diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -2,9 +2,9 @@
   "best_global_step": null,
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.5465717757333797,
+  "epoch": 0.6887693108835272,
   "eval_steps": 2951,
-  "global_step": 5904,
+  "global_step": 7440,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -41352,6 +41352,10758 @@
       "learning_rate": 0.02,
       "loss": 1.565,
       "step": 5904
+    },
+    {
+      "epoch": 0.5466643522536596,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.02,
+      "loss": 1.5874,
+      "step": 5905
+    },
+    {
+      "epoch": 0.5467569287739397,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.595,
+      "step": 5906
+    },
+    {
+      "epoch": 0.5468495052942197,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5294,
+      "step": 5907
+    },
+    {
+      "epoch": 0.5469420818144998,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5086,
+      "step": 5908
+    },
+    {
+      "epoch": 0.5470346583347798,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5447,
+      "step": 5909
+    },
+    {
+      "epoch": 0.5471272348550599,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.02,
+      "loss": 1.5581,
+      "step": 5910
+    },
+    {
+      "epoch": 0.5472198113753399,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.5498,
+      "step": 5911
+    },
+    {
+      "epoch": 0.54731238789562,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.6479,
+      "step": 5912
+    },
+    {
+      "epoch": 0.5474049644159,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.581,
+      "step": 5913
+    },
+    {
+      "epoch": 0.5474975409361801,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.569,
+      "step": 5914
+    },
+    {
+      "epoch": 0.5475901174564601,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.02,
+      "loss": 1.5129,
+      "step": 5915
+    },
+    {
+      "epoch": 0.5476826939767402,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.02,
+      "loss": 1.5106,
+      "step": 5916
+    },
+    {
+      "epoch": 0.5477752704970202,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.543,
+      "step": 5917
+    },
+    {
+      "epoch": 0.5478678470173003,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.6299,
+      "step": 5918
+    },
+    {
+      "epoch": 0.5479604235375802,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5737,
+      "step": 5919
+    },
+    {
+      "epoch": 0.5480530000578603,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.02,
+      "loss": 1.5662,
+      "step": 5920
+    },
+    {
+      "epoch": 0.5481455765781403,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5858,
+      "step": 5921
+    },
+    {
+      "epoch": 0.5482381530984204,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5562,
+      "step": 5922
+    },
+    {
+      "epoch": 0.5483307296187004,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.566,
+      "step": 5923
+    },
+    {
+      "epoch": 0.5484233061389805,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.02,
+      "loss": 1.5291,
+      "step": 5924
+    },
+    {
+      "epoch": 0.5485158826592605,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5176,
+      "step": 5925
+    },
+    {
+      "epoch": 0.5486084591795406,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.5948,
+      "step": 5926
+    },
+    {
+      "epoch": 0.5487010356998206,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.562,
+      "step": 5927
+    },
+    {
+      "epoch": 0.5487936122201007,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5255,
+      "step": 5928
+    },
+    {
+      "epoch": 0.5488861887403808,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.4922,
+      "step": 5929
+    },
+    {
+      "epoch": 0.5489787652606608,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5217,
+      "step": 5930
+    },
+    {
+      "epoch": 0.5490713417809409,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.5319,
+      "step": 5931
+    },
+    {
+      "epoch": 0.5491639183012208,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.5527,
+      "step": 5932
+    },
+    {
+      "epoch": 0.5492564948215009,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5471,
+      "step": 5933
+    },
+    {
+      "epoch": 0.5493490713417809,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.02,
+      "loss": 1.545,
+      "step": 5934
+    },
+    {
+      "epoch": 0.549441647862061,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5043,
+      "step": 5935
+    },
+    {
+      "epoch": 0.549534224382341,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.02,
+      "loss": 1.5489,
+      "step": 5936
+    },
+    {
+      "epoch": 0.5496268009026211,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5497,
+      "step": 5937
+    },
+    {
+      "epoch": 0.5497193774229011,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5397,
+      "step": 5938
+    },
+    {
+      "epoch": 0.5498119539431812,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.519,
+      "step": 5939
+    },
+    {
+      "epoch": 0.5499045304634612,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.593,
+      "step": 5940
+    },
+    {
+      "epoch": 0.5499971069837413,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5034,
+      "step": 5941
+    },
+    {
+      "epoch": 0.5500896835040213,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5427,
+      "step": 5942
+    },
+    {
+      "epoch": 0.5501822600243014,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.5985,
+      "step": 5943
+    },
+    {
+      "epoch": 0.5502748365445814,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.5973,
+      "step": 5944
+    },
+    {
+      "epoch": 0.5503674130648615,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5879,
+      "step": 5945
+    },
+    {
+      "epoch": 0.5504599895851414,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.5132,
+      "step": 5946
+    },
+    {
+      "epoch": 0.5505525661054215,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5282,
+      "step": 5947
+    },
+    {
+      "epoch": 0.5506451426257015,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.02,
+      "loss": 1.5109,
+      "step": 5948
+    },
+    {
+      "epoch": 0.5507377191459816,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5365,
+      "step": 5949
+    },
+    {
+      "epoch": 0.5508302956662616,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.02,
+      "loss": 1.5245,
+      "step": 5950
+    },
+    {
+      "epoch": 0.5509228721865417,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.02,
+      "loss": 1.5004,
+      "step": 5951
+    },
+    {
+      "epoch": 0.5510154487068217,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.5682,
+      "step": 5952
+    },
+    {
+      "epoch": 0.5511080252271018,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.02,
+      "loss": 1.5883,
+      "step": 5953
+    },
+    {
+      "epoch": 0.5512006017473818,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.02,
+      "loss": 1.5506,
+      "step": 5954
+    },
+    {
+      "epoch": 0.5512931782676619,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.02,
+      "loss": 1.5621,
+      "step": 5955
+    },
+    {
+      "epoch": 0.5513857547879419,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.5737,
+      "step": 5956
+    },
+    {
+      "epoch": 0.551478331308222,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.02,
+      "loss": 1.5726,
+      "step": 5957
+    },
+    {
+      "epoch": 0.551570907828502,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5612,
+      "step": 5958
+    },
+    {
+      "epoch": 0.551663484348782,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.448,
+      "step": 5959
+    },
+    {
+      "epoch": 0.551756060869062,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.02,
+      "loss": 1.5272,
+      "step": 5960
+    },
+    {
+      "epoch": 0.5518486373893421,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.5679,
+      "step": 5961
+    },
+    {
+      "epoch": 0.5519412139096221,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.4825,
+      "step": 5962
+    },
+    {
+      "epoch": 0.5520337904299022,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.5138,
+      "step": 5963
+    },
+    {
+      "epoch": 0.5521263669501822,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5615,
+      "step": 5964
+    },
+    {
+      "epoch": 0.5522189434704623,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5261,
+      "step": 5965
+    },
+    {
+      "epoch": 0.5523115199907424,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.5633,
+      "step": 5966
+    },
+    {
+      "epoch": 0.5524040965110224,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.5021,
+      "step": 5967
+    },
+    {
+      "epoch": 0.5524966730313025,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.6194,
+      "step": 5968
+    },
+    {
+      "epoch": 0.5525892495515825,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5489,
+      "step": 5969
+    },
+    {
+      "epoch": 0.5526818260718626,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5773,
+      "step": 5970
+    },
+    {
+      "epoch": 0.5527744025921426,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.6275,
+      "step": 5971
+    },
+    {
+      "epoch": 0.5528669791124227,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5864,
+      "step": 5972
+    },
+    {
+      "epoch": 0.5529595556327026,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5235,
+      "step": 5973
+    },
+    {
+      "epoch": 0.5530521321529827,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.574,
+      "step": 5974
+    },
+    {
+      "epoch": 0.5531447086732627,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5496,
+      "step": 5975
+    },
+    {
+      "epoch": 0.5532372851935428,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.5547,
+      "step": 5976
+    },
+    {
+      "epoch": 0.5533298617138228,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5179,
+      "step": 5977
+    },
+    {
+      "epoch": 0.5534224382341029,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.6324,
+      "step": 5978
+    },
+    {
+      "epoch": 0.5535150147543829,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.6043,
+      "step": 5979
+    },
+    {
+      "epoch": 0.553607591274663,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5972,
+      "step": 5980
+    },
+    {
+      "epoch": 0.553700167794943,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5851,
+      "step": 5981
+    },
+    {
+      "epoch": 0.5537927443152231,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5137,
+      "step": 5982
+    },
+    {
+      "epoch": 0.5538853208355031,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5073,
+      "step": 5983
+    },
+    {
+      "epoch": 0.5539778973557832,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.5631,
+      "step": 5984
+    },
+    {
+      "epoch": 0.5540704738760632,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5708,
+      "step": 5985
+    },
+    {
+      "epoch": 0.5541630503963433,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.505,
+      "step": 5986
+    },
+    {
+      "epoch": 0.5542556269166232,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.02,
+      "loss": 1.513,
+      "step": 5987
+    },
+    {
+      "epoch": 0.5543482034369033,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5705,
+      "step": 5988
+    },
+    {
+      "epoch": 0.5544407799571833,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.6026,
+      "step": 5989
+    },
+    {
+      "epoch": 0.5545333564774634,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.505,
+      "step": 5990
+    },
+    {
+      "epoch": 0.5546259329977434,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5464,
+      "step": 5991
+    },
+    {
+      "epoch": 0.5547185095180235,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5789,
+      "step": 5992
+    },
+    {
+      "epoch": 0.5548110860383035,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.6512,
+      "step": 5993
+    },
+    {
+      "epoch": 0.5549036625585836,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.6024,
+      "step": 5994
+    },
+    {
+      "epoch": 0.5549962390788636,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5744,
+      "step": 5995
+    },
+    {
+      "epoch": 0.5550888155991437,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.5413,
+      "step": 5996
+    },
+    {
+      "epoch": 0.5551813921194237,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5813,
+      "step": 5997
+    },
+    {
+      "epoch": 0.5552739686397038,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.55,
+      "step": 5998
+    },
+    {
+      "epoch": 0.5553665451599838,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.6071,
+      "step": 5999
+    },
+    {
+      "epoch": 0.5554591216802638,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.02,
+      "loss": 1.6078,
+      "step": 6000
+    },
+    {
+      "epoch": 0.5555516982005438,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5247,
+      "step": 6001
+    },
+    {
+      "epoch": 0.5556442747208239,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.6026,
+      "step": 6002
+    },
+    {
+      "epoch": 0.555736851241104,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5895,
+      "step": 6003
+    },
+    {
+      "epoch": 0.555829427761384,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.4875,
+      "step": 6004
+    },
+    {
+      "epoch": 0.555922004281664,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5848,
+      "step": 6005
+    },
+    {
+      "epoch": 0.5560145808019441,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.523,
+      "step": 6006
+    },
+    {
+      "epoch": 0.5561071573222242,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.6117,
+      "step": 6007
+    },
+    {
+      "epoch": 0.5561997338425042,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.5251,
+      "step": 6008
+    },
+    {
+      "epoch": 0.5562923103627843,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5458,
+      "step": 6009
+    },
+    {
+      "epoch": 0.5563848868830643,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.6022,
+      "step": 6010
+    },
+    {
+      "epoch": 0.5564774634033444,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5169,
+      "step": 6011
+    },
+    {
+      "epoch": 0.5565700399236244,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5442,
+      "step": 6012
+    },
+    {
+      "epoch": 0.5566626164439045,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.5905,
+      "step": 6013
+    },
+    {
+      "epoch": 0.5567551929641844,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.5151,
+      "step": 6014
+    },
+    {
+      "epoch": 0.5568477694844645,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.6318,
+      "step": 6015
+    },
+    {
+      "epoch": 0.5569403460047445,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.507,
+      "step": 6016
+    },
+    {
+      "epoch": 0.5570329225250246,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5276,
+      "step": 6017
+    },
+    {
+      "epoch": 0.5571254990453046,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.552,
+      "step": 6018
+    },
+    {
+      "epoch": 0.5572180755655847,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5458,
+      "step": 6019
+    },
+    {
+      "epoch": 0.5573106520858647,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.02,
+      "loss": 1.6208,
+      "step": 6020
+    },
+    {
+      "epoch": 0.5574032286061448,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5786,
+      "step": 6021
+    },
+    {
+      "epoch": 0.5574958051264248,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.559,
+      "step": 6022
+    },
+    {
+      "epoch": 0.5575883816467049,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.5314,
+      "step": 6023
+    },
+    {
+      "epoch": 0.5576809581669849,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.526,
+      "step": 6024
+    },
+    {
+      "epoch": 0.557773534687265,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.6247,
+      "step": 6025
+    },
+    {
+      "epoch": 0.557866111207545,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.6031,
+      "step": 6026
+    },
+    {
+      "epoch": 0.557958687727825,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.521,
+      "step": 6027
+    },
+    {
+      "epoch": 0.558051264248105,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5293,
+      "step": 6028
+    },
+    {
+      "epoch": 0.5581438407683851,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.02,
+      "loss": 1.5642,
+      "step": 6029
+    },
+    {
+      "epoch": 0.5582364172886651,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.6067,
+      "step": 6030
+    },
+    {
+      "epoch": 0.5583289938089452,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.6204,
+      "step": 6031
+    },
+    {
+      "epoch": 0.5584215703292252,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.02,
+      "loss": 1.5232,
+      "step": 6032
+    },
+    {
+      "epoch": 0.5585141468495053,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.02,
+      "loss": 1.6223,
+      "step": 6033
+    },
+    {
+      "epoch": 0.5586067233697853,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.02,
+      "loss": 1.5342,
+      "step": 6034
+    },
+    {
+      "epoch": 0.5586992998900654,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5241,
+      "step": 6035
+    },
+    {
+      "epoch": 0.5587918764103454,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5571,
+      "step": 6036
+    },
+    {
+      "epoch": 0.5588844529306255,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5282,
+      "step": 6037
+    },
+    {
+      "epoch": 0.5589770294509055,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5159,
+      "step": 6038
+    },
+    {
+      "epoch": 0.5590696059711856,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5565,
+      "step": 6039
+    },
+    {
+      "epoch": 0.5591621824914657,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.02,
+      "loss": 1.5704,
+      "step": 6040
+    },
+    {
+      "epoch": 0.5592547590117456,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.02,
+      "loss": 1.5175,
+      "step": 6041
+    },
+    {
+      "epoch": 0.5593473355320256,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.4938,
+      "step": 6042
+    },
+    {
+      "epoch": 0.5594399120523057,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.5651,
+      "step": 6043
+    },
+    {
+      "epoch": 0.5595324885725858,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.5619,
+      "step": 6044
+    },
+    {
+      "epoch": 0.5596250650928658,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5835,
+      "step": 6045
+    },
+    {
+      "epoch": 0.5597176416131459,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.6105,
+      "step": 6046
+    },
+    {
+      "epoch": 0.5598102181334259,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5772,
+      "step": 6047
+    },
+    {
+      "epoch": 0.559902794653706,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5302,
+      "step": 6048
+    },
+    {
+      "epoch": 0.559995371173986,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5118,
+      "step": 6049
+    },
+    {
+      "epoch": 0.5600879476942661,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.02,
+      "loss": 1.5045,
+      "step": 6050
+    },
+    {
+      "epoch": 0.5601805242145461,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5051,
+      "step": 6051
+    },
+    {
+      "epoch": 0.5602731007348262,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.6011,
+      "step": 6052
+    },
+    {
+      "epoch": 0.5603656772551062,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.02,
+      "loss": 1.5594,
+      "step": 6053
+    },
+    {
+      "epoch": 0.5604582537753863,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.5485,
+      "step": 6054
+    },
+    {
+      "epoch": 0.5605508302956662,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5687,
+      "step": 6055
+    },
+    {
+      "epoch": 0.5606434068159463,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.5773,
+      "step": 6056
+    },
+    {
+      "epoch": 0.5607359833362263,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5983,
+      "step": 6057
+    },
+    {
+      "epoch": 0.5608285598565064,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5267,
+      "step": 6058
+    },
+    {
+      "epoch": 0.5609211363767864,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.6068,
+      "step": 6059
+    },
+    {
+      "epoch": 0.5610137128970665,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.4765,
+      "step": 6060
+    },
+    {
+      "epoch": 0.5611062894173465,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5653,
+      "step": 6061
+    },
+    {
+      "epoch": 0.5611988659376266,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.02,
+      "loss": 1.5231,
+      "step": 6062
+    },
+    {
+      "epoch": 0.5612914424579066,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5525,
+      "step": 6063
+    },
+    {
+      "epoch": 0.5613840189781867,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5029,
+      "step": 6064
+    },
+    {
+      "epoch": 0.5614765954984667,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5684,
+      "step": 6065
+    },
+    {
+      "epoch": 0.5615691720187468,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.02,
+      "loss": 1.6028,
+      "step": 6066
+    },
+    {
+      "epoch": 0.5616617485390268,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5625,
+      "step": 6067
+    },
+    {
+      "epoch": 0.5617543250593068,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5099,
+      "step": 6068
+    },
+    {
+      "epoch": 0.5618469015795868,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.6117,
+      "step": 6069
+    },
+    {
+      "epoch": 0.5619394780998669,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.558,
+      "step": 6070
+    },
+    {
+      "epoch": 0.5620320546201469,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.02,
+      "loss": 1.5471,
+      "step": 6071
+    },
+    {
+      "epoch": 0.562124631140427,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5803,
+      "step": 6072
+    },
+    {
+      "epoch": 0.562217207660707,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.02,
+      "loss": 1.5761,
+      "step": 6073
+    },
+    {
+      "epoch": 0.5623097841809871,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.6077,
+      "step": 6074
+    },
+    {
+      "epoch": 0.5624023607012671,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5395,
+      "step": 6075
+    },
+    {
+      "epoch": 0.5624949372215472,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5342,
+      "step": 6076
+    },
+    {
+      "epoch": 0.5625875137418272,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5355,
+      "step": 6077
+    },
+    {
+      "epoch": 0.5626800902621073,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.02,
+      "loss": 1.4983,
+      "step": 6078
+    },
+    {
+      "epoch": 0.5627726667823874,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5737,
+      "step": 6079
+    },
+    {
+      "epoch": 0.5628652433026674,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5429,
+      "step": 6080
+    },
+    {
+      "epoch": 0.5629578198229475,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.528,
+      "step": 6081
+    },
+    {
+      "epoch": 0.5630503963432274,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.562,
+      "step": 6082
+    },
+    {
+      "epoch": 0.5631429728635075,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5798,
+      "step": 6083
+    },
+    {
+      "epoch": 0.5632355493837875,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5786,
+      "step": 6084
+    },
+    {
+      "epoch": 0.5633281259040676,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5148,
+      "step": 6085
+    },
+    {
+      "epoch": 0.5634207024243476,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5807,
+      "step": 6086
+    },
+    {
+      "epoch": 0.5635132789446277,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5962,
+      "step": 6087
+    },
+    {
+      "epoch": 0.5636058554649077,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.4903,
+      "step": 6088
+    },
+    {
+      "epoch": 0.5636984319851878,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5515,
+      "step": 6089
+    },
+    {
+      "epoch": 0.5637910085054678,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5646,
+      "step": 6090
+    },
+    {
+      "epoch": 0.5638835850257479,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.02,
+      "loss": 1.4961,
+      "step": 6091
+    },
+    {
+      "epoch": 0.5639761615460279,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5592,
+      "step": 6092
+    },
+    {
+      "epoch": 0.564068738066308,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5387,
+      "step": 6093
+    },
+    {
+      "epoch": 0.564161314586588,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5201,
+      "step": 6094
+    },
+    {
+      "epoch": 0.564253891106868,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5454,
+      "step": 6095
+    },
+    {
