "loss": 0.1763,
+      "loss": 0.7122,
       "step": 1889
     },
     {
-      "epoch": 1.859251968503937,
-      "grad_norm": 12.026067733764648,
+      "epoch": 1.860236220472441,
+      "grad_norm": 9.233994483947754,
       "learning_rate": 6.096712790513222e-07,
-      "loss": 0.6163,
+      "loss": 0.4032,
       "step": 1890
     },
     {
-      "epoch": 1.860236220472441,
-      "grad_norm": 11.317314147949219,
+      "epoch": 1.861220472440945,
+      "grad_norm": 11.49265193939209,
       "learning_rate": 6.08922694683481e-07,
-      "loss": 0.5629,
+      "loss": 0.5003,
       "step": 1891
     },
     {
-      "epoch": 1.861220472440945,
-      "grad_norm": 13.000326156616211,
+      "epoch": 1.8622047244094488,
+      "grad_norm": 11.93250846862793,
       "learning_rate": 6.081738564465103e-07,
-      "loss": 0.7518,
+      "loss": 0.5014,
       "step": 1892
     },
     {
-      "epoch": 1.8622047244094488,
-      "grad_norm": 13.375837326049805,
+      "epoch": 1.8631889763779528,
+      "grad_norm": 11.877928733825684,
       "learning_rate": 6.07424766112869e-07,
-      "loss": 0.6893,
+      "loss": 0.3998,
       "step": 1893
     },
     {
-      "epoch": 1.8631889763779528,
-      "grad_norm": 8.774999618530273,
+      "epoch": 1.8641732283464567,
+      "grad_norm": 10.758837699890137,
       "learning_rate": 6.066754254556127e-07,
-      "loss": 0.3519,
+      "loss": 0.3397,
       "step": 1894
     },
     {
-      "epoch": 1.8641732283464567,
-      "grad_norm": 5.026043891906738,
+      "epoch": 1.8651574803149606,
+      "grad_norm": 9.902132987976074,
       "learning_rate": 6.059258362483897e-07,
-      "loss": 0.121,
+      "loss": 0.4174,
       "step": 1895
     },
     {
-      "epoch": 1.8651574803149606,
-      "grad_norm": 14.182245254516602,
+      "epoch": 1.8661417322834646,
+      "grad_norm": 9.253582000732422,
       "learning_rate": 6.051760002654364e-07,
-      "loss": 0.5967,
+      "loss": 0.5333,
       "step": 1896
     },
     {
-      "epoch": 1.8661417322834646,
-      "grad_norm": 8.58065128326416,
+      "epoch": 1.8671259842519685,
+      "grad_norm": 11.772601127624512,
       "learning_rate": 6.044259192815735e-07,
-      "loss": 0.3536,
+      "loss": 0.575,
       "step": 1897
     },
     {
-      "epoch": 1.8671259842519685,
-      "grad_norm": 11.549651145935059,
+      "epoch": 1.8681102362204725,
+      "grad_norm": 10.690569877624512,
       "learning_rate": 6.036755950722014e-07,
-      "loss": 0.6354,
+      "loss": 0.5142,
       "step": 1898
     },
     {
-      "epoch": 1.8681102362204725,
-      "grad_norm": 12.778814315795898,
+      "epoch": 1.8690944881889764,
+      "grad_norm": 13.561893463134766,
       "learning_rate": 6.029250294132965e-07,
-      "loss": 0.6448,
+      "loss": 0.6661,
       "step": 1899
     },
     {
-      "epoch": 1.8690944881889764,
-      "grad_norm": 9.952596664428711,
+      "epoch": 1.8700787401574803,
+      "grad_norm": 13.437638282775879,
       "learning_rate": 6.021742240814067e-07,
-      "loss": 0.3805,
+      "loss": 0.5965,
       "step": 1900
     },
     {
-      "epoch": 1.8700787401574803,
-      "grad_norm": 7.297644138336182,
+      "epoch": 1.8710629921259843,
+      "grad_norm": 10.452093124389648,
       "learning_rate": 6.014231808536463e-07,
-      "loss": 0.2854,
+      "loss": 0.3209,
       "step": 1901
     },
     {
-      "epoch": 1.8710629921259843,
-      "grad_norm": 22.722822189331055,
+      "epoch": 1.8720472440944882,
+      "grad_norm": 9.1297607421875,
       "learning_rate": 6.006719015076938e-07,
-      "loss": 1.1512,
+      "loss": 0.3278,
       "step": 1902
     },
     {
-      "epoch": 1.8720472440944882,
-      "grad_norm": 10.664011001586914,
+      "epoch": 1.8730314960629921,
+      "grad_norm": 11.424860000610352,
       "learning_rate": 5.999203878217861e-07,
-      "loss": 0.3825,
+      "loss": 0.6277,
       "step": 1903
     },
     {
-      "epoch": 1.8730314960629921,
-      "grad_norm": 9.881097793579102,
+      "epoch": 1.874015748031496,
+      "grad_norm": 12.790457725524902,
       "learning_rate": 5.99168641574715e-07,
-      "loss": 0.4156,
+      "loss": 0.7779,
       "step": 1904
     },
     {
-      "epoch": 1.874015748031496,
-      "grad_norm": 8.037322044372559,
+      "epoch": 1.875,
+      "grad_norm": 10.565644264221191,
       "learning_rate": 5.984166645458224e-07,
-      "loss": 0.2155,
+      "loss": 0.3863,
       "step": 1905
     },
     {
-      "epoch": 1.875,
-      "grad_norm": 21.00752830505371,
+      "epoch": 1.875984251968504,
+      "grad_norm": 10.955162048339844,
       "learning_rate": 5.976644585149967e-07,
-      "loss": 0.9292,
+      "loss": 0.4736,
       "step": 1906
     },
     {
-      "epoch": 1.875984251968504,
-      "grad_norm": 15.958588600158691,
+      "epoch": 1.8769685039370079,
+      "grad_norm": 8.910943031311035,
       "learning_rate": 5.969120252626684e-07,
-      "loss": 0.6131,
+      "loss": 0.3502,
       "step": 1907
     },
     {
-      "epoch": 1.8769685039370079,
-      "grad_norm": 13.447348594665527,
+      "epoch": 1.8779527559055118,
+      "grad_norm": 14.70986270904541,
       "learning_rate": 5.961593665698055e-07,
-      "loss": 0.516,
+      "loss": 0.8825,
       "step": 1908
     },
     {
-      "epoch": 1.8779527559055118,
-      "grad_norm": 11.870959281921387,
+      "epoch": 1.8789370078740157,
+      "grad_norm": 15.559196472167969,
       "learning_rate": 5.954064842179099e-07,
-      "loss": 0.829,
+      "loss": 0.8411,
       "step": 1909
     },
     {
-      "epoch": 1.8789370078740157,
-      "grad_norm": 14.993748664855957,
+      "epoch": 1.8799212598425197,
+      "grad_norm": 6.268329620361328,
       "learning_rate": 5.946533799890127e-07,
-      "loss": 0.7785,
+      "loss": 0.1829,
       "step": 1910
     },
     {
-      "epoch": 1.8799212598425197,
-      "grad_norm": 10.761703491210938,
+      "epoch": 1.8809055118110236,
+      "grad_norm": 15.748644828796387,
       "learning_rate": 5.939000556656704e-07,
-      "loss": 0.4304,
+      "loss": 1.0817,
       "step": 1911
     },
     {
-      "epoch": 1.8809055118110236,
-      "grad_norm": 11.223014831542969,
+      "epoch": 1.8818897637795275,
+      "grad_norm": 10.739591598510742,
       "learning_rate": 5.931465130309604e-07,
-      "loss": 0.4163,
+      "loss": 0.6255,
       "step": 1912
     },
     {
-      "epoch": 1.8818897637795275,
-      "grad_norm": 10.408675193786621,
+      "epoch": 1.8828740157480315,
+      "grad_norm": 11.86316967010498,
       "learning_rate": 5.923927538684764e-07,
-      "loss": 0.4314,
+      "loss": 0.4679,
       "step": 1913
     },
     {
-      "epoch": 1.8828740157480315,
-      "grad_norm": 10.93159294128418,
+      "epoch": 1.8838582677165354,
+      "grad_norm": 15.708845138549805,
       "learning_rate": 5.916387799623253e-07,
-      "loss": 0.5409,
+      "loss": 0.7575,
       "step": 1914
     },
     {
-      "epoch": 1.8838582677165354,
-      "grad_norm": 5.786235809326172,
+      "epoch": 1.8848425196850394,
+      "grad_norm": 18.941633224487305,
       "learning_rate": 5.908845930971217e-07,
-      "loss": 0.1406,
+      "loss": 1.062,
       "step": 1915
     },
     {
-      "epoch": 1.8848425196850394,
-      "grad_norm": 10.518360137939453,
+      "epoch": 1.8858267716535433,
+      "grad_norm": 11.85770320892334,
       "learning_rate": 5.901301950579845e-07,
-      "loss": 0.4662,
+      "loss": 0.6546,
       "step": 1916
     },
     {
-      "epoch": 1.8858267716535433,
-      "grad_norm": 12.29217529296875,
+      "epoch": 1.8868110236220472,
+      "grad_norm": 10.091187477111816,
       "learning_rate": 5.893755876305328e-07,
-      "loss": 0.5706,
+      "loss": 0.5099,
       "step": 1917
     },
     {
-      "epoch": 1.8868110236220472,
-      "grad_norm": 14.452948570251465,
+      "epoch": 1.8877952755905512,
+      "grad_norm": 11.632383346557617,
       "learning_rate": 5.886207726008805e-07,
-      "loss": 0.6524,
+      "loss": 0.5189,
       "step": 1918
     },
     {
-      "epoch": 1.8877952755905512,
-      "grad_norm": 13.0665283203125,
+      "epoch": 1.888779527559055,
+      "grad_norm": 7.834343910217285,
       "learning_rate": 5.878657517556336e-07,
-      "loss": 0.5334,
+      "loss": 0.3031,
       "step": 1919
     },
     {
-      "epoch": 1.888779527559055,
-      "grad_norm": 9.819680213928223,
+      "epoch": 1.889763779527559,
+      "grad_norm": 11.638785362243652,
       "learning_rate": 5.871105268818848e-07,
-      "loss": 0.2941,
+      "loss": 0.6165,
       "step": 1920
     },
     {
-      "epoch": 1.889763779527559,
-      "grad_norm": 13.505655288696289,
+      "epoch": 1.890748031496063,
+      "grad_norm": 13.849247932434082,
       "learning_rate": 5.863550997672101e-07,
-      "loss": 1.0042,
+      "loss": 0.8606,
       "step": 1921
     },
     {
-      "epoch": 1.890748031496063,
-      "grad_norm": 11.612969398498535,
+      "epoch": 1.891732283464567,
+      "grad_norm": 8.213641166687012,
       "learning_rate": 5.855994721996638e-07,
-      "loss": 0.6607,
+      "loss": 0.4711,
       "step": 1922
     },
     {
-      "epoch": 1.891732283464567,
-      "grad_norm": 13.760869979858398,
+      "epoch": 1.8927165354330708,
+      "grad_norm": 11.465242385864258,
       "learning_rate": 5.848436459677752e-07,
-      "loss": 0.6232,
+      "loss": 0.6318,
       "step": 1923
     },
     {
-      "epoch": 1.8927165354330708,
-      "grad_norm": 11.171491622924805,
+      "epoch": 1.8937007874015748,
+      "grad_norm": 9.324861526489258,
       "learning_rate": 5.840876228605432e-07,
-      "loss": 0.7116,
+      "loss": 0.4974,
       "step": 1924
     },
     {
-      "epoch": 1.8937007874015748,
-      "grad_norm": 10.05971908569336,
+      "epoch": 1.8946850393700787,
+      "grad_norm": 13.302372932434082,
       "learning_rate": 5.833314046674329e-07,
-      "loss": 0.5706,
+      "loss": 0.7208,
       "step": 1925
     },
     {
-      "epoch": 1.8946850393700787,
-      "grad_norm": 12.375988960266113,
+      "epoch": 1.8956692913385826,
+      "grad_norm": 11.545489311218262,
       "learning_rate": 5.825749931783715e-07,
-      "loss": 0.5863,
+      "loss": 0.5613,
       "step": 1926
     },
     {
-      "epoch": 1.8956692913385826,
-      "grad_norm": 7.188908576965332,
+      "epoch": 1.8966535433070866,
+      "grad_norm": 10.206701278686523,
       "learning_rate": 5.818183901837435e-07,
-      "loss": 0.2272,
+      "loss": 0.4673,
       "step": 1927
     },
     {
-      "epoch": 1.8966535433070866,
-      "grad_norm": 9.860791206359863,
+      "epoch": 1.8976377952755905,
+      "grad_norm": 8.953893661499023,
       "learning_rate": 5.810615974743864e-07,
-      "loss": 0.3337,
+      "loss": 0.4062,
       "step": 1928
     },
     {
-      "epoch": 1.8976377952755905,
-      "grad_norm": 11.774974822998047,
+      "epoch": 1.8986220472440944,
+      "grad_norm": 10.11854362487793,
       "learning_rate": 5.803046168415872e-07,
-      "loss": 0.7918,
+      "loss": 0.4713,
       "step": 1929
     },
     {
-      "epoch": 1.8986220472440944,
-      "grad_norm": 8.987882614135742,
+      "epoch": 1.8996062992125984,
+      "grad_norm": 11.15442180633545,
       "learning_rate": 5.795474500770772e-07,
-      "loss": 0.3992,
+      "loss": 0.4277,
       "step": 1930
     },
     {
-      "epoch": 1.8996062992125984,
-      "grad_norm": 7.925570487976074,
+      "epoch": 1.9005905511811023,
+      "grad_norm": 13.043636322021484,
       "learning_rate": 5.787900989730291e-07,
-      "loss": 0.2711,
+      "loss": 0.9161,
       "step": 1931
     },
     {
-      "epoch": 1.9005905511811023,
-      "grad_norm": 21.005599975585938,
+      "epoch": 1.9015748031496063,
+      "grad_norm": 10.691286087036133,
       "learning_rate": 5.780325653220509e-07,
-      "loss": 0.7199,
+      "loss": 0.4953,
       "step": 1932
     },
     {
-      "epoch": 1.9015748031496063,
-      "grad_norm": 9.796464920043945,
+      "epoch": 1.9025590551181102,
+      "grad_norm": 8.616952896118164,
       "learning_rate": 5.772748509171833e-07,
-      "loss": 0.3781,
+      "loss": 0.393,
       "step": 1933
     },
     {
-      "epoch": 1.9025590551181102,
-      "grad_norm": 5.327077865600586,
+      "epoch": 1.9035433070866141,
+      "grad_norm": 6.288427829742432,
       "learning_rate": 5.765169575518948e-07,
-      "loss": 0.1524,
+      "loss": 0.4689,
       "step": 1934
     },
     {
-      "epoch": 1.9035433070866141,
-      "grad_norm": 3.7077765464782715,
+      "epoch": 1.904527559055118,
+      "grad_norm": 13.286681175231934,
       "learning_rate": 5.757588870200772e-07,
-      "loss": 0.2347,
+      "loss": 0.9634,
       "step": 1935
     },
     {
-      "epoch": 1.904527559055118,
-      "grad_norm": 14.269500732421875,
+      "epoch": 1.905511811023622,
+      "grad_norm": 12.330071449279785,
       "learning_rate": 5.750006411160421e-07,
-      "loss": 0.8348,
+      "loss": 0.9026,
       "step": 1936
     },
     {
-      "epoch": 1.905511811023622,
-      "grad_norm": 12.124266624450684,
+      "epoch": 1.906496062992126,
+      "grad_norm": 8.335945129394531,
       "learning_rate": 5.742422216345157e-07,
-      "loss": 0.5498,
+      "loss": 0.3855,
       "step": 1937
     },
     {
-      "epoch": 1.906496062992126,
-      "grad_norm": 6.518699645996094,
+      "epoch": 1.90748031496063,
+      "grad_norm": 9.104451179504395,
       "learning_rate": 5.734836303706352e-07,
-      "loss": 0.226,
+      "loss": 0.3751,
       "step": 1938
     },
     {
-      "epoch": 1.90748031496063,
-      "grad_norm": 12.600378036499023,
+      "epoch": 1.9084645669291338,
+      "grad_norm": 10.96930980682373,
       "learning_rate": 5.727248691199446e-07,
-      "loss": 0.719,
+      "loss": 0.7743,
       "step": 1939
     },
     {
-      "epoch": 1.9084645669291338,
-      "grad_norm": 6.58778190612793,
+      "epoch": 1.909448818897638,
+      "grad_norm": 11.329811096191406,
       "learning_rate": 5.719659396783902e-07,
-      "loss": 0.2547,
+      "loss": 0.5837,
       "step": 1940
     },
     {
-      "epoch": 1.909448818897638,
-      "grad_norm": 11.544054985046387,
+      "epoch": 1.9104330708661417,
+      "grad_norm": 2.9978647232055664,
       "learning_rate": 5.712068438423161e-07,
-      "loss": 0.5611,
+      "loss": 0.0811,
       "step": 1941
     },
     {
-      "epoch": 1.9104330708661417,
-      "grad_norm": 5.600453853607178,
+      "epoch": 1.9114173228346458,
+      "grad_norm": 10.107305526733398,
       "learning_rate": 5.704475834084608e-07,
-      "loss": 0.1426,
+      "loss": 0.3471,
       "step": 1942
     },
     {
-      "epoch": 1.9114173228346458,
-      "grad_norm": 14.013601303100586,
+      "epoch": 1.9124015748031495,
+      "grad_norm": 11.733490943908691,
       "learning_rate": 5.696881601739517e-07,
-      "loss": 0.743,
+      "loss": 0.6189,
       "step": 1943
     },
     {
-      "epoch": 1.9124015748031495,
-      "grad_norm": 12.552058219909668,
+      "epoch": 1.9133858267716537,
+      "grad_norm": 8.555854797363281,
       "learning_rate": 5.689285759363022e-07,
-      "loss": 0.5874,
+      "loss": 0.35,
       "step": 1944
     },
     {
-      "epoch": 1.9133858267716537,
-      "grad_norm": 12.306783676147461,
+      "epoch": 1.9143700787401574,
+      "grad_norm": 11.3431396484375,
       "learning_rate": 5.681688324934064e-07,
-      "loss": 0.3334,
+      "loss": 0.4165,
       "step": 1945
     },
     {
-      "epoch": 1.9143700787401574,
-      "grad_norm": 12.582601547241211,
+      "epoch": 1.9153543307086616,
+      "grad_norm": 16.661283493041992,
       "learning_rate": 5.674089316435353e-07,
-      "loss": 0.4916,
+      "loss": 0.9556,
       "step": 1946
     },
     {
-      "epoch": 1.9153543307086616,
-      "grad_norm": 13.343052864074707,
+      "epoch": 1.9163385826771653,
+      "grad_norm": 9.4197416305542,
       "learning_rate": 5.666488751853321e-07,
-      "loss": 0.7915,
+      "loss": 0.4205,
       "step": 1947
     },
     {
-      "epoch": 1.9163385826771653,
-      "grad_norm": 4.297492027282715,
+      "epoch": 1.9173228346456694,
+      "grad_norm": 11.713933944702148,
       "learning_rate": 5.658886649178096e-07,
-      "loss": 0.1582,
+      "loss": 0.4595,
       "step": 1948
     },
     {
-      "epoch": 1.9173228346456694,
-      "grad_norm": 12.203116416931152,
+      "epoch": 1.9183070866141732,
+      "grad_norm": 15.105450630187988,
       "learning_rate": 5.651283026403427e-07,
-      "loss": 0.408,
+      "loss": 0.4687,
       "step": 1949
     },
     {
-      "epoch": 1.9183070866141732,
-      "grad_norm": 11.420016288757324,
+      "epoch": 1.9192913385826773,
+      "grad_norm": 10.329501152038574,
       "learning_rate": 5.64367790152668e-07,
-      "loss": 0.4246,
+      "loss": 0.3396,
       "step": 1950
     },
     {
-      "epoch": 1.9192913385826773,
-      "grad_norm": 11.461502075195312,
+      "epoch": 1.920275590551181,
+      "grad_norm": 13.009134292602539,
       "learning_rate": 5.636071292548766e-07,
-      "loss": 0.5638,
+      "loss": 0.689,
       "step": 1951
     },
     {
-      "epoch": 1.920275590551181,
-      "grad_norm": 10.541890144348145,
+      "epoch": 1.9212598425196852,
+      "grad_norm": 10.723394393920898,
       "learning_rate": 5.628463217474108e-07,
-      "loss": 0.4243,
+      "loss": 0.43,
       "step": 1952
     },
     {
-      "epoch": 1.9212598425196852,
-      "grad_norm": 9.319199562072754,
+      "epoch": 1.922244094488189,
+      "grad_norm": 5.5873565673828125,
       "learning_rate": 5.620853694310604e-07,
-      "loss": 0.4148,
+      "loss": 0.1506,
       "step": 1953
     },
     {
-      "epoch": 1.922244094488189,
-      "grad_norm": 9.032517433166504,
+      "epoch": 1.923228346456693,
+      "grad_norm": 10.47590160369873,
       "learning_rate": 5.613242741069578e-07,
-      "loss": 0.268,
+      "loss": 0.5124,
       "step": 1954
     },
     {
-      "epoch": 1.923228346456693,
-      "grad_norm": 6.920299053192139,
+      "epoch": 1.9242125984251968,
+      "grad_norm": 12.383049011230469,
       "learning_rate": 5.605630375765739e-07,
-      "loss": 0.2287,
+      "loss": 0.5966,
       "step": 1955
     },
     {
-      "epoch": 1.9242125984251968,
-      "grad_norm": 8.552000999450684,
+      "epoch": 1.925196850393701,
+      "grad_norm": 13.749089241027832,
       "learning_rate": 5.598016616417136e-07,
-      "loss": 0.1981,
+      "loss": 0.9889,
       "step": 1956
     },
     {
-      "epoch": 1.925196850393701,
-      "grad_norm": 28.06435775756836,
+      "epoch": 1.9261811023622046,
+      "grad_norm": 10.72147274017334,
       "learning_rate": 5.590401481045118e-07,
-      "loss": 1.661,
+      "loss": 0.7009,
       "step": 1957
     },
     {
-      "epoch": 1.9261811023622046,
-      "grad_norm": 17.66850471496582,
+      "epoch": 1.9271653543307088,
+      "grad_norm": 14.116695404052734,
       "learning_rate": 5.582784987674296e-07,
-      "loss": 0.6172,
+      "loss": 0.5476,
       "step": 1958
     },
     {
-      "epoch": 1.9271653543307088,
-      "grad_norm": 10.603317260742188,
+      "epoch": 1.9281496062992125,
+      "grad_norm": 11.435696601867676,
       "learning_rate": 5.575167154332489e-07,
-      "loss": 0.7265,
+      "loss": 0.4462,
       "step": 1959
     },
     {
-      "epoch": 1.9281496062992125,
-      "grad_norm": 10.895527839660645,
+      "epoch": 1.9291338582677167,
+      "grad_norm": 10.497963905334473,
       "learning_rate": 5.567547999050692e-07,
-      "loss": 0.4045,
+      "loss": 0.465,
       "step": 1960
     },
     {
-      "epoch": 1.9291338582677167,
-      "grad_norm": 12.122705459594727,
+      "epoch": 1.9301181102362204,
+      "grad_norm": 11.568528175354004,
       "learning_rate": 5.559927539863024e-07,
-      "loss": 0.3756,
+      "loss": 0.6114,
       "step": 1961
     },
     {
-      "epoch": 1.9301181102362204,
-      "grad_norm": 12.188031196594238,
+      "epoch": 1.9311023622047245,
+      "grad_norm": 5.820825099945068,
       "learning_rate": 5.552305794806699e-07,
-      "loss": 0.5601,
+      "loss": 0.1944,
       "step": 1962
     },
     {
-      "epoch": 1.9311023622047245,
-      "grad_norm": 8.183180809020996,
+      "epoch": 1.9320866141732282,
+      "grad_norm": 10.468626022338867,
       "learning_rate": 5.544682781921964e-07,
-      "loss": 0.2951,
+      "loss": 0.4095,
       "step": 1963
     },
     {
-      "epoch": 1.9320866141732282,
-      "grad_norm": 20.317163467407227,
+      "epoch": 1.9330708661417324,
+      "grad_norm": 10.831794738769531,
       "learning_rate": 5.537058519252076e-07,
-      "loss": 0.8282,
+      "loss": 0.5479,
       "step": 1964
     },
     {
-      "epoch": 1.9330708661417324,
-      "grad_norm": 10.017477989196777,
+      "epoch": 1.9340551181102361,
+      "grad_norm": 14.258051872253418,
       "learning_rate": 5.529433024843244e-07,
-      "loss": 0.3984,
+      "loss": 0.5858,
       "step": 1965
     },
     {
-      "epoch": 1.9340551181102361,
-      "grad_norm": 12.640592575073242,
+      "epoch": 1.9350393700787403,
+      "grad_norm": 7.533563137054443,
       "learning_rate": 5.521806316744594e-07,
-      "loss": 0.6055,
+      "loss": 0.2531,
       "step": 1966
     },
     {
-      "epoch": 1.9350393700787403,
-      "grad_norm": 10.652737617492676,
+      "epoch": 1.936023622047244,
+      "grad_norm": 8.989212036132812,
       "learning_rate": 5.514178413008125e-07,
-      "loss": 0.5407,
+      "loss": 0.4321,
       "step": 1967
     },
     {
-      "epoch": 1.936023622047244,
-      "grad_norm": 7.8703837394714355,
+      "epoch": 1.9370078740157481,
+      "grad_norm": 9.187650680541992,
       "learning_rate": 5.506549331688669e-07,
-      "loss": 0.2173,
+      "loss": 0.5694,
       "step": 1968
     },
     {
-      "epoch": 1.9370078740157481,
-      "grad_norm": 16.282611846923828,
+      "epoch": 1.9379921259842519,
+      "grad_norm": 8.983762741088867,
       "learning_rate": 5.49891909084384e-07,
-      "loss": 0.5948,
+      "loss": 0.2313,
       "step": 1969
     },
     {
-      "epoch": 1.9379921259842519,
-      "grad_norm": 9.560543060302734,
+      "epoch": 1.938976377952756,
+      "grad_norm": 11.622753143310547,
       "learning_rate": 5.491287708534e-07,
-      "loss": 0.4105,
+      "loss": 0.4807,
       "step": 1970
     },
     {
-      "epoch": 1.938976377952756,
-      "grad_norm": 6.578395843505859,
+      "epoch": 1.9399606299212597,
+      "grad_norm": 8.603584289550781,
       "learning_rate": 5.483655202822214e-07,
-      "loss": 0.1847,
+      "loss": 0.2989,
       "step": 1971
     },
     {
-      "epoch": 1.9399606299212597,
-      "grad_norm": 9.591692924499512,
+      "epoch": 1.9409448818897639,
+      "grad_norm": 11.923957824707031,
       "learning_rate": 5.4760215917742e-07,
-      "loss": 0.4112,
+      "loss": 0.6416,
       "step": 1972
     },
     {
-      "epoch": 1.9409448818897639,
-      "grad_norm": 5.674706935882568,
+      "epoch": 1.9419291338582676,
+      "grad_norm": 11.62790584564209,
       "learning_rate": 5.4683868934583e-07,
-      "loss": 0.1703,
+      "loss": 0.4995,
       "step": 1973
     },
     {
-      "epoch": 1.9419291338582676,
-      "grad_norm": 10.246027946472168,
+      "epoch": 1.9429133858267718,
+      "grad_norm": 12.349837303161621,
       "learning_rate": 5.460751125945424e-07,
-      "loss": 0.4364,
+      "loss": 0.8427,
       "step": 1974
     },
     {
-      "epoch": 1.9429133858267718,
-      "grad_norm": 15.110301971435547,
+      "epoch": 1.9438976377952755,
+      "grad_norm": 8.440821647644043,
       "learning_rate": 5.453114307309015e-07,
-      "loss": 0.4748,
+      "loss": 0.281,
       "step": 1975
     },
     {
-      "epoch": 1.9438976377952755,
-      "grad_norm": 7.355463981628418,
+      "epoch": 1.9448818897637796,
+      "grad_norm": 7.2062506675720215,
       "learning_rate": 5.445476455625002e-07,
-      "loss": 0.224,
+      "loss": 0.2401,
       "step": 1976
     },
     {
-      "epoch": 1.9448818897637796,
-      "grad_norm": 13.137723922729492,
+      "epoch": 1.9458661417322833,
+      "grad_norm": 9.149591445922852,
       "learning_rate": 5.437837588971761e-07,
-      "loss": 0.8408,
+      "loss": 0.347,
       "step": 1977
     },
     {
-      "epoch": 1.9458661417322833,
-      "grad_norm": 14.898588180541992,
+      "epoch": 1.9468503937007875,
+      "grad_norm": 11.855673789978027,
       "learning_rate": 5.430197725430071e-07,
-      "loss": 0.5296,
+      "loss": 0.6268,
       "step": 1978
     },
     {
-      "epoch": 1.9468503937007875,
-      "grad_norm": 15.721670150756836,
+      "epoch": 1.9478346456692912,
+      "grad_norm": 13.50779914855957,
       "learning_rate": 5.422556883083069e-07,
-      "loss": 0.7688,
+      "loss": 0.7361,
       "step": 1979
     },
     {
-      "epoch": 1.9478346456692912,
-      "grad_norm": 13.495804786682129,
+      "epoch": 1.9488188976377954,
+      "grad_norm": 9.99897289276123,
       "learning_rate": 5.414915080016209e-07,
-      "loss": 1.0069,
+      "loss": 0.676,
       "step": 1980
     },
     {
-      "epoch": 1.9488188976377954,
-      "grad_norm": 22.801025390625,
+      "epoch": 1.949803149606299,
+      "grad_norm": 10.191953659057617,
       "learning_rate": 5.407272334317221e-07,
-      "loss": 1.1509,
+      "loss": 0.4061,
       "step": 1981
     },
     {
-      "epoch": 1.949803149606299,
-      "grad_norm": 14.018686294555664,
+      "epoch": 1.9507874015748032,
+      "grad_norm": 12.090611457824707,
       "learning_rate": 5.399628664076062e-07,
-      "loss": 0.5735,
+      "loss": 0.8464,
       "step": 1982
     },
     {
-      "epoch": 1.9507874015748032,
-      "grad_norm": 12.201386451721191,
+      "epoch": 1.951771653543307,
+      "grad_norm": 13.784215927124023,
       "learning_rate": 5.391984087384882e-07,
-      "loss": 0.5137,
+      "loss": 0.8171,
       "step": 1983
     },
     {
-      "epoch": 1.951771653543307,
-      "grad_norm": 11.208210945129395,
+      "epoch": 1.952755905511811,
+      "grad_norm": 12.421671867370605,
       "learning_rate": 5.384338622337973e-07,
-      "loss": 0.3373,
+      "loss": 0.796,
       "step": 1984
     },
     {
-      "epoch": 1.952755905511811,
-      "grad_norm": 10.17148494720459,
+      "epoch": 1.9537401574803148,
+      "grad_norm": 8.954252243041992,
       "learning_rate": 5.37669228703173e-07,
-      "loss": 0.3546,
+      "loss": 0.4251,
       "step": 1985
     },
     {
-      "epoch": 1.9537401574803148,
-      "grad_norm": 12.111408233642578,
+      "epoch": 1.954724409448819,
+      "grad_norm": 11.040847778320312,
       "learning_rate": 5.369045099564613e-07,
-      "loss": 0.5453,
+      "loss": 0.5535,
       "step": 1986
     },
     {
-      "epoch": 1.954724409448819,
-      "grad_norm": 18.0567569732666,
+      "epoch": 1.9557086614173227,
+      "grad_norm": 8.30056381225586,
       "learning_rate": 5.361397078037089e-07,
-      "loss": 0.8836,
+      "loss": 0.3494,
       "step": 1987
     },
     {
-      "epoch": 1.9557086614173227,
-      "grad_norm": 8.216286659240723,
+      "epoch": 1.9566929133858268,
+      "grad_norm": 10.169771194458008,
       "learning_rate": 5.353748240551612e-07,
-      "loss": 0.2828,
+      "loss": 0.5363,
       "step": 1988
     },
     {
-      "epoch": 1.9566929133858268,
-      "grad_norm": 9.901942253112793,
+      "epoch": 1.9576771653543306,
+      "grad_norm": 9.978471755981445,
       "learning_rate": 5.346098605212553e-07,
-      "loss": 0.3481,
+      "loss": 0.5541,
       "step": 1989
     },
     {
-      "epoch": 1.9566929133858268,
-      "eval_Qnli-dev_cosine_accuracy": 0.658203125,
-      "eval_Qnli-dev_cosine_accuracy_threshold": 0.8023818731307983,
-      "eval_Qnli-dev_cosine_ap": 0.6886786340579703,
-      "eval_Qnli-dev_cosine_f1": 0.6916802610114192,
-      "eval_Qnli-dev_cosine_f1_threshold": 0.7310751676559448,
-      "eval_Qnli-dev_cosine_precision": 0.5623342175066313,
-      "eval_Qnli-dev_cosine_recall": 0.8983050847457628,
-      "eval_Qnli-dev_dot_accuracy": 0.66015625,
-      "eval_Qnli-dev_dot_accuracy_threshold": 615.7371826171875,
-      "eval_Qnli-dev_dot_ap": 0.6875993155505726,
-      "eval_Qnli-dev_dot_f1": 0.6916802610114192,
-      "eval_Qnli-dev_dot_f1_threshold": 558.3238525390625,
-      "eval_Qnli-dev_dot_precision": 0.5623342175066313,
-      "eval_Qnli-dev_dot_recall": 0.8983050847457628,
-      "eval_Qnli-dev_euclidean_accuracy": 0.662109375,
-      "eval_Qnli-dev_euclidean_accuracy_threshold": 17.46044158935547,
-      "eval_Qnli-dev_euclidean_ap": 0.6887027301873778,
-      "eval_Qnli-dev_euclidean_f1": 0.6906710310965629,
-      "eval_Qnli-dev_euclidean_f1_threshold": 20.146350860595703,
-      "eval_Qnli-dev_euclidean_precision": 0.5626666666666666,
-      "eval_Qnli-dev_euclidean_recall": 0.8940677966101694,
-      "eval_Qnli-dev_manhattan_accuracy": 0.677734375,
-      "eval_Qnli-dev_manhattan_accuracy_threshold": 363.1766052246094,
-      "eval_Qnli-dev_manhattan_ap": 0.7003574987662966,
-      "eval_Qnli-dev_manhattan_f1": 0.6951026856240127,
-      "eval_Qnli-dev_manhattan_f1_threshold": 436.8890075683594,
-      "eval_Qnli-dev_manhattan_precision": 0.5541561712846348,
-      "eval_Qnli-dev_manhattan_recall": 0.9322033898305084,
-      "eval_Qnli-dev_max_accuracy": 0.677734375,
-      "eval_Qnli-dev_max_accuracy_threshold": 615.7371826171875,
-      "eval_Qnli-dev_max_ap": 0.7003574987662966,
-      "eval_Qnli-dev_max_f1": 0.6951026856240127,
-      "eval_Qnli-dev_max_f1_threshold": 558.3238525390625,
-      "eval_Qnli-dev_max_precision": 0.5626666666666666,
-      "eval_Qnli-dev_max_recall": 0.9322033898305084,
-      "eval_allNLI-dev_cosine_accuracy": 0.701171875,
-      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9141938090324402,
-      "eval_allNLI-dev_cosine_ap": 0.5178189449379189,
-      "eval_allNLI-dev_cosine_f1": 0.5469728601252609,
-      "eval_allNLI-dev_cosine_f1_threshold": 0.8089991807937622,
-      "eval_allNLI-dev_cosine_precision": 0.42810457516339867,
-      "eval_allNLI-dev_cosine_recall": 0.7572254335260116,
-      "eval_allNLI-dev_dot_accuracy": 0.701171875,
-      "eval_allNLI-dev_dot_accuracy_threshold": 701.9563598632812,
-      "eval_allNLI-dev_dot_ap": 0.5185120023725823,
-      "eval_allNLI-dev_dot_f1": 0.5492662473794548,
-      "eval_allNLI-dev_dot_f1_threshold": 619.0263671875,
-      "eval_allNLI-dev_dot_precision": 0.4309210526315789,
-      "eval_allNLI-dev_dot_recall": 0.7572254335260116,
-      "eval_allNLI-dev_euclidean_accuracy": 0.701171875,
-      "eval_allNLI-dev_euclidean_accuracy_threshold": 11.46473503112793,
-      "eval_allNLI-dev_euclidean_ap": 0.5178582923548423,
-      "eval_allNLI-dev_euclidean_f1": 0.5458422174840084,
-      "eval_allNLI-dev_euclidean_f1_threshold": 16.86907958984375,
-      "eval_allNLI-dev_euclidean_precision": 0.43243243243243246,
-      "eval_allNLI-dev_euclidean_recall": 0.7398843930635838,
-      "eval_allNLI-dev_manhattan_accuracy": 0.697265625,
-      "eval_allNLI-dev_manhattan_accuracy_threshold": 243.82118225097656,
-      "eval_allNLI-dev_manhattan_ap": 0.5208045922745452,
-      "eval_allNLI-dev_manhattan_f1": 0.5536480686695279,
-      "eval_allNLI-dev_manhattan_f1_threshold": 353.4998474121094,
-      "eval_allNLI-dev_manhattan_precision": 0.4402730375426621,
-      "eval_allNLI-dev_manhattan_recall": 0.7456647398843931,
-      "eval_allNLI-dev_max_accuracy": 0.701171875,
-      "eval_allNLI-dev_max_accuracy_threshold": 701.9563598632812,
-      "eval_allNLI-dev_max_ap": 0.5208045922745452,
-      "eval_allNLI-dev_max_f1": 0.5536480686695279,
-      "eval_allNLI-dev_max_f1_threshold": 619.0263671875,
-      "eval_allNLI-dev_max_precision": 0.4402730375426621,
-      "eval_allNLI-dev_max_recall": 0.7572254335260116,
-      "eval_loss": 0.9706022143363953,
-      "eval_runtime": 54.1103,
-      "eval_samples_per_second": 30.752,
-      "eval_sequential_score": 0.7003574987662966,
-      "eval_steps_per_second": 0.129,
-      "eval_sts-test_pearson_cosine": 0.6620027075959773,
-      "eval_sts-test_pearson_dot": 0.6601968030872636,
-      "eval_sts-test_pearson_euclidean": 0.6760063290032935,
-      "eval_sts-test_pearson_manhattan": 0.6792900086102078,
-      "eval_sts-test_pearson_max": 0.6792900086102078,
-      "eval_sts-test_spearman_cosine": 0.6712350889173807,
-      "eval_sts-test_spearman_dot": 0.6700788953280725,
-      "eval_sts-test_spearman_euclidean": 0.6714677059740932,
-      "eval_sts-test_spearman_manhattan": 0.6752350909143282,
-      "eval_sts-test_spearman_max": 0.6752350909143282,
+      "epoch": 1.9576771653543306,
+      "eval_Qnli-dev_cosine_accuracy": 0.673828125,
+      "eval_Qnli-dev_cosine_accuracy_threshold": 0.7969574928283691,
+      "eval_Qnli-dev_cosine_ap": 0.6978522990819034,
+      "eval_Qnli-dev_cosine_f1": 0.6907216494845361,
+      "eval_Qnli-dev_cosine_f1_threshold": 0.759760856628418,
+      "eval_Qnli-dev_cosine_precision": 0.5809248554913294,
+      "eval_Qnli-dev_cosine_recall": 0.8516949152542372,
+      "eval_Qnli-dev_dot_accuracy": 0.66796875,
+      "eval_Qnli-dev_dot_accuracy_threshold": 604.6656494140625,
+      "eval_Qnli-dev_dot_ap": 0.6961098574484125,
+      "eval_Qnli-dev_dot_f1": 0.6900175131348512,
+      "eval_Qnli-dev_dot_f1_threshold": 584.50830078125,
+      "eval_Qnli-dev_dot_precision": 0.5880597014925373,
+      "eval_Qnli-dev_dot_recall": 0.8347457627118644,
+      "eval_Qnli-dev_euclidean_accuracy": 0.673828125,
+      "eval_Qnli-dev_euclidean_accuracy_threshold": 17.583419799804688,
+      "eval_Qnli-dev_euclidean_ap": 0.6979529403637269,
+      "eval_Qnli-dev_euclidean_f1": 0.6917808219178082,
+      "eval_Qnli-dev_euclidean_f1_threshold": 19.166725158691406,
+      "eval_Qnli-dev_euclidean_precision": 0.5804597701149425,
+      "eval_Qnli-dev_euclidean_recall": 0.8559322033898306,
+      "eval_Qnli-dev_manhattan_accuracy": 0.6796875,
+      "eval_Qnli-dev_manhattan_accuracy_threshold": 363.2380065917969,
+      "eval_Qnli-dev_manhattan_ap": 0.7071976338822723,
+      "eval_Qnli-dev_manhattan_f1": 0.6950092421441774,
+      "eval_Qnli-dev_manhattan_f1_threshold": 384.6904296875,
+      "eval_Qnli-dev_manhattan_precision": 0.6163934426229508,
+      "eval_Qnli-dev_manhattan_recall": 0.7966101694915254,
+      "eval_Qnli-dev_max_accuracy": 0.6796875,
+      "eval_Qnli-dev_max_accuracy_threshold": 604.6656494140625,
+      "eval_Qnli-dev_max_ap": 0.7071976338822723,
+      "eval_Qnli-dev_max_f1": 0.6950092421441774,
+      "eval_Qnli-dev_max_f1_threshold": 584.50830078125,
+      "eval_Qnli-dev_max_precision": 0.6163934426229508,
+      "eval_Qnli-dev_max_recall": 0.8559322033898306,
+      "eval_allNLI-dev_cosine_accuracy": 0.693359375,
+      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9265936613082886,
+      "eval_allNLI-dev_cosine_ap": 0.5135607855999773,
+      "eval_allNLI-dev_cosine_f1": 0.5537848605577689,
+      "eval_allNLI-dev_cosine_f1_threshold": 0.8162949085235596,
+      "eval_allNLI-dev_cosine_precision": 0.42249240121580545,
+      "eval_allNLI-dev_cosine_recall": 0.8034682080924855,
+      "eval_allNLI-dev_dot_accuracy": 0.693359375,
+      "eval_allNLI-dev_dot_accuracy_threshold": 708.295166015625,
+      "eval_allNLI-dev_dot_ap": 0.5122623119148267,
+      "eval_allNLI-dev_dot_f1": 0.5542168674698795,
+      "eval_allNLI-dev_dot_f1_threshold": 624.698974609375,
+      "eval_allNLI-dev_dot_precision": 0.4246153846153846,
+      "eval_allNLI-dev_dot_recall": 0.7976878612716763,
+      "eval_allNLI-dev_euclidean_accuracy": 0.693359375,
+      "eval_allNLI-dev_euclidean_accuracy_threshold": 10.592409133911133,
+      "eval_allNLI-dev_euclidean_ap": 0.5133305497190137,
+      "eval_allNLI-dev_euclidean_f1": 0.5537848605577689,
+      "eval_allNLI-dev_euclidean_f1_threshold": 16.733966827392578,
+      "eval_allNLI-dev_euclidean_precision": 0.42249240121580545,
+      "eval_allNLI-dev_euclidean_recall": 0.8034682080924855,
+      "eval_allNLI-dev_manhattan_accuracy": 0.689453125,
+      "eval_allNLI-dev_manhattan_accuracy_threshold": 221.28228759765625,
+      "eval_allNLI-dev_manhattan_ap": 0.5111987819682724,
+      "eval_allNLI-dev_manhattan_f1": 0.5553235908141962,
+      "eval_allNLI-dev_manhattan_f1_threshold": 342.06512451171875,
+      "eval_allNLI-dev_manhattan_precision": 0.434640522875817,
+      "eval_allNLI-dev_manhattan_recall": 0.7687861271676301,
+      "eval_allNLI-dev_max_accuracy": 0.693359375,
+      "eval_allNLI-dev_max_accuracy_threshold": 708.295166015625,
+      "eval_allNLI-dev_max_ap": 0.5135607855999773,
+      "eval_allNLI-dev_max_f1": 0.5553235908141962,
+      "eval_allNLI-dev_max_f1_threshold": 624.698974609375,
+      "eval_allNLI-dev_max_precision": 0.434640522875817,
+      "eval_allNLI-dev_max_recall": 0.8034682080924855,
+      "eval_loss": 0.8407362103462219,
+      "eval_runtime": 52.229,
+      "eval_samples_per_second": 31.86,
+      "eval_sequential_score": 0.7071976338822723,
+      "eval_steps_per_second": 0.134,
+      "eval_sts-test_pearson_cosine": 0.6486131885836244,
+      "eval_sts-test_pearson_dot": 0.6455690772643841,
+      "eval_sts-test_pearson_euclidean": 0.6614666062012153,
+      "eval_sts-test_pearson_manhattan": 0.6665880327992885,
+      "eval_sts-test_pearson_max": 0.6665880327992885,
+      "eval_sts-test_spearman_cosine": 0.6575633100547399,
+      "eval_sts-test_spearman_dot": 0.6538646630520021,
+      "eval_sts-test_spearman_euclidean": 0.6579358195545618,
+      "eval_sts-test_spearman_manhattan": 0.6619746656463971,
+      "eval_sts-test_spearman_max": 0.6619746656463971,
       "step": 1989
     },
     {
-      "epoch": 1.9576771653543306,
-      "grad_norm": 10.558905601501465,
+      "epoch": 1.9586614173228347,
+      "grad_norm": 12.40316104888916,
       "learning_rate": 5.338448190126184e-07,
-      "loss": 0.5158,
+      "loss": 0.9368,
       "step": 1990
     },
     {
-      "epoch": 1.9586614173228347,
-      "grad_norm": 16.4892520904541,
+      "epoch": 1.9596456692913384,
+      "grad_norm": 7.745510578155518,
       "learning_rate": 5.330797013400614e-07,
-      "loss": 1.1472,
+      "loss": 0.2367,
       "step": 1991
     },
     {
-      "epoch": 1.9596456692913384,
-      "grad_norm": 13.120809555053711,
+      "epoch": 1.9606299212598426,
+      "grad_norm": 14.807564735412598,
       "learning_rate": 5.323145093145763e-07,
-      "loss": 0.5507,
+      "loss": 0.7361,
       "step": 1992
     },
     {
-      "epoch": 1.9606299212598426,
-      "grad_norm": 13.758267402648926,
+      "epoch": 1.9616141732283463,
+      "grad_norm": 11.049567222595215,
       "learning_rate": 5.3154924474733e-07,
-      "loss": 0.4904,
+      "loss": 0.5095,
       "step": 1993
     },
     {
-      "epoch": 1.9616141732283463,
-      "grad_norm": 10.835561752319336,
+      "epoch": 1.9625984251968505,
+      "grad_norm": 7.133857727050781,
       "learning_rate": 5.307839094496621e-07,
-      "loss": 0.2959,
+      "loss": 0.3207,
       "step": 1994
     },
     {
-      "epoch": 1.9625984251968505,
-      "grad_norm": 7.2871222496032715,
+      "epoch": 1.9635826771653542,
+      "grad_norm": 10.536307334899902,
       "learning_rate": 5.300185052330793e-07,
-      "loss": 0.1981,
+      "loss": 0.4538,
       "step": 1995
     },
     {
-      "epoch": 1.9635826771653542,
-      "grad_norm": 9.152851104736328,
+      "epoch": 1.9645669291338583,
+      "grad_norm": 8.809112548828125,
       "learning_rate": 5.292530339092513e-07,
-      "loss": 0.332,
+      "loss": 0.2839,
       "step": 1996
     },
     {
-      "epoch": 1.9645669291338583,
-      "grad_norm": 10.531471252441406,
+      "epoch": 1.965551181102362,
+      "grad_norm": 8.962004661560059,
       "learning_rate": 5.284874972900064e-07,
-      "loss": 0.3562,
+      "loss": 0.3786,
       "step": 1997
     },
     {
-      "epoch": 1.965551181102362,
-      "grad_norm": 21.480953216552734,
+      "epoch": 1.9665354330708662,
+      "grad_norm": 14.268336296081543,
       "learning_rate": 5.277218971873281e-07,
-      "loss": 0.7395,
+      "loss": 0.7311,
       "step": 1998
     },
     {
-      "epoch": 1.9665354330708662,
-      "grad_norm": 10.496049880981445,
+      "epoch": 1.96751968503937,
+      "grad_norm": 6.8707756996154785,
       "learning_rate": 5.269562354133496e-07,
-      "loss": 0.5546,
+      "loss": 0.199,
       "step": 1999
     },
     {
-      "epoch": 1.96751968503937,
-      "grad_norm": 12.824313163757324,
+      "epoch": 1.968503937007874,
+      "grad_norm": 12.054238319396973,
       "learning_rate": 5.261905137803504e-07,
-      "loss": 0.6003,
+      "loss": 0.5929,
       "step": 2000
     },
     {
-      "epoch": 1.968503937007874,
-      "grad_norm": 14.431049346923828,
+      "epoch": 1.969488188976378,
+      "grad_norm": 11.269950866699219,
       "learning_rate": 5.254247341007514e-07,
-      "loss": 0.784,
+      "loss": 0.4542,
       "step": 2001
     },
     {
-      "epoch": 1.969488188976378,
-      "grad_norm": 11.358637809753418,
+      "epoch": 1.970472440944882,
+      "grad_norm": 8.575435638427734,
       "learning_rate": 5.246588981871113e-07,
-      "loss": 0.5101,
+      "loss": 0.4191,
       "step": 2002
     },
     {
-      "epoch": 1.970472440944882,
-      "grad_norm": 10.88247013092041,
+      "epoch": 1.9714566929133859,
+      "grad_norm": 10.653684616088867,
       "learning_rate": 5.238930078521215e-07,
-      "loss": 0.5691,
+      "loss": 0.4298,
       "step": 2003
     },
     {
-      "epoch": 1.9714566929133859,
-      "grad_norm": 6.141389846801758,
+      "epoch": 1.9724409448818898,
+      "grad_norm": 8.076311111450195,
       "learning_rate": 5.231270649086023e-07,
-      "loss": 0.2075,
+      "loss": 0.3283,
       "step": 2004
     },
     {
-      "epoch": 1.9724409448818898,
-      "grad_norm": 11.258003234863281,
+      "epoch": 1.9734251968503937,
+      "grad_norm": 13.3706636428833,
       "learning_rate": 5.22361071169499e-07,
-      "loss": 0.4202,
+      "loss": 0.9176,
       "step": 2005
     },
     {
-      "epoch": 1.9734251968503937,
-      "grad_norm": 9.83862590789795,
+      "epoch": 1.9744094488188977,
+      "grad_norm": 11.212428092956543,
       "learning_rate": 5.215950284478763e-07,
-      "loss": 0.3815,
+      "loss": 0.6424,
       "step": 2006
     },
     {
-      "epoch": 1.9744094488188977,
-      "grad_norm": 6.295627593994141,
+      "epoch": 1.9753937007874016,
+      "grad_norm": 10.307910919189453,
       "learning_rate": 5.208289385569154e-07,
-      "loss": 0.2197,
+      "loss": 0.4727,
       "step": 2007
     },
     {
-      "epoch": 1.9753937007874016,
-      "grad_norm": 8.904109954833984,
+      "epoch": 1.9763779527559056,
+      "grad_norm": 9.23723030090332,
       "learning_rate": 5.20062803309909e-07,
-      "loss": 0.3602,
+      "loss": 0.3531,
       "step": 2008
     },
     {
-      "epoch": 1.9763779527559056,
-      "grad_norm": 10.298103332519531,
+      "epoch": 1.9773622047244095,
+      "grad_norm": 8.683030128479004,
       "learning_rate": 5.192966245202571e-07,
-      "loss": 0.295,
+      "loss": 0.3506,
       "step": 2009
     },
     {
-      "epoch": 1.9773622047244095,
-      "grad_norm": 9.491402626037598,
+      "epoch": 1.9783464566929134,
+      "grad_norm": 10.536169052124023,
       "learning_rate": 5.18530404001463e-07,
-      "loss": 0.3633,
+      "loss": 0.5331,
       "step": 2010
     },
     {
-      "epoch": 1.9783464566929134,
-      "grad_norm": 9.721626281738281,
+      "epoch": 1.9793307086614174,
+      "grad_norm": 13.709545135498047,
       "learning_rate": 5.177641435671284e-07,
-      "loss": 0.4982,
+      "loss": 0.7927,
       "step": 2011
     },
     {
-      "epoch": 1.9793307086614174,
-      "grad_norm": 16.09177017211914,
+      "epoch": 1.9803149606299213,
+      "grad_norm": 12.91935920715332,
       "learning_rate": 5.169978450309498e-07,
-      "loss": 0.8649,
+      "loss": 0.6007,
       "step": 2012
     },
     {
-      "epoch": 1.9803149606299213,
-      "grad_norm": 14.924250602722168,
+      "epoch": 1.9812992125984252,
+      "grad_norm": 10.19194221496582,
       "learning_rate": 5.162315102067139e-07,
-      "loss": 0.8593,
+      "loss": 0.4699,
       "step": 2013
     },
     {
-      "epoch": 1.9812992125984252,
-      "grad_norm": 11.977181434631348,
+      "epoch": 1.9822834645669292,
+      "grad_norm": 12.63509464263916,
       "learning_rate": 5.154651409082927e-07,
-      "loss": 0.5689,
+      "loss": 0.757,
       "step": 2014
     },
     {
-      "epoch": 1.9822834645669292,
-      "grad_norm": 20.18109893798828,
+      "epoch": 1.983267716535433,
+      "grad_norm": 9.417659759521484,
       "learning_rate": 5.146987389496408e-07,
-      "loss": 0.9294,
+      "loss": 0.3937,
       "step": 2015
     },
     {
-      "epoch": 1.983267716535433,
-      "grad_norm": 9.459802627563477,
+      "epoch": 1.984251968503937,
+      "grad_norm": 11.925963401794434,
       "learning_rate": 5.139323061447892e-07,
-      "loss": 0.3826,
+      "loss": 0.9555,
       "step": 2016
     },
     {
-      "epoch": 1.984251968503937,
-      "grad_norm": 8.496149063110352,
+      "epoch": 1.985236220472441,
+      "grad_norm": 7.745391368865967,
       "learning_rate": 5.131658443078426e-07,
-      "loss": 0.3136,
+      "loss": 0.3785,
       "step": 2017
     },
     {
-      "epoch": 1.985236220472441,
-      "grad_norm": 11.517538070678711,
+      "epoch": 1.986220472440945,
+      "grad_norm": 12.12795639038086,
       "learning_rate": 5.123993552529736e-07,
-      "loss": 0.6924,
+      "loss": 0.5992,
       "step": 2018
     },
     {
-      "epoch": 1.986220472440945,
-      "grad_norm": 11.961481094360352,
+      "epoch": 1.9872047244094488,
+      "grad_norm": 9.798077583312988,
       "learning_rate": 5.116328407944201e-07,
-      "loss": 0.6278,
+      "loss": 0.2856,
       "step": 2019
     },
     {
-      "epoch": 1.9872047244094488,
-      "grad_norm": 11.670297622680664,
+      "epoch": 1.9881889763779528,
+      "grad_norm": 12.854366302490234,
       "learning_rate": 5.108663027464795e-07,
-      "loss": 0.615,
+      "loss": 0.8852,
       "step": 2020
     },
     {
-      "epoch": 1.9881889763779528,
-      "grad_norm": 11.576616287231445,
+      "epoch": 1.9891732283464567,
+      "grad_norm": 7.777148723602295,
       "learning_rate": 5.100997429235055e-07,
-      "loss": 0.5819,
+      "loss": 0.2864,
       "step": 2021
     },
     {
-      "epoch": 1.9891732283464567,
-      "grad_norm": 9.68857192993164,
+      "epoch": 1.9901574803149606,
+      "grad_norm": 12.080920219421387,
       "learning_rate": 5.093331631399029e-07,
-      "loss": 0.5012,
+      "loss": 0.4794,
       "step": 2022
     },
     {
-      "epoch": 1.9901574803149606,
-      "grad_norm": 11.992721557617188,
+      "epoch": 1.9911417322834646,
+      "grad_norm": 10.177239418029785,
       "learning_rate": 5.085665652101244e-07,
-      "loss": 0.4416,
+      "loss": 0.5617,
       "step": 2023
     },
     {
-      "epoch": 1.9911417322834646,
-      "grad_norm": 9.953713417053223,
+      "epoch": 1.9921259842519685,
+      "grad_norm": 11.578356742858887,
       "learning_rate": 5.077999509486645e-07,
-      "loss": 0.4151,
+      "loss": 0.5764,
       "step": 2024
     },
     {
-      "epoch": 1.9921259842519685,
-      "grad_norm": 17.759462356567383,
+      "epoch": 1.9931102362204725,
+      "grad_norm": 12.756275177001953,
       "learning_rate": 5.070333221700575e-07,
-      "loss": 0.6352,
+      "loss": 0.5486,
       "step": 2025
     },
     {
-      "epoch": 1.9931102362204725,
-      "grad_norm": 4.798762321472168,
+      "epoch": 1.9940944881889764,
+      "grad_norm": 9.579181671142578,
       "learning_rate": 5.062666806888715e-07,
-      "loss": 0.1386,
+      "loss": 0.4615,
       "step": 2026
     },
     {
-      "epoch": 1.9940944881889764,
-      "grad_norm": 9.868159294128418,
+      "epoch": 1.9950787401574803,
+      "grad_norm": 10.681057929992676,
       "learning_rate": 5.055000283197051e-07,
-      "loss": 0.4451,
+      "loss": 0.4842,
       "step": 2027
     },
     {
-      "epoch": 1.9950787401574803,
-      "grad_norm": 6.54140567779541,
+      "epoch": 1.9960629921259843,
+      "grad_norm": 11.068312644958496,
       "learning_rate": 5.047333668771821e-07,
-      "loss": 0.2217,
+      "loss": 0.4858,
       "step": 2028
     },
     {
-      "epoch": 1.9960629921259843,
-      "grad_norm": 9.168428421020508,
+      "epoch": 1.9970472440944882,
+      "grad_norm": 7.881600379943848,
