diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,9 +1,9 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.19469457288878073,
+  "epoch": 0.7787782915551229,
   "eval_steps": 500,
-  "global_step": 800,
+  "global_step": 3200,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -4807,6 +4807,14406 @@
       "learning_rate": 1.8613233568836164e-05,
       "loss": 0.8159,
       "step": 800
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.860922560125301e-05,
+      "loss": 0.814,
+      "step": 801
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8605212283012263e-05,
+      "loss": 0.8206,
+      "step": 802
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860119361660821e-05,
+      "loss": 0.8262,
+      "step": 803
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859716960453846e-05,
+      "loss": 0.8196,
+      "step": 804
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8593140249303957e-05,
+      "loss": 0.7827,
+      "step": 805
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858910555340895e-05,
+      "loss": 0.7998,
+      "step": 806
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8585065519361017e-05,
+      "loss": 0.8589,
+      "step": 807
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8581020149671046e-05,
+      "loss": 0.7922,
+      "step": 808
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8576969446853248e-05,
+      "loss": 0.8652,
+      "step": 809
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8572913413425145e-05,
+      "loss": 0.7852,
+      "step": 810
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8568852051907575e-05,
+      "loss": 0.7988,
+      "step": 811
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8564785364824684e-05,
+      "loss": 0.8267,
+      "step": 812
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8560713354703933e-05,
+      "loss": 0.8181,
+      "step": 813
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855663602407608e-05,
+      "loss": 0.8506,
+      "step": 814
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8552553375475205e-05,
+      "loss": 0.804,
+      "step": 815
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854846541143868e-05,
+      "loss": 0.8328,
+      "step": 816
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8544372134507184e-05,
+      "loss": 0.8506,
+      "step": 817
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854027354722471e-05,
+      "loss": 0.8057,
+      "step": 818
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8536169652138532e-05,
+      "loss": 0.8066,
+      "step": 819
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8532060451799238e-05,
+      "loss": 0.8091,
+      "step": 820
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8527945948760702e-05,
+      "loss": 0.8083,
+      "step": 821
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8523826145580103e-05,
+      "loss": 0.833,
+      "step": 822
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851970104481791e-05,
+      "loss": 0.783,
+      "step": 823
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8515570649037885e-05,
+      "loss": 0.8501,
+      "step": 824
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8511434960807077e-05,
+      "loss": 0.8647,
+      "step": 825
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850729398269583e-05,
+      "loss": 0.8132,
+      "step": 826
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8503147717277774e-05,
+      "loss": 0.8403,
+      "step": 827
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8498996167129825e-05,
+      "loss": 0.8474,
+      "step": 828
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8494839334832176e-05,
+      "loss": 0.8232,
+      "step": 829
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849067722296832e-05,
+      "loss": 0.8081,
+      "step": 830
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8486509834125013e-05,
+      "loss": 0.8499,
+      "step": 831
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8482337170892308e-05,
+      "loss": 0.7856,
+      "step": 832
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8478159235863514e-05,
+      "loss": 0.8401,
+      "step": 833
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847397603163524e-05,
+      "loss": 0.8352,
+      "step": 834
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8469787560807357e-05,
+      "loss": 0.8008,
+      "step": 835
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8465593825983005e-05,
+      "loss": 0.8062,
+      "step": 836
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846139482976861e-05,
+      "loss": 0.8076,
+      "step": 837
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8457190574773854e-05,
+      "loss": 0.7815,
+      "step": 838
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8452981063611696e-05,
+      "loss": 0.8152,
+      "step": 839
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8448766298898356e-05,
+      "loss": 0.8032,
+      "step": 840
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8444546283253325e-05,
+      "loss": 0.8203,
+      "step": 841
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8440321019299355e-05,
+      "loss": 0.7712,
+      "step": 842
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8436090509662456e-05,
+      "loss": 0.8557,
+      "step": 843
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8431854756971905e-05,
+      "loss": 0.8418,
+      "step": 844
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842761376386023e-05,
+      "loss": 0.7876,
+      "step": 845
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8423367532963224e-05,
+      "loss": 0.8142,
+      "step": 846
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8419116066919928e-05,
+      "loss": 0.8499,
+      "step": 847
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8414859368372643e-05,
+      "loss": 0.8042,
+      "step": 848
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8410597439966915e-05,
+      "loss": 0.7361,
+      "step": 849
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8406330284351548e-05,
+      "loss": 0.7937,
+      "step": 850
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840205790417859e-05,
+      "loss": 0.7932,
+      "step": 851
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8397780302103337e-05,
+      "loss": 0.8413,
+      "step": 852
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839349748078433e-05,
+      "loss": 0.8257,
+      "step": 853
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8389209442883355e-05,
+      "loss": 0.8867,
+      "step": 854
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838491619106544e-05,
+      "loss": 0.7827,
+      "step": 855
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838061772799885e-05,
+      "loss": 0.8936,
+      "step": 856
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8376314056355095e-05,
+      "loss": 0.8494,
+      "step": 857
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837200517880892e-05,
+      "loss": 0.8174,
+      "step": 858
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83676910980383e-05,
+      "loss": 0.8096,
+      "step": 859