+      "epoch": 0.564346467627148,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.6085,
+      "step": 6096
+    },
+    {
+      "epoch": 0.5644390441474281,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5805,
+      "step": 6097
+    },
+    {
+      "epoch": 0.5645316206677081,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5672,
+      "step": 6098
+    },
+    {
+      "epoch": 0.5646241971879882,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5656,
+      "step": 6099
+    },
+    {
+      "epoch": 0.5647167737082682,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5225,
+      "step": 6100
+    },
+    {
+      "epoch": 0.5648093502285483,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5923,
+      "step": 6101
+    },
+    {
+      "epoch": 0.5649019267488283,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5151,
+      "step": 6102
+    },
+    {
+      "epoch": 0.5649945032691084,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.6001,
+      "step": 6103
+    },
+    {
+      "epoch": 0.5650870797893884,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.6102,
+      "step": 6104
+    },
+    {
+      "epoch": 0.5651796563096685,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5519,
+      "step": 6105
+    },
+    {
+      "epoch": 0.5652722328299485,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.02,
+      "loss": 1.5612,
+      "step": 6106
+    },
+    {
+      "epoch": 0.5653648093502286,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5942,
+      "step": 6107
+    },
+    {
+      "epoch": 0.5654573858705086,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.02,
+      "loss": 1.574,
+      "step": 6108
+    },
+    {
+      "epoch": 0.5655499623907886,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5701,
+      "step": 6109
+    },
+    {
+      "epoch": 0.5656425389110686,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.5858,
+      "step": 6110
+    },
+    {
+      "epoch": 0.5657351154313487,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.02,
+      "loss": 1.5318,
+      "step": 6111
+    },
+    {
+      "epoch": 0.5658276919516287,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5587,
+      "step": 6112
+    },
+    {
+      "epoch": 0.5659202684719088,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.571,
+      "step": 6113
+    },
+    {
+      "epoch": 0.5660128449921888,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5667,
+      "step": 6114
+    },
+    {
+      "epoch": 0.5661054215124689,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.5175,
+      "step": 6115
+    },
+    {
+      "epoch": 0.566197998032749,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.4651,
+      "step": 6116
+    },
+    {
+      "epoch": 0.566290574553029,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5533,
+      "step": 6117
+    },
+    {
+      "epoch": 0.566383151073309,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.02,
+      "loss": 1.6063,
+      "step": 6118
+    },
+    {
+      "epoch": 0.5664757275935891,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5644,
+      "step": 6119
+    },
+    {
+      "epoch": 0.5665683041138692,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5103,
+      "step": 6120
+    },
+    {
+      "epoch": 0.5666608806341492,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.5295,
+      "step": 6121
+    },
+    {
+      "epoch": 0.5667534571544293,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.4538,
+      "step": 6122
+    },
+    {
+      "epoch": 0.5668460336747092,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.02,
+      "loss": 1.5419,
+      "step": 6123
+    },
+    {
+      "epoch": 0.5669386101949893,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.02,
+      "loss": 1.5419,
+      "step": 6124
+    },
+    {
+      "epoch": 0.5670311867152693,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.6051,
+      "step": 6125
+    },
+    {
+      "epoch": 0.5671237632355494,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.02,
+      "loss": 1.5779,
+      "step": 6126
+    },
+    {
+      "epoch": 0.5672163397558294,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5425,
+      "step": 6127
+    },
+    {
+      "epoch": 0.5673089162761095,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.6006,
+      "step": 6128
+    },
+    {
+      "epoch": 0.5674014927963895,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5269,
+      "step": 6129
+    },
+    {
+      "epoch": 0.5674940693166696,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.02,
+      "loss": 1.6145,
+      "step": 6130
+    },
+    {
+      "epoch": 0.5675866458369496,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.552,
+      "step": 6131
+    },
+    {
+      "epoch": 0.5676792223572297,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.4956,
+      "step": 6132
+    },
+    {
+      "epoch": 0.5677717988775097,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5368,
+      "step": 6133
+    },
+    {
+      "epoch": 0.5678643753977898,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.02,
+      "loss": 1.552,
+      "step": 6134
+    },
+    {
+      "epoch": 0.5679569519180698,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.6023,
+      "step": 6135
+    },
+    {
+      "epoch": 0.5680495284383498,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5595,
+      "step": 6136
+    },
+    {
+      "epoch": 0.5681421049586298,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.621,
+      "step": 6137
+    },
+    {
+      "epoch": 0.5682346814789099,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.6304,
+      "step": 6138
+    },
+    {
+      "epoch": 0.5683272579991899,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.5273,
+      "step": 6139
+    },
+    {
+      "epoch": 0.56841983451947,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.459,
+      "step": 6140
+    },
+    {
+      "epoch": 0.56851241103975,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5311,
+      "step": 6141
+    },
+    {
+      "epoch": 0.5686049875600301,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.02,
+      "loss": 1.5889,
+      "step": 6142
+    },
+    {
+      "epoch": 0.5686975640803101,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.5465,
+      "step": 6143
+    },
+    {
+      "epoch": 0.5687901406005902,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.02,
+      "loss": 1.5221,
+      "step": 6144
+    },
+    {
+      "epoch": 0.5688827171208702,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.6692,
+      "step": 6145
+    },
+    {
+      "epoch": 0.5689752936411503,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.02,
+      "loss": 1.5939,
+      "step": 6146
+    },
+    {
+      "epoch": 0.5690678701614303,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.6172,
+      "step": 6147
+    },
+    {
+      "epoch": 0.5691604466817104,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5864,
+      "step": 6148
+    },
+    {
+      "epoch": 0.5692530232019904,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.02,
+      "loss": 1.4831,
+      "step": 6149
+    },
+    {
+      "epoch": 0.5693455997222704,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5976,
+      "step": 6150
+    },
+    {
+      "epoch": 0.5694381762425504,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.02,
+      "loss": 1.5434,
+      "step": 6151
+    },
+    {
+      "epoch": 0.5695307527628305,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.02,
+      "loss": 1.4685,
+      "step": 6152
+    },
+    {
+      "epoch": 0.5696233292831105,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5528,
+      "step": 6153
+    },
+    {
+      "epoch": 0.5697159058033906,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5527,
+      "step": 6154
+    },
+    {
+      "epoch": 0.5698084823236707,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.656,
+      "step": 6155
+    },
+    {
+      "epoch": 0.5699010588439507,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5508,
+      "step": 6156
+    },
+    {
+      "epoch": 0.5699936353642308,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5532,
+      "step": 6157
+    },
+    {
+      "epoch": 0.5700862118845108,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5807,
+      "step": 6158
+    },
+    {
+      "epoch": 0.5701787884047909,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5246,
+      "step": 6159
+    },
+    {
+      "epoch": 0.5702713649250709,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5369,
+      "step": 6160
+    },
+    {
+      "epoch": 0.570363941445351,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5218,
+      "step": 6161
+    },
+    {
+      "epoch": 0.570456517965631,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.02,
+      "loss": 1.535,
+      "step": 6162
+    },
+    {
+      "epoch": 0.570549094485911,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5569,
+      "step": 6163
+    },
+    {
+      "epoch": 0.570641671006191,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5727,
+      "step": 6164
+    },
+    {
+      "epoch": 0.5707342475264711,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5428,
+      "step": 6165
+    },
+    {
+      "epoch": 0.5708268240467511,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5897,
+      "step": 6166
+    },
+    {
+      "epoch": 0.5709194005670312,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5386,
+      "step": 6167
+    },
+    {
+      "epoch": 0.5710119770873112,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.5456,
+      "step": 6168
+    },
+    {
+      "epoch": 0.5711045536075913,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.54,
+      "step": 6169
+    },
+    {
+      "epoch": 0.5711971301278713,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.02,
+      "loss": 1.4929,
+      "step": 6170
+    },
+    {
+      "epoch": 0.5712897066481514,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.02,
+      "loss": 1.5632,
+      "step": 6171
+    },
+    {
+      "epoch": 0.5713822831684314,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.6497,
+      "step": 6172
+    },
+    {
+      "epoch": 0.5714748596887115,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.02,
+      "loss": 1.5095,
+      "step": 6173
+    },
+    {
+      "epoch": 0.5715674362089915,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.5955,
+      "step": 6174
+    },
+    {
+      "epoch": 0.5716600127292716,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.577,
+      "step": 6175
+    },
+    {
+      "epoch": 0.5717525892495516,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5497,
+      "step": 6176
+    },
+    {
+      "epoch": 0.5718451657698316,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5829,
+      "step": 6177
+    },
+    {
+      "epoch": 0.5719377422901116,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.6288,
+      "step": 6178
+    },
+    {
+      "epoch": 0.5720303188103917,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.527,
+      "step": 6179
+    },
+    {
+      "epoch": 0.5721228953306717,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5059,
+      "step": 6180
+    },
+    {
+      "epoch": 0.5722154718509518,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5368,
+      "step": 6181
+    },
+    {
+      "epoch": 0.5723080483712318,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.02,
+      "loss": 1.6413,
+      "step": 6182
+    },
+    {
+      "epoch": 0.5724006248915119,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5558,
+      "step": 6183
+    },
+    {
+      "epoch": 0.5724932014117919,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.4733,
+      "step": 6184
+    },
+    {
+      "epoch": 0.572585777932072,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5474,
+      "step": 6185
+    },
+    {
+      "epoch": 0.572678354452352,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5813,
+      "step": 6186
+    },
+    {
+      "epoch": 0.5727709309726321,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.02,
+      "loss": 1.6214,
+      "step": 6187
+    },
+    {
+      "epoch": 0.5728635074929121,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.5323,
+      "step": 6188
+    },
+    {
+      "epoch": 0.5729560840131922,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5501,
+      "step": 6189
+    },
+    {
+      "epoch": 0.5730486605334721,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.02,
+      "loss": 1.5082,
+      "step": 6190
+    },
+    {
+      "epoch": 0.5731412370537522,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5008,
+      "step": 6191
+    },
+    {
+      "epoch": 0.5732338135740322,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.02,
+      "loss": 1.5606,
+      "step": 6192
+    },
+    {
+      "epoch": 0.5733263900943123,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5407,
+      "step": 6193
+    },
+    {
+      "epoch": 0.5734189666145924,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5331,
+      "step": 6194
+    },
+    {
+      "epoch": 0.5735115431348724,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.4957,
+      "step": 6195
+    },
+    {
+      "epoch": 0.5736041196551525,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5677,
+      "step": 6196
+    },
+    {
+      "epoch": 0.5736966961754325,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.6088,
+      "step": 6197
+    },
+    {
+      "epoch": 0.5737892726957126,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5389,
+      "step": 6198
+    },
+    {
+      "epoch": 0.5738818492159926,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5088,
+      "step": 6199
+    },
+    {
+      "epoch": 0.5739744257362727,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5406,
+      "step": 6200
+    },
+    {
+      "epoch": 0.5740670022565527,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.4772,
+      "step": 6201
+    },
+    {
+      "epoch": 0.5741595787768328,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5772,
+      "step": 6202
+    },
+    {
+      "epoch": 0.5742521552971128,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.558,
+      "step": 6203
+    },
+    {
+      "epoch": 0.5743447318173928,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.02,
+      "loss": 1.5736,
+      "step": 6204
+    },
+    {
+      "epoch": 0.5744373083376728,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.4973,
+      "step": 6205
+    },
+    {
+      "epoch": 0.5745298848579529,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5749,
+      "step": 6206
+    },
+    {
+      "epoch": 0.5746224613782329,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.6057,
+      "step": 6207
+    },
+    {
+      "epoch": 0.574715037898513,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.6144,
+      "step": 6208
+    },
+    {
+      "epoch": 0.574807614418793,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5222,
+      "step": 6209
+    },
+    {
+      "epoch": 0.5749001909390731,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5857,
+      "step": 6210
+    },
+    {
+      "epoch": 0.5749927674593531,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.02,
+      "loss": 1.6274,
+      "step": 6211
+    },
+    {
+      "epoch": 0.5750853439796332,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.02,
+      "loss": 1.5724,
+      "step": 6212
+    },
+    {
+      "epoch": 0.5751779204999132,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5634,
+      "step": 6213
+    },
+    {
+      "epoch": 0.5752704970201933,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5589,
+      "step": 6214
+    },
+    {
+      "epoch": 0.5753630735404733,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5267,
+      "step": 6215
+    },
+    {
+      "epoch": 0.5754556500607534,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.58,
+      "step": 6216
+    },
+    {
+      "epoch": 0.5755482265810334,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5863,
+      "step": 6217
+    },
+    {
+      "epoch": 0.5756408031013134,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5335,
+      "step": 6218
+    },
+    {
+      "epoch": 0.5757333796215934,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5579,
+      "step": 6219
+    },
+    {
+      "epoch": 0.5758259561418735,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.02,
+      "loss": 1.5358,
+      "step": 6220
+    },
+    {
+      "epoch": 0.5759185326621535,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5837,
+      "step": 6221
+    },
+    {
+      "epoch": 0.5760111091824336,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5617,
+      "step": 6222
+    },
+    {
+      "epoch": 0.5761036857027136,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.6017,
+      "step": 6223
+    },
+    {
+      "epoch": 0.5761962622229937,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.02,
+      "loss": 1.5398,
+      "step": 6224
+    },
+    {
+      "epoch": 0.5762888387432737,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5637,
+      "step": 6225
+    },
+    {
+      "epoch": 0.5763814152635538,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5417,
+      "step": 6226
+    },
+    {
+      "epoch": 0.5764739917838339,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5244,
+      "step": 6227
+    },
+    {
+      "epoch": 0.5765665683041139,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.634,
+      "step": 6228
+    },
+    {
+      "epoch": 0.576659144824394,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.5735,
+      "step": 6229
+    },
+    {
+      "epoch": 0.576751721344674,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.02,
+      "loss": 1.5315,
+      "step": 6230
+    },
+    {
+      "epoch": 0.576844297864954,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5709,
+      "step": 6231
+    },
+    {
+      "epoch": 0.576936874385234,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.627,
+      "step": 6232
+    },
+    {
+      "epoch": 0.577029450905514,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.5224,
+      "step": 6233
+    },
+    {
+      "epoch": 0.5771220274257941,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.02,
+      "loss": 1.527,
+      "step": 6234
+    },
+    {
+      "epoch": 0.5772146039460742,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.593,
+      "step": 6235
+    },
+    {
+      "epoch": 0.5773071804663542,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.5215,
+      "step": 6236
+    },
+    {
+      "epoch": 0.5773997569866343,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.489,
+      "step": 6237
+    },
+    {
+      "epoch": 0.5774923335069143,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.5513,
+      "step": 6238
+    },
+    {
+      "epoch": 0.5775849100271944,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.5078,
+      "step": 6239
+    },
+    {
+      "epoch": 0.5776774865474744,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.4865,
+      "step": 6240
+    },
+    {
+      "epoch": 0.5777700630677545,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.529,
+      "step": 6241
+    },
+    {
+      "epoch": 0.5778626395880345,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.02,
+      "loss": 1.5794,
+      "step": 6242
+    },
+    {
+      "epoch": 0.5779552161083146,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5606,
+      "step": 6243
+    },
+    {
+      "epoch": 0.5780477926285946,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.5319,
+      "step": 6244
+    },
+    {
+      "epoch": 0.5781403691488746,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.02,
+      "loss": 1.5463,
+      "step": 6245
+    },
+    {
+      "epoch": 0.5782329456691546,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.567,
+      "step": 6246
+    },
+    {
+      "epoch": 0.5783255221894347,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5152,
+      "step": 6247
+    },
+    {
+      "epoch": 0.5784180987097147,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5863,
+      "step": 6248
+    },
+    {
+      "epoch": 0.5785106752299948,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5728,
+      "step": 6249
+    },
+    {
+      "epoch": 0.5786032517502748,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.5268,
+      "step": 6250
+    },
+    {
+      "epoch": 0.5786958282705549,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.5605,
+      "step": 6251
+    },
+    {
+      "epoch": 0.5787884047908349,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.5085,
+      "step": 6252
+    },
+    {
+      "epoch": 0.578880981311115,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.5225,
+      "step": 6253
+    },
+    {
+      "epoch": 0.578973557831395,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.02,
+      "loss": 1.5207,
+      "step": 6254
+    },
+    {
+      "epoch": 0.5790661343516751,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.02,
+      "loss": 1.5876,
+      "step": 6255
+    },
+    {
+      "epoch": 0.5791587108719551,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.6537,
+      "step": 6256
+    },
+    {
+      "epoch": 0.5792512873922352,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.5429,
+      "step": 6257
+    },
+    {
+      "epoch": 0.5793438639125151,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5922,
+      "step": 6258
+    },
+    {
+      "epoch": 0.5794364404327952,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5318,
+      "step": 6259
+    },
+    {
+      "epoch": 0.5795290169530752,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5117,
+      "step": 6260
+    },
+    {
+      "epoch": 0.5796215934733553,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.02,
+      "loss": 1.5841,
+      "step": 6261
+    },
+    {
+      "epoch": 0.5797141699936353,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.02,
+      "loss": 1.541,
+      "step": 6262
+    },
+    {
+      "epoch": 0.5798067465139154,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.02,
+      "loss": 1.5798,
+      "step": 6263
+    },
+    {
+      "epoch": 0.5798993230341954,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.02,
+      "loss": 1.551,
+      "step": 6264
+    },
+    {
+      "epoch": 0.5799918995544755,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.537,
+      "step": 6265
+    },
+    {
+      "epoch": 0.5800844760747556,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5439,
+      "step": 6266
+    },
+    {
+      "epoch": 0.5801770525950356,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.4923,
+      "step": 6267
+    },
+    {
+      "epoch": 0.5802696291153157,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.02,
+      "loss": 1.4585,
+      "step": 6268
+    },
+    {
+      "epoch": 0.5803622056355957,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5752,
+      "step": 6269
+    },
+    {
+      "epoch": 0.5804547821558758,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.02,
+      "loss": 1.5792,
+      "step": 6270
+    },
+    {
+      "epoch": 0.5805473586761558,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.5695,
+      "step": 6271
+    },
+    {
+      "epoch": 0.5806399351964358,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.02,
+      "loss": 1.538,
+      "step": 6272
+    },
+    {
+      "epoch": 0.5807325117167158,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5323,
+      "step": 6273
+    },
+    {
+      "epoch": 0.5808250882369959,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.02,
+      "loss": 1.5639,
+      "step": 6274
+    },
+    {
+      "epoch": 0.5809176647572759,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5445,
+      "step": 6275
+    },
+    {
+      "epoch": 0.581010241277556,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5044,
+      "step": 6276
+    },
+    {
+      "epoch": 0.581102817797836,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.4582,
+      "step": 6277
+    },
+    {
+      "epoch": 0.5811953943181161,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5625,
+      "step": 6278
+    },
+    {
+      "epoch": 0.5812879708383961,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.02,
+      "loss": 1.553,
+      "step": 6279
+    },
+    {
+      "epoch": 0.5813805473586762,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5976,
+      "step": 6280
+    },
+    {
+      "epoch": 0.5814731238789562,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5379,
+      "step": 6281
+    },
+    {
+      "epoch": 0.5815657003992363,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5477,
+      "step": 6282
+    },
+    {
+      "epoch": 0.5816582769195163,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.02,
+      "loss": 1.5659,
+      "step": 6283
+    },
+    {
+      "epoch": 0.5817508534397964,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.4831,
+      "step": 6284
+    },
+    {
+      "epoch": 0.5818434299600764,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5283,
+      "step": 6285
+    },
+    {
+      "epoch": 0.5819360064803564,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.518,
+      "step": 6286
+    },
+    {
+      "epoch": 0.5820285830006364,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5323,
+      "step": 6287
+    },
+    {
+      "epoch": 0.5821211595209165,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.5101,
+      "step": 6288
+    },
+    {