       "learning_rate": 5.03966698175948e-07,
-      "loss": 0.3559,
+      "loss": 0.3759,
       "step": 2029
     },
     {
-      "epoch": 1.9970472440944882,
-      "grad_norm": 6.069472312927246,
+      "epoch": 1.9980314960629921,
+      "grad_norm": 12.428053855895996,
       "learning_rate": 5.032000240306655e-07,
-      "loss": 0.3077,
+      "loss": 0.8044,
       "step": 2030
     },
     {
-      "epoch": 1.9980314960629921,
-      "grad_norm": 18.002717971801758,
+      "epoch": 1.999015748031496,
+      "grad_norm": 10.837055206298828,
       "learning_rate": 5.024333462560104e-07,
-      "loss": 1.0025,
+      "loss": 0.4664,
       "step": 2031
     },
     {
-      "epoch": 1.999015748031496,
-      "grad_norm": 15.968050956726074,
+      "epoch": 2.0,
+      "grad_norm": 13.234967231750488,
       "learning_rate": 5.016666666666668e-07,
-      "loss": 0.9233,
+      "loss": 0.5267,
       "step": 2032
     },
     {
-      "epoch": 2.0,
-      "grad_norm": 21.219402313232422,
+      "epoch": 2.000984251968504,
+      "grad_norm": 11.927884101867676,
       "learning_rate": 5.008999870773232e-07,
-      "loss": 0.52,
+      "loss": 0.7186,
       "step": 2033
     },
     {
-      "epoch": 2.0009842519685037,
-      "grad_norm": 0.0,
+      "epoch": 2.001968503937008,
+      "grad_norm": 10.080817222595215,
       "learning_rate": 5.001333093026681e-07,
-      "loss": 0.0,
+      "loss": 0.5132,
       "step": 2034
     },
     {
-      "epoch": 2.000984251968504,
-      "grad_norm": 19.841766357421875,
+      "epoch": 2.002952755905512,
+      "grad_norm": 8.70809268951416,
       "learning_rate": 4.993666351573856e-07,
-      "loss": 1.0536,
+      "loss": 0.3771,
       "step": 2035
     },
     {
-      "epoch": 2.001968503937008,
-      "grad_norm": 12.783218383789062,
+      "epoch": 2.0039370078740157,
+      "grad_norm": 8.891264915466309,
       "learning_rate": 4.985999664561514e-07,
-      "loss": 0.5749,
+      "loss": 0.3564,
       "step": 2036
     },
     {
-      "epoch": 2.002952755905512,
-      "grad_norm": 21.010698318481445,
+      "epoch": 2.00492125984252,
+      "grad_norm": 5.313558578491211,
       "learning_rate": 4.978333050136285e-07,
-      "loss": 0.9689,
+      "loss": 0.199,
       "step": 2037
     },
     {
-      "epoch": 2.0039370078740157,
-      "grad_norm": 10.027554512023926,
+      "epoch": 2.0059055118110236,
+      "grad_norm": 7.508504390716553,
       "learning_rate": 4.970666526444619e-07,
-      "loss": 0.4037,
+      "loss": 0.2808,
       "step": 2038
     },
     {
-      "epoch": 2.00492125984252,
-      "grad_norm": 14.6770658493042,
+      "epoch": 2.0068897637795278,
+      "grad_norm": 11.430014610290527,
       "learning_rate": 4.96300011163276e-07,
-      "loss": 0.6314,
+      "loss": 0.5424,
       "step": 2039
     },
     {
-      "epoch": 2.0059055118110236,
-      "grad_norm": 6.943289279937744,
+      "epoch": 2.0078740157480315,
+      "grad_norm": 9.434273719787598,
       "learning_rate": 4.955333823846691e-07,
-      "loss": 0.1926,
+      "loss": 0.4058,
       "step": 2040
     },
     {
-      "epoch": 2.0068897637795278,
-      "grad_norm": 13.570602416992188,
+      "epoch": 2.0088582677165356,
+      "grad_norm": 6.240298748016357,
       "learning_rate": 4.947667681232094e-07,
-      "loss": 0.8417,
+      "loss": 0.2607,
       "step": 2041
     },
     {
-      "epoch": 2.0078740157480315,
-      "grad_norm": 12.345090866088867,
+      "epoch": 2.0098425196850394,
+      "grad_norm": 11.556763648986816,
       "learning_rate": 4.940001701934305e-07,
-      "loss": 0.6205,
+      "loss": 0.592,
       "step": 2042
     },
     {
-      "epoch": 2.0088582677165356,
-      "grad_norm": 9.181805610656738,
+      "epoch": 2.0108267716535435,
+      "grad_norm": 13.08253002166748,
       "learning_rate": 4.932335904098282e-07,
-      "loss": 0.2801,
+      "loss": 0.5805,
       "step": 2043
     },
     {
-      "epoch": 2.0098425196850394,
-      "grad_norm": 17.511737823486328,
+      "epoch": 2.0118110236220472,
+      "grad_norm": 8.079017639160156,
       "learning_rate": 4.924670305868541e-07,
-      "loss": 0.6844,
+      "loss": 0.253,
       "step": 2044
     },
     {
-      "epoch": 2.0108267716535435,
-      "grad_norm": 12.344738960266113,
+      "epoch": 2.0127952755905514,
+      "grad_norm": 6.124202251434326,
       "learning_rate": 4.917004925389137e-07,
-      "loss": 0.6534,
+      "loss": 0.252,
       "step": 2045
     },
     {
-      "epoch": 2.0118110236220472,
-      "grad_norm": 14.163546562194824,
+      "epoch": 2.013779527559055,
+      "grad_norm": 6.105528831481934,
       "learning_rate": 4.909339780803601e-07,
-      "loss": 0.5133,
+      "loss": 0.2473,
       "step": 2046
     },
     {
-      "epoch": 2.0127952755905514,
-      "grad_norm": 10.767459869384766,
+      "epoch": 2.0147637795275593,
+      "grad_norm": 10.853846549987793,
       "learning_rate": 4.901674890254911e-07,
-      "loss": 0.4383,
+      "loss": 0.386,
       "step": 2047
     },
     {
-      "epoch": 2.013779527559055,
-      "grad_norm": 14.008377075195312,
+      "epoch": 2.015748031496063,
+      "grad_norm": 11.963359832763672,
       "learning_rate": 4.894010271885444e-07,
-      "loss": 0.7045,
+      "loss": 0.4424,
       "step": 2048
     },
     {
-      "epoch": 2.0147637795275593,
-      "grad_norm": 11.19058895111084,
+      "epoch": 2.016732283464567,
+      "grad_norm": 8.942184448242188,
       "learning_rate": 4.886345943836928e-07,
-      "loss": 0.4953,
+      "loss": 0.5241,
       "step": 2049
     },
     {
-      "epoch": 2.015748031496063,
-      "grad_norm": 6.641456127166748,
+      "epoch": 2.017716535433071,
+      "grad_norm": 12.437562942504883,
       "learning_rate": 4.878681924250408e-07,
-      "loss": 0.2567,
+      "loss": 0.5726,
       "step": 2050
     },
     {
-      "epoch": 2.016732283464567,
-      "grad_norm": 8.315643310546875,
+      "epoch": 2.018700787401575,
+      "grad_norm": 13.252161979675293,
       "learning_rate": 4.871018231266197e-07,
-      "loss": 0.4574,
+      "loss": 0.8175,
       "step": 2051
     },
     {
-      "epoch": 2.017716535433071,
-      "grad_norm": 9.133217811584473,
+      "epoch": 2.0196850393700787,
+      "grad_norm": 15.371834754943848,
       "learning_rate": 4.863354883023838e-07,
-      "loss": 0.3211,
+      "loss": 1.3487,
       "step": 2052
     },
     {
-      "epoch": 2.018700787401575,
-      "grad_norm": 15.33566951751709,
+      "epoch": 2.020669291338583,
+      "grad_norm": 14.541483879089355,
       "learning_rate": 4.85569189766205e-07,
-      "loss": 0.7028,
+      "loss": 0.5315,
       "step": 2053
     },
     {
-      "epoch": 2.0196850393700787,
-      "grad_norm": 11.80440616607666,
+      "epoch": 2.0216535433070866,
+      "grad_norm": 9.403450012207031,
       "learning_rate": 4.848029293318705e-07,
-      "loss": 0.5107,
+      "loss": 0.2093,
       "step": 2054
     },
     {
-      "epoch": 2.020669291338583,
-      "grad_norm": 11.949298858642578,
+      "epoch": 2.0226377952755907,
+      "grad_norm": 7.741646766662598,
       "learning_rate": 4.840367088130764e-07,
-      "loss": 0.327,
+      "loss": 0.3055,
       "step": 2055
     },
     {
-      "epoch": 2.0216535433070866,
-      "grad_norm": 14.780426979064941,
+      "epoch": 2.0236220472440944,
+      "grad_norm": 10.87693977355957,
       "learning_rate": 4.832705300234245e-07,
-      "loss": 0.7758,
+      "loss": 0.6388,
       "step": 2056
     },
     {
-      "epoch": 2.0226377952755907,
-      "grad_norm": 8.046977996826172,
+      "epoch": 2.0246062992125986,
+      "grad_norm": 12.780906677246094,
       "learning_rate": 4.825043947764181e-07,
-      "loss": 0.2169,
+      "loss": 0.5827,
       "step": 2057
     },
     {
-      "epoch": 2.0236220472440944,
-      "grad_norm": 4.532480716705322,
+      "epoch": 2.0255905511811023,
+      "grad_norm": 11.055867195129395,
       "learning_rate": 4.817383048854572e-07,
-      "loss": 0.0995,
+      "loss": 0.4087,
       "step": 2058
     },
     {
-      "epoch": 2.0246062992125986,
-      "grad_norm": 10.037430763244629,
+      "epoch": 2.0265748031496065,
+      "grad_norm": 15.772777557373047,
       "learning_rate": 4.809722621638345e-07,
-      "loss": 0.4255,
+      "loss": 0.7724,
       "step": 2059
     },
     {
-      "epoch": 2.0255905511811023,
-      "grad_norm": 14.573445320129395,
+      "epoch": 2.02755905511811,
+      "grad_norm": 7.676803112030029,
       "learning_rate": 4.802062684247311e-07,
-      "loss": 0.6974,
+      "loss": 0.2678,
       "step": 2060
     },
     {
-      "epoch": 2.0265748031496065,
-      "grad_norm": 9.875505447387695,
+      "epoch": 2.0285433070866143,
+      "grad_norm": 12.137404441833496,
       "learning_rate": 4.79440325481212e-07,
-      "loss": 0.3386,
+      "loss": 0.5643,
       "step": 2061
     },
     {
-      "epoch": 2.02755905511811,
-      "grad_norm": 11.617470741271973,
+      "epoch": 2.029527559055118,
+      "grad_norm": 10.45738697052002,
       "learning_rate": 4.786744351462222e-07,
-      "loss": 0.4871,
+      "loss": 0.3502,
       "step": 2062
     },
     {
-      "epoch": 2.0285433070866143,
-      "grad_norm": 15.038314819335938,
+      "epoch": 2.030511811023622,
+      "grad_norm": 8.989094734191895,
       "learning_rate": 4.779085992325821e-07,
-      "loss": 0.7656,
+      "loss": 0.2656,
       "step": 2063
     },
     {
-      "epoch": 2.029527559055118,
-      "grad_norm": 10.356962203979492,
+      "epoch": 2.031496062992126,
+      "grad_norm": 13.004704475402832,
       "learning_rate": 4.771428195529833e-07,
-      "loss": 0.3288,
+      "loss": 0.8536,
       "step": 2064
     },
     {
-      "epoch": 2.030511811023622,
-      "grad_norm": 7.781601905822754,
+      "epoch": 2.03248031496063,
+      "grad_norm": 8.035222053527832,
       "learning_rate": 4.763770979199841e-07,
-      "loss": 0.3246,
+      "loss": 0.278,
       "step": 2065
     },
     {
-      "epoch": 2.031496062992126,
-      "grad_norm": 10.731339454650879,
+      "epoch": 2.033464566929134,
+      "grad_norm": 10.170429229736328,
       "learning_rate": 4.756114361460057e-07,
-      "loss": 0.5359,
+      "loss": 0.5795,
       "step": 2066
     },
     {
-      "epoch": 2.03248031496063,
-      "grad_norm": 13.668349266052246,
+      "epoch": 2.034448818897638,
+      "grad_norm": 6.464893817901611,
       "learning_rate": 4.748458360433273e-07,
-      "loss": 0.5731,
+      "loss": 0.228,
       "step": 2067
     },
     {
-      "epoch": 2.033464566929134,
-      "grad_norm": 12.049600601196289,
+      "epoch": 2.0354330708661417,
+      "grad_norm": 8.70262622833252,
       "learning_rate": 4.7408029942408244e-07,
-      "loss": 0.6563,
+      "loss": 0.3636,
       "step": 2068
     },
     {
-      "epoch": 2.034448818897638,
-      "grad_norm": 10.32591724395752,
+      "epoch": 2.036417322834646,
+      "grad_norm": 10.14581298828125,
       "learning_rate": 4.733148281002543e-07,
-      "loss": 0.434,
+      "loss": 0.4357,
       "step": 2069
     },
     {
-      "epoch": 2.0354330708661417,
-      "grad_norm": 14.998777389526367,
+      "epoch": 2.0374015748031495,
+      "grad_norm": 12.38016414642334,
       "learning_rate": 4.7254942388367147e-07,
-      "loss": 0.8234,
+      "loss": 0.8257,
       "step": 2070
     },
     {
-      "epoch": 2.036417322834646,
-      "grad_norm": 12.361584663391113,
+      "epoch": 2.0383858267716537,
+      "grad_norm": 5.428321838378906,
       "learning_rate": 4.717840885860036e-07,
-      "loss": 0.7725,
+      "loss": 0.1714,
       "step": 2071
     },
     {
-      "epoch": 2.0374015748031495,
-      "grad_norm": 9.835253715515137,
+      "epoch": 2.0393700787401574,
+      "grad_norm": 11.890104293823242,
       "learning_rate": 4.7101882401875745e-07,
-      "loss": 0.5631,
+      "loss": 0.7014,
       "step": 2072
     },
     {
-      "epoch": 2.0383858267716537,
-      "grad_norm": 7.7654709815979,
+      "epoch": 2.0403543307086616,
+      "grad_norm": 6.636118412017822,
       "learning_rate": 4.7025363199327213e-07,
-      "loss": 0.3587,
+      "loss": 0.264,
       "step": 2073
     },
     {
-      "epoch": 2.0393700787401574,
-      "grad_norm": 11.994633674621582,
+      "epoch": 2.0413385826771653,
+      "grad_norm": 9.366705894470215,
       "learning_rate": 4.694885143207153e-07,
-      "loss": 0.6894,
+      "loss": 0.3574,
       "step": 2074
     },
     {
-      "epoch": 2.0403543307086616,
-      "grad_norm": 15.344325065612793,
+      "epoch": 2.0423228346456694,
+      "grad_norm": 9.188568115234375,
       "learning_rate": 4.687234728120783e-07,
-      "loss": 0.8536,
+      "loss": 0.2038,
       "step": 2075
     },
     {
-      "epoch": 2.0413385826771653,
-      "grad_norm": 6.544132232666016,
+      "epoch": 2.043307086614173,
+      "grad_norm": 16.774993896484375,
       "learning_rate": 4.6795850927817256e-07,
-      "loss": 0.2449,
+      "loss": 1.1227,
       "step": 2076
     },
     {
-      "epoch": 2.0423228346456694,
-      "grad_norm": 12.171512603759766,
+      "epoch": 2.0442913385826773,
+      "grad_norm": 11.76972484588623,
       "learning_rate": 4.6719362552962457e-07,
-      "loss": 0.5641,
+      "loss": 0.6045,
       "step": 2077
     },
     {
-      "epoch": 2.043307086614173,
-      "grad_norm": 9.727547645568848,
+      "epoch": 2.045275590551181,
+      "grad_norm": 5.630438804626465,
       "learning_rate": 4.664288233768724e-07,
-      "loss": 0.4836,
+      "loss": 0.1255,
       "step": 2078
     },
     {
-      "epoch": 2.0442913385826773,
-      "grad_norm": 17.973478317260742,
+      "epoch": 2.046259842519685,
+      "grad_norm": 8.554719924926758,
       "learning_rate": 4.6566410463016044e-07,
-      "loss": 0.905,
+      "loss": 0.2404,
       "step": 2079
     },
     {
-      "epoch": 2.045275590551181,
-      "grad_norm": 10.42161750793457,
+      "epoch": 2.047244094488189,
+      "grad_norm": 16.28089714050293,
       "learning_rate": 4.648994710995363e-07,
-      "loss": 0.4315,
+      "loss": 0.9607,
       "step": 2080
     },
     {
-      "epoch": 2.046259842519685,
-      "grad_norm": 10.346402168273926,
+      "epoch": 2.048228346456693,
+      "grad_norm": 7.879763603210449,
       "learning_rate": 4.641349245948454e-07,
-      "loss": 0.3829,
+      "loss": 0.2796,
       "step": 2081
     },
     {
-      "epoch": 2.047244094488189,
-      "grad_norm": 8.672287940979004,
+      "epoch": 2.0492125984251968,
+      "grad_norm": 9.846636772155762,
       "learning_rate": 4.633704669257273e-07,
-      "loss": 0.366,
+      "loss": 0.307,
       "step": 2082
     },
     {
-      "epoch": 2.048228346456693,
-      "grad_norm": 8.586081504821777,
+      "epoch": 2.050196850393701,
+      "grad_norm": 14.373556137084961,
       "learning_rate": 4.626060999016115e-07,
-      "loss": 0.3392,
+      "loss": 0.6566,
       "step": 2083
     },
     {
-      "epoch": 2.0492125984251968,
-      "grad_norm": 11.210050582885742,
+      "epoch": 2.0511811023622046,
+      "grad_norm": 12.32565689086914,
       "learning_rate": 4.6184182533171257e-07,
-      "loss": 0.3751,
+      "loss": 0.5244,
       "step": 2084
     },
     {
-      "epoch": 2.050196850393701,
-      "grad_norm": 7.204981803894043,
+      "epoch": 2.052165354330709,
+      "grad_norm": 11.996628761291504,
       "learning_rate": 4.6107764502502684e-07,
-      "loss": 0.2075,
+      "loss": 0.5113,
       "step": 2085
     },
     {
-      "epoch": 2.0511811023622046,
-      "grad_norm": 15.422869682312012,
+      "epoch": 2.0531496062992125,
+      "grad_norm": 6.542557716369629,
       "learning_rate": 4.603135607903266e-07,
-      "loss": 0.8522,
+      "loss": 0.18,
       "step": 2086
     },
     {
-      "epoch": 2.052165354330709,
-      "grad_norm": 9.124427795410156,
+      "epoch": 2.0541338582677167,
+      "grad_norm": 8.335312843322754,
       "learning_rate": 4.5954957443615765e-07,
-      "loss": 0.4735,
+      "loss": 0.2831,
       "step": 2087
     },
     {
-      "epoch": 2.0531496062992125,
-      "grad_norm": 9.447388648986816,
+      "epoch": 2.0551181102362204,
+      "grad_norm": 9.850923538208008,
       "learning_rate": 4.5878568777083356e-07,
-      "loss": 0.4227,
+      "loss": 0.4851,
       "step": 2088
     },
     {
-      "epoch": 2.0541338582677167,
-      "grad_norm": 12.309717178344727,
+      "epoch": 2.0561023622047245,
+      "grad_norm": 11.451051712036133,
       "learning_rate": 4.5802190260243226e-07,
-      "loss": 0.5437,
+      "loss": 0.4294,
       "step": 2089
     },
     {
-      "epoch": 2.0551181102362204,
-      "grad_norm": 9.313911437988281,
+      "epoch": 2.0570866141732282,
+      "grad_norm": 8.916075706481934,
       "learning_rate": 4.572582207387912e-07,
-      "loss": 0.33,
+      "loss": 0.3259,
       "step": 2090
     },
     {
-      "epoch": 2.0561023622047245,
-      "grad_norm": 11.166501998901367,
+      "epoch": 2.0580708661417324,
+      "grad_norm": 12.524399757385254,
       "learning_rate": 4.5649464398750356e-07,
-      "loss": 0.3904,
+      "loss": 0.6698,
       "step": 2091
     },
     {
-      "epoch": 2.0570866141732282,
-      "grad_norm": 8.26967716217041,
+      "epoch": 2.059055118110236,
+      "grad_norm": 13.372047424316406,
       "learning_rate": 4.557311741559135e-07,
-      "loss": 0.3672,
+      "loss": 0.637,
       "step": 2092
     },
     {
-      "epoch": 2.0580708661417324,
-      "grad_norm": 16.655155181884766,
+      "epoch": 2.0600393700787403,
+      "grad_norm": 9.822489738464355,
       "learning_rate": 4.5496781305111215e-07,
-      "loss": 1.0232,
+      "loss": 0.5197,
       "step": 2093
     },
     {
-      "epoch": 2.059055118110236,
-      "grad_norm": 7.771897315979004,
+      "epoch": 2.061023622047244,
+      "grad_norm": 9.007411003112793,
       "learning_rate": 4.5420456247993354e-07,
-      "loss": 0.3397,
+      "loss": 0.3809,
       "step": 2094
     },
     {
-      "epoch": 2.0600393700787403,
-      "grad_norm": 14.735974311828613,
+      "epoch": 2.062007874015748,
+      "grad_norm": 10.58594799041748,
       "learning_rate": 4.5344142424894954e-07,
-      "loss": 0.6183,
+      "loss": 0.451,
       "step": 2095
     },
     {
-      "epoch": 2.061023622047244,
-      "grad_norm": 7.392879009246826,
+      "epoch": 2.062992125984252,
+      "grad_norm": 6.462528705596924,
       "learning_rate": 4.526784001644667e-07,
-      "loss": 0.282,
+      "loss": 0.1733,
       "step": 2096
     },
     {
-      "epoch": 2.062007874015748,
-      "grad_norm": 8.864421844482422,
+      "epoch": 2.063976377952756,
+      "grad_norm": 11.029929161071777,
       "learning_rate": 4.5191549203252103e-07,
-      "loss": 0.2667,
+      "loss": 0.4961,
       "step": 2097
     },
     {
-      "epoch": 2.062992125984252,
-      "grad_norm": 10.327754974365234,
+      "epoch": 2.0649606299212597,
+      "grad_norm": 10.875728607177734,
       "learning_rate": 4.511527016588742e-07,
-      "loss": 0.5652,
+      "loss": 0.4755,
       "step": 2098
     },
     {
-      "epoch": 2.063976377952756,
-      "grad_norm": 14.026025772094727,
+      "epoch": 2.065944881889764,
+      "grad_norm": 9.324095726013184,
       "learning_rate": 4.503900308490092e-07,
-      "loss": 1.0643,
+      "loss": 0.3887,
       "step": 2099
     },
     {
-      "epoch": 2.0649606299212597,
-      "grad_norm": 11.10537052154541,
+      "epoch": 2.0669291338582676,
+      "grad_norm": 15.244876861572266,
       "learning_rate": 4.496274814081259e-07,
-      "loss": 0.6203,
+      "loss": 0.9013,
       "step": 2100
     },
     {
-      "epoch": 2.065944881889764,
-      "grad_norm": 9.276914596557617,
+      "epoch": 2.0679133858267718,
+      "grad_norm": 13.389811515808105,
       "learning_rate": 4.4886505514113695e-07,
-      "loss": 0.4321,
+      "loss": 0.5693,
       "step": 2101
     },
     {
-      "epoch": 2.0669291338582676,
-      "grad_norm": 14.24113655090332,
+      "epoch": 2.0688976377952755,
+      "grad_norm": 14.510342597961426,
       "learning_rate": 4.481027538526636e-07,
-      "loss": 0.8522,
+      "loss": 0.8441,
       "step": 2102
     },
     {
-      "epoch": 2.0679133858267718,
-      "grad_norm": 12.71141529083252,
+      "epoch": 2.0698818897637796,
+      "grad_norm": 11.707121849060059,
       "learning_rate": 4.4734057934703095e-07,
-      "loss": 0.4217,
+      "loss": 0.5542,
       "step": 2103
     },
     {
-      "epoch": 2.0688976377952755,
-      "grad_norm": 17.116039276123047,
+      "epoch": 2.0708661417322833,
+      "grad_norm": 9.184391021728516,
       "learning_rate": 4.4657853342826435e-07,
-      "loss": 0.9329,
+      "loss": 0.3462,
       "step": 2104
     },
     {
-      "epoch": 2.0698818897637796,
-      "grad_norm": 9.301265716552734,
+      "epoch": 2.0718503937007875,
+      "grad_norm": 8.211544036865234,
       "learning_rate": 4.458166179000848e-07,
-      "loss": 0.4836,
+      "loss": 0.2953,
       "step": 2105
     },
     {
-      "epoch": 2.0708661417322833,
-      "grad_norm": 8.274540901184082,
+      "epoch": 2.072834645669291,
+      "grad_norm": 14.863567352294922,
       "learning_rate": 4.4505483456590413e-07,
-      "loss": 0.1996,
+      "loss": 0.5272,
       "step": 2106
     },
     {
-      "epoch": 2.0718503937007875,
-      "grad_norm": 14.8345308303833,
+      "epoch": 2.0738188976377954,
+      "grad_norm": 8.472002029418945,
       "learning_rate": 4.442931852288219e-07,
-      "loss": 0.7449,
+      "loss": 0.2177,
       "step": 2107
     },
     {
-      "epoch": 2.072834645669291,
-      "grad_norm": 10.62204360961914,
+      "epoch": 2.074803149606299,
+      "grad_norm": 9.405238151550293,
       "learning_rate": 4.435316716916202e-07,
-      "loss": 0.4782,
+      "loss": 0.4521,
       "step": 2108
     },
     {
-      "epoch": 2.0738188976377954,
-      "grad_norm": 6.196303367614746,
+      "epoch": 2.0757874015748032,
+      "grad_norm": 11.434769630432129,
       "learning_rate": 4.4277029575675983e-07,
-      "loss": 0.2508,
+      "loss": 0.8877,
       "step": 2109
     },
     {
-      "epoch": 2.074803149606299,
-      "grad_norm": 10.151144981384277,
+      "epoch": 2.076771653543307,
+      "grad_norm": 8.168601036071777,
       "learning_rate": 4.4200905922637575e-07,
-      "loss": 0.1712,
+      "loss": 0.3271,
       "step": 2110
     },
     {
-      "epoch": 2.0757874015748032,
-      "grad_norm": 13.606684684753418,
+      "epoch": 2.077755905511811,
+      "grad_norm": 6.86898136138916,
       "learning_rate": 4.4124796390227313e-07,
-      "loss": 0.9656,
+      "loss": 0.3271,
       "step": 2111
     },
     {
-      "epoch": 2.076771653543307,
-      "grad_norm": 11.874922752380371,
+      "epoch": 2.078740157480315,
+      "grad_norm": 7.66591215133667,
       "learning_rate": 4.4048701158592277e-07,
-      "loss": 0.4058,
+      "loss": 0.2575,
       "step": 2112
     },
     {
-      "epoch": 2.077755905511811,
-      "grad_norm": 12.032085418701172,
+      "epoch": 2.079724409448819,
+      "grad_norm": 11.074723243713379,
       "learning_rate": 4.3972620407845714e-07,
-      "loss": 0.3136,
+      "loss": 0.4542,
       "step": 2113
     },
     {
-      "epoch": 2.078740157480315,
-      "grad_norm": 12.020431518554688,
+      "epoch": 2.0807086614173227,
+      "grad_norm": 10.790759086608887,
       "learning_rate": 4.3896554318066556e-07,
-      "loss": 0.6674,
+      "loss": 0.4742,
       "step": 2114
     },
     {
-      "epoch": 2.079724409448819,
-      "grad_norm": 11.953283309936523,
+      "epoch": 2.081692913385827,
+      "grad_norm": 10.077275276184082,
       "learning_rate": 4.3820503069299077e-07,
-      "loss": 0.5842,
+      "loss": 0.4831,
       "step": 2115
     },
     {
-      "epoch": 2.0807086614173227,
-      "grad_norm": 10.502331733703613,
+      "epoch": 2.0826771653543306,
+      "grad_norm": 11.153060913085938,
       "learning_rate": 4.3744466841552416e-07,
-      "loss": 0.5443,
+      "loss": 0.5033,
       "step": 2116
     },
     {
-      "epoch": 2.081692913385827,
-      "grad_norm": 12.981659889221191,
+      "epoch": 2.0836614173228347,
+      "grad_norm": 12.802962303161621,
       "learning_rate": 4.3668445814800137e-07,
-      "loss": 0.678,
+      "loss": 0.7284,
       "step": 2117
     },
     {
-      "epoch": 2.0826771653543306,
-      "grad_norm": 7.417089939117432,
+      "epoch": 2.0846456692913384,
+      "grad_norm": 14.490029335021973,
       "learning_rate": 4.359244016897984e-07,
-      "loss": 0.237,
+      "loss": 0.9458,
       "step": 2118
     },
     {
-      "epoch": 2.0836614173228347,
-      "grad_norm": 9.361095428466797,
+      "epoch": 2.0856299212598426,
+      "grad_norm": 9.921537399291992,
       "learning_rate": 4.3516450083992714e-07,
-      "loss": 0.4916,
+      "loss": 0.5841,
       "step": 2119
     },
     {
-      "epoch": 2.0846456692913384,
-      "grad_norm": 12.709978103637695,
+      "epoch": 2.0866141732283463,
+      "grad_norm": 9.66409969329834,
       "learning_rate": 4.344047573970313e-07,
-      "loss": 0.7586,
+      "loss": 0.3642,
       "step": 2120
     },
     {
-      "epoch": 2.0856299212598426,
-      "grad_norm": 13.901938438415527,
+      "epoch": 2.0875984251968505,
+      "grad_norm": 8.675952911376953,
       "learning_rate": 4.336451731593817e-07,
-      "loss": 0.5489,
+      "loss": 0.2957,
       "step": 2121
     },
     {
-      "epoch": 2.0866141732283463,
-      "grad_norm": 13.167049407958984,
+      "epoch": 2.088582677165354,
+      "grad_norm": 6.787413597106934,
       "learning_rate": 4.328857499248727e-07,
-      "loss": 0.4912,
+      "loss": 0.2844,
       "step": 2122
     },
     {
-      "epoch": 2.0875984251968505,
-      "grad_norm": 13.484986305236816,
+      "epoch": 2.0895669291338583,
+      "grad_norm": 4.220664978027344,
       "learning_rate": 4.321264894910173e-07,
-      "loss": 0.7887,
+      "loss": 0.0965,
       "step": 2123
     },
     {
-      "epoch": 2.088582677165354,
-      "grad_norm": 11.721290588378906,
+      "epoch": 2.090551181102362,
+      "grad_norm": 9.611949920654297,
       "learning_rate": 4.3136739365494334e-07,
-      "loss": 0.5329,
+      "loss": 0.3975,
       "step": 2124
     },
     {
-      "epoch": 2.0895669291338583,
-      "grad_norm": 5.6164093017578125,
+      "epoch": 2.091535433070866,
+      "grad_norm": 12.23066520690918,
       "learning_rate": 4.306084642133889e-07,
-      "loss": 0.1016,
+      "loss": 0.7468,
       "step": 2125
     },
     {
-      "epoch": 2.090551181102362,
-      "grad_norm": 10.29283618927002,
+      "epoch": 2.09251968503937,
+      "grad_norm": 8.946391105651855,
       "learning_rate": 4.2984970296269856e-07,
-      "loss": 0.3978,
+      "loss": 0.4677,
       "step": 2126
     },
     {
-      "epoch": 2.091535433070866,
-      "grad_norm": 12.174274444580078,
+      "epoch": 2.093503937007874,
+      "grad_norm": 9.884770393371582,
       "learning_rate": 4.290911116988181e-07,
-      "loss": 0.4734,
+      "loss": 0.4633,
       "step": 2127
     },
     {
-      "epoch": 2.09251968503937,
-      "grad_norm": 8.400107383728027,
+      "epoch": 2.094488188976378,
+      "grad_norm": 11.755213737487793,
       "learning_rate": 4.2833269221729153e-07,
-      "loss": 0.3846,
+      "loss": 0.6358,
       "step": 2128
     },
     {
-      "epoch": 2.093503937007874,
-      "grad_norm": 6.006407260894775,
+      "epoch": 2.095472440944882,
+      "grad_norm": 8.471250534057617,
       "learning_rate": 4.275744463132564e-07,
-      "loss": 0.1579,
+      "loss": 0.2283,
       "step": 2129
     },
     {
-      "epoch": 2.094488188976378,
-      "grad_norm": 11.806699752807617,
+      "epoch": 2.0964566929133857,
+      "grad_norm": 8.962312698364258,
       "learning_rate": 4.268163757814388e-07,
-      "loss": 0.5939,
+      "loss": 0.5378,
       "step": 2130
     },
     {
-      "epoch": 2.095472440944882,
-      "grad_norm": 12.883883476257324,
+      "epoch": 2.09744094488189,
+      "grad_norm": 11.992077827453613,
       "learning_rate": 4.2605848241615035e-07,
-      "loss": 0.5173,
+      "loss": 0.7019,
       "step": 2131
     },
     {
-      "epoch": 2.0964566929133857,
-      "grad_norm": 8.410216331481934,
+      "epoch": 2.0984251968503935,
+      "grad_norm": 10.937454223632812,
       "learning_rate": 4.2530076801128275e-07,
-      "loss": 0.3282,
+      "loss": 0.5093,
       "step": 2132
     },
     {
-      "epoch": 2.09744094488189,
-      "grad_norm": 11.412123680114746,
+      "epoch": 2.0994094488188977,
+      "grad_norm": 6.808335304260254,
       "learning_rate": 4.2454323436030457e-07,
-      "loss": 0.5281,
+      "loss": 0.2462,
       "step": 2133
     },
     {
-      "epoch": 2.0984251968503935,
-      "grad_norm": 10.594966888427734,
+      "epoch": 2.1003937007874014,
+      "grad_norm": 12.572240829467773,
       "learning_rate": 4.2378588325625627e-07,
-      "loss": 0.4023,
+      "loss": 0.4929,
       "step": 2134
     },
     {
-      "epoch": 2.0994094488188977,
-      "grad_norm": 8.408379554748535,
+      "epoch": 2.1013779527559056,
+      "grad_norm": 9.318580627441406,
       "learning_rate": 4.2302871649174646e-07,
-      "loss": 0.3061,
+      "loss": 0.4297,
       "step": 2135
     },
     {
-      "epoch": 2.1003937007874014,
-      "grad_norm": 8.434245109558105,
+      "epoch": 2.1023622047244093,
+      "grad_norm": 8.599699974060059,
       "learning_rate": 4.2227173585894724e-07,
-      "loss": 0.4158,
+      "loss": 0.3427,
       "step": 2136
     },
     {
-      "epoch": 2.1013779527559056,
-      "grad_norm": 11.54025936126709,
+      "epoch": 2.1033464566929134,
+      "grad_norm": 12.969562530517578,
       "learning_rate": 4.2151494314959015e-07,
-      "loss": 0.5952,
+      "loss": 0.7622,
       "step": 2137
     },
     {
-      "epoch": 2.1023622047244093,
-      "grad_norm": 10.130776405334473,
+      "epoch": 2.104330708661417,
+      "grad_norm": 10.069689750671387,
       "learning_rate": 4.20758340154962e-07,
-      "loss": 0.3903,
+      "loss": 0.6178,
       "step": 2138
     },
     {
-      "epoch": 2.1033464566929134,
-      "grad_norm": 10.995735168457031,
+      "epoch": 2.1053149606299213,
+      "grad_norm": 11.031486511230469,
       "learning_rate": 4.200019286659007e-07,
-      "loss": 0.3577,
+      "loss": 0.478,
       "step": 2139
     },
     {
-      "epoch": 2.104330708661417,
-      "grad_norm": 6.9837327003479,
+      "epoch": 2.106299212598425,
+      "grad_norm": 7.9008469581604,
       "learning_rate": 4.192457104727905e-07,
-      "loss": 0.196,
+      "loss": 0.4509,
       "step": 2140
     },
     {
-      "epoch": 2.1053149606299213,
-      "grad_norm": 9.005616188049316,
+      "epoch": 2.107283464566929,
+      "grad_norm": 7.731719017028809,
       "learning_rate": 4.184896873655584e-07,
-      "loss": 0.38,
+      "loss": 0.3435,
       "step": 2141
     },
     {
-      "epoch": 2.106299212598425,
-      "grad_norm": 15.84337329864502,
+      "epoch": 2.108267716535433,
+      "grad_norm": 7.314538955688477,
       "learning_rate": 4.177338611336697e-07,
-      "loss": 0.6768,
+      "loss": 0.3346,
       "step": 2142
     },
     {
-      "epoch": 2.106299212598425,
-      "eval_Qnli-dev_cosine_accuracy": 0.671875,
-      "eval_Qnli-dev_cosine_accuracy_threshold": 0.8106182813644409,
-      "eval_Qnli-dev_cosine_ap": 0.6921761205800739,
-      "eval_Qnli-dev_cosine_f1": 0.6859903381642513,
-      "eval_Qnli-dev_cosine_f1_threshold": 0.7057304382324219,
-      "eval_Qnli-dev_cosine_precision": 0.5532467532467532,
-      "eval_Qnli-dev_cosine_recall": 0.902542372881356,
+      "epoch": 2.108267716535433,
+      "eval_Qnli-dev_cosine_accuracy": 0.67578125,
+      "eval_Qnli-dev_cosine_accuracy_threshold": 0.7934461832046509,
+      "eval_Qnli-dev_cosine_ap": 0.6989545476819907,
+      "eval_Qnli-dev_cosine_f1": 0.6877076411960132,
+      "eval_Qnli-dev_cosine_f1_threshold": 0.7384255528450012,
+      "eval_Qnli-dev_cosine_precision": 0.5655737704918032,
+      "eval_Qnli-dev_cosine_recall": 0.8771186440677966,
       "eval_Qnli-dev_dot_accuracy": 0.673828125,
-      "eval_Qnli-dev_dot_accuracy_threshold": 613.2439575195312,
-      "eval_Qnli-dev_dot_ap": 0.6888309953492245,
-      "eval_Qnli-dev_dot_f1": 0.685064935064935,
-      "eval_Qnli-dev_dot_f1_threshold": 541.5089721679688,
-      "eval_Qnli-dev_dot_precision": 0.5552631578947368,
-      "eval_Qnli-dev_dot_recall": 0.8940677966101694,
-      "eval_Qnli-dev_euclidean_accuracy": 0.673828125,
-      "eval_Qnli-dev_euclidean_accuracy_threshold": 17.046558380126953,
-      "eval_Qnli-dev_euclidean_ap": 0.692753086975512,
-      "eval_Qnli-dev_euclidean_f1": 0.6838709677419356,
-      "eval_Qnli-dev_euclidean_f1_threshold": 21.16459083557129,
-      "eval_Qnli-dev_euclidean_precision": 0.5520833333333334,
-      "eval_Qnli-dev_euclidean_recall": 0.8983050847457628,
-      "eval_Qnli-dev_manhattan_accuracy": 0.6875,
-      "eval_Qnli-dev_manhattan_accuracy_threshold": 371.18255615234375,
-      "eval_Qnli-dev_manhattan_ap": 0.7050924683830799,
-      "eval_Qnli-dev_manhattan_f1": 0.686106346483705,
-      "eval_Qnli-dev_manhattan_f1_threshold": 425.8172607421875,
-      "eval_Qnli-dev_manhattan_precision": 0.5763688760806917,
-      "eval_Qnli-dev_manhattan_recall": 0.847457627118644,
-      "eval_Qnli-dev_max_accuracy": 0.6875,
-      "eval_Qnli-dev_max_accuracy_threshold": 613.2439575195312,
-      "eval_Qnli-dev_max_ap": 0.7050924683830799,
-      "eval_Qnli-dev_max_f1": 0.686106346483705,
-      "eval_Qnli-dev_max_f1_threshold": 541.5089721679688,
-      "eval_Qnli-dev_max_precision": 0.5763688760806917,
-      "eval_Qnli-dev_max_recall": 0.902542372881356,
-      "eval_allNLI-dev_cosine_accuracy": 0.689453125,
-      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9146234393119812,
-      "eval_allNLI-dev_cosine_ap": 0.5060845831275185,
-      "eval_allNLI-dev_cosine_f1": 0.5430210325047802,
-      "eval_allNLI-dev_cosine_f1_threshold": 0.7860702276229858,
-      "eval_allNLI-dev_cosine_precision": 0.4057142857142857,
-      "eval_allNLI-dev_cosine_recall": 0.8208092485549133,
-      "eval_allNLI-dev_dot_accuracy": 0.689453125,
-      "eval_allNLI-dev_dot_accuracy_threshold": 700.458740234375,
-      "eval_allNLI-dev_dot_ap": 0.5043834893000478,
-      "eval_allNLI-dev_dot_f1": 0.5440613026819924,
-      "eval_allNLI-dev_dot_f1_threshold": 601.576904296875,
-      "eval_allNLI-dev_dot_precision": 0.4068767908309456,
-      "eval_allNLI-dev_dot_recall": 0.8208092485549133,
-      "eval_allNLI-dev_euclidean_accuracy": 0.689453125,
-      "eval_allNLI-dev_euclidean_accuracy_threshold": 11.43293571472168,
-      "eval_allNLI-dev_euclidean_ap": 0.505756126438367,
-      "eval_allNLI-dev_euclidean_f1": 0.5430210325047802,
-      "eval_allNLI-dev_euclidean_f1_threshold": 18.10074234008789,
-      "eval_allNLI-dev_euclidean_precision": 0.4057142857142857,
-      "eval_allNLI-dev_euclidean_recall": 0.8208092485549133,
-      "eval_allNLI-dev_manhattan_accuracy": 0.681640625,
-      "eval_allNLI-dev_manhattan_accuracy_threshold": 216.44842529296875,
-      "eval_allNLI-dev_manhattan_ap": 0.5042689696590352,
-      "eval_allNLI-dev_manhattan_f1": 0.5437616387337059,
-      "eval_allNLI-dev_manhattan_f1_threshold": 386.01776123046875,
-      "eval_allNLI-dev_manhattan_precision": 0.4010989010989011,
-      "eval_allNLI-dev_manhattan_recall": 0.8439306358381503,
-      "eval_allNLI-dev_max_accuracy": 0.689453125,
-      "eval_allNLI-dev_max_accuracy_threshold": 700.458740234375,
-      "eval_allNLI-dev_max_ap": 0.5060845831275185,
-      "eval_allNLI-dev_max_f1": 0.5440613026819924,
-      "eval_allNLI-dev_max_f1_threshold": 601.576904296875,
-      "eval_allNLI-dev_max_precision": 0.4068767908309456,
-      "eval_allNLI-dev_max_recall": 0.8439306358381503,
-      "eval_loss": 0.9628643989562988,
-      "eval_runtime": 54.1521,
-      "eval_samples_per_second": 30.728,
-      "eval_sequential_score": 0.7050924683830799,
-      "eval_steps_per_second": 0.129,
-      "eval_sts-test_pearson_cosine": 0.6202459810611565,
-      "eval_sts-test_pearson_dot": 0.6186786427489253,
-      "eval_sts-test_pearson_euclidean": 0.6307823993253823,
-      "eval_sts-test_pearson_manhattan": 0.63666039543781,
-      "eval_sts-test_pearson_max": 0.63666039543781,
-      "eval_sts-test_spearman_cosine": 0.6312596821416715,
-      "eval_sts-test_spearman_dot": 0.630321964210476,
-      "eval_sts-test_spearman_euclidean": 0.6316041773518624,
-      "eval_sts-test_spearman_manhattan": 0.6356551062842734,
-      "eval_sts-test_spearman_max": 0.6356551062842734,
+      "eval_Qnli-dev_dot_accuracy_threshold": 598.9378662109375,
+      "eval_Qnli-dev_dot_ap": 0.6981506758362209,
+      "eval_Qnli-dev_dot_f1": 0.6869409660107335,
+      "eval_Qnli-dev_dot_f1_threshold": 579.2918701171875,
+      "eval_Qnli-dev_dot_precision": 0.5944272445820433,
+      "eval_Qnli-dev_dot_recall": 0.8135593220338984,
+      "eval_Qnli-dev_euclidean_accuracy": 0.67578125,
+      "eval_Qnli-dev_euclidean_accuracy_threshold": 17.748992919921875,
+      "eval_Qnli-dev_euclidean_ap": 0.6994649561937168,
+      "eval_Qnli-dev_euclidean_f1": 0.6880570409982174,
+      "eval_Qnli-dev_euclidean_f1_threshold": 19.142541885375977,
+      "eval_Qnli-dev_euclidean_precision": 0.5938461538461538,
+      "eval_Qnli-dev_euclidean_recall": 0.8177966101694916,
+      "eval_Qnli-dev_manhattan_accuracy": 0.685546875,
+      "eval_Qnli-dev_manhattan_accuracy_threshold": 379.6617431640625,
+      "eval_Qnli-dev_manhattan_ap": 0.7046887593063369,
+      "eval_Qnli-dev_manhattan_f1": 0.6867924528301886,
+      "eval_Qnli-dev_manhattan_f1_threshold": 386.82403564453125,
+      "eval_Qnli-dev_manhattan_precision": 0.6190476190476191,
+      "eval_Qnli-dev_manhattan_recall": 0.7711864406779662,
+      "eval_Qnli-dev_max_accuracy": 0.685546875,
+      "eval_Qnli-dev_max_accuracy_threshold": 598.9378662109375,
+      "eval_Qnli-dev_max_ap": 0.7046887593063369,
+      "eval_Qnli-dev_max_f1": 0.6880570409982174,
+      "eval_Qnli-dev_max_f1_threshold": 579.2918701171875,
+      "eval_Qnli-dev_max_precision": 0.6190476190476191,
+      "eval_Qnli-dev_max_recall": 0.8771186440677966,
+      "eval_allNLI-dev_cosine_accuracy": 0.6953125,
+      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9204268455505371,
+      "eval_allNLI-dev_cosine_ap": 0.508913349349435,
+      "eval_allNLI-dev_cosine_f1": 0.5450733752620545,
+      "eval_allNLI-dev_cosine_f1_threshold": 0.8241016864776611,
+      "eval_allNLI-dev_cosine_precision": 0.4276315789473684,
+      "eval_allNLI-dev_cosine_recall": 0.7514450867052023,
+      "eval_allNLI-dev_dot_accuracy": 0.693359375,
+      "eval_allNLI-dev_dot_accuracy_threshold": 707.6104125976562,
+      "eval_allNLI-dev_dot_ap": 0.5119631017815847,
+      "eval_allNLI-dev_dot_f1": 0.5438596491228072,
+      "eval_allNLI-dev_dot_f1_threshold": 634.8453369140625,
+      "eval_allNLI-dev_dot_precision": 0.4381625441696113,
+      "eval_allNLI-dev_dot_recall": 0.7167630057803468,
+      "eval_allNLI-dev_euclidean_accuracy": 0.6953125,
+      "eval_allNLI-dev_euclidean_accuracy_threshold": 11.027429580688477,
+      "eval_allNLI-dev_euclidean_ap": 0.5085079595360542,
+      "eval_allNLI-dev_euclidean_f1": 0.5450733752620545,
+      "eval_allNLI-dev_euclidean_f1_threshold": 16.372371673583984,
+      "eval_allNLI-dev_euclidean_precision": 0.4276315789473684,
+      "eval_allNLI-dev_euclidean_recall": 0.7514450867052023,
+      "eval_allNLI-dev_manhattan_accuracy": 0.69140625,
+      "eval_allNLI-dev_manhattan_accuracy_threshold": 239.98931884765625,
+      "eval_allNLI-dev_manhattan_ap": 0.5060172612660457,
+      "eval_allNLI-dev_manhattan_f1": 0.5454545454545454,
+      "eval_allNLI-dev_manhattan_f1_threshold": 342.2213134765625,
+      "eval_allNLI-dev_manhattan_precision": 0.43,
+      "eval_allNLI-dev_manhattan_recall": 0.7456647398843931,
+      "eval_allNLI-dev_max_accuracy": 0.6953125,
+      "eval_allNLI-dev_max_accuracy_threshold": 707.6104125976562,
+      "eval_allNLI-dev_max_ap": 0.5119631017815847,
+      "eval_allNLI-dev_max_f1": 0.5454545454545454,
+      "eval_allNLI-dev_max_f1_threshold": 634.8453369140625,
+      "eval_allNLI-dev_max_precision": 0.4381625441696113,
+      "eval_allNLI-dev_max_recall": 0.7514450867052023,
+      "eval_loss": 0.8402745723724365,
+      "eval_runtime": 52.7757,
+      "eval_samples_per_second": 31.53,
+      "eval_sequential_score": 0.7046887593063369,
+      "eval_steps_per_second": 0.133,
+      "eval_sts-test_pearson_cosine": 0.6574174242381561,
+      "eval_sts-test_pearson_dot": 0.6552793345605479,
+      "eval_sts-test_pearson_euclidean": 0.6709799048687015,
+      "eval_sts-test_pearson_manhattan": 0.6751941182586493,
+      "eval_sts-test_pearson_max": 0.6751941182586493,
+      "eval_sts-test_spearman_cosine": 0.6651209478297808,
+      "eval_sts-test_spearman_dot": 0.6628165263756921,
+      "eval_sts-test_spearman_euclidean": 0.6654525099254297,
+      "eval_sts-test_spearman_manhattan": 0.6697684110765391,
+      "eval_sts-test_spearman_max": 0.6697684110765391,
       "step": 2142
     },
     {
-      "epoch": 2.107283464566929,
-      "grad_norm": 11.44278335571289,
+      "epoch": 2.109251968503937,
+      "grad_norm": 12.002259254455566,
       "learning_rate": 4.169782335661235e-07,
-      "loss": 0.5215,
+      "loss": 0.4465,
       "step": 2143
     },
     {
-      "epoch": 2.108267716535433,
-      "grad_norm": 10.876325607299805,
+      "epoch": 2.1102362204724407,
+      "grad_norm": 12.813706398010254,
       "learning_rate": 4.1622280645144883e-07,
-      "loss": 0.3134,
+      "loss": 0.8581,
       "step": 2144
     },
     {
-      "epoch": 2.109251968503937,
-      "grad_norm": 9.747419357299805,
+      "epoch": 2.111220472440945,
+      "grad_norm": 11.165507316589355,
       "learning_rate": 4.1546758157770004e-07,
-      "loss": 0.3255,
+      "loss": 0.4412,
       "step": 2145
     },
     {
-      "epoch": 2.1102362204724407,
-      "grad_norm": 15.33596420288086,
+      "epoch": 2.1122047244094486,
+      "grad_norm": 12.227032661437988,
       "learning_rate": 4.1471256073245306e-07,
-      "loss": 0.8023,
+      "loss": 0.4847,
       "step": 2146
     },
     {
-      "epoch": 2.111220472440945,
-      "grad_norm": 23.52260398864746,
+      "epoch": 2.1131889763779528,
+      "grad_norm": 10.416069984436035,
       "learning_rate": 4.139577457028009e-07,
-      "loss": 1.2468,
+      "loss": 0.3782,
       "step": 2147
     },
     {
-      "epoch": 2.1122047244094486,
-      "grad_norm": 13.40678882598877,
+      "epoch": 2.1141732283464565,
+      "grad_norm": 12.957719802856445,
       "learning_rate": 4.13203138275349e-07,
-      "loss": 0.7636,
+      "loss": 0.4406,
       "step": 2148
     },
     {
-      "epoch": 2.1131889763779528,
-      "grad_norm": 12.064020156860352,
+      "epoch": 2.1151574803149606,
+      "grad_norm": 11.170369148254395,
       "learning_rate": 4.124487402362119e-07,
-      "loss": 0.3759,
+      "loss": 0.4634,
       "step": 2149
     },
     {
-      "epoch": 2.1141732283464565,
-      "grad_norm": 13.580906867980957,
+      "epoch": 2.1161417322834644,
+      "grad_norm": 12.000916481018066,
       "learning_rate": 4.116945533710084e-07,
-      "loss": 0.4563,
+      "loss": 0.4512,
       "step": 2150
     },
     {
-      "epoch": 2.1151574803149606,
-      "grad_norm": 16.30206871032715,
+      "epoch": 2.1171259842519685,
+      "grad_norm": 10.328460693359375,
       "learning_rate": 4.1094057946485717e-07,
-      "loss": 0.7122,
+      "loss": 0.3702,
       "step": 2151
     },
     {
-      "epoch": 2.1161417322834644,
-      "grad_norm": 11.582755088806152,
+      "epoch": 2.1181102362204722,
+      "grad_norm": 4.709320068359375,
       "learning_rate": 4.1018682030237324e-07,
-      "loss": 0.5501,
+      "loss": 0.1144,
       "step": 2152
     },
     {
-      "epoch": 2.1171259842519685,
-      "grad_norm": 9.636466979980469,
+      "epoch": 2.1190944881889764,
+      "grad_norm": 8.625373840332031,
       "learning_rate": 4.0943327766766314e-07,
-      "loss": 0.3651,
+      "loss": 0.3481,
       "step": 2153
     },
     {
-      "epoch": 2.1181102362204722,
-      "grad_norm": 10.754108428955078,
+      "epoch": 2.12007874015748,
+      "grad_norm": 10.802593231201172,
       "learning_rate": 4.086799533443208e-07,
-      "loss": 0.3793,
+      "loss": 0.4423,
       "step": 2154
     },
     {
-      "epoch": 2.1190944881889764,
-      "grad_norm": 13.00988483428955,
+      "epoch": 2.1210629921259843,
+      "grad_norm": 13.430143356323242,
       "learning_rate": 4.0792684911542373e-07,
-      "loss": 0.6461,
+      "loss": 0.7575,
       "step": 2155
     },
     {
-      "epoch": 2.12007874015748,
-      "grad_norm": 11.099771499633789,
+      "epoch": 2.122047244094488,
+      "grad_norm": 9.047517776489258,
       "learning_rate": 4.071739667635281e-07,
-      "loss": 0.5457,
+      "loss": 0.3417,
       "step": 2156
     },
     {
-      "epoch": 2.1210629921259843,
-      "grad_norm": 6.3068623542785645,
+      "epoch": 2.123031496062992,
+      "grad_norm": 12.781135559082031,
       "learning_rate": 4.064213080706652e-07,
-      "loss": 0.1725,
+      "loss": 0.4928,
       "step": 2157
     },
     {
-      "epoch": 2.122047244094488,
-      "grad_norm": 6.355545520782471,
+      "epoch": 2.1240157480314963,
+      "grad_norm": 9.351764678955078,
       "learning_rate": 4.056688748183368e-07,
-      "loss": 0.227,
+      "loss": 0.4311,
       "step": 2158
     },
     {
-      "epoch": 2.123031496062992,
-      "grad_norm": 14.778956413269043,
+      "epoch": 2.125,
+      "grad_norm": 9.538174629211426,
       "learning_rate": 4.049166687875111e-07,
-      "loss": 0.529,
+      "loss": 0.4382,
       "step": 2159
     },
     {
-      "epoch": 2.1240157480314963,
-      "grad_norm": 11.180927276611328,
+      "epoch": 2.1259842519685037,
+      "grad_norm": 9.007962226867676,
       "learning_rate": 4.0416469175861855e-07,
-      "loss": 0.3569,
+      "loss": 0.3946,
       "step": 2160
     },
     {
-      "epoch": 2.125,
-      "grad_norm": 8.885268211364746,
+      "epoch": 2.126968503937008,
+      "grad_norm": 6.888474464416504,
       "learning_rate": 4.034129455115474e-07,
-      "loss": 0.3879,
+      "loss": 0.2044,
       "step": 2161
     },
     {
-      "epoch": 2.1259842519685037,
-      "grad_norm": 9.933160781860352,
+      "epoch": 2.127952755905512,
+      "grad_norm": 12.265450477600098,
       "learning_rate": 4.0266143182563974e-07,
-      "loss": 0.3999,
+      "loss": 0.6667,
       "step": 2162
     },
     {
-      "epoch": 2.126968503937008,
-      "grad_norm": 16.638206481933594,
+      "epoch": 2.1289370078740157,
+      "grad_norm": 8.379030227661133,
       "learning_rate": 4.019101524796874e-07,
-      "loss": 0.3085,
+      "loss": 0.2457,
       "step": 2163
     },
     {
-      "epoch": 2.127952755905512,
-      "grad_norm": 17.310379028320312,
+      "epoch": 2.1299212598425195,
+      "grad_norm": 14.336709022521973,
       "learning_rate": 4.0115910925192705e-07,
-      "loss": 0.7374,
+      "loss": 0.9209,
       "step": 2164
     },
     {
-      "epoch": 2.1289370078740157,
-      "grad_norm": 10.217885971069336,
+      "epoch": 2.1309055118110236,
+      "grad_norm": 9.62060260772705,
       "learning_rate": 4.004083039200369e-07,
-      "loss": 0.2956,
+      "loss": 0.5682,
       "step": 2165
     },
     {
-      "epoch": 2.1299212598425195,
-      "grad_norm": 10.47181510925293,
+      "epoch": 2.1318897637795278,
+      "grad_norm": 12.849811553955078,
       "learning_rate": 3.99657738261132e-07,
-      "loss": 0.703,
+      "loss": 0.5466,
       "step": 2166
     },
     {
-      "epoch": 2.1309055118110236,
-      "grad_norm": 17.75139808654785,
+      "epoch": 2.1328740157480315,
+      "grad_norm": 8.40816879272461,
       "learning_rate": 3.9890741405176e-07,
-      "loss": 0.5395,
+      "loss": 0.3111,
       "step": 2167
     },
     {
-      "epoch": 2.1318897637795278,
-      "grad_norm": 12.336260795593262,
+      "epoch": 2.1338582677165356,
+      "grad_norm": 10.843924522399902,
       "learning_rate": 3.981573330678973e-07,
-      "loss": 0.5119,
+      "loss": 0.6332,
       "step": 2168
     },
     {
-      "epoch": 2.1328740157480315,
-      "grad_norm": 9.175243377685547,
+      "epoch": 2.1348425196850394,
+      "grad_norm": 9.696345329284668,
       "learning_rate": 3.9740749708494405e-07,
-      "loss": 0.3965,
+      "loss": 0.3745,
       "step": 2169
     },
     {