+      "epoch": 0.5822137360411965,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5626,
+      "step": 6289
+    },
+    {
+      "epoch": 0.5823063125614766,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.548,
+      "step": 6290
+    },
+    {
+      "epoch": 0.5823988890817566,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.4608,
+      "step": 6291
+    },
+    {
+      "epoch": 0.5824914656020367,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5382,
+      "step": 6292
+    },
+    {
+      "epoch": 0.5825840421223167,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5461,
+      "step": 6293
+    },
+    {
+      "epoch": 0.5826766186425968,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5345,
+      "step": 6294
+    },
+    {
+      "epoch": 0.5827691951628768,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5537,
+      "step": 6295
+    },
+    {
+      "epoch": 0.5828617716831569,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5838,
+      "step": 6296
+    },
+    {
+      "epoch": 0.582954348203437,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.02,
+      "loss": 1.5383,
+      "step": 6297
+    },
+    {
+      "epoch": 0.583046924723717,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5583,
+      "step": 6298
+    },
+    {
+      "epoch": 0.5831395012439969,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5919,
+      "step": 6299
+    },
+    {
+      "epoch": 0.583232077764277,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.5756,
+      "step": 6300
+    },
+    {
+      "epoch": 0.583324654284557,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.591,
+      "step": 6301
+    },
+    {
+      "epoch": 0.5834172308048371,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.02,
+      "loss": 1.4678,
+      "step": 6302
+    },
+    {
+      "epoch": 0.5835098073251171,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5689,
+      "step": 6303
+    },
+    {
+      "epoch": 0.5836023838453972,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5426,
+      "step": 6304
+    },
+    {
+      "epoch": 0.5836949603656773,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5045,
+      "step": 6305
+    },
+    {
+      "epoch": 0.5837875368859573,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.02,
+      "loss": 1.5852,
+      "step": 6306
+    },
+    {
+      "epoch": 0.5838801134062374,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.02,
+      "loss": 1.5393,
+      "step": 6307
+    },
+    {
+      "epoch": 0.5839726899265174,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5541,
+      "step": 6308
+    },
+    {
+      "epoch": 0.5840652664467975,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5501,
+      "step": 6309
+    },
+    {
+      "epoch": 0.5841578429670775,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5962,
+      "step": 6310
+    },
+    {
+      "epoch": 0.5842504194873576,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.5288,
+      "step": 6311
+    },
+    {
+      "epoch": 0.5843429960076376,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.4677,
+      "step": 6312
+    },
+    {
+      "epoch": 0.5844355725279176,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5353,
+      "step": 6313
+    },
+    {
+      "epoch": 0.5845281490481976,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.4701,
+      "step": 6314
+    },
+    {
+      "epoch": 0.5846207255684777,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.5017,
+      "step": 6315
+    },
+    {
+      "epoch": 0.5847133020887577,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.4944,
+      "step": 6316
+    },
+    {
+      "epoch": 0.5848058786090378,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5353,
+      "step": 6317
+    },
+    {
+      "epoch": 0.5848984551293178,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.02,
+      "loss": 1.6036,
+      "step": 6318
+    },
+    {
+      "epoch": 0.5849910316495979,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5741,
+      "step": 6319
+    },
+    {
+      "epoch": 0.5850836081698779,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5497,
+      "step": 6320
+    },
+    {
+      "epoch": 0.585176184690158,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5268,
+      "step": 6321
+    },
+    {
+      "epoch": 0.585268761210438,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5545,
+      "step": 6322
+    },
+    {
+      "epoch": 0.5853613377307181,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.6326,
+      "step": 6323
+    },
+    {
+      "epoch": 0.5854539142509981,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.572,
+      "step": 6324
+    },
+    {
+      "epoch": 0.5855464907712782,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.553,
+      "step": 6325
+    },
+    {
+      "epoch": 0.5856390672915581,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.5608,
+      "step": 6326
+    },
+    {
+      "epoch": 0.5857316438118382,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.02,
+      "loss": 1.528,
+      "step": 6327
+    },
+    {
+      "epoch": 0.5858242203321182,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.02,
+      "loss": 1.5681,
+      "step": 6328
+    },
+    {
+      "epoch": 0.5859167968523983,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.5855,
+      "step": 6329
+    },
+    {
+      "epoch": 0.5860093733726783,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.5622,
+      "step": 6330
+    },
+    {
+      "epoch": 0.5861019498929584,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.02,
+      "loss": 1.5444,
+      "step": 6331
+    },
+    {
+      "epoch": 0.5861945264132384,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5229,
+      "step": 6332
+    },
+    {
+      "epoch": 0.5862871029335185,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.02,
+      "loss": 1.5173,
+      "step": 6333
+    },
+    {
+      "epoch": 0.5863796794537985,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5245,
+      "step": 6334
+    },
+    {
+      "epoch": 0.5864722559740786,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.4832,
+      "step": 6335
+    },
+    {
+      "epoch": 0.5865648324943586,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.543,
+      "step": 6336
+    },
+    {
+      "epoch": 0.5866574090146387,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.5048,
+      "step": 6337
+    },
+    {
+      "epoch": 0.5867499855349187,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.02,
+      "loss": 1.5277,
+      "step": 6338
+    },
+    {
+      "epoch": 0.5868425620551988,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5372,
+      "step": 6339
+    },
+    {
+      "epoch": 0.5869351385754787,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5939,
+      "step": 6340
+    },
+    {
+      "epoch": 0.5870277150957588,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5363,
+      "step": 6341
+    },
+    {
+      "epoch": 0.5871202916160388,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5968,
+      "step": 6342
+    },
+    {
+      "epoch": 0.5872128681363189,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.02,
+      "loss": 1.5293,
+      "step": 6343
+    },
+    {
+      "epoch": 0.587305444656599,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5396,
+      "step": 6344
+    },
+    {
+      "epoch": 0.587398021176879,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.02,
+      "loss": 1.5327,
+      "step": 6345
+    },
+    {
+      "epoch": 0.5874905976971591,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5035,
+      "step": 6346
+    },
+    {
+      "epoch": 0.5875831742174391,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5403,
+      "step": 6347
+    },
+    {
+      "epoch": 0.5876757507377192,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5447,
+      "step": 6348
+    },
+    {
+      "epoch": 0.5877683272579992,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5663,
+      "step": 6349
+    },
+    {
+      "epoch": 0.5878609037782793,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5374,
+      "step": 6350
+    },
+    {
+      "epoch": 0.5879534802985593,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.587,
+      "step": 6351
+    },
+    {
+      "epoch": 0.5880460568188394,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5679,
+      "step": 6352
+    },
+    {
+      "epoch": 0.5881386333391194,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.02,
+      "loss": 1.5836,
+      "step": 6353
+    },
+    {
+      "epoch": 0.5882312098593994,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.6134,
+      "step": 6354
+    },
+    {
+      "epoch": 0.5883237863796794,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.4913,
+      "step": 6355
+    },
+    {
+      "epoch": 0.5884163628999595,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5116,
+      "step": 6356
+    },
+    {
+      "epoch": 0.5885089394202395,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.4725,
+      "step": 6357
+    },
+    {
+      "epoch": 0.5886015159405196,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5865,
+      "step": 6358
+    },
+    {
+      "epoch": 0.5886940924607996,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.4726,
+      "step": 6359
+    },
+    {
+      "epoch": 0.5887866689810797,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.6126,
+      "step": 6360
+    },
+    {
+      "epoch": 0.5888792455013597,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5546,
+      "step": 6361
+    },
+    {
+      "epoch": 0.5889718220216398,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5451,
+      "step": 6362
+    },
+    {
+      "epoch": 0.5890643985419198,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5333,
+      "step": 6363
+    },
+    {
+      "epoch": 0.5891569750621999,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5465,
+      "step": 6364
+    },
+    {
+      "epoch": 0.5892495515824799,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5933,
+      "step": 6365
+    },
+    {
+      "epoch": 0.58934212810276,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.4986,
+      "step": 6366
+    },
+    {
+      "epoch": 0.5894347046230399,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.02,
+      "loss": 1.5465,
+      "step": 6367
+    },
+    {
+      "epoch": 0.58952728114332,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5526,
+      "step": 6368
+    },
+    {
+      "epoch": 0.5896198576636,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5368,
+      "step": 6369
+    },
+    {
+      "epoch": 0.5897124341838801,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5293,
+      "step": 6370
+    },
+    {
+      "epoch": 0.5898050107041601,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5985,
+      "step": 6371
+    },
+    {
+      "epoch": 0.5898975872244402,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5128,
+      "step": 6372
+    },
+    {
+      "epoch": 0.5899901637447202,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5541,
+      "step": 6373
+    },
+    {
+      "epoch": 0.5900827402650003,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.5625,
+      "step": 6374
+    },
+    {
+      "epoch": 0.5901753167852803,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5565,
+      "step": 6375
+    },
+    {
+      "epoch": 0.5902678933055604,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.561,
+      "step": 6376
+    },
+    {
+      "epoch": 0.5903604698258405,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5212,
+      "step": 6377
+    },
+    {
+      "epoch": 0.5904530463461205,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.02,
+      "loss": 1.5104,
+      "step": 6378
+    },
+    {
+      "epoch": 0.5905456228664006,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5835,
+      "step": 6379
+    },
+    {
+      "epoch": 0.5906381993866806,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5764,
+      "step": 6380
+    },
+    {
+      "epoch": 0.5907307759069605,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.02,
+      "loss": 1.5509,
+      "step": 6381
+    },
+    {
+      "epoch": 0.5908233524272406,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5665,
+      "step": 6382
+    },
+    {
+      "epoch": 0.5909159289475207,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.4999,
+      "step": 6383
+    },
+    {
+      "epoch": 0.5910085054678007,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5412,
+      "step": 6384
+    },
+    {
+      "epoch": 0.5911010819880808,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5162,
+      "step": 6385
+    },
+    {
+      "epoch": 0.5911936585083608,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.4715,
+      "step": 6386
+    },
+    {
+      "epoch": 0.5912862350286409,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.5593,
+      "step": 6387
+    },
+    {
+      "epoch": 0.5913788115489209,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.5979,
+      "step": 6388
+    },
+    {
+      "epoch": 0.591471388069201,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5641,
+      "step": 6389
+    },
+    {
+      "epoch": 0.591563964589481,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.5768,
+      "step": 6390
+    },
+    {
+      "epoch": 0.5916565411097611,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5787,
+      "step": 6391
+    },
+    {
+      "epoch": 0.5917491176300411,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.543,
+      "step": 6392
+    },
+    {
+      "epoch": 0.5918416941503212,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.02,
+      "loss": 1.5481,
+      "step": 6393
+    },
+    {
+      "epoch": 0.5919342706706011,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5603,
+      "step": 6394
+    },
+    {
+      "epoch": 0.5920268471908812,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.02,
+      "loss": 1.5628,
+      "step": 6395
+    },
+    {
+      "epoch": 0.5921194237111612,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.5432,
+      "step": 6396
+    },
+    {
+      "epoch": 0.5922120002314413,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5468,
+      "step": 6397
+    },
+    {
+      "epoch": 0.5923045767517213,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5261,
+      "step": 6398
+    },
+    {
+      "epoch": 0.5923971532720014,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5705,
+      "step": 6399
+    },
+    {
+      "epoch": 0.5924897297922814,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5513,
+      "step": 6400
+    },
+    {
+      "epoch": 0.5925823063125615,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5644,
+      "step": 6401
+    },
+    {
+      "epoch": 0.5926748828328415,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5196,
+      "step": 6402
+    },
+    {
+      "epoch": 0.5927674593531216,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5728,
+      "step": 6403
+    },
+    {
+      "epoch": 0.5928600358734016,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.5213,
+      "step": 6404
+    },
+    {
+      "epoch": 0.5929526123936817,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5383,
+      "step": 6405
+    },
+    {
+      "epoch": 0.5930451889139617,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.02,
+      "loss": 1.5566,
+      "step": 6406
+    },
+    {
+      "epoch": 0.5931377654342418,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.02,
+      "loss": 1.5459,
+      "step": 6407
+    },
+    {
+      "epoch": 0.5932303419545217,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.4937,
+      "step": 6408
+    },
+    {
+      "epoch": 0.5933229184748018,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.02,
+      "loss": 1.6202,
+      "step": 6409
+    },
+    {
+      "epoch": 0.5934154949950818,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5716,
+      "step": 6410
+    },
+    {
+      "epoch": 0.5935080715153619,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5845,
+      "step": 6411
+    },
+    {
+      "epoch": 0.5936006480356419,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5689,
+      "step": 6412
+    },
+    {
+      "epoch": 0.593693224555922,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.02,
+      "loss": 1.5358,
+      "step": 6413
+    },
+    {
+      "epoch": 0.593785801076202,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.02,
+      "loss": 1.6157,
+      "step": 6414
+    },
+    {
+      "epoch": 0.5938783775964821,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5641,
+      "step": 6415
+    },
+    {
+      "epoch": 0.5939709541167622,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.6275,
+      "step": 6416
+    },
+    {
+      "epoch": 0.5940635306370422,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.02,
+      "loss": 1.5707,
+      "step": 6417
+    },
+    {
+      "epoch": 0.5941561071573223,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5931,
+      "step": 6418
+    },
+    {
+      "epoch": 0.5942486836776023,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5819,
+      "step": 6419
+    },
+    {
+      "epoch": 0.5943412601978824,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.02,
+      "loss": 1.5657,
+      "step": 6420
+    },
+    {
+      "epoch": 0.5944338367181623,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5176,
+      "step": 6421
+    },
+    {
+      "epoch": 0.5945264132384424,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5463,
+      "step": 6422
+    },
+    {
+      "epoch": 0.5946189897587224,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5355,
+      "step": 6423
+    },
+    {
+      "epoch": 0.5947115662790025,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.6074,
+      "step": 6424
+    },
+    {
+      "epoch": 0.5948041427992825,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5958,
+      "step": 6425
+    },
+    {
+      "epoch": 0.5948967193195626,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5317,
+      "step": 6426
+    },
+    {
+      "epoch": 0.5949892958398426,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.02,
+      "loss": 1.5086,
+      "step": 6427
+    },
+    {
+      "epoch": 0.5950818723601227,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.5121,
+      "step": 6428
+    },
+    {
+      "epoch": 0.5951744488804027,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5949,
+      "step": 6429
+    },
+    {
+      "epoch": 0.5952670254006828,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5292,
+      "step": 6430
+    },
+    {
+      "epoch": 0.5953596019209628,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.6059,
+      "step": 6431
+    },
+    {
+      "epoch": 0.5954521784412429,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.02,
+      "loss": 1.5515,
+      "step": 6432
+    },
+    {
+      "epoch": 0.5955447549615229,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.502,
+      "step": 6433
+    },
+    {
+      "epoch": 0.595637331481803,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.5245,
+      "step": 6434
+    },
+    {
+      "epoch": 0.5957299080020829,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.6093,
+      "step": 6435
+    },
+    {
+      "epoch": 0.595822484522363,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5204,
+      "step": 6436
+    },
+    {
+      "epoch": 0.595915061042643,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.02,
+      "loss": 1.5549,
+      "step": 6437
+    },
+    {
+      "epoch": 0.5960076375629231,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.59,
+      "step": 6438
+    },
+    {
+      "epoch": 0.5961002140832031,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5688,
+      "step": 6439
+    },
+    {
+      "epoch": 0.5961927906034832,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.02,
+      "loss": 1.5686,
+      "step": 6440
+    },
+    {
+      "epoch": 0.5962853671237632,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.02,
+      "loss": 1.5391,
+      "step": 6441
+    },
+    {
+      "epoch": 0.5963779436440433,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.02,
+      "loss": 1.5544,
+      "step": 6442
+    },
+    {
+      "epoch": 0.5964705201643233,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.02,
+      "loss": 1.5486,
+      "step": 6443
+    },
+    {
+      "epoch": 0.5965630966846034,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5564,
+      "step": 6444
+    },
+    {
+      "epoch": 0.5966556732048834,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5502,
+      "step": 6445
+    },
+    {
+      "epoch": 0.5967482497251635,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5162,
+      "step": 6446
+    },
+    {
+      "epoch": 0.5968408262454435,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5785,
+      "step": 6447
+    },
+    {
+      "epoch": 0.5969334027657236,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5395,
+      "step": 6448
+    },
+    {
+      "epoch": 0.5970259792860035,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5465,
+      "step": 6449
+    },
+    {
+      "epoch": 0.5971185558062836,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5495,
+      "step": 6450
+    },
+    {
+      "epoch": 0.5972111323265636,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5932,
+      "step": 6451
+    },
+    {
+      "epoch": 0.5973037088468437,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5477,
+      "step": 6452
+    },
+    {
+      "epoch": 0.5973962853671237,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.5092,
+      "step": 6453
+    },
+    {
+      "epoch": 0.5974888618874038,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.02,
+      "loss": 1.5084,
+      "step": 6454
+    },
+    {
+      "epoch": 0.5975814384076839,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5462,
+      "step": 6455
+    },
+    {
+      "epoch": 0.5976740149279639,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.54,
+      "step": 6456
+    },
+    {
+      "epoch": 0.597766591448244,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5101,
+      "step": 6457
+    },
+    {
+      "epoch": 0.597859167968524,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5885,
+      "step": 6458
+    },
+    {
+      "epoch": 0.5979517444888041,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5022,
+      "step": 6459
+    },
+    {
+      "epoch": 0.5980443210090841,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.02,
+      "loss": 1.6019,
+      "step": 6460
+    },
+    {
+      "epoch": 0.5981368975293642,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.6054,
+      "step": 6461
+    },
+    {
+      "epoch": 0.5982294740496441,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5466,
+      "step": 6462
+    },
+    {
+      "epoch": 0.5983220505699242,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.4979,
+      "step": 6463
+    },
+    {
+      "epoch": 0.5984146270902042,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.4881,
+      "step": 6464
+    },
+    {
+      "epoch": 0.5985072036104843,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.5608,
+      "step": 6465
+    },
+    {
+      "epoch": 0.5985997801307643,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5846,
+      "step": 6466
+    },
+    {
+      "epoch": 0.5986923566510444,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.02,
+      "loss": 1.535,
+      "step": 6467
+    },
+    {
+      "epoch": 0.5987849331713244,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.4723,
+      "step": 6468
+    },
+    {
+      "epoch": 0.5988775096916045,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.6076,
+      "step": 6469
+    },
+    {
+      "epoch": 0.5989700862118845,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5674,
+      "step": 6470
+    },
+    {
+      "epoch": 0.5990626627321646,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.535,
+      "step": 6471
+    },
+    {
+      "epoch": 0.5991552392524446,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.6081,
+      "step": 6472
+    },
+    {
+      "epoch": 0.5992478157727247,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.4759,
+      "step": 6473
+    },
+    {
+      "epoch": 0.5993403922930047,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5287,
+      "step": 6474
+    },
+    {
+      "epoch": 0.5994329688132848,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.471,
+      "step": 6475
+    },
+    {
+      "epoch": 0.5995255453335647,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5687,
+      "step": 6476
+    },
+    {
+      "epoch": 0.5996181218538448,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.4771,
+      "step": 6477
+    },
+    {
+      "epoch": 0.5997106983741248,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5557,
+      "step": 6478
+    },
+    {
+      "epoch": 0.5998032748944049,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5703,
+      "step": 6479
+    },
+    {
+      "epoch": 0.5998958514146849,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5532,
+      "step": 6480
+    },
+    {
+      "epoch": 0.599988427934965,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.552,
+      "step": 6481
+    },
+    {
+      "epoch": 0.600081004455245,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5465,
+      "step": 6482
+    },
+    {
+      "epoch": 0.6001735809755251,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5658,
+      "step": 6483
+    },
+    {
+      "epoch": 0.6002661574958051,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.5777,
+      "step": 6484
+    },
+    {
+      "epoch": 0.6003587340160852,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.02,
+      "loss": 1.5446,
+      "step": 6485
+    },
+    {
+      "epoch": 0.6004513105363652,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.5848,
+      "step": 6486
+    },
+    {
+      "epoch": 0.6005438870566453,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5498,
+      "step": 6487
+    },
+    {
+      "epoch": 0.6006364635769254,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5528,
+      "step": 6488
+    },
+    {
+      "epoch": 0.6007290400972053,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.5257,
+      "step": 6489
+    },
+    {
+      "epoch": 0.6008216166174853,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5368,
+      "step": 6490
+    },
+    {
+      "epoch": 0.6009141931377654,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.4677,
+      "step": 6491
+    },
+    {
+      "epoch": 0.6010067696580454,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5882,
+      "step": 6492
+    },
+    {
+      "epoch": 0.6010993461783255,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.5558,
+      "step": 6493
+    },
+    {
+      "epoch": 0.6011919226986056,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.4939,
+      "step": 6494
+    },
+    {
+      "epoch": 0.6012844992188856,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.02,
+      "loss": 1.5687,
+      "step": 6495
+    },
+    {
+      "epoch": 0.6013770757391657,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.5171,
+      "step": 6496
+    },
+    {
+      "epoch": 0.6014696522594457,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.4823,
+      "step": 6497
+    },
+    {
+      "epoch": 0.6015622287797258,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.02,
+      "loss": 1.6103,
+      "step": 6498
+    },
+    {
+      "epoch": 0.6016548053000058,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5202,
+      "step": 6499
+    },
+    {
+      "epoch": 0.6017473818202859,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.02,
+      "loss": 1.519,
+      "step": 6500
+    },
+    {
+      "epoch": 0.6018399583405659,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.02,
+      "loss": 1.5342,
+      "step": 6501
+    },
+    {
+      "epoch": 0.601932534860846,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.515,
+      "step": 6502
+    },
+    {
+      "epoch": 0.6020251113811259,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5204,
+      "step": 6503
+    },
+    {
+      "epoch": 0.602117687901406,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5829,
+      "step": 6504
+    },
+    {
+      "epoch": 0.602210264421686,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.582,
+      "step": 6505
+    },
+    {
+      "epoch": 0.6023028409419661,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5545,
+      "step": 6506
+    },
+    {
+      "epoch": 0.6023954174622461,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.6172,
+      "step": 6507
+    },
+    {
+      "epoch": 0.6024879939825262,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5134,
+      "step": 6508
+    },
+    {
+      "epoch": 0.6025805705028062,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5258,
+      "step": 6509
+    },
+    {
+      "epoch": 0.6026731470230863,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5799,
+      "step": 6510
+    },
+    {
+      "epoch": 0.6027657235433663,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5681,
+      "step": 6511
+    },
+    {
+      "epoch": 0.6028583000636464,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.02,
+      "loss": 1.5795,
+      "step": 6512
+    },
+    {
+      "epoch": 0.6029508765839264,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5378,
+      "step": 6513
+    },
+    {
+      "epoch": 0.6030434531042065,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.02,
+      "loss": 1.5804,
+      "step": 6514
+    },
+    {
+      "epoch": 0.6031360296244865,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.02,
+      "loss": 1.4941,
+      "step": 6515
+    },
+    {
+      "epoch": 0.6032286061447666,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5601,
+      "step": 6516
+    },
+    {
+      "epoch": 0.6033211826650465,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5465,
+      "step": 6517
+    },
+    {
+      "epoch": 0.6034137591853266,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.5613,
+      "step": 6518
+    },
+    {
+      "epoch": 0.6035063357056066,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5093,
+      "step": 6519
+    },
+    {
+      "epoch": 0.6035989122258867,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5122,
+      "step": 6520
+    },
+    {
+      "epoch": 0.6036914887461667,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.6274,
+      "step": 6521
+    },
+    {
+      "epoch": 0.6037840652664468,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.555,
+      "step": 6522
+    },
+    {
+      "epoch": 0.6038766417867268,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5507,
+      "step": 6523
+    },
+    {
+      "epoch": 0.6039692183070069,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5681,
+      "step": 6524
+    },
+    {
+      "epoch": 0.604061794827287,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.5563,
+      "step": 6525
+    },
+    {
+      "epoch": 0.604154371347567,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.02,
+      "loss": 1.5846,
+      "step": 6526
+    },
+    {
+      "epoch": 0.604246947867847,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.5819,
+      "step": 6527
+    },
+    {
+      "epoch": 0.6043395243881271,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5792,
+      "step": 6528
+    },
+    {
+      "epoch": 0.6044321009084072,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5261,
+      "step": 6529
+    },
+    {
+      "epoch": 0.6045246774286871,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.5948,
+      "step": 6530
+    },
+    {
+      "epoch": 0.6046172539489671,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.02,
+      "loss": 1.5082,
+      "step": 6531
+    },
+    {
+      "epoch": 0.6047098304692472,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5443,
+      "step": 6532
+    },
+    {
+      "epoch": 0.6048024069895273,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.4872,
+      "step": 6533
+    },
+    {
+      "epoch": 0.6048949835098073,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.6424,
+      "step": 6534
+    },
+    {
+      "epoch": 0.6049875600300874,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.496,
+      "step": 6535
+    },
+    {
+      "epoch": 0.6050801365503674,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.4944,
+      "step": 6536
+    },
+    {
+      "epoch": 0.6051727130706475,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.507,
+      "step": 6537
+    },
+    {
+      "epoch": 0.6052652895909275,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5267,
+      "step": 6538
+    },
+    {
+      "epoch": 0.6053578661112076,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.02,
+      "loss": 1.5416,
+      "step": 6539
+    },
+    {
+      "epoch": 0.6054504426314876,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5959,
+      "step": 6540
+    },
+    {
+      "epoch": 0.6055430191517677,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5292,
+      "step": 6541
+    },
+    {
+      "epoch": 0.6056355956720477,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.5418,
+      "step": 6542
+    },
+    {
+      "epoch": 0.6057281721923278,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.5231,
+      "step": 6543
+    },
+    {
+      "epoch": 0.6058207487126077,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.5853,
+      "step": 6544
+    },
+    {
+      "epoch": 0.6059133252328878,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.5003,
+      "step": 6545
+    },
+    {
+      "epoch": 0.6060059017531678,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5482,
+      "step": 6546
+    },
+    {
+      "epoch": 0.6060984782734479,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.6247,
+      "step": 6547
+    },
+    {
+      "epoch": 0.6061910547937279,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.4844,
+      "step": 6548
+    },
+    {
+      "epoch": 0.606283631314008,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5107,
+      "step": 6549
+    },
+    {
+      "epoch": 0.606376207834288,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5626,
+      "step": 6550
+    },
+    {
+      "epoch": 0.6064687843545681,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5299,
+      "step": 6551
+    },
+    {
+      "epoch": 0.6065613608748481,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.02,
+      "loss": 1.5385,
+      "step": 6552
+    },
+    {
+      "epoch": 0.6066539373951282,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.4735,
+      "step": 6553
+    },
+    {
+      "epoch": 0.6067465139154082,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.5624,
+      "step": 6554
+    },
+    {
+      "epoch": 0.6068390904356883,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5498,
+      "step": 6555
+    },
+    {
+      "epoch": 0.6069316669559683,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.4612,
+      "step": 6556
+    },
+    {
+      "epoch": 0.6070242434762483,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5409,
+      "step": 6557
+    },
+    {
+      "epoch": 0.6071168199965283,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5862,
+      "step": 6558
+    },
+    {
+      "epoch": 0.6072093965168084,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5758,
+      "step": 6559
+    },
+    {
+      "epoch": 0.6073019730370884,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.557,
+      "step": 6560
+    },
+    {
+      "epoch": 0.6073945495573685,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5748,
+      "step": 6561
+    },
+    {
+      "epoch": 0.6074871260776485,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.02,
+      "loss": 1.565,
+      "step": 6562
+    },
+    {
+      "epoch": 0.6075797025979286,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.02,
+      "loss": 1.5265,
+      "step": 6563
+    },
+    {
+      "epoch": 0.6076722791182086,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.02,
+      "loss": 1.5407,
+      "step": 6564
+    },
+    {
+      "epoch": 0.6077648556384887,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.02,
+      "loss": 1.5846,
+      "step": 6565
+    },
+    {
+      "epoch": 0.6078574321587688,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5542,
+      "step": 6566
+    },
+    {
+      "epoch": 0.6079500086790488,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.02,
+      "loss": 1.5203,
+      "step": 6567
+    },
+    {
+      "epoch": 0.6080425851993289,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.02,
+      "loss": 1.5503,
+      "step": 6568
+    },
+    {
+      "epoch": 0.6081351617196089,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.02,
+      "loss": 1.5956,
+      "step": 6569
+    },
+    {
+      "epoch": 0.608227738239889,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.02,
+      "loss": 1.6191,
+      "step": 6570
+    },
+    {
+      "epoch": 0.6083203147601689,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.5918,
+      "step": 6571
+    },
+    {
+      "epoch": 0.608412891280449,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5376,
+      "step": 6572
+    },
+    {
+      "epoch": 0.608505467800729,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5655,
+      "step": 6573
+    },
+    {
+      "epoch": 0.6085980443210091,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5514,
+      "step": 6574
+    },
+    {
+      "epoch": 0.6086906208412891,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.4902,
+      "step": 6575
+    },
+    {
+      "epoch": 0.6087831973615692,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.4581,
+      "step": 6576
+    },
+    {
+      "epoch": 0.6088757738818492,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5505,
+      "step": 6577
+    },
+    {
+      "epoch": 0.6089683504021293,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.4879,
+      "step": 6578
+    },
+    {
+      "epoch": 0.6090609269224093,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5769,
+      "step": 6579
+    },
+    {
+      "epoch": 0.6091535034426894,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5708,
+      "step": 6580
+    },
+    {
+      "epoch": 0.6092460799629694,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.4736,
+      "step": 6581
+    },
+    {
+      "epoch": 0.6093386564832495,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5186,
+      "step": 6582
+    },
+    {
+      "epoch": 0.6094312330035295,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5759,
+      "step": 6583
+    },
+    {
+      "epoch": 0.6095238095238096,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5327,
+      "step": 6584
+    },
+    {
+      "epoch": 0.6096163860440895,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5683,
+      "step": 6585
+    },
+    {
+      "epoch": 0.6097089625643696,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5488,
+      "step": 6586
+    },
+    {
+      "epoch": 0.6098015390846496,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5267,
+      "step": 6587
+    },
+    {
+      "epoch": 0.6098941156049297,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5775,
+      "step": 6588
+    },
+    {
+      "epoch": 0.6099866921252097,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5455,
+      "step": 6589
+    },
+    {
+      "epoch": 0.6100792686454898,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5672,
+      "step": 6590
+    },
+    {
+      "epoch": 0.6101718451657698,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.5174,
+      "step": 6591
+    },
+    {
+      "epoch": 0.6102644216860499,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5988,
+      "step": 6592
+    },
+    {
+      "epoch": 0.6103569982063299,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.4693,
+      "step": 6593
+    },
+    {
+      "epoch": 0.61044957472661,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.5315,
+      "step": 6594
+    },
+    {
+      "epoch": 0.61054215124689,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.02,
+      "loss": 1.5557,
+      "step": 6595
+    },
+    {
+      "epoch": 0.6106347277671701,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5256,
+      "step": 6596
+    },
+    {
+      "epoch": 0.6107273042874501,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5602,
+      "step": 6597
+    },
+    {
+      "epoch": 0.6108198808077301,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5272,
+      "step": 6598
+    },
+    {
+      "epoch": 0.6109124573280101,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5499,
+      "step": 6599
+    },
+    {
+      "epoch": 0.6110050338482902,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.5372,
+      "step": 6600
+    },
+    {
+      "epoch": 0.6110976103685702,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5684,
+      "step": 6601
+    },
+    {
+      "epoch": 0.6111901868888503,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.4804,
+      "step": 6602
+    },
+    {
+      "epoch": 0.6112827634091303,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.534,
+      "step": 6603
+    },
+    {
+      "epoch": 0.6113753399294104,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5451,
+      "step": 6604
+    },
+    {
+      "epoch": 0.6114679164496905,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.6211,
+      "step": 6605
+    },
+    {
+      "epoch": 0.6115604929699705,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5822,
+      "step": 6606
+    },
+    {
+      "epoch": 0.6116530694902506,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.547,
+      "step": 6607
+    },
+    {
+      "epoch": 0.6117456460105306,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.5704,
+      "step": 6608
+    },
+    {
+      "epoch": 0.6118382225308107,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5664,
+      "step": 6609
+    },
+    {
+      "epoch": 0.6119307990510907,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.02,
+      "loss": 1.5201,
+      "step": 6610
+    },
+    {
+      "epoch": 0.6120233755713708,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.02,
+      "loss": 1.4944,
+      "step": 6611
+    },
+    {
+      "epoch": 0.6121159520916507,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5013,
+      "step": 6612
+    },
+    {
+      "epoch": 0.6122085286119308,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.4498,
+      "step": 6613
+    },
+    {
+      "epoch": 0.6123011051322108,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.5259,
+      "step": 6614
+    },
+    {
+      "epoch": 0.6123936816524909,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.02,
+      "loss": 1.5289,
+      "step": 6615
+    },
+    {
+      "epoch": 0.6124862581727709,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.02,
+      "loss": 1.5128,
+      "step": 6616
+    },
+    {
+      "epoch": 0.612578834693051,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.5374,
+      "step": 6617
+    },
+    {
+      "epoch": 0.612671411213331,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.5595,
+      "step": 6618
+    },
+    {
+      "epoch": 0.6127639877336111,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.02,
+      "loss": 1.4861,
+      "step": 6619
+    },
+    {
+      "epoch": 0.6128565642538911,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.02,
+      "loss": 1.482,
+      "step": 6620
+    },
+    {
+      "epoch": 0.6129491407741712,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5353,
+      "step": 6621
+    },
+    {
+      "epoch": 0.6130417172944512,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.02,
+      "loss": 1.5561,
+      "step": 6622
+    },
+    {
+      "epoch": 0.6131342938147313,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.6124,
+      "step": 6623
+    },
+    {
+      "epoch": 0.6132268703350113,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.02,
+      "loss": 1.5961,
+      "step": 6624
+    },
+    {
+      "epoch": 0.6133194468552913,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.02,
+      "loss": 1.5311,
+      "step": 6625
+    },
+    {
+      "epoch": 0.6134120233755713,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5189,
+      "step": 6626
+    },
+    {
+      "epoch": 0.6135045998958514,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5266,
+      "step": 6627
+    },
+    {
+      "epoch": 0.6135971764161314,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5757,
+      "step": 6628
+    },
+    {
+      "epoch": 0.6136897529364115,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.6004,
+      "step": 6629
+    },
+    {
+      "epoch": 0.6137823294566915,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5896,
+      "step": 6630
+    },
+    {
+      "epoch": 0.6138749059769716,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5633,
+      "step": 6631
+    },
+    {
+      "epoch": 0.6139674824972516,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.4798,
+      "step": 6632
+    },
+    {
+      "epoch": 0.6140600590175317,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.02,
+      "loss": 1.5741,
+      "step": 6633
+    },
+    {
+      "epoch": 0.6141526355378117,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5684,
+      "step": 6634
+    },
+    {
+      "epoch": 0.6142452120580918,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5334,
+      "step": 6635
+    },
+    {
+      "epoch": 0.6143377885783718,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.02,
+      "loss": 1.4836,
+      "step": 6636
+    },
+    {
+      "epoch": 0.6144303650986519,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.57,
+      "step": 6637
+    },
+    {
+      "epoch": 0.614522941618932,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.02,
+      "loss": 1.5456,
+      "step": 6638
+    },
+    {
+      "epoch": 0.6146155181392119,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.528,
+      "step": 6639
+    },
+    {
+      "epoch": 0.6147080946594919,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5032,
+      "step": 6640
+    },
+    {
+      "epoch": 0.614800671179772,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.02,
+      "loss": 1.5767,
+      "step": 6641
+    },
+    {
+      "epoch": 0.614893247700052,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5037,
+      "step": 6642
+    },
+    {
+      "epoch": 0.6149858242203321,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5108,
+      "step": 6643
+    },
+    {
+      "epoch": 0.6150784007406122,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5914,
+      "step": 6644
+    },
+    {
+      "epoch": 0.6151709772608922,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5933,
+      "step": 6645
+    },
+    {
+      "epoch": 0.6152635537811723,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5199,
+      "step": 6646
+    },
+    {
+      "epoch": 0.6153561303014523,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5621,
+      "step": 6647
+    },
+    {
+      "epoch": 0.6154487068217324,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.5604,
+      "step": 6648
+    },
+    {
+      "epoch": 0.6155412833420124,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.4952,
+      "step": 6649
+    },
+    {
+      "epoch": 0.6156338598622925,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5301,
+      "step": 6650
+    },
+    {
+      "epoch": 0.6157264363825725,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5563,
+      "step": 6651
+    },
+    {
+      "epoch": 0.6158190129028526,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5388,
+      "step": 6652
+    },
+    {
+      "epoch": 0.6159115894231325,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.6154,
+      "step": 6653
+    },
+    {
+      "epoch": 0.6160041659434126,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.5262,
+      "step": 6654
+    },
+    {
+      "epoch": 0.6160967424636926,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.02,
+      "loss": 1.5928,
+      "step": 6655
+    },
+    {
+      "epoch": 0.6161893189839727,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.02,
+      "loss": 1.5204,
+      "step": 6656
+    },
+    {
+      "epoch": 0.6162818955042527,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5998,
+      "step": 6657
+    },
+    {
+      "epoch": 0.6163744720245328,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5215,
+      "step": 6658
+    },
+    {
+      "epoch": 0.6164670485448128,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.4974,
+      "step": 6659
+    },
+    {
+      "epoch": 0.6165596250650929,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.02,
+      "loss": 1.5237,
+      "step": 6660
+    },
+    {
+      "epoch": 0.6166522015853729,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5332,
+      "step": 6661
+    },
+    {
+      "epoch": 0.616744778105653,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.02,
+      "loss": 1.582,
+      "step": 6662
+    },
+    {
+      "epoch": 0.616837354625933,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.5228,
+      "step": 6663
+    },
+    {
+      "epoch": 0.6169299311462131,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5022,
+      "step": 6664
+    },
+    {
+      "epoch": 0.6170225076664931,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5566,
+      "step": 6665
+    },
+    {
+      "epoch": 0.6171150841867731,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5247,
+      "step": 6666
+    },
+    {
+      "epoch": 0.6172076607070531,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.546,
+      "step": 6667
+    },
+    {
+      "epoch": 0.6173002372273332,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5824,
+      "step": 6668
+    },
+    {
+      "epoch": 0.6173928137476132,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.5199,
+      "step": 6669
+    },
+    {
+      "epoch": 0.6174853902678933,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.02,
+      "loss": 1.6074,
+      "step": 6670
+    },
+    {
+      "epoch": 0.6175779667881733,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.5608,
+      "step": 6671
+    },
+    {
+      "epoch": 0.6176705433084534,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.5506,
+      "step": 6672
+    },
+    {
+      "epoch": 0.6177631198287334,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.4967,
+      "step": 6673
+    },
+    {
+      "epoch": 0.6178556963490135,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5419,
+      "step": 6674
+    },
+    {
+      "epoch": 0.6179482728692935,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.5208,
+      "step": 6675
+    },
+    {
+      "epoch": 0.6180408493895736,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.4989,
+      "step": 6676
+    },
+    {