-      "epoch": 2.1338582677165356,
-      "grad_norm": 11.417914390563965,
+      "epoch": 2.1358267716535435,
+      "grad_norm": 7.173345565795898,
       "learning_rate": 3.96657907877721e-07,
-      "loss": 0.6092,
+      "loss": 0.2525,
       "step": 2170
     },
     {
-      "epoch": 2.1348425196850394,
-      "grad_norm": 10.03918170928955,
+      "epoch": 2.1368110236220472,
+      "grad_norm": 7.381486415863037,
       "learning_rate": 3.959085672204647e-07,
-      "loss": 0.4268,
+      "loss": 0.3447,
       "step": 2171
     },
     {
-      "epoch": 2.1358267716535435,
-      "grad_norm": 15.49741268157959,
+      "epoch": 2.1377952755905514,
+      "grad_norm": 10.929116249084473,
       "learning_rate": 3.951594768868233e-07,
-      "loss": 0.5907,
+      "loss": 0.3985,
       "step": 2172
     },
     {
-      "epoch": 2.1368110236220472,
-      "grad_norm": 10.845100402832031,
+      "epoch": 2.138779527559055,
+      "grad_norm": 9.900410652160645,
       "learning_rate": 3.9441063864985255e-07,
-      "loss": 0.4374,
+      "loss": 0.402,
       "step": 2173
     },
     {
-      "epoch": 2.1377952755905514,
-      "grad_norm": 7.834804058074951,
+      "epoch": 2.1397637795275593,
+      "grad_norm": 16.318891525268555,
       "learning_rate": 3.936620542820114e-07,
-      "loss": 0.2971,
+      "loss": 0.7393,
       "step": 2174
     },
     {
-      "epoch": 2.138779527559055,
-      "grad_norm": 10.065025329589844,
+      "epoch": 2.140748031496063,
+      "grad_norm": 7.925287246704102,
       "learning_rate": 3.9291372555515794e-07,
-      "loss": 0.4096,
+      "loss": 0.2344,
       "step": 2175
     },
     {
-      "epoch": 2.1397637795275593,
-      "grad_norm": 10.566422462463379,
+      "epoch": 2.141732283464567,
+      "grad_norm": 11.985597610473633,
       "learning_rate": 3.921656542405452e-07,
-      "loss": 0.627,
+      "loss": 0.4252,
       "step": 2176
     },
     {
-      "epoch": 2.140748031496063,
-      "grad_norm": 10.61729621887207,
+      "epoch": 2.142716535433071,
+      "grad_norm": 10.37873649597168,
       "learning_rate": 3.9141784210881695e-07,
-      "loss": 0.4026,
+      "loss": 0.3969,
       "step": 2177
     },
     {
-      "epoch": 2.141732283464567,
-      "grad_norm": 10.74808120727539,
+      "epoch": 2.143700787401575,
+      "grad_norm": 11.185521125793457,
       "learning_rate": 3.9067029093000335e-07,
-      "loss": 0.4056,
+      "loss": 0.547,
       "step": 2178
     },
     {
-      "epoch": 2.142716535433071,
-      "grad_norm": 9.267818450927734,
+      "epoch": 2.1446850393700787,
+      "grad_norm": 7.473401069641113,
       "learning_rate": 3.899230024735171e-07,
-      "loss": 0.3453,
+      "loss": 0.3339,
       "step": 2179
     },
     {
-      "epoch": 2.143700787401575,
-      "grad_norm": 6.095523357391357,
+      "epoch": 2.145669291338583,
+      "grad_norm": 7.62275505065918,
       "learning_rate": 3.8917597850814885e-07,
-      "loss": 0.2895,
+      "loss": 0.2952,
       "step": 2180
     },
     {
-      "epoch": 2.1446850393700787,
-      "grad_norm": 11.208436012268066,
+      "epoch": 2.1466535433070866,
+      "grad_norm": 8.859162330627441,
       "learning_rate": 3.884292208020635e-07,
-      "loss": 0.5194,
+      "loss": 0.2556,
       "step": 2181
     },
     {
-      "epoch": 2.145669291338583,
-      "grad_norm": 15.45287799835205,
+      "epoch": 2.1476377952755907,
+      "grad_norm": 10.127699851989746,
       "learning_rate": 3.876827311227954e-07,
-      "loss": 0.8513,
+      "loss": 0.5002,
       "step": 2182
     },
     {
-      "epoch": 2.1466535433070866,
-      "grad_norm": 9.89079761505127,
+      "epoch": 2.1486220472440944,
+      "grad_norm": 10.91092586517334,
       "learning_rate": 3.8693651123724455e-07,
-      "loss": 0.4421,
+      "loss": 0.5223,
       "step": 2183
     },
     {
-      "epoch": 2.1476377952755907,
-      "grad_norm": 8.460564613342285,
+      "epoch": 2.1496062992125986,
+      "grad_norm": 2.6207144260406494,
       "learning_rate": 3.861905629116725e-07,
-      "loss": 0.486,
+      "loss": 0.0491,
       "step": 2184
     },
     {
-      "epoch": 2.1486220472440944,
-      "grad_norm": 12.875810623168945,
+      "epoch": 2.1505905511811023,
+      "grad_norm": 13.004858016967773,
       "learning_rate": 3.854448879116981e-07,
-      "loss": 1.1001,
+      "loss": 0.4989,
       "step": 2185
     },
     {
-      "epoch": 2.1496062992125986,
-      "grad_norm": 8.028280258178711,
+      "epoch": 2.1515748031496065,
+      "grad_norm": 12.66700267791748,
       "learning_rate": 3.84699488002293e-07,
-      "loss": 0.2871,
+      "loss": 0.6368,
       "step": 2186
     },
     {
-      "epoch": 2.1505905511811023,
-      "grad_norm": 13.88902759552002,
+      "epoch": 2.15255905511811,
+      "grad_norm": 9.519246101379395,
       "learning_rate": 3.83954364947778e-07,
-      "loss": 0.7672,
+      "loss": 0.1781,
       "step": 2187
     },
     {
-      "epoch": 2.1515748031496065,
-      "grad_norm": 12.074015617370605,
+      "epoch": 2.1535433070866143,
+      "grad_norm": 8.556527137756348,
       "learning_rate": 3.832095205118183e-07,
-      "loss": 0.8712,
+      "loss": 0.2386,
       "step": 2188
     },
     {
-      "epoch": 2.15255905511811,
-      "grad_norm": 9.419001579284668,
+      "epoch": 2.154527559055118,
+      "grad_norm": 15.067211151123047,
       "learning_rate": 3.8246495645742006e-07,
-      "loss": 0.2191,
+      "loss": 0.424,
       "step": 2189
     },
     {
-      "epoch": 2.1535433070866143,
-      "grad_norm": 12.924214363098145,
+      "epoch": 2.155511811023622,
+      "grad_norm": 10.784927368164062,
       "learning_rate": 3.8172067454692517e-07,
-      "loss": 0.617,
+      "loss": 0.4195,
       "step": 2190
     },
     {
-      "epoch": 2.154527559055118,
-      "grad_norm": 10.700763702392578,
+      "epoch": 2.156496062992126,
+      "grad_norm": 5.950012683868408,
       "learning_rate": 3.8097667654200826e-07,
-      "loss": 0.4812,
+      "loss": 0.13,
       "step": 2191
     },
     {
-      "epoch": 2.155511811023622,
-      "grad_norm": 12.375061988830566,
+      "epoch": 2.15748031496063,
+      "grad_norm": 8.245309829711914,
       "learning_rate": 3.802329642036716e-07,
-      "loss": 0.5425,
+      "loss": 0.2525,
       "step": 2192
     },
     {
-      "epoch": 2.156496062992126,
-      "grad_norm": 11.396371841430664,
+      "epoch": 2.158464566929134,
+      "grad_norm": 5.700161933898926,
       "learning_rate": 3.7948953929224155e-07,
-      "loss": 0.6849,
+      "loss": 0.1913,
       "step": 2193
     },
     {
-      "epoch": 2.15748031496063,
-      "grad_norm": 11.016363143920898,
+      "epoch": 2.159448818897638,
+      "grad_norm": 12.569465637207031,
       "learning_rate": 3.787464035673641e-07,
-      "loss": 0.4039,
+      "loss": 0.4215,
       "step": 2194
     },
     {
-      "epoch": 2.158464566929134,
-      "grad_norm": 8.921324729919434,
+      "epoch": 2.1604330708661417,
+      "grad_norm": 11.401483535766602,
       "learning_rate": 3.780035587880007e-07,
-      "loss": 0.2696,
+      "loss": 0.3393,
       "step": 2195
     },
     {
-      "epoch": 2.159448818897638,
-      "grad_norm": 12.214681625366211,
+      "epoch": 2.161417322834646,
+      "grad_norm": 17.855337142944336,
       "learning_rate": 3.7726100671242414e-07,
-      "loss": 0.6889,
+      "loss": 1.3127,
       "step": 2196
     },
     {
-      "epoch": 2.1604330708661417,
-      "grad_norm": 12.57135009765625,
+      "epoch": 2.1624015748031495,
+      "grad_norm": 13.412671089172363,
       "learning_rate": 3.765187490982144e-07,
-      "loss": 0.5543,
+      "loss": 0.6814,
       "step": 2197
     },
     {
-      "epoch": 2.161417322834646,
-      "grad_norm": 14.028517723083496,
+      "epoch": 2.1633858267716537,
+      "grad_norm": 7.205867767333984,
       "learning_rate": 3.757767877022545e-07,
-      "loss": 0.8829,
+      "loss": 0.263,
       "step": 2198
     },
     {
-      "epoch": 2.1624015748031495,
-      "grad_norm": 15.3035249710083,
+      "epoch": 2.1643700787401574,
+      "grad_norm": 9.04843521118164,
       "learning_rate": 3.750351242807264e-07,
-      "loss": 0.7397,
+      "loss": 0.3647,
       "step": 2199
     },
     {
-      "epoch": 2.1633858267716537,
-      "grad_norm": 8.669970512390137,
+      "epoch": 2.1653543307086616,
+      "grad_norm": 11.996747016906738,
       "learning_rate": 3.7429376058910674e-07,
-      "loss": 0.4366,
+      "loss": 0.5629,
       "step": 2200
     },
     {
-      "epoch": 2.1643700787401574,
-      "grad_norm": 4.055088520050049,
+      "epoch": 2.1663385826771653,
+      "grad_norm": 12.483857154846191,
       "learning_rate": 3.735526983821624e-07,
-      "loss": 0.0933,
+      "loss": 0.6589,
       "step": 2201
     },
     {
-      "epoch": 2.1653543307086616,
-      "grad_norm": 8.365303039550781,
+      "epoch": 2.1673228346456694,
+      "grad_norm": 10.485642433166504,
       "learning_rate": 3.728119394139472e-07,
-      "loss": 0.285,
+      "loss": 0.3113,
       "step": 2202
     },
     {
-      "epoch": 2.1663385826771653,
-      "grad_norm": 9.966198921203613,
+      "epoch": 2.168307086614173,
+      "grad_norm": 12.677536964416504,
       "learning_rate": 3.7207148543779677e-07,
-      "loss": 0.5301,
+      "loss": 0.6811,
       "step": 2203
     },
     {
-      "epoch": 2.1673228346456694,
-      "grad_norm": 14.930325508117676,
+      "epoch": 2.1692913385826773,
+      "grad_norm": 11.008816719055176,
       "learning_rate": 3.713313382063253e-07,
-      "loss": 0.6857,
+      "loss": 0.5942,
       "step": 2204
     },
     {
-      "epoch": 2.168307086614173,
-      "grad_norm": 11.771857261657715,
+      "epoch": 2.170275590551181,
+      "grad_norm": 8.265438079833984,
       "learning_rate": 3.705914994714203e-07,
-      "loss": 0.7281,
+      "loss": 0.2195,
       "step": 2205
     },
     {
-      "epoch": 2.1692913385826773,
-      "grad_norm": 10.2894287109375,
+      "epoch": 2.171259842519685,
+      "grad_norm": 16.324005126953125,
       "learning_rate": 3.6985197098423957e-07,
-      "loss": 0.4846,
+      "loss": 0.9414,
       "step": 2206
     },
     {
-      "epoch": 2.170275590551181,
-      "grad_norm": 9.719088554382324,
+      "epoch": 2.172244094488189,
+      "grad_norm": 10.122859954833984,
       "learning_rate": 3.691127544952065e-07,
-      "loss": 0.3703,
+      "loss": 0.2976,
       "step": 2207
     },
     {
-      "epoch": 2.171259842519685,
-      "grad_norm": 12.411601066589355,
+      "epoch": 2.173228346456693,
+      "grad_norm": 11.006841659545898,
       "learning_rate": 3.6837385175400597e-07,
-      "loss": 0.589,
+      "loss": 0.3168,
       "step": 2208
     },
     {
-      "epoch": 2.172244094488189,
-      "grad_norm": 10.245353698730469,
+      "epoch": 2.1742125984251968,
+      "grad_norm": 8.506003379821777,
       "learning_rate": 3.6763526450958005e-07,
-      "loss": 0.6338,
+      "loss": 0.3567,
       "step": 2209
     },
     {
-      "epoch": 2.173228346456693,
-      "grad_norm": 10.593074798583984,
+      "epoch": 2.175196850393701,
+      "grad_norm": 13.206911087036133,
       "learning_rate": 3.668969945101245e-07,
-      "loss": 0.5374,
+      "loss": 0.7402,
       "step": 2210
     },
     {
-      "epoch": 2.1742125984251968,
-      "grad_norm": 6.299219131469727,
+      "epoch": 2.1761811023622046,
+      "grad_norm": 12.434174537658691,
       "learning_rate": 3.6615904350308343e-07,
-      "loss": 0.2392,
+      "loss": 0.617,
       "step": 2211
     },
     {
-      "epoch": 2.175196850393701,
-      "grad_norm": 6.071145534515381,
+      "epoch": 2.177165354330709,
+      "grad_norm": 11.357221603393555,
       "learning_rate": 3.6542141323514666e-07,
-      "loss": 0.2405,
+      "loss": 0.5497,
       "step": 2212
     },
     {
-      "epoch": 2.1761811023622046,
-      "grad_norm": 15.841425895690918,
+      "epoch": 2.1781496062992125,
+      "grad_norm": 11.165534019470215,
       "learning_rate": 3.6468410545224436e-07,
-      "loss": 0.8154,
+      "loss": 0.5301,
       "step": 2213
     },
     {
-      "epoch": 2.177165354330709,
-      "grad_norm": 11.276870727539062,
+      "epoch": 2.1791338582677167,
+      "grad_norm": 7.299598217010498,
       "learning_rate": 3.6394712189954365e-07,
-      "loss": 0.6533,
+      "loss": 0.2226,
       "step": 2214
     },
     {
-      "epoch": 2.1781496062992125,
-      "grad_norm": 12.441000938415527,
+      "epoch": 2.1801181102362204,
+      "grad_norm": 10.696553230285645,
       "learning_rate": 3.632104643214442e-07,
-      "loss": 0.5278,
+      "loss": 0.2863,
       "step": 2215
     },
     {
-      "epoch": 2.1791338582677167,
-      "grad_norm": 14.340779304504395,
+      "epoch": 2.1811023622047245,
+      "grad_norm": 9.571880340576172,
       "learning_rate": 3.62474134461574e-07,
-      "loss": 0.907,
+      "loss": 0.2799,
       "step": 2216
     },
     {
-      "epoch": 2.1801181102362204,
-      "grad_norm": 6.19671106338501,
+      "epoch": 2.1820866141732282,
+      "grad_norm": 9.28178596496582,
       "learning_rate": 3.6173813406278543e-07,
-      "loss": 0.1621,
+      "loss": 0.2728,
       "step": 2217
     },
     {
-      "epoch": 2.1811023622047245,
-      "grad_norm": 10.040095329284668,
+      "epoch": 2.1830708661417324,
+      "grad_norm": 10.238334655761719,
       "learning_rate": 3.610024648671511e-07,
-      "loss": 0.4738,
+      "loss": 0.4322,
       "step": 2218
     },
     {
-      "epoch": 2.1820866141732282,
-      "grad_norm": 16.7701416015625,
+      "epoch": 2.184055118110236,
+      "grad_norm": 9.688760757446289,
       "learning_rate": 3.602671286159594e-07,
-      "loss": 0.8227,
+      "loss": 0.6777,
       "step": 2219
     },
     {
-      "epoch": 2.1830708661417324,
-      "grad_norm": 15.204830169677734,
+      "epoch": 2.1850393700787403,
+      "grad_norm": 13.491595268249512,
       "learning_rate": 3.5953212704971096e-07,
-      "loss": 0.668,
+      "loss": 0.6416,
       "step": 2220
     },
     {
-      "epoch": 2.184055118110236,
-      "grad_norm": 10.306490898132324,
+      "epoch": 2.186023622047244,
+      "grad_norm": 13.086960792541504,
       "learning_rate": 3.5879746190811415e-07,
-      "loss": 0.5098,
+      "loss": 0.6656,
       "step": 2221
     },
     {
-      "epoch": 2.1850393700787403,
-      "grad_norm": 6.649641513824463,
+      "epoch": 2.187007874015748,
+      "grad_norm": 11.892463684082031,
       "learning_rate": 3.5806313493008104e-07,
-      "loss": 0.2362,
+      "loss": 0.3136,
       "step": 2222
     },
     {
-      "epoch": 2.186023622047244,
-      "grad_norm": 13.032939910888672,
+      "epoch": 2.187992125984252,
+      "grad_norm": 11.732651710510254,
       "learning_rate": 3.573291478537231e-07,
-      "loss": 0.8769,
+      "loss": 0.7484,
       "step": 2223
     },
     {
-      "epoch": 2.187007874015748,
-      "grad_norm": 9.772322654724121,
+      "epoch": 2.188976377952756,
+      "grad_norm": 10.257271766662598,
       "learning_rate": 3.565955024163477e-07,
-      "loss": 0.479,
+      "loss": 0.6143,
       "step": 2224
     },
     {
-      "epoch": 2.187992125984252,
-      "grad_norm": 11.886706352233887,
+      "epoch": 2.1899606299212597,
+      "grad_norm": 10.35676383972168,
       "learning_rate": 3.5586220035445327e-07,
-      "loss": 0.6157,
+      "loss": 0.3557,
       "step": 2225
     },
     {
-      "epoch": 2.188976377952756,
-      "grad_norm": 8.725859642028809,
+      "epoch": 2.190944881889764,
+      "grad_norm": 10.930755615234375,
       "learning_rate": 3.551292434037254e-07,
-      "loss": 0.3198,
+      "loss": 0.4548,
       "step": 2226
     },
     {
-      "epoch": 2.1899606299212597,
-      "grad_norm": 6.983432292938232,
+      "epoch": 2.1919291338582676,
+      "grad_norm": 9.970959663391113,
       "learning_rate": 3.54396633299033e-07,
-      "loss": 0.2217,
+      "loss": 0.4907,
       "step": 2227
     },
     {
-      "epoch": 2.190944881889764,
-      "grad_norm": 10.190155982971191,
+      "epoch": 2.1929133858267718,
+      "grad_norm": 9.224796295166016,
       "learning_rate": 3.5366437177442394e-07,
-      "loss": 0.5194,
+      "loss": 0.2768,
       "step": 2228
     },
     {
-      "epoch": 2.1919291338582676,
-      "grad_norm": 10.641023635864258,
+      "epoch": 2.1938976377952755,
+      "grad_norm": 12.519915580749512,
       "learning_rate": 3.529324605631211e-07,
-      "loss": 0.413,
+      "loss": 0.57,
       "step": 2229
     },
     {
-      "epoch": 2.1929133858267718,
-      "grad_norm": 7.027321815490723,
+      "epoch": 2.1948818897637796,
+      "grad_norm": 18.26183319091797,
       "learning_rate": 3.522009013975184e-07,
-      "loss": 0.242,
+      "loss": 0.86,
       "step": 2230
     },
     {
-      "epoch": 2.1938976377952755,
-      "grad_norm": 11.656682968139648,
+      "epoch": 2.1958661417322833,
+      "grad_norm": 13.520648956298828,
       "learning_rate": 3.514696960091756e-07,
-      "loss": 0.4697,
+      "loss": 0.6951,
       "step": 2231
     },
     {
-      "epoch": 2.1948818897637796,
-      "grad_norm": 8.944205284118652,
+      "epoch": 2.1968503937007875,
+      "grad_norm": 7.12697172164917,
       "learning_rate": 3.5073884612881616e-07,
-      "loss": 0.4982,
+      "loss": 0.2213,
       "step": 2232
     },
     {
-      "epoch": 2.1958661417322833,
-      "grad_norm": 9.124181747436523,
+      "epoch": 2.197834645669291,
+      "grad_norm": 11.289865493774414,
       "learning_rate": 3.500083534863215e-07,
-      "loss": 0.4052,
+      "loss": 0.3834,
       "step": 2233
     },
     {
-      "epoch": 2.1968503937007875,
-      "grad_norm": 5.858971118927002,
+      "epoch": 2.1988188976377954,
+      "grad_norm": 13.221308708190918,
       "learning_rate": 3.492782198107275e-07,
-      "loss": 0.1339,
+      "loss": 0.5838,
       "step": 2234
     },
     {
-      "epoch": 2.197834645669291,
-      "grad_norm": 11.764811515808105,
+      "epoch": 2.199803149606299,
+      "grad_norm": 12.246597290039062,
       "learning_rate": 3.485484468302207e-07,
-      "loss": 0.488,
+      "loss": 0.6818,
       "step": 2235
     },
     {
-      "epoch": 2.1988188976377954,
-      "grad_norm": 14.925984382629395,
+      "epoch": 2.2007874015748032,
+      "grad_norm": 11.04542064666748,
       "learning_rate": 3.4781903627213347e-07,
-      "loss": 0.7147,
+      "loss": 0.7577,
       "step": 2236
     },
     {
-      "epoch": 2.199803149606299,
-      "grad_norm": 10.232291221618652,
+      "epoch": 2.201771653543307,
+      "grad_norm": 9.378790855407715,
       "learning_rate": 3.470899898629408e-07,
-      "loss": 0.6582,
+      "loss": 0.4476,
       "step": 2237
     },
     {
-      "epoch": 2.2007874015748032,
-      "grad_norm": 12.924386024475098,
+      "epoch": 2.202755905511811,
+      "grad_norm": 9.48752212524414,
       "learning_rate": 3.463613093282552e-07,
-      "loss": 0.7598,
+      "loss": 0.2899,
       "step": 2238
     },
     {
-      "epoch": 2.201771653543307,
-      "grad_norm": 9.899399757385254,
+      "epoch": 2.203740157480315,
+      "grad_norm": 10.349443435668945,
       "learning_rate": 3.456329963928237e-07,
-      "loss": 0.4642,
+      "loss": 0.4989,
       "step": 2239
     },
     {
-      "epoch": 2.202755905511811,
-      "grad_norm": 9.124329566955566,
+      "epoch": 2.204724409448819,
+      "grad_norm": 14.239435195922852,
       "learning_rate": 3.4490505278052295e-07,
-      "loss": 0.4712,
+      "loss": 0.7162,
       "step": 2240
     },
     {
-      "epoch": 2.203740157480315,
-      "grad_norm": 7.940446376800537,
+      "epoch": 2.2057086614173227,
+      "grad_norm": 8.705245971679688,
       "learning_rate": 3.4417748021435563e-07,
-      "loss": 0.3164,
+      "loss": 0.4761,
       "step": 2241
     },
     {
-      "epoch": 2.204724409448819,
-      "grad_norm": 9.96937370300293,
+      "epoch": 2.206692913385827,
+      "grad_norm": 11.865876197814941,
       "learning_rate": 3.4345028041644595e-07,
-      "loss": 0.4444,
+      "loss": 0.4957,
       "step": 2242
     },
     {
-      "epoch": 2.2057086614173227,
-      "grad_norm": 9.700082778930664,
+      "epoch": 2.2076771653543306,
+      "grad_norm": 12.22436237335205,
       "learning_rate": 3.427234551080358e-07,
-      "loss": 0.4983,
+      "loss": 0.7497,
       "step": 2243
     },
     {
-      "epoch": 2.206692913385827,
-      "grad_norm": 14.01784896850586,
+      "epoch": 2.2086614173228347,
+      "grad_norm": 8.873125076293945,
       "learning_rate": 3.41997006009481e-07,
-      "loss": 0.5059,
+      "loss": 0.3968,
       "step": 2244
     },
     {
-      "epoch": 2.2076771653543306,
-      "grad_norm": 19.83002281188965,
+      "epoch": 2.2096456692913384,
+      "grad_norm": 13.000483512878418,
       "learning_rate": 3.4127093484024656e-07,
-      "loss": 1.0876,
+      "loss": 0.7304,
       "step": 2245
     },
     {
-      "epoch": 2.2086614173228347,
-      "grad_norm": 5.920734882354736,
+      "epoch": 2.2106299212598426,
+      "grad_norm": 8.905461311340332,
       "learning_rate": 3.4054524331890316e-07,
-      "loss": 0.1743,
+      "loss": 0.3713,
       "step": 2246
     },
     {
-      "epoch": 2.2096456692913384,
-      "grad_norm": 11.091670036315918,
+      "epoch": 2.2116141732283463,
+      "grad_norm": 10.071168899536133,
       "learning_rate": 3.3981993316312277e-07,
-      "loss": 0.7035,
+      "loss": 0.3308,
       "step": 2247
     },
     {
-      "epoch": 2.2106299212598426,
-      "grad_norm": 8.12314224243164,
+      "epoch": 2.2125984251968505,
+      "grad_norm": 14.614319801330566,
       "learning_rate": 3.390950060896746e-07,
-      "loss": 0.2316,
+      "loss": 0.581,
       "step": 2248
     },
     {
-      "epoch": 2.2116141732283463,
-      "grad_norm": 4.970713138580322,
+      "epoch": 2.213582677165354,
+      "grad_norm": 5.161202430725098,
       "learning_rate": 3.3837046381442135e-07,
-      "loss": 0.1159,
+      "loss": 0.1899,
       "step": 2249
     },
     {
-      "epoch": 2.2125984251968505,
-      "grad_norm": 8.45492172241211,
+      "epoch": 2.2145669291338583,
+      "grad_norm": 9.152812957763672,
       "learning_rate": 3.376463080523147e-07,
-      "loss": 0.3093,
+      "loss": 0.4058,
       "step": 2250
     },
     {
-      "epoch": 2.213582677165354,
-      "grad_norm": 8.577178955078125,
+      "epoch": 2.215551181102362,
+      "grad_norm": 13.240073204040527,
       "learning_rate": 3.369225405173918e-07,
-      "loss": 0.2272,
+      "loss": 0.686,
       "step": 2251
     },
     {
-      "epoch": 2.2145669291338583,
-      "grad_norm": 13.585640907287598,
+      "epoch": 2.216535433070866,
+      "grad_norm": 11.227309226989746,
       "learning_rate": 3.361991629227704e-07,
-      "loss": 0.8034,
+      "loss": 0.348,
       "step": 2252
     },
     {
-      "epoch": 2.215551181102362,
-      "grad_norm": 11.29826831817627,
+      "epoch": 2.21751968503937,
+      "grad_norm": 7.6174702644348145,
       "learning_rate": 3.354761769806456e-07,
-      "loss": 0.4926,
+      "loss": 0.3332,
       "step": 2253
     },
     {
-      "epoch": 2.216535433070866,
-      "grad_norm": 8.566338539123535,
+      "epoch": 2.218503937007874,
+      "grad_norm": 10.43615436553955,
       "learning_rate": 3.3475358440228547e-07,
-      "loss": 0.2633,
+      "loss": 0.5503,
       "step": 2254
     },
     {
-      "epoch": 2.21751968503937,
-      "grad_norm": 7.11183500289917,
+      "epoch": 2.219488188976378,
+      "grad_norm": 8.086403846740723,
       "learning_rate": 3.340313868980271e-07,
-      "loss": 0.2974,
+      "loss": 0.2385,
       "step": 2255
     },
     {
-      "epoch": 2.218503937007874,
-      "grad_norm": 12.137798309326172,
+      "epoch": 2.220472440944882,
+      "grad_norm": 12.876132011413574,
       "learning_rate": 3.3330958617727213e-07,
-      "loss": 0.6339,
+      "loss": 0.8728,
       "step": 2256
     },
     {
-      "epoch": 2.219488188976378,
-      "grad_norm": 13.24455738067627,
+      "epoch": 2.2214566929133857,
+      "grad_norm": 12.77263355255127,
       "learning_rate": 3.325881839484833e-07,
-      "loss": 0.4699,
+      "loss": 0.7366,
       "step": 2257
     },
     {
-      "epoch": 2.220472440944882,
-      "grad_norm": 10.560152053833008,
+      "epoch": 2.22244094488189,
+      "grad_norm": 9.329802513122559,
       "learning_rate": 3.318671819191802e-07,
-      "loss": 0.5012,
+      "loss": 0.3359,
       "step": 2258
     },
     {
-      "epoch": 2.2214566929133857,
-      "grad_norm": 13.34184741973877,
+      "epoch": 2.2234251968503935,
+      "grad_norm": 10.418845176696777,
       "learning_rate": 3.3114658179593487e-07,
-      "loss": 0.5572,
+      "loss": 0.6247,
       "step": 2259
     },
     {
-      "epoch": 2.22244094488189,
-      "grad_norm": 16.4458065032959,
+      "epoch": 2.2244094488188977,
+      "grad_norm": 10.72793197631836,
       "learning_rate": 3.304263852843683e-07,
-      "loss": 0.9996,
+      "loss": 0.501,
       "step": 2260
     },
     {
-      "epoch": 2.2234251968503935,
-      "grad_norm": 11.808619499206543,
+      "epoch": 2.2253937007874014,
+      "grad_norm": 12.984864234924316,
       "learning_rate": 3.297065940891462e-07,
-      "loss": 0.7209,
+      "loss": 0.6602,
       "step": 2261
     },
     {
-      "epoch": 2.2244094488188977,
-      "grad_norm": 9.634578704833984,
+      "epoch": 2.2263779527559056,
+      "grad_norm": 12.649776458740234,
       "learning_rate": 3.289872099139746e-07,
-      "loss": 0.3005,
+      "loss": 0.4829,
       "step": 2262
     },
     {
-      "epoch": 2.2253937007874014,
-      "grad_norm": 11.289079666137695,
+      "epoch": 2.2273622047244093,
+      "grad_norm": 6.961759567260742,
       "learning_rate": 3.282682344615966e-07,
-      "loss": 0.7379,
+      "loss": 0.1947,
       "step": 2263
     },
     {
-      "epoch": 2.2263779527559056,
-      "grad_norm": 5.66681432723999,
+      "epoch": 2.2283464566929134,
+      "grad_norm": 8.726760864257812,
       "learning_rate": 3.275496694337874e-07,
-      "loss": 0.1233,
+      "loss": 0.2411,
       "step": 2264
     },
     {
-      "epoch": 2.2273622047244093,
-      "grad_norm": 10.479454040527344,
+      "epoch": 2.229330708661417,
+      "grad_norm": 9.40872859954834,
       "learning_rate": 3.2683151653135104e-07,
-      "loss": 0.4234,
+      "loss": 0.362,
       "step": 2265
     },
     {
-      "epoch": 2.2283464566929134,
-      "grad_norm": 8.537384986877441,
+      "epoch": 2.2303149606299213,
+      "grad_norm": 8.935779571533203,
       "learning_rate": 3.261137774541161e-07,
-      "loss": 0.2912,
+      "loss": 0.3329,
       "step": 2266
     },
     {
-      "epoch": 2.229330708661417,
-      "grad_norm": 14.457018852233887,
+      "epoch": 2.231299212598425,
+      "grad_norm": 8.23385238647461,
       "learning_rate": 3.2539645390093155e-07,
-      "loss": 0.5686,
+      "loss": 0.2588,
       "step": 2267
     },
     {
-      "epoch": 2.2303149606299213,
-      "grad_norm": 8.997966766357422,
+      "epoch": 2.232283464566929,
+      "grad_norm": 10.827224731445312,
       "learning_rate": 3.2467954756966276e-07,
-      "loss": 0.3039,
+      "loss": 0.4144,
       "step": 2268
     },
     {
-      "epoch": 2.231299212598425,
-      "grad_norm": 9.181035995483398,
+      "epoch": 2.233267716535433,
+      "grad_norm": 3.3559486865997314,
       "learning_rate": 3.2396306015718775e-07,
-      "loss": 0.3685,
+      "loss": 0.0592,
       "step": 2269
     },
     {
-      "epoch": 2.232283464566929,
-      "grad_norm": 10.959527969360352,
+      "epoch": 2.234251968503937,
+      "grad_norm": 15.330273628234863,
       "learning_rate": 3.2324699335939273e-07,
-      "loss": 0.4984,
+      "loss": 0.5764,
       "step": 2270
     },
     {
-      "epoch": 2.233267716535433,
-      "grad_norm": 6.941189765930176,
+      "epoch": 2.2352362204724407,
+      "grad_norm": 10.293482780456543,
       "learning_rate": 3.225313488711686e-07,
-      "loss": 0.2326,
+      "loss": 0.4594,
       "step": 2271
     },
     {
-      "epoch": 2.234251968503937,
-      "grad_norm": 13.026391983032227,
+      "epoch": 2.236220472440945,
+      "grad_norm": 8.733466148376465,
       "learning_rate": 3.2181612838640673e-07,
-      "loss": 0.5876,
+      "loss": 0.3137,
       "step": 2272
     },
     {
-      "epoch": 2.2352362204724407,
-      "grad_norm": 10.644556999206543,
+      "epoch": 2.2372047244094486,
+      "grad_norm": 7.797464847564697,
       "learning_rate": 3.2110133359799457e-07,
-      "loss": 0.4002,
+      "loss": 0.2659,
       "step": 2273
     },
     {
-      "epoch": 2.236220472440945,
-      "grad_norm": 12.20460319519043,
+      "epoch": 2.2381889763779528,
+      "grad_norm": 10.667586326599121,
       "learning_rate": 3.203869661978121e-07,
-      "loss": 0.5853,
+      "loss": 0.4802,
       "step": 2274
     },
     {
-      "epoch": 2.2372047244094486,
-      "grad_norm": 8.994668006896973,
+      "epoch": 2.2391732283464565,
+      "grad_norm": 10.651825904846191,
       "learning_rate": 3.196730278767277e-07,
-      "loss": 0.2947,
+      "loss": 0.5744,
       "step": 2275
     },
     {
-      "epoch": 2.2381889763779528,
-      "grad_norm": 12.746809959411621,
+      "epoch": 2.2401574803149606,
+      "grad_norm": 11.770074844360352,
       "learning_rate": 3.189595203245944e-07,
-      "loss": 0.5385,
+      "loss": 0.5319,
       "step": 2276
     },
     {
-      "epoch": 2.2391732283464565,
-      "grad_norm": 8.311347961425781,
+      "epoch": 2.2411417322834644,
+      "grad_norm": 13.193926811218262,
       "learning_rate": 3.182464452302452e-07,
-      "loss": 0.3142,
+      "loss": 0.5163,
       "step": 2277
     },
     {
-      "epoch": 2.2401574803149606,
-      "grad_norm": 9.598487854003906,
+      "epoch": 2.2421259842519685,
+      "grad_norm": 12.137622833251953,
       "learning_rate": 3.175338042814898e-07,
-      "loss": 0.383,
+      "loss": 0.5062,
       "step": 2278
     },
     {
-      "epoch": 2.2411417322834644,
-      "grad_norm": 11.807975769042969,
+      "epoch": 2.2431102362204722,
+      "grad_norm": 9.126914024353027,
       "learning_rate": 3.168215991651101e-07,
-      "loss": 0.4688,
+      "loss": 0.4614,
       "step": 2279
     },
     {
-      "epoch": 2.2421259842519685,
-      "grad_norm": 8.984016418457031,
+      "epoch": 2.2440944881889764,
+      "grad_norm": 5.195245265960693,
       "learning_rate": 3.1610983156685664e-07,
-      "loss": 0.4124,
+      "loss": 0.0751,
       "step": 2280
     },
     {
-      "epoch": 2.2431102362204722,
-      "grad_norm": 15.54113483428955,
+      "epoch": 2.2450787401574805,
+      "grad_norm": 8.36727523803711,
       "learning_rate": 3.153985031714442e-07,
-      "loss": 0.538,
+      "loss": 0.3802,
       "step": 2281
     },
     {
-      "epoch": 2.2440944881889764,
-      "grad_norm": 12.840856552124023,
+      "epoch": 2.2460629921259843,
+      "grad_norm": 13.25330638885498,
       "learning_rate": 3.14687615662548e-07,
-      "loss": 0.6009,
+      "loss": 0.6284,
       "step": 2282
     },
     {
-      "epoch": 2.2450787401574805,
-      "grad_norm": 7.0070271492004395,
+      "epoch": 2.247047244094488,
+      "grad_norm": 11.463336944580078,
       "learning_rate": 3.139771707227997e-07,
-      "loss": 0.2344,
+      "loss": 0.6046,
       "step": 2283
     },
     {
-      "epoch": 2.2460629921259843,
-      "grad_norm": 10.138955116271973,
+      "epoch": 2.248031496062992,
+      "grad_norm": 11.328153610229492,
       "learning_rate": 3.1326717003378365e-07,
-      "loss": 0.3602,
+      "loss": 0.4421,
       "step": 2284
     },
     {
-      "epoch": 2.247047244094488,
-      "grad_norm": 11.75495719909668,
+      "epoch": 2.2490157480314963,
+      "grad_norm": 8.789440155029297,
       "learning_rate": 3.1255761527603233e-07,
-      "loss": 0.7008,
+      "loss": 0.4261,
       "step": 2285
     },
     {
-      "epoch": 2.248031496062992,
-      "grad_norm": 10.067399978637695,
+      "epoch": 2.25,
+      "grad_norm": 3.675967216491699,
       "learning_rate": 3.1184850812902304e-07,
-      "loss": 0.2619,
+      "loss": 0.0917,
       "step": 2286
     },
     {
-      "epoch": 2.2490157480314963,
-      "grad_norm": 13.368266105651855,
+      "epoch": 2.2509842519685037,
+      "grad_norm": 10.393106460571289,
       "learning_rate": 3.111398502711733e-07,
-      "loss": 0.5225,
+      "loss": 0.5027,
       "step": 2287
     },
     {
-      "epoch": 2.25,
-      "grad_norm": 5.223156452178955,
+      "epoch": 2.251968503937008,
+      "grad_norm": 8.672551155090332,
       "learning_rate": 3.104316433798374e-07,
-      "loss": 0.1657,
+      "loss": 0.3014,
       "step": 2288
     },
     {
-      "epoch": 2.2509842519685037,
-      "grad_norm": 10.531564712524414,
+      "epoch": 2.252952755905512,
+      "grad_norm": 14.485270500183105,
       "learning_rate": 3.0972388913130215e-07,
-      "loss": 0.3965,
+      "loss": 0.9846,
       "step": 2289
     },
     {
-      "epoch": 2.251968503937008,
-      "grad_norm": 14.430265426635742,
+      "epoch": 2.2539370078740157,
+      "grad_norm": 8.181056022644043,
       "learning_rate": 3.0901658920078314e-07,
-      "loss": 0.8387,
+      "loss": 0.3314,
       "step": 2290
     },
     {
-      "epoch": 2.252952755905512,
-      "grad_norm": 13.575235366821289,
+      "epoch": 2.2549212598425195,
+      "grad_norm": 9.592098236083984,
       "learning_rate": 3.083097452624203e-07,
-      "loss": 0.6505,
+      "loss": 0.2861,
       "step": 2291
     },
     {
-      "epoch": 2.2539370078740157,
-      "grad_norm": 15.854157447814941,
+      "epoch": 2.2559055118110236,
+      "grad_norm": 16.158761978149414,
       "learning_rate": 3.076033589892745e-07,
-      "loss": 0.8453,
+      "loss": 1.0061,
       "step": 2292
     },
     {
-      "epoch": 2.2549212598425195,
-      "grad_norm": 10.565400123596191,
+      "epoch": 2.2568897637795278,
+      "grad_norm": 2.54319429397583,
       "learning_rate": 3.0689743205332345e-07,
-      "loss": 0.3865,
+      "loss": 0.0463,
       "step": 2293
     },
     {
-      "epoch": 2.2559055118110236,
-      "grad_norm": 11.114446640014648,
+      "epoch": 2.2578740157480315,
+      "grad_norm": 6.582190990447998,
       "learning_rate": 3.061919661254571e-07,
-      "loss": 0.7424,
+      "loss": 0.1619,
       "step": 2294
     },
     {
-      "epoch": 2.2568897637795278,
-      "grad_norm": 9.461173057556152,
+      "epoch": 2.258858267716535,
+      "grad_norm": 11.633143424987793,
       "learning_rate": 3.054869628754748e-07,
-      "loss": 0.2301,
+      "loss": 0.6535,
       "step": 2295
     },
     {
-      "epoch": 2.2568897637795278,
-      "eval_Qnli-dev_cosine_accuracy": 0.67578125,
-      "eval_Qnli-dev_cosine_accuracy_threshold": 0.8218872547149658,
-      "eval_Qnli-dev_cosine_ap": 0.6982356648607996,
-      "eval_Qnli-dev_cosine_f1": 0.6914378029079159,
-      "eval_Qnli-dev_cosine_f1_threshold": 0.7249837517738342,
-      "eval_Qnli-dev_cosine_precision": 0.5587467362924282,
-      "eval_Qnli-dev_cosine_recall": 0.9067796610169492,
-      "eval_Qnli-dev_dot_accuracy": 0.681640625,
-      "eval_Qnli-dev_dot_accuracy_threshold": 619.56884765625,
-      "eval_Qnli-dev_dot_ap": 0.694508632045775,
-      "eval_Qnli-dev_dot_f1": 0.6903225806451613,
-      "eval_Qnli-dev_dot_f1_threshold": 551.586669921875,
-      "eval_Qnli-dev_dot_precision": 0.5572916666666666,
-      "eval_Qnli-dev_dot_recall": 0.9067796610169492,
-      "eval_Qnli-dev_euclidean_accuracy": 0.677734375,
-      "eval_Qnli-dev_euclidean_accuracy_threshold": 16.517976760864258,
-      "eval_Qnli-dev_euclidean_ap": 0.698306251672151,
-      "eval_Qnli-dev_euclidean_f1": 0.6914378029079159,
-      "eval_Qnli-dev_euclidean_f1_threshold": 20.45012664794922,
-      "eval_Qnli-dev_euclidean_precision": 0.5587467362924282,
-      "eval_Qnli-dev_euclidean_recall": 0.9067796610169492,
+      "epoch": 2.258858267716535,
+      "eval_Qnli-dev_cosine_accuracy": 0.681640625,
+      "eval_Qnli-dev_cosine_accuracy_threshold": 0.7885782718658447,
+      "eval_Qnli-dev_cosine_ap": 0.7002236719885953,
+      "eval_Qnli-dev_cosine_f1": 0.6894197952218429,
+      "eval_Qnli-dev_cosine_f1_threshold": 0.7466418743133545,
+      "eval_Qnli-dev_cosine_precision": 0.5771428571428572,
+      "eval_Qnli-dev_cosine_recall": 0.8559322033898306,
+      "eval_Qnli-dev_dot_accuracy": 0.6796875,
+      "eval_Qnli-dev_dot_accuracy_threshold": 598.7392578125,
+      "eval_Qnli-dev_dot_ap": 0.6980666596688423,
+      "eval_Qnli-dev_dot_f1": 0.6869712351945855,
+      "eval_Qnli-dev_dot_f1_threshold": 566.3282470703125,
+      "eval_Qnli-dev_dot_precision": 0.571830985915493,
+      "eval_Qnli-dev_dot_recall": 0.8601694915254238,
+      "eval_Qnli-dev_euclidean_accuracy": 0.68359375,
+      "eval_Qnli-dev_euclidean_accuracy_threshold": 18.127986907958984,
+      "eval_Qnli-dev_euclidean_ap": 0.7003054512799944,
+      "eval_Qnli-dev_euclidean_f1": 0.6903553299492386,
+      "eval_Qnli-dev_euclidean_f1_threshold": 19.754405975341797,
+      "eval_Qnli-dev_euclidean_precision": 0.5746478873239437,
+      "eval_Qnli-dev_euclidean_recall": 0.864406779661017,
       "eval_Qnli-dev_manhattan_accuracy": 0.685546875,
-      "eval_Qnli-dev_manhattan_accuracy_threshold": 360.5780029296875,
-      "eval_Qnli-dev_manhattan_ap": 0.709700437823957,
-      "eval_Qnli-dev_manhattan_f1": 0.6902086677367576,
-      "eval_Qnli-dev_manhattan_f1_threshold": 433.43255615234375,
-      "eval_Qnli-dev_manhattan_precision": 0.5555555555555556,
-      "eval_Qnli-dev_manhattan_recall": 0.9110169491525424,
+      "eval_Qnli-dev_manhattan_accuracy_threshold": 369.89447021484375,
+      "eval_Qnli-dev_manhattan_ap": 0.7083138482854554,
+      "eval_Qnli-dev_manhattan_f1": 0.6902356902356903,
+      "eval_Qnli-dev_manhattan_f1_threshold": 421.6947021484375,
+      "eval_Qnli-dev_manhattan_precision": 0.5726256983240223,
+      "eval_Qnli-dev_manhattan_recall": 0.8686440677966102,
       "eval_Qnli-dev_max_accuracy": 0.685546875,
-      "eval_Qnli-dev_max_accuracy_threshold": 619.56884765625,
-      "eval_Qnli-dev_max_ap": 0.709700437823957,
-      "eval_Qnli-dev_max_f1": 0.6914378029079159,
-      "eval_Qnli-dev_max_f1_threshold": 551.586669921875,
-      "eval_Qnli-dev_max_precision": 0.5587467362924282,
-      "eval_Qnli-dev_max_recall": 0.9110169491525424,
-      "eval_allNLI-dev_cosine_accuracy": 0.693359375,
-      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9163593053817749,
-      "eval_allNLI-dev_cosine_ap": 0.5156280201012031,
-      "eval_allNLI-dev_cosine_f1": 0.5478424015009381,
-      "eval_allNLI-dev_cosine_f1_threshold": 0.7880935668945312,
-      "eval_allNLI-dev_cosine_precision": 0.40555555555555556,
-      "eval_allNLI-dev_cosine_recall": 0.8439306358381503,
-      "eval_allNLI-dev_dot_accuracy": 0.693359375,
-      "eval_allNLI-dev_dot_accuracy_threshold": 698.921142578125,
-      "eval_allNLI-dev_dot_ap": 0.5147433660036644,
-      "eval_allNLI-dev_dot_f1": 0.5457943925233645,
-      "eval_allNLI-dev_dot_f1_threshold": 599.6630859375,
-      "eval_allNLI-dev_dot_precision": 0.40331491712707185,
-      "eval_allNLI-dev_dot_recall": 0.8439306358381503,
-      "eval_allNLI-dev_euclidean_accuracy": 0.693359375,
-      "eval_allNLI-dev_euclidean_accuracy_threshold": 11.29690933227539,
-      "eval_allNLI-dev_euclidean_ap": 0.5155161734618426,
-      "eval_allNLI-dev_euclidean_f1": 0.5478424015009381,
-      "eval_allNLI-dev_euclidean_f1_threshold": 17.960678100585938,
-      "eval_allNLI-dev_euclidean_precision": 0.40555555555555556,
-      "eval_allNLI-dev_euclidean_recall": 0.8439306358381503,
-      "eval_allNLI-dev_manhattan_accuracy": 0.69140625,
-      "eval_allNLI-dev_manhattan_accuracy_threshold": 247.3362274169922,
-      "eval_allNLI-dev_manhattan_ap": 0.5162044992927799,
-      "eval_allNLI-dev_manhattan_f1": 0.5513307984790875,
-      "eval_allNLI-dev_manhattan_f1_threshold": 376.1918640136719,
-      "eval_allNLI-dev_manhattan_precision": 0.41076487252124644,
-      "eval_allNLI-dev_manhattan_recall": 0.838150289017341,
-      "eval_allNLI-dev_max_accuracy": 0.693359375,
-      "eval_allNLI-dev_max_accuracy_threshold": 698.921142578125,
-      "eval_allNLI-dev_max_ap": 0.5162044992927799,
-      "eval_allNLI-dev_max_f1": 0.5513307984790875,
-      "eval_allNLI-dev_max_f1_threshold": 599.6630859375,
-      "eval_allNLI-dev_max_precision": 0.41076487252124644,
-      "eval_allNLI-dev_max_recall": 0.8439306358381503,
-      "eval_loss": 0.9295917749404907,
-      "eval_runtime": 54.1496,
-      "eval_samples_per_second": 30.73,
-      "eval_sequential_score": 0.709700437823957,
-      "eval_steps_per_second": 0.129,
-      "eval_sts-test_pearson_cosine": 0.6552902390696351,
-      "eval_sts-test_pearson_dot": 0.6546625093787101,
-      "eval_sts-test_pearson_euclidean": 0.6686392487670392,
-      "eval_sts-test_pearson_manhattan": 0.6720928689273424,
-      "eval_sts-test_pearson_max": 0.6720928689273424,
-      "eval_sts-test_spearman_cosine": 0.6644033613533696,
-      "eval_sts-test_spearman_dot": 0.6645467891461292,
-      "eval_sts-test_spearman_euclidean": 0.6643764210940122,
-      "eval_sts-test_spearman_manhattan": 0.6672981861989953,
-      "eval_sts-test_spearman_max": 0.6672981861989953,
+      "eval_Qnli-dev_max_accuracy_threshold": 598.7392578125,
+      "eval_Qnli-dev_max_ap": 0.7083138482854554,
+      "eval_Qnli-dev_max_f1": 0.6903553299492386,
+      "eval_Qnli-dev_max_f1_threshold": 566.3282470703125,
+      "eval_Qnli-dev_max_precision": 0.5771428571428572,
+      "eval_Qnli-dev_max_recall": 0.8686440677966102,
+      "eval_allNLI-dev_cosine_accuracy": 0.697265625,
+      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9179866909980774,
+      "eval_allNLI-dev_cosine_ap": 0.5023677428864256,
+      "eval_allNLI-dev_cosine_f1": 0.5533769063180829,
+      "eval_allNLI-dev_cosine_f1_threshold": 0.8282407522201538,
+      "eval_allNLI-dev_cosine_precision": 0.44405594405594406,
+      "eval_allNLI-dev_cosine_recall": 0.7341040462427746,
+      "eval_allNLI-dev_dot_accuracy": 0.6953125,
+      "eval_allNLI-dev_dot_accuracy_threshold": 702.9041748046875,
+      "eval_allNLI-dev_dot_ap": 0.5005776281820205,
+      "eval_allNLI-dev_dot_f1": 0.5533769063180829,
+      "eval_allNLI-dev_dot_f1_threshold": 631.9176635742188,
+      "eval_allNLI-dev_dot_precision": 0.44405594405594406,
+      "eval_allNLI-dev_dot_recall": 0.7341040462427746,
+      "eval_allNLI-dev_euclidean_accuracy": 0.697265625,
+      "eval_allNLI-dev_euclidean_accuracy_threshold": 11.191230773925781,
+      "eval_allNLI-dev_euclidean_ap": 0.5028917478226242,
+      "eval_allNLI-dev_euclidean_f1": 0.5533769063180829,
+      "eval_allNLI-dev_euclidean_f1_threshold": 16.181026458740234,
+      "eval_allNLI-dev_euclidean_precision": 0.44405594405594406,
+      "eval_allNLI-dev_euclidean_recall": 0.7341040462427746,
+      "eval_allNLI-dev_manhattan_accuracy": 0.689453125,
+      "eval_allNLI-dev_manhattan_accuracy_threshold": 237.93911743164062,
+      "eval_allNLI-dev_manhattan_ap": 0.502842113525952,
+      "eval_allNLI-dev_manhattan_f1": 0.5550660792951542,
+      "eval_allNLI-dev_manhattan_f1_threshold": 339.4214782714844,
+      "eval_allNLI-dev_manhattan_precision": 0.4483985765124555,
+      "eval_allNLI-dev_manhattan_recall": 0.7283236994219653,
+      "eval_allNLI-dev_max_accuracy": 0.697265625,
+      "eval_allNLI-dev_max_accuracy_threshold": 702.9041748046875,
+      "eval_allNLI-dev_max_ap": 0.5028917478226242,
+      "eval_allNLI-dev_max_f1": 0.5550660792951542,
+      "eval_allNLI-dev_max_f1_threshold": 631.9176635742188,
+      "eval_allNLI-dev_max_precision": 0.4483985765124555,
+      "eval_allNLI-dev_max_recall": 0.7341040462427746,
+      "eval_loss": 0.8167487382888794,
+      "eval_runtime": 52.9224,
+      "eval_samples_per_second": 31.442,
+      "eval_sequential_score": 0.7083138482854554,
+      "eval_steps_per_second": 0.132,
+      "eval_sts-test_pearson_cosine": 0.6631509770031139,
+      "eval_sts-test_pearson_dot": 0.6607555540270089,
+      "eval_sts-test_pearson_euclidean": 0.6756013936756754,
+      "eval_sts-test_pearson_manhattan": 0.6807913115534869,
+      "eval_sts-test_pearson_max": 0.6807913115534869,
+      "eval_sts-test_spearman_cosine": 0.6695385238467715,
+      "eval_sts-test_spearman_dot": 0.6666498299239565,
+      "eval_sts-test_spearman_euclidean": 0.6697092051577183,
+      "eval_sts-test_spearman_manhattan": 0.6746454482942351,
+      "eval_sts-test_spearman_max": 0.6746454482942351,
       "step": 2295
     },
     {
-      "epoch": 2.2578740157480315,
-      "grad_norm": 5.400763988494873,
+      "epoch": 2.2598425196850394,
+      "grad_norm": 7.000857353210449,
       "learning_rate": 3.047824239720804e-07,
-      "loss": 0.128,
+      "loss": 0.2378,
       "step": 2296
     },
     {
-      "epoch": 2.258858267716535,
-      "grad_norm": 14.74106502532959,
+      "epoch": 2.2608267716535435,
+      "grad_norm": 9.137283325195312,
       "learning_rate": 3.0407835108287885e-07,
-      "loss": 0.7932,
+      "loss": 0.4016,
       "step": 2297
     },
     {
-      "epoch": 2.2598425196850394,
-      "grad_norm": 9.304885864257812,
+      "epoch": 2.2618110236220472,
+      "grad_norm": 12.972397804260254,
       "learning_rate": 3.03374745874372e-07,
-      "loss": 0.3471,
+      "loss": 0.7567,
       "step": 2298
     },
     {
-      "epoch": 2.2608267716535435,
-      "grad_norm": 13.81615161895752,
+      "epoch": 2.262795275590551,
+      "grad_norm": 11.782989501953125,
       "learning_rate": 3.0267161001195493e-07,
-      "loss": 0.5931,
+      "loss": 0.4916,
       "step": 2299
     },
     {
-      "epoch": 2.2618110236220472,
-      "grad_norm": 12.656591415405273,
+      "epoch": 2.263779527559055,
+      "grad_norm": 15.624208450317383,
       "learning_rate": 3.019689451599114e-07,
-      "loss": 0.3816,
+      "loss": 0.6291,
       "step": 2300
     },
     {
-      "epoch": 2.262795275590551,
-      "grad_norm": 8.846150398254395,
+      "epoch": 2.2647637795275593,
+      "grad_norm": 12.483575820922852,
       "learning_rate": 3.0126675298141056e-07,
-      "loss": 0.2919,
+      "loss": 0.4626,
       "step": 2301
     },
     {
-      "epoch": 2.263779527559055,
-      "grad_norm": 5.267794609069824,
+      "epoch": 2.265748031496063,
+      "grad_norm": 12.61329174041748,
       "learning_rate": 3.005650351385029e-07,
-      "loss": 0.1231,
+      "loss": 0.6925,
       "step": 2302
     },
     {
-      "epoch": 2.2647637795275593,
-      "grad_norm": 9.467386245727539,
+      "epoch": 2.266732283464567,
+      "grad_norm": 7.855734348297119,
       "learning_rate": 2.998637932921159e-07,
-      "loss": 0.3372,
+      "loss": 0.2438,
       "step": 2303
     },
     {
-      "epoch": 2.265748031496063,
-      "grad_norm": 8.390028953552246,
+      "epoch": 2.267716535433071,
+      "grad_norm": 7.524454593658447,
       "learning_rate": 2.991630291020506e-07,
-      "loss": 0.2717,
+      "loss": 0.3313,
       "step": 2304
     },
     {
-      "epoch": 2.266732283464567,
-      "grad_norm": 9.141326904296875,
+      "epoch": 2.268700787401575,
+      "grad_norm": 13.126644134521484,
       "learning_rate": 2.9846274422697734e-07,
-      "loss": 0.2641,
+      "loss": 0.7253,
       "step": 2305
     },
     {
-      "epoch": 2.267716535433071,
-      "grad_norm": 7.371496200561523,
+      "epoch": 2.2696850393700787,
+      "grad_norm": 12.794411659240723,
       "learning_rate": 2.9776294032443184e-07,
-      "loss": 0.1901,
+      "loss": 0.7217,
       "step": 2306
     },
     {
-      "epoch": 2.268700787401575,
-      "grad_norm": 9.045897483825684,
+      "epoch": 2.270669291338583,
+      "grad_norm": 8.721311569213867,
       "learning_rate": 2.9706361905081164e-07,
-      "loss": 0.409,
+      "loss": 0.258,
       "step": 2307
     },
     {
-      "epoch": 2.2696850393700787,
-      "grad_norm": 18.923290252685547,
+      "epoch": 2.2716535433070866,
+      "grad_norm": 10.186891555786133,
       "learning_rate": 2.9636478206137174e-07,
-      "loss": 0.9497,
+      "loss": 0.4069,
       "step": 2308
     },
     {
-      "epoch": 2.270669291338583,
-      "grad_norm": 16.945598602294922,
+      "epoch": 2.2726377952755907,
+      "grad_norm": 10.59805965423584,
       "learning_rate": 2.95666431010221e-07,
-      "loss": 0.7417,
+      "loss": 0.4181,
       "step": 2309
     },
     {
-      "epoch": 2.2716535433070866,
-      "grad_norm": 8.197480201721191,
+      "epoch": 2.2736220472440944,