+      "epoch": 0.6181334259098537,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5445,
+      "step": 6677
+    },
+    {
+      "epoch": 0.6182260024301337,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.5214,
+      "step": 6678
+    },
+    {
+      "epoch": 0.6183185789504138,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5326,
+      "step": 6679
+    },
+    {
+      "epoch": 0.6184111554706937,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5612,
+      "step": 6680
+    },
+    {
+      "epoch": 0.6185037319909737,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.5966,
+      "step": 6681
+    },
+    {
+      "epoch": 0.6185963085112538,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.5649,
+      "step": 6682
+    },
+    {
+      "epoch": 0.6186888850315339,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5808,
+      "step": 6683
+    },
+    {
+      "epoch": 0.6187814615518139,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5585,
+      "step": 6684
+    },
+    {
+      "epoch": 0.618874038072094,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.02,
+      "loss": 1.5065,
+      "step": 6685
+    },
+    {
+      "epoch": 0.618966614592374,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.503,
+      "step": 6686
+    },
+    {
+      "epoch": 0.6190591911126541,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.02,
+      "loss": 1.4825,
+      "step": 6687
+    },
+    {
+      "epoch": 0.6191517676329341,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.02,
+      "loss": 1.5768,
+      "step": 6688
+    },
+    {
+      "epoch": 0.6192443441532142,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.02,
+      "loss": 1.5553,
+      "step": 6689
+    },
+    {
+      "epoch": 0.6193369206734942,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.02,
+      "loss": 1.6236,
+      "step": 6690
+    },
+    {
+      "epoch": 0.6194294971937743,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.4929,
+      "step": 6691
+    },
+    {
+      "epoch": 0.6195220737140543,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.4727,
+      "step": 6692
+    },
+    {
+      "epoch": 0.6196146502343343,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.02,
+      "loss": 1.4949,
+      "step": 6693
+    },
+    {
+      "epoch": 0.6197072267546143,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.02,
+      "loss": 1.5192,
+      "step": 6694
+    },
+    {
+      "epoch": 0.6197998032748944,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.02,
+      "loss": 1.5407,
+      "step": 6695
+    },
+    {
+      "epoch": 0.6198923797951744,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.4978,
+      "step": 6696
+    },
+    {
+      "epoch": 0.6199849563154545,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.5381,
+      "step": 6697
+    },
+    {
+      "epoch": 0.6200775328357345,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.02,
+      "loss": 1.5135,
+      "step": 6698
+    },
+    {
+      "epoch": 0.6201701093560146,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.02,
+      "loss": 1.5321,
+      "step": 6699
+    },
+    {
+      "epoch": 0.6202626858762946,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.02,
+      "loss": 1.5232,
+      "step": 6700
+    },
+    {
+      "epoch": 0.6203552623965747,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.02,
+      "loss": 1.495,
+      "step": 6701
+    },
+    {
+      "epoch": 0.6204478389168547,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5319,
+      "step": 6702
+    },
+    {
+      "epoch": 0.6205404154371348,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.02,
+      "loss": 1.5022,
+      "step": 6703
+    },
+    {
+      "epoch": 0.6206329919574148,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5856,
+      "step": 6704
+    },
+    {
+      "epoch": 0.6207255684776949,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.02,
+      "loss": 1.5209,
+      "step": 6705
+    },
+    {
+      "epoch": 0.6208181449979749,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.02,
+      "loss": 1.5802,
+      "step": 6706
+    },
+    {
+      "epoch": 0.6209107215182549,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.538,
+      "step": 6707
+    },
+    {
+      "epoch": 0.6210032980385349,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5421,
+      "step": 6708
+    },
+    {
+      "epoch": 0.621095874558815,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.482,
+      "step": 6709
+    },
+    {
+      "epoch": 0.621188451079095,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.4862,
+      "step": 6710
+    },
+    {
+      "epoch": 0.6212810275993751,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5777,
+      "step": 6711
+    },
+    {
+      "epoch": 0.6213736041196551,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.02,
+      "loss": 1.4995,
+      "step": 6712
+    },
+    {
+      "epoch": 0.6214661806399352,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.6032,
+      "step": 6713
+    },
+    {
+      "epoch": 0.6215587571602152,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.02,
+      "loss": 1.5751,
+      "step": 6714
+    },
+    {
+      "epoch": 0.6216513336804953,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5646,
+      "step": 6715
+    },
+    {
+      "epoch": 0.6217439102007754,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5763,
+      "step": 6716
+    },
+    {
+      "epoch": 0.6218364867210554,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5474,
+      "step": 6717
+    },
+    {
+      "epoch": 0.6219290632413355,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.6049,
+      "step": 6718
+    },
+    {
+      "epoch": 0.6220216397616155,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5419,
+      "step": 6719
+    },
+    {
+      "epoch": 0.6221142162818954,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.02,
+      "loss": 1.5154,
+      "step": 6720
+    },
+    {
+      "epoch": 0.6222067928021755,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5755,
+      "step": 6721
+    },
+    {
+      "epoch": 0.6222993693224556,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5325,
+      "step": 6722
+    },
+    {
+      "epoch": 0.6223919458427356,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.6207,
+      "step": 6723
+    },
+    {
+      "epoch": 0.6224845223630157,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5891,
+      "step": 6724
+    },
+    {
+      "epoch": 0.6225770988832957,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5534,
+      "step": 6725
+    },
+    {
+      "epoch": 0.6226696754035758,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.02,
+      "loss": 1.5288,
+      "step": 6726
+    },
+    {
+      "epoch": 0.6227622519238558,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.4771,
+      "step": 6727
+    },
+    {
+      "epoch": 0.6228548284441359,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.02,
+      "loss": 1.5371,
+      "step": 6728
+    },
+    {
+      "epoch": 0.6229474049644159,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.4744,
+      "step": 6729
+    },
+    {
+      "epoch": 0.623039981484696,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.4472,
+      "step": 6730
+    },
+    {
+      "epoch": 0.623132558004976,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.02,
+      "loss": 1.5206,
+      "step": 6731
+    },
+    {
+      "epoch": 0.6232251345252561,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.02,
+      "loss": 1.4971,
+      "step": 6732
+    },
+    {
+      "epoch": 0.6233177110455361,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5427,
+      "step": 6733
+    },
+    {
+      "epoch": 0.6234102875658161,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5065,
+      "step": 6734
+    },
+    {
+      "epoch": 0.6235028640860961,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.02,
+      "loss": 1.5552,
+      "step": 6735
+    },
+    {
+      "epoch": 0.6235954406063762,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.4611,
+      "step": 6736
+    },
+    {
+      "epoch": 0.6236880171266562,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.4739,
+      "step": 6737
+    },
+    {
+      "epoch": 0.6237805936469363,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5504,
+      "step": 6738
+    },
+    {
+      "epoch": 0.6238731701672163,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.511,
+      "step": 6739
+    },
+    {
+      "epoch": 0.6239657466874964,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.4979,
+      "step": 6740
+    },
+    {
+      "epoch": 0.6240583232077764,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.02,
+      "loss": 1.5451,
+      "step": 6741
+    },
+    {
+      "epoch": 0.6241508997280565,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.5801,
+      "step": 6742
+    },
+    {
+      "epoch": 0.6242434762483365,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5192,
+      "step": 6743
+    },
+    {
+      "epoch": 0.6243360527686166,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.02,
+      "loss": 1.5191,
+      "step": 6744
+    },
+    {
+      "epoch": 0.6244286292888966,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5168,
+      "step": 6745
+    },
+    {
+      "epoch": 0.6245212058091767,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.02,
+      "loss": 1.531,
+      "step": 6746
+    },
+    {
+      "epoch": 0.6246137823294567,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5246,
+      "step": 6747
+    },
+    {
+      "epoch": 0.6247063588497367,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.6065,
+      "step": 6748
+    },
+    {
+      "epoch": 0.6247989353700167,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.4978,
+      "step": 6749
+    },
+    {
+      "epoch": 0.6248915118902968,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.02,
+      "loss": 1.5058,
+      "step": 6750
+    },
+    {
+      "epoch": 0.6249840884105768,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.4946,
+      "step": 6751
+    },
+    {
+      "epoch": 0.6250766649308569,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.6439,
+      "step": 6752
+    },
+    {
+      "epoch": 0.625169241451137,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5023,
+      "step": 6753
+    },
+    {
+      "epoch": 0.625261817971417,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.02,
+      "loss": 1.5575,
+      "step": 6754
+    },
+    {
+      "epoch": 0.625354394491697,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5247,
+      "step": 6755
+    },
+    {
+      "epoch": 0.6254469710119771,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5078,
+      "step": 6756
+    },
+    {
+      "epoch": 0.6255395475322572,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5392,
+      "step": 6757
+    },
+    {
+      "epoch": 0.6256321240525372,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5691,
+      "step": 6758
+    },
+    {
+      "epoch": 0.6257247005728173,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.4953,
+      "step": 6759
+    },
+    {
+      "epoch": 0.6258172770930973,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5074,
+      "step": 6760
+    },
+    {
+      "epoch": 0.6259098536133773,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5939,
+      "step": 6761
+    },
+    {
+      "epoch": 0.6260024301336573,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.557,
+      "step": 6762
+    },
+    {
+      "epoch": 0.6260950066539374,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.02,
+      "loss": 1.6182,
+      "step": 6763
+    },
+    {
+      "epoch": 0.6261875831742174,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.6222,
+      "step": 6764
+    },
+    {
+      "epoch": 0.6262801596944975,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.5027,
+      "step": 6765
+    },
+    {
+      "epoch": 0.6263727362147775,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5019,
+      "step": 6766
+    },
+    {
+      "epoch": 0.6264653127350576,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5826,
+      "step": 6767
+    },
+    {
+      "epoch": 0.6265578892553376,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5582,
+      "step": 6768
+    },
+    {
+      "epoch": 0.6266504657756177,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5937,
+      "step": 6769
+    },
+    {
+      "epoch": 0.6267430422958977,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.6126,
+      "step": 6770
+    },
+    {
+      "epoch": 0.6268356188161778,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.527,
+      "step": 6771
+    },
+    {
+      "epoch": 0.6269281953364578,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5487,
+      "step": 6772
+    },
+    {
+      "epoch": 0.6270207718567379,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5721,
+      "step": 6773
+    },
+    {
+      "epoch": 0.6271133483770179,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5953,
+      "step": 6774
+    },
+    {
+      "epoch": 0.6272059248972979,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.494,
+      "step": 6775
+    },
+    {
+      "epoch": 0.6272985014175779,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.02,
+      "loss": 1.5298,
+      "step": 6776
+    },
+    {
+      "epoch": 0.627391077937858,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5538,
+      "step": 6777
+    },
+    {
+      "epoch": 0.627483654458138,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5399,
+      "step": 6778
+    },
+    {
+      "epoch": 0.6275762309784181,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5315,
+      "step": 6779
+    },
+    {
+      "epoch": 0.6276688074986981,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5445,
+      "step": 6780
+    },
+    {
+      "epoch": 0.6277613840189782,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5176,
+      "step": 6781
+    },
+    {
+      "epoch": 0.6278539605392582,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5561,
+      "step": 6782
+    },
+    {
+      "epoch": 0.6279465370595383,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.6013,
+      "step": 6783
+    },
+    {
+      "epoch": 0.6280391135798183,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5188,
+      "step": 6784
+    },
+    {
+      "epoch": 0.6281316901000984,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5482,
+      "step": 6785
+    },
+    {
+      "epoch": 0.6282242666203784,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5199,
+      "step": 6786
+    },
+    {
+      "epoch": 0.6283168431406585,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.02,
+      "loss": 1.5347,
+      "step": 6787
+    },
+    {
+      "epoch": 0.6284094196609384,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.6074,
+      "step": 6788
+    },
+    {
+      "epoch": 0.6285019961812185,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.02,
+      "loss": 1.5224,
+      "step": 6789
+    },
+    {
+      "epoch": 0.6285945727014985,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.566,
+      "step": 6790
+    },
+    {
+      "epoch": 0.6286871492217786,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.4844,
+      "step": 6791
+    },
+    {
+      "epoch": 0.6287797257420586,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5236,
+      "step": 6792
+    },
+    {
+      "epoch": 0.6288723022623387,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5125,
+      "step": 6793
+    },
+    {
+      "epoch": 0.6289648787826188,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.02,
+      "loss": 1.5127,
+      "step": 6794
+    },
+    {
+      "epoch": 0.6290574553028988,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5138,
+      "step": 6795
+    },
+    {
+      "epoch": 0.6291500318231789,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.02,
+      "loss": 1.4401,
+      "step": 6796
+    },
+    {
+      "epoch": 0.6292426083434589,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5157,
+      "step": 6797
+    },
+    {
+      "epoch": 0.629335184863739,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.02,
+      "loss": 1.5402,
+      "step": 6798
+    },
+    {
+      "epoch": 0.629427761384019,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5397,
+      "step": 6799
+    },
+    {
+      "epoch": 0.6295203379042991,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.634,
+      "step": 6800
+    },
+    {
+      "epoch": 0.6296129144245791,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5347,
+      "step": 6801
+    },
+    {
+      "epoch": 0.6297054909448591,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5449,
+      "step": 6802
+    },
+    {
+      "epoch": 0.6297980674651391,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.02,
+      "loss": 1.5431,
+      "step": 6803
+    },
+    {
+      "epoch": 0.6298906439854192,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5944,
+      "step": 6804
+    },
+    {
+      "epoch": 0.6299832205056992,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5464,
+      "step": 6805
+    },
+    {
+      "epoch": 0.6300757970259793,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.5359,
+      "step": 6806
+    },
+    {
+      "epoch": 0.6301683735462593,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5277,
+      "step": 6807
+    },
+    {
+      "epoch": 0.6302609500665394,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.02,
+      "loss": 1.5286,
+      "step": 6808
+    },
+    {
+      "epoch": 0.6303535265868194,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.02,
+      "loss": 1.5568,
+      "step": 6809
+    },
+    {
+      "epoch": 0.6304461031070995,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.4955,
+      "step": 6810
+    },
+    {
+      "epoch": 0.6305386796273795,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.6033,
+      "step": 6811
+    },
+    {
+      "epoch": 0.6306312561476596,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5297,
+      "step": 6812
+    },
+    {
+      "epoch": 0.6307238326679396,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5824,
+      "step": 6813
+    },
+    {
+      "epoch": 0.6308164091882197,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.4772,
+      "step": 6814
+    },
+    {
+      "epoch": 0.6309089857084997,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.02,
+      "loss": 1.4861,
+      "step": 6815
+    },
+    {
+      "epoch": 0.6310015622287797,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5484,
+      "step": 6816
+    },
+    {
+      "epoch": 0.6310941387490597,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5725,
+      "step": 6817
+    },
+    {
+      "epoch": 0.6311867152693398,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.02,
+      "loss": 1.5253,
+      "step": 6818
+    },
+    {
+      "epoch": 0.6312792917896198,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.02,
+      "loss": 1.5673,
+      "step": 6819
+    },
+    {
+      "epoch": 0.6313718683098999,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.518,
+      "step": 6820
+    },
+    {
+      "epoch": 0.6314644448301799,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5634,
+      "step": 6821
+    },
+    {
+      "epoch": 0.63155702135046,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.5053,
+      "step": 6822
+    },
+    {
+      "epoch": 0.63164959787074,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.6209,
+      "step": 6823
+    },
+    {
+      "epoch": 0.6317421743910201,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5405,
+      "step": 6824
+    },
+    {
+      "epoch": 0.6318347509113001,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.02,
+      "loss": 1.5051,
+      "step": 6825
+    },
+    {
+      "epoch": 0.6319273274315802,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.6131,
+      "step": 6826
+    },
+    {
+      "epoch": 0.6320199039518603,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5352,
+      "step": 6827
+    },
+    {
+      "epoch": 0.6321124804721403,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5225,
+      "step": 6828
+    },
+    {
+      "epoch": 0.6322050569924202,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.5132,
+      "step": 6829
+    },
+    {
+      "epoch": 0.6322976335127003,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.02,
+      "loss": 1.5845,
+      "step": 6830
+    },
+    {
+      "epoch": 0.6323902100329803,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5778,
+      "step": 6831
+    },
+    {
+      "epoch": 0.6324827865532604,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5354,
+      "step": 6832
+    },
+    {
+      "epoch": 0.6325753630735405,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.5108,
+      "step": 6833
+    },
+    {
+      "epoch": 0.6326679395938205,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5193,
+      "step": 6834
+    },
+    {
+      "epoch": 0.6327605161141006,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.02,
+      "loss": 1.5823,
+      "step": 6835
+    },
+    {
+      "epoch": 0.6328530926343806,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.4716,
+      "step": 6836
+    },
+    {
+      "epoch": 0.6329456691546607,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.531,
+      "step": 6837
+    },
+    {
+      "epoch": 0.6330382456749407,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.02,
+      "loss": 1.5429,
+      "step": 6838
+    },
+    {
+      "epoch": 0.6331308221952208,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5171,
+      "step": 6839
+    },
+    {
+      "epoch": 0.6332233987155008,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.4949,
+      "step": 6840
+    },
+    {
+      "epoch": 0.6333159752357809,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5696,
+      "step": 6841
+    },
+    {
+      "epoch": 0.6334085517560609,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5396,
+      "step": 6842
+    },
+    {
+      "epoch": 0.6335011282763409,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5314,
+      "step": 6843
+    },
+    {
+      "epoch": 0.6335937047966209,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.02,
+      "loss": 1.6416,
+      "step": 6844
+    },
+    {
+      "epoch": 0.633686281316901,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.02,
+      "loss": 1.5146,
+      "step": 6845
+    },
+    {
+      "epoch": 0.633778857837181,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.02,
+      "loss": 1.5109,
+      "step": 6846
+    },
+    {
+      "epoch": 0.6338714343574611,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.5465,
+      "step": 6847
+    },
+    {
+      "epoch": 0.6339640108777411,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.4715,
+      "step": 6848
+    },
+    {
+      "epoch": 0.6340565873980212,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5156,
+      "step": 6849
+    },
+    {
+      "epoch": 0.6341491639183012,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.5612,
+      "step": 6850
+    },
+    {
+      "epoch": 0.6342417404385813,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5648,
+      "step": 6851
+    },
+    {
+      "epoch": 0.6343343169588613,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5548,
+      "step": 6852
+    },
+    {
+      "epoch": 0.6344268934791414,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5017,
+      "step": 6853
+    },
+    {
+      "epoch": 0.6345194699994214,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5028,
+      "step": 6854
+    },
+    {
+      "epoch": 0.6346120465197015,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5272,
+      "step": 6855
+    },
+    {
+      "epoch": 0.6347046230399814,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.5115,
+      "step": 6856
+    },
+    {
+      "epoch": 0.6347971995602615,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5362,
+      "step": 6857
+    },
+    {
+      "epoch": 0.6348897760805415,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5531,
+      "step": 6858
+    },
+    {
+      "epoch": 0.6349823526008216,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.02,
+      "loss": 1.5135,
+      "step": 6859
+    },
+    {
+      "epoch": 0.6350749291211016,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5148,
+      "step": 6860
+    },
+    {
+      "epoch": 0.6351675056413817,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5144,
+      "step": 6861
+    },
+    {
+      "epoch": 0.6352600821616617,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5273,
+      "step": 6862
+    },
+    {
+      "epoch": 0.6353526586819418,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5307,
+      "step": 6863
+    },
+    {
+      "epoch": 0.6354452352022218,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5497,
+      "step": 6864
+    },
+    {
+      "epoch": 0.6355378117225019,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.02,
+      "loss": 1.5532,
+      "step": 6865
+    },
+    {
+      "epoch": 0.635630388242782,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5032,
+      "step": 6866
+    },
+    {
+      "epoch": 0.635722964763062,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.484,
+      "step": 6867
+    },
+    {
+      "epoch": 0.6358155412833421,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5252,
+      "step": 6868
+    },
+    {