+      "grad_norm": 12.393839836120605,
       "learning_rate": 2.949685675503179e-07,
-      "loss": 0.3104,
+      "loss": 0.5408,
       "step": 2310
     },
     {
-      "epoch": 2.2726377952755907,
-      "grad_norm": 8.451308250427246,
+      "epoch": 2.2746062992125986,
+      "grad_norm": 9.25861644744873,
       "learning_rate": 2.9427119333346706e-07,
-      "loss": 0.3535,
+      "loss": 0.426,
       "step": 2311
     },
     {
-      "epoch": 2.2736220472440944,
-      "grad_norm": 9.730972290039062,
+      "epoch": 2.2755905511811023,
+      "grad_norm": 8.95864200592041,
       "learning_rate": 2.935743100103149e-07,
-      "loss": 0.2776,
+      "loss": 0.5295,
       "step": 2312
     },
     {
-      "epoch": 2.2746062992125986,
-      "grad_norm": 9.99947452545166,
+      "epoch": 2.2765748031496065,
+      "grad_norm": 8.777239799499512,
       "learning_rate": 2.9287791923034593e-07,
-      "loss": 0.4681,
+      "loss": 0.2723,
       "step": 2313
     },
     {
-      "epoch": 2.2755905511811023,
-      "grad_norm": 9.144072532653809,
+      "epoch": 2.27755905511811,
+      "grad_norm": 11.933242797851562,
       "learning_rate": 2.921820226418793e-07,
-      "loss": 0.3353,
+      "loss": 0.5554,
       "step": 2314
     },
     {
-      "epoch": 2.2765748031496065,
-      "grad_norm": 12.268770217895508,
+      "epoch": 2.2785433070866143,
+      "grad_norm": 8.490767478942871,
       "learning_rate": 2.9148662189206344e-07,
-      "loss": 0.3965,
+      "loss": 0.3267,
       "step": 2315
     },
     {
-      "epoch": 2.27755905511811,
-      "grad_norm": 11.276354789733887,
+      "epoch": 2.279527559055118,
+      "grad_norm": 10.99681282043457,
       "learning_rate": 2.907917186268739e-07,
-      "loss": 0.4975,
+      "loss": 0.6567,
       "step": 2316
     },
     {
-      "epoch": 2.2785433070866143,
-      "grad_norm": 12.892839431762695,
+      "epoch": 2.280511811023622,
+      "grad_norm": 11.940130233764648,
       "learning_rate": 2.900973144911088e-07,
-      "loss": 0.6786,
+      "loss": 0.44,
       "step": 2317
     },
     {
-      "epoch": 2.279527559055118,
-      "grad_norm": 6.2306342124938965,
+      "epoch": 2.281496062992126,
+      "grad_norm": 5.789243221282959,
       "learning_rate": 2.894034111283843e-07,
-      "loss": 0.3268,
+      "loss": 0.1803,
       "step": 2318
     },
     {
-      "epoch": 2.280511811023622,
-      "grad_norm": 10.548524856567383,
+      "epoch": 2.28248031496063,
+      "grad_norm": 9.700678825378418,
       "learning_rate": 2.8871001018113183e-07,
-      "loss": 0.3992,
+      "loss": 0.3852,
       "step": 2319
     },
     {
-      "epoch": 2.281496062992126,
-      "grad_norm": 7.72267484664917,
+      "epoch": 2.283464566929134,
+      "grad_norm": 10.271327018737793,
       "learning_rate": 2.880171132905932e-07,
-      "loss": 0.2394,
+      "loss": 0.6622,
       "step": 2320
     },
     {
-      "epoch": 2.28248031496063,
-      "grad_norm": 13.049686431884766,
+      "epoch": 2.284448818897638,
+      "grad_norm": 12.159454345703125,
       "learning_rate": 2.8732472209681733e-07,
-      "loss": 0.4223,
+      "loss": 0.6441,
       "step": 2321
     },
     {
-      "epoch": 2.283464566929134,
-      "grad_norm": 3.6191823482513428,
+      "epoch": 2.2854330708661417,
+      "grad_norm": 12.692157745361328,
       "learning_rate": 2.866328382386562e-07,
-      "loss": 0.096,
+      "loss": 0.7071,
       "step": 2322
     },
     {
-      "epoch": 2.284448818897638,
-      "grad_norm": 12.548807144165039,
+      "epoch": 2.286417322834646,
+      "grad_norm": 10.10477352142334,
       "learning_rate": 2.859414633537609e-07,
-      "loss": 0.6566,
+      "loss": 0.3546,
       "step": 2323
     },
     {
-      "epoch": 2.2854330708661417,
-      "grad_norm": 16.991758346557617,
+      "epoch": 2.2874015748031495,
+      "grad_norm": 6.0948591232299805,
       "learning_rate": 2.8525059907857786e-07,
-      "loss": 0.9653,
+      "loss": 0.2224,
       "step": 2324
     },
     {
-      "epoch": 2.286417322834646,
-      "grad_norm": 8.746471405029297,
+      "epoch": 2.2883858267716537,
+      "grad_norm": 10.294443130493164,
       "learning_rate": 2.8456024704834483e-07,
-      "loss": 0.4092,
+      "loss": 0.4985,
       "step": 2325
     },
     {
-      "epoch": 2.2874015748031495,
-      "grad_norm": 6.317548751831055,
+      "epoch": 2.2893700787401574,
+      "grad_norm": 12.439338684082031,
       "learning_rate": 2.8387040889708715e-07,
-      "loss": 0.1846,
+      "loss": 0.6903,
       "step": 2326
     },
     {
-      "epoch": 2.2883858267716537,
-      "grad_norm": 13.521329879760742,
+      "epoch": 2.2903543307086616,
+      "grad_norm": 7.961695194244385,
       "learning_rate": 2.8318108625761395e-07,
-      "loss": 0.6888,
+      "loss": 0.2445,
       "step": 2327
     },
     {
-      "epoch": 2.2893700787401574,
-      "grad_norm": 18.330942153930664,
+      "epoch": 2.2913385826771653,
+      "grad_norm": 11.010169982910156,
       "learning_rate": 2.8249228076151396e-07,
-      "loss": 0.7732,
+      "loss": 0.5646,
       "step": 2328
     },
     {
-      "epoch": 2.2903543307086616,
-      "grad_norm": 5.730623245239258,
+      "epoch": 2.2923228346456694,
+      "grad_norm": 9.052175521850586,
       "learning_rate": 2.81803994039152e-07,
-      "loss": 0.1397,
+      "loss": 0.4209,
       "step": 2329
     },
     {
-      "epoch": 2.2913385826771653,
-      "grad_norm": 9.373048782348633,
+      "epoch": 2.293307086614173,
+      "grad_norm": 9.717872619628906,
       "learning_rate": 2.8111622771966523e-07,
-      "loss": 0.5713,
+      "loss": 0.4754,
       "step": 2330
     },
     {
-      "epoch": 2.2923228346456694,
-      "grad_norm": 8.271077156066895,
+      "epoch": 2.2942913385826773,
+      "grad_norm": 8.112227439880371,
       "learning_rate": 2.804289834309583e-07,
-      "loss": 0.3005,
+      "loss": 0.2935,
       "step": 2331
     },
     {
-      "epoch": 2.293307086614173,
-      "grad_norm": 10.701303482055664,
+      "epoch": 2.295275590551181,
+      "grad_norm": 10.873849868774414,
       "learning_rate": 2.797422627997009e-07,
-      "loss": 0.5604,
+      "loss": 0.6359,
       "step": 2332
     },
     {
-      "epoch": 2.2942913385826773,
-      "grad_norm": 10.093025207519531,
+      "epoch": 2.296259842519685,
+      "grad_norm": 8.873044967651367,
       "learning_rate": 2.7905606745132307e-07,
-      "loss": 0.4826,
+      "loss": 0.3648,
       "step": 2333
     },
     {
-      "epoch": 2.295275590551181,
-      "grad_norm": 15.213040351867676,
+      "epoch": 2.297244094488189,
+      "grad_norm": 8.419299125671387,
       "learning_rate": 2.7837039901001155e-07,
-      "loss": 0.747,
+      "loss": 0.5006,
       "step": 2334
     },
     {
-      "epoch": 2.296259842519685,
-      "grad_norm": 10.684081077575684,
+      "epoch": 2.298228346456693,
+      "grad_norm": 6.847270965576172,
       "learning_rate": 2.776852590987061e-07,
-      "loss": 0.4473,
+      "loss": 0.2887,
       "step": 2335
     },
     {
-      "epoch": 2.297244094488189,
-      "grad_norm": 10.946547508239746,
+      "epoch": 2.2992125984251968,
+      "grad_norm": 6.187161445617676,
       "learning_rate": 2.7700064933909474e-07,
-      "loss": 0.5294,
+      "loss": 0.2313,
       "step": 2336
     },
     {
-      "epoch": 2.298228346456693,
-      "grad_norm": 9.167648315429688,
+      "epoch": 2.300196850393701,
+      "grad_norm": 12.033468246459961,
       "learning_rate": 2.7631657135161156e-07,
-      "loss": 0.439,
+      "loss": 0.5629,
       "step": 2337
     },
     {
-      "epoch": 2.2992125984251968,
-      "grad_norm": 7.027797222137451,
+      "epoch": 2.3011811023622046,
+      "grad_norm": 16.42943572998047,
       "learning_rate": 2.756330267554316e-07,
-      "loss": 0.2158,
+      "loss": 0.8398,
       "step": 2338
     },
     {
-      "epoch": 2.300196850393701,
-      "grad_norm": 9.155388832092285,
+      "epoch": 2.302165354330709,
+      "grad_norm": 8.346458435058594,
       "learning_rate": 2.749500171684673e-07,
-      "loss": 0.4406,
+      "loss": 0.3545,
       "step": 2339
     },
     {
-      "epoch": 2.3011811023622046,
-      "grad_norm": 7.610925197601318,
+      "epoch": 2.3031496062992125,
+      "grad_norm": 7.371239185333252,
       "learning_rate": 2.7426754420736485e-07,
-      "loss": 0.219,
+      "loss": 0.3029,
       "step": 2340
     },
     {
-      "epoch": 2.302165354330709,
-      "grad_norm": 12.610198974609375,
+      "epoch": 2.3041338582677167,
+      "grad_norm": 11.532629013061523,
       "learning_rate": 2.735856094875003e-07,
-      "loss": 0.4623,
+      "loss": 0.5024,
       "step": 2341
     },
     {
-      "epoch": 2.3031496062992125,
-      "grad_norm": 10.615156173706055,
+      "epoch": 2.3051181102362204,
+      "grad_norm": 9.633743286132812,
       "learning_rate": 2.7290421462297574e-07,
-      "loss": 0.2466,
+      "loss": 0.322,
       "step": 2342
     },
     {
-      "epoch": 2.3041338582677167,
-      "grad_norm": 12.127360343933105,
+      "epoch": 2.3061023622047245,
+      "grad_norm": 8.252344131469727,
       "learning_rate": 2.7222336122661535e-07,
-      "loss": 0.6863,
+      "loss": 0.3881,
       "step": 2343
     },
     {
-      "epoch": 2.3051181102362204,
-      "grad_norm": 9.351911544799805,
+      "epoch": 2.3070866141732282,
+      "grad_norm": 10.416007041931152,
       "learning_rate": 2.715430509099618e-07,
-      "loss": 0.3583,
+      "loss": 0.5629,
       "step": 2344
     },
     {
-      "epoch": 2.3061023622047245,
-      "grad_norm": 11.017844200134277,
+      "epoch": 2.3080708661417324,
+      "grad_norm": 5.372657299041748,
       "learning_rate": 2.7086328528327215e-07,
-      "loss": 0.5925,
+      "loss": 0.1312,
       "step": 2345
     },
     {
-      "epoch": 2.3070866141732282,
-      "grad_norm": 12.029109954833984,
+      "epoch": 2.309055118110236,
+      "grad_norm": 12.414799690246582,
       "learning_rate": 2.7018406595551434e-07,
-      "loss": 0.552,
+      "loss": 0.5804,
       "step": 2346
     },
     {
-      "epoch": 2.3080708661417324,
-      "grad_norm": 13.67851448059082,
+      "epoch": 2.3100393700787403,
+      "grad_norm": 8.874005317687988,
       "learning_rate": 2.695053945343633e-07,
-      "loss": 0.8135,
+      "loss": 0.2675,
       "step": 2347
     },
     {
-      "epoch": 2.309055118110236,
-      "grad_norm": 12.758377075195312,
+      "epoch": 2.311023622047244,
+      "grad_norm": 14.968682289123535,
       "learning_rate": 2.6882727262619696e-07,
-      "loss": 0.5165,
+      "loss": 0.5829,
       "step": 2348
     },
     {
-      "epoch": 2.3100393700787403,
-      "grad_norm": 9.64584732055664,
+      "epoch": 2.312007874015748,
+      "grad_norm": 11.825992584228516,
       "learning_rate": 2.681497018360927e-07,
-      "loss": 0.3632,
+      "loss": 0.5274,
       "step": 2349
     },
     {
-      "epoch": 2.311023622047244,
-      "grad_norm": 10.910879135131836,
+      "epoch": 2.312992125984252,
+      "grad_norm": 12.583436965942383,
       "learning_rate": 2.6747268376782334e-07,
-      "loss": 0.4364,
+      "loss": 0.5471,
       "step": 2350
     },
     {
-      "epoch": 2.312007874015748,
-      "grad_norm": 7.449069023132324,
+      "epoch": 2.313976377952756,
+      "grad_norm": 6.978092193603516,
       "learning_rate": 2.6679622002385354e-07,
-      "loss": 0.2048,
+      "loss": 0.1937,
       "step": 2351
     },
     {
-      "epoch": 2.312992125984252,
-      "grad_norm": 9.218055725097656,
+      "epoch": 2.3149606299212597,
+      "grad_norm": 17.644004821777344,
       "learning_rate": 2.6612031220533574e-07,
-      "loss": 0.5,
+      "loss": 1.4522,
       "step": 2352
     },
     {
-      "epoch": 2.313976377952756,
-      "grad_norm": 10.417744636535645,
+      "epoch": 2.315944881889764,
+      "grad_norm": 15.869028091430664,
       "learning_rate": 2.654449619121067e-07,
-      "loss": 0.4657,
+      "loss": 0.8113,
       "step": 2353
     },
     {
-      "epoch": 2.3149606299212597,
-      "grad_norm": 17.5401554107666,
+      "epoch": 2.3169291338582676,
+      "grad_norm": 16.225513458251953,
       "learning_rate": 2.647701707426835e-07,
-      "loss": 0.8455,
+      "loss": 0.6974,
       "step": 2354
     },
     {
-      "epoch": 2.315944881889764,
-      "grad_norm": 11.498607635498047,
+      "epoch": 2.3179133858267718,
+      "grad_norm": 9.975135803222656,
       "learning_rate": 2.6409594029425986e-07,
-      "loss": 0.703,
+      "loss": 0.3138,
       "step": 2355
     },
     {
-      "epoch": 2.3169291338582676,
-      "grad_norm": 10.016690254211426,
+      "epoch": 2.3188976377952755,
+      "grad_norm": 16.176620483398438,
       "learning_rate": 2.6342227216270205e-07,
-      "loss": 0.4548,
+      "loss": 0.8778,
       "step": 2356
     },
     {
-      "epoch": 2.3179133858267718,
-      "grad_norm": 8.504423141479492,
+      "epoch": 2.3198818897637796,
+      "grad_norm": 9.9556884765625,
       "learning_rate": 2.6274916794254554e-07,
-      "loss": 0.3097,
+      "loss": 0.4305,
       "step": 2357
     },
     {
-      "epoch": 2.3188976377952755,
-      "grad_norm": 7.540285110473633,
+      "epoch": 2.3208661417322833,
+      "grad_norm": 12.0952730178833,
       "learning_rate": 2.6207662922699125e-07,
-      "loss": 0.2965,
+      "loss": 0.6552,
       "step": 2358
     },
     {
-      "epoch": 2.3198818897637796,
-      "grad_norm": 8.403088569641113,
+      "epoch": 2.3218503937007875,
+      "grad_norm": 12.304922103881836,
       "learning_rate": 2.6140465760790126e-07,
-      "loss": 0.2303,
+      "loss": 0.5439,
       "step": 2359
     },
     {
-      "epoch": 2.3208661417322833,
-      "grad_norm": 8.636588096618652,
+      "epoch": 2.322834645669291,
+      "grad_norm": 6.5663042068481445,
       "learning_rate": 2.6073325467579553e-07,
-      "loss": 0.2937,
+      "loss": 0.1782,
       "step": 2360
     },
     {
-      "epoch": 2.3218503937007875,
-      "grad_norm": 10.3641996383667,
+      "epoch": 2.3238188976377954,
+      "grad_norm": 11.106794357299805,
       "learning_rate": 2.60062422019848e-07,
-      "loss": 0.6068,
+      "loss": 0.4607,
       "step": 2361
     },
     {
-      "epoch": 2.322834645669291,
-      "grad_norm": 5.669926166534424,
+      "epoch": 2.324803149606299,
+      "grad_norm": 9.776447296142578,
       "learning_rate": 2.5939216122788267e-07,
-      "loss": 0.169,
+      "loss": 0.5819,
       "step": 2362
     },
     {
-      "epoch": 2.3238188976377954,
-      "grad_norm": 6.79951810836792,
+      "epoch": 2.3257874015748032,
+      "grad_norm": 12.052081108093262,
       "learning_rate": 2.5872247388637e-07,
-      "loss": 0.1574,
+      "loss": 0.3315,
       "step": 2363
     },
     {
-      "epoch": 2.324803149606299,
-      "grad_norm": 8.884796142578125,
+      "epoch": 2.326771653543307,
+      "grad_norm": 5.742453098297119,
       "learning_rate": 2.5805336158042327e-07,
-      "loss": 0.3442,
+      "loss": 0.147,
       "step": 2364
     },
     {
-      "epoch": 2.3257874015748032,
-      "grad_norm": 6.176455974578857,
+      "epoch": 2.327755905511811,
+      "grad_norm": 10.463750839233398,
       "learning_rate": 2.573848258937945e-07,
-      "loss": 0.2219,
+      "loss": 0.6971,
       "step": 2365
     },
     {
-      "epoch": 2.326771653543307,
-      "grad_norm": 5.007375240325928,
+      "epoch": 2.328740157480315,
+      "grad_norm": 8.848584175109863,
       "learning_rate": 2.56716868408871e-07,
-      "loss": 0.1165,
+      "loss": 0.2906,
       "step": 2366
     },
     {
-      "epoch": 2.327755905511811,
-      "grad_norm": 12.076241493225098,
+      "epoch": 2.329724409448819,
+      "grad_norm": 16.255966186523438,
       "learning_rate": 2.5604949070667156e-07,
-      "loss": 0.3702,
+      "loss": 0.4248,
       "step": 2367
     },
     {
-      "epoch": 2.328740157480315,
-      "grad_norm": 9.106301307678223,
+      "epoch": 2.3307086614173227,
+      "grad_norm": 5.990625858306885,
       "learning_rate": 2.553826943668423e-07,
-      "loss": 0.3726,
+      "loss": 0.2883,
       "step": 2368
     },
     {
-      "epoch": 2.329724409448819,
-      "grad_norm": 10.722580909729004,
+      "epoch": 2.331692913385827,
+      "grad_norm": 10.7510986328125,
       "learning_rate": 2.5471648096765365e-07,
-      "loss": 0.5213,
+      "loss": 0.564,
       "step": 2369
     },
     {
-      "epoch": 2.3307086614173227,
-      "grad_norm": 9.431106567382812,
+      "epoch": 2.3326771653543306,
+      "grad_norm": 13.148714065551758,
       "learning_rate": 2.5405085208599623e-07,
-      "loss": 0.3472,
+      "loss": 0.7244,
       "step": 2370
     },
     {
-      "epoch": 2.331692913385827,
-      "grad_norm": 7.851450443267822,
+      "epoch": 2.3336614173228347,
+      "grad_norm": 9.19403076171875,
       "learning_rate": 2.53385809297377e-07,
-      "loss": 0.3008,
+      "loss": 0.3938,
       "step": 2371
     },
     {
-      "epoch": 2.3326771653543306,
-      "grad_norm": 17.107131958007812,
+      "epoch": 2.3346456692913384,
+      "grad_norm": 11.909695625305176,
       "learning_rate": 2.527213541759156e-07,
-      "loss": 0.7413,
+      "loss": 0.7018,
       "step": 2372
     },
     {
-      "epoch": 2.3336614173228347,
-      "grad_norm": 10.417817115783691,
+      "epoch": 2.3356299212598426,
+      "grad_norm": 8.899336814880371,
       "learning_rate": 2.520574882943411e-07,
-      "loss": 0.6197,
+      "loss": 0.316,
       "step": 2373
     },
     {
-      "epoch": 2.3346456692913384,
-      "grad_norm": 7.435096263885498,
+      "epoch": 2.3366141732283463,
+      "grad_norm": 10.192708015441895,
       "learning_rate": 2.5139421322398734e-07,
-      "loss": 0.2909,
+      "loss": 0.4577,
       "step": 2374
     },
     {
-      "epoch": 2.3356299212598426,
-      "grad_norm": 9.948948860168457,
+      "epoch": 2.3375984251968505,
+      "grad_norm": 7.061481952667236,
       "learning_rate": 2.5073153053479025e-07,
-      "loss": 0.383,
+      "loss": 0.3306,
       "step": 2375
     },
     {
-      "epoch": 2.3366141732283463,
-      "grad_norm": 15.071062088012695,
+      "epoch": 2.338582677165354,
+      "grad_norm": 16.46156883239746,
       "learning_rate": 2.5006944179528326e-07,
-      "loss": 0.6285,
+      "loss": 1.0778,
       "step": 2376
     },
     {
-      "epoch": 2.3375984251968505,
-      "grad_norm": 15.67939281463623,
+      "epoch": 2.3395669291338583,
+      "grad_norm": 10.59173583984375,
       "learning_rate": 2.494079485725939e-07,
-      "loss": 0.5903,
+      "loss": 0.3178,
       "step": 2377
     },
     {
-      "epoch": 2.338582677165354,
-      "grad_norm": 5.176926612854004,
+      "epoch": 2.340551181102362,
+      "grad_norm": 9.605278968811035,
       "learning_rate": 2.4874705243244044e-07,
-      "loss": 0.3165,
+      "loss": 0.396,
       "step": 2378
     },
     {
-      "epoch": 2.3395669291338583,
-      "grad_norm": 11.611159324645996,
+      "epoch": 2.341535433070866,
+      "grad_norm": 11.687631607055664,
       "learning_rate": 2.480867549391278e-07,
-      "loss": 0.4497,
+      "loss": 0.4328,
       "step": 2379
     },
     {
-      "epoch": 2.340551181102362,
-      "grad_norm": 11.762500762939453,
+      "epoch": 2.34251968503937,
+      "grad_norm": 12.270319938659668,
       "learning_rate": 2.4742705765554384e-07,
-      "loss": 0.6278,
+      "loss": 0.5537,
       "step": 2380
     },
     {
-      "epoch": 2.341535433070866,
-      "grad_norm": 9.781825065612793,
+      "epoch": 2.343503937007874,
+      "grad_norm": 5.425745010375977,
       "learning_rate": 2.467679621431558e-07,
-      "loss": 0.4037,
+      "loss": 0.244,
       "step": 2381
     },
     {
-      "epoch": 2.34251968503937,
-      "grad_norm": 9.960203170776367,
+      "epoch": 2.344488188976378,
+      "grad_norm": 6.1567583084106445,
       "learning_rate": 2.461094699620066e-07,
-      "loss": 0.555,
+      "loss": 0.2966,
       "step": 2382
     },
     {
-      "epoch": 2.343503937007874,
-      "grad_norm": 6.709632873535156,
+      "epoch": 2.345472440944882,
+      "grad_norm": 11.766680717468262,
       "learning_rate": 2.454515826707111e-07,
-      "loss": 0.1986,
+      "loss": 0.5565,
       "step": 2383
     },
     {
-      "epoch": 2.344488188976378,
-      "grad_norm": 13.113850593566895,
+      "epoch": 2.3464566929133857,
+      "grad_norm": 5.525933265686035,
       "learning_rate": 2.4479430182645237e-07,
-      "loss": 0.6334,
+      "loss": 0.1748,
       "step": 2384
     },
     {
-      "epoch": 2.345472440944882,
-      "grad_norm": 9.094974517822266,
+      "epoch": 2.34744094488189,
+      "grad_norm": 11.15112590789795,
       "learning_rate": 2.441376289849781e-07,
-      "loss": 0.4468,
+      "loss": 0.5972,
       "step": 2385
     },
     {
-      "epoch": 2.3464566929133857,
-      "grad_norm": 10.722009658813477,
+      "epoch": 2.3484251968503935,
+      "grad_norm": 11.578911781311035,
       "learning_rate": 2.4348156570059685e-07,
-      "loss": 0.3903,
+      "loss": 0.5639,
       "step": 2386
     },
     {
-      "epoch": 2.34744094488189,
-      "grad_norm": 7.4452223777771,
+      "epoch": 2.3494094488188977,
+      "grad_norm": 10.115438461303711,
       "learning_rate": 2.428261135261743e-07,
-      "loss": 0.4445,
+      "loss": 0.7857,
       "step": 2387
     },
     {
-      "epoch": 2.3484251968503935,
-      "grad_norm": 21.786439895629883,
+      "epoch": 2.3503937007874014,
+      "grad_norm": 9.496044158935547,
       "learning_rate": 2.4217127401312993e-07,
-      "loss": 1.0605,
+      "loss": 0.4016,
       "step": 2388
     },
     {
-      "epoch": 2.3494094488188977,
-      "grad_norm": 10.825016975402832,
+      "epoch": 2.3513779527559056,
+      "grad_norm": 10.8843412399292,
       "learning_rate": 2.4151704871143276e-07,
-      "loss": 0.4604,
+      "loss": 0.4808,
       "step": 2389
     },
     {
-      "epoch": 2.3503937007874014,
-      "grad_norm": 10.936182975769043,
+      "epoch": 2.3523622047244093,
+      "grad_norm": 8.956517219543457,
       "learning_rate": 2.4086343916959827e-07,
-      "loss": 0.5446,
+      "loss": 0.4922,
       "step": 2390
     },
     {
-      "epoch": 2.3513779527559056,
-      "grad_norm": 4.6355462074279785,
+      "epoch": 2.3533464566929134,
+      "grad_norm": 11.153104782104492,
       "learning_rate": 2.402104469346844e-07,
-      "loss": 0.127,
+      "loss": 0.4503,
       "step": 2391
     },
     {
-      "epoch": 2.3523622047244093,
-      "grad_norm": 8.731232643127441,
+      "epoch": 2.354330708661417,
+      "grad_norm": 12.932547569274902,
       "learning_rate": 2.3955807355228767e-07,
-      "loss": 0.457,
+      "loss": 0.9119,
       "step": 2392
     },
     {
-      "epoch": 2.3533464566929134,
-      "grad_norm": 7.753507614135742,
+      "epoch": 2.3553149606299213,
+      "grad_norm": 9.553558349609375,
       "learning_rate": 2.3890632056654027e-07,
-      "loss": 0.2166,
+      "loss": 0.405,
       "step": 2393
     },
     {
-      "epoch": 2.354330708661417,
-      "grad_norm": 10.839706420898438,
+      "epoch": 2.356299212598425,
+      "grad_norm": 5.514533519744873,
       "learning_rate": 2.3825518952010577e-07,
-      "loss": 0.4146,
+      "loss": 0.1421,
       "step": 2394
     },
     {
-      "epoch": 2.3553149606299213,
-      "grad_norm": 12.403355598449707,
+      "epoch": 2.357283464566929,
+      "grad_norm": 10.408130645751953,
       "learning_rate": 2.3760468195417546e-07,
-      "loss": 0.6237,
+      "loss": 0.5206,
       "step": 2395
     },
     {
-      "epoch": 2.356299212598425,
-      "grad_norm": 13.206016540527344,
+      "epoch": 2.358267716535433,
+      "grad_norm": 13.569435119628906,
       "learning_rate": 2.3695479940846513e-07,
-      "loss": 0.6888,
+      "loss": 0.3862,
       "step": 2396
     },
     {
-      "epoch": 2.357283464566929,
-      "grad_norm": 14.071619987487793,
+      "epoch": 2.359251968503937,
+      "grad_norm": 8.522723197937012,
       "learning_rate": 2.363055434212112e-07,
-      "loss": 0.9276,
+      "loss": 0.397,
       "step": 2397
     },
     {
-      "epoch": 2.358267716535433,
-      "grad_norm": 10.460234642028809,
+      "epoch": 2.3602362204724407,
+      "grad_norm": 9.208047866821289,
       "learning_rate": 2.3565691552916664e-07,
-      "loss": 0.4188,
+      "loss": 0.3575,
       "step": 2398
     },
     {
-      "epoch": 2.359251968503937,
-      "grad_norm": 13.995847702026367,
+      "epoch": 2.361220472440945,
+      "grad_norm": 6.657994747161865,
       "learning_rate": 2.3500891726759822e-07,
-      "loss": 0.7293,
+      "loss": 0.2154,
       "step": 2399
     },
     {
-      "epoch": 2.3602362204724407,
-      "grad_norm": 7.614719867706299,
+      "epoch": 2.362204724409449,
+      "grad_norm": 9.61835765838623,
       "learning_rate": 2.3436155017028224e-07,
-      "loss": 0.3724,
+      "loss": 0.5393,
       "step": 2400
     },
     {
-      "epoch": 2.361220472440945,
-      "grad_norm": 8.675552368164062,
+      "epoch": 2.3631889763779528,
+      "grad_norm": 12.659372329711914,
       "learning_rate": 2.3371481576950107e-07,
-      "loss": 0.3509,
+      "loss": 0.77,
       "step": 2401
     },
     {
-      "epoch": 2.362204724409449,
-      "grad_norm": 11.069519996643066,
+      "epoch": 2.3641732283464565,
+      "grad_norm": 10.12260627746582,
       "learning_rate": 2.3306871559603962e-07,
-      "loss": 0.4966,
+      "loss": 0.5329,
       "step": 2402
     },
     {
-      "epoch": 2.3631889763779528,
-      "grad_norm": 11.743781089782715,
+      "epoch": 2.3651574803149606,
+      "grad_norm": 12.133002281188965,
       "learning_rate": 2.324232511791814e-07,
-      "loss": 0.5803,
+      "loss": 0.6572,
       "step": 2403
     },
     {
-      "epoch": 2.3641732283464565,
-      "grad_norm": 12.193204879760742,
+      "epoch": 2.366141732283465,
+      "grad_norm": 10.349122047424316,
       "learning_rate": 2.3177842404670548e-07,
-      "loss": 0.6173,
+      "loss": 0.3386,
       "step": 2404
     },
     {
-      "epoch": 2.3651574803149606,
-      "grad_norm": 14.182714462280273,
+      "epoch": 2.3671259842519685,
+      "grad_norm": 9.186944007873535,
       "learning_rate": 2.3113423572488192e-07,
-      "loss": 0.3544,
+      "loss": 0.417,
       "step": 2405
     },
     {
-      "epoch": 2.366141732283465,
-      "grad_norm": 9.212885856628418,
+      "epoch": 2.3681102362204722,
+      "grad_norm": 10.587055206298828,
       "learning_rate": 2.304906877384692e-07,
-      "loss": 0.3204,
+      "loss": 0.5592,
       "step": 2406
     },
     {
-      "epoch": 2.3671259842519685,
-      "grad_norm": 16.641233444213867,
+      "epoch": 2.3690944881889764,
+      "grad_norm": 9.575271606445312,
       "learning_rate": 2.298477816107103e-07,
-      "loss": 0.7335,
+      "loss": 0.3127,
       "step": 2407
     },
     {
-      "epoch": 2.3681102362204722,
-      "grad_norm": 6.9422383308410645,
+      "epoch": 2.3700787401574805,
+      "grad_norm": 5.813891887664795,
       "learning_rate": 2.292055188633285e-07,
-      "loss": 0.2462,
+      "loss": 0.1558,
       "step": 2408
     },
     {
-      "epoch": 2.3690944881889764,
-      "grad_norm": 7.835033893585205,
+      "epoch": 2.3710629921259843,
+      "grad_norm": 9.62125301361084,
       "learning_rate": 2.2856390101652452e-07,
-      "loss": 0.2921,
+      "loss": 0.3328,
       "step": 2409
     },
     {
-      "epoch": 2.3700787401574805,
-      "grad_norm": 11.217126846313477,
+      "epoch": 2.372047244094488,
+      "grad_norm": 9.054393768310547,
       "learning_rate": 2.2792292958897266e-07,
-      "loss": 0.5388,
+      "loss": 0.3298,
       "step": 2410
     },
     {
-      "epoch": 2.3710629921259843,
-      "grad_norm": 10.122322082519531,
+      "epoch": 2.373031496062992,
+      "grad_norm": 11.466174125671387,
       "learning_rate": 2.2728260609781702e-07,
-      "loss": 0.5551,
+      "loss": 0.7329,
       "step": 2411
     },
     {
-      "epoch": 2.372047244094488,
-      "grad_norm": 11.416410446166992,
+      "epoch": 2.3740157480314963,
+      "grad_norm": 10.68238639831543,
       "learning_rate": 2.2664293205866817e-07,
-      "loss": 0.4502,
+      "loss": 0.4631,
       "step": 2412
     },
     {
-      "epoch": 2.373031496062992,
-      "grad_norm": 18.35036849975586,
+      "epoch": 2.375,
+      "grad_norm": 15.685314178466797,
       "learning_rate": 2.2600390898559944e-07,
-      "loss": 0.9301,
+      "loss": 0.7865,
       "step": 2413
     },
     {
-      "epoch": 2.3740157480314963,
-      "grad_norm": 10.62098503112793,
+      "epoch": 2.3759842519685037,
+      "grad_norm": 8.211750030517578,
       "learning_rate": 2.2536553839114338e-07,
-      "loss": 0.4878,
+      "loss": 0.4219,
       "step": 2414
     },
     {
-      "epoch": 2.375,
-      "grad_norm": 9.163315773010254,
+      "epoch": 2.376968503937008,
+      "grad_norm": 8.674180030822754,
       "learning_rate": 2.2472782178628807e-07,
-      "loss": 0.3984,
+      "loss": 0.3302,
       "step": 2415
     },
     {
-      "epoch": 2.3759842519685037,
-      "grad_norm": 7.131898880004883,
+      "epoch": 2.377952755905512,
+      "grad_norm": 8.48555850982666,
       "learning_rate": 2.2409076068047373e-07,
-      "loss": 0.2384,
+      "loss": 0.4047,
       "step": 2416
     },
     {
-      "epoch": 2.376968503937008,
-      "grad_norm": 8.367667198181152,
+      "epoch": 2.3789370078740157,
+      "grad_norm": 8.057952880859375,
       "learning_rate": 2.2345435658158908e-07,
-      "loss": 0.3365,
+      "loss": 0.2258,
       "step": 2417
     },
     {
-      "epoch": 2.377952755905512,
-      "grad_norm": 11.66543960571289,
+      "epoch": 2.3799212598425195,
+      "grad_norm": 7.399155139923096,
       "learning_rate": 2.2281861099596783e-07,
-      "loss": 0.5808,
+      "loss": 0.2736,
       "step": 2418
     },
     {
-      "epoch": 2.3789370078740157,
-      "grad_norm": 10.385133743286133,
+      "epoch": 2.3809055118110236,
+      "grad_norm": 12.442201614379883,
       "learning_rate": 2.2218352542838447e-07,
-      "loss": 0.2567,
+      "loss": 0.5218,
       "step": 2419
     },
     {
-      "epoch": 2.3799212598425195,
-      "grad_norm": 7.88507080078125,
+      "epoch": 2.3818897637795278,
+      "grad_norm": 12.58726692199707,
       "learning_rate": 2.215491013820519e-07,
-      "loss": 0.2747,
+      "loss": 0.6613,
       "step": 2420
     },
     {
-      "epoch": 2.3809055118110236,
-      "grad_norm": 6.82606315612793,
+      "epoch": 2.3828740157480315,
+      "grad_norm": 7.496347427368164,
       "learning_rate": 2.2091534035861703e-07,
-      "loss": 0.2653,
+      "loss": 0.2613,
       "step": 2421
     },
     {
-      "epoch": 2.3818897637795278,
-      "grad_norm": 13.949604034423828,
+      "epoch": 2.383858267716535,
+      "grad_norm": 13.048212051391602,
       "learning_rate": 2.2028224385815726e-07,
-      "loss": 0.7345,
+      "loss": 0.7631,
       "step": 2422
     },
     {
-      "epoch": 2.3828740157480315,
-      "grad_norm": 8.991510391235352,
+      "epoch": 2.3848425196850394,
+      "grad_norm": 9.532183647155762,
       "learning_rate": 2.1964981337917744e-07,
-      "loss": 0.3903,
+      "loss": 0.3823,
       "step": 2423
     },
     {
-      "epoch": 2.383858267716535,
-      "grad_norm": 17.700889587402344,
+      "epoch": 2.3858267716535435,
+      "grad_norm": 10.271509170532227,
       "learning_rate": 2.1901805041860567e-07,
-      "loss": 0.9409,
+      "loss": 0.456,
       "step": 2424
     },
     {
-      "epoch": 2.3848425196850394,
-      "grad_norm": 19.267620086669922,
+      "epoch": 2.3868110236220472,
+      "grad_norm": 14.456857681274414,
       "learning_rate": 2.1838695647179024e-07,
-      "loss": 0.7465,
+      "loss": 1.1117,
       "step": 2425
     },
     {
-      "epoch": 2.3858267716535435,
-      "grad_norm": 15.618047714233398,
+      "epoch": 2.387795275590551,
+      "grad_norm": 8.841161727905273,
       "learning_rate": 2.1775653303249587e-07,
-      "loss": 0.7577,
+      "loss": 0.4854,
       "step": 2426
     },
     {
-      "epoch": 2.3868110236220472,
-      "grad_norm": 5.0363450050354,
+      "epoch": 2.388779527559055,
+      "grad_norm": 5.9602460861206055,
       "learning_rate": 2.1712678159290023e-07,
-      "loss": 0.1028,
+      "loss": 0.1749,
       "step": 2427
     },
     {
-      "epoch": 2.387795275590551,
-      "grad_norm": 9.236956596374512,
+      "epoch": 2.3897637795275593,
+      "grad_norm": 9.90578842163086,
       "learning_rate": 2.164977036435904e-07,
-      "loss": 0.4326,
+      "loss": 0.2807,
       "step": 2428
     },
     {
-      "epoch": 2.388779527559055,
-      "grad_norm": 3.6530535221099854,
+      "epoch": 2.390748031496063,
+      "grad_norm": 11.956024169921875,
       "learning_rate": 2.1586930067355932e-07,
-      "loss": 0.0553,
+      "loss": 0.8104,
       "step": 2429
     },
     {
-      "epoch": 2.3897637795275593,
-      "grad_norm": 6.015213966369629,
+      "epoch": 2.391732283464567,
+      "grad_norm": 7.525900840759277,
       "learning_rate": 2.152415741702023e-07,
-      "loss": 0.1769,
+      "loss": 0.2549,
       "step": 2430
     },
     {
-      "epoch": 2.390748031496063,
-      "grad_norm": 9.619933128356934,
+      "epoch": 2.392716535433071,
+      "grad_norm": 10.520135879516602,
       "learning_rate": 2.146145256193136e-07,
-      "loss": 0.2792,
+      "loss": 0.3055,
       "step": 2431
     },
     {
-      "epoch": 2.391732283464567,
-      "grad_norm": 11.973121643066406,
+      "epoch": 2.393700787401575,
+      "grad_norm": 13.294878005981445,
       "learning_rate": 2.1398815650508263e-07,
-      "loss": 0.5102,
+      "loss": 0.7992,
       "step": 2432
     },
     {
-      "epoch": 2.392716535433071,
-      "grad_norm": 9.242534637451172,
+      "epoch": 2.3946850393700787,
+      "grad_norm": 10.733575820922852,
       "learning_rate": 2.1336246831009075e-07,
-      "loss": 0.2668,
+      "loss": 0.4635,
       "step": 2433
     },
     {
-      "epoch": 2.393700787401575,
-      "grad_norm": 11.018208503723145,
+      "epoch": 2.395669291338583,
+      "grad_norm": 8.763347625732422,
       "learning_rate": 2.1273746251530752e-07,
-      "loss": 0.5544,
+      "loss": 0.2588,
       "step": 2434
     },
     {
-      "epoch": 2.3946850393700787,
-      "grad_norm": 8.128729820251465,
+      "epoch": 2.3966535433070866,
+      "grad_norm": 8.571314811706543,
       "learning_rate": 2.1211314060008755e-07,
-      "loss": 0.2821,
+      "loss": 0.329,
       "step": 2435
     },
     {
-      "epoch": 2.395669291338583,
-      "grad_norm": 11.07681941986084,
+      "epoch": 2.3976377952755907,
+      "grad_norm": 9.651409149169922,
       "learning_rate": 2.1148950404216617e-07,
-      "loss": 0.3712,
+      "loss": 0.5157,
       "step": 2436
     },
     {
-      "epoch": 2.3966535433070866,
-      "grad_norm": 7.771712303161621,
+      "epoch": 2.3986220472440944,
+      "grad_norm": 5.080859184265137,
       "learning_rate": 2.10866554317657e-07,
-      "loss": 0.2709,
+      "loss": 0.1639,
       "step": 2437
     },
     {
-      "epoch": 2.3976377952755907,
-      "grad_norm": 9.18880558013916,
+      "epoch": 2.3996062992125986,
+      "grad_norm": 7.444197654724121,
       "learning_rate": 2.1024429290104796e-07,
-      "loss": 0.3861,
+      "loss": 0.2695,
       "step": 2438
     },
     {
-      "epoch": 2.3986220472440944,
-      "grad_norm": 8.970224380493164,
+      "epoch": 2.4005905511811023,
+      "grad_norm": 7.8818359375,
       "learning_rate": 2.0962272126519756e-07,
-      "loss": 0.2303,
+      "loss": 0.2459,
       "step": 2439
     },
     {
-      "epoch": 2.3996062992125986,
-      "grad_norm": 13.390796661376953,
+      "epoch": 2.4015748031496065,
+      "grad_norm": 11.374838829040527,
       "learning_rate": 2.0900184088133173e-07,
-      "loss": 0.7863,
+      "loss": 0.4676,
       "step": 2440
     },
     {
-      "epoch": 2.4005905511811023,
-      "grad_norm": 9.731471061706543,
+      "epoch": 2.40255905511811,
+      "grad_norm": 5.920230388641357,
       "learning_rate": 2.0838165321904045e-07,
-      "loss": 0.2954,
+      "loss": 0.1643,
       "step": 2441
     },
     {
-      "epoch": 2.4015748031496065,
-      "grad_norm": 8.255941390991211,
+      "epoch": 2.4035433070866143,
+      "grad_norm": 8.242584228515625,
       "learning_rate": 2.0776215974627359e-07,
-      "loss": 0.3938,
+      "loss": 0.2919,
       "step": 2442
     },
     {
-      "epoch": 2.40255905511811,
-      "grad_norm": 10.414751052856445,
+      "epoch": 2.404527559055118,
+      "grad_norm": 11.445818901062012,
       "learning_rate": 2.0714336192933824e-07,
-      "loss": 0.5809,
+      "loss": 0.4056,
       "step": 2443
     },
     {
-      "epoch": 2.4035433070866143,
-      "grad_norm": 9.516477584838867,
+      "epoch": 2.405511811023622,
+      "grad_norm": 9.079263687133789,
       "learning_rate": 2.0652526123289488e-07,
-      "loss": 0.3192,
+      "loss": 0.3375,
       "step": 2444
     },
     {
-      "epoch": 2.404527559055118,
-      "grad_norm": 8.152141571044922,
+      "epoch": 2.406496062992126,
+      "grad_norm": 9.128870010375977,
       "learning_rate": 2.0590785911995406e-07,
-      "loss": 0.3214,
+      "loss": 0.3404,
       "step": 2445
     },
     {
-      "epoch": 2.405511811023622,
-      "grad_norm": 12.904752731323242,
+      "epoch": 2.40748031496063,
+      "grad_norm": 12.104570388793945,
       "learning_rate": 2.052911570518725e-07,
-      "loss": 0.6149,
+      "loss": 0.4788,
       "step": 2446
     },
     {
-      "epoch": 2.406496062992126,
-      "grad_norm": 11.767104148864746,
+      "epoch": 2.408464566929134,
+      "grad_norm": 9.512703895568848,
       "learning_rate": 2.046751564883503e-07,
-      "loss": 0.6265,
+      "loss": 0.3727,
       "step": 2447
     },
     {
-      "epoch": 2.40748031496063,
-      "grad_norm": 15.724270820617676,
+      "epoch": 2.409448818897638,
+      "grad_norm": 13.60953426361084,
       "learning_rate": 2.040598588874268e-07,
-      "loss": 0.5605,
+      "loss": 0.5221,
       "step": 2448
     },
     {
-      "epoch": 2.40748031496063,
-      "eval_Qnli-dev_cosine_accuracy": 0.685546875,
-      "eval_Qnli-dev_cosine_accuracy_threshold": 0.7995734214782715,
-      "eval_Qnli-dev_cosine_ap": 0.708134684015663,
-      "eval_Qnli-dev_cosine_f1": 0.6960784313725491,
-      "eval_Qnli-dev_cosine_f1_threshold": 0.7281224131584167,
-      "eval_Qnli-dev_cosine_precision": 0.5664893617021277,
-      "eval_Qnli-dev_cosine_recall": 0.902542372881356,
-      "eval_Qnli-dev_dot_accuracy": 0.681640625,
-      "eval_Qnli-dev_dot_accuracy_threshold": 609.6793212890625,
-      "eval_Qnli-dev_dot_ap": 0.7047383707764411,
-      "eval_Qnli-dev_dot_f1": 0.6929392446633826,
-      "eval_Qnli-dev_dot_f1_threshold": 555.753173828125,
-      "eval_Qnli-dev_dot_precision": 0.5656836461126006,
-      "eval_Qnli-dev_dot_recall": 0.8940677966101694,
-      "eval_Qnli-dev_euclidean_accuracy": 0.685546875,
-      "eval_Qnli-dev_euclidean_accuracy_threshold": 17.47189712524414,
-      "eval_Qnli-dev_euclidean_ap": 0.7083737231857451,
-      "eval_Qnli-dev_euclidean_f1": 0.6950819672131149,
-      "eval_Qnli-dev_euclidean_f1_threshold": 20.271848678588867,
-      "eval_Qnli-dev_euclidean_precision": 0.5668449197860963,
-      "eval_Qnli-dev_euclidean_recall": 0.8983050847457628,
-      "eval_Qnli-dev_manhattan_accuracy": 0.693359375,
-      "eval_Qnli-dev_manhattan_accuracy_threshold": 365.6907958984375,
-      "eval_Qnli-dev_manhattan_ap": 0.7170160615271656,
-      "eval_Qnli-dev_manhattan_f1": 0.7000000000000001,
-      "eval_Qnli-dev_manhattan_f1_threshold": 435.0411376953125,
-      "eval_Qnli-dev_manhattan_precision": 0.5651041666666666,
-      "eval_Qnli-dev_manhattan_recall": 0.9194915254237288,
-      "eval_Qnli-dev_max_accuracy": 0.693359375,
-      "eval_Qnli-dev_max_accuracy_threshold": 609.6793212890625,
-      "eval_Qnli-dev_max_ap": 0.7170160615271656,
-      "eval_Qnli-dev_max_f1": 0.7000000000000001,
-      "eval_Qnli-dev_max_f1_threshold": 555.753173828125,
-      "eval_Qnli-dev_max_precision": 0.5668449197860963,
-      "eval_Qnli-dev_max_recall": 0.9194915254237288,
-      "eval_allNLI-dev_cosine_accuracy": 0.697265625,
-      "eval_allNLI-dev_cosine_accuracy_threshold": 0.918925404548645,
-      "eval_allNLI-dev_cosine_ap": 0.520334459841962,
-      "eval_allNLI-dev_cosine_f1": 0.557919621749409,
-      "eval_allNLI-dev_cosine_f1_threshold": 0.8274284601211548,
-      "eval_allNLI-dev_cosine_precision": 0.472,
-      "eval_allNLI-dev_cosine_recall": 0.6820809248554913,
-      "eval_allNLI-dev_dot_accuracy": 0.697265625,
-      "eval_allNLI-dev_dot_accuracy_threshold": 701.4763793945312,
-      "eval_allNLI-dev_dot_ap": 0.5215384128413414,
-      "eval_allNLI-dev_dot_f1": 0.557919621749409,
-      "eval_allNLI-dev_dot_f1_threshold": 631.42578125,
-      "eval_allNLI-dev_dot_precision": 0.472,
-      "eval_allNLI-dev_dot_recall": 0.6820809248554913,
-      "eval_allNLI-dev_euclidean_accuracy": 0.697265625,
-      "eval_allNLI-dev_euclidean_accuracy_threshold": 11.115690231323242,
-      "eval_allNLI-dev_euclidean_ap": 0.5206070999404473,
-      "eval_allNLI-dev_euclidean_f1": 0.557919621749409,
-      "eval_allNLI-dev_euclidean_f1_threshold": 16.241830825805664,
-      "eval_allNLI-dev_euclidean_precision": 0.472,
-      "eval_allNLI-dev_euclidean_recall": 0.6820809248554913,
-      "eval_allNLI-dev_manhattan_accuracy": 0.69140625,
-      "eval_allNLI-dev_manhattan_accuracy_threshold": 240.7887725830078,
-      "eval_allNLI-dev_manhattan_ap": 0.5210401368161633,
-      "eval_allNLI-dev_manhattan_f1": 0.5545454545454546,
-      "eval_allNLI-dev_manhattan_f1_threshold": 348.866455078125,
-      "eval_allNLI-dev_manhattan_precision": 0.45692883895131087,
-      "eval_allNLI-dev_manhattan_recall": 0.7052023121387283,
-      "eval_allNLI-dev_max_accuracy": 0.697265625,
-      "eval_allNLI-dev_max_accuracy_threshold": 701.4763793945312,
-      "eval_allNLI-dev_max_ap": 0.5215384128413414,
-      "eval_allNLI-dev_max_f1": 0.557919621749409,
-      "eval_allNLI-dev_max_f1_threshold": 631.42578125,
-      "eval_allNLI-dev_max_precision": 0.472,
-      "eval_allNLI-dev_max_recall": 0.7052023121387283,
-      "eval_loss": 0.9221314787864685,
-      "eval_runtime": 54.3282,
-      "eval_samples_per_second": 30.629,
-      "eval_sequential_score": 0.7170160615271656,
-      "eval_steps_per_second": 0.129,
-      "eval_sts-test_pearson_cosine": 0.6584277178847996,
-      "eval_sts-test_pearson_dot": 0.657756195496527,
-      "eval_sts-test_pearson_euclidean": 0.6707158629188267,
-      "eval_sts-test_pearson_manhattan": 0.6745148890301079,
-      "eval_sts-test_pearson_max": 0.6745148890301079,
-      "eval_sts-test_spearman_cosine": 0.6639480172687122,
-      "eval_sts-test_spearman_dot": 0.6642357678794589,
-      "eval_sts-test_spearman_euclidean": 0.6641763382043312,
-      "eval_sts-test_spearman_manhattan": 0.6678039202039794,
-      "eval_sts-test_spearman_max": 0.6678039202039794,
+      "epoch": 2.409448818897638,
+      "eval_Qnli-dev_cosine_accuracy": 0.6796875,
+      "eval_Qnli-dev_cosine_accuracy_threshold": 0.7952160835266113,
+      "eval_Qnli-dev_cosine_ap": 0.7017112308062804,
+      "eval_Qnli-dev_cosine_f1": 0.6917808219178082,
+      "eval_Qnli-dev_cosine_f1_threshold": 0.747643232345581,
+      "eval_Qnli-dev_cosine_precision": 0.5804597701149425,
+      "eval_Qnli-dev_cosine_recall": 0.8559322033898306,
+      "eval_Qnli-dev_dot_accuracy": 0.6796875,
+      "eval_Qnli-dev_dot_accuracy_threshold": 602.139404296875,
+      "eval_Qnli-dev_dot_ap": 0.7001697001554149,
+      "eval_Qnli-dev_dot_f1": 0.6914893617021277,
+      "eval_Qnli-dev_dot_f1_threshold": 577.031005859375,
+      "eval_Qnli-dev_dot_precision": 0.5945121951219512,
+      "eval_Qnli-dev_dot_recall": 0.826271186440678,
+      "eval_Qnli-dev_euclidean_accuracy": 0.681640625,
+      "eval_Qnli-dev_euclidean_accuracy_threshold": 17.874900817871094,
+      "eval_Qnli-dev_euclidean_ap": 0.7016931617573394,
+      "eval_Qnli-dev_euclidean_f1": 0.6916221033868093,
+      "eval_Qnli-dev_euclidean_f1_threshold": 19.12659454345703,
+      "eval_Qnli-dev_euclidean_precision": 0.5969230769230769,
+      "eval_Qnli-dev_euclidean_recall": 0.8220338983050848,
+      "eval_Qnli-dev_manhattan_accuracy": 0.68359375,
+      "eval_Qnli-dev_manhattan_accuracy_threshold": 366.3169860839844,
+      "eval_Qnli-dev_manhattan_ap": 0.7099970111283378,
+      "eval_Qnli-dev_manhattan_f1": 0.6940170940170939,
+      "eval_Qnli-dev_manhattan_f1_threshold": 416.5448913574219,
+      "eval_Qnli-dev_manhattan_precision": 0.5816618911174785,
+      "eval_Qnli-dev_manhattan_recall": 0.8601694915254238,
+      "eval_Qnli-dev_max_accuracy": 0.68359375,
+      "eval_Qnli-dev_max_accuracy_threshold": 602.139404296875,
+      "eval_Qnli-dev_max_ap": 0.7099970111283378,
+      "eval_Qnli-dev_max_f1": 0.6940170940170939,
+      "eval_Qnli-dev_max_f1_threshold": 577.031005859375,
+      "eval_Qnli-dev_max_precision": 0.5969230769230769,
+      "eval_Qnli-dev_max_recall": 0.8601694915254238,
+      "eval_allNLI-dev_cosine_accuracy": 0.689453125,
+      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9271423816680908,
+      "eval_allNLI-dev_cosine_ap": 0.5030647458582906,
+      "eval_allNLI-dev_cosine_f1": 0.5476673427991887,
+      "eval_allNLI-dev_cosine_f1_threshold": 0.816743791103363,
+      "eval_allNLI-dev_cosine_precision": 0.421875,
+      "eval_allNLI-dev_cosine_recall": 0.7803468208092486,
+      "eval_allNLI-dev_dot_accuracy": 0.689453125,
+      "eval_allNLI-dev_dot_accuracy_threshold": 709.7324829101562,
+      "eval_allNLI-dev_dot_ap": 0.49999248536953744,
+      "eval_allNLI-dev_dot_f1": 0.5481171548117154,
+      "eval_allNLI-dev_dot_f1_threshold": 626.8201904296875,
+      "eval_allNLI-dev_dot_precision": 0.42950819672131146,
+      "eval_allNLI-dev_dot_recall": 0.7572254335260116,
+      "eval_allNLI-dev_euclidean_accuracy": 0.689453125,
+      "eval_allNLI-dev_euclidean_accuracy_threshold": 10.533082962036133,
+      "eval_allNLI-dev_euclidean_ap": 0.503026837479258,
+      "eval_allNLI-dev_euclidean_f1": 0.5476673427991887,
+      "eval_allNLI-dev_euclidean_f1_threshold": 16.725177764892578,
+      "eval_allNLI-dev_euclidean_precision": 0.421875,
+      "eval_allNLI-dev_euclidean_recall": 0.7803468208092486,
+      "eval_allNLI-dev_manhattan_accuracy": 0.685546875,
+      "eval_allNLI-dev_manhattan_accuracy_threshold": 225.1102752685547,
+      "eval_allNLI-dev_manhattan_ap": 0.5010991653362263,
+      "eval_allNLI-dev_manhattan_f1": 0.5478260869565218,
+      "eval_allNLI-dev_manhattan_f1_threshold": 342.0830383300781,
+      "eval_allNLI-dev_manhattan_precision": 0.43902439024390244,
+      "eval_allNLI-dev_manhattan_recall": 0.7283236994219653,
+      "eval_allNLI-dev_max_accuracy": 0.689453125,
+      "eval_allNLI-dev_max_accuracy_threshold": 709.7324829101562,
+      "eval_allNLI-dev_max_ap": 0.5030647458582906,
+      "eval_allNLI-dev_max_f1": 0.5481171548117154,
+      "eval_allNLI-dev_max_f1_threshold": 626.8201904296875,
+      "eval_allNLI-dev_max_precision": 0.43902439024390244,
+      "eval_allNLI-dev_max_recall": 0.7803468208092486,
+      "eval_loss": 0.8089768886566162,
+      "eval_runtime": 53.3682,
+      "eval_samples_per_second": 31.18,
+      "eval_sequential_score": 0.7099970111283378,
+      "eval_steps_per_second": 0.131,
+      "eval_sts-test_pearson_cosine": 0.6657219579047406,
+      "eval_sts-test_pearson_dot": 0.6635298297483249,
+      "eval_sts-test_pearson_euclidean": 0.677320071607842,
+      "eval_sts-test_pearson_manhattan": 0.68025341564508,
+      "eval_sts-test_pearson_max": 0.68025341564508,
+      "eval_sts-test_spearman_cosine": 0.6710335739873353,
+      "eval_sts-test_spearman_dot": 0.6670371073400377,
+      "eval_sts-test_spearman_euclidean": 0.671354485282773,
+      "eval_sts-test_spearman_manhattan": 0.6729487489698417,
+      "eval_sts-test_spearman_max": 0.6729487489698417,
       "step": 2448
     },
     {
-      "epoch": 2.408464566929134,
-      "grad_norm": 15.768280982971191,
+      "epoch": 2.4104330708661417,
+      "grad_norm": 11.234091758728027,
       "learning_rate": 2.0344526570547774e-07,
-      "loss": 0.6515,
+      "loss": 0.5299,
       "step": 2449
     },
     {
-      "epoch": 2.409448818897638,
-      "grad_norm": 7.4104814529418945,
+      "epoch": 2.411417322834646,
+      "grad_norm": 4.555033206939697,
       "learning_rate": 2.0283137839721143e-07,
-      "loss": 0.258,
+      "loss": 0.1552,
       "step": 2450
     },
     {
-      "epoch": 2.4104330708661417,
-      "grad_norm": 10.254469871520996,
+      "epoch": 2.4124015748031495,
+      "grad_norm": 13.280162811279297,
       "learning_rate": 2.022181984156654e-07,
-      "loss": 0.5888,
+      "loss": 0.446,
       "step": 2451
     },
     {
-      "epoch": 2.411417322834646,
-      "grad_norm": 9.213778495788574,
+      "epoch": 2.4133858267716537,
+      "grad_norm": 7.345797538757324,
       "learning_rate": 2.0160572721220312e-07,
-      "loss": 0.3751,
+      "loss": 0.2375,
       "step": 2452
     },
     {
-      "epoch": 2.4124015748031495,
-      "grad_norm": 6.009047985076904,
+      "epoch": 2.4143700787401574,
+      "grad_norm": 6.918489456176758,
       "learning_rate": 2.0099396623651012e-07,
-      "loss": 0.1577,
+      "loss": 0.2357,
       "step": 2453
     },
     {
-      "epoch": 2.4133858267716537,
-      "grad_norm": 18.305646896362305,
+      "epoch": 2.4153543307086616,
+      "grad_norm": 10.824198722839355,
       "learning_rate": 2.0038291693659113e-07,
-      "loss": 0.8158,
+      "loss": 0.3831,
       "step": 2454
     },
     {
-      "epoch": 2.4143700787401574,
-      "grad_norm": 14.949423789978027,
+      "epoch": 2.4163385826771653,
+      "grad_norm": 11.411164283752441,
       "learning_rate": 1.997725807587663e-07,
-      "loss": 0.4467,
+      "loss": 0.6244,
       "step": 2455
     },
     {
-      "epoch": 2.4153543307086616,
-      "grad_norm": 8.558846473693848,
+      "epoch": 2.4173228346456694,
+      "grad_norm": 10.078960418701172,
       "learning_rate": 1.991629591476679e-07,
-      "loss": 0.3115,
+      "loss": 0.544,
       "step": 2456
     },
     {
-      "epoch": 2.4163385826771653,
-      "grad_norm": 9.066378593444824,
+      "epoch": 2.418307086614173,
+      "grad_norm": 8.428912162780762,
       "learning_rate": 1.9855405354623664e-07,
-      "loss": 0.3204,
+      "loss": 0.458,
       "step": 2457
     },
     {
-      "epoch": 2.4173228346456694,
-      "grad_norm": 10.729354858398438,
+      "epoch": 2.4192913385826773,
+      "grad_norm": 8.138558387756348,
       "learning_rate": 1.9794586539571887e-07,
-      "loss": 0.4332,
+      "loss": 0.2782,
       "step": 2458
     },
     {
-      "epoch": 2.418307086614173,
-      "grad_norm": 13.04471206665039,
+      "epoch": 2.420275590551181,
+      "grad_norm": 5.636626243591309,
       "learning_rate": 1.9733839613566244e-07,
-      "loss": 0.7696,
+      "loss": 0.2767,
       "step": 2459
     },
     {
-      "epoch": 2.4192913385826773,
-      "grad_norm": 7.856568813323975,
+      "epoch": 2.421259842519685,
+      "grad_norm": 9.14765453338623,
       "learning_rate": 1.9673164720391387e-07,
-      "loss": 0.2334,
+      "loss": 0.4177,
       "step": 2460
     },
     {
-      "epoch": 2.420275590551181,
-      "grad_norm": 7.954514503479004,
+      "epoch": 2.422244094488189,
+      "grad_norm": 6.205542087554932,
       "learning_rate": 1.9612562003661423e-07,
-      "loss": 0.346,
+      "loss": 0.2191,
       "step": 2461
     },
     {
-      "epoch": 2.421259842519685,
-      "grad_norm": 6.173386096954346,
+      "epoch": 2.423228346456693,
+      "grad_norm": 9.531805038452148,
       "learning_rate": 1.9552031606819685e-07,
-      "loss": 0.1947,
+      "loss": 0.2996,
       "step": 2462
     },
     {
-      "epoch": 2.422244094488189,
-      "grad_norm": 9.832446098327637,
+      "epoch": 2.4242125984251968,
+      "grad_norm": 8.64529800415039,
       "learning_rate": 1.9491573673138287e-07,
-      "loss": 0.4588,
+      "loss": 0.3109,
       "step": 2463
     },
     {
-      "epoch": 2.423228346456693,
-      "grad_norm": 9.947352409362793,
+      "epoch": 2.425196850393701,
+      "grad_norm": 16.158525466918945,
       "learning_rate": 1.9431188345717844e-07,
-      "loss": 0.3671,
+      "loss": 0.7066,
       "step": 2464
     },
     {
-      "epoch": 2.4242125984251968,
-      "grad_norm": 11.656517028808594,
+      "epoch": 2.4261811023622046,
+      "grad_norm": 10.459314346313477,
       "learning_rate": 1.9370875767487104e-07,
-      "loss": 0.3214,
+      "loss": 0.6101,
       "step": 2465
     },
     {
-      "epoch": 2.425196850393701,
-      "grad_norm": 14.011076927185059,
+      "epoch": 2.427165354330709,
+      "grad_norm": 11.172530174255371,
       "learning_rate": 1.9310636081202636e-07,
-      "loss": 0.5728,
+      "loss": 0.6249,
       "step": 2466
     },
     {
-      "epoch": 2.4261811023622046,
-      "grad_norm": 12.770512580871582,
+      "epoch": 2.4281496062992125,
+      "grad_norm": 8.52651596069336,
       "learning_rate": 1.925046942944847e-07,
-      "loss": 0.7554,
+      "loss": 0.2996,
       "step": 2467
     },
     {
-      "epoch": 2.427165354330709,
-      "grad_norm": 9.351787567138672,
+      "epoch": 2.4291338582677167,
+      "grad_norm": 9.142427444458008,
       "learning_rate": 1.919037595463576e-07,
-      "loss": 0.3476,
+      "loss": 0.4268,
       "step": 2468
     },
     {
-      "epoch": 2.4281496062992125,
-      "grad_norm": 9.923174858093262,
+      "epoch": 2.4301181102362204,
+      "grad_norm": 9.976550102233887,
       "learning_rate": 1.9130355799002467e-07,
-      "loss": 0.4924,
+      "loss": 0.5168,
       "step": 2469
     },
     {
-      "epoch": 2.4291338582677167,
-      "grad_norm": 12.662055969238281,
+      "epoch": 2.4311023622047245,
+      "grad_norm": 7.142285346984863,
       "learning_rate": 1.9070409104613e-07,
-      "loss": 0.6821,
+      "loss": 0.1912,
       "step": 2470
     },
     {
-      "epoch": 2.4301181102362204,
-      "grad_norm": 12.125091552734375,
+      "epoch": 2.4320866141732282,
+      "grad_norm": 9.933122634887695,
       "learning_rate": 1.90105360133579e-07,
-      "loss": 0.5663,
+      "loss": 0.3655,
       "step": 2471
     },
     {
-      "epoch": 2.4311023622047245,
-      "grad_norm": 7.01002311706543,
+      "epoch": 2.4330708661417324,
+      "grad_norm": 11.601430892944336,
       "learning_rate": 1.89507366669535e-07,
-      "loss": 0.2414,
+      "loss": 0.5847,
       "step": 2472
     },
     {
-      "epoch": 2.4320866141732282,
-      "grad_norm": 7.311014175415039,
+      "epoch": 2.434055118110236,
+      "grad_norm": 10.952865600585938,
       "learning_rate": 1.8891011206941535e-07,
-      "loss": 0.1729,
+      "loss": 0.5454,
       "step": 2473
     },
     {
-      "epoch": 2.4330708661417324,
-      "grad_norm": 13.780094146728516,
+      "epoch": 2.4350393700787403,
+      "grad_norm": 8.046846389770508,
       "learning_rate": 1.883135977468891e-07,
-      "loss": 0.8102,
+      "loss": 0.2352,
       "step": 2474
     },
     {
-      "epoch": 2.434055118110236,
-      "grad_norm": 11.154767036437988,
+      "epoch": 2.436023622047244,
+      "grad_norm": 7.111355304718018,
       "learning_rate": 1.8771782511387285e-07,
-      "loss": 0.5511,
+      "loss": 0.2123,
       "step": 2475
     },
     {
-      "epoch": 2.4350393700787403,
-      "grad_norm": 13.085041999816895,
+      "epoch": 2.437007874015748,
+      "grad_norm": 11.884984970092773,
       "learning_rate": 1.8712279558052777e-07,
-      "loss": 0.5231,
+      "loss": 0.6476,
       "step": 2476
     },
     {
-      "epoch": 2.436023622047244,
-      "grad_norm": 2.9605321884155273,
+      "epoch": 2.437992125984252,
+      "grad_norm": 12.956525802612305,
       "learning_rate": 1.8652851055525605e-07,
-      "loss": 0.0735,
+      "loss": 0.6335,
       "step": 2477
     },
     {
-      "epoch": 2.437007874015748,
-      "grad_norm": 9.01956844329834,
+      "epoch": 2.438976377952756,
+      "grad_norm": 9.851030349731445,
       "learning_rate": 1.859349714446977e-07,
-      "loss": 0.344,
+      "loss": 0.2991,
       "step": 2478
     },
     {
-      "epoch": 2.437992125984252,
-      "grad_norm": 10.247904777526855,
+      "epoch": 2.4399606299212597,
+      "grad_norm": 9.243252754211426,
       "learning_rate": 1.853421796537273e-07,
-      "loss": 0.5571,
+      "loss": 0.3649,
       "step": 2479
     },
     {
-      "epoch": 2.438976377952756,
-      "grad_norm": 9.704607009887695,
+      "epoch": 2.440944881889764,
+      "grad_norm": 10.924264907836914,
       "learning_rate": 1.8475013658545054e-07,
-      "loss": 0.3844,
+      "loss": 0.5475,
       "step": 2480
     },
     {
-      "epoch": 2.4399606299212597,
-      "grad_norm": 7.756630897521973,
+      "epoch": 2.4419291338582676,
+      "grad_norm": 8.867262840270996,
       "learning_rate": 1.841588436412006e-07,
-      "loss": 0.2894,
+      "loss": 0.3319,
       "step": 2481
     },
     {
-      "epoch": 2.440944881889764,
-      "grad_norm": 10.263578414916992,
+      "epoch": 2.4429133858267718,
+      "grad_norm": 10.920964241027832,
       "learning_rate": 1.8356830222053546e-07,
-      "loss": 0.2938,
+      "loss": 0.4839,
       "step": 2482
     },
     {
-      "epoch": 2.4419291338582676,
-      "grad_norm": 7.927118301391602,
+      "epoch": 2.4438976377952755,
+      "grad_norm": 11.578985214233398,
       "learning_rate": 1.8297851372123444e-07,
-      "loss": 0.2579,
+      "loss": 0.649,
       "step": 2483
     },
     {
-      "epoch": 2.4429133858267718,
-      "grad_norm": 12.571815490722656,
+      "epoch": 2.4448818897637796,
+      "grad_norm": 15.51160717010498,
       "learning_rate": 1.8238947953929436e-07,
-      "loss": 0.6039,
+      "loss": 1.0672,
       "step": 2484
     },
     {
-      "epoch": 2.4438976377952755,
-      "grad_norm": 10.60606575012207,
+      "epoch": 2.4458661417322833,
+      "grad_norm": 16.37097930908203,
       "learning_rate": 1.8180120106892685e-07,
-      "loss": 0.4512,
+      "loss": 0.7989,
       "step": 2485
     },
     {
-      "epoch": 2.4448818897637796,
-      "grad_norm": 9.76247787475586,
+      "epoch": 2.4468503937007875,
+      "grad_norm": 7.007961273193359,
       "learning_rate": 1.8121367970255475e-07,
-      "loss": 0.2862,
+      "loss": 0.1751,
       "step": 2486
     },
     {
-      "epoch": 2.4458661417322833,
-      "grad_norm": 11.170296669006348,
+      "epoch": 2.447834645669291,
+      "grad_norm": 14.998355865478516,
       "learning_rate": 1.806269168308089e-07,
-      "loss": 0.3739,
+      "loss": 0.8791,
       "step": 2487
     },
     {
-      "epoch": 2.4468503937007875,
-      "grad_norm": 9.177827835083008,
+      "epoch": 2.4488188976377954,
+      "grad_norm": 9.001073837280273,
       "learning_rate": 1.800409138425248e-07,
-      "loss": 0.328,
+      "loss": 0.2708,
       "step": 2488
     },
     {
-      "epoch": 2.447834645669291,
-      "grad_norm": 11.069591522216797,
+      "epoch": 2.449803149606299,
+      "grad_norm": 5.734395980834961,
       "learning_rate": 1.794556721247393e-07,
-      "loss": 0.344,
+      "loss": 0.1751,
       "step": 2489
     },
     {
-      "epoch": 2.4488188976377954,
-      "grad_norm": 13.294142723083496,
+      "epoch": 2.4507874015748032,
+      "grad_norm": 4.852680683135986,
       "learning_rate": 1.7887119306268743e-07,
-      "loss": 0.7,
+      "loss": 0.0918,
       "step": 2490
     },
     {
-      "epoch": 2.449803149606299,
-      "grad_norm": 11.407720565795898,
+      "epoch": 2.451771653543307,
+      "grad_norm": 7.260805606842041,
       "learning_rate": 1.7828747803979914e-07,
-      "loss": 0.6593,
+      "loss": 0.2444,
       "step": 2491
     },
     {
-      "epoch": 2.4507874015748032,
-      "grad_norm": 10.573090553283691,
+      "epoch": 2.452755905511811,
+      "grad_norm": 11.61798095703125,
       "learning_rate": 1.7770452843769567e-07,
-      "loss": 0.3137,
+      "loss": 0.5802,
       "step": 2492
     },
     {
-      "epoch": 2.451771653543307,
-      "grad_norm": 8.411659240722656,
+      "epoch": 2.453740157480315,
+      "grad_norm": 13.109360694885254,
       "learning_rate": 1.771223456361868e-07,
-      "loss": 0.2765,
+      "loss": 0.5564,
       "step": 2493
     },
     {
-      "epoch": 2.452755905511811,
-      "grad_norm": 8.66166877746582,
+      "epoch": 2.454724409448819,
+      "grad_norm": 5.175988674163818,
       "learning_rate": 1.765409310132673e-07,
-      "loss": 0.5268,
+      "loss": 0.1286,
       "step": 2494
     },
     {
-      "epoch": 2.453740157480315,
-      "grad_norm": 10.782097816467285,
+      "epoch": 2.4557086614173227,
+      "grad_norm": 7.814208030700684,
       "learning_rate": 1.7596028594511364e-07,
-      "loss": 0.5169,
+      "loss": 0.2139,
       "step": 2495
     },
     {
-      "epoch": 2.454724409448819,
-      "grad_norm": 11.39598274230957,
+      "epoch": 2.456692913385827,
+      "grad_norm": 12.347286224365234,
       "learning_rate": 1.753804118060808e-07,
-      "loss": 0.3784,
+      "loss": 0.6699,
       "step": 2496
     },
     {
-      "epoch": 2.4557086614173227,
-      "grad_norm": 8.448685646057129,
+      "epoch": 2.4576771653543306,
+      "grad_norm": 11.43953800201416,
       "learning_rate": 1.7480130996869905e-07,
-      "loss": 0.1408,
+      "loss": 0.4542,
       "step": 2497
     },
     {
-      "epoch": 2.456692913385827,
-      "grad_norm": 6.417379379272461,
+      "epoch": 2.4586614173228347,
+      "grad_norm": 9.194092750549316,
       "learning_rate": 1.7422298180367063e-07,
-      "loss": 0.2154,
+      "loss": 0.1878,
       "step": 2498
     },
     {
-      "epoch": 2.4576771653543306,
-      "grad_norm": 10.22805118560791,
+      "epoch": 2.4596456692913384,
+      "grad_norm": 5.263253688812256,
       "learning_rate": 1.736454286798665e-07,
-      "loss": 0.4219,
+      "loss": 0.1191,
       "step": 2499
     },
     {
-      "epoch": 2.4586614173228347,
-      "grad_norm": 9.10873031616211,
+      "epoch": 2.4606299212598426,
+      "grad_norm": 7.27897834777832,
       "learning_rate": 1.730686519643233e-07,
-      "loss": 0.3657,
+      "loss": 0.3498,
       "step": 2500
     },
     {
-      "epoch": 2.4596456692913384,
-      "grad_norm": 8.159028053283691,
+      "epoch": 2.4616141732283463,
+      "grad_norm": 7.4611310958862305,
       "learning_rate": 1.7249265302223998e-07,
-      "loss": 0.2133,
+      "loss": 0.2625,
       "step": 2501
     },
     {
-      "epoch": 2.4606299212598426,
-      "grad_norm": 13.752241134643555,
+      "epoch": 2.4625984251968505,
+      "grad_norm": 10.757491111755371,
       "learning_rate": 1.7191743321697403e-07,
-      "loss": 0.9923,
+      "loss": 0.4147,
       "step": 2502
     },
     {
-      "epoch": 2.4616141732283463,
-      "grad_norm": 5.2988786697387695,
+      "epoch": 2.463582677165354,
+      "grad_norm": 12.465801239013672,
       "learning_rate": 1.713429939100393e-07,
-      "loss": 0.1073,
+      "loss": 0.5618,
       "step": 2503
     },
     {
-      "epoch": 2.4625984251968505,
-      "grad_norm": 9.656152725219727,
+      "epoch": 2.4645669291338583,
+      "grad_norm": 8.395308494567871,
       "learning_rate": 1.7076933646110222e-07,
-      "loss": 0.2467,
+      "loss": 0.2314,
       "step": 2504
     },
     {
-      "epoch": 2.463582677165354,
-      "grad_norm": 12.059698104858398,
+      "epoch": 2.465551181102362,
+      "grad_norm": 9.32253360748291,
       "learning_rate": 1.7019646222797834e-07,
-      "loss": 0.609,
+      "loss": 0.4154,
       "step": 2505
     },
     {
-      "epoch": 2.4645669291338583,
-      "grad_norm": 9.394807815551758,
+      "epoch": 2.466535433070866,
+      "grad_norm": 8.11479377746582,
       "learning_rate": 1.696243725666296e-07,
-      "loss": 0.2827,
+      "loss": 0.2686,
       "step": 2506
     },
     {
-      "epoch": 2.465551181102362,
-      "grad_norm": 7.839410781860352,
+      "epoch": 2.46751968503937,
+      "grad_norm": 10.989008903503418,
       "learning_rate": 1.6905306883116077e-07,
-      "loss": 0.3833,
+      "loss": 0.385,
       "step": 2507
     },
     {
-      "epoch": 2.466535433070866,
-      "grad_norm": 8.78386116027832,
+      "epoch": 2.468503937007874,
+      "grad_norm": 11.128430366516113,
       "learning_rate": 1.684825523738165e-07,
-      "loss": 0.3798,
+      "loss": 0.3911,
       "step": 2508
     },
     {
-      "epoch": 2.46751968503937,
-      "grad_norm": 16.044845581054688,
+      "epoch": 2.469488188976378,
+      "grad_norm": 5.646144390106201,
       "learning_rate": 1.67912824544978e-07,
-      "loss": 0.5413,
+      "loss": 0.1549,
       "step": 2509
     },
     {
-      "epoch": 2.468503937007874,
-      "grad_norm": 6.673086643218994,
+      "epoch": 2.470472440944882,
+      "grad_norm": 6.215924263000488,
       "learning_rate": 1.6734388669315944e-07,
-      "loss": 0.2506,
+      "loss": 0.2425,
       "step": 2510
     },
     {
-      "epoch": 2.469488188976378,
-      "grad_norm": 5.23464298248291,
+      "epoch": 2.4714566929133857,
+      "grad_norm": 11.394367218017578,
       "learning_rate": 1.667757401650057e-07,
-      "loss": 0.102,
+      "loss": 0.402,
       "step": 2511
     },
     {
-      "epoch": 2.470472440944882,
-      "grad_norm": 8.43914794921875,
+      "epoch": 2.47244094488189,
+      "grad_norm": 12.046913146972656,
       "learning_rate": 1.6620838630528833e-07,
-      "loss": 0.223,
+      "loss": 0.5896,
       "step": 2512
     },
     {
-      "epoch": 2.4714566929133857,
-      "grad_norm": 14.900171279907227,
+      "epoch": 2.4734251968503935,
+      "grad_norm": 11.846915245056152,
       "learning_rate": 1.6564182645690283e-07,
-      "loss": 0.5648,
+      "loss": 0.5268,
       "step": 2513
     },
     {
-      "epoch": 2.47244094488189,
-      "grad_norm": 12.017776489257812,
+      "epoch": 2.4744094488188977,
+      "grad_norm": 8.131032943725586,
       "learning_rate": 1.650760619608653e-07,
-      "loss": 0.6887,
+      "loss": 0.3167,
       "step": 2514
     },
     {
-      "epoch": 2.4734251968503935,
-      "grad_norm": 12.223630905151367,
+      "epoch": 2.4753937007874014,
+      "grad_norm": 6.810969352722168,
       "learning_rate": 1.645110941563092e-07,
-      "loss": 0.5032,
+      "loss": 0.1611,
       "step": 2515
     },
     {
-      "epoch": 2.4744094488188977,
-      "grad_norm": 14.946356773376465,
+      "epoch": 2.4763779527559056,
+      "grad_norm": 12.168174743652344,
       "learning_rate": 1.639469243804822e-07,
-      "loss": 0.6455,
+      "loss": 0.6922,
       "step": 2516
     },
     {
-      "epoch": 2.4753937007874014,
-      "grad_norm": 6.877208232879639,
+      "epoch": 2.4773622047244093,
+      "grad_norm": 12.65621280670166,
       "learning_rate": 1.6338355396874332e-07,
-      "loss": 0.1955,
+      "loss": 0.5503,
       "step": 2517
     },
     {
-      "epoch": 2.4763779527559056,
-      "grad_norm": 14.754609107971191,
+      "epoch": 2.4783464566929134,
+      "grad_norm": 9.902356147766113,
       "learning_rate": 1.6282098425455936e-07,
-      "loss": 0.9092,
+      "loss": 0.4106,
       "step": 2518
     },
     {
-      "epoch": 2.4773622047244093,
-      "grad_norm": 7.898383140563965,
+      "epoch": 2.479330708661417,
+      "grad_norm": 8.353038787841797,
       "learning_rate": 1.622592165695019e-07,
-      "loss": 0.2389,
+      "loss": 0.4554,
       "step": 2519
     },
     {
-      "epoch": 2.4783464566929134,
-      "grad_norm": 10.59056568145752,
+      "epoch": 2.4803149606299213,
+      "grad_norm": 10.984722137451172,
       "learning_rate": 1.616982522432443e-07,
-      "loss": 0.5426,
+      "loss": 0.3079,
       "step": 2520
     },
     {
-      "epoch": 2.479330708661417,
-      "grad_norm": 9.697310447692871,
+      "epoch": 2.481299212598425,
+      "grad_norm": 10.451424598693848,
       "learning_rate": 1.6113809260355824e-07,
-      "loss": 0.6541,
+      "loss": 0.3354,
       "step": 2521
     },
     {
-      "epoch": 2.4803149606299213,
-      "grad_norm": 12.979537010192871,
+      "epoch": 2.482283464566929,
+      "grad_norm": 9.425240516662598,
       "learning_rate": 1.6057873897631103e-07,
-      "loss": 0.5123,
+      "loss": 0.288,
       "step": 2522
     },
     {
-      "epoch": 2.481299212598425,
-      "grad_norm": 11.808232307434082,
+      "epoch": 2.483267716535433,
+      "grad_norm": 5.637824058532715,
       "learning_rate": 1.600201926854616e-07,
-      "loss": 0.5188,
+      "loss": 0.1446,
       "step": 2523
     },
     {
-      "epoch": 2.482283464566929,
-      "grad_norm": 9.810900688171387,
+      "epoch": 2.484251968503937,
+      "grad_norm": 15.202845573425293,
       "learning_rate": 1.5946245505305863e-07,
-      "loss": 0.3183,
+      "loss": 0.8988,
       "step": 2524
     },
     {
-      "epoch": 2.483267716535433,
-      "grad_norm": 10.592032432556152,
+      "epoch": 2.4852362204724407,
+      "grad_norm": 13.908705711364746,
       "learning_rate": 1.5890552739923647e-07,
-      "loss": 0.2874,
+      "loss": 0.7265,
       "step": 2525
     },
     {
-      "epoch": 2.484251968503937,
-      "grad_norm": 11.300968170166016,
+      "epoch": 2.486220472440945,
+      "grad_norm": 6.362472057342529,
       "learning_rate": 1.5834941104221217e-07,
-      "loss": 0.6203,
+      "loss": 0.1949,
       "step": 2526
     },
     {
-      "epoch": 2.4852362204724407,
-      "grad_norm": 9.015886306762695,
+      "epoch": 2.487204724409449,
+      "grad_norm": 11.73279094696045,
       "learning_rate": 1.577941072982827e-07,
-      "loss": 0.3392,
+      "loss": 0.8494,
       "step": 2527
     },
     {
-      "epoch": 2.486220472440945,
-      "grad_norm": 5.7811689376831055,
+      "epoch": 2.4881889763779528,
+      "grad_norm": 8.031981468200684,
       "learning_rate": 1.5723961748182163e-07,
-      "loss": 0.1664,
+      "loss": 0.3557,
       "step": 2528
     },
     {
-      "epoch": 2.487204724409449,
-      "grad_norm": 11.204570770263672,
+      "epoch": 2.4891732283464565,
+      "grad_norm": 11.21584415435791,
       "learning_rate": 1.5668594290527568e-07,
-      "loss": 0.5371,
+      "loss": 0.4228,
       "step": 2529
     },
     {
-      "epoch": 2.4881889763779528,
-      "grad_norm": 16.38753890991211,
+      "epoch": 2.4901574803149606,
+      "grad_norm": 8.426931381225586,
       "learning_rate": 1.5613308487916242e-07,
-      "loss": 0.7918,
+      "loss": 0.259,
       "step": 2530
     },
     {
-      "epoch": 2.4891732283464565,
-      "grad_norm": 6.875202178955078,
+      "epoch": 2.491141732283465,
+      "grad_norm": 8.10767650604248,
       "learning_rate": 1.5558104471206626e-07,
-      "loss": 0.2345,
+      "loss": 0.2982,
       "step": 2531
     },
     {
-      "epoch": 2.4901574803149606,
-      "grad_norm": 8.88460636138916,
+      "epoch": 2.4921259842519685,
+      "grad_norm": 13.152838706970215,
       "learning_rate": 1.5502982371063602e-07,
-      "loss": 0.2692,
+      "loss": 0.6837,
       "step": 2532
     },
     {
-      "epoch": 2.491141732283465,
-      "grad_norm": 12.828963279724121,
+      "epoch": 2.4931102362204722,
+      "grad_norm": 9.704029083251953,
       "learning_rate": 1.5447942317958161e-07,
-      "loss": 0.4839,
+      "loss": 0.4417,
       "step": 2533
     },
     {
-      "epoch": 2.4921259842519685,
-      "grad_norm": 9.021425247192383,
+      "epoch": 2.4940944881889764,
+      "grad_norm": 8.404932975769043,
       "learning_rate": 1.539298444216708e-07,
-      "loss": 0.3797,
+      "loss": 0.4822,
       "step": 2534
     },
     {
-      "epoch": 2.4931102362204722,
-      "grad_norm": 7.733816623687744,
+      "epoch": 2.4950787401574805,
+      "grad_norm": 8.78940486907959,
       "learning_rate": 1.5338108873772638e-07,
-      "loss": 0.5387,
+      "loss": 0.3461,
       "step": 2535
     },
     {
-      "epoch": 2.4940944881889764,
-      "grad_norm": 12.049376487731934,
+      "epoch": 2.4960629921259843,
+      "grad_norm": 6.783337593078613,
       "learning_rate": 1.5283315742662288e-07,
-      "loss": 0.3812,
+      "loss": 0.2396,
       "step": 2536
     },
     {
-      "epoch": 2.4950787401574805,
-      "grad_norm": 7.660175323486328,
+      "epoch": 2.497047244094488,
+      "grad_norm": 10.388428688049316,
       "learning_rate": 1.5228605178528368e-07,
-      "loss": 0.2541,
+      "loss": 0.4991,
       "step": 2537
     },
     {
-      "epoch": 2.4960629921259843,
-      "grad_norm": 6.382974624633789,
+      "epoch": 2.498031496062992,
+      "grad_norm": 6.733151912689209,
       "learning_rate": 1.5173977310867775e-07,
-      "loss": 0.1971,
+      "loss": 0.3198,
       "step": 2538
     },
     {
-      "epoch": 2.497047244094488,
-      "grad_norm": 12.477048873901367,
+      "epoch": 2.4990157480314963,
+      "grad_norm": 8.143392562866211,
       "learning_rate": 1.5119432268981675e-07,
-      "loss": 0.5732,
+      "loss": 0.3665,
       "step": 2539
     },
     {
-      "epoch": 2.498031496062992,
-      "grad_norm": 15.173735618591309,
+      "epoch": 2.5,
+      "grad_norm": 6.186513423919678,
       "learning_rate": 1.5064970181975204e-07,
-      "loss": 0.6801,
+      "loss": 0.176,
       "step": 2540
     },
     {
-      "epoch": 2.4990157480314963,
-      "grad_norm": 13.453434944152832,
+      "epoch": 2.5009842519685037,
+      "grad_norm": 8.67770767211914,
       "learning_rate": 1.5010591178757092e-07,
-      "loss": 0.7655,
+      "loss": 0.2136,
       "step": 2541
     },
     {
-      "epoch": 2.5,
-      "grad_norm": 7.837796211242676,
+      "epoch": 2.501968503937008,
+      "grad_norm": 11.920140266418457,
       "learning_rate": 1.4956295388039484e-07,
-      "loss": 0.1766,
+      "loss": 0.5891,
       "step": 2542
     },
     {
-      "epoch": 2.5009842519685037,
-      "grad_norm": 14.00476360321045,
+      "epoch": 2.502952755905512,
+      "grad_norm": 11.652994155883789,
       "learning_rate": 1.4902082938337527e-07,
-      "loss": 0.3666,
+      "loss": 0.5309,
       "step": 2543
     },
     {
-      "epoch": 2.501968503937008,
-      "grad_norm": 9.707633972167969,
+      "epoch": 2.5039370078740157,
+      "grad_norm": 9.029703140258789,
       "learning_rate": 1.4847953957969112e-07,
-      "loss": 0.2594,
+      "loss": 0.3781,
       "step": 2544
     },
     {
-      "epoch": 2.502952755905512,
-      "grad_norm": 10.913497924804688,
+      "epoch": 2.5049212598425195,
+      "grad_norm": 10.146659851074219,
       "learning_rate": 1.479390857505457e-07,
-      "loss": 0.5756,
+      "loss": 0.7324,
       "step": 2545
     },
     {
-      "epoch": 2.5039370078740157,
-      "grad_norm": 5.065262794494629,
+      "epoch": 2.5059055118110236,
+      "grad_norm": 8.75976848602295,
       "learning_rate": 1.4739946917516362e-07,
-      "loss": 0.1573,
+      "loss": 0.433,
       "step": 2546
     },
     {
-      "epoch": 2.5049212598425195,
-      "grad_norm": 12.768359184265137,
+      "epoch": 2.5068897637795278,
+      "grad_norm": 13.880877494812012,
       "learning_rate": 1.4686069113078744e-07,
-      "loss": 0.7746,
+      "loss": 0.7656,
       "step": 2547
     },
     {
-      "epoch": 2.5059055118110236,
-      "grad_norm": 25.458555221557617,
+      "epoch": 2.5078740157480315,
+      "grad_norm": 13.18405818939209,
       "learning_rate": 1.4632275289267535e-07,
-      "loss": 1.303,
+      "loss": 0.6324,
       "step": 2548
     },
     {
-      "epoch": 2.5068897637795278,
-      "grad_norm": 7.535705089569092,
+      "epoch": 2.508858267716535,
+      "grad_norm": 4.782319068908691,
       "learning_rate": 1.457856557340976e-07,
-      "loss": 0.2377,
+      "loss": 0.1181,
       "step": 2549
     },
     {
-      "epoch": 2.5078740157480315,
-      "grad_norm": 6.729396820068359,
+      "epoch": 2.5098425196850394,
+      "grad_norm": 7.828704357147217,
       "learning_rate": 1.4524940092633376e-07,
-      "loss": 0.2366,
+      "loss": 0.3459,
       "step": 2550
     },
     {
-      "epoch": 2.508858267716535,
-      "grad_norm": 6.5007429122924805,
+      "epoch": 2.5108267716535435,
+      "grad_norm": 5.572178363800049,
       "learning_rate": 1.447139897386694e-07,
-      "loss": 0.1585,
+      "loss": 0.1759,
       "step": 2551
     },
     {
-      "epoch": 2.5098425196850394,
-      "grad_norm": 13.314377784729004,
+      "epoch": 2.5118110236220472,
+      "grad_norm": 6.532750606536865,
       "learning_rate": 1.4417942343839348e-07,
-      "loss": 0.8265,
+      "loss": 0.1898,
       "step": 2552
     },
     {
-      "epoch": 2.5108267716535435,
-      "grad_norm": 9.079033851623535,
+      "epoch": 2.512795275590551,
+      "grad_norm": 5.581180095672607,
       "learning_rate": 1.4364570329079504e-07,
-      "loss": 0.3276,
+      "loss": 0.1363,
       "step": 2553
     },
     {
-      "epoch": 2.5118110236220472,
-      "grad_norm": 10.1290864944458,
+      "epoch": 2.513779527559055,
+      "grad_norm": 6.385770320892334,
       "learning_rate": 1.4311283055916042e-07,
-      "loss": 0.437,
+      "loss": 0.1972,
       "step": 2554
     },
     {
-      "epoch": 2.512795275590551,
-      "grad_norm": 11.950562477111816,
+      "epoch": 2.5147637795275593,
+      "grad_norm": 8.028898239135742,
       "learning_rate": 1.4258080650477008e-07,
-      "loss": 0.5382,
+      "loss": 0.23,
       "step": 2555
     },
     {
-      "epoch": 2.513779527559055,
-      "grad_norm": 14.388399124145508,
+      "epoch": 2.515748031496063,
+      "grad_norm": 12.586973190307617,
       "learning_rate": 1.4204963238689582e-07,
-      "loss": 0.3736,
+      "loss": 0.6073,
       "step": 2556
     },
     {
-      "epoch": 2.5147637795275593,
-      "grad_norm": 6.537032127380371,
+      "epoch": 2.5167322834645667,
+      "grad_norm": 7.130866050720215,
       "learning_rate": 1.4151930946279746e-07,
-      "loss": 0.1608,
+      "loss": 0.2138,
       "step": 2557
     },
     {
-      "epoch": 2.515748031496063,
-      "grad_norm": 10.22529125213623,
+      "epoch": 2.517716535433071,
+      "grad_norm": 8.065289497375488,
       "learning_rate": 1.4098983898772037e-07,
-      "loss": 0.4933,
+      "loss": 0.3059,
       "step": 2558
     },
     {
-      "epoch": 2.5167322834645667,
-      "grad_norm": 8.023038864135742,
+      "epoch": 2.518700787401575,
+      "grad_norm": 7.807585716247559,
       "learning_rate": 1.404612222148921e-07,
-      "loss": 0.3567,
+      "loss": 0.37,
       "step": 2559
     },
     {
-      "epoch": 2.517716535433071,
-      "grad_norm": 10.37587833404541,
+      "epoch": 2.5196850393700787,
+      "grad_norm": 13.064702987670898,
       "learning_rate": 1.3993346039551943e-07,
-      "loss": 0.5998,
+      "loss": 0.4976,
       "step": 2560
     },
     {
-      "epoch": 2.518700787401575,
-      "grad_norm": 5.052724361419678,
+      "epoch": 2.5206692913385824,
+      "grad_norm": 7.2660698890686035,
       "learning_rate": 1.3940655477878572e-07,
-      "loss": 0.1126,
+      "loss": 0.2644,
       "step": 2561
     },
     {
-      "epoch": 2.5196850393700787,
-      "grad_norm": 9.678995132446289,
+      "epoch": 2.5216535433070866,
+      "grad_norm": 8.346268653869629,
       "learning_rate": 1.3888050661184755e-07,
-      "loss": 0.4245,
+      "loss": 0.3422,
       "step": 2562
     },
     {
-      "epoch": 2.5206692913385824,
-      "grad_norm": 8.200692176818848,
+      "epoch": 2.5226377952755907,
+      "grad_norm": 6.177550792694092,
       "learning_rate": 1.3835531713983213e-07,
-      "loss": 0.2628,
+      "loss": 0.203,
       "step": 2563
     },
     {
-      "epoch": 2.5216535433070866,
-      "grad_norm": 8.967484474182129,
+      "epoch": 2.5236220472440944,
+      "grad_norm": 8.378596305847168,
       "learning_rate": 1.3783098760583417e-07,
-      "loss": 0.3958,
+      "loss": 0.2347,
       "step": 2564
     },
     {
-      "epoch": 2.5226377952755907,
-      "grad_norm": 8.796217918395996,
+      "epoch": 2.524606299212598,
+      "grad_norm": 10.247299194335938,
       "learning_rate": 1.3730751925091263e-07,
-      "loss": 0.2905,
+      "loss": 0.4776,
       "step": 2565
     },
     {
-      "epoch": 2.5236220472440944,
-      "grad_norm": 6.851153373718262,
+      "epoch": 2.5255905511811023,
+      "grad_norm": 8.838934898376465,
       "learning_rate": 1.3678491331408854e-07,
-      "loss": 0.2747,
+      "loss": 0.3452,
       "step": 2566
     },
     {
-      "epoch": 2.524606299212598,
-      "grad_norm": 7.8298258781433105,
+      "epoch": 2.5265748031496065,
+      "grad_norm": 10.743462562561035,
       "learning_rate": 1.3626317103234146e-07,
-      "loss": 0.2156,
+      "loss": 0.5831,
       "step": 2567
     },
     {
-      "epoch": 2.5255905511811023,
-      "grad_norm": 9.772421836853027,
+      "epoch": 2.52755905511811,
+      "grad_norm": 8.789103507995605,
       "learning_rate": 1.3574229364060663e-07,
-      "loss": 0.4541,
+      "loss": 0.4885,
       "step": 2568
     },
     {
-      "epoch": 2.5265748031496065,
-      "grad_norm": 8.752959251403809,
+      "epoch": 2.528543307086614,
+      "grad_norm": 5.7821173667907715,
       "learning_rate": 1.3522228237177238e-07,
-      "loss": 0.4792,
+      "loss": 0.1946,
       "step": 2569
     },
     {
-      "epoch": 2.52755905511811,
-      "grad_norm": 8.197463989257812,
+      "epoch": 2.529527559055118,
+      "grad_norm": 12.357575416564941,
       "learning_rate": 1.347031384566767e-07,
-      "loss": 0.2525,
+      "loss": 0.6301,
       "step": 2570
     },
     {
-      "epoch": 2.528543307086614,
-      "grad_norm": 6.956662178039551,
+      "epoch": 2.530511811023622,
+      "grad_norm": 12.54269790649414,
       "learning_rate": 1.3418486312410484e-07,
-      "loss": 0.1934,
+      "loss": 0.5759,
       "step": 2571
     },
     {
-      "epoch": 2.529527559055118,
-      "grad_norm": 5.082679271697998,
+      "epoch": 2.531496062992126,
+      "grad_norm": 8.5524320602417,
       "learning_rate": 1.3366745760078604e-07,
-      "loss": 0.1325,
+      "loss": 0.352,
       "step": 2572
     },
     {
-      "epoch": 2.530511811023622,
-      "grad_norm": 11.066061019897461,
+      "epoch": 2.53248031496063,
+      "grad_norm": 9.425482749938965,
       "learning_rate": 1.3315092311139072e-07,
-      "loss": 0.3603,
+      "loss": 0.3132,
       "step": 2573
     },
     {
-      "epoch": 2.531496062992126,
-      "grad_norm": 7.229071617126465,
+      "epoch": 2.533464566929134,
+      "grad_norm": 11.823163986206055,
       "learning_rate": 1.3263526087852766e-07,
-      "loss": 0.3196,
+      "loss": 0.4923,
       "step": 2574
     },
     {
-      "epoch": 2.53248031496063,
-      "grad_norm": 10.58609390258789,
+      "epoch": 2.534448818897638,
+      "grad_norm": 11.727007865905762,
       "learning_rate": 1.321204721227411e-07,
-      "loss": 0.4006,
+      "loss": 0.7069,
       "step": 2575
     },
     {
-      "epoch": 2.533464566929134,
-      "grad_norm": 11.721096992492676,
+      "epoch": 2.5354330708661417,
+      "grad_norm": 9.601505279541016,
       "learning_rate": 1.316065580625077e-07,
-      "loss": 0.6041,
+      "loss": 0.3848,
       "step": 2576
     },
     {
-      "epoch": 2.534448818897638,
-      "grad_norm": 11.265833854675293,
+      "epoch": 2.536417322834646,
+      "grad_norm": 14.605145454406738,
       "learning_rate": 1.3109351991423402e-07,
-      "loss": 0.6093,
+      "loss": 0.8719,
       "step": 2577
     },
     {
-      "epoch": 2.5354330708661417,
-      "grad_norm": 19.454679489135742,
+      "epoch": 2.5374015748031495,
+      "grad_norm": 7.809975624084473,
       "learning_rate": 1.3058135889225286e-07,
-      "loss": 1.0304,
+      "loss": 0.4992,
       "step": 2578
     },
     {
-      "epoch": 2.536417322834646,
-      "grad_norm": 12.229068756103516,
+      "epoch": 2.5383858267716537,
+      "grad_norm": 9.553762435913086,
       "learning_rate": 1.3007007620882145e-07,
-      "loss": 0.591,
+      "loss": 0.4379,
       "step": 2579
     },
     {
-      "epoch": 2.5374015748031495,
-      "grad_norm": 12.774563789367676,
+      "epoch": 2.5393700787401574,
+      "grad_norm": 9.366168975830078,
       "learning_rate": 1.2955967307411777e-07,
-      "loss": 0.5078,
+      "loss": 0.3223,
       "step": 2580
     },
     {
-      "epoch": 2.5383858267716537,
-      "grad_norm": 10.360858917236328,
+      "epoch": 2.5403543307086616,
+      "grad_norm": 9.658661842346191,
       "learning_rate": 1.290501506962381e-07,
-      "loss": 0.4428,
+      "loss": 0.4615,
       "step": 2581
     },
     {
-      "epoch": 2.5393700787401574,
-      "grad_norm": 14.671048164367676,
+      "epoch": 2.5413385826771653,
+      "grad_norm": 7.113828659057617,
       "learning_rate": 1.2854151028119375e-07,
-      "loss": 0.5756,
+      "loss": 0.2853,
       "step": 2582
     },
     {
-      "epoch": 2.5403543307086616,
-      "grad_norm": 8.14897632598877,
+      "epoch": 2.5423228346456694,
+      "grad_norm": 10.615381240844727,
       "learning_rate": 1.2803375303290884e-07,
-      "loss": 0.2725,
+      "loss": 0.5365,
       "step": 2583
     },
     {
-      "epoch": 2.5413385826771653,
-      "grad_norm": 14.281302452087402,
+      "epoch": 2.543307086614173,
+      "grad_norm": 10.920248031616211,
       "learning_rate": 1.2752688015321684e-07,
-      "loss": 0.6336,
+      "loss": 0.4741,
       "step": 2584
     },
     {
-      "epoch": 2.5423228346456694,
-      "grad_norm": 9.67027759552002,
+      "epoch": 2.5442913385826773,
+      "grad_norm": 4.072298526763916,
       "learning_rate": 1.2702089284185817e-07,
-      "loss": 0.3588,
+      "loss": 0.1163,
       "step": 2585
     },
     {
-      "epoch": 2.543307086614173,
-      "grad_norm": 14.037250518798828,
+      "epoch": 2.545275590551181,
+      "grad_norm": 6.3231658935546875,
       "learning_rate": 1.265157922964767e-07,
-      "loss": 0.4981,
+      "loss": 0.1334,
       "step": 2586
     },
     {
-      "epoch": 2.5442913385826773,
-      "grad_norm": 6.424685955047607,
+      "epoch": 2.546259842519685,
+      "grad_norm": 8.722452163696289,
       "learning_rate": 1.2601157971261788e-07,
-      "loss": 0.1576,
+      "loss": 0.2915,
       "step": 2587
     },
     {
-      "epoch": 2.545275590551181,
-      "grad_norm": 6.243099212646484,
+      "epoch": 2.547244094488189,
+      "grad_norm": 8.990911483764648,
       "learning_rate": 1.255082562837252e-07,
-      "loss": 0.2362,
+      "loss": 0.3424,
       "step": 2588
     },
     {
-      "epoch": 2.546259842519685,
-      "grad_norm": 8.598806381225586,
+      "epoch": 2.548228346456693,
+      "grad_norm": 7.959079265594482,
       "learning_rate": 1.250058232011375e-07,
-      "loss": 0.3347,
+      "loss": 0.2587,
       "step": 2589
     },
     {
-      "epoch": 2.547244094488189,
-      "grad_norm": 8.309163093566895,
+      "epoch": 2.5492125984251968,
+      "grad_norm": 9.749112129211426,
       "learning_rate": 1.2450428165408641e-07,
-      "loss": 0.3335,
+      "loss": 0.3412,
       "step": 2590
     },
     {
-      "epoch": 2.548228346456693,
-      "grad_norm": 9.962733268737793,
+      "epoch": 2.550196850393701,
+      "grad_norm": 13.236554145812988,
       "learning_rate": 1.2400363282969322e-07,
-      "loss": 0.4493,
+      "loss": 0.7519,
       "step": 2591
     },
     {
-      "epoch": 2.5492125984251968,
-      "grad_norm": 8.566224098205566,
+      "epoch": 2.5511811023622046,
+      "grad_norm": 10.591288566589355,
       "learning_rate": 1.2350387791296617e-07,
-      "loss": 0.2456,
+      "loss": 0.6034,
       "step": 2592
     },
     {
-      "epoch": 2.550196850393701,
-      "grad_norm": 10.342599868774414,
+      "epoch": 2.552165354330709,
+      "grad_norm": 9.161969184875488,
       "learning_rate": 1.2300501808679776e-07,
-      "loss": 0.3954,
+      "loss": 0.2687,
       "step": 2593
     },
     {
-      "epoch": 2.5511811023622046,
-      "grad_norm": 5.752167701721191,
+      "epoch": 2.5531496062992125,
+      "grad_norm": 8.445438385009766,
       "learning_rate": 1.2250705453196177e-07,
-      "loss": 0.1741,
+      "loss": 0.2823,
       "step": 2594
     },
     {
-      "epoch": 2.552165354330709,
-      "grad_norm": 8.814495086669922,
+      "epoch": 2.5541338582677167,
+      "grad_norm": 8.941618919372559,
       "learning_rate": 1.2200998842711062e-07,
-      "loss": 0.3343,
+      "loss": 0.2531,
       "step": 2595
     },
     {
-      "epoch": 2.5531496062992125,
-      "grad_norm": 7.947423934936523,
+      "epoch": 2.5551181102362204,
+      "grad_norm": 13.971638679504395,
       "learning_rate": 1.2151382094877245e-07,
-      "loss": 0.1977,
+      "loss": 0.748,
       "step": 2596
     },
     {
-      "epoch": 2.5541338582677167,
-      "grad_norm": 16.38239288330078,
+      "epoch": 2.5561023622047245,
+      "grad_norm": 6.199235439300537,
       "learning_rate": 1.2101855327134849e-07,
-      "loss": 0.8805,
+      "loss": 0.1596,
       "step": 2597
     },
     {
-      "epoch": 2.5551181102362204,
-      "grad_norm": 12.04304313659668,
+      "epoch": 2.5570866141732282,
+      "grad_norm": 13.273075103759766,
       "learning_rate": 1.2052418656711007e-07,
-      "loss": 0.5718,
+      "loss": 0.7275,
       "step": 2598
     },
     {
-      "epoch": 2.5561023622047245,
-      "grad_norm": 7.0162787437438965,
+      "epoch": 2.5580708661417324,
+      "grad_norm": 8.330723762512207,
       "learning_rate": 1.200307220061961e-07,
-      "loss": 0.1922,
+      "loss": 0.2608,
       "step": 2599
     },
     {
-      "epoch": 2.5570866141732282,
-      "grad_norm": 8.008231163024902,
+      "epoch": 2.559055118110236,
+      "grad_norm": 10.253194808959961,
       "learning_rate": 1.1953816075661008e-07,
-      "loss": 0.307,
+      "loss": 0.4607,
       "step": 2600
     },
     {
-      "epoch": 2.5580708661417324,
-      "grad_norm": 7.720792293548584,
+      "epoch": 2.5600393700787403,
+      "grad_norm": 13.266863822937012,
       "learning_rate": 1.1904650398421742e-07,
-      "loss": 0.1893,
+      "loss": 0.7355,
       "step": 2601
     },
     {
-      "epoch": 2.5580708661417324,
-      "eval_Qnli-dev_cosine_accuracy": 0.6953125,
-      "eval_Qnli-dev_cosine_accuracy_threshold": 0.8067620396614075,
-      "eval_Qnli-dev_cosine_ap": 0.7043241142333805,
-      "eval_Qnli-dev_cosine_f1": 0.6953748006379585,
-      "eval_Qnli-dev_cosine_f1_threshold": 0.7095954418182373,
-      "eval_Qnli-dev_cosine_precision": 0.5575447570332481,
-      "eval_Qnli-dev_cosine_recall": 0.923728813559322,
-      "eval_Qnli-dev_dot_accuracy": 0.6875,
-      "eval_Qnli-dev_dot_accuracy_threshold": 612.6456298828125,
-      "eval_Qnli-dev_dot_ap": 0.7015226184100491,
-      "eval_Qnli-dev_dot_f1": 0.6943999999999998,
-      "eval_Qnli-dev_dot_f1_threshold": 541.830322265625,
-      "eval_Qnli-dev_dot_precision": 0.5578406169665809,
-      "eval_Qnli-dev_dot_recall": 0.9194915254237288,
-      "eval_Qnli-dev_euclidean_accuracy": 0.701171875,
-      "eval_Qnli-dev_euclidean_accuracy_threshold": 17.201496124267578,
-      "eval_Qnli-dev_euclidean_ap": 0.7050333505348279,
-      "eval_Qnli-dev_euclidean_f1": 0.6953748006379585,
-      "eval_Qnli-dev_euclidean_f1_threshold": 20.963668823242188,
-      "eval_Qnli-dev_euclidean_precision": 0.5575447570332481,
-      "eval_Qnli-dev_euclidean_recall": 0.923728813559322,
-      "eval_Qnli-dev_manhattan_accuracy": 0.6953125,
-      "eval_Qnli-dev_manhattan_accuracy_threshold": 364.14117431640625,
-      "eval_Qnli-dev_manhattan_ap": 0.7137495271732022,
-      "eval_Qnli-dev_manhattan_f1": 0.694136291600634,
-      "eval_Qnli-dev_manhattan_f1_threshold": 444.42218017578125,
-      "eval_Qnli-dev_manhattan_precision": 0.5544303797468354,
-      "eval_Qnli-dev_manhattan_recall": 0.9279661016949152,
-      "eval_Qnli-dev_max_accuracy": 0.701171875,
-      "eval_Qnli-dev_max_accuracy_threshold": 612.6456298828125,
-      "eval_Qnli-dev_max_ap": 0.7137495271732022,
-      "eval_Qnli-dev_max_f1": 0.6953748006379585,
-      "eval_Qnli-dev_max_f1_threshold": 541.830322265625,
-      "eval_Qnli-dev_max_precision": 0.5578406169665809,
-      "eval_Qnli-dev_max_recall": 0.9279661016949152,
-      "eval_allNLI-dev_cosine_accuracy": 0.697265625,
-      "eval_allNLI-dev_cosine_accuracy_threshold": 0.912941038608551,
-      "eval_allNLI-dev_cosine_ap": 0.5170912462717843,
-      "eval_allNLI-dev_cosine_f1": 0.558891454965358,
-      "eval_allNLI-dev_cosine_f1_threshold": 0.826492190361023,
-      "eval_allNLI-dev_cosine_precision": 0.4653846153846154,
-      "eval_allNLI-dev_cosine_recall": 0.6994219653179191,
-      "eval_allNLI-dev_dot_accuracy": 0.6953125,
-      "eval_allNLI-dev_dot_accuracy_threshold": 699.905517578125,
-      "eval_allNLI-dev_dot_ap": 0.5173897942840021,
-      "eval_allNLI-dev_dot_f1": 0.5568445475638051,
-      "eval_allNLI-dev_dot_f1_threshold": 630.53271484375,
-      "eval_allNLI-dev_dot_precision": 0.46511627906976744,
-      "eval_allNLI-dev_dot_recall": 0.6936416184971098,
-      "eval_allNLI-dev_euclidean_accuracy": 0.697265625,
-      "eval_allNLI-dev_euclidean_accuracy_threshold": 11.538026809692383,
-      "eval_allNLI-dev_euclidean_ap": 0.5171319156824372,
-      "eval_allNLI-dev_euclidean_f1": 0.5576036866359447,
-      "eval_allNLI-dev_euclidean_f1_threshold": 16.292882919311523,
-      "eval_allNLI-dev_euclidean_precision": 0.46360153256704983,
-      "eval_allNLI-dev_euclidean_recall": 0.6994219653179191,
-      "eval_allNLI-dev_manhattan_accuracy": 0.689453125,
-      "eval_allNLI-dev_manhattan_accuracy_threshold": 248.03265380859375,
-      "eval_allNLI-dev_manhattan_ap": 0.5172164569460806,
-      "eval_allNLI-dev_manhattan_f1": 0.5596330275229358,
-      "eval_allNLI-dev_manhattan_f1_threshold": 344.65313720703125,
-      "eval_allNLI-dev_manhattan_precision": 0.46387832699619774,
-      "eval_allNLI-dev_manhattan_recall": 0.7052023121387283,
-      "eval_allNLI-dev_max_accuracy": 0.697265625,
-      "eval_allNLI-dev_max_accuracy_threshold": 699.905517578125,
-      "eval_allNLI-dev_max_ap": 0.5173897942840021,
-      "eval_allNLI-dev_max_f1": 0.5596330275229358,
-      "eval_allNLI-dev_max_f1_threshold": 630.53271484375,
-      "eval_allNLI-dev_max_precision": 0.4653846153846154,
-      "eval_allNLI-dev_max_recall": 0.7052023121387283,
-      "eval_loss": 0.901435136795044,
-      "eval_runtime": 54.1536,
-      "eval_samples_per_second": 30.727,
-      "eval_sequential_score": 0.7137495271732022,
-      "eval_steps_per_second": 0.129,
-      "eval_sts-test_pearson_cosine": 0.6591049782717581,
-      "eval_sts-test_pearson_dot": 0.6585371125233661,
-      "eval_sts-test_pearson_euclidean": 0.6710226989754035,
-      "eval_sts-test_pearson_manhattan": 0.674239366272805,
-      "eval_sts-test_pearson_max": 0.674239366272805,
-      "eval_sts-test_spearman_cosine": 0.6655085832559531,
-      "eval_sts-test_spearman_dot": 0.6647672338597431,
-      "eval_sts-test_spearman_euclidean": 0.6657110379624879,
-      "eval_sts-test_spearman_manhattan": 0.6690548074623918,
-      "eval_sts-test_spearman_max": 0.6690548074623918,
+      "epoch": 2.5600393700787403,
+      "eval_Qnli-dev_cosine_accuracy": 0.67578125,
+      "eval_Qnli-dev_cosine_accuracy_threshold": 0.7981740236282349,
+      "eval_Qnli-dev_cosine_ap": 0.7024537673431017,
+      "eval_Qnli-dev_cosine_f1": 0.6933797909407665,
+      "eval_Qnli-dev_cosine_f1_threshold": 0.744655430316925,
+      "eval_Qnli-dev_cosine_precision": 0.5887573964497042,
+      "eval_Qnli-dev_cosine_recall": 0.8432203389830508,
+      "eval_Qnli-dev_dot_accuracy": 0.669921875,
+      "eval_Qnli-dev_dot_accuracy_threshold": 610.97509765625,
+      "eval_Qnli-dev_dot_ap": 0.7000264168874522,
+      "eval_Qnli-dev_dot_f1": 0.6900175131348512,
+      "eval_Qnli-dev_dot_f1_threshold": 568.671875,
+      "eval_Qnli-dev_dot_precision": 0.5880597014925373,
+      "eval_Qnli-dev_dot_recall": 0.8347457627118644,
+      "eval_Qnli-dev_euclidean_accuracy": 0.673828125,
+      "eval_Qnli-dev_euclidean_accuracy_threshold": 17.699954986572266,
+      "eval_Qnli-dev_euclidean_ap": 0.7027015569550685,
+      "eval_Qnli-dev_euclidean_f1": 0.6933797909407665,
+      "eval_Qnli-dev_euclidean_f1_threshold": 19.75226402282715,
+      "eval_Qnli-dev_euclidean_precision": 0.5887573964497042,
+      "eval_Qnli-dev_euclidean_recall": 0.8432203389830508,
+      "eval_Qnli-dev_manhattan_accuracy": 0.69140625,
+      "eval_Qnli-dev_manhattan_accuracy_threshold": 383.7911682128906,
+      "eval_Qnli-dev_manhattan_ap": 0.710276735549099,
+      "eval_Qnli-dev_manhattan_f1": 0.6901408450704225,
+      "eval_Qnli-dev_manhattan_f1_threshold": 413.4296875,
+      "eval_Qnli-dev_manhattan_precision": 0.5903614457831325,
+      "eval_Qnli-dev_manhattan_recall": 0.8305084745762712,
+      "eval_Qnli-dev_max_accuracy": 0.69140625,
+      "eval_Qnli-dev_max_accuracy_threshold": 610.97509765625,
+      "eval_Qnli-dev_max_ap": 0.710276735549099,