+      "epoch": 0.6359081178036221,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.02,
+      "loss": 1.6132,
+      "step": 6869
+    },
+    {
+      "epoch": 0.636000694323902,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.02,
+      "loss": 1.5779,
+      "step": 6870
+    },
+    {
+      "epoch": 0.6360932708441821,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5395,
+      "step": 6871
+    },
+    {
+      "epoch": 0.6361858473644622,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.02,
+      "loss": 1.4735,
+      "step": 6872
+    },
+    {
+      "epoch": 0.6362784238847422,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.6134,
+      "step": 6873
+    },
+    {
+      "epoch": 0.6363710004050223,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.02,
+      "loss": 1.5436,
+      "step": 6874
+    },
+    {
+      "epoch": 0.6364635769253023,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.4479,
+      "step": 6875
+    },
+    {
+      "epoch": 0.6365561534455824,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5501,
+      "step": 6876
+    },
+    {
+      "epoch": 0.6366487299658624,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5602,
+      "step": 6877
+    },
+    {
+      "epoch": 0.6367413064861425,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.02,
+      "loss": 1.5308,
+      "step": 6878
+    },
+    {
+      "epoch": 0.6368338830064225,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5616,
+      "step": 6879
+    },
+    {
+      "epoch": 0.6369264595267026,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.02,
+      "loss": 1.4953,
+      "step": 6880
+    },
+    {
+      "epoch": 0.6370190360469826,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5021,
+      "step": 6881
+    },
+    {
+      "epoch": 0.6371116125672627,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.5274,
+      "step": 6882
+    },
+    {
+      "epoch": 0.6372041890875427,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.5295,
+      "step": 6883
+    },
+    {
+      "epoch": 0.6372967656078227,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5061,
+      "step": 6884
+    },
+    {
+      "epoch": 0.6373893421281027,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5102,
+      "step": 6885
+    },
+    {
+      "epoch": 0.6374819186483828,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5372,
+      "step": 6886
+    },
+    {
+      "epoch": 0.6375744951686628,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5279,
+      "step": 6887
+    },
+    {
+      "epoch": 0.6376670716889429,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5596,
+      "step": 6888
+    },
+    {
+      "epoch": 0.6377596482092229,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5539,
+      "step": 6889
+    },
+    {
+      "epoch": 0.637852224729503,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.02,
+      "loss": 1.5243,
+      "step": 6890
+    },
+    {
+      "epoch": 0.637944801249783,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.02,
+      "loss": 1.559,
+      "step": 6891
+    },
+    {
+      "epoch": 0.6380373777700631,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.4824,
+      "step": 6892
+    },
+    {
+      "epoch": 0.6381299542903431,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5057,
+      "step": 6893
+    },
+    {
+      "epoch": 0.6382225308106232,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.5187,
+      "step": 6894
+    },
+    {
+      "epoch": 0.6383151073309032,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5176,
+      "step": 6895
+    },
+    {
+      "epoch": 0.6384076838511833,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5905,
+      "step": 6896
+    },
+    {
+      "epoch": 0.6385002603714632,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5079,
+      "step": 6897
+    },
+    {
+      "epoch": 0.6385928368917433,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5887,
+      "step": 6898
+    },
+    {
+      "epoch": 0.6386854134120233,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5667,
+      "step": 6899
+    },
+    {
+      "epoch": 0.6387779899323034,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.4797,
+      "step": 6900
+    },
+    {
+      "epoch": 0.6388705664525834,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.02,
+      "loss": 1.5768,
+      "step": 6901
+    },
+    {
+      "epoch": 0.6389631429728635,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5428,
+      "step": 6902
+    },
+    {
+      "epoch": 0.6390557194931435,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5882,
+      "step": 6903
+    },
+    {
+      "epoch": 0.6391482960134236,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5279,
+      "step": 6904
+    },
+    {
+      "epoch": 0.6392408725337037,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5541,
+      "step": 6905
+    },
+    {
+      "epoch": 0.6393334490539837,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.5721,
+      "step": 6906
+    },
+    {
+      "epoch": 0.6394260255742638,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.4988,
+      "step": 6907
+    },
+    {
+      "epoch": 0.6395186020945438,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.02,
+      "loss": 1.5483,
+      "step": 6908
+    },
+    {
+      "epoch": 0.6396111786148239,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5715,
+      "step": 6909
+    },
+    {
+      "epoch": 0.6397037551351039,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.6095,
+      "step": 6910
+    },
+    {
+      "epoch": 0.6397963316553839,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5479,
+      "step": 6911
+    },
+    {
+      "epoch": 0.6398889081756639,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.5332,
+      "step": 6912
+    },
+    {
+      "epoch": 0.639981484695944,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5467,
+      "step": 6913
+    },
+    {
+      "epoch": 0.640074061216224,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5142,
+      "step": 6914
+    },
+    {
+      "epoch": 0.6401666377365041,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.4879,
+      "step": 6915
+    },
+    {
+      "epoch": 0.6402592142567841,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.517,
+      "step": 6916
+    },
+    {
+      "epoch": 0.6403517907770642,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5169,
+      "step": 6917
+    },
+    {
+      "epoch": 0.6404443672973442,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.02,
+      "loss": 1.5689,
+      "step": 6918
+    },
+    {
+      "epoch": 0.6405369438176243,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5503,
+      "step": 6919
+    },
+    {
+      "epoch": 0.6406295203379043,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.02,
+      "loss": 1.5423,
+      "step": 6920
+    },
+    {
+      "epoch": 0.6407220968581844,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.02,
+      "loss": 1.5662,
+      "step": 6921
+    },
+    {
+      "epoch": 0.6408146733784644,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5408,
+      "step": 6922
+    },
+    {
+      "epoch": 0.6409072498987445,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5496,
+      "step": 6923
+    },
+    {
+      "epoch": 0.6409998264190244,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.4823,
+      "step": 6924
+    },
+    {
+      "epoch": 0.6410924029393045,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.4633,
+      "step": 6925
+    },
+    {
+      "epoch": 0.6411849794595845,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5403,
+      "step": 6926
+    },
+    {
+      "epoch": 0.6412775559798646,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5125,
+      "step": 6927
+    },
+    {
+      "epoch": 0.6413701325001446,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5201,
+      "step": 6928
+    },
+    {
+      "epoch": 0.6414627090204247,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.02,
+      "loss": 1.4283,
+      "step": 6929
+    },
+    {
+      "epoch": 0.6415552855407047,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.5329,
+      "step": 6930
+    },
+    {
+      "epoch": 0.6416478620609848,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5394,
+      "step": 6931
+    },
+    {
+      "epoch": 0.6417404385812648,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5008,
+      "step": 6932
+    },
+    {
+      "epoch": 0.6418330151015449,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.5371,
+      "step": 6933
+    },
+    {
+      "epoch": 0.6419255916218249,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.565,
+      "step": 6934
+    },
+    {
+      "epoch": 0.642018168142105,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.02,
+      "loss": 1.5065,
+      "step": 6935
+    },
+    {
+      "epoch": 0.642110744662385,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5359,
+      "step": 6936
+    },
+    {
+      "epoch": 0.6422033211826651,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5331,
+      "step": 6937
+    },
+    {
+      "epoch": 0.642295897702945,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5694,
+      "step": 6938
+    },
+    {
+      "epoch": 0.6423884742232251,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5365,
+      "step": 6939
+    },
+    {
+      "epoch": 0.6424810507435051,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.02,
+      "loss": 1.5305,
+      "step": 6940
+    },
+    {
+      "epoch": 0.6425736272637852,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5389,
+      "step": 6941
+    },
+    {
+      "epoch": 0.6426662037840652,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.4914,
+      "step": 6942
+    },
+    {
+      "epoch": 0.6427587803043453,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.5213,
+      "step": 6943
+    },
+    {
+      "epoch": 0.6428513568246254,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5706,
+      "step": 6944
+    },
+    {
+      "epoch": 0.6429439333449054,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5141,
+      "step": 6945
+    },
+    {
+      "epoch": 0.6430365098651855,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.4908,
+      "step": 6946
+    },
+    {
+      "epoch": 0.6431290863854655,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5179,
+      "step": 6947
+    },
+    {
+      "epoch": 0.6432216629057456,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.509,
+      "step": 6948
+    },
+    {
+      "epoch": 0.6433142394260256,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.02,
+      "loss": 1.5446,
+      "step": 6949
+    },
+    {
+      "epoch": 0.6434068159463057,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.02,
+      "loss": 1.5493,
+      "step": 6950
+    },
+    {
+      "epoch": 0.6434993924665856,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5626,
+      "step": 6951
+    },
+    {
+      "epoch": 0.6435919689868657,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.5388,
+      "step": 6952
+    },
+    {
+      "epoch": 0.6436845455071457,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.02,
+      "loss": 1.5022,
+      "step": 6953
+    },
+    {
+      "epoch": 0.6437771220274258,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.532,
+      "step": 6954
+    },
+    {
+      "epoch": 0.6438696985477058,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5086,
+      "step": 6955
+    },
+    {
+      "epoch": 0.6439622750679859,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.5256,
+      "step": 6956
+    },
+    {
+      "epoch": 0.6440548515882659,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.02,
+      "loss": 1.545,
+      "step": 6957
+    },
+    {
+      "epoch": 0.644147428108546,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.02,
+      "loss": 1.5534,
+      "step": 6958
+    },
+    {
+      "epoch": 0.644240004628826,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5293,
+      "step": 6959
+    },
+    {
+      "epoch": 0.6443325811491061,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.4989,
+      "step": 6960
+    },
+    {
+      "epoch": 0.6444251576693861,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5141,
+      "step": 6961
+    },
+    {
+      "epoch": 0.6445177341896662,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.02,
+      "loss": 1.5295,
+      "step": 6962
+    },
+    {
+      "epoch": 0.6446103107099462,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.4915,
+      "step": 6963
+    },
+    {
+      "epoch": 0.6447028872302263,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.02,
+      "loss": 1.537,
+      "step": 6964
+    },
+    {
+      "epoch": 0.6447954637505062,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.6043,
+      "step": 6965
+    },
+    {
+      "epoch": 0.6448880402707863,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.513,
+      "step": 6966
+    },
+    {
+      "epoch": 0.6449806167910663,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.513,
+      "step": 6967
+    },
+    {
+      "epoch": 0.6450731933113464,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5594,
+      "step": 6968
+    },
+    {
+      "epoch": 0.6451657698316264,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.5712,
+      "step": 6969
+    },
+    {
+      "epoch": 0.6452583463519065,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.5551,
+      "step": 6970
+    },
+    {
+      "epoch": 0.6453509228721865,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.02,
+      "loss": 1.5477,
+      "step": 6971
+    },
+    {
+      "epoch": 0.6454434993924666,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.02,
+      "loss": 1.4643,
+      "step": 6972
+    },
+    {
+      "epoch": 0.6455360759127466,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.5483,
+      "step": 6973
+    },
+    {
+      "epoch": 0.6456286524330267,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.02,
+      "loss": 1.4795,
+      "step": 6974
+    },
+    {
+      "epoch": 0.6457212289533067,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5788,
+      "step": 6975
+    },
+    {
+      "epoch": 0.6458138054735868,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5272,
+      "step": 6976
+    },
+    {
+      "epoch": 0.6459063819938669,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5188,
+      "step": 6977
+    },
+    {
+      "epoch": 0.6459989585141469,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.02,
+      "loss": 1.5237,
+      "step": 6978
+    },
+    {
+      "epoch": 0.6460915350344268,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.5556,
+      "step": 6979
+    },
+    {
+      "epoch": 0.6461841115547069,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.6078,
+      "step": 6980
+    },
+    {
+      "epoch": 0.646276688074987,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5068,
+      "step": 6981
+    },
+    {
+      "epoch": 0.646369264595267,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5594,
+      "step": 6982
+    },
+    {
+      "epoch": 0.646461841115547,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.02,
+      "loss": 1.5739,
+      "step": 6983
+    },
+    {
+      "epoch": 0.6465544176358271,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5648,
+      "step": 6984
+    },
+    {
+      "epoch": 0.6466469941561072,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5095,
+      "step": 6985
+    },
+    {
+      "epoch": 0.6467395706763872,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5381,
+      "step": 6986
+    },
+    {
+      "epoch": 0.6468321471966673,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.02,
+      "loss": 1.592,
+      "step": 6987
+    },
+    {
+      "epoch": 0.6469247237169473,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5577,
+      "step": 6988
+    },
+    {
+      "epoch": 0.6470173002372274,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.5469,
+      "step": 6989
+    },
+    {
+      "epoch": 0.6471098767575074,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.5236,
+      "step": 6990
+    },
+    {
+      "epoch": 0.6472024532777875,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.496,
+      "step": 6991
+    },
+    {
+      "epoch": 0.6472950297980674,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.5682,
+      "step": 6992
+    },
+    {
+      "epoch": 0.6473876063183475,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.02,
+      "loss": 1.5051,
+      "step": 6993
+    },
+    {
+      "epoch": 0.6474801828386275,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.4882,
+      "step": 6994
+    },
+    {
+      "epoch": 0.6475727593589076,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.02,
+      "loss": 1.5555,
+      "step": 6995
+    },
+    {
+      "epoch": 0.6476653358791876,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.5144,
+      "step": 6996
+    },
+    {
+      "epoch": 0.6477579123994677,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.02,
+      "loss": 1.5738,
+      "step": 6997
+    },
+    {
+      "epoch": 0.6478504889197477,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.5481,
+      "step": 6998
+    },
+    {
+      "epoch": 0.6479430654400278,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5341,
+      "step": 6999
+    },
+    {
+      "epoch": 0.6480356419603078,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.02,
+      "loss": 1.57,
+      "step": 7000
+    },
+    {
+      "epoch": 0.6481282184805879,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.02,
+      "loss": 1.5522,
+      "step": 7001
+    },
+    {
+      "epoch": 0.6482207950008679,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.4615,
+      "step": 7002
+    },
+    {
+      "epoch": 0.648313371521148,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.02,
+      "loss": 1.4901,
+      "step": 7003
+    },
+    {
+      "epoch": 0.648405948041428,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5291,
+      "step": 7004
+    },
+    {
+      "epoch": 0.6484985245617081,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5162,
+      "step": 7005
+    },
+    {
+      "epoch": 0.648591101081988,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.02,
+      "loss": 1.55,
+      "step": 7006
+    },
+    {
+      "epoch": 0.6486836776022681,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.02,
+      "loss": 1.503,
+      "step": 7007
+    },
+    {
+      "epoch": 0.6487762541225481,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5728,
+      "step": 7008
+    },
+    {
+      "epoch": 0.6488688306428282,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5504,
+      "step": 7009
+    },
+    {
+      "epoch": 0.6489614071631082,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.5112,
+      "step": 7010
+    },
+    {
+      "epoch": 0.6490539836833883,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.498,
+      "step": 7011
+    },
+    {
+      "epoch": 0.6491465602036683,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5504,
+      "step": 7012
+    },
+    {
+      "epoch": 0.6492391367239484,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5146,
+      "step": 7013
+    },
+    {
+      "epoch": 0.6493317132442284,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.02,
+      "loss": 1.5434,
+      "step": 7014
+    },
+    {
+      "epoch": 0.6494242897645085,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.6207,
+      "step": 7015
+    },
+    {
+      "epoch": 0.6495168662847886,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5758,
+      "step": 7016
+    },
+    {
+      "epoch": 0.6496094428050686,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.02,
+      "loss": 1.5382,
+      "step": 7017
+    },
+    {
+      "epoch": 0.6497020193253487,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.51,
+      "step": 7018
+    },
+    {
+      "epoch": 0.6497945958456286,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5833,
+      "step": 7019
+    },
+    {
+      "epoch": 0.6498871723659086,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.02,
+      "loss": 1.5017,
+      "step": 7020
+    },
+    {
+      "epoch": 0.6499797488861887,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.6002,
+      "step": 7021
+    },
+    {
+      "epoch": 0.6500723254064688,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.5729,
+      "step": 7022
+    },
+    {
+      "epoch": 0.6501649019267488,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.02,
+      "loss": 1.5552,
+      "step": 7023
+    },
+    {
+      "epoch": 0.6502574784470289,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5883,
+      "step": 7024
+    },
+    {
+      "epoch": 0.6503500549673089,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.02,
+      "loss": 1.5193,
+      "step": 7025
+    },
+    {
+      "epoch": 0.650442631487589,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.02,
+      "loss": 1.5085,
+      "step": 7026
+    },
+    {
+      "epoch": 0.650535208007869,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5219,
+      "step": 7027
+    },
+    {
+      "epoch": 0.6506277845281491,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5924,
+      "step": 7028
+    },
+    {
+      "epoch": 0.6507203610484291,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.02,
+      "loss": 1.5353,
+      "step": 7029
+    },
+    {
+      "epoch": 0.6508129375687092,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.498,
+      "step": 7030
+    },
+    {
+      "epoch": 0.6509055140889892,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.5127,
+      "step": 7031
+    },
+    {
+      "epoch": 0.6509980906092693,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.5929,
+      "step": 7032
+    },
+    {
+      "epoch": 0.6510906671295492,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5752,
+      "step": 7033
+    },
+    {
+      "epoch": 0.6511832436498293,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5528,
+      "step": 7034
+    },
+    {
+      "epoch": 0.6512758201701093,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5595,
+      "step": 7035
+    },
+    {
+      "epoch": 0.6513683966903894,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5401,
+      "step": 7036
+    },
+    {
+      "epoch": 0.6514609732106694,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.02,
+      "loss": 1.6001,
+      "step": 7037
+    },
+    {
+      "epoch": 0.6515535497309495,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5089,
+      "step": 7038
+    },
+    {
+      "epoch": 0.6516461262512295,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5124,
+      "step": 7039
+    },
+    {
+      "epoch": 0.6517387027715096,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.5675,
+      "step": 7040
+    },
+    {
+      "epoch": 0.6518312792917896,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.4882,
+      "step": 7041
+    },
+    {
+      "epoch": 0.6519238558120697,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5613,
+      "step": 7042
+    },
+    {
+      "epoch": 0.6520164323323497,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5192,
+      "step": 7043
+    },
+    {
+      "epoch": 0.6521090088526298,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5245,
+      "step": 7044
+    },
+    {
+      "epoch": 0.6522015853729098,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5628,
+      "step": 7045
+    },
+    {
+      "epoch": 0.6522941618931899,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.02,
+      "loss": 1.5165,
+      "step": 7046
+    },
+    {
+      "epoch": 0.6523867384134698,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.02,
+      "loss": 1.4986,
+      "step": 7047
+    },
+    {
+      "epoch": 0.6524793149337499,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.02,
+      "loss": 1.5836,
+      "step": 7048
+    },
+    {
+      "epoch": 0.6525718914540299,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5569,
+      "step": 7049
+    },
+    {
+      "epoch": 0.65266446797431,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.02,
+      "loss": 1.5142,
+      "step": 7050
+    },
+    {
+      "epoch": 0.65275704449459,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.6124,
+      "step": 7051
+    },
+    {
+      "epoch": 0.6528496210148701,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.02,
+      "loss": 1.4804,
+      "step": 7052
+    },
+    {
+      "epoch": 0.6529421975351501,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.534,
+      "step": 7053
+    },
+    {
+      "epoch": 0.6530347740554302,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.02,
+      "loss": 1.5432,
+      "step": 7054
+    },
+    {
+      "epoch": 0.6531273505757103,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.4559,
+      "step": 7055
+    },
+    {
+      "epoch": 0.6532199270959903,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5517,
+      "step": 7056
+    },
+    {
+      "epoch": 0.6533125036162704,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5942,
+      "step": 7057
+    },
+    {
+      "epoch": 0.6534050801365504,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.4725,
+      "step": 7058
+    },
+    {
+      "epoch": 0.6534976566568305,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5225,
+      "step": 7059
+    },
+    {
+      "epoch": 0.6535902331771104,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5129,
+      "step": 7060