+      "eval_Qnli-dev_max_f1": 0.6933797909407665,
+      "eval_Qnli-dev_max_f1_threshold": 568.671875,
+      "eval_Qnli-dev_max_precision": 0.5903614457831325,
+      "eval_Qnli-dev_max_recall": 0.8432203389830508,
+      "eval_allNLI-dev_cosine_accuracy": 0.689453125,
+      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9239224195480347,
+      "eval_allNLI-dev_cosine_ap": 0.5100510670205529,
+      "eval_allNLI-dev_cosine_f1": 0.5515695067264574,
+      "eval_allNLI-dev_cosine_f1_threshold": 0.8346847295761108,
+      "eval_allNLI-dev_cosine_precision": 0.45054945054945056,
+      "eval_allNLI-dev_cosine_recall": 0.7109826589595376,
+      "eval_allNLI-dev_dot_accuracy": 0.69140625,
+      "eval_allNLI-dev_dot_accuracy_threshold": 704.034912109375,
+      "eval_allNLI-dev_dot_ap": 0.5088577654112999,
+      "eval_allNLI-dev_dot_f1": 0.5517241379310345,
+      "eval_allNLI-dev_dot_f1_threshold": 620.714599609375,
+      "eval_allNLI-dev_dot_precision": 0.425,
+      "eval_allNLI-dev_dot_recall": 0.7861271676300579,
+      "eval_allNLI-dev_euclidean_accuracy": 0.69140625,
+      "eval_allNLI-dev_euclidean_accuracy_threshold": 11.195505142211914,
+      "eval_allNLI-dev_euclidean_ap": 0.5098181318339068,
+      "eval_allNLI-dev_euclidean_f1": 0.5491071428571428,
+      "eval_allNLI-dev_euclidean_f1_threshold": 15.894262313842773,
+      "eval_allNLI-dev_euclidean_precision": 0.44727272727272727,
+      "eval_allNLI-dev_euclidean_recall": 0.7109826589595376,
+      "eval_allNLI-dev_manhattan_accuracy": 0.6875,
+      "eval_allNLI-dev_manhattan_accuracy_threshold": 227.51451110839844,
+      "eval_allNLI-dev_manhattan_ap": 0.5073092702655073,
+      "eval_allNLI-dev_manhattan_f1": 0.5520169851380042,
+      "eval_allNLI-dev_manhattan_f1_threshold": 344.68267822265625,
+      "eval_allNLI-dev_manhattan_precision": 0.436241610738255,
+      "eval_allNLI-dev_manhattan_recall": 0.7514450867052023,
+      "eval_allNLI-dev_max_accuracy": 0.69140625,
+      "eval_allNLI-dev_max_accuracy_threshold": 704.034912109375,
+      "eval_allNLI-dev_max_ap": 0.5100510670205529,
+      "eval_allNLI-dev_max_f1": 0.5520169851380042,
+      "eval_allNLI-dev_max_f1_threshold": 620.714599609375,
+      "eval_allNLI-dev_max_precision": 0.45054945054945056,
+      "eval_allNLI-dev_max_recall": 0.7861271676300579,
+      "eval_loss": 0.8038363456726074,
+      "eval_runtime": 52.9315,
+      "eval_samples_per_second": 31.437,
+      "eval_sequential_score": 0.710276735549099,
+      "eval_steps_per_second": 0.132,
+      "eval_sts-test_pearson_cosine": 0.6581325546921425,
+      "eval_sts-test_pearson_dot": 0.655652438684661,
+      "eval_sts-test_pearson_euclidean": 0.6699927428669137,
+      "eval_sts-test_pearson_manhattan": 0.6737449201833172,
+      "eval_sts-test_pearson_max": 0.6737449201833172,
+      "eval_sts-test_spearman_cosine": 0.663227790467879,
+      "eval_sts-test_spearman_dot": 0.6598153725330006,
+      "eval_sts-test_spearman_euclidean": 0.6639387537576042,
+      "eval_sts-test_spearman_manhattan": 0.6677674926772068,
+      "eval_sts-test_spearman_max": 0.6677674926772068,
       "step": 2601
     },
     {
-      "epoch": 2.559055118110236,
-      "grad_norm": 5.273887634277344,
+      "epoch": 2.561023622047244,
+      "grad_norm": 7.018889427185059,
       "learning_rate": 1.185557528527428e-07,
-      "loss": 0.1929,
+      "loss": 0.2174,
       "step": 2602
     },
     {
-      "epoch": 2.5600393700787403,
-      "grad_norm": 10.100863456726074,
+      "epoch": 2.562007874015748,
+      "grad_norm": 11.365270614624023,
       "learning_rate": 1.1806590852376707e-07,
-      "loss": 0.5182,
+      "loss": 0.6878,
       "step": 2603
     },
     {
-      "epoch": 2.561023622047244,
-      "grad_norm": 4.793201923370361,
+      "epoch": 2.562992125984252,
+      "grad_norm": 6.989991188049316,
       "learning_rate": 1.1757697215672498e-07,
-      "loss": 0.1183,
+      "loss": 0.2113,
       "step": 2604
     },
     {
-      "epoch": 2.562007874015748,
-      "grad_norm": 15.92398452758789,
+      "epoch": 2.563976377952756,
+      "grad_norm": 9.395386695861816,
       "learning_rate": 1.17088944908902e-07,
-      "loss": 0.9434,
+      "loss": 0.4043,
       "step": 2605
     },
     {
-      "epoch": 2.562992125984252,
-      "grad_norm": 9.905328750610352,
+      "epoch": 2.5649606299212597,
+      "grad_norm": 6.393521785736084,
       "learning_rate": 1.1660182793543199e-07,
-      "loss": 0.4401,
+      "loss": 0.2495,
       "step": 2606
     },
     {
-      "epoch": 2.563976377952756,
-      "grad_norm": 8.067452430725098,
+      "epoch": 2.565944881889764,
+      "grad_norm": 7.525991439819336,
       "learning_rate": 1.161156223892938e-07,
-      "loss": 0.4079,
+      "loss": 0.2753,
       "step": 2607
     },
     {
-      "epoch": 2.5649606299212597,
-      "grad_norm": 4.889436721801758,
+      "epoch": 2.5669291338582676,
+      "grad_norm": 9.084732055664062,
       "learning_rate": 1.1563032942130937e-07,
-      "loss": 0.2161,
+      "loss": 0.3233,
       "step": 2608
     },
     {
-      "epoch": 2.565944881889764,
-      "grad_norm": 4.1334686279296875,
+      "epoch": 2.5679133858267718,
+      "grad_norm": 9.213691711425781,
       "learning_rate": 1.1514595018014061e-07,
-      "loss": 0.0959,
+      "loss": 0.4387,
       "step": 2609
     },
     {
-      "epoch": 2.5669291338582676,
-      "grad_norm": 9.387733459472656,
+      "epoch": 2.5688976377952755,
+      "grad_norm": 8.93049144744873,
       "learning_rate": 1.1466248581228649e-07,
-      "loss": 0.2877,
+      "loss": 0.4123,
       "step": 2610
     },
     {
-      "epoch": 2.5679133858267718,
-      "grad_norm": 5.032553672790527,
+      "epoch": 2.5698818897637796,
+      "grad_norm": 6.442656517028809,
       "learning_rate": 1.1417993746208068e-07,
-      "loss": 0.1447,
+      "loss": 0.2335,
       "step": 2611
     },
     {
-      "epoch": 2.5688976377952755,
-      "grad_norm": 12.307943344116211,
+      "epoch": 2.5708661417322833,
+      "grad_norm": 9.612574577331543,
       "learning_rate": 1.136983062716886e-07,
-      "loss": 0.5285,
+      "loss": 0.4015,
       "step": 2612
     },
     {
-      "epoch": 2.5698818897637796,
-      "grad_norm": 15.693968772888184,
+      "epoch": 2.5718503937007875,
+      "grad_norm": 7.650222301483154,
       "learning_rate": 1.1321759338110483e-07,
-      "loss": 0.4332,
+      "loss": 0.3037,
       "step": 2613
     },
     {
-      "epoch": 2.5708661417322833,
-      "grad_norm": 7.844369411468506,
+      "epoch": 2.572834645669291,
+      "grad_norm": 11.224543571472168,
       "learning_rate": 1.1273779992815057e-07,
-      "loss": 0.1841,
+      "loss": 0.3129,
       "step": 2614
     },
     {
-      "epoch": 2.5718503937007875,
-      "grad_norm": 10.694293022155762,
+      "epoch": 2.5738188976377954,
+      "grad_norm": 9.04195499420166,
       "learning_rate": 1.1225892704847027e-07,
-      "loss": 0.4788,
+      "loss": 0.5355,
       "step": 2615
     },
     {
-      "epoch": 2.572834645669291,
-      "grad_norm": 11.026900291442871,
+      "epoch": 2.574803149606299,
+      "grad_norm": 10.610971450805664,
       "learning_rate": 1.1178097587552988e-07,
-      "loss": 0.4086,
+      "loss": 0.3987,
       "step": 2616
     },
     {
-      "epoch": 2.5738188976377954,
-      "grad_norm": 5.624032020568848,
+      "epoch": 2.5757874015748032,
+      "grad_norm": 14.198867797851562,
       "learning_rate": 1.113039475406136e-07,
-      "loss": 0.1467,
+      "loss": 0.7301,
       "step": 2617
     },
     {
-      "epoch": 2.574803149606299,
-      "grad_norm": 8.110433578491211,
+      "epoch": 2.576771653543307,
+      "grad_norm": 8.334237098693848,
       "learning_rate": 1.1082784317282129e-07,
-      "loss": 0.389,
+      "loss": 0.2253,
       "step": 2618
     },
     {
-      "epoch": 2.5757874015748032,
-      "grad_norm": 8.149201393127441,
+      "epoch": 2.577755905511811,
+      "grad_norm": 11.503011703491211,
       "learning_rate": 1.1035266389906588e-07,
-      "loss": 0.2413,
+      "loss": 0.3705,
       "step": 2619
     },
     {
-      "epoch": 2.576771653543307,
-      "grad_norm": 10.457308769226074,
+      "epoch": 2.578740157480315,
+      "grad_norm": 5.937502861022949,
       "learning_rate": 1.0987841084407059e-07,
-      "loss": 0.3814,
+      "loss": 0.1456,
       "step": 2620
     },
     {
-      "epoch": 2.577755905511811,
-      "grad_norm": 19.59735107421875,
+      "epoch": 2.579724409448819,
+      "grad_norm": 5.737178325653076,
       "learning_rate": 1.0940508513036641e-07,
-      "loss": 1.1875,
+      "loss": 0.1329,
       "step": 2621
     },
     {
-      "epoch": 2.578740157480315,
-      "grad_norm": 11.51667594909668,
+      "epoch": 2.5807086614173227,
+      "grad_norm": 8.81822681427002,
       "learning_rate": 1.0893268787828929e-07,
-      "loss": 0.5347,
+      "loss": 0.277,
       "step": 2622
     },
     {
-      "epoch": 2.579724409448819,
-      "grad_norm": 7.69096040725708,
+      "epoch": 2.581692913385827,
+      "grad_norm": 8.412044525146484,
       "learning_rate": 1.0846122020597759e-07,
-      "loss": 0.3168,
+      "loss": 0.2787,
       "step": 2623
     },
     {
-      "epoch": 2.5807086614173227,
-      "grad_norm": 9.124616622924805,
+      "epoch": 2.5826771653543306,
+      "grad_norm": 11.592912673950195,
       "learning_rate": 1.0799068322936935e-07,
-      "loss": 0.3111,
+      "loss": 0.5661,
       "step": 2624
     },
     {
-      "epoch": 2.581692913385827,
-      "grad_norm": 7.774921894073486,
+      "epoch": 2.5836614173228347,
+      "grad_norm": 7.069263935089111,
       "learning_rate": 1.0752107806219986e-07,
-      "loss": 0.2107,
+      "loss": 0.1745,
       "step": 2625
     },
     {
-      "epoch": 2.5826771653543306,
-      "grad_norm": 11.955632209777832,
+      "epoch": 2.5846456692913384,
+      "grad_norm": 11.376214981079102,
       "learning_rate": 1.0705240581599881e-07,
-      "loss": 0.4728,
+      "loss": 0.5826,
       "step": 2626
     },
     {
-      "epoch": 2.5836614173228347,
-      "grad_norm": 8.494672775268555,
+      "epoch": 2.5856299212598426,
+      "grad_norm": 11.338303565979004,
       "learning_rate": 1.0658466760008749e-07,
-      "loss": 0.4143,
+      "loss": 0.3865,
       "step": 2627
     },
     {
-      "epoch": 2.5846456692913384,
-      "grad_norm": 15.70314884185791,
+      "epoch": 2.5866141732283463,
+      "grad_norm": 10.733962059020996,
       "learning_rate": 1.0611786452157668e-07,
-      "loss": 0.8418,
+      "loss": 0.7072,
       "step": 2628
     },
     {
-      "epoch": 2.5856299212598426,
-      "grad_norm": 4.958100318908691,
+      "epoch": 2.5875984251968505,
+      "grad_norm": 11.317181587219238,
       "learning_rate": 1.0565199768536355e-07,
-      "loss": 0.1177,
+      "loss": 0.4822,
       "step": 2629
     },
     {
-      "epoch": 2.5866141732283463,
-      "grad_norm": 11.516602516174316,
+      "epoch": 2.588582677165354,
+      "grad_norm": 9.927474021911621,
       "learning_rate": 1.0518706819412948e-07,
-      "loss": 0.5185,
+      "loss": 0.5733,
       "step": 2630
     },
     {
-      "epoch": 2.5875984251968505,
-      "grad_norm": 16.29201889038086,
+      "epoch": 2.5895669291338583,
+      "grad_norm": 9.038625717163086,
       "learning_rate": 1.0472307714833692e-07,
-      "loss": 0.5703,
+      "loss": 0.364,
       "step": 2631
     },
     {
-      "epoch": 2.588582677165354,
-      "grad_norm": 6.621307373046875,
+      "epoch": 2.590551181102362,
+      "grad_norm": 13.401626586914062,
       "learning_rate": 1.042600256462273e-07,
-      "loss": 0.1895,
+      "loss": 0.6186,
       "step": 2632
     },
     {
-      "epoch": 2.5895669291338583,
-      "grad_norm": 8.731075286865234,
+      "epoch": 2.591535433070866,
+      "grad_norm": 12.441668510437012,
       "learning_rate": 1.0379791478381805e-07,
-      "loss": 0.3434,
+      "loss": 0.6711,
       "step": 2633
     },
     {
-      "epoch": 2.590551181102362,
-      "grad_norm": 13.496260643005371,
+      "epoch": 2.59251968503937,
+      "grad_norm": 12.678735733032227,
       "learning_rate": 1.0333674565490014e-07,
-      "loss": 0.6885,
+      "loss": 0.4297,
       "step": 2634
     },
     {
-      "epoch": 2.591535433070866,
-      "grad_norm": 12.25928020477295,
+      "epoch": 2.593503937007874,
+      "grad_norm": 13.716216087341309,
       "learning_rate": 1.0287651935103572e-07,
-      "loss": 0.5406,
+      "loss": 0.6323,
       "step": 2635
     },
     {
-      "epoch": 2.59251968503937,
-      "grad_norm": 7.45269250869751,
+      "epoch": 2.594488188976378,
+      "grad_norm": 3.8192131519317627,
       "learning_rate": 1.024172369615551e-07,
-      "loss": 0.3012,
+      "loss": 0.0794,
       "step": 2636
     },
     {
-      "epoch": 2.593503937007874,
-      "grad_norm": 17.18060874938965,
+      "epoch": 2.595472440944882,
+      "grad_norm": 4.939494609832764,
       "learning_rate": 1.0195889957355454e-07,
-      "loss": 0.9385,
+      "loss": 0.1299,
       "step": 2637
     },
     {
-      "epoch": 2.594488188976378,
-      "grad_norm": 9.831514358520508,
+      "epoch": 2.596456692913386,
+      "grad_norm": 12.66039752960205,
       "learning_rate": 1.0150150827189337e-07,
-      "loss": 0.3235,
+      "loss": 0.7791,
       "step": 2638
     },
     {
-      "epoch": 2.595472440944882,
-      "grad_norm": 7.493289947509766,
+      "epoch": 2.59744094488189,
+      "grad_norm": 12.604268074035645,
       "learning_rate": 1.0104506413919173e-07,
-      "loss": 0.2036,
+      "loss": 0.7766,
       "step": 2639
     },
     {
-      "epoch": 2.596456692913386,
-      "grad_norm": 9.814960479736328,
+      "epoch": 2.5984251968503935,
+      "grad_norm": 7.212848663330078,
       "learning_rate": 1.0058956825582792e-07,
-      "loss": 0.3403,
+      "loss": 0.4644,
       "step": 2640
     },
     {
-      "epoch": 2.59744094488189,
-      "grad_norm": 8.145853042602539,
+      "epoch": 2.5994094488188977,
+      "grad_norm": 8.909222602844238,
       "learning_rate": 1.0013502169993552e-07,
-      "loss": 0.3688,
+      "loss": 0.3575,
       "step": 2641
     },
     {
-      "epoch": 2.5984251968503935,
-      "grad_norm": 5.9386773109436035,
+      "epoch": 2.600393700787402,
+      "grad_norm": 7.3771443367004395,
       "learning_rate": 9.96814255474014e-08,
-      "loss": 0.3998,
+      "loss": 0.2899,
       "step": 2642
     },
     {
-      "epoch": 2.5994094488188977,
-      "grad_norm": 10.198760986328125,
+      "epoch": 2.6013779527559056,
+      "grad_norm": 9.995150566101074,
       "learning_rate": 9.922878087186273e-08,
-      "loss": 0.3149,
+      "loss": 0.3339,
       "step": 2643
     },
     {
-      "epoch": 2.600393700787402,
-      "grad_norm": 8.756124496459961,
+      "epoch": 2.6023622047244093,
+      "grad_norm": 4.899702072143555,
       "learning_rate": 9.877708874470455e-08,
-      "loss": 0.2704,
+      "loss": 0.1267,
       "step": 2644
     },
     {
-      "epoch": 2.6013779527559056,
-      "grad_norm": 6.515359401702881,
+      "epoch": 2.6033464566929134,
+      "grad_norm": 8.763663291931152,
       "learning_rate": 9.832635023505755e-08,
-      "loss": 0.158,
+      "loss": 0.2784,
       "step": 2645
     },
     {
-      "epoch": 2.6023622047244093,
-      "grad_norm": 8.050736427307129,
+      "epoch": 2.6043307086614176,
+      "grad_norm": 10.163912773132324,
       "learning_rate": 9.787656640979466e-08,
-      "loss": 0.2982,
+      "loss": 0.3755,
       "step": 2646
     },
     {
-      "epoch": 2.6033464566929134,
-      "grad_norm": 7.078582286834717,
+      "epoch": 2.6053149606299213,
+      "grad_norm": 9.669671058654785,
       "learning_rate": 9.74277383335297e-08,
-      "loss": 0.2121,
+      "loss": 0.296,
       "step": 2647
     },
     {
-      "epoch": 2.6043307086614176,
-      "grad_norm": 12.721890449523926,
+      "epoch": 2.606299212598425,
+      "grad_norm": 9.506949424743652,
       "learning_rate": 9.697986706861403e-08,
-      "loss": 0.426,
+      "loss": 0.4438,
       "step": 2648
     },
     {
-      "epoch": 2.6053149606299213,
-      "grad_norm": 5.437317371368408,
+      "epoch": 2.607283464566929,
+      "grad_norm": 9.876251220703125,
       "learning_rate": 9.653295367513442e-08,
-      "loss": 0.1346,
+      "loss": 0.3387,
       "step": 2649
     },
     {
-      "epoch": 2.606299212598425,
-      "grad_norm": 10.403167724609375,
+      "epoch": 2.6082677165354333,
+      "grad_norm": 9.776094436645508,
       "learning_rate": 9.608699921091029e-08,
-      "loss": 0.5152,
+      "loss": 0.3702,
       "step": 2650
     },
     {
-      "epoch": 2.607283464566929,
-      "grad_norm": 15.584598541259766,
+      "epoch": 2.609251968503937,
+      "grad_norm": 10.888093948364258,
       "learning_rate": 9.564200473149156e-08,
-      "loss": 0.705,
+      "loss": 0.5393,
       "step": 2651
     },
     {
-      "epoch": 2.6082677165354333,
-      "grad_norm": 10.282527923583984,
+      "epoch": 2.6102362204724407,
+      "grad_norm": 8.191256523132324,
       "learning_rate": 9.519797129015536e-08,
-      "loss": 0.3815,
+      "loss": 0.2475,
       "step": 2652
     },
     {
-      "epoch": 2.609251968503937,
-      "grad_norm": 8.903995513916016,
+      "epoch": 2.611220472440945,
+      "grad_norm": 7.278419494628906,
       "learning_rate": 9.475489993790467e-08,
-      "loss": 0.3878,
+      "loss": 0.366,
       "step": 2653
     },
     {
-      "epoch": 2.6102362204724407,
-      "grad_norm": 11.657297134399414,
+      "epoch": 2.612204724409449,
+      "grad_norm": 12.221094131469727,
       "learning_rate": 9.431279172346506e-08,
-      "loss": 0.2256,
+      "loss": 0.557,
       "step": 2654
     },
     {
-      "epoch": 2.611220472440945,
-      "grad_norm": 13.781116485595703,
+      "epoch": 2.6131889763779528,
+      "grad_norm": 8.53677749633789,
       "learning_rate": 9.387164769328237e-08,
-      "loss": 0.6192,
+      "loss": 0.2491,
       "step": 2655
     },
     {
-      "epoch": 2.612204724409449,
-      "grad_norm": 16.146852493286133,
+      "epoch": 2.6141732283464565,
+      "grad_norm": 8.400619506835938,
       "learning_rate": 9.343146889152028e-08,
-      "loss": 0.6693,
+      "loss": 0.2431,
       "step": 2656
     },
     {
-      "epoch": 2.6131889763779528,
-      "grad_norm": 10.342763900756836,
+      "epoch": 2.6151574803149606,
+      "grad_norm": 6.362612247467041,
       "learning_rate": 9.299225636005791e-08,
-      "loss": 0.2787,
+      "loss": 0.1967,
       "step": 2657
     },
     {
-      "epoch": 2.6141732283464565,
-      "grad_norm": 6.541711807250977,
+      "epoch": 2.616141732283465,
+      "grad_norm": 7.944909572601318,
       "learning_rate": 9.25540111384872e-08,
-      "loss": 0.2554,
+      "loss": 0.3272,
       "step": 2658
     },
     {
-      "epoch": 2.6151574803149606,
-      "grad_norm": 6.099989414215088,
+      "epoch": 2.6171259842519685,
+      "grad_norm": 8.696026802062988,
       "learning_rate": 9.211673426411052e-08,
-      "loss": 0.1382,
+      "loss": 0.5326,
       "step": 2659
     },
     {
-      "epoch": 2.616141732283465,
-      "grad_norm": 9.739631652832031,
+      "epoch": 2.6181102362204722,
+      "grad_norm": 11.001298904418945,
       "learning_rate": 9.168042677193818e-08,
-      "loss": 0.4049,
+      "loss": 0.4241,
       "step": 2660
     },
     {
-      "epoch": 2.6171259842519685,
-      "grad_norm": 8.16472339630127,
+      "epoch": 2.6190944881889764,
+      "grad_norm": 8.780117988586426,
       "learning_rate": 9.124508969468606e-08,
-      "loss": 0.4021,
+      "loss": 0.381,
       "step": 2661
     },
     {
-      "epoch": 2.6181102362204722,
-      "grad_norm": 7.355652809143066,
+      "epoch": 2.6200787401574805,
+      "grad_norm": 13.57796573638916,
       "learning_rate": 9.08107240627732e-08,
-      "loss": 0.2101,
+      "loss": 0.5542,
       "step": 2662
     },
     {
-      "epoch": 2.6190944881889764,
-      "grad_norm": 14.677267074584961,
+      "epoch": 2.6210629921259843,
+      "grad_norm": 6.780984878540039,
       "learning_rate": 9.037733090431912e-08,
-      "loss": 0.4881,
+      "loss": 0.1807,
       "step": 2663
     },
     {
-      "epoch": 2.6200787401574805,
-      "grad_norm": 8.316322326660156,
+      "epoch": 2.622047244094488,
+      "grad_norm": 9.28343677520752,
       "learning_rate": 8.994491124514166e-08,
-      "loss": 0.2375,
+      "loss": 0.4598,
       "step": 2664
     },
     {
-      "epoch": 2.6210629921259843,
-      "grad_norm": 18.017789840698242,
+      "epoch": 2.623031496062992,
+      "grad_norm": 9.147518157958984,
       "learning_rate": 8.951346610875441e-08,
-      "loss": 0.7727,
+      "loss": 0.37,
       "step": 2665
     },
     {
-      "epoch": 2.622047244094488,
-      "grad_norm": 10.476648330688477,
+      "epoch": 2.6240157480314963,
+      "grad_norm": 8.63045883178711,
       "learning_rate": 8.908299651636436e-08,
-      "loss": 0.489,
+      "loss": 0.388,
       "step": 2666
     },
     {
-      "epoch": 2.623031496062992,
-      "grad_norm": 14.659133911132812,
+      "epoch": 2.625,
+      "grad_norm": 7.269915580749512,
       "learning_rate": 8.865350348686934e-08,
-      "loss": 0.7648,
+      "loss": 0.4206,
       "step": 2667
     },
     {
-      "epoch": 2.6240157480314963,
-      "grad_norm": 11.297569274902344,
+      "epoch": 2.6259842519685037,
+      "grad_norm": 7.627872943878174,
       "learning_rate": 8.822498803685595e-08,
-      "loss": 0.3395,
+      "loss": 0.3069,
       "step": 2668
     },
     {
-      "epoch": 2.625,
-      "grad_norm": 2.457162380218506,
+      "epoch": 2.626968503937008,
+      "grad_norm": 10.175527572631836,
       "learning_rate": 8.779745118059643e-08,
-      "loss": 0.0534,
+      "loss": 0.496,
       "step": 2669
     },
     {
-      "epoch": 2.6259842519685037,
-      "grad_norm": 7.145143032073975,
+      "epoch": 2.627952755905512,
+      "grad_norm": 13.628823280334473,
       "learning_rate": 8.737089393004722e-08,
-      "loss": 0.3496,
+      "loss": 0.7739,
       "step": 2670
     },
     {
-      "epoch": 2.626968503937008,
-      "grad_norm": 12.202618598937988,
+      "epoch": 2.6289370078740157,
+      "grad_norm": 9.814640998840332,
       "learning_rate": 8.694531729484584e-08,
-      "loss": 0.7193,
+      "loss": 0.527,
       "step": 2671
     },
     {
-      "epoch": 2.627952755905512,
-      "grad_norm": 8.030990600585938,
+      "epoch": 2.6299212598425195,
+      "grad_norm": 12.2955961227417,
       "learning_rate": 8.65207222823089e-08,
-      "loss": 0.2831,
+      "loss": 0.8247,
       "step": 2672
     },
     {
-      "epoch": 2.6289370078740157,
-      "grad_norm": 8.702947616577148,
+      "epoch": 2.6309055118110236,
+      "grad_norm": 4.090799331665039,
       "learning_rate": 8.609710989742948e-08,
-      "loss": 0.4369,
+      "loss": 0.0693,
       "step": 2673
     },
     {
-      "epoch": 2.6299212598425195,
-      "grad_norm": 11.548181533813477,
+      "epoch": 2.6318897637795278,
+      "grad_norm": 8.672216415405273,
       "learning_rate": 8.567448114287475e-08,
-      "loss": 0.4547,
+      "loss": 0.2947,
       "step": 2674
     },
     {
-      "epoch": 2.6309055118110236,
-      "grad_norm": 8.138830184936523,
+      "epoch": 2.6328740157480315,
+      "grad_norm": 4.160043239593506,
       "learning_rate": 8.525283701898387e-08,
-      "loss": 0.1804,
+      "loss": 0.1196,
       "step": 2675
     },
     {
-      "epoch": 2.6318897637795278,
-      "grad_norm": 8.969474792480469,
+      "epoch": 2.633858267716535,
+      "grad_norm": 6.367125988006592,
       "learning_rate": 8.483217852376527e-08,
-      "loss": 0.3648,
+      "loss": 0.29,
       "step": 2676
     },
     {
-      "epoch": 2.6328740157480315,
-      "grad_norm": 7.652669429779053,
+      "epoch": 2.6348425196850394,
+      "grad_norm": 12.123324394226074,
       "learning_rate": 8.441250665289457e-08,
-      "loss": 0.255,
+      "loss": 0.6652,
       "step": 2677
     },
     {
-      "epoch": 2.633858267716535,
-      "grad_norm": 6.664198875427246,
+      "epoch": 2.6358267716535435,
+      "grad_norm": 11.956300735473633,
       "learning_rate": 8.399382239971196e-08,
-      "loss": 0.2309,
+      "loss": 0.6153,
       "step": 2678
     },
     {
-      "epoch": 2.6348425196850394,
-      "grad_norm": 11.025222778320312,
+      "epoch": 2.6368110236220472,
+      "grad_norm": 6.050601005554199,
       "learning_rate": 8.357612675522011e-08,
-      "loss": 0.7033,
+      "loss": 0.1655,
       "step": 2679
     },
     {
-      "epoch": 2.6358267716535435,
-      "grad_norm": 17.735673904418945,
+      "epoch": 2.637795275590551,
+      "grad_norm": 5.100040912628174,
       "learning_rate": 8.315942070808176e-08,
-      "loss": 0.8678,
+      "loss": 0.152,
       "step": 2680
     },
     {
-      "epoch": 2.6368110236220472,
-      "grad_norm": 12.674147605895996,
+      "epoch": 2.638779527559055,
+      "grad_norm": 10.340904235839844,
       "learning_rate": 8.274370524461712e-08,
-      "loss": 0.5771,
+      "loss": 0.4414,
       "step": 2681
     },
     {
-      "epoch": 2.637795275590551,
-      "grad_norm": 8.278904914855957,
+      "epoch": 2.6397637795275593,
+      "grad_norm": 10.77912425994873,
       "learning_rate": 8.232898134880213e-08,
-      "loss": 0.3597,
+      "loss": 0.423,
       "step": 2682
     },
     {
-      "epoch": 2.638779527559055,
-      "grad_norm": 3.23533034324646,
+      "epoch": 2.640748031496063,
+      "grad_norm": 9.533746719360352,
       "learning_rate": 8.19152500022651e-08,
-      "loss": 0.0554,
+      "loss": 0.181,
       "step": 2683
     },
     {
-      "epoch": 2.6397637795275593,
-      "grad_norm": 8.985905647277832,
+      "epoch": 2.6417322834645667,
+      "grad_norm": 11.64153003692627,
       "learning_rate": 8.150251218428557e-08,
-      "loss": 0.2783,
+      "loss": 0.4381,
       "step": 2684
     },
     {
-      "epoch": 2.640748031496063,
-      "grad_norm": 6.580881595611572,
+      "epoch": 2.642716535433071,
+      "grad_norm": 8.37061882019043,
       "learning_rate": 8.109076887179138e-08,
-      "loss": 0.2926,
+      "loss": 0.309,
       "step": 2685
     },
     {
-      "epoch": 2.6417322834645667,
-      "grad_norm": 6.928036689758301,
+      "epoch": 2.643700787401575,
+      "grad_norm": 9.878986358642578,
       "learning_rate": 8.068002103935624e-08,
-      "loss": 0.2097,
+      "loss": 0.3099,
       "step": 2686
     },
     {
-      "epoch": 2.642716535433071,
-      "grad_norm": 17.22882843017578,
+      "epoch": 2.6446850393700787,
+      "grad_norm": 8.295797348022461,
       "learning_rate": 8.027026965919774e-08,
-      "loss": 1.0672,
+      "loss": 0.2231,
       "step": 2687
     },
     {
-      "epoch": 2.643700787401575,
-      "grad_norm": 6.01613712310791,
+      "epoch": 2.6456692913385824,
+      "grad_norm": 10.967564582824707,
       "learning_rate": 7.986151570117498e-08,
-      "loss": 0.1549,
+      "loss": 0.5459,
       "step": 2688
     },
     {
-      "epoch": 2.6446850393700787,
-      "grad_norm": 2.506317377090454,
+      "epoch": 2.6466535433070866,
+      "grad_norm": 4.851083278656006,
       "learning_rate": 7.945376013278619e-08,
-      "loss": 0.0484,
+      "loss": 0.1033,
       "step": 2689
     },
     {
-      "epoch": 2.6456692913385824,
-      "grad_norm": 3.891319513320923,
+      "epoch": 2.6476377952755907,
+      "grad_norm": 9.955401420593262,
       "learning_rate": 7.904700391916614e-08,
-      "loss": 0.0606,
+      "loss": 0.4474,
       "step": 2690
     },
     {
-      "epoch": 2.6466535433070866,
-      "grad_norm": 7.721035957336426,
+      "epoch": 2.6486220472440944,
+      "grad_norm": 6.133537769317627,
       "learning_rate": 7.864124802308461e-08,
-      "loss": 0.2239,
+      "loss": 0.1934,
       "step": 2691
     },
     {
-      "epoch": 2.6476377952755907,
-      "grad_norm": 10.72614574432373,
+      "epoch": 2.649606299212598,
+      "grad_norm": 9.899557113647461,
       "learning_rate": 7.823649340494358e-08,
-      "loss": 0.3587,
+      "loss": 0.356,
       "step": 2692
     },
     {
-      "epoch": 2.6486220472440944,
-      "grad_norm": 7.484883785247803,
+      "epoch": 2.6505905511811023,
+      "grad_norm": 5.93404483795166,
       "learning_rate": 7.783274102277492e-08,
-      "loss": 0.1615,
+      "loss": 0.2524,
       "step": 2693
     },
     {
-      "epoch": 2.649606299212598,
-      "grad_norm": 13.634208679199219,
+      "epoch": 2.6515748031496065,
+      "grad_norm": 4.263927936553955,
       "learning_rate": 7.742999183223844e-08,
-      "loss": 0.9166,
+      "loss": 0.0994,
       "step": 2694
     },
     {
-      "epoch": 2.6505905511811023,
-      "grad_norm": 13.837207794189453,
+      "epoch": 2.65255905511811,
+      "grad_norm": 9.903258323669434,
       "learning_rate": 7.702824678661923e-08,
-      "loss": 0.9832,
+      "loss": 0.3365,
       "step": 2695
     },
     {
-      "epoch": 2.6515748031496065,
-      "grad_norm": 13.681209564208984,
+      "epoch": 2.653543307086614,
+      "grad_norm": 8.988314628601074,
       "learning_rate": 7.662750683682593e-08,
-      "loss": 0.788,
+      "loss": 0.3263,
       "step": 2696
     },
     {
-      "epoch": 2.65255905511811,
-      "grad_norm": 10.74668025970459,
+      "epoch": 2.654527559055118,
+      "grad_norm": 6.7262396812438965,
       "learning_rate": 7.622777293138787e-08,
-      "loss": 0.4497,
+      "loss": 0.2474,
       "step": 2697
     },
     {
-      "epoch": 2.653543307086614,
-      "grad_norm": 8.555436134338379,
+      "epoch": 2.655511811023622,
+      "grad_norm": 6.978824138641357,
       "learning_rate": 7.582904601645333e-08,
-      "loss": 0.2578,
+      "loss": 0.2176,
       "step": 2698
     },
     {
-      "epoch": 2.654527559055118,
-      "grad_norm": 7.200206279754639,
+      "epoch": 2.656496062992126,
+      "grad_norm": 15.484535217285156,
       "learning_rate": 7.543132703578697e-08,
-      "loss": 0.2575,
+      "loss": 1.0462,
       "step": 2699
     },
     {
-      "epoch": 2.655511811023622,
-      "grad_norm": 7.164278984069824,
+      "epoch": 2.65748031496063,
+      "grad_norm": 5.7792205810546875,
       "learning_rate": 7.503461693076783e-08,
-      "loss": 0.2621,
+      "loss": 0.1482,
       "step": 2700
     },
     {
-      "epoch": 2.656496062992126,
-      "grad_norm": 7.4781036376953125,
+      "epoch": 2.658464566929134,
+      "grad_norm": 9.764116287231445,
       "learning_rate": 7.463891664038691e-08,
-      "loss": 0.2647,
+      "loss": 0.3416,
       "step": 2701
     },
     {
-      "epoch": 2.65748031496063,
-      "grad_norm": 9.039846420288086,
+      "epoch": 2.659448818897638,
+      "grad_norm": 7.243642807006836,
       "learning_rate": 7.424422710124503e-08,
-      "loss": 0.3083,
+      "loss": 0.2477,
       "step": 2702
     },
     {
-      "epoch": 2.658464566929134,
-      "grad_norm": 11.146456718444824,
+      "epoch": 2.6604330708661417,
+      "grad_norm": 11.70655345916748,
       "learning_rate": 7.385054924755081e-08,
-      "loss": 0.326,
+      "loss": 0.7111,
       "step": 2703
     },
     {
-      "epoch": 2.659448818897638,
-      "grad_norm": 7.017238140106201,
+      "epoch": 2.661417322834646,
+      "grad_norm": 9.447341918945312,
       "learning_rate": 7.345788401111793e-08,
-      "loss": 0.3751,
+      "loss": 0.3614,
       "step": 2704
     },
     {
-      "epoch": 2.6604330708661417,
-      "grad_norm": 15.564183235168457,
+      "epoch": 2.6624015748031495,
+      "grad_norm": 6.129354476928711,
       "learning_rate": 7.306623232136358e-08,
-      "loss": 1.0587,
+      "loss": 0.1872,
       "step": 2705
     },
     {
-      "epoch": 2.661417322834646,
-      "grad_norm": 5.695704460144043,
+      "epoch": 2.6633858267716537,
+      "grad_norm": 8.463857650756836,
       "learning_rate": 7.26755951053058e-08,
-      "loss": 0.1722,
+      "loss": 0.3295,
       "step": 2706
     },
     {
-      "epoch": 2.6624015748031495,
-      "grad_norm": 8.566353797912598,
+      "epoch": 2.6643700787401574,
+      "grad_norm": 4.36069917678833,
       "learning_rate": 7.228597328756136e-08,
-      "loss": 0.2922,
+      "loss": 0.089,
       "step": 2707
     },
     {
-      "epoch": 2.6633858267716537,
-      "grad_norm": 11.269908905029297,
+      "epoch": 2.6653543307086616,
+      "grad_norm": 4.437070369720459,
       "learning_rate": 7.189736779034375e-08,
-      "loss": 0.409,
+      "loss": 0.0955,
       "step": 2708
     },
     {
-      "epoch": 2.6643700787401574,
-      "grad_norm": 11.922059059143066,
+      "epoch": 2.6663385826771653,
+      "grad_norm": 8.272147178649902,
       "learning_rate": 7.150977953346082e-08,
-      "loss": 0.2721,
+      "loss": 0.2971,
       "step": 2709
     },
     {
-      "epoch": 2.6653543307086616,
-      "grad_norm": 9.26468563079834,
+      "epoch": 2.6673228346456694,
+      "grad_norm": 11.063708305358887,
       "learning_rate": 7.112320943431287e-08,
-      "loss": 0.2416,
+      "loss": 0.4915,
       "step": 2710
     },
     {
-      "epoch": 2.6663385826771653,
-      "grad_norm": 9.859065055847168,
+      "epoch": 2.668307086614173,
+      "grad_norm": 10.723077774047852,
       "learning_rate": 7.07376584078898e-08,
-      "loss": 0.3148,
+      "loss": 0.4308,
       "step": 2711
     },
     {
-      "epoch": 2.6673228346456694,
-      "grad_norm": 10.863545417785645,
+      "epoch": 2.6692913385826773,
+      "grad_norm": 4.664196014404297,
       "learning_rate": 7.035312736676993e-08,
-      "loss": 0.3821,
+      "loss": 0.1033,
       "step": 2712
     },
     {
-      "epoch": 2.668307086614173,
-      "grad_norm": 13.722128868103027,
+      "epoch": 2.670275590551181,
+      "grad_norm": 7.82461404800415,
       "learning_rate": 6.996961722111701e-08,
-      "loss": 0.6524,
+      "loss": 0.328,
       "step": 2713
     },
     {
-      "epoch": 2.6692913385826773,
-      "grad_norm": 12.660308837890625,
+      "epoch": 2.671259842519685,
+      "grad_norm": 7.769363880157471,
       "learning_rate": 6.958712887867856e-08,
-      "loss": 0.5353,
+      "loss": 0.1923,
       "step": 2714
     },
     {
-      "epoch": 2.670275590551181,
-      "grad_norm": 15.21914291381836,
+      "epoch": 2.672244094488189,
+      "grad_norm": 15.456931114196777,
       "learning_rate": 6.920566324478348e-08,
-      "loss": 0.4884,
+      "loss": 1.0202,
       "step": 2715
     },
     {
-      "epoch": 2.671259842519685,
-      "grad_norm": 6.661035060882568,
+      "epoch": 2.673228346456693,
+      "grad_norm": 12.189972877502441,
       "learning_rate": 6.882522122234002e-08,
-      "loss": 0.1514,
+      "loss": 0.5382,
       "step": 2716
     },
     {
-      "epoch": 2.672244094488189,
-      "grad_norm": 8.941335678100586,
+      "epoch": 2.6742125984251968,
+      "grad_norm": 9.517648696899414,
       "learning_rate": 6.844580371183347e-08,
-      "loss": 0.3488,
+      "loss": 0.4475,
       "step": 2717
     },
     {
-      "epoch": 2.673228346456693,
-      "grad_norm": 8.471234321594238,
+      "epoch": 2.675196850393701,
+      "grad_norm": 10.647794723510742,
       "learning_rate": 6.806741161132428e-08,
-      "loss": 0.2628,
+      "loss": 0.6444,
       "step": 2718
     },
     {
-      "epoch": 2.6742125984251968,
-      "grad_norm": 13.892240524291992,
+      "epoch": 2.6761811023622046,
+      "grad_norm": 11.130318641662598,
       "learning_rate": 6.769004581644596e-08,
-      "loss": 0.8696,
+      "loss": 0.4964,
       "step": 2719
     },
     {
-      "epoch": 2.675196850393701,
-      "grad_norm": 7.2624921798706055,
+      "epoch": 2.677165354330709,
+      "grad_norm": 6.473448753356934,
       "learning_rate": 6.73137072204023e-08,
-      "loss": 0.199,
+      "loss": 0.1781,
       "step": 2720
     },
     {
-      "epoch": 2.6761811023622046,
-      "grad_norm": 15.945473670959473,
+      "epoch": 2.6781496062992125,
+      "grad_norm": 11.058150291442871,
       "learning_rate": 6.693839671396629e-08,
-      "loss": 0.8202,
+      "loss": 0.6226,
       "step": 2721
     },
     {
-      "epoch": 2.677165354330709,
-      "grad_norm": 12.784812927246094,
+      "epoch": 2.6791338582677167,
+      "grad_norm": 11.366744995117188,
       "learning_rate": 6.656411518547725e-08,
-      "loss": 0.905,
+      "loss": 0.4835,
       "step": 2722
     },
     {
-      "epoch": 2.6781496062992125,
-      "grad_norm": 10.727723121643066,
+      "epoch": 2.6801181102362204,
+      "grad_norm": 6.459019660949707,
       "learning_rate": 6.619086352083902e-08,
-      "loss": 0.4115,
+      "loss": 0.2242,
       "step": 2723
     },
     {
-      "epoch": 2.6791338582677167,
-      "grad_norm": 8.653194427490234,
+      "epoch": 2.6811023622047245,
+      "grad_norm": 4.640380859375,
       "learning_rate": 6.581864260351775e-08,
-      "loss": 0.2155,
+      "loss": 0.1288,
       "step": 2724
     },
     {
-      "epoch": 2.6801181102362204,
-      "grad_norm": 10.35145378112793,
+      "epoch": 2.6820866141732282,
+      "grad_norm": 12.007043838500977,
       "learning_rate": 6.544745331454003e-08,
-      "loss": 0.408,
+      "loss": 0.5609,
       "step": 2725
     },
     {
-      "epoch": 2.6811023622047245,
-      "grad_norm": 4.746224880218506,
+      "epoch": 2.6830708661417324,
+      "grad_norm": 8.815008163452148,
       "learning_rate": 6.50772965324904e-08,
-      "loss": 0.104,
+      "loss": 0.3572,
       "step": 2726
     },
     {
-      "epoch": 2.6820866141732282,
-      "grad_norm": 9.364527702331543,
+      "epoch": 2.684055118110236,
+      "grad_norm": 9.34145450592041,
       "learning_rate": 6.470817313350971e-08,
-      "loss": 0.5343,
+      "loss": 0.2987,
       "step": 2727
     },
     {
-      "epoch": 2.6830708661417324,
-      "grad_norm": 8.605332374572754,
+      "epoch": 2.6850393700787403,
+      "grad_norm": 10.670259475708008,
       "learning_rate": 6.434008399129279e-08,
-      "loss": 0.258,
+      "loss": 0.4154,
       "step": 2728
     },
     {
-      "epoch": 2.684055118110236,
-      "grad_norm": 4.468958854675293,
+      "epoch": 2.686023622047244,
+      "grad_norm": 5.689862251281738,
       "learning_rate": 6.397302997708641e-08,
-      "loss": 0.133,
+      "loss": 0.1642,
       "step": 2729
     },
     {
-      "epoch": 2.6850393700787403,
-      "grad_norm": 7.568987846374512,
+      "epoch": 2.687007874015748,
+      "grad_norm": 7.3171467781066895,
       "learning_rate": 6.360701195968723e-08,
-      "loss": 0.24,
+      "loss": 0.2572,
       "step": 2730
     },
     {
-      "epoch": 2.686023622047244,
-      "grad_norm": 6.589425086975098,
+      "epoch": 2.687992125984252,
+      "grad_norm": 4.065371990203857,
       "learning_rate": 6.324203080543998e-08,
-      "loss": 0.2113,
+      "loss": 0.0667,
       "step": 2731
     },
     {
-      "epoch": 2.687007874015748,
-      "grad_norm": 11.292271614074707,
+      "epoch": 2.688976377952756,
+      "grad_norm": 7.206827640533447,
       "learning_rate": 6.287808737823473e-08,
-      "loss": 0.5365,
+      "loss": 0.1979,
       "step": 2732
     },
     {
-      "epoch": 2.687992125984252,
-      "grad_norm": 2.5217807292938232,
+      "epoch": 2.6899606299212597,
+      "grad_norm": 5.508035659790039,
       "learning_rate": 6.251518253950574e-08,
-      "loss": 0.0353,
+      "loss": 0.1383,
       "step": 2733
     },
     {
-      "epoch": 2.688976377952756,
-      "grad_norm": 6.140711307525635,
+      "epoch": 2.690944881889764,
+      "grad_norm": 11.834595680236816,
       "learning_rate": 6.215331714822883e-08,
-      "loss": 0.145,
+      "loss": 0.6743,
       "step": 2734
     },
     {
-      "epoch": 2.6899606299212597,
-      "grad_norm": 9.195682525634766,
+      "epoch": 2.6919291338582676,
+      "grad_norm": 9.458779335021973,
       "learning_rate": 6.179249206091951e-08,
-      "loss": 0.3852,
+      "loss": 0.4473,
       "step": 2735
     },
     {
-      "epoch": 2.690944881889764,
-      "grad_norm": 11.93599796295166,
+      "epoch": 2.6929133858267718,
+      "grad_norm": 8.913199424743652,
       "learning_rate": 6.143270813163099e-08,
-      "loss": 0.5938,
+      "loss": 0.3028,
       "step": 2736
     },
     {
-      "epoch": 2.6919291338582676,
-      "grad_norm": 10.107711791992188,
+      "epoch": 2.6938976377952755,
+      "grad_norm": 5.683338165283203,
       "learning_rate": 6.107396621195201e-08,
-      "loss": 0.4196,
+      "loss": 0.1625,
       "step": 2737
     },
     {
-      "epoch": 2.6929133858267718,
-      "grad_norm": 14.341265678405762,
+      "epoch": 2.6948818897637796,
+      "grad_norm": 9.772565841674805,
       "learning_rate": 6.071626715100503e-08,
-      "loss": 0.7104,
+      "loss": 0.4893,
       "step": 2738
     },
     {
-      "epoch": 2.6938976377952755,
-      "grad_norm": 8.13415241241455,
+      "epoch": 2.6958661417322833,
+      "grad_norm": 6.35415506362915,
       "learning_rate": 6.035961179544409e-08,
-      "loss": 0.2068,
+      "loss": 0.195,
       "step": 2739
     },
     {
-      "epoch": 2.6948818897637796,
-      "grad_norm": 7.959664821624756,
+      "epoch": 2.6968503937007875,
+      "grad_norm": 8.442315101623535,
       "learning_rate": 6.000400098945283e-08,
-      "loss": 0.2763,
+      "loss": 0.2517,
       "step": 2740
     },
     {
-      "epoch": 2.6958661417322833,
-      "grad_norm": 11.592156410217285,
+      "epoch": 2.697834645669291,
+      "grad_norm": 6.984649181365967,
       "learning_rate": 5.96494355747425e-08,
-      "loss": 0.4907,
+      "loss": 0.3004,
       "step": 2741
     },
     {
-      "epoch": 2.6968503937007875,
-      "grad_norm": 16.181486129760742,
+      "epoch": 2.6988188976377954,
+      "grad_norm": 8.231756210327148,
       "learning_rate": 5.9295916390550034e-08,
-      "loss": 0.5446,
+      "loss": 0.3545,
       "step": 2742
     },
     {
-      "epoch": 2.697834645669291,
-      "grad_norm": 14.021303176879883,
+      "epoch": 2.699803149606299,
+      "grad_norm": 11.05482006072998,
       "learning_rate": 5.89434442736358e-08,
-      "loss": 0.694,
+      "loss": 0.3757,
       "step": 2743
     },
     {
-      "epoch": 2.6988188976377954,
-      "grad_norm": 8.919441223144531,
+      "epoch": 2.7007874015748032,
+      "grad_norm": 13.126968383789062,
       "learning_rate": 5.859202005828208e-08,
-      "loss": 0.3986,
+      "loss": 0.6278,
       "step": 2744
     },
     {
-      "epoch": 2.699803149606299,
-      "grad_norm": 14.602743148803711,
+      "epoch": 2.701771653543307,
+      "grad_norm": 12.401497840881348,
       "learning_rate": 5.824164457629061e-08,
-      "loss": 0.6666,
+      "loss": 0.5391,
       "step": 2745
     },
     {
-      "epoch": 2.7007874015748032,
-      "grad_norm": 6.111942768096924,
+      "epoch": 2.702755905511811,
+      "grad_norm": 5.602319717407227,
       "learning_rate": 5.789231865698098e-08,
-      "loss": 0.1457,
+      "loss": 0.2004,
       "step": 2746
     },
     {
-      "epoch": 2.701771653543307,
-      "grad_norm": 10.023294448852539,
+      "epoch": 2.703740157480315,
+      "grad_norm": 8.838926315307617,
       "learning_rate": 5.754404312718839e-08,
-      "loss": 0.3554,
+      "loss": 0.3201,
       "step": 2747
     },
     {
-      "epoch": 2.702755905511811,
-      "grad_norm": 7.863895893096924,
+      "epoch": 2.704724409448819,
+      "grad_norm": 7.979492664337158,
       "learning_rate": 5.719681881126195e-08,
-      "loss": 0.235,
+      "loss": 0.2066,
       "step": 2748
     },
     {
-      "epoch": 2.703740157480315,
-      "grad_norm": 8.779541969299316,
+      "epoch": 2.7057086614173227,
+      "grad_norm": 6.89271879196167,
       "learning_rate": 5.685064653106249e-08,
-      "loss": 0.3047,
+      "loss": 0.1971,
       "step": 2749
     },
     {
-      "epoch": 2.704724409448819,
-      "grad_norm": 5.974859237670898,
+      "epoch": 2.706692913385827,
+      "grad_norm": 9.40224838256836,
       "learning_rate": 5.6505527105960925e-08,
-      "loss": 0.2739,
+      "loss": 0.662,
       "step": 2750
     },
     {
-      "epoch": 2.7057086614173227,
-      "grad_norm": 11.043965339660645,
+      "epoch": 2.7076771653543306,
+      "grad_norm": 6.6087327003479,
       "learning_rate": 5.6161461352835975e-08,
-      "loss": 0.4118,
+      "loss": 0.2182,
       "step": 2751
     },
     {
-      "epoch": 2.706692913385827,
-      "grad_norm": 17.099946975708008,
+      "epoch": 2.7086614173228347,
+      "grad_norm": 12.50594711303711,
       "learning_rate": 5.5818450086072214e-08,
-      "loss": 1.0672,
+      "loss": 0.7571,
       "step": 2752
     },
     {
-      "epoch": 2.7076771653543306,
-      "grad_norm": 7.933254241943359,
+      "epoch": 2.7096456692913384,
+      "grad_norm": 9.776381492614746,
       "learning_rate": 5.547649411755852e-08,
-      "loss": 0.3133,
+      "loss": 0.443,
       "step": 2753
     },
     {
-      "epoch": 2.7086614173228347,
-      "grad_norm": 13.105121612548828,
+      "epoch": 2.7106299212598426,
+      "grad_norm": 8.475297927856445,
       "learning_rate": 5.5135594256686e-08,
-      "loss": 0.6007,
+      "loss": 0.3281,
       "step": 2754
     },
     {
-      "epoch": 2.7086614173228347,
-      "eval_Qnli-dev_cosine_accuracy": 0.67578125,
-      "eval_Qnli-dev_cosine_accuracy_threshold": 0.7977303266525269,
-      "eval_Qnli-dev_cosine_ap": 0.7019567607393539,
-      "eval_Qnli-dev_cosine_f1": 0.6915584415584415,
-      "eval_Qnli-dev_cosine_f1_threshold": 0.7133994102478027,
-      "eval_Qnli-dev_cosine_precision": 0.5605263157894737,
-      "eval_Qnli-dev_cosine_recall": 0.902542372881356,
-      "eval_Qnli-dev_dot_accuracy": 0.671875,
-      "eval_Qnli-dev_dot_accuracy_threshold": 608.2194213867188,
-      "eval_Qnli-dev_dot_ap": 0.6989171470401194,
-      "eval_Qnli-dev_dot_f1": 0.6918032786885246,
-      "eval_Qnli-dev_dot_f1_threshold": 548.5459594726562,
-      "eval_Qnli-dev_dot_precision": 0.5641711229946524,
-      "eval_Qnli-dev_dot_recall": 0.8940677966101694,
+      "epoch": 2.7106299212598426,
+      "eval_Qnli-dev_cosine_accuracy": 0.673828125,
+      "eval_Qnli-dev_cosine_accuracy_threshold": 0.8004011511802673,
+      "eval_Qnli-dev_cosine_ap": 0.7020060295913004,
+      "eval_Qnli-dev_cosine_f1": 0.6906710310965629,
+      "eval_Qnli-dev_cosine_f1_threshold": 0.7151964902877808,
+      "eval_Qnli-dev_cosine_precision": 0.5626666666666666,
+      "eval_Qnli-dev_cosine_recall": 0.8940677966101694,
+      "eval_Qnli-dev_dot_accuracy": 0.66796875,
+      "eval_Qnli-dev_dot_accuracy_threshold": 610.5162963867188,
+      "eval_Qnli-dev_dot_ap": 0.6990298974021447,
+      "eval_Qnli-dev_dot_f1": 0.6892109500805152,
+      "eval_Qnli-dev_dot_f1_threshold": 540.068603515625,
+      "eval_Qnli-dev_dot_precision": 0.5558441558441558,
+      "eval_Qnli-dev_dot_recall": 0.9067796610169492,
       "eval_Qnli-dev_euclidean_accuracy": 0.67578125,
-      "eval_Qnli-dev_euclidean_accuracy_threshold": 17.582195281982422,
-      "eval_Qnli-dev_euclidean_ap": 0.7027766922467029,
-      "eval_Qnli-dev_euclidean_f1": 0.6926829268292684,
-      "eval_Qnli-dev_euclidean_f1_threshold": 20.844341278076172,
-      "eval_Qnli-dev_euclidean_precision": 0.5620052770448549,
-      "eval_Qnli-dev_euclidean_recall": 0.902542372881356,
-      "eval_Qnli-dev_manhattan_accuracy": 0.69140625,
-      "eval_Qnli-dev_manhattan_accuracy_threshold": 378.83184814453125,
-      "eval_Qnli-dev_manhattan_ap": 0.7116177157083026,
-      "eval_Qnli-dev_manhattan_f1": 0.6935483870967742,
-      "eval_Qnli-dev_manhattan_f1_threshold": 444.3184814453125,
-      "eval_Qnli-dev_manhattan_precision": 0.5598958333333334,
-      "eval_Qnli-dev_manhattan_recall": 0.9110169491525424,
-      "eval_Qnli-dev_max_accuracy": 0.69140625,
-      "eval_Qnli-dev_max_accuracy_threshold": 608.2194213867188,
-      "eval_Qnli-dev_max_ap": 0.7116177157083026,
-      "eval_Qnli-dev_max_f1": 0.6935483870967742,
-      "eval_Qnli-dev_max_f1_threshold": 548.5459594726562,
-      "eval_Qnli-dev_max_precision": 0.5641711229946524,
-      "eval_Qnli-dev_max_recall": 0.9110169491525424,
-      "eval_allNLI-dev_cosine_accuracy": 0.6953125,
-      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9163441061973572,
-      "eval_allNLI-dev_cosine_ap": 0.5218088604536244,
-      "eval_allNLI-dev_cosine_f1": 0.5505882352941177,
-      "eval_allNLI-dev_cosine_f1_threshold": 0.8232982158660889,
-      "eval_allNLI-dev_cosine_precision": 0.4642857142857143,
-      "eval_allNLI-dev_cosine_recall": 0.6763005780346821,
+      "eval_Qnli-dev_euclidean_accuracy_threshold": 18.270591735839844,