+    },
+    {
+      "epoch": 0.6536828096973905,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5477,
+      "step": 7061
+    },
+    {
+      "epoch": 0.6537753862176705,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.4667,
+      "step": 7062
+    },
+    {
+      "epoch": 0.6538679627379506,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5612,
+      "step": 7063
+    },
+    {
+      "epoch": 0.6539605392582306,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.02,
+      "loss": 1.524,
+      "step": 7064
+    },
+    {
+      "epoch": 0.6540531157785107,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.02,
+      "loss": 1.5737,
+      "step": 7065
+    },
+    {
+      "epoch": 0.6541456922987907,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.02,
+      "loss": 1.5636,
+      "step": 7066
+    },
+    {
+      "epoch": 0.6542382688190708,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.538,
+      "step": 7067
+    },
+    {
+      "epoch": 0.6543308453393508,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.02,
+      "loss": 1.4911,
+      "step": 7068
+    },
+    {
+      "epoch": 0.6544234218596309,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5825,
+      "step": 7069
+    },
+    {
+      "epoch": 0.6545159983799109,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.489,
+      "step": 7070
+    },
+    {
+      "epoch": 0.654608574900191,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.02,
+      "loss": 1.4667,
+      "step": 7071
+    },
+    {
+      "epoch": 0.654701151420471,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5188,
+      "step": 7072
+    },
+    {
+      "epoch": 0.6547937279407511,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.4967,
+      "step": 7073
+    },
+    {
+      "epoch": 0.654886304461031,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.502,
+      "step": 7074
+    },
+    {
+      "epoch": 0.6549788809813111,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.538,
+      "step": 7075
+    },
+    {
+      "epoch": 0.6550714575015911,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5159,
+      "step": 7076
+    },
+    {
+      "epoch": 0.6551640340218712,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5382,
+      "step": 7077
+    },
+    {
+      "epoch": 0.6552566105421512,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5602,
+      "step": 7078
+    },
+    {
+      "epoch": 0.6553491870624313,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.02,
+      "loss": 1.4543,
+      "step": 7079
+    },
+    {
+      "epoch": 0.6554417635827113,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.02,
+      "loss": 1.543,
+      "step": 7080
+    },
+    {
+      "epoch": 0.6555343401029914,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5178,
+      "step": 7081
+    },
+    {
+      "epoch": 0.6556269166232714,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.02,
+      "loss": 1.5675,
+      "step": 7082
+    },
+    {
+      "epoch": 0.6557194931435515,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5709,
+      "step": 7083
+    },
+    {
+      "epoch": 0.6558120696638315,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.5332,
+      "step": 7084
+    },
+    {
+      "epoch": 0.6559046461841116,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.6065,
+      "step": 7085
+    },
+    {
+      "epoch": 0.6559972227043916,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.02,
+      "loss": 1.5203,
+      "step": 7086
+    },
+    {
+      "epoch": 0.6560897992246716,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.02,
+      "loss": 1.5032,
+      "step": 7087
+    },
+    {
+      "epoch": 0.6561823757449516,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5191,
+      "step": 7088
+    },
+    {
+      "epoch": 0.6562749522652317,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.02,
+      "loss": 1.5865,
+      "step": 7089
+    },
+    {
+      "epoch": 0.6563675287855117,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.02,
+      "loss": 1.5691,
+      "step": 7090
+    },
+    {
+      "epoch": 0.6564601053057918,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5632,
+      "step": 7091
+    },
+    {
+      "epoch": 0.6565526818260718,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.5852,
+      "step": 7092
+    },
+    {
+      "epoch": 0.6566452583463519,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5442,
+      "step": 7093
+    },
+    {
+      "epoch": 0.656737834866632,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5066,
+      "step": 7094
+    },
+    {
+      "epoch": 0.656830411386912,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.02,
+      "loss": 1.5269,
+      "step": 7095
+    },
+    {
+      "epoch": 0.6569229879071921,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5695,
+      "step": 7096
+    },
+    {
+      "epoch": 0.6570155644274721,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5025,
+      "step": 7097
+    },
+    {
+      "epoch": 0.6571081409477522,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5173,
+      "step": 7098
+    },
+    {
+      "epoch": 0.6572007174680322,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5416,
+      "step": 7099
+    },
+    {
+      "epoch": 0.6572932939883123,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.02,
+      "loss": 1.513,
+      "step": 7100
+    },
+    {
+      "epoch": 0.6573858705085922,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5878,
+      "step": 7101
+    },
+    {
+      "epoch": 0.6574784470288723,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5321,
+      "step": 7102
+    },
+    {
+      "epoch": 0.6575710235491523,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.4953,
+      "step": 7103
+    },
+    {
+      "epoch": 0.6576636000694324,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5587,
+      "step": 7104
+    },
+    {
+      "epoch": 0.6577561765897124,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.5255,
+      "step": 7105
+    },
+    {
+      "epoch": 0.6578487531099925,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5729,
+      "step": 7106
+    },
+    {
+      "epoch": 0.6579413296302725,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.4722,
+      "step": 7107
+    },
+    {
+      "epoch": 0.6580339061505526,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.02,
+      "loss": 1.5141,
+      "step": 7108
+    },
+    {
+      "epoch": 0.6581264826708326,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.53,
+      "step": 7109
+    },
+    {
+      "epoch": 0.6582190591911127,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5049,
+      "step": 7110
+    },
+    {
+      "epoch": 0.6583116357113927,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.5695,
+      "step": 7111
+    },
+    {
+      "epoch": 0.6584042122316728,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5702,
+      "step": 7112
+    },
+    {
+      "epoch": 0.6584967887519528,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5384,
+      "step": 7113
+    },
+    {
+      "epoch": 0.6585893652722329,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.02,
+      "loss": 1.5777,
+      "step": 7114
+    },
+    {
+      "epoch": 0.6586819417925128,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5142,
+      "step": 7115
+    },
+    {
+      "epoch": 0.6587745183127929,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5431,
+      "step": 7116
+    },
+    {
+      "epoch": 0.6588670948330729,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.4982,
+      "step": 7117
+    },
+    {
+      "epoch": 0.658959671353353,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5792,
+      "step": 7118
+    },
+    {
+      "epoch": 0.659052247873633,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5506,
+      "step": 7119
+    },
+    {
+      "epoch": 0.6591448243939131,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.5503,
+      "step": 7120
+    },
+    {
+      "epoch": 0.6592374009141931,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5406,
+      "step": 7121
+    },
+    {
+      "epoch": 0.6593299774344732,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.5276,
+      "step": 7122
+    },
+    {
+      "epoch": 0.6594225539547532,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5232,
+      "step": 7123
+    },
+    {
+      "epoch": 0.6595151304750333,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.562,
+      "step": 7124
+    },
+    {
+      "epoch": 0.6596077069953133,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5392,
+      "step": 7125
+    },
+    {
+      "epoch": 0.6597002835155934,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5357,
+      "step": 7126
+    },
+    {
+      "epoch": 0.6597928600358735,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5559,
+      "step": 7127
+    },
+    {
+      "epoch": 0.6598854365561534,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.02,
+      "loss": 1.4738,
+      "step": 7128
+    },
+    {
+      "epoch": 0.6599780130764334,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.5174,
+      "step": 7129
+    },
+    {
+      "epoch": 0.6600705895967135,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.4874,
+      "step": 7130
+    },
+    {
+      "epoch": 0.6601631661169935,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.02,
+      "loss": 1.5331,
+      "step": 7131
+    },
+    {
+      "epoch": 0.6602557426372736,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5166,
+      "step": 7132
+    },
+    {
+      "epoch": 0.6603483191575537,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5238,
+      "step": 7133
+    },
+    {
+      "epoch": 0.6604408956778337,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.02,
+      "loss": 1.5368,
+      "step": 7134
+    },
+    {
+      "epoch": 0.6605334721981138,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.4943,
+      "step": 7135
+    },
+    {
+      "epoch": 0.6606260487183938,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.02,
+      "loss": 1.5634,
+      "step": 7136
+    },
+    {
+      "epoch": 0.6607186252386739,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5243,
+      "step": 7137
+    },
+    {
+      "epoch": 0.6608112017589539,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.4649,
+      "step": 7138
+    },
+    {
+      "epoch": 0.660903778279234,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.02,
+      "loss": 1.5548,
+      "step": 7139
+    },
+    {
+      "epoch": 0.660996354799514,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.546,
+      "step": 7140
+    },
+    {
+      "epoch": 0.6610889313197941,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5231,
+      "step": 7141
+    },
+    {
+      "epoch": 0.661181507840074,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.4466,
+      "step": 7142
+    },
+    {
+      "epoch": 0.6612740843603541,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.6078,
+      "step": 7143
+    },
+    {
+      "epoch": 0.6613666608806341,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5178,
+      "step": 7144
+    },
+    {
+      "epoch": 0.6614592374009142,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.02,
+      "loss": 1.5519,
+      "step": 7145
+    },
+    {
+      "epoch": 0.6615518139211942,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5726,
+      "step": 7146
+    },
+    {
+      "epoch": 0.6616443904414743,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5071,
+      "step": 7147
+    },
+    {
+      "epoch": 0.6617369669617543,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.5506,
+      "step": 7148
+    },
+    {
+      "epoch": 0.6618295434820344,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.5557,
+      "step": 7149
+    },
+    {
+      "epoch": 0.6619221200023144,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.02,
+      "loss": 1.598,
+      "step": 7150
+    },
+    {
+      "epoch": 0.6620146965225945,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5964,
+      "step": 7151
+    },
+    {
+      "epoch": 0.6621072730428745,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.02,
+      "loss": 1.5408,
+      "step": 7152
+    },
+    {
+      "epoch": 0.6621998495631546,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.5969,
+      "step": 7153
+    },
+    {
+      "epoch": 0.6622924260834346,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.02,
+      "loss": 1.5948,
+      "step": 7154
+    },
+    {
+      "epoch": 0.6623850026037146,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.02,
+      "loss": 1.5563,
+      "step": 7155
+    },
+    {
+      "epoch": 0.6624775791239946,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.02,
+      "loss": 1.5378,
+      "step": 7156
+    },
+    {
+      "epoch": 0.6625701556442747,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.02,
+      "loss": 1.5333,
+      "step": 7157
+    },
+    {
+      "epoch": 0.6626627321645547,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5619,
+      "step": 7158
+    },
+    {
+      "epoch": 0.6627553086848348,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5354,
+      "step": 7159
+    },
+    {
+      "epoch": 0.6628478852051148,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.5298,
+      "step": 7160
+    },
+    {
+      "epoch": 0.6629404617253949,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.5279,
+      "step": 7161
+    },
+    {
+      "epoch": 0.6630330382456749,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.5337,
+      "step": 7162
+    },
+    {
+      "epoch": 0.663125614765955,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.02,
+      "loss": 1.5593,
+      "step": 7163
+    },
+    {
+      "epoch": 0.663218191286235,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.5441,
+      "step": 7164
+    },
+    {
+      "epoch": 0.6633107678065151,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.02,
+      "loss": 1.6139,
+      "step": 7165
+    },
+    {
+      "epoch": 0.6634033443267952,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.02,
+      "loss": 1.4896,
+      "step": 7166
+    },
+    {
+      "epoch": 0.6634959208470752,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.6001,
+      "step": 7167
+    },
+    {
+      "epoch": 0.6635884973673553,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5397,
+      "step": 7168
+    },
+    {
+      "epoch": 0.6636810738876352,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.5601,
+      "step": 7169
+    },
+    {
+      "epoch": 0.6637736504079152,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5898,
+      "step": 7170
+    },
+    {
+      "epoch": 0.6638662269281953,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.02,
+      "loss": 1.5099,
+      "step": 7171
+    },
+    {
+      "epoch": 0.6639588034484754,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.5629,
+      "step": 7172
+    },
+    {
+      "epoch": 0.6640513799687554,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.4783,
+      "step": 7173
+    },
+    {
+      "epoch": 0.6641439564890355,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.02,
+      "loss": 1.4947,
+      "step": 7174
+    },
+    {
+      "epoch": 0.6642365330093155,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5053,
+      "step": 7175
+    },
+    {
+      "epoch": 0.6643291095295956,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5149,
+      "step": 7176
+    },
+    {
+      "epoch": 0.6644216860498756,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.5256,
+      "step": 7177
+    },
+    {
+      "epoch": 0.6645142625701557,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5664,
+      "step": 7178
+    },
+    {
+      "epoch": 0.6646068390904357,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.5148,
+      "step": 7179
+    },
+    {
+      "epoch": 0.6646994156107158,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5602,
+      "step": 7180
+    },
+    {
+      "epoch": 0.6647919921309958,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5878,
+      "step": 7181
+    },
+    {
+      "epoch": 0.6648845686512758,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5171,
+      "step": 7182
+    },
+    {
+      "epoch": 0.6649771451715558,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.02,
+      "loss": 1.5691,
+      "step": 7183
+    },
+    {
+      "epoch": 0.6650697216918359,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5745,
+      "step": 7184
+    },
+    {
+      "epoch": 0.6651622982121159,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.4759,
+      "step": 7185
+    },
+    {
+      "epoch": 0.665254874732396,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5892,
+      "step": 7186
+    },
+    {
+      "epoch": 0.665347451252676,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.5543,
+      "step": 7187
+    },
+    {
+      "epoch": 0.6654400277729561,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.502,
+      "step": 7188
+    },
+    {
+      "epoch": 0.6655326042932361,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.474,
+      "step": 7189
+    },
+    {
+      "epoch": 0.6656251808135162,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5121,
+      "step": 7190
+    },
+    {
+      "epoch": 0.6657177573337962,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.4957,
+      "step": 7191
+    },
+    {
+      "epoch": 0.6658103338540763,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.02,
+      "loss": 1.4405,
+      "step": 7192
+    },
+    {
+      "epoch": 0.6659029103743563,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.5286,
+      "step": 7193
+    },
+    {
+      "epoch": 0.6659954868946364,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5325,
+      "step": 7194
+    },
+    {
+      "epoch": 0.6660880634149164,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.02,
+      "loss": 1.5783,
+      "step": 7195
+    },
+    {
+      "epoch": 0.6661806399351964,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5453,
+      "step": 7196
+    },
+    {
+      "epoch": 0.6662732164554764,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5387,
+      "step": 7197
+    },
+    {
+      "epoch": 0.6663657929757565,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.571,
+      "step": 7198
+    },
+    {
+      "epoch": 0.6664583694960365,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.5366,
+      "step": 7199
+    },
+    {
+      "epoch": 0.6665509460163166,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5379,
+      "step": 7200
+    },
+    {
+      "epoch": 0.6666435225365966,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.4347,
+      "step": 7201
+    },
+    {
+      "epoch": 0.6667360990568767,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5078,
+      "step": 7202
+    },
+    {
+      "epoch": 0.6668286755771567,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.02,
+      "loss": 1.4877,
+      "step": 7203
+    },
+    {
+      "epoch": 0.6669212520974368,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.577,
+      "step": 7204
+    },
+    {
+      "epoch": 0.6670138286177169,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5395,
+      "step": 7205
+    },
+    {
+      "epoch": 0.6671064051379969,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5582,
+      "step": 7206
+    },
+    {
+      "epoch": 0.667198981658277,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.02,
+      "loss": 1.6177,
+      "step": 7207
+    },
+    {
+      "epoch": 0.667291558178557,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.02,
+      "loss": 1.5635,
+      "step": 7208
+    },
+    {
+      "epoch": 0.6673841346988371,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.02,
+      "loss": 1.4763,
+      "step": 7209
+    },
+    {
+      "epoch": 0.667476711219117,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.02,
+      "loss": 1.5029,
+      "step": 7210
+    },
+    {
+      "epoch": 0.667569287739397,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5508,
+      "step": 7211
+    },
+    {
+      "epoch": 0.6676618642596771,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.02,
+      "loss": 1.5336,
+      "step": 7212
+    },
+    {
+      "epoch": 0.6677544407799572,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.4986,
+      "step": 7213
+    },
+    {
+      "epoch": 0.6678470173002372,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.02,
+      "loss": 1.5035,
+      "step": 7214
+    },
+    {
+      "epoch": 0.6679395938205173,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.56,
+      "step": 7215
+    },
+    {
+      "epoch": 0.6680321703407973,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5499,
+      "step": 7216
+    },
+    {
+      "epoch": 0.6681247468610774,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.02,
+      "loss": 1.6123,
+      "step": 7217
+    },
+    {
+      "epoch": 0.6682173233813574,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.5434,
+      "step": 7218
+    },
+    {
+      "epoch": 0.6683098999016375,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.4776,
+      "step": 7219
+    },
+    {
+      "epoch": 0.6684024764219175,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.524,
+      "step": 7220
+    },
+    {
+      "epoch": 0.6684950529421976,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.525,
+      "step": 7221
+    },
+    {
+      "epoch": 0.6685876294624776,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.5257,
+      "step": 7222
+    },
+    {
+      "epoch": 0.6686802059827576,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5541,
+      "step": 7223
+    },
+    {
+      "epoch": 0.6687727825030376,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.5456,
+      "step": 7224
+    },
+    {
+      "epoch": 0.6688653590233177,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.02,
+      "loss": 1.4785,
+      "step": 7225
+    },
+    {
+      "epoch": 0.6689579355435977,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.5146,
+      "step": 7226
+    },
+    {
+      "epoch": 0.6690505120638778,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.4814,
+      "step": 7227
+    },
+    {
+      "epoch": 0.6691430885841578,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.4842,
+      "step": 7228
+    },
+    {
+      "epoch": 0.6692356651044379,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.4826,
+      "step": 7229
+    },
+    {
+      "epoch": 0.6693282416247179,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.02,
+      "loss": 1.5381,
+      "step": 7230
+    },
+    {
+      "epoch": 0.669420818144998,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.6165,
+      "step": 7231
+    },
+    {
+      "epoch": 0.669513394665278,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.4714,
+      "step": 7232
+    },
+    {
+      "epoch": 0.6696059711855581,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5282,
+      "step": 7233
+    },
+    {
+      "epoch": 0.6696985477058381,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.6161,
+      "step": 7234
+    },
+    {
+      "epoch": 0.6697911242261182,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.498,
+      "step": 7235
+    },
+    {
+      "epoch": 0.6698837007463982,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.465,
+      "step": 7236
+    },
+    {
+      "epoch": 0.6699762772666782,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5868,
+      "step": 7237
+    },
+    {
+      "epoch": 0.6700688537869582,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5231,
+      "step": 7238
+    },
+    {
+      "epoch": 0.6701614303072383,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5378,
+      "step": 7239
+    },
+    {
+      "epoch": 0.6702540068275183,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.5046,
+      "step": 7240
+    },
+    {
+      "epoch": 0.6703465833477984,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5005,
+      "step": 7241
+    },
+    {
+      "epoch": 0.6704391598680784,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.4253,
+      "step": 7242
+    },
+    {
+      "epoch": 0.6705317363883585,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.4826,
+      "step": 7243
+    },
+    {
+      "epoch": 0.6706243129086386,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.4633,
+      "step": 7244
+    },
+    {
+      "epoch": 0.6707168894289186,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.4599,
+      "step": 7245
+    },
+    {
+      "epoch": 0.6708094659491987,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5515,
+      "step": 7246
+    },
+    {
+      "epoch": 0.6709020424694787,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5312,
+      "step": 7247
+    },
+    {
+      "epoch": 0.6709946189897588,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.02,
+      "loss": 1.488,
+      "step": 7248
+    },
+    {
+      "epoch": 0.6710871955100388,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5267,
+      "step": 7249
+    },
+    {
+      "epoch": 0.6711797720303188,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.4875,
+      "step": 7250
+    },
+    {
+      "epoch": 0.6712723485505988,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5055,
+      "step": 7251
+    },
+    {
+      "epoch": 0.6713649250708789,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.5251,
+      "step": 7252
+    },
+    {
+      "epoch": 0.6714575015911589,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.02,
+      "loss": 1.4931,
+      "step": 7253
+    },
+    {
+      "epoch": 0.671550078111439,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5863,
+      "step": 7254
+    },
+    {
+      "epoch": 0.671642654631719,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.02,
+      "loss": 1.5178,
+      "step": 7255
+    },
+    {
+      "epoch": 0.6717352311519991,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5436,
+      "step": 7256
+    },
+    {
+      "epoch": 0.6718278076722791,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.4701,
+      "step": 7257
+    },
+    {
+      "epoch": 0.6719203841925592,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5031,
+      "step": 7258
+    },
+    {
+      "epoch": 0.6720129607128392,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5146,
+      "step": 7259
+    },
+    {
+      "epoch": 0.6721055372331193,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5435,
+      "step": 7260
+    },
+    {
+      "epoch": 0.6721981137533993,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.4942,
+      "step": 7261
+    },
+    {
+      "epoch": 0.6722906902736794,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.4986,
+      "step": 7262
+    },
+    {
+      "epoch": 0.6723832667939594,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.4925,
+      "step": 7263
+    },
+    {
+      "epoch": 0.6724758433142394,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5009,
+      "step": 7264
+    },
+    {
+      "epoch": 0.6725684198345194,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5492,
+      "step": 7265
+    },
+    {
+      "epoch": 0.6726609963547995,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.529,
+      "step": 7266
+    },
+    {
+      "epoch": 0.6727535728750795,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.4179,
+      "step": 7267
+    },
+    {
+      "epoch": 0.6728461493953596,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.02,
+      "loss": 1.4933,
+      "step": 7268
+    },
+    {
+      "epoch": 0.6729387259156396,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.5538,
+      "step": 7269
+    },
+    {
+      "epoch": 0.6730313024359197,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.02,
+      "loss": 1.5521,
+      "step": 7270
+    },
+    {
+      "epoch": 0.6731238789561997,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5315,
+      "step": 7271
+    },
+    {
+      "epoch": 0.6732164554764798,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5084,
+      "step": 7272
+    },
+    {
+      "epoch": 0.6733090319967598,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.5228,
+      "step": 7273
+    },
+    {
+      "epoch": 0.6734016085170399,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5578,
+      "step": 7274
+    },
+    {
+      "epoch": 0.67349418503732,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.4632,
+      "step": 7275
+    },
+    {
+      "epoch": 0.6735867615576,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5971,
+      "step": 7276
+    },
+    {
+      "epoch": 0.67367933807788,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5007,
+      "step": 7277
+    },
+    {
+      "epoch": 0.67377191459816,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.02,
+      "loss": 1.5587,
+      "step": 7278
+    },
+    {
+      "epoch": 0.67386449111844,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.4848,
+      "step": 7279
+    },
+    {
+      "epoch": 0.6739570676387201,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.4997,
+      "step": 7280
+    },
+    {
+      "epoch": 0.6740496441590001,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.5154,
+      "step": 7281
+    },
+    {
+      "epoch": 0.6741422206792802,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.5275,
+      "step": 7282
+    },
+    {
+      "epoch": 0.6742347971995603,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5133,
+      "step": 7283
+    },
+    {
+      "epoch": 0.6743273737198403,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.02,
+      "loss": 1.502,
+      "step": 7284
+    },
+    {
+      "epoch": 0.6744199502401204,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5601,
+      "step": 7285
+    },
+    {
+      "epoch": 0.6745125267604004,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5227,
+      "step": 7286
+    },
+    {
+      "epoch": 0.6746051032806805,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.4801,
+      "step": 7287
+    },
+    {
+      "epoch": 0.6746976798009605,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5126,
+      "step": 7288
+    },
+    {
+      "epoch": 0.6747902563212406,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5353,
+      "step": 7289
+    },
+    {
+      "epoch": 0.6748828328415206,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5646,
+      "step": 7290
+    },
+    {
+      "epoch": 0.6749754093618006,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.4658,
+      "step": 7291
+    },
+    {
+      "epoch": 0.6750679858820806,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5118,
+      "step": 7292
+    },
+    {
+      "epoch": 0.6751605624023607,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.596,
+      "step": 7293
+    },
+    {
+      "epoch": 0.6752531389226407,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5276,
+      "step": 7294
+    },
+    {
+      "epoch": 0.6753457154429208,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5659,
+      "step": 7295
+    },
+    {
+      "epoch": 0.6754382919632008,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5002,
+      "step": 7296
+    },
+    {
+      "epoch": 0.6755308684834809,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.02,
+      "loss": 1.5878,
+      "step": 7297
+    },
+    {
+      "epoch": 0.6756234450037609,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5248,
+      "step": 7298
+    },
+    {
+      "epoch": 0.675716021524041,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.4757,
+      "step": 7299
+    },
+    {
+      "epoch": 0.675808598044321,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.02,
+      "loss": 1.477,
+      "step": 7300
+    },
+    {
+      "epoch": 0.6759011745646011,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5331,
+      "step": 7301
+    },
+    {
+      "epoch": 0.6759937510848811,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.4789,
+      "step": 7302
+    },
+    {
+      "epoch": 0.6760863276051612,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5093,
+      "step": 7303
+    },
+    {
+      "epoch": 0.6761789041254412,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.5175,
+      "step": 7304
+    },
+    {
+      "epoch": 0.6762714806457212,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.4642,
+      "step": 7305
+    },
+    {
+      "epoch": 0.6763640571660012,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5742,
+      "step": 7306
+    },
+    {
+      "epoch": 0.6764566336862813,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.6081,
+      "step": 7307
+    },
+    {
+      "epoch": 0.6765492102065613,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5655,
+      "step": 7308
+    },
+    {
+      "epoch": 0.6766417867268414,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.02,
+      "loss": 1.5124,
+      "step": 7309
+    },
+    {
+      "epoch": 0.6767343632471214,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.4942,
+      "step": 7310
+    },
+    {
+      "epoch": 0.6768269397674015,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5616,
+      "step": 7311
+    },
+    {
+      "epoch": 0.6769195162876815,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.5361,
+      "step": 7312
+    },
+    {
+      "epoch": 0.6770120928079616,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5343,
+      "step": 7313
+    },
+    {
+      "epoch": 0.6771046693282416,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5188,
+      "step": 7314
+    },
+    {
+      "epoch": 0.6771972458485217,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.56,
+      "step": 7315
+    },
+    {
+      "epoch": 0.6772898223688018,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.02,
+      "loss": 1.5331,
+      "step": 7316
+    },
+    {
+      "epoch": 0.6773823988890818,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.5047,
+      "step": 7317
+    },
+    {
+      "epoch": 0.6774749754093617,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.4803,
+      "step": 7318
+    },
+    {
+      "epoch": 0.6775675519296418,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5665,
+      "step": 7319
+    },
+    {
+      "epoch": 0.6776601284499219,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.02,
+      "loss": 1.5727,
+      "step": 7320
+    },
+    {
+      "epoch": 0.6777527049702019,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5347,
+      "step": 7321
+    },
+    {
+      "epoch": 0.677845281490482,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.4872,
+      "step": 7322
+    },
+    {
+      "epoch": 0.677937858010762,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.02,
+      "loss": 1.5092,
+      "step": 7323
+    },
+    {
+      "epoch": 0.6780304345310421,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.02,
+      "loss": 1.4781,
+      "step": 7324
+    },
+    {
+      "epoch": 0.6781230110513221,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5573,
+      "step": 7325
+    },
+    {
+      "epoch": 0.6782155875716022,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.02,
+      "loss": 1.4438,
+      "step": 7326
+    },
+    {
+      "epoch": 0.6783081640918822,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.02,
+      "loss": 1.552,
+      "step": 7327
+    },
+    {
+      "epoch": 0.6784007406121623,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.02,
+      "loss": 1.5302,
+      "step": 7328
+    },
+    {
+      "epoch": 0.6784933171324423,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5429,
+      "step": 7329
+    },
+    {
+      "epoch": 0.6785858936527224,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.4792,
+      "step": 7330
+    },
+    {
+      "epoch": 0.6786784701730024,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.5632,
+      "step": 7331
+    },
+    {
+      "epoch": 0.6787710466932824,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5327,
+      "step": 7332
+    },
+    {
+      "epoch": 0.6788636232135624,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.5624,
+      "step": 7333
+    },
+    {
+      "epoch": 0.6789561997338425,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5891,
+      "step": 7334
+    },
+    {
+      "epoch": 0.6790487762541225,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5289,
+      "step": 7335
+    },
+    {
+      "epoch": 0.6791413527744026,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.4554,
+      "step": 7336
+    },
+    {
+      "epoch": 0.6792339292946826,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.525,
+      "step": 7337
+    },
+    {
+      "epoch": 0.6793265058149627,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.5914,
+      "step": 7338
+    },
+    {
+      "epoch": 0.6794190823352427,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5697,
+      "step": 7339
+    },
+    {
+      "epoch": 0.6795116588555228,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.4911,
+      "step": 7340
+    },
+    {
+      "epoch": 0.6796042353758028,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5078,
+      "step": 7341
+    },
+    {
+      "epoch": 0.6796968118960829,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5713,
+      "step": 7342
+    },
+    {
+      "epoch": 0.6797893884163629,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.02,
+      "loss": 1.4377,
+      "step": 7343
+    },
+    {
+      "epoch": 0.679881964936643,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.4304,
+      "step": 7344
+    },
+    {
+      "epoch": 0.679974541456923,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.02,
+      "loss": 1.6018,
+      "step": 7345
+    },
+    {
+      "epoch": 0.680067117977203,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.6202,
+      "step": 7346
+    },
+    {
+      "epoch": 0.680159694497483,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.02,
+      "loss": 1.489,
+      "step": 7347
+    },
+    {
+      "epoch": 0.6802522710177631,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5354,
+      "step": 7348
+    },
+    {
+      "epoch": 0.6803448475380431,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.4541,
+      "step": 7349
+    },
+    {
+      "epoch": 0.6804374240583232,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5599,
+      "step": 7350
+    },
+    {
+      "epoch": 0.6805300005786032,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5078,
+      "step": 7351
+    },
+    {
+      "epoch": 0.6806225770988833,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.4506,
+      "step": 7352
+    },
+    {
+      "epoch": 0.6807151536191633,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5107,
+      "step": 7353
+    },
+    {
+      "epoch": 0.6808077301394434,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5214,
+      "step": 7354
+    },
+    {
+      "epoch": 0.6809003066597235,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.4983,
+      "step": 7355
+    },
+    {
+      "epoch": 0.6809928831800035,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.4641,
+      "step": 7356
+    },
+    {
+      "epoch": 0.6810854597002836,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5502,
+      "step": 7357
+    },
+    {
+      "epoch": 0.6811780362205636,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5399,
+      "step": 7358
+    },
+    {
+      "epoch": 0.6812706127408436,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.5021,
+      "step": 7359
+    },
+    {
+      "epoch": 0.6813631892611236,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.469,
+      "step": 7360
+    },
+    {
+      "epoch": 0.6814557657814037,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5576,
+      "step": 7361
+    },
+    {
+      "epoch": 0.6815483423016837,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5761,
+      "step": 7362
+    },
+    {
+      "epoch": 0.6816409188219638,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.488,
+      "step": 7363
+    },
+    {
+      "epoch": 0.6817334953422438,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.4215,
+      "step": 7364
+    },
+    {
+      "epoch": 0.6818260718625239,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5353,
+      "step": 7365
+    },
+    {
+      "epoch": 0.6819186483828039,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5073,
+      "step": 7366
+    },
+    {
+      "epoch": 0.682011224903084,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.5642,
+      "step": 7367
+    },
+    {
+      "epoch": 0.682103801423364,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.4755,
+      "step": 7368
+    },
+    {
+      "epoch": 0.6821963779436441,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.02,
+      "loss": 1.5806,
+      "step": 7369
+    },
+    {
+      "epoch": 0.6822889544639241,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5347,
+      "step": 7370
+    },
+    {
+      "epoch": 0.6823815309842042,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.5303,
+      "step": 7371
+    },
+    {
+      "epoch": 0.6824741075044842,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5137,
+      "step": 7372
+    },
+    {
+      "epoch": 0.6825666840247642,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5442,
+      "step": 7373
+    },
+    {
+      "epoch": 0.6826592605450442,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5166,
+      "step": 7374
+    },
+    {
+      "epoch": 0.6827518370653243,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.4808,
+      "step": 7375
+    },
+    {
+      "epoch": 0.6828444135856043,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.497,
+      "step": 7376
+    },
+    {
+      "epoch": 0.6829369901058844,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.02,
+      "loss": 1.4695,
+      "step": 7377
+    },
+    {
+      "epoch": 0.6830295666261644,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.521,
+      "step": 7378
+    },
+    {
+      "epoch": 0.6831221431464445,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.02,
+      "loss": 1.4887,
+      "step": 7379
+    },
+    {
+      "epoch": 0.6832147196667245,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.02,
+      "loss": 1.4984,
+      "step": 7380
+    },
+    {
+      "epoch": 0.6833072961870046,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5001,
+      "step": 7381
+    },
+    {
+      "epoch": 0.6833998727072846,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5183,
+      "step": 7382
+    },
+    {
+      "epoch": 0.6834924492275647,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.495,
+      "step": 7383
+    },
+    {
+      "epoch": 0.6835850257478447,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5766,
+      "step": 7384
+    },
+    {
+      "epoch": 0.6836776022681248,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.02,
+      "loss": 1.5788,
+      "step": 7385
+    },
+    {
+      "epoch": 0.6837701787884047,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.5495,
+      "step": 7386
+    },
+    {
+      "epoch": 0.6838627553086848,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.4891,
+      "step": 7387
+    },
+    {
+      "epoch": 0.6839553318289648,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.5912,
+      "step": 7388
+    },
+    {
+      "epoch": 0.6840479083492449,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5436,
+      "step": 7389
+    },
+    {
+      "epoch": 0.684140484869525,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.02,
+      "loss": 1.5307,
+      "step": 7390
+    },
+    {
+      "epoch": 0.684233061389805,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5441,
+      "step": 7391
+    },
+    {
+      "epoch": 0.684325637910085,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.5218,
+      "step": 7392
+    },
+    {
+      "epoch": 0.6844182144303651,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.02,
+      "loss": 1.6006,
+      "step": 7393
+    },
+    {
+      "epoch": 0.6845107909506452,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.02,
+      "loss": 1.5419,
+      "step": 7394
+    },
+    {
+      "epoch": 0.6846033674709252,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.4775,
+      "step": 7395
+    },
+    {
+      "epoch": 0.6846959439912053,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.545,
+      "step": 7396
+    },
+    {
+      "epoch": 0.6847885205114853,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.5628,
+      "step": 7397
+    },
+    {
+      "epoch": 0.6848810970317654,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5521,
+      "step": 7398
+    },
+    {
+      "epoch": 0.6849736735520454,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.514,
+      "step": 7399
+    },
+    {
+      "epoch": 0.6850662500723254,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5556,
+      "step": 7400
+    },
+    {
+      "epoch": 0.6851588265926054,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5129,
+      "step": 7401
+    },
+    {
+      "epoch": 0.6852514031128855,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.5262,
+      "step": 7402
+    },
+    {
+      "epoch": 0.6853439796331655,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.02,
+      "loss": 1.5519,
+      "step": 7403
+    },
+    {
+      "epoch": 0.6854365561534456,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.02,
+      "loss": 1.5364,
+      "step": 7404
+    },
+    {
+      "epoch": 0.6855291326737256,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.02,
+      "loss": 1.4768,
+      "step": 7405
+    },
+    {
+      "epoch": 0.6856217091940057,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.02,
+      "loss": 1.5421,
+      "step": 7406
+    },
+    {
+      "epoch": 0.6857142857142857,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.02,
+      "loss": 1.5112,
+      "step": 7407
+    },
+    {
+      "epoch": 0.6858068622345658,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.5195,
+      "step": 7408
+    },
+    {
+      "epoch": 0.6858994387548458,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.02,
+      "loss": 1.5407,
+      "step": 7409
+    },
+    {
+      "epoch": 0.6859920152751259,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.02,
+      "loss": 1.4589,
+      "step": 7410
+    },
+    {
+      "epoch": 0.6860845917954059,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.02,
+      "loss": 1.5148,
+      "step": 7411
+    },
+    {
+      "epoch": 0.686177168315686,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.02,
+      "loss": 1.4883,
+      "step": 7412
+    },
+    {
+      "epoch": 0.686269744835966,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.5213,
+      "step": 7413
+    },
+    {
+      "epoch": 0.686362321356246,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.02,
+      "loss": 1.5104,
+      "step": 7414
+    },
+    {
+      "epoch": 0.686454897876526,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.548,
+      "step": 7415
+    },
+    {
+      "epoch": 0.6865474743968061,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.5317,
+      "step": 7416
+    },
+    {
+      "epoch": 0.6866400509170861,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.02,
+      "loss": 1.516,
+      "step": 7417
+    },
+    {
+      "epoch": 0.6867326274373662,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5046,
+      "step": 7418
+    },
+    {
+      "epoch": 0.6868252039576462,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.02,
+      "loss": 1.4812,
+      "step": 7419
+    },
+    {
+      "epoch": 0.6869177804779263,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5018,
+      "step": 7420
+    },
+    {
+      "epoch": 0.6870103569982063,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.5047,
+      "step": 7421
+    },
+    {
+      "epoch": 0.6871029335184864,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.02,
+      "loss": 1.5227,
+      "step": 7422
+    },
+    {
+      "epoch": 0.6871955100387664,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.02,
+      "loss": 1.5506,
+      "step": 7423
+    },
+    {
+      "epoch": 0.6872880865590465,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.4786,
+      "step": 7424
+    },
+    {
+      "epoch": 0.6873806630793265,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.02,
+      "loss": 1.4945,
+      "step": 7425
+    },
+    {
+      "epoch": 0.6874732395996066,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.02,
+      "loss": 1.4975,
+      "step": 7426
+    },
+    {
+      "epoch": 0.6875658161198865,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.6003,
+      "step": 7427
+    },
+    {
+      "epoch": 0.6876583926401666,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.02,
+      "loss": 1.4866,
+      "step": 7428
+    },
+    {
+      "epoch": 0.6877509691604466,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.02,
+      "loss": 1.4946,
+      "step": 7429
+    },
+    {
+      "epoch": 0.6878435456807267,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.02,
+      "loss": 1.501,
+      "step": 7430
+    },
+    {
+      "epoch": 0.6879361222010068,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5101,
+      "step": 7431
+    },
+    {
+      "epoch": 0.6880286987212868,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.02,
+      "loss": 1.4444,
+      "step": 7432
+    },
+    {
+      "epoch": 0.6881212752415669,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.5034,
+      "step": 7433
+    },
+    {
+      "epoch": 0.6882138517618469,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.02,
+      "loss": 1.5182,
+      "step": 7434
+    },
+    {
+      "epoch": 0.688306428282127,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.02,
+      "loss": 1.6076,
+      "step": 7435
+    },
+    {
+      "epoch": 0.688399004802407,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.02,
+      "loss": 1.4451,
+      "step": 7436
+    },
+    {
+      "epoch": 0.6884915813226871,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.513,
+      "step": 7437
+    },
+    {
+      "epoch": 0.6885841578429671,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.02,
+      "loss": 1.5914,
+      "step": 7438
+    },
+    {
+      "epoch": 0.6886767343632472,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.02,
+      "loss": 1.5765,
+      "step": 7439
+    },
+    {
+      "epoch": 0.6887693108835272,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.02,
+      "loss": 1.5209,
+      "step": 7440
     }
   ],
   "logging_steps": 1,
@@ -41365,13 +52117,13 @@
         "should_epoch_stop": false,
         "should_evaluate": false,
         "should_log": false,
-        "should_save": true,
+        "should_save": false,
         "should_training_stop": false
       },
       "attributes": {}
     }
   },
-  "total_flos": 5.200473320974384e+18,
+  "total_flos": 6.554468380893837e+18,
   "train_batch_size": 8,
   "trial_name": null,
   "trial_params": null