+      "eval_Qnli-dev_euclidean_ap": 0.7016708232291158,
+      "eval_Qnli-dev_euclidean_f1": 0.6895424836601307,
+      "eval_Qnli-dev_euclidean_f1_threshold": 20.884302139282227,
+      "eval_Qnli-dev_euclidean_precision": 0.5611702127659575,
+      "eval_Qnli-dev_euclidean_recall": 0.8940677966101694,
+      "eval_Qnli-dev_manhattan_accuracy": 0.6875,
+      "eval_Qnli-dev_manhattan_accuracy_threshold": 385.834716796875,
+      "eval_Qnli-dev_manhattan_ap": 0.7101346716654798,
+      "eval_Qnli-dev_manhattan_f1": 0.6893203883495146,
+      "eval_Qnli-dev_manhattan_f1_threshold": 442.5292663574219,
+      "eval_Qnli-dev_manhattan_precision": 0.5575916230366492,
+      "eval_Qnli-dev_manhattan_recall": 0.902542372881356,
+      "eval_Qnli-dev_max_accuracy": 0.6875,
+      "eval_Qnli-dev_max_accuracy_threshold": 610.5162963867188,
+      "eval_Qnli-dev_max_ap": 0.7101346716654798,
+      "eval_Qnli-dev_max_f1": 0.6906710310965629,
+      "eval_Qnli-dev_max_f1_threshold": 540.068603515625,
+      "eval_Qnli-dev_max_precision": 0.5626666666666666,
+      "eval_Qnli-dev_max_recall": 0.9067796610169492,
+      "eval_allNLI-dev_cosine_accuracy": 0.697265625,
+      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9155057668685913,
+      "eval_allNLI-dev_cosine_ap": 0.5121080209650164,
+      "eval_allNLI-dev_cosine_f1": 0.5514223194748359,
+      "eval_allNLI-dev_cosine_f1_threshold": 0.8262423276901245,
+      "eval_allNLI-dev_cosine_precision": 0.44366197183098594,
+      "eval_allNLI-dev_cosine_recall": 0.7283236994219653,
       "eval_allNLI-dev_dot_accuracy": 0.693359375,
-      "eval_allNLI-dev_dot_accuracy_threshold": 701.0006103515625,
-      "eval_allNLI-dev_dot_ap": 0.5213952861248943,
-      "eval_allNLI-dev_dot_f1": 0.5509259259259259,
-      "eval_allNLI-dev_dot_f1_threshold": 625.916259765625,
-      "eval_allNLI-dev_dot_precision": 0.4594594594594595,
-      "eval_allNLI-dev_dot_recall": 0.6878612716763006,
-      "eval_allNLI-dev_euclidean_accuracy": 0.6953125,
-      "eval_allNLI-dev_euclidean_accuracy_threshold": 11.311955451965332,
-      "eval_allNLI-dev_euclidean_ap": 0.5215694226758205,
-      "eval_allNLI-dev_euclidean_f1": 0.5505882352941177,
-      "eval_allNLI-dev_euclidean_f1_threshold": 16.409137725830078,
-      "eval_allNLI-dev_euclidean_precision": 0.4642857142857143,
-      "eval_allNLI-dev_euclidean_recall": 0.6763005780346821,
-      "eval_allNLI-dev_manhattan_accuracy": 0.693359375,
-      "eval_allNLI-dev_manhattan_accuracy_threshold": 245.81846618652344,
-      "eval_allNLI-dev_manhattan_ap": 0.5220504800061738,
-      "eval_allNLI-dev_manhattan_f1": 0.5555555555555556,
-      "eval_allNLI-dev_manhattan_f1_threshold": 383.66192626953125,
-      "eval_allNLI-dev_manhattan_precision": 0.4154727793696275,
-      "eval_allNLI-dev_manhattan_recall": 0.838150289017341,
-      "eval_allNLI-dev_max_accuracy": 0.6953125,
-      "eval_allNLI-dev_max_accuracy_threshold": 701.0006103515625,
-      "eval_allNLI-dev_max_ap": 0.5220504800061738,
-      "eval_allNLI-dev_max_f1": 0.5555555555555556,
-      "eval_allNLI-dev_max_f1_threshold": 625.916259765625,
-      "eval_allNLI-dev_max_precision": 0.4642857142857143,
-      "eval_allNLI-dev_max_recall": 0.838150289017341,
-      "eval_loss": 0.8913472294807434,
-      "eval_runtime": 53.9182,
-      "eval_samples_per_second": 30.862,
-      "eval_sequential_score": 0.7116177157083026,
-      "eval_steps_per_second": 0.13,
-      "eval_sts-test_pearson_cosine": 0.6576157964894995,
-      "eval_sts-test_pearson_dot": 0.6569358484152671,
-      "eval_sts-test_pearson_euclidean": 0.6698101275286956,
-      "eval_sts-test_pearson_manhattan": 0.6729782820829205,
-      "eval_sts-test_pearson_max": 0.6729782820829205,
-      "eval_sts-test_spearman_cosine": 0.6645270090885942,
-      "eval_sts-test_spearman_dot": 0.6641536045635444,
-      "eval_sts-test_spearman_euclidean": 0.6648468463537585,
-      "eval_sts-test_spearman_manhattan": 0.6680530952274067,
-      "eval_sts-test_spearman_max": 0.6680530952274067,
+      "eval_allNLI-dev_dot_accuracy_threshold": 701.3142700195312,
+      "eval_allNLI-dev_dot_ap": 0.5104494310044695,
+      "eval_allNLI-dev_dot_f1": 0.5524193548387097,
+      "eval_allNLI-dev_dot_f1_threshold": 616.7257080078125,
+      "eval_allNLI-dev_dot_precision": 0.4241486068111455,
+      "eval_allNLI-dev_dot_recall": 0.791907514450867,
+      "eval_allNLI-dev_euclidean_accuracy": 0.697265625,
+      "eval_allNLI-dev_euclidean_accuracy_threshold": 11.370841979980469,
+      "eval_allNLI-dev_euclidean_ap": 0.5117166235619579,
+      "eval_allNLI-dev_euclidean_f1": 0.5514223194748359,
+      "eval_allNLI-dev_euclidean_f1_threshold": 16.276464462280273,
+      "eval_allNLI-dev_euclidean_precision": 0.44366197183098594,
+      "eval_allNLI-dev_euclidean_recall": 0.7283236994219653,
+      "eval_allNLI-dev_manhattan_accuracy": 0.69140625,
+      "eval_allNLI-dev_manhattan_accuracy_threshold": 230.18153381347656,
+      "eval_allNLI-dev_manhattan_ap": 0.5085875515850806,
+      "eval_allNLI-dev_manhattan_f1": 0.5567928730512249,
+      "eval_allNLI-dev_manhattan_f1_threshold": 344.5216064453125,
+      "eval_allNLI-dev_manhattan_precision": 0.4528985507246377,
+      "eval_allNLI-dev_manhattan_recall": 0.7225433526011561,
+      "eval_allNLI-dev_max_accuracy": 0.697265625,
+      "eval_allNLI-dev_max_accuracy_threshold": 701.3142700195312,
+      "eval_allNLI-dev_max_ap": 0.5121080209650164,
+      "eval_allNLI-dev_max_f1": 0.5567928730512249,
+      "eval_allNLI-dev_max_f1_threshold": 616.7257080078125,
+      "eval_allNLI-dev_max_precision": 0.4528985507246377,
+      "eval_allNLI-dev_max_recall": 0.791907514450867,
+      "eval_loss": 0.7940033674240112,
+      "eval_runtime": 53.0242,
+      "eval_samples_per_second": 31.382,
+      "eval_sequential_score": 0.7101346716654798,
+      "eval_steps_per_second": 0.132,
+      "eval_sts-test_pearson_cosine": 0.6675668552689955,
+      "eval_sts-test_pearson_dot": 0.665684271163508,
+      "eval_sts-test_pearson_euclidean": 0.6793524734927204,
+      "eval_sts-test_pearson_manhattan": 0.682946521113783,
+      "eval_sts-test_pearson_max": 0.682946521113783,
+      "eval_sts-test_spearman_cosine": 0.6729811936343505,
+      "eval_sts-test_spearman_dot": 0.6700007148744218,
+      "eval_sts-test_spearman_euclidean": 0.6734100449722642,
+      "eval_sts-test_spearman_manhattan": 0.6768540125464235,
+      "eval_sts-test_spearman_max": 0.6768540125464235,
       "step": 2754
     },
     {
-      "epoch": 2.7096456692913384,
-      "grad_norm": 9.957964897155762,
+      "epoch": 2.7116141732283463,
+      "grad_norm": 6.073899269104004,
       "learning_rate": 5.4795751310345864e-08,
-      "loss": 0.4171,
+      "loss": 0.1828,
       "step": 2755
     },
     {
-      "epoch": 2.7106299212598426,
-      "grad_norm": 5.921916961669922,
+      "epoch": 2.7125984251968505,
+      "grad_norm": 6.912418842315674,
       "learning_rate": 5.445696608292781e-08,
-      "loss": 0.1617,
+      "loss": 0.2148,
       "step": 2756
     },
     {
-      "epoch": 2.7116141732283463,
-      "grad_norm": 6.3721137046813965,
+      "epoch": 2.713582677165354,
+      "grad_norm": 7.220956325531006,
       "learning_rate": 5.4119239376317644e-08,
-      "loss": 0.2103,
+      "loss": 0.3059,
       "step": 2757
     },
     {
-      "epoch": 2.7125984251968505,
-      "grad_norm": 9.366639137268066,
+      "epoch": 2.7145669291338583,
+      "grad_norm": 9.336836814880371,
       "learning_rate": 5.378257198989609e-08,
-      "loss": 0.3573,
+      "loss": 0.2714,
       "step": 2758
     },
     {
-      "epoch": 2.713582677165354,
-      "grad_norm": 9.169002532958984,
+      "epoch": 2.715551181102362,
+      "grad_norm": 6.493551254272461,
       "learning_rate": 5.344696472053635e-08,
-      "loss": 0.3494,
+      "loss": 0.1713,
       "step": 2759
     },
     {
-      "epoch": 2.7145669291338583,
-      "grad_norm": 15.052176475524902,
+      "epoch": 2.716535433070866,
+      "grad_norm": 11.62325382232666,
       "learning_rate": 5.3112418362602364e-08,
-      "loss": 0.6991,
+      "loss": 0.3999,
       "step": 2760
     },
     {
-      "epoch": 2.715551181102362,
-      "grad_norm": 10.323905944824219,
+      "epoch": 2.71751968503937,
+      "grad_norm": 10.040081977844238,
       "learning_rate": 5.2778933707947035e-08,
-      "loss": 0.5728,
+      "loss": 0.4077,
       "step": 2761
     },
     {
-      "epoch": 2.716535433070866,
-      "grad_norm": 7.4978346824646,
+      "epoch": 2.718503937007874,
+      "grad_norm": 7.886175632476807,
       "learning_rate": 5.2446511545910186e-08,
-      "loss": 0.1708,
+      "loss": 0.3149,
       "step": 2762
     },
     {
-      "epoch": 2.71751968503937,
-      "grad_norm": 11.339003562927246,
+      "epoch": 2.719488188976378,
+      "grad_norm": 6.3333210945129395,
       "learning_rate": 5.2115152663316917e-08,
-      "loss": 0.5428,
+      "loss": 0.2239,
       "step": 2763
     },
     {
-      "epoch": 2.718503937007874,
-      "grad_norm": 7.155446529388428,
+      "epoch": 2.720472440944882,
+      "grad_norm": 6.320239067077637,
       "learning_rate": 5.1784857844475405e-08,
-      "loss": 0.249,
+      "loss": 0.2417,
       "step": 2764
     },
     {
-      "epoch": 2.719488188976378,
-      "grad_norm": 8.376879692077637,
+      "epoch": 2.721456692913386,
+      "grad_norm": 7.927882194519043,
       "learning_rate": 5.145562787117536e-08,
-      "loss": 0.3401,
+      "loss": 0.1812,
       "step": 2765
     },
     {
-      "epoch": 2.720472440944882,
-      "grad_norm": 9.51554012298584,
+      "epoch": 2.72244094488189,
+      "grad_norm": 9.089080810546875,
       "learning_rate": 5.112746352268614e-08,
-      "loss": 0.3281,
+      "loss": 0.3691,
       "step": 2766
     },
     {
-      "epoch": 2.721456692913386,
-      "grad_norm": 11.976239204406738,
+      "epoch": 2.7234251968503935,
+      "grad_norm": 10.0004243850708,
       "learning_rate": 5.0800365575754697e-08,
-      "loss": 0.3198,
+      "loss": 0.2731,
       "step": 2767
     },
     {
-      "epoch": 2.72244094488189,
-      "grad_norm": 4.413916110992432,
+      "epoch": 2.7244094488188977,
+      "grad_norm": 7.451738357543945,
       "learning_rate": 5.0474334804603915e-08,
-      "loss": 0.1,
+      "loss": 0.1939,
       "step": 2768
     },
     {
-      "epoch": 2.7234251968503935,
-      "grad_norm": 11.406291007995605,
+      "epoch": 2.725393700787402,
+      "grad_norm": 6.70740270614624,
       "learning_rate": 5.0149371980930695e-08,
-      "loss": 0.4922,
+      "loss": 0.185,
       "step": 2769
     },
     {
-      "epoch": 2.7244094488188977,
-      "grad_norm": 10.862181663513184,
+      "epoch": 2.7263779527559056,
+      "grad_norm": 6.52256965637207,
       "learning_rate": 4.9825477873904235e-08,
-      "loss": 0.449,
+      "loss": 0.1714,
       "step": 2770
     },
     {
-      "epoch": 2.725393700787402,
-      "grad_norm": 11.480317115783691,
+      "epoch": 2.7273622047244093,
+      "grad_norm": 11.203923225402832,
       "learning_rate": 4.950265325016404e-08,
-      "loss": 0.3628,
+      "loss": 0.4337,
       "step": 2771
     },
     {
-      "epoch": 2.7263779527559056,
-      "grad_norm": 4.006394863128662,
+      "epoch": 2.7283464566929134,
+      "grad_norm": 6.011223793029785,
       "learning_rate": 4.9180898873818494e-08,
-      "loss": 0.1053,
+      "loss": 0.2208,
       "step": 2772
     },
     {
-      "epoch": 2.7273622047244093,
-      "grad_norm": 6.242520809173584,
+      "epoch": 2.7293307086614176,
+      "grad_norm": 7.703699111938477,
       "learning_rate": 4.88602155064422e-08,
-      "loss": 0.2412,
+      "loss": 0.1848,
       "step": 2773
     },
     {
-      "epoch": 2.7283464566929134,
-      "grad_norm": 9.83320140838623,
+      "epoch": 2.7303149606299213,
+      "grad_norm": 12.34919261932373,
       "learning_rate": 4.8540603907075335e-08,
-      "loss": 0.3419,
+      "loss": 0.4455,
       "step": 2774
     },
     {
-      "epoch": 2.7293307086614176,
-      "grad_norm": 8.112334251403809,
+      "epoch": 2.731299212598425,
+      "grad_norm": 8.315437316894531,
       "learning_rate": 4.822206483222088e-08,
-      "loss": 0.2806,
+      "loss": 0.1928,
       "step": 2775
     },
     {
-      "epoch": 2.7303149606299213,
-      "grad_norm": 11.555301666259766,
+      "epoch": 2.732283464566929,
+      "grad_norm": 6.745145320892334,
       "learning_rate": 4.790459903584333e-08,
-      "loss": 0.4546,
+      "loss": 0.2259,
       "step": 2776
     },
     {
-      "epoch": 2.731299212598425,
-      "grad_norm": 11.073164939880371,
+      "epoch": 2.7332677165354333,
+      "grad_norm": 5.3009514808654785,
       "learning_rate": 4.7588207269366905e-08,
-      "loss": 0.3591,
+      "loss": 0.1529,
       "step": 2777
     },
     {
-      "epoch": 2.732283464566929,
-      "grad_norm": 4.647140026092529,
+      "epoch": 2.734251968503937,
+      "grad_norm": 8.33968734741211,
       "learning_rate": 4.7272890281673495e-08,
-      "loss": 0.097,
+      "loss": 0.4687,
       "step": 2778
     },
     {
-      "epoch": 2.7332677165354333,
-      "grad_norm": 8.033677101135254,
+      "epoch": 2.7352362204724407,
+      "grad_norm": 9.36735725402832,
       "learning_rate": 4.695864881910104e-08,
-      "loss": 0.3075,
+      "loss": 0.4754,
       "step": 2779
     },
     {
-      "epoch": 2.734251968503937,
-      "grad_norm": 7.941624164581299,
+      "epoch": 2.736220472440945,
+      "grad_norm": 9.189268112182617,
       "learning_rate": 4.6645483625441925e-08,
-      "loss": 0.2836,
+      "loss": 0.2371,
       "step": 2780
     },
     {
-      "epoch": 2.7352362204724407,
-      "grad_norm": 12.40319538116455,
+      "epoch": 2.737204724409449,
+      "grad_norm": 8.296624183654785,
       "learning_rate": 4.6333395441940915e-08,
-      "loss": 0.5892,
+      "loss": 0.4237,
       "step": 2781
     },
     {
-      "epoch": 2.736220472440945,
-      "grad_norm": 7.96142053604126,
+      "epoch": 2.7381889763779528,
+      "grad_norm": 9.132102012634277,
       "learning_rate": 4.60223850072936e-08,
-      "loss": 0.2919,
+      "loss": 0.2798,
       "step": 2782
     },
     {
-      "epoch": 2.737204724409449,
-      "grad_norm": 8.30141544342041,
+      "epoch": 2.7391732283464565,
+      "grad_norm": 11.105525970458984,
       "learning_rate": 4.571245305764458e-08,
-      "loss": 0.307,
+      "loss": 0.5882,
       "step": 2783
     },
     {
-      "epoch": 2.7381889763779528,
-      "grad_norm": 8.26657485961914,
+      "epoch": 2.7401574803149606,
+      "grad_norm": 10.983388900756836,
       "learning_rate": 4.540360032658586e-08,
-      "loss": 0.2358,
+      "loss": 0.5746,
       "step": 2784
     },
     {
-      "epoch": 2.7391732283464565,
-      "grad_norm": 22.45551872253418,
+      "epoch": 2.741141732283465,
+      "grad_norm": 12.174887657165527,
       "learning_rate": 4.5095827545154776e-08,
-      "loss": 1.1891,
+      "loss": 0.4695,
       "step": 2785
     },
     {
-      "epoch": 2.7401574803149606,
-      "grad_norm": 17.988201141357422,
+      "epoch": 2.7421259842519685,
+      "grad_norm": 7.856240272521973,
       "learning_rate": 4.478913544183271e-08,
-      "loss": 0.6927,
+      "loss": 0.2319,
       "step": 2786
     },
     {
-      "epoch": 2.741141732283465,
-      "grad_norm": 13.452264785766602,
+      "epoch": 2.7431102362204722,
+      "grad_norm": 5.020266532897949,
       "learning_rate": 4.4483524742542986e-08,
-      "loss": 0.2876,
+      "loss": 0.129,
       "step": 2787
     },
     {
-      "epoch": 2.7421259842519685,
-      "grad_norm": 10.717510223388672,
+      "epoch": 2.7440944881889764,
+      "grad_norm": 8.512626647949219,
       "learning_rate": 4.417899617064943e-08,
-      "loss": 0.2945,
+      "loss": 0.365,
       "step": 2788
     },
     {
-      "epoch": 2.7431102362204722,
-      "grad_norm": 8.089428901672363,
+      "epoch": 2.7450787401574805,
+      "grad_norm": 13.706979751586914,
       "learning_rate": 4.387555044695425e-08,
-      "loss": 0.2065,
+      "loss": 0.9308,
       "step": 2789
     },
     {
-      "epoch": 2.7440944881889764,
-      "grad_norm": 8.970033645629883,
+      "epoch": 2.7460629921259843,
+      "grad_norm": 13.032624244689941,
       "learning_rate": 4.3573188289696984e-08,
-      "loss": 0.4644,
+      "loss": 0.5642,
       "step": 2790
     },
     {
-      "epoch": 2.7450787401574805,
-      "grad_norm": 8.521092414855957,
+      "epoch": 2.747047244094488,
+      "grad_norm": 8.541265487670898,
       "learning_rate": 4.3271910414552155e-08,
-      "loss": 0.4232,
+      "loss": 0.3723,
       "step": 2791
     },
     {
-      "epoch": 2.7460629921259843,
-      "grad_norm": 15.873388290405273,
+      "epoch": 2.748031496062992,
+      "grad_norm": 6.123878002166748,
       "learning_rate": 4.297171753462808e-08,
-      "loss": 0.6449,
+      "loss": 0.1993,
       "step": 2792
     },
     {
-      "epoch": 2.747047244094488,
-      "grad_norm": 8.041421890258789,
+      "epoch": 2.7490157480314963,
+      "grad_norm": 8.669310569763184,
       "learning_rate": 4.267261036046481e-08,
-      "loss": 0.3795,
+      "loss": 0.3233,
       "step": 2793
     },
     {
-      "epoch": 2.748031496062992,
-      "grad_norm": 8.81619644165039,
+      "epoch": 2.75,
+      "grad_norm": 12.549506187438965,
       "learning_rate": 4.2374589600032454e-08,
-      "loss": 0.2358,
+      "loss": 0.5217,
       "step": 2794
     },
     {
-      "epoch": 2.7490157480314963,
-      "grad_norm": 10.297800064086914,
+      "epoch": 2.7509842519685037,
+      "grad_norm": 12.702281951904297,
       "learning_rate": 4.207765595872984e-08,
-      "loss": 0.4759,
+      "loss": 0.6836,
       "step": 2795
     },
     {
-      "epoch": 2.75,
-      "grad_norm": 18.464712142944336,
+      "epoch": 2.751968503937008,
+      "grad_norm": 5.9549946784973145,
       "learning_rate": 4.1781810139382675e-08,
-      "loss": 0.7407,
+      "loss": 0.159,
       "step": 2796
     },
     {
-      "epoch": 2.7509842519685037,
-      "grad_norm": 12.736725807189941,
+      "epoch": 2.752952755905512,
+      "grad_norm": 14.522462844848633,
       "learning_rate": 4.148705284224176e-08,
-      "loss": 0.37,
+      "loss": 0.7725,
       "step": 2797
     },
     {
-      "epoch": 2.751968503937008,
-      "grad_norm": 9.105003356933594,
+      "epoch": 2.7539370078740157,
+      "grad_norm": 6.126986503601074,
       "learning_rate": 4.1193384764981336e-08,
-      "loss": 0.2381,
+      "loss": 0.2326,
       "step": 2798
     },
     {
-      "epoch": 2.752952755905512,
-      "grad_norm": 8.26710319519043,
+      "epoch": 2.7549212598425195,
+      "grad_norm": 6.900618076324463,
       "learning_rate": 4.090080660269774e-08,
-      "loss": 0.3177,
+      "loss": 0.1674,
       "step": 2799
     },
     {
-      "epoch": 2.7539370078740157,
-      "grad_norm": 8.486204147338867,
+      "epoch": 2.7559055118110236,
+      "grad_norm": 9.16586971282959,
       "learning_rate": 4.060931904790737e-08,
-      "loss": 0.2014,
+      "loss": 0.5721,
       "step": 2800
     },
     {
-      "epoch": 2.7549212598425195,
-      "grad_norm": 7.135037899017334,
+      "epoch": 2.7568897637795278,
+      "grad_norm": 6.684321403503418,
       "learning_rate": 4.0318922790545247e-08,
-      "loss": 0.1755,
+      "loss": 0.1241,
       "step": 2801
     },
     {
-      "epoch": 2.7559055118110236,
-      "grad_norm": 11.886211395263672,
+      "epoch": 2.7578740157480315,
+      "grad_norm": 8.893206596374512,
       "learning_rate": 4.0029618517963386e-08,
-      "loss": 0.5556,
+      "loss": 0.3849,
       "step": 2802
     },
     {
-      "epoch": 2.7568897637795278,
-      "grad_norm": 8.875014305114746,
+      "epoch": 2.758858267716535,
+      "grad_norm": 9.859920501708984,
       "learning_rate": 3.974140691492909e-08,
-      "loss": 0.292,
+      "loss": 0.4502,
       "step": 2803
     },
     {
-      "epoch": 2.7578740157480315,
-      "grad_norm": 8.197360038757324,
+      "epoch": 2.7598425196850394,
+      "grad_norm": 5.285876274108887,
       "learning_rate": 3.9454288663623483e-08,
-      "loss": 0.2805,
+      "loss": 0.1952,
       "step": 2804
     },
     {
-      "epoch": 2.758858267716535,
-      "grad_norm": 7.326289653778076,
+      "epoch": 2.7608267716535435,
+      "grad_norm": 7.1930155754089355,
       "learning_rate": 3.916826444363958e-08,
-      "loss": 0.361,
+      "loss": 0.2037,
       "step": 2805
     },
     {
-      "epoch": 2.7598425196850394,
-      "grad_norm": 9.661887168884277,
+      "epoch": 2.7618110236220472,
+      "grad_norm": 9.012578010559082,
       "learning_rate": 3.888333493198107e-08,
-      "loss": 0.4121,
+      "loss": 0.4054,
       "step": 2806
     },
     {
-      "epoch": 2.7608267716535435,
-      "grad_norm": 7.319786548614502,
+      "epoch": 2.762795275590551,
+      "grad_norm": 7.291186332702637,
       "learning_rate": 3.859950080306051e-08,
-      "loss": 0.2068,
+      "loss": 0.2229,
       "step": 2807
     },
     {
-      "epoch": 2.7618110236220472,
-      "grad_norm": 12.52310562133789,
+      "epoch": 2.763779527559055,
+      "grad_norm": 5.994788646697998,
       "learning_rate": 3.8316762728697635e-08,
-      "loss": 0.548,
+      "loss": 0.1633,
       "step": 2808
     },
     {
-      "epoch": 2.762795275590551,
-      "grad_norm": 8.638126373291016,
+      "epoch": 2.7647637795275593,
+      "grad_norm": 10.90695858001709,
       "learning_rate": 3.8035121378117895e-08,
-      "loss": 0.3264,
+      "loss": 0.3005,
       "step": 2809
     },
     {
-      "epoch": 2.763779527559055,
-      "grad_norm": 13.303729057312012,
+      "epoch": 2.765748031496063,
+      "grad_norm": 8.864053726196289,
       "learning_rate": 3.7754577417951074e-08,
-      "loss": 0.5705,
+      "loss": 0.347,
       "step": 2810
     },
     {
-      "epoch": 2.7647637795275593,
-      "grad_norm": 8.288713455200195,
+      "epoch": 2.7667322834645667,
+      "grad_norm": 6.2091755867004395,
       "learning_rate": 3.747513151222915e-08,
-      "loss": 0.3383,
+      "loss": 0.1762,
       "step": 2811
     },
     {
-      "epoch": 2.765748031496063,
-      "grad_norm": 9.315450668334961,
+      "epoch": 2.767716535433071,
+      "grad_norm": 8.503474235534668,
       "learning_rate": 3.719678432238533e-08,
-      "loss": 0.4793,
+      "loss": 0.3815,
       "step": 2812
     },
     {
-      "epoch": 2.7667322834645667,
-      "grad_norm": 8.430915832519531,
+      "epoch": 2.768700787401575,
+      "grad_norm": 9.022677421569824,
       "learning_rate": 3.691953650725206e-08,
-      "loss": 0.3474,
+      "loss": 0.2376,
       "step": 2813
     },
     {
-      "epoch": 2.767716535433071,
-      "grad_norm": 9.265016555786133,
+      "epoch": 2.7696850393700787,
+      "grad_norm": 9.856568336486816,
       "learning_rate": 3.664338872305991e-08,
-      "loss": 0.3937,
+      "loss": 0.2882,
       "step": 2814
     },
     {
-      "epoch": 2.768700787401575,
-      "grad_norm": 13.440638542175293,
+      "epoch": 2.7706692913385824,
+      "grad_norm": 11.792848587036133,
       "learning_rate": 3.6368341623435226e-08,
-      "loss": 0.4636,
+      "loss": 0.6411,
       "step": 2815
     },
     {
-      "epoch": 2.7696850393700787,
-      "grad_norm": 9.833179473876953,
+      "epoch": 2.7716535433070866,
+      "grad_norm": 4.4309539794921875,
       "learning_rate": 3.609439585939952e-08,
-      "loss": 0.4123,
+      "loss": 0.0842,
       "step": 2816
     },
     {
-      "epoch": 2.7706692913385824,
-      "grad_norm": 6.852646827697754,
+      "epoch": 2.7726377952755907,
+      "grad_norm": 7.737056732177734,
       "learning_rate": 3.5821552079367375e-08,
-      "loss": 0.3588,
+      "loss": 0.2796,
       "step": 2817
     },
     {
-      "epoch": 2.7716535433070866,
-      "grad_norm": 7.504284381866455,
+      "epoch": 2.7736220472440944,
+      "grad_norm": 11.905488967895508,
       "learning_rate": 3.554981092914507e-08,
-      "loss": 0.1521,
+      "loss": 0.4634,
       "step": 2818
     },
     {
-      "epoch": 2.7726377952755907,
-      "grad_norm": 10.694111824035645,
+      "epoch": 2.774606299212598,
+      "grad_norm": 9.010229110717773,
       "learning_rate": 3.527917305192892e-08,
-      "loss": 0.4246,
+      "loss": 0.3445,
       "step": 2819
     },
     {
-      "epoch": 2.7736220472440944,
-      "grad_norm": 11.322754859924316,
+      "epoch": 2.7755905511811023,
+      "grad_norm": 12.911606788635254,
       "learning_rate": 3.500963908830402e-08,
-      "loss": 0.4689,
+      "loss": 0.7675,
       "step": 2820
     },
     {
-      "epoch": 2.774606299212598,
-      "grad_norm": 7.798206806182861,
+      "epoch": 2.7765748031496065,
+      "grad_norm": 9.425992965698242,
       "learning_rate": 3.474120967624232e-08,
-      "loss": 0.2204,
+      "loss": 0.5084,
       "step": 2821
     },
     {
-      "epoch": 2.7755905511811023,
-      "grad_norm": 16.072729110717773,
+      "epoch": 2.77755905511811,
+      "grad_norm": 6.954996585845947,
       "learning_rate": 3.4473885451101665e-08,
-      "loss": 0.4716,
+      "loss": 0.2283,
       "step": 2822
     },
     {
-      "epoch": 2.7765748031496065,
-      "grad_norm": 8.112894058227539,
+      "epoch": 2.778543307086614,
+      "grad_norm": 12.67591381072998,
       "learning_rate": 3.4207667045623697e-08,
-      "loss": 0.2358,
+      "loss": 0.7626,
       "step": 2823
     },
     {
-      "epoch": 2.77755905511811,
-      "grad_norm": 11.253445625305176,
+      "epoch": 2.779527559055118,
+      "grad_norm": 10.049227714538574,
       "learning_rate": 3.394255508993286e-08,
-      "loss": 0.5598,
+      "loss": 0.4696,
       "step": 2824
     },
     {
-      "epoch": 2.778543307086614,
-      "grad_norm": 5.37000036239624,
+      "epoch": 2.780511811023622,
+      "grad_norm": 8.578816413879395,
       "learning_rate": 3.3678550211534556e-08,
-      "loss": 0.1785,
+      "loss": 0.3666,
       "step": 2825
     },
     {
-      "epoch": 2.779527559055118,
-      "grad_norm": 5.042856216430664,
+      "epoch": 2.781496062992126,
+      "grad_norm": 10.245843887329102,
       "learning_rate": 3.341565303531378e-08,
-      "loss": 0.1104,
+      "loss": 0.3069,
       "step": 2826
     },
     {
-      "epoch": 2.780511811023622,
-      "grad_norm": 11.312214851379395,
+      "epoch": 2.78248031496063,
+      "grad_norm": 10.3125581741333,
       "learning_rate": 3.315386418353376e-08,
-      "loss": 0.5949,
+      "loss": 0.5439,
       "step": 2827
     },
     {
-      "epoch": 2.781496062992126,
-      "grad_norm": 5.814931392669678,
+      "epoch": 2.783464566929134,
+      "grad_norm": 8.867962837219238,
       "learning_rate": 3.289318427583434e-08,
-      "loss": 0.1517,
+      "loss": 0.2623,
       "step": 2828
     },
     {
-      "epoch": 2.78248031496063,
-      "grad_norm": 13.6033353805542,
+      "epoch": 2.784448818897638,
+      "grad_norm": 7.080619812011719,
       "learning_rate": 3.263361392923056e-08,
-      "loss": 0.5286,
+      "loss": 0.1852,
       "step": 2829
     },
     {
-      "epoch": 2.783464566929134,
-      "grad_norm": 7.42643928527832,
+      "epoch": 2.7854330708661417,
+      "grad_norm": 11.38115119934082,
       "learning_rate": 3.2375153758111094e-08,
-      "loss": 0.1825,
+      "loss": 0.6098,
       "step": 2830
     },
     {
-      "epoch": 2.784448818897638,
-      "grad_norm": 8.158123970031738,
+      "epoch": 2.786417322834646,
+      "grad_norm": 7.123397350311279,
       "learning_rate": 3.211780437423711e-08,
-      "loss": 0.1949,
+      "loss": 0.2909,
       "step": 2831
     },
     {
-      "epoch": 2.7854330708661417,
-      "grad_norm": 6.828018665313721,
+      "epoch": 2.7874015748031495,
+      "grad_norm": 10.4204683303833,
       "learning_rate": 3.186156638674044e-08,
-      "loss": 0.2943,
+      "loss": 0.3049,
       "step": 2832
     },
     {
-      "epoch": 2.786417322834646,
-      "grad_norm": 8.297696113586426,
+      "epoch": 2.7883858267716537,
+      "grad_norm": 5.15102481842041,
       "learning_rate": 3.160644040212225e-08,
-      "loss": 0.4348,
+      "loss": 0.1428,
       "step": 2833
     },
     {
-      "epoch": 2.7874015748031495,
-      "grad_norm": 12.7998046875,
+      "epoch": 2.7893700787401574,
+      "grad_norm": 13.16177749633789,
       "learning_rate": 3.135242702425176e-08,
-      "loss": 0.7606,
+      "loss": 0.6295,
       "step": 2834
     },
     {
-      "epoch": 2.7883858267716537,
-      "grad_norm": 4.100052356719971,
+      "epoch": 2.7903543307086616,
+      "grad_norm": 10.833467483520508,
       "learning_rate": 3.109952685436487e-08,
-      "loss": 0.0896,
+      "loss": 0.345,
       "step": 2835
     },
     {
-      "epoch": 2.7893700787401574,
-      "grad_norm": 11.113504409790039,
+      "epoch": 2.7913385826771653,
+      "grad_norm": 7.641319751739502,
       "learning_rate": 3.0847740491062106e-08,
-      "loss": 0.5316,
+      "loss": 0.1657,
       "step": 2836
     },
     {
-      "epoch": 2.7903543307086616,
-      "grad_norm": 7.46032190322876,
+      "epoch": 2.7923228346456694,
+      "grad_norm": 13.410679817199707,
       "learning_rate": 3.059706853030815e-08,
-      "loss": 0.2567,
+      "loss": 0.903,
       "step": 2837
     },
     {
-      "epoch": 2.7913385826771653,
-      "grad_norm": 12.460062980651855,
+      "epoch": 2.793307086614173,
+      "grad_norm": 7.318542957305908,
       "learning_rate": 3.034751156542968e-08,
-      "loss": 0.288,
+      "loss": 0.2197,
       "step": 2838
     },
     {
-      "epoch": 2.7923228346456694,
-      "grad_norm": 8.528483390808105,
+      "epoch": 2.7942913385826773,
+      "grad_norm": 7.567079544067383,
       "learning_rate": 3.009907018711438e-08,
-      "loss": 0.3627,
+      "loss": 0.2575,
       "step": 2839
     },
     {
-      "epoch": 2.793307086614173,
-      "grad_norm": 11.319576263427734,
+      "epoch": 2.795275590551181,
+      "grad_norm": 6.560040473937988,
       "learning_rate": 2.985174498340938e-08,
-      "loss": 0.5699,
+      "loss": 0.1947,
       "step": 2840
     },
     {
-      "epoch": 2.7942913385826773,
-      "grad_norm": 12.464033126831055,
+      "epoch": 2.796259842519685,
+      "grad_norm": 9.053120613098145,
       "learning_rate": 2.9605536539719827e-08,
-      "loss": 0.5515,
+      "loss": 0.2109,
       "step": 2841
     },
     {
-      "epoch": 2.795275590551181,
-      "grad_norm": 11.264272689819336,
+      "epoch": 2.797244094488189,
+      "grad_norm": 6.420078277587891,
       "learning_rate": 2.9360445438807588e-08,
-      "loss": 0.5343,
+      "loss": 0.1597,
       "step": 2842
     },
     {
-      "epoch": 2.796259842519685,
-      "grad_norm": 5.43027925491333,
+      "epoch": 2.798228346456693,
+      "grad_norm": 10.127596855163574,
       "learning_rate": 2.9116472260789813e-08,
-      "loss": 0.1174,
+      "loss": 0.4367,
       "step": 2843
     },
     {
-      "epoch": 2.797244094488189,
-      "grad_norm": 7.383944034576416,
+      "epoch": 2.7992125984251968,
+      "grad_norm": 6.173708438873291,
       "learning_rate": 2.887361758313769e-08,
-      "loss": 0.1913,
+      "loss": 0.169,
       "step": 2844
     },
     {
-      "epoch": 2.798228346456693,
-      "grad_norm": 10.380624771118164,
+      "epoch": 2.800196850393701,
+      "grad_norm": 10.47236156463623,
       "learning_rate": 2.8631881980674824e-08,
-      "loss": 0.3796,
+      "loss": 0.4285,
       "step": 2845
     },
     {
-      "epoch": 2.7992125984251968,
-      "grad_norm": 7.057868957519531,
+      "epoch": 2.8011811023622046,
+      "grad_norm": 10.154450416564941,
       "learning_rate": 2.83912660255762e-08,
-      "loss": 0.1732,
+      "loss": 0.5736,
       "step": 2846
     },
     {
-      "epoch": 2.800196850393701,
-      "grad_norm": 11.689227104187012,
+      "epoch": 2.802165354330709,
+      "grad_norm": 8.472715377807617,
       "learning_rate": 2.815177028736651e-08,
-      "loss": 0.5418,
+      "loss": 0.3591,
       "step": 2847
     },
     {
-      "epoch": 2.8011811023622046,
-      "grad_norm": 8.381189346313477,
+      "epoch": 2.8031496062992125,
+      "grad_norm": 7.559357166290283,
       "learning_rate": 2.7913395332919e-08,
-      "loss": 0.3664,
+      "loss": 0.2364,
       "step": 2848
     },
     {
-      "epoch": 2.802165354330709,
-      "grad_norm": 16.76746940612793,
+      "epoch": 2.8041338582677167,
+      "grad_norm": 11.218221664428711,
       "learning_rate": 2.7676141726454192e-08,
-      "loss": 1.0863,
+      "loss": 0.3626,
       "step": 2849
     },
     {
-      "epoch": 2.8031496062992125,
-      "grad_norm": 6.873603343963623,
+      "epoch": 2.8051181102362204,
+      "grad_norm": 5.1599884033203125,
       "learning_rate": 2.744001002953829e-08,
-      "loss": 0.2157,
+      "loss": 0.1216,
       "step": 2850
     },
     {
-      "epoch": 2.8041338582677167,
-      "grad_norm": 7.641900062561035,
+      "epoch": 2.8061023622047245,
+      "grad_norm": 8.414644241333008,
       "learning_rate": 2.7205000801082003e-08,
-      "loss": 0.2521,
+      "loss": 0.3223,
       "step": 2851
     },
     {
-      "epoch": 2.8051181102362204,
-      "grad_norm": 7.553186893463135,
+      "epoch": 2.8070866141732282,
+      "grad_norm": 6.221158981323242,
       "learning_rate": 2.6971114597339394e-08,
-      "loss": 0.2675,
+      "loss": 0.2612,
       "step": 2852
     },
     {
-      "epoch": 2.8061023622047245,
-      "grad_norm": 5.070723056793213,
+      "epoch": 2.8080708661417324,
+      "grad_norm": 5.908691883087158,
       "learning_rate": 2.6738351971906158e-08,
-      "loss": 0.123,
+      "loss": 0.1934,
       "step": 2853
     },
     {
-      "epoch": 2.8070866141732282,
-      "grad_norm": 9.046456336975098,
+      "epoch": 2.809055118110236,
+      "grad_norm": 8.381319046020508,
       "learning_rate": 2.6506713475718692e-08,
-      "loss": 0.2632,
+      "loss": 0.3137,
       "step": 2854
     },
     {
-      "epoch": 2.8080708661417324,
-      "grad_norm": 8.117990493774414,
+      "epoch": 2.8100393700787403,
+      "grad_norm": 8.013070106506348,
       "learning_rate": 2.6276199657052583e-08,
-      "loss": 0.2046,
+      "loss": 0.2874,
       "step": 2855
     },
     {
-      "epoch": 2.809055118110236,
-      "grad_norm": 7.5084228515625,
+      "epoch": 2.811023622047244,
+      "grad_norm": 8.19188404083252,
       "learning_rate": 2.604681106152153e-08,
-      "loss": 0.2534,
+      "loss": 0.2845,
       "step": 2856
     },
     {
-      "epoch": 2.8100393700787403,
-      "grad_norm": 1.9132944345474243,
+      "epoch": 2.812007874015748,
+      "grad_norm": 10.451369285583496,
       "learning_rate": 2.581854823207548e-08,
-      "loss": 0.0277,
+      "loss": 0.4606,
       "step": 2857
     },
     {
-      "epoch": 2.811023622047244,
-      "grad_norm": 7.12310791015625,
+      "epoch": 2.812992125984252,
+      "grad_norm": 9.580084800720215,
       "learning_rate": 2.5591411709000163e-08,
-      "loss": 0.2174,
+      "loss": 0.4252,
       "step": 2858
     },
     {
-      "epoch": 2.812007874015748,
-      "grad_norm": 8.81682014465332,
+      "epoch": 2.813976377952756,
+      "grad_norm": 6.034160137176514,
       "learning_rate": 2.5365402029915256e-08,
-      "loss": 0.2608,
+      "loss": 0.1587,
       "step": 2859
     },
     {
-      "epoch": 2.812992125984252,
-      "grad_norm": 9.629202842712402,
+      "epoch": 2.8149606299212597,
+      "grad_norm": 7.840694904327393,
       "learning_rate": 2.5140519729773275e-08,
-      "loss": 0.3077,
+      "loss": 0.2167,
       "step": 2860
     },
     {
-      "epoch": 2.813976377952756,
-      "grad_norm": 10.051465034484863,
+      "epoch": 2.815944881889764,
+      "grad_norm": 7.9547858238220215,
       "learning_rate": 2.4916765340858292e-08,
-      "loss": 0.3801,
+      "loss": 0.248,
       "step": 2861
     },
     {
-      "epoch": 2.8149606299212597,
-      "grad_norm": 7.95008659362793,
+      "epoch": 2.8169291338582676,
+      "grad_norm": 13.104825019836426,
       "learning_rate": 2.4694139392784567e-08,
-      "loss": 0.1731,
+      "loss": 0.7752,
       "step": 2862
     },
     {
-      "epoch": 2.815944881889764,
-      "grad_norm": 11.5671968460083,
+      "epoch": 2.8179133858267718,
+      "grad_norm": 12.998180389404297,
       "learning_rate": 2.4472642412495554e-08,
-      "loss": 0.4511,
+      "loss": 0.4594,
       "step": 2863
     },
     {
-      "epoch": 2.8169291338582676,
-      "grad_norm": 18.472042083740234,
+      "epoch": 2.8188976377952755,
+      "grad_norm": 11.027125358581543,
       "learning_rate": 2.425227492426245e-08,
-      "loss": 1.1221,
+      "loss": 0.6002,
       "step": 2864
     },
     {
-      "epoch": 2.8179133858267718,
-      "grad_norm": 8.366148948669434,
+      "epoch": 2.8198818897637796,
+      "grad_norm": 6.999696254730225,
       "learning_rate": 2.403303744968298e-08,
-      "loss": 0.2557,
+      "loss": 0.1635,
       "step": 2865
     },
     {
-      "epoch": 2.8188976377952755,
-      "grad_norm": 9.588556289672852,
+      "epoch": 2.8208661417322833,
+      "grad_norm": 6.121972560882568,
       "learning_rate": 2.3814930507680188e-08,
-      "loss": 0.377,
+      "loss": 0.2643,
       "step": 2866
     },
     {
-      "epoch": 2.8198818897637796,
-      "grad_norm": 17.30539894104004,
+      "epoch": 2.8218503937007875,
+      "grad_norm": 10.438579559326172,
       "learning_rate": 2.359795461450133e-08,
-      "loss": 0.5314,
+      "loss": 0.4415,
       "step": 2867
     },
     {
-      "epoch": 2.8208661417322833,
-      "grad_norm": 9.442021369934082,
+      "epoch": 2.822834645669291,
+      "grad_norm": 11.241499900817871,
       "learning_rate": 2.3382110283716312e-08,
-      "loss": 0.227,
+      "loss": 0.5654,
       "step": 2868
     },
     {
-      "epoch": 2.8218503937007875,
-      "grad_norm": 18.870290756225586,
+      "epoch": 2.8238188976377954,
+      "grad_norm": 10.601723670959473,
       "learning_rate": 2.3167398026216926e-08,
-      "loss": 0.7812,
+      "loss": 0.3358,
       "step": 2869
     },
     {
-      "epoch": 2.822834645669291,
-      "grad_norm": 9.909849166870117,
+      "epoch": 2.824803149606299,
+      "grad_norm": 9.787238121032715,
       "learning_rate": 2.2953818350215195e-08,
-      "loss": 0.4969,
+      "loss": 0.3752,
       "step": 2870
     },
     {
-      "epoch": 2.8238188976377954,
-      "grad_norm": 9.456997871398926,
+      "epoch": 2.8257874015748032,
+      "grad_norm": 7.7390241622924805,
       "learning_rate": 2.2741371761242526e-08,
-      "loss": 0.2511,
+      "loss": 0.2576,
       "step": 2871
     },
     {
-      "epoch": 2.824803149606299,
-      "grad_norm": 11.477304458618164,
+      "epoch": 2.826771653543307,
+      "grad_norm": 8.766083717346191,
       "learning_rate": 2.253005876214835e-08,
-      "loss": 0.4915,
+      "loss": 0.3116,
       "step": 2872
     },
     {
-      "epoch": 2.8257874015748032,
-      "grad_norm": 12.824424743652344,
+      "epoch": 2.827755905511811,
+      "grad_norm": 7.964391708374023,
       "learning_rate": 2.2319879853098933e-08,
-      "loss": 0.4664,
+      "loss": 0.2287,
       "step": 2873
     },
     {
-      "epoch": 2.826771653543307,
-      "grad_norm": 12.857833862304688,
+      "epoch": 2.828740157480315,
+      "grad_norm": 9.628897666931152,
       "learning_rate": 2.2110835531576244e-08,
-      "loss": 0.3976,
+      "loss": 0.5022,
       "step": 2874
     },
     {
-      "epoch": 2.827755905511811,
-      "grad_norm": 14.371920585632324,
+      "epoch": 2.829724409448819,
+      "grad_norm": 11.806326866149902,
       "learning_rate": 2.1902926292376597e-08,
-      "loss": 0.7253,
+      "loss": 0.7434,
       "step": 2875
     },
     {
-      "epoch": 2.828740157480315,
-      "grad_norm": 11.818841934204102,
+      "epoch": 2.8307086614173227,
+      "grad_norm": 10.130755424499512,
       "learning_rate": 2.16961526276099e-08,
-      "loss": 0.6665,
+      "loss": 0.3239,
       "step": 2876
     },
     {
-      "epoch": 2.829724409448819,
-      "grad_norm": 14.680107116699219,
+      "epoch": 2.831692913385827,
+      "grad_norm": 8.444489479064941,
       "learning_rate": 2.1490515026697924e-08,
-      "loss": 0.8738,
+      "loss": 0.3038,
       "step": 2877
     },
     {
-      "epoch": 2.8307086614173227,
-      "grad_norm": 10.799333572387695,
+      "epoch": 2.8326771653543306,
+      "grad_norm": 10.616978645324707,
       "learning_rate": 2.128601397637366e-08,
-      "loss": 0.3708,
+      "loss": 0.6397,
       "step": 2878
     },
     {
-      "epoch": 2.831692913385827,
-      "grad_norm": 7.399778366088867,
+      "epoch": 2.8336614173228347,
+      "grad_norm": 11.629962921142578,
       "learning_rate": 2.1082649960679856e-08,
-      "loss": 0.2079,
+      "loss": 0.262,
       "step": 2879
     },
     {
-      "epoch": 2.8326771653543306,
-      "grad_norm": 15.631341934204102,
+      "epoch": 2.8346456692913384,
+      "grad_norm": 2.5758774280548096,
       "learning_rate": 2.0880423460967977e-08,
-      "loss": 0.7036,
+      "loss": 0.0492,
       "step": 2880
     },
     {
-      "epoch": 2.8336614173228347,
-      "grad_norm": 10.37580394744873,
+      "epoch": 2.8356299212598426,
+      "grad_norm": 8.437071800231934,
       "learning_rate": 2.0679334955896987e-08,
-      "loss": 0.3334,
+      "loss": 0.3264,
       "step": 2881
     },
     {
-      "epoch": 2.8346456692913384,
-      "grad_norm": 9.839747428894043,
+      "epoch": 2.8366141732283463,
+      "grad_norm": 6.413364410400391,
       "learning_rate": 2.0479384921432415e-08,
-      "loss": 0.2204,
+      "loss": 0.1855,
       "step": 2882
     },
     {
-      "epoch": 2.8356299212598426,
-      "grad_norm": 6.21480131149292,
+      "epoch": 2.8375984251968505,
+      "grad_norm": 4.362641334533691,
       "learning_rate": 2.0280573830844957e-08,
-      "loss": 0.1601,
+      "loss": 0.1059,
       "step": 2883
     },
     {
-      "epoch": 2.8366141732283463,
-      "grad_norm": 8.333520889282227,
+      "epoch": 2.838582677165354,
+      "grad_norm": 14.029305458068848,
       "learning_rate": 2.0082902154709605e-08,
-      "loss": 0.2936,
+      "loss": 1.0493,
       "step": 2884
     },
     {
-      "epoch": 2.8375984251968505,
-      "grad_norm": 12.826237678527832,
+      "epoch": 2.8395669291338583,
+      "grad_norm": 13.094476699829102,
       "learning_rate": 1.9886370360904313e-08,
-      "loss": 0.6482,
+      "loss": 0.5651,
       "step": 2885
     },
     {
-      "epoch": 2.838582677165354,
-      "grad_norm": 6.790856838226318,
+      "epoch": 2.840551181102362,
+      "grad_norm": 8.946066856384277,
       "learning_rate": 1.9690978914609107e-08,
-      "loss": 0.3446,
+      "loss": 0.246,
       "step": 2886
     },
     {
-      "epoch": 2.8395669291338583,
-      "grad_norm": 14.885942459106445,
+      "epoch": 2.841535433070866,
+      "grad_norm": 8.751285552978516,
       "learning_rate": 1.949672827830472e-08,
-      "loss": 0.3579,
+      "loss": 0.3116,
       "step": 2887
     },
     {
-      "epoch": 2.840551181102362,
-      "grad_norm": 15.45215892791748,
+      "epoch": 2.84251968503937,
+      "grad_norm": 13.891740798950195,
       "learning_rate": 1.9303618911771794e-08,
-      "loss": 0.379,
+      "loss": 0.9535,
       "step": 2888
     },
     {
-      "epoch": 2.841535433070866,
-      "grad_norm": 7.229466915130615,
+      "epoch": 2.843503937007874,
+      "grad_norm": 8.203922271728516,
       "learning_rate": 1.911165127208967e-08,
-      "loss": 0.2504,
+      "loss": 0.3018,
       "step": 2889
     },
     {
-      "epoch": 2.84251968503937,
-      "grad_norm": 18.67596435546875,
+      "epoch": 2.844488188976378,
+      "grad_norm": 10.783148765563965,
       "learning_rate": 1.8920825813635075e-08,
-      "loss": 0.8369,
+      "loss": 0.477,
       "step": 2890
     },
     {
-      "epoch": 2.843503937007874,
-      "grad_norm": 10.370317459106445,
+      "epoch": 2.845472440944882,
+      "grad_norm": 11.584412574768066,
       "learning_rate": 1.8731142988081442e-08,
-      "loss": 0.3911,
+      "loss": 0.4697,
       "step": 2891
     },
     {
-      "epoch": 2.844488188976378,
-      "grad_norm": 8.032980918884277,
+      "epoch": 2.846456692913386,
+      "grad_norm": 8.882328033447266,
       "learning_rate": 1.8542603244397694e-08,
-      "loss": 0.3052,
+      "loss": 0.3028,
       "step": 2892
     },
     {
-      "epoch": 2.845472440944882,
-      "grad_norm": 9.2338228225708,
+      "epoch": 2.84744094488189,
+      "grad_norm": 5.329285621643066,
       "learning_rate": 1.835520702884698e-08,
-      "loss": 0.2668,
+      "loss": 0.1494,
       "step": 2893
     },
     {
-      "epoch": 2.846456692913386,
-      "grad_norm": 5.767791271209717,
+      "epoch": 2.8484251968503935,
+      "grad_norm": 9.194082260131836,
       "learning_rate": 1.8168954784985783e-08,
-      "loss": 0.185,
+      "loss": 0.3779,
       "step": 2894
     },
     {
-      "epoch": 2.84744094488189,
-      "grad_norm": 3.9710404872894287,
+      "epoch": 2.8494094488188977,
+      "grad_norm": 10.444098472595215,
       "learning_rate": 1.7983846953663087e-08,
-      "loss": 0.0768,
+      "loss": 0.436,
       "step": 2895
     },
     {
-      "epoch": 2.8484251968503935,
-      "grad_norm": 10.550639152526855,
+      "epoch": 2.850393700787402,
+      "grad_norm": 10.215253829956055,
       "learning_rate": 1.7799883973018788e-08,
-      "loss": 0.3886,
+      "loss": 0.4104,
       "step": 2896
     },
     {
-      "epoch": 2.8494094488188977,
-      "grad_norm": 19.47537612915039,
+      "epoch": 2.8513779527559056,
+      "grad_norm": 8.82308292388916,
       "learning_rate": 1.7617066278483343e-08,
-      "loss": 0.7719,
+      "loss": 0.3295,
       "step": 2897
     },
     {
-      "epoch": 2.850393700787402,
-      "grad_norm": 14.099431037902832,
+      "epoch": 2.8523622047244093,
+      "grad_norm": 5.599417209625244,
       "learning_rate": 1.7435394302776123e-08,
-      "loss": 0.6041,
+      "loss": 0.128,
       "step": 2898
     },
     {
-      "epoch": 2.8513779527559056,
-      "grad_norm": 6.4440107345581055,
+      "epoch": 2.8533464566929134,
+      "grad_norm": 13.941229820251465,
       "learning_rate": 1.725486847590469e-08,
-      "loss": 0.2142,
+      "loss": 0.9193,
       "step": 2899
     },
     {
-      "epoch": 2.8523622047244093,
-      "grad_norm": 11.775022506713867,
+      "epoch": 2.8543307086614176,
+      "grad_norm": 6.804110527038574,
       "learning_rate": 1.7075489225163737e-08,
-      "loss": 0.5572,
+      "loss": 0.2787,
       "step": 2900
     },
     {
-      "epoch": 2.8533464566929134,
-      "grad_norm": 16.148359298706055,
+      "epoch": 2.8553149606299213,
+      "grad_norm": 8.19189167022705,
       "learning_rate": 1.689725697513417e-08,
-      "loss": 0.9624,
+      "loss": 0.2746,
       "step": 2901
     },
     {
-      "epoch": 2.8543307086614176,
-      "grad_norm": 10.461920738220215,
+      "epoch": 2.856299212598425,
+      "grad_norm": 13.617401123046875,
       "learning_rate": 1.6720172147681977e-08,
-      "loss": 0.4482,
+      "loss": 0.5146,
       "step": 2902
     },
     {
-      "epoch": 2.8553149606299213,
-      "grad_norm": 11.626091957092285,
+      "epoch": 2.857283464566929,
+      "grad_norm": 13.147882461547852,
       "learning_rate": 1.654423516195707e-08,
-      "loss": 0.4551,
+      "loss": 0.6607,
       "step": 2903
     },
     {
-      "epoch": 2.856299212598425,
-      "grad_norm": 11.768128395080566,
+      "epoch": 2.8582677165354333,
+      "grad_norm": 9.217270851135254,
       "learning_rate": 1.636944643439282e-08,
-      "loss": 0.5805,
+      "loss": 0.3297,
       "step": 2904
     },
     {
-      "epoch": 2.857283464566929,
-      "grad_norm": 9.375350952148438,
+      "epoch": 2.859251968503937,
+      "grad_norm": 8.807034492492676,
       "learning_rate": 1.619580637870439e-08,
-      "loss": 0.3688,
+      "loss": 0.2913,
       "step": 2905
     },
     {
-      "epoch": 2.8582677165354333,
-      "grad_norm": 7.829890727996826,
+      "epoch": 2.8602362204724407,
+      "grad_norm": 10.446438789367676,
       "learning_rate": 1.602331540588836e-08,
-      "loss": 0.1731,
+      "loss": 0.4855,
       "step": 2906
     },
     {
-      "epoch": 2.859251968503937,
-      "grad_norm": 17.9921932220459,
+      "epoch": 2.861220472440945,
+      "grad_norm": 7.405145168304443,
       "learning_rate": 1.5851973924221346e-08,
-      "loss": 0.5437,
+      "loss": 0.3315,
       "step": 2907
     },
     {
-      "epoch": 2.859251968503937,
-      "eval_Qnli-dev_cosine_accuracy": 0.6796875,
-      "eval_Qnli-dev_cosine_accuracy_threshold": 0.7960593700408936,
-      "eval_Qnli-dev_cosine_ap": 0.7039872572986636,
-      "eval_Qnli-dev_cosine_f1": 0.6924315619967794,
-      "eval_Qnli-dev_cosine_f1_threshold": 0.7081151008605957,
-      "eval_Qnli-dev_cosine_precision": 0.5584415584415584,
-      "eval_Qnli-dev_cosine_recall": 0.9110169491525424,
-      "eval_Qnli-dev_dot_accuracy": 0.677734375,
-      "eval_Qnli-dev_dot_accuracy_threshold": 606.1690063476562,
-      "eval_Qnli-dev_dot_ap": 0.7003540936136716,
-      "eval_Qnli-dev_dot_f1": 0.6939443535188216,
-      "eval_Qnli-dev_dot_f1_threshold": 547.9747314453125,
-      "eval_Qnli-dev_dot_precision": 0.5653333333333334,
-      "eval_Qnli-dev_dot_recall": 0.8983050847457628,
-      "eval_Qnli-dev_euclidean_accuracy": 0.6796875,
-      "eval_Qnli-dev_euclidean_accuracy_threshold": 17.620656967163086,
-      "eval_Qnli-dev_euclidean_ap": 0.7045153090257912,
-      "eval_Qnli-dev_euclidean_f1": 0.6935483870967742,
-      "eval_Qnli-dev_euclidean_f1_threshold": 20.996856689453125,
-      "eval_Qnli-dev_euclidean_precision": 0.5598958333333334,
-      "eval_Qnli-dev_euclidean_recall": 0.9110169491525424,
-      "eval_Qnli-dev_manhattan_accuracy": 0.697265625,
-      "eval_Qnli-dev_manhattan_accuracy_threshold": 376.2118835449219,
-      "eval_Qnli-dev_manhattan_ap": 0.7133804101994428,
-      "eval_Qnli-dev_manhattan_f1": 0.6949429037520392,
-      "eval_Qnli-dev_manhattan_f1_threshold": 439.9166564941406,
-      "eval_Qnli-dev_manhattan_precision": 0.5649867374005305,
-      "eval_Qnli-dev_manhattan_recall": 0.902542372881356,
-      "eval_Qnli-dev_max_accuracy": 0.697265625,
-      "eval_Qnli-dev_max_accuracy_threshold": 606.1690063476562,
-      "eval_Qnli-dev_max_ap": 0.7133804101994428,
-      "eval_Qnli-dev_max_f1": 0.6949429037520392,
-      "eval_Qnli-dev_max_f1_threshold": 547.9747314453125,
-      "eval_Qnli-dev_max_precision": 0.5653333333333334,
-      "eval_Qnli-dev_max_recall": 0.9110169491525424,
+      "epoch": 2.861220472440945,
+      "eval_Qnli-dev_cosine_accuracy": 0.67578125,
+      "eval_Qnli-dev_cosine_accuracy_threshold": 0.7804532051086426,
+      "eval_Qnli-dev_cosine_ap": 0.7024806206418001,
+      "eval_Qnli-dev_cosine_f1": 0.6888111888111887,
+      "eval_Qnli-dev_cosine_f1_threshold": 0.7391257286071777,
+      "eval_Qnli-dev_cosine_precision": 0.5863095238095238,
+      "eval_Qnli-dev_cosine_recall": 0.8347457627118644,
+      "eval_Qnli-dev_dot_accuracy": 0.671875,
+      "eval_Qnli-dev_dot_accuracy_threshold": 589.8578491210938,
+      "eval_Qnli-dev_dot_ap": 0.6993628011203765,
+      "eval_Qnli-dev_dot_f1": 0.6887417218543046,
+      "eval_Qnli-dev_dot_f1_threshold": 549.0986328125,
+      "eval_Qnli-dev_dot_precision": 0.5652173913043478,
+      "eval_Qnli-dev_dot_recall": 0.8813559322033898,
+      "eval_Qnli-dev_euclidean_accuracy": 0.677734375,
+      "eval_Qnli-dev_euclidean_accuracy_threshold": 18.17205238342285,
+      "eval_Qnli-dev_euclidean_ap": 0.7030782182119286,
+      "eval_Qnli-dev_euclidean_f1": 0.6893039049235994,
+      "eval_Qnli-dev_euclidean_f1_threshold": 20.12609100341797,
+      "eval_Qnli-dev_euclidean_precision": 0.5750708215297451,
+      "eval_Qnli-dev_euclidean_recall": 0.8601694915254238,
+      "eval_Qnli-dev_manhattan_accuracy": 0.689453125,
+      "eval_Qnli-dev_manhattan_accuracy_threshold": 383.89593505859375,
+      "eval_Qnli-dev_manhattan_ap": 0.7092608588336894,
+      "eval_Qnli-dev_manhattan_f1": 0.6876090750436299,
+      "eval_Qnli-dev_manhattan_f1_threshold": 419.38165283203125,
+      "eval_Qnli-dev_manhattan_precision": 0.5845697329376854,
+      "eval_Qnli-dev_manhattan_recall": 0.8347457627118644,
+      "eval_Qnli-dev_max_accuracy": 0.689453125,
+      "eval_Qnli-dev_max_accuracy_threshold": 589.8578491210938,
+      "eval_Qnli-dev_max_ap": 0.7092608588336894,
+      "eval_Qnli-dev_max_f1": 0.6893039049235994,
+      "eval_Qnli-dev_max_f1_threshold": 549.0986328125,
+      "eval_Qnli-dev_max_precision": 0.5863095238095238,
+      "eval_Qnli-dev_max_recall": 0.8813559322033898,
       "eval_allNLI-dev_cosine_accuracy": 0.6953125,
-      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9145020246505737,
-      "eval_allNLI-dev_cosine_ap": 0.5212810979375195,
-      "eval_allNLI-dev_cosine_f1": 0.560364464692483,
-      "eval_allNLI-dev_cosine_f1_threshold": 0.8175112009048462,
-      "eval_allNLI-dev_cosine_precision": 0.462406015037594,
-      "eval_allNLI-dev_cosine_recall": 0.7109826589595376,
+      "eval_allNLI-dev_cosine_accuracy_threshold": 0.9157084226608276,
+      "eval_allNLI-dev_cosine_ap": 0.5141008524462632,
+      "eval_allNLI-dev_cosine_f1": 0.5553145336225597,
+      "eval_allNLI-dev_cosine_f1_threshold": 0.8242306113243103,
+      "eval_allNLI-dev_cosine_precision": 0.4444444444444444,
+      "eval_allNLI-dev_cosine_recall": 0.7398843930635838,
       "eval_allNLI-dev_dot_accuracy": 0.693359375,
-      "eval_allNLI-dev_dot_accuracy_threshold": 700.8072509765625,
-      "eval_allNLI-dev_dot_ap": 0.5186569295484549,
-      "eval_allNLI-dev_dot_f1": 0.5540540540540541,
-      "eval_allNLI-dev_dot_f1_threshold": 623.1869506835938,
-      "eval_allNLI-dev_dot_precision": 0.45387453874538747,
-      "eval_allNLI-dev_dot_recall": 0.7109826589595376,
+      "eval_allNLI-dev_dot_accuracy_threshold": 701.5110473632812,
+      "eval_allNLI-dev_dot_ap": 0.5138197724822148,
+      "eval_allNLI-dev_dot_f1": 0.5533769063180829,
+      "eval_allNLI-dev_dot_f1_threshold": 629.531982421875,
+      "eval_allNLI-dev_dot_precision": 0.44405594405594406,
+      "eval_allNLI-dev_dot_recall": 0.7341040462427746,
       "eval_allNLI-dev_euclidean_accuracy": 0.6953125,
-      "eval_allNLI-dev_euclidean_accuracy_threshold": 11.445773124694824,
-      "eval_allNLI-dev_euclidean_ap": 0.5209866197717024,
-      "eval_allNLI-dev_euclidean_f1": 0.560364464692483,
-      "eval_allNLI-dev_euclidean_f1_threshold": 16.688749313354492,
-      "eval_allNLI-dev_euclidean_precision": 0.462406015037594,
-      "eval_allNLI-dev_euclidean_recall": 0.7109826589595376,
+      "eval_allNLI-dev_euclidean_accuracy_threshold": 11.348623275756836,
+      "eval_allNLI-dev_euclidean_ap": 0.5140792717603242,
+      "eval_allNLI-dev_euclidean_f1": 0.5541125541125541,
+      "eval_allNLI-dev_euclidean_f1_threshold": 16.391315460205078,
+      "eval_allNLI-dev_euclidean_precision": 0.4429065743944637,
+      "eval_allNLI-dev_euclidean_recall": 0.7398843930635838,
       "eval_allNLI-dev_manhattan_accuracy": 0.69140625,
-      "eval_allNLI-dev_manhattan_accuracy_threshold": 245.1360626220703,
-      "eval_allNLI-dev_manhattan_ap": 0.5193808341055445,
-      "eval_allNLI-dev_manhattan_f1": 0.5584415584415584,
-      "eval_allNLI-dev_manhattan_f1_threshold": 360.3310546875,
-      "eval_allNLI-dev_manhattan_precision": 0.4463667820069204,
-      "eval_allNLI-dev_manhattan_recall": 0.7456647398843931,
+      "eval_allNLI-dev_manhattan_accuracy_threshold": 229.85313415527344,
+      "eval_allNLI-dev_manhattan_ap": 0.5124222610401596,
+      "eval_allNLI-dev_manhattan_f1": 0.5631929046563192,
+      "eval_allNLI-dev_manhattan_f1_threshold": 344.6240539550781,
+      "eval_allNLI-dev_manhattan_precision": 0.4568345323741007,
+      "eval_allNLI-dev_manhattan_recall": 0.7341040462427746,
       "eval_allNLI-dev_max_accuracy": 0.6953125,
-      "eval_allNLI-dev_max_accuracy_threshold": 700.8072509765625,
-      "eval_allNLI-dev_max_ap": 0.5212810979375195,
-      "eval_allNLI-dev_max_f1": 0.560364464692483,
-      "eval_allNLI-dev_max_f1_threshold": 623.1869506835938,
-      "eval_allNLI-dev_max_precision": 0.462406015037594,
-      "eval_allNLI-dev_max_recall": 0.7456647398843931,
-      "eval_loss": 0.887163519859314,
-      "eval_runtime": 54.0962,
-      "eval_samples_per_second": 30.76,
-      "eval_sequential_score": 0.7133804101994428,
-      "eval_steps_per_second": 0.129,
-      "eval_sts-test_pearson_cosine": 0.6549476499152822,
-      "eval_sts-test_pearson_dot": 0.654094600699473,
-      "eval_sts-test_pearson_euclidean": 0.6669531190153849,
-      "eval_sts-test_pearson_manhattan": 0.6703075464471974,
-      "eval_sts-test_pearson_max": 0.6703075464471974,
-      "eval_sts-test_spearman_cosine": 0.6617045917839,
-      "eval_sts-test_spearman_dot": 0.6611669948808051,
-      "eval_sts-test_spearman_euclidean": 0.6619002890499646,
-      "eval_sts-test_spearman_manhattan": 0.664562675843923,
-      "eval_sts-test_spearman_max": 0.664562675843923,
+      "eval_allNLI-dev_max_accuracy_threshold": 701.5110473632812,
+      "eval_allNLI-dev_max_ap": 0.5141008524462632,
+      "eval_allNLI-dev_max_f1": 0.5631929046563192,
+      "eval_allNLI-dev_max_f1_threshold": 629.531982421875,
+      "eval_allNLI-dev_max_precision": 0.4568345323741007,
+      "eval_allNLI-dev_max_recall": 0.7398843930635838,
+      "eval_loss": 0.7887536883354187,
+      "eval_runtime": 52.7503,
+      "eval_samples_per_second": 31.545,
+      "eval_sequential_score": 0.7092608588336894,
+      "eval_steps_per_second": 0.133,
+      "eval_sts-test_pearson_cosine": 0.6709134867652243,
+      "eval_sts-test_pearson_dot": 0.6691751223113676,
+      "eval_sts-test_pearson_euclidean": 0.682301685365551,
+      "eval_sts-test_pearson_manhattan": 0.6857895258959846,
+      "eval_sts-test_pearson_max": 0.6857895258959846,
+      "eval_sts-test_spearman_cosine": 0.6758937399794869,
+      "eval_sts-test_spearman_dot": 0.672994753266552,
+      "eval_sts-test_spearman_euclidean": 0.6764461943013637,
+      "eval_sts-test_spearman_manhattan": 0.6790772999636097,
+      "eval_sts-test_spearman_max": 0.6790772999636097,
       "step": 2907
     },
     {
-      "epoch": 2.8602362204724407,
-      "grad_norm": 12.062636375427246,
+      "epoch": 2.862204724409449,
+      "grad_norm": 8.543414115905762,
       "learning_rate": 1.5681782339259265e-08,
-      "loss": 0.555,
+      "loss": 0.3065,
       "step": 2908
     },
     {
-      "epoch": 2.861220472440945,
-      "grad_norm": 12.004955291748047,
+      "epoch": 2.8631889763779528,
+      "grad_norm": 6.582826137542725,
       "learning_rate": 1.5512741053836244e-08,
-      "loss": 0.6113,
+      "loss": 0.1963,
       "step": 2909
     },
     {
-      "epoch": 2.862204724409449,
-      "grad_norm": 7.830421447753906,
+      "epoch": 2.8641732283464565,
+      "grad_norm": 7.284085750579834,
       "learning_rate": 1.5344850468063668e-08,
-      "loss": 0.3816,
+      "loss": 0.1822,
       "step": 2910
     },
     {
-      "epoch": 2.8631889763779528,
-      "grad_norm": 6.853997707366943,
+      "epoch": 2.8651574803149606,
+      "grad_norm": 10.267645835876465,
       "learning_rate": 1.5178110979329423e-08,
-      "loss": 0.1588,
+      "loss": 0.5729,
       "step": 2911
     },
     {
-      "epoch": 2.8641732283464565,
-      "grad_norm": 7.042568683624268,
+      "epoch": 2.866141732283465,
+      "grad_norm": 9.939976692199707,
       "learning_rate": 1.5012522982296648e-08,
-      "loss": 0.1333,
+      "loss": 0.4219,
       "step": 2912
     },
     {
-      "epoch": 2.8651574803149606,
-      "grad_norm": 8.45995044708252,
+      "epoch": 2.8671259842519685,
+      "grad_norm": 7.504786491394043,
       "learning_rate": 1.4848086868903112e-08,
-      "loss": 0.2772,
+      "loss": 0.2837,
       "step": 2913
     },
     {
-      "epoch": 2.866141732283465,
-      "grad_norm": 8.339113235473633,
+      "epoch": 2.8681102362204722,
+      "grad_norm": 5.3694024085998535,
       "learning_rate": 1.4684803028360009e-08,
-      "loss": 0.3424,
+      "loss": 0.1562,
       "step": 2914
     },
     {
-      "epoch": 2.8671259842519685,
-      "grad_norm": 17.337478637695312,
+      "epoch": 2.8690944881889764,
+      "grad_norm": 9.125850677490234,
       "learning_rate": 1.4522671847151226e-08,
-      "loss": 0.7045,
+      "loss": 0.4343,
       "step": 2915
     },
     {
-      "epoch": 2.8681102362204722,
-      "grad_norm": 11.440997123718262,
+      "epoch": 2.8700787401574805,
+      "grad_norm": 7.746906757354736,
       "learning_rate": 1.436169370903241e-08,
-      "loss": 0.4514,
+      "loss": 0.287,
       "step": 2916
     },
     {
-      "epoch": 2.8690944881889764,
-      "grad_norm": 11.359492301940918,
+      "epoch": 2.8710629921259843,
+      "grad_norm": 9.241104125976562,
       "learning_rate": 1.420186899502992e-08,
-      "loss": 0.2715,
+      "loss": 0.2159,
       "step": 2917
     },
     {
-      "epoch": 2.8700787401574805,
-      "grad_norm": 6.252547264099121,
+      "epoch": 2.872047244094488,
+      "grad_norm": 10.579710960388184,
       "learning_rate": 1.4043198083440207e-08,
-      "loss": 0.1755,
+      "loss": 0.421,
       "step": 2918
     },
     {
-      "epoch": 2.8710629921259843,
-      "grad_norm": 13.951661109924316,
+      "epoch": 2.873031496062992,
+      "grad_norm": 11.039432525634766,
       "learning_rate": 1.3885681349828442e-08,
-      "loss": 0.5026,
+      "loss": 0.4408,
       "step": 2919
     },
     {
-      "epoch": 2.872047244094488,
-      "grad_norm": 17.206926345825195,
+      "epoch": 2.8740157480314963,
+      "grad_norm": 10.788917541503906,
       "learning_rate": 1.3729319167028123e-08,
-      "loss": 0.6616,
+      "loss": 0.4993,
       "step": 2920
     },
     {
-      "epoch": 2.873031496062992,
-      "grad_norm": 10.049877166748047,
+      "epoch": 2.875,
+      "grad_norm": 10.828028678894043,
       "learning_rate": 1.3574111905140026e-08,
-      "loss": 0.4262,
+      "loss": 0.5581,
       "step": 2921
     },
     {
-      "epoch": 2.8740157480314963,
-      "grad_norm": 8.150603294372559,
+      "epoch": 2.8759842519685037,
+      "grad_norm": 11.519461631774902,
       "learning_rate": 1.3420059931531155e-08,
-      "loss": 0.2671,
+      "loss": 0.4127,
       "step": 2922
     },
     {
-      "epoch": 2.875,
-      "grad_norm": 11.366801261901855,
+      "epoch": 2.876968503937008,
+      "grad_norm": 8.185306549072266,
       "learning_rate": 1.3267163610834127e-08,
-      "loss": 0.471,
+      "loss": 0.2388,
       "step": 2923
     },
     {
-      "epoch": 2.8759842519685037,
-      "grad_norm": 10.024985313415527,
+      "epoch": 2.877952755905512,
+      "grad_norm": 11.209269523620605,
       "learning_rate": 1.3115423304946073e-08,
-      "loss": 0.3942,
+      "loss": 0.6897,
       "step": 2924
     },
     {
-      "epoch": 2.876968503937008,
-      "grad_norm": 9.747432708740234,
+      "epoch": 2.8789370078740157,
+      "grad_norm": 13.304403305053711,
       "learning_rate": 1.2964839373027969e-08,
-      "loss": 0.3706,
+      "loss": 0.4772,
       "step": 2925
     },
     {
-      "epoch": 2.877952755905512,
-      "grad_norm": 9.726685523986816,
+      "epoch": 2.8799212598425195,
+      "grad_norm": 5.14035177230835,
       "learning_rate": 1.2815412171503812e-08,
-      "loss": 0.4969,
+      "loss": 0.14,
       "step": 2926
     },
     {
-      "epoch": 2.8789370078740157,
-      "grad_norm": 12.810749053955078,
+      "epoch": 2.8809055118110236,
+      "grad_norm": 13.388592720031738,
       "learning_rate": 1.2667142054059505e-08,
-      "loss": 0.7252,
+      "loss": 0.5986,
       "step": 2927
     },
     {
-      "epoch": 2.8799212598425195,
-      "grad_norm": 6.0586090087890625,
+      "epoch": 2.8818897637795278,
+      "grad_norm": 8.109567642211914,
       "learning_rate": 1.2520029371642255e-08,
-      "loss": 0.1062,
+      "loss": 0.3224,
       "step": 2928
     },
     {
-      "epoch": 2.8809055118110236,
-      "grad_norm": 9.592218399047852,
+      "epoch": 2.8828740157480315,
+      "grad_norm": 10.926013946533203,
       "learning_rate": 1.2374074472459742e-08,
-      "loss": 0.3599,
+      "loss": 0.4514,
       "step": 2929
     },
     {
-      "epoch": 2.8818897637795278,
-      "grad_norm": 9.726683616638184,
+      "epoch": 2.883858267716535,
+      "grad_norm": 11.717981338500977,
       "learning_rate": 1.2229277701979232e-08,
-      "loss": 0.2928,
+      "loss": 0.4632,
       "step": 2930
     },
     {
-      "epoch": 2.8828740157480315,
-      "grad_norm": 12.128030776977539,
+      "epoch": 2.8848425196850394,
+      "grad_norm": 13.883950233459473,
       "learning_rate": 1.2085639402926691e-08,
-      "loss": 0.4123,
+      "loss": 0.7036,
       "step": 2931
     },
     {
-      "epoch": 2.883858267716535,
-      "grad_norm": 3.233788251876831,
+      "epoch": 2.8858267716535435,
+      "grad_norm": 8.726773262023926,
       "learning_rate": 1.1943159915286018e-08,
-      "loss": 0.0589,
+      "loss": 0.3824,
       "step": 2932
     },
     {
-      "epoch": 2.8848425196850394,
-      "grad_norm": 16.795368194580078,
+      "epoch": 2.8868110236220472,
+      "grad_norm": 9.846919059753418,
       "learning_rate": 1.1801839576298478e-08,
-      "loss": 0.6296,
+      "loss": 0.335,
       "step": 2933
     },
     {
-      "epoch": 2.8858267716535435,
-      "grad_norm": 12.347762107849121,
+      "epoch": 2.887795275590551,
+      "grad_norm": 7.308245658874512,
       "learning_rate": 1.1661678720461443e-08,
-      "loss": 0.5993,
+      "loss": 0.2227,
       "step": 2934
     },
     {
-      "epoch": 2.8868110236220472,
-      "grad_norm": 9.521347999572754,
+      "epoch": 2.888779527559055,
+      "grad_norm": 8.39908218383789,
       "learning_rate": 1.1522677679527994e-08,
-      "loss": 0.3544,
+      "loss": 0.3978,
       "step": 2935
     },
     {
-      "epoch": 2.887795275590551,
-      "grad_norm": 14.060378074645996,
+      "epoch": 2.8897637795275593,
+      "grad_norm": 13.679941177368164,
       "learning_rate": 1.138483678250605e-08,
-      "loss": 0.5743,
+      "loss": 0.6791,
       "step": 2936
     },
     {
-      "epoch": 2.888779527559055,
-      "grad_norm": 7.655622959136963,
+      "epoch": 2.890748031496063,
+      "grad_norm": 12.295086860656738,
       "learning_rate": 1.1248156355657297e-08,
-      "loss": 0.2369,
+      "loss": 0.6434,
       "step": 2937
     },
     {
-      "epoch": 2.8897637795275593,
-      "grad_norm": 13.059883117675781,
+      "epoch": 2.8917322834645667,
+      "grad_norm": 7.051697731018066,
       "learning_rate": 1.1112636722496874e-08,
-      "loss": 0.8943,
+      "loss": 0.3197,
       "step": 2938
     },
     {
-      "epoch": 2.890748031496063,
-      "grad_norm": 9.250969886779785,
+      "epoch": 2.892716535433071,
+      "grad_norm": 8.776548385620117,
       "learning_rate": 1.0978278203792367e-08,
-      "loss": 0.3377,
+      "loss": 0.2692,
       "step": 2939
     },
     {
-      "epoch": 2.8917322834645667,
-      "grad_norm": 12.578452110290527,
+      "epoch": 2.893700787401575,
+      "grad_norm": 8.831172943115234,
       "learning_rate": 1.0845081117562924e-08,
-      "loss": 0.4964,
+      "loss": 0.4412,
       "step": 2940
     },
     {
-      "epoch": 2.892716535433071,
-      "grad_norm": 8.722951889038086,
+      "epoch": 2.8946850393700787,
+      "grad_norm": 10.295771598815918,
       "learning_rate": 1.0713045779078767e-08,
-      "loss": 0.3776,
+      "loss": 0.4292,
       "step": 2941
     },
     {
-      "epoch": 2.893700787401575,
-      "grad_norm": 9.845478057861328,
+      "epoch": 2.8956692913385824,
+      "grad_norm": 12.517659187316895,
       "learning_rate": 1.0582172500860295e-08,
-      "loss": 0.3713,
+      "loss": 0.4835,
       "step": 2942
     },
     {
-      "epoch": 2.8946850393700787,
-      "grad_norm": 12.618115425109863,
+      "epoch": 2.8966535433070866,
+      "grad_norm": 10.806954383850098,
       "learning_rate": 1.0452461592677421e-08,
-      "loss": 0.4148,
+      "loss": 0.3929,
       "step": 2943
     },
     {
-      "epoch": 2.8956692913385824,
-      "grad_norm": 6.505346775054932,
+      "epoch": 2.8976377952755907,
+      "grad_norm": 7.610433101654053,
       "learning_rate": 1.0323913361548642e-08,
-      "loss": 0.1323,
+      "loss": 0.2987,
       "step": 2944
     },
     {
-      "epoch": 2.8966535433070866,
-      "grad_norm": 9.50572395324707,
+      "epoch": 2.8986220472440944,
+      "grad_norm": 10.908812522888184,
       "learning_rate": 1.0196528111740585e-08,
-      "loss": 0.3221,
+      "loss": 0.4143,
       "step": 2945
     },
     {
-      "epoch": 2.8976377952755907,
-      "grad_norm": 10.917316436767578,
+      "epoch": 2.899606299212598,
+      "grad_norm": 6.618739604949951,
       "learning_rate": 1.007030614476724e-08,
-      "loss": 0.5279,
+      "loss": 0.1872,
       "step": 2946
     },
     {
-      "epoch": 2.8986220472440944,
-      "grad_norm": 7.913192272186279,
+      "epoch": 2.9005905511811023,
+      "grad_norm": 13.165253639221191,
       "learning_rate": 9.945247759388956e-09,
-      "loss": 0.3671,
+      "loss": 0.6787,
       "step": 2947
     },
     {
-      "epoch": 2.899606299212598,
-      "grad_norm": 13.66878604888916,
+      "epoch": 2.9015748031496065,
+      "grad_norm": 10.658306121826172,
       "learning_rate": 9.821353251612172e-09,
-      "loss": 0.5808,
+      "loss": 0.3208,
       "step": 2948
     },
     {
-      "epoch": 2.9005905511811023,
-      "grad_norm": 10.809479713439941,
+      "epoch": 2.90255905511811,
+      "grad_norm": 7.823882102966309,
       "learning_rate": 9.69862291468831e-09,
-      "loss": 0.2977,
+      "loss": 0.2856,
       "step": 2949
     },
     {
-      "epoch": 2.9015748031496065,
-      "grad_norm": 9.589229583740234,
+      "epoch": 2.903543307086614,
+      "grad_norm": 5.899032115936279,
       "learning_rate": 9.57705703911333e-09,
-      "loss": 0.3059,
+      "loss": 0.412,
       "step": 2950
     },
     {
-      "epoch": 2.90255905511811,
-      "grad_norm": 5.2587571144104,
+      "epoch": 2.904527559055118,
+      "grad_norm": 9.451079368591309,
       "learning_rate": 9.456655912627006e-09,
-      "loss": 0.113,
+      "loss": 0.4724,
       "step": 2951
     },
     {
-      "epoch": 2.903543307086614,
-      "grad_norm": 3.4376654624938965,
+      "epoch": 2.905511811023622,
+      "grad_norm": 13.56018352508545,
       "learning_rate": 9.337419820212217e-09,
-      "loss": 0.2043,
+      "loss": 0.7545,
       "step": 2952
     },
     {
-      "epoch": 2.904527559055118,
-      "grad_norm": 10.453415870666504,
+      "epoch": 2.906496062992126,
+      "grad_norm": 7.16195821762085,
       "learning_rate": 9.219349044094277e-09,
-      "loss": 0.4614,
+      "loss": 0.2691,
       "step": 2953
     },
     {
-      "epoch": 2.905511811023622,
-      "grad_norm": 9.700399398803711,
+      "epoch": 2.90748031496063,
+      "grad_norm": 9.492947578430176,
       "learning_rate": 9.102443863740158e-09,
-      "loss": 0.3697,
+      "loss": 0.3376,
       "step": 2954
     },
     {
-      "epoch": 2.906496062992126,
-      "grad_norm": 7.6696295738220215,
+      "epoch": 2.908464566929134,
+      "grad_norm": 11.326608657836914,
       "learning_rate": 8.986704555857998e-09,
-      "loss": 0.2648,
+      "loss": 0.6738,
       "step": 2955
     },
     {
-      "epoch": 2.90748031496063,
-      "grad_norm": 8.789196968078613,
+      "epoch": 2.909448818897638,
+      "grad_norm": 13.294182777404785,
       "learning_rate": 8.872131394396434e-09,
-      "loss": 0.3086,
+      "loss": 0.4907,
       "step": 2956
     },
     {
-      "epoch": 2.908464566929134,
-      "grad_norm": 5.326946258544922,
+      "epoch": 2.9104330708661417,
+      "grad_norm": 2.137803792953491,
       "learning_rate": 8.75872465054383e-09,
-      "loss": 0.1898,
+      "loss": 0.0425,
       "step": 2957
     },
     {
-      "epoch": 2.909448818897638,
-      "grad_norm": 9.471596717834473,
+      "epoch": 2.911417322834646,
+      "grad_norm": 7.782170295715332,
       "learning_rate": 8.646484592727771e-09,
-      "loss": 0.2795,
+      "loss": 0.2025,
       "step": 2958
     },
     {
-      "epoch": 2.9104330708661417,
-      "grad_norm": 3.430964231491089,
+      "epoch": 2.9124015748031495,
+      "grad_norm": 12.268577575683594,
       "learning_rate": 8.535411486614188e-09,
-      "loss": 0.0734,
+      "loss": 0.5872,
       "step": 2959
     },
     {
-      "epoch": 2.911417322834646,
-      "grad_norm": 12.03944206237793,
+      "epoch": 2.9133858267716537,
+      "grad_norm": 7.75424861907959,
       "learning_rate": 8.425505595107189e-09,
-      "loss": 0.7132,
+      "loss": 0.2862,
       "step": 2960
     },
     {
-      "epoch": 2.9124015748031495,
-      "grad_norm": 12.02408504486084,
+      "epoch": 2.9143700787401574,
+      "grad_norm": 4.5719499588012695,
       "learning_rate": 8.316767178347839e-09,
-      "loss": 0.5367,
+      "loss": 0.1192,
       "step": 2961
     },
     {
-      "epoch": 2.9133858267716537,
-      "grad_norm": 9.92115592956543,
+      "epoch": 2.9153543307086616,
+      "grad_norm": 16.893280029296875,
       "learning_rate": 8.209196493714103e-09,
-      "loss": 0.3531,
+      "loss": 0.7957,
       "step": 2962
     },
     {
-      "epoch": 2.9143700787401574,
-      "grad_norm": 9.805298805236816,
+      "epoch": 2.9163385826771653,
+      "grad_norm": 12.421287536621094,
       "learning_rate": 8.102793795819803e-09,
-      "loss": 0.3368,
+      "loss": 0.5745,
       "step": 2963
     },
     {
-      "epoch": 2.9153543307086616,
-      "grad_norm": 10.747846603393555,
+      "epoch": 2.9173228346456694,
+      "grad_norm": 10.711176872253418,
       "learning_rate": 7.997559336514277e-09,
-      "loss": 0.5955,
+      "loss": 0.3342,
       "step": 2964
     },
     {
-      "epoch": 2.9163385826771653,
-      "grad_norm": 7.148336887359619,
+      "epoch": 2.918307086614173,
+      "grad_norm": 14.741097450256348,
       "learning_rate": 7.893493364881835e-09,
-      "loss": 0.2534,
+      "loss": 0.5756,
       "step": 2965
     },
     {
-      "epoch": 2.9173228346456694,
-      "grad_norm": 13.780056953430176,
+      "epoch": 2.9192913385826773,
+      "grad_norm": 9.382786750793457,
       "learning_rate": 7.790596127240808e-09,
-      "loss": 0.6842,
+      "loss": 0.3301,
       "step": 2966
     },
     {
-      "epoch": 2.918307086614173,
-      "grad_norm": 5.862837314605713,
+      "epoch": 2.920275590551181,
+      "grad_norm": 12.23287582397461,
       "learning_rate": 7.688867867143388e-09,
-      "loss": 0.1778,
+      "loss": 0.5396,
       "step": 2967
     },
     {
-      "epoch": 2.9192913385826773,
-      "grad_norm": 8.07844352722168,
+      "epoch": 2.921259842519685,
+      "grad_norm": 7.210506916046143,
       "learning_rate": 7.58830882537469e-09,
-      "loss": 0.2702,
+      "loss": 0.248,
       "step": 2968
     },
     {
-      "epoch": 2.920275590551181,
-      "grad_norm": 5.911746978759766,
+      "epoch": 2.922244094488189,
+      "grad_norm": 7.3606743812561035,
       "learning_rate": 7.488919239952469e-09,
-      "loss": 0.1802,
+      "loss": 0.2135,
       "step": 2969
     },
     {
-      "epoch": 2.921259842519685,
-      "grad_norm": 8.986471176147461,
+      "epoch": 2.923228346456693,
+      "grad_norm": 9.690980911254883,
       "learning_rate": 7.3906993461264595e-09,
-      "loss": 0.3735,
+      "loss": 0.2297,
       "step": 2970
     },
     {
-      "epoch": 2.922244094488189,
-      "grad_norm": 12.880712509155273,
+      "epoch": 2.9242125984251968,
+      "grad_norm": 8.869536399841309,
       "learning_rate": 7.2936493763777695e-09,
-      "loss": 0.4372,
+      "loss": 0.3205,
       "step": 2971
     },
     {
-      "epoch": 2.923228346456693,
-      "grad_norm": 4.7115302085876465,
+      "epoch": 2.925196850393701,
+      "grad_norm": 9.908832550048828,
       "learning_rate": 7.197769560418265e-09,
-      "loss": 0.1008,
+      "loss": 0.4763,
       "step": 2972
     },
     {
-      "epoch": 2.9242125984251968,
-      "grad_norm": 5.608097076416016,
+      "epoch": 2.9261811023622046,
+      "grad_norm": 9.375164985656738,
       "learning_rate": 7.103060125190245e-09,
-      "loss": 0.0808,
+      "loss": 0.3079,
       "step": 2973
     },
     {
-      "epoch": 2.925196850393701,
-      "grad_norm": 18.1824893951416,
+      "epoch": 2.927165354330709,
+      "grad_norm": 6.768562316894531,
       "learning_rate": 7.009521294865773e-09,
-      "loss": 1.1305,
+      "loss": 0.2189,
       "step": 2974
     },
     {
-      "epoch": 2.9261811023622046,
-      "grad_norm": 7.591930866241455,
+      "epoch": 2.9281496062992125,
+      "grad_norm": 10.663924217224121,
       "learning_rate": 6.917153290846126e-09,
-      "loss": 0.2259,
+      "loss": 0.3928,
       "step": 2975
     },
     {
-      "epoch": 2.927165354330709,
-      "grad_norm": 12.07780933380127,
+      "epoch": 2.9291338582677167,
+      "grad_norm": 8.191192626953125,
       "learning_rate": 6.82595633176124e-09,
-      "loss": 0.542,
+      "loss": 0.3047,
       "step": 2976
     },
     {
-      "epoch": 2.9281496062992125,
-      "grad_norm": 7.355630874633789,
+      "epoch": 2.9301181102362204,
+      "grad_norm": 9.06370735168457,
       "learning_rate": 6.735930633469381e-09,
-      "loss": 0.2336,
+      "loss": 0.4384,
       "step": 2977
     },
     {
-      "epoch": 2.9291338582677167,
-      "grad_norm": 9.989481925964355,
+      "epoch": 2.9311023622047245,
+      "grad_norm": 9.023050308227539,
       "learning_rate": 6.647076409056365e-09,
-      "loss": 0.2918,
+      "loss": 0.2607,
       "step": 2978
     },
     {
-      "epoch": 2.9301181102362204,
-      "grad_norm": 15.82009220123291,
+      "epoch": 2.9320866141732282,
+      "grad_norm": 7.34676456451416,
       "learning_rate": 6.559393868835341e-09,
-      "loss": 0.4221,
+      "loss": 0.2344,
       "step": 2979
     },
     {
-      "epoch": 2.9311023622047245,
-      "grad_norm": 5.613612651824951,
+      "epoch": 2.9330708661417324,
+      "grad_norm": 8.566618919372559,
       "learning_rate": 6.472883220346015e-09,
-      "loss": 0.166,
+      "loss": 0.2591,
       "step": 2980
     },
     {
-      "epoch": 2.9320866141732282,
-      "grad_norm": 6.4554033279418945,
+      "epoch": 2.934055118110236,
+      "grad_norm": 9.22314167022705,
       "learning_rate": 6.387544668354376e-09,
-      "loss": 0.1594,
+      "loss": 0.2901,
       "step": 2981
     },
     {
-      "epoch": 2.9330708661417324,
-      "grad_norm": 9.34963321685791,
+      "epoch": 2.9350393700787403,
+      "grad_norm": 4.9475626945495605,
       "learning_rate": 6.3033784148520255e-09,
-      "loss": 0.4577,
+      "loss": 0.0912,
       "step": 2982
     },
     {
-      "epoch": 2.934055118110236,
-      "grad_norm": 20.08962631225586,
+      "epoch": 2.936023622047244,
+      "grad_norm": 12.423126220703125,
       "learning_rate": 6.220384659055909e-09,
-      "loss": 0.9842,
+      "loss": 0.6679,
       "step": 2983
     },
     {
-      "epoch": 2.9350393700787403,
-      "grad_norm": 9.738802909851074,
+      "epoch": 2.937007874015748,
+      "grad_norm": 10.94775104522705,
       "learning_rate": 6.138563597407699e-09,
-      "loss": 0.4133,
+      "loss": 0.4532,
       "step": 2984
     },
     {
-      "epoch": 2.936023622047244,
-      "grad_norm": 5.940000534057617,
+      "epoch": 2.937992125984252,
+      "grad_norm": 7.819056510925293,
       "learning_rate": 6.057915423573362e-09,
-      "loss": 0.1749,
+      "loss": 0.2569,
       "step": 2985
     },
     {
-      "epoch": 2.937007874015748,
-      "grad_norm": 5.507705211639404,
+      "epoch": 2.938976377952756,
+      "grad_norm": 6.427585601806641,
       "learning_rate": 5.9784403284426504e-09,
-      "loss": 0.151,
+      "loss": 0.183,
       "step": 2986
     },
     {
-      "epoch": 2.937992125984252,
-      "grad_norm": 5.60882568359375,
+      "epoch": 2.9399606299212597,
+      "grad_norm": 6.900952339172363,
       "learning_rate": 5.9001385001287815e-09,
-      "loss": 0.1375,
+      "loss": 0.1677,
       "step": 2987
     },
     {
-      "epoch": 2.938976377952756,
-      "grad_norm": 8.243809700012207,
+      "epoch": 2.940944881889764,
+      "grad_norm": 9.287033081054688,
       "learning_rate": 5.8230101239678736e-09,
-      "loss": 0.1525,
+      "loss": 0.2964,
       "step": 2988
     },
     {
-      "epoch": 2.9399606299212597,
-      "grad_norm": 8.408968925476074,
+      "epoch": 2.9419291338582676,
+      "grad_norm": 8.828174591064453,
       "learning_rate": 5.747055382518618e-09,
-      "loss": 0.3028,
+      "loss": 0.3214,
       "step": 2989
     },
     {
-      "epoch": 2.940944881889764,
-      "grad_norm": 7.637701511383057,
+      "epoch": 2.9429133858267718,
+      "grad_norm": 9.51991081237793,
       "learning_rate": 5.672274455561619e-09,
-      "loss": 0.2416,
+      "loss": 0.4345,
       "step": 2990
     },
     {
-      "epoch": 2.9419291338582676,
-      "grad_norm": 8.669809341430664,
+      "epoch": 2.9438976377952755,
+      "grad_norm": 8.270242691040039,
       "learning_rate": 5.5986675200992805e-09,
-      "loss": 0.2551,
+      "loss": 0.2537,
       "step": 2991
     },
     {
-      "epoch": 2.9429133858267718,
-      "grad_norm": 10.099055290222168,
+      "epoch": 2.9448818897637796,
+      "grad_norm": 7.7813262939453125,
       "learning_rate": 5.526234750355195e-09,
-      "loss": 0.3194,
+      "loss": 0.187,
       "step": 2992
     },
     {
-      "epoch": 2.9438976377952755,
-      "grad_norm": 16.009340286254883,
+      "epoch": 2.9458661417322833,
+      "grad_norm": 8.690130233764648,
       "learning_rate": 5.4549763177737016e-09,
-      "loss": 0.3568,
+      "loss": 0.2577,
       "step": 2993
     },
     {
-      "epoch": 2.9448818897637796,
-      "grad_norm": 10.001502990722656,
+      "epoch": 2.9468503937007875,
+      "grad_norm": 10.886865615844727,
       "learning_rate": 5.384892391019614e-09,
-      "loss": 0.9329,
+      "loss": 0.5357,
       "step": 2994
     },
     {
-      "epoch": 2.9458661417322833,
-      "grad_norm": 6.539050102233887,
+      "epoch": 2.947834645669291,
+      "grad_norm": 14.08293628692627,
       "learning_rate": 5.315983135977773e-09,
-      "loss": 0.1931,
+      "loss": 0.6319,
       "step": 2995
     },
     {
-      "epoch": 2.9468503937007875,
-      "grad_norm": 7.452674865722656,
+      "epoch": 2.9488188976377954,
+      "grad_norm": 11.071148872375488,
       "learning_rate": 5.248248715752606e-09,
-      "loss": 0.2403,
+      "loss": 0.4426,
       "step": 2996
     },
     {
-      "epoch": 2.947834645669291,
-      "grad_norm": 13.421844482421875,
+      "epoch": 2.949803149606299,
+      "grad_norm": 11.40675163269043,
       "learning_rate": 5.181689290667684e-09,
-      "loss": 0.8228,
+      "loss": 0.3953,
       "step": 2997
     },
     {
-      "epoch": 2.9488188976377954,
-      "grad_norm": 11.040371894836426,
+      "epoch": 2.9507874015748032,
+      "grad_norm": 16.47882652282715,
       "learning_rate": 5.11630501826561e-09,
-      "loss": 0.478,
+      "loss": 0.6083,
       "step": 2998
     },
     {
-      "epoch": 2.949803149606299,
-      "grad_norm": 11.840859413146973,
+      "epoch": 2.951771653543307,
+      "grad_norm": 12.269733428955078,
       "learning_rate": 5.052096053307301e-09,
-      "loss": 0.4418,
+      "loss": 0.6614,
       "step": 2999
     },
     {
-      "epoch": 2.9507874015748032,
-      "grad_norm": 9.078227996826172,
+      "epoch": 2.952755905511811,
+      "grad_norm": 15.693445205688477,
       "learning_rate": 4.989062547771878e-09,
-      "loss": 0.3721,
+      "loss": 0.7353,
       "step": 3000
     },
     {
-      "epoch": 2.951771653543307,
-      "grad_norm": 18.081226348876953,
+      "epoch": 2.953740157480315,
+      "grad_norm": 9.95142650604248,
       "learning_rate": 4.927204650856165e-09,
-      "loss": 0.388,
+      "loss": 0.3747,
       "step": 3001
     },
     {
-      "epoch": 2.952755905511811,
-      "grad_norm": 13.961949348449707,
+      "epoch": 2.954724409448819,
+      "grad_norm": 9.600488662719727,
       "learning_rate": 4.866522508974473e-09,
-      "loss": 0.6466,
+      "loss": 0.4253,
       "step": 3002
     },
     {
-      "epoch": 2.953740157480315,
-      "grad_norm": 12.137185096740723,
+      "epoch": 2.9557086614173227,
+      "grad_norm": 12.95174503326416,
       "learning_rate": 4.807016265757984e-09,
-      "loss": 0.378,
+      "loss": 0.3911,
       "step": 3003
     },
     {
-      "epoch": 2.954724409448819,
-      "grad_norm": 12.590106964111328,
+      "epoch": 2.956692913385827,
+      "grad_norm": 7.742303371429443,
       "learning_rate": 4.7486860620548696e-09,
-      "loss": 0.4389,
+      "loss": 0.2605,
       "step": 3004
     },
     {
-      "epoch": 2.9557086614173227,
-      "grad_norm": 4.588067531585693,
+      "epoch": 2.9576771653543306,
+      "grad_norm": 7.132505416870117,
       "learning_rate": 4.691532035929345e-09,
-      "loss": 0.1211,
+      "loss": 0.2481,
       "step": 3005
     },
     {
-      "epoch": 2.956692913385827,
-      "grad_norm": 8.0169095993042,
+      "epoch": 2.9586614173228347,
+      "grad_norm": 12.884316444396973,
       "learning_rate": 4.635554322662e-09,
-      "loss": 0.2726,
+      "loss": 0.9352,
       "step": 3006
     },
     {
-      "epoch": 2.9576771653543306,
-      "grad_norm": 11.216535568237305,
+      "epoch": 2.9596456692913384,
+      "grad_norm": 4.859545707702637,
       "learning_rate": 4.580753054748868e-09,
-      "loss": 0.3852,
+      "loss": 0.0924,
       "step": 3007
     },
     {
-      "epoch": 2.9586614173228347,
-      "grad_norm": 13.945511817932129,
+      "epoch": 2.9606299212598426,
+      "grad_norm": 7.548928260803223,
       "learning_rate": 4.527128361901688e-09,
-      "loss": 0.8051,
+      "loss": 0.2654,
       "step": 3008
     },
     {
-      "epoch": 2.9596456692913384,
-      "grad_norm": 10.2655611038208,
+      "epoch": 2.9616141732283463,
+      "grad_norm": 9.318865776062012,
       "learning_rate": 4.4746803710470335e-09,
-      "loss": 0.3938,
+      "loss": 0.3855,
       "step": 3009
     },
     {
-      "epoch": 2.9606299212598426,
-      "grad_norm": 9.719473838806152,
+      "epoch": 2.9625984251968505,
+      "grad_norm": 7.037998199462891,
       "learning_rate": 4.423409206326521e-09,
-      "loss": 0.3433,
+      "loss": 0.2321,
       "step": 3010
     },
     {
-      "epoch": 2.9616141732283463,
-      "grad_norm": 7.4270734786987305,
+      "epoch": 2.963582677165354,
+      "grad_norm": 8.737883567810059,
       "learning_rate": 4.37331498909616e-09,
-      "loss": 0.2285,
+      "loss": 0.3513,
       "step": 3011
     },
     {
-      "epoch": 2.9625984251968505,
-      "grad_norm": 4.804949760437012,
+      "epoch": 2.9645669291338583,
+      "grad_norm": 7.727255821228027,
       "learning_rate": 4.324397837926115e-09,
-      "loss": 0.0872,
+      "loss": 0.1755,
       "step": 3012
     },
     {
-      "epoch": 2.963582677165354,
-      "grad_norm": 9.397032737731934,
+      "epoch": 2.965551181102362,
+      "grad_norm": 7.446308612823486,
       "learning_rate": 4.2766578686007245e-09,
-      "loss": 0.3572,
+      "loss": 0.231,
       "step": 3013
     },
     {
-      "epoch": 2.9645669291338583,
-      "grad_norm": 9.023151397705078,
+      "epoch": 2.966535433070866,
+      "grad_norm": 13.526126861572266,
       "learning_rate": 4.230095194117821e-09,
-      "loss": 0.2511,
+      "loss": 0.7136,
       "step": 3014
     },
     {
-      "epoch": 2.965551181102362,
-      "grad_norm": 8.523262023925781,
+      "epoch": 2.96751968503937,
+      "grad_norm": 5.387681007385254,
       "learning_rate": 4.184709924688737e-09,
-      "loss": 0.3336,
+      "loss": 0.1341,
       "step": 3015
     },
     {
-      "epoch": 2.966535433070866,
-      "grad_norm": 9.497230529785156,
+      "epoch": 2.968503937007874,
+      "grad_norm": 10.890751838684082,
       "learning_rate": 4.140502167737812e-09,
-      "loss": 0.4178,
+      "loss": 0.5539,
       "step": 3016
     },
     {
-      "epoch": 2.96751968503937,
-      "grad_norm": 22.946258544921875,
+      "epoch": 2.969488188976378,
+      "grad_norm": 12.355507850646973,
       "learning_rate": 4.097472027902494e-09,
-      "loss": 0.7104,
+      "loss": 0.498,
       "step": 3017
     },
     {
-      "epoch": 2.968503937007874,
-      "grad_norm": 12.379600524902344,
+      "epoch": 2.970472440944882,
+      "grad_norm": 5.778040409088135,
       "learning_rate": 4.055619607032737e-09,
-      "loss": 0.5916,
+      "loss": 0.2991,
       "step": 3018
     },
     {
-      "epoch": 2.969488188976378,
-      "grad_norm": 9.416070938110352,
+      "epoch": 2.971456692913386,
+      "grad_norm": 6.504598140716553,
       "learning_rate": 4.014945004190834e-09,
-      "loss": 0.4608,
+      "loss": 0.2015,
       "step": 3019
     },
     {
-      "epoch": 2.970472440944882,
-      "grad_norm": 16.8037052154541,
+      "epoch": 2.97244094488189,
+      "grad_norm": 9.223751068115234,
       "learning_rate": 3.975448315651416e-09,
-      "loss": 0.728,
+      "loss": 0.2851,
       "step": 3020
     },
     {
-      "epoch": 2.971456692913386,
-      "grad_norm": 4.428066730499268,
+      "epoch": 2.9734251968503935,
+      "grad_norm": 11.1191987991333,
       "learning_rate": 3.937129634901008e-09,
-      "loss": 0.1237,
+      "loss": 0.5233,
       "step": 3021
     },
     {
-      "epoch": 2.97244094488189,
-      "grad_norm": 8.113018989562988,
+      "epoch": 2.9744094488188977,
+      "grad_norm": 7.762978553771973,
       "learning_rate": 3.899989052637758e-09,
-      "loss": 0.2183,
+      "loss": 0.3245,
       "step": 3022
     },
     {
-      "epoch": 2.9734251968503935,
-      "grad_norm": 6.563396453857422,
+      "epoch": 2.975393700787402,
+      "grad_norm": 12.028802871704102,
       "learning_rate": 3.864026656771375e-09,
-      "loss": 0.2081,
+      "loss": 0.4117,
       "step": 3023
     },
     {
-      "epoch": 2.9744094488188977,
-      "grad_norm": 7.403074741363525,
+      "epoch": 2.9763779527559056,
+      "grad_norm": 7.198526859283447,
       "learning_rate": 3.8292425324229665e-09,
-      "loss": 0.2921,
+      "loss": 0.1747,
       "step": 3024
     },
     {
-      "epoch": 2.975393700787402,
-      "grad_norm": 7.093117713928223,
+      "epoch": 2.9773622047244093,
+      "grad_norm": 10.538836479187012,
       "learning_rate": 3.79563676192454e-09,
-      "loss": 0.3556,
+      "loss": 0.4728,
       "step": 3025
     },
     {
-      "epoch": 2.9763779527559056,
-      "grad_norm": 7.03363037109375,
+      "epoch": 2.9783464566929134,
+      "grad_norm": 11.64247989654541,
       "learning_rate": 3.763209424819168e-09,
-      "loss": 0.189,
+      "loss": 0.658,
       "step": 3026
     },
     {
-      "epoch": 2.9773622047244093,
-      "grad_norm": 8.17770767211914,
+      "epoch": 2.9793307086614176,
+      "grad_norm": 13.467778205871582,
       "learning_rate": 3.731960597860549e-09,
-      "loss": 0.3349,
+      "loss": 0.7424,
       "step": 3027
     },
     {
-      "epoch": 2.9783464566929134,
-      "grad_norm": 7.777528285980225,
+      "epoch": 2.9803149606299213,
+      "grad_norm": 13.620326042175293,
       "learning_rate": 3.7018903550130016e-09,
-      "loss": 0.3051,
+      "loss": 0.5564,
       "step": 3028
     },
     {
-      "epoch": 2.9793307086614176,
-      "grad_norm": 8.266366004943848,
+      "epoch": 2.981299212598425,
+      "grad_norm": 8.670527458190918,
       "learning_rate": 3.6729987674510274e-09,
-      "loss": 0.3591,
+      "loss": 0.3174,
       "step": 3029
     },
     {
-      "epoch": 2.9803149606299213,
-      "grad_norm": 10.026298522949219,
+      "epoch": 2.982283464566929,
+      "grad_norm": 11.121075630187988,
       "learning_rate": 3.6452859035595284e-09,
-      "loss": 0.4902,
+      "loss": 0.5552,
       "step": 3030
     },
     {
-      "epoch": 2.981299212598425,
-      "grad_norm": 9.655516624450684,
+      "epoch": 2.9832677165354333,
+      "grad_norm": 5.212947845458984,
       "learning_rate": 3.6187518289333124e-09,
-      "loss": 0.4644,
+      "loss": 0.1804,
       "step": 3031
     },
     {
-      "epoch": 2.982283464566929,
-      "grad_norm": 18.602052688598633,
+      "epoch": 2.984251968503937,
+      "grad_norm": 10.453910827636719,
       "learning_rate": 3.5933966063770337e-09,
-      "loss": 0.7464,
+      "loss": 0.5327,
       "step": 3032
     },
     {
-      "epoch": 2.9832677165354333,
-      "grad_norm": 16.17316246032715,
+      "epoch": 2.9852362204724407,
+      "grad_norm": 7.186238765716553,
       "learning_rate": 3.5692202959051424e-09,
-      "loss": 0.7023,
+      "loss": 0.3662,
       "step": 3033
     },
     {
-      "epoch": 2.984251968503937,
-      "grad_norm": 8.509237289428711,
+      "epoch": 2.986220472440945,
+      "grad_norm": 10.561626434326172,
       "learning_rate": 3.5462229547416585e-09,
-      "loss": 0.2988,
+      "loss": 0.452,
       "step": 3034
     },
     {
-      "epoch": 2.9852362204724407,
-      "grad_norm": 11.711136817932129,
+      "epoch": 2.987204724409449,
+      "grad_norm": 4.8978190422058105,
       "learning_rate": 3.5244046373200104e-09,
-      "loss": 0.6121,
+      "loss": 0.105,
       "step": 3035
     },
     {
-      "epoch": 2.986220472440945,
-      "grad_norm": 11.348662376403809,
+      "epoch": 2.9881889763779528,
+      "grad_norm": 10.981940269470215,
       "learning_rate": 3.50376539528292e-09,
-      "loss": 0.5366,
+      "loss": 0.6204,
       "step": 3036
     },
     {
-      "epoch": 2.987204724409449,
-      "grad_norm": 12.44474983215332,
+      "epoch": 2.9891732283464565,
+      "grad_norm": 7.311155319213867,
       "learning_rate": 3.48430527748235e-09,
-      "loss": 0.6555,
+      "loss": 0.2357,
       "step": 3037
     },
     {
-      "epoch": 2.9881889763779528,
-      "grad_norm": 8.161657333374023,
+      "epoch": 2.9901574803149606,
+      "grad_norm": 10.446833610534668,
       "learning_rate": 3.4660243299793937e-09,
-      "loss": 0.2812,
+      "loss": 0.3682,
       "step": 3038
     },
     {
-      "epoch": 2.9891732283464565,
-      "grad_norm": 13.011687278747559,
+      "epoch": 2.991141732283465,
+      "grad_norm": 6.968286514282227,
       "learning_rate": 3.448922596043995e-09,
-      "loss": 0.6854,
+      "loss": 0.3431,
       "step": 3039
     },
     {
-      "epoch": 2.9901574803149606,
-      "grad_norm": 9.701081275939941,
+      "epoch": 2.9921259842519685,
+      "grad_norm": 13.282569885253906,
       "learning_rate": 3.4330001161550637e-09,
-      "loss": 0.3585,
+      "loss": 0.4461,
       "step": 3040
     },
     {
-      "epoch": 2.991141732283465,
-      "grad_norm": 7.371988296508789,
+      "epoch": 2.9931102362204722,
+      "grad_norm": 9.187698364257812,
       "learning_rate": 3.4182569280001402e-09,
-      "loss": 0.2875,
+      "loss": 0.2229,
       "step": 3041
     },
     {
-      "epoch": 2.9921259842519685,
-      "grad_norm": 6.704482078552246,
+      "epoch": 2.9940944881889764,
+      "grad_norm": 9.882904052734375,
       "learning_rate": 3.404693066475672e-09,
-      "loss": 0.1937,
+      "loss": 0.369,
       "step": 3042
     },
     {
-      "epoch": 2.9931102362204722,
-      "grad_norm": 5.843419551849365,
+      "epoch": 2.9950787401574805,
+      "grad_norm": 9.664177894592285,
       "learning_rate": 3.392308563686463e-09,
-      "loss": 0.1825,
+      "loss": 0.2761,
       "step": 3043
     },
     {
-      "epoch": 2.9940944881889764,
-      "grad_norm": 12.087967872619629,
+      "epoch": 2.9960629921259843,
+      "grad_norm": 7.984935283660889,
       "learning_rate": 3.381103448946004e-09,
-      "loss": 0.6199,
+      "loss": 0.1884,
       "step": 3044
     },
     {
-      "epoch": 2.9950787401574805,
-      "grad_norm": 8.321404457092285,
+      "epoch": 2.997047244094488,
+      "grad_norm": 9.106756210327148,
       "learning_rate": 3.3710777487761954e-09,
-      "loss": 0.2884,
+      "loss": 0.5279,
       "step": 3045
     },
     {
-      "epoch": 2.9960629921259843,
-      "grad_norm": 7.362402439117432,
+      "epoch": 2.998031496062992,
+      "grad_norm": 12.392125129699707,
       "learning_rate": 3.3622314869072926e-09,
-      "loss": 0.2596,
+      "loss": 0.5891,
       "step": 3046
     },
     {
-      "epoch": 2.997047244094488,
-      "grad_norm": 6.557697296142578,
+      "epoch": 2.9990157480314963,
+      "grad_norm": 8.469876289367676,
       "learning_rate": 3.3545646842779058e-09,
-      "loss": 0.2532,
+      "loss": 0.2687,
       "step": 3047
     },
     {
-      "epoch": 2.998031496062992,
-      "grad_norm": 14.144463539123535,
+      "epoch": 3.0,
+      "grad_norm": 4.075329780578613,
       "learning_rate": 3.348077359034945e-09,
-      "loss": 0.8036,
+      "loss": 0.0841,
       "step": 3048
     }
   ],