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836337181672445e-05,
+      "loss": 0.8418,
+      "step": 860
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8359047337551815e-05,
+      "loss": 0.824,
+      "step": 861
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835471766320807e-05,
+      "loss": 0.7776,
+      "step": 862
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8350382796384125e-05,
+      "loss": 0.8552,
+      "step": 863
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8346042739774105e-05,
+      "loss": 0.7698,
+      "step": 864
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8341697496075372e-05,
+      "loss": 0.8257,
+      "step": 865
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8337347067988506e-05,
+      "loss": 0.8386,
+      "step": 866
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8332991458217304e-05,
+      "loss": 0.8098,
+      "step": 867
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83286306694688e-05,
+      "loss": 0.8401,
+      "step": 868
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8324264704453228e-05,
+      "loss": 0.7864,
+      "step": 869
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8319893565884048e-05,
+      "loss": 0.8149,
+      "step": 870
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8315517256477942e-05,
+      "loss": 0.7883,
+      "step": 871
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831113577895479e-05,
+      "loss": 0.7925,
+      "step": 872
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83067491360377e-05,
+      "loss": 0.7847,
+      "step": 873
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8302357330452972e-05,
+      "loss": 0.8093,
+      "step": 874
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8297960364930136e-05,
+      "loss": 0.8008,
+      "step": 875
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8293558242201914e-05,
+      "loss": 0.8398,
+      "step": 876
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828915096500424e-05,
+      "loss": 0.7949,
+      "step": 877
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8284738536076244e-05,
+      "loss": 0.7913,
+      "step": 878
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8280320958160268e-05,
+      "loss": 0.8152,
+      "step": 879
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8275898234001844e-05,
+      "loss": 0.8264,
+      "step": 880
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827147036634971e-05,
+      "loss": 0.7429,
+      "step": 881
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.82670373579558e-05,
+      "loss": 0.8562,
+      "step": 882
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8262599211575235e-05,
+      "loss": 0.8489,
+      "step": 883
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8258155929966344e-05,
+      "loss": 0.8743,
+      "step": 884
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825370751589063e-05,
+      "loss": 0.8564,
+      "step": 885
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8249253972112805e-05,
+      "loss": 0.8501,
+      "step": 886
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8244795301400745e-05,
+      "loss": 0.8381,
+      "step": 887
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8240331506525537e-05,
+      "loss": 0.8381,
+      "step": 888
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823586259026144e-05,
+      "loss": 0.8059,
+      "step": 889
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8231388555385897e-05,
+      "loss": 0.7776,
+      "step": 890
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8226909404679526e-05,
+      "loss": 0.8486,
+      "step": 891
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8222425140926145e-05,
+      "loss": 0.7549,
+      "step": 892
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8217935766912727e-05,
+      "loss": 0.8354,
+      "step": 893
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8213441285429435e-05,
+      "loss": 0.7766,
+      "step": 894
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8208941699269598e-05,
+      "loss": 0.8501,
+      "step": 895
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820443701122972e-05,
+      "loss": 0.7942,
+      "step": 896
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8199927224109487e-05,
+      "loss": 0.7795,
+      "step": 897
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819541234071173e-05,
+      "loss": 0.8591,
+      "step": 898
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8190892363842474e-05,
+      "loss": 0.8518,
+      "step": 899
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818636729631089e-05,
+      "loss": 0.8054,
+      "step": 900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8181837140929322e-05,
+      "loss": 0.8237,
+      "step": 901
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8177301900513275e-05,
+      "loss": 0.7744,
+      "step": 902
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817276157788141e-05,
+      "loss": 0.8047,
+      "step": 903
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8168216175855557e-05,
+      "loss": 0.8367,
+      "step": 904
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816366569726069e-05,
+      "loss": 0.8474,
+      "step": 905
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8159110144924943e-05,
+      "loss": 0.7988,
+      "step": 906
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815454952167961e-05,
+      "loss": 0.812,
+      "step": 907
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8149983830359128e-05,
+      "loss": 0.7825,
+      "step": 908
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814541307380109e-05,
+      "loss": 0.8386,
+      "step": 909
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8140837254846226e-05,
+      "loss": 0.762,
+      "step": 910
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8136256376338427e-05,
+      "loss": 0.7681,
+      "step": 911
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813167044112472e-05,
+      "loss": 0.7871,
+      "step": 912
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8127079452055274e-05,
+      "loss": 0.8652,
+      "step": 913
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8122483411983403e-05,
+      "loss": 0.8503,
+      "step": 914
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8117882323765563e-05,
+      "loss": 0.8093,
+      "step": 915
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811327619026134e-05,
+      "loss": 0.8347,
+      "step": 916
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8108665014333457e-05,
+      "loss": 0.8345,
+      "step": 917
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810404879884777e-05,
+      "loss": 0.7732,
+      "step": 918
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8099427546673277e-05,
+      "loss": 0.8269,
+      "step": 919
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8094801260682095e-05,
+      "loss": 0.8269,
+      "step": 920
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 0.8389,
+      "step": 921
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8085533598753792e-05,
+      "loss": 0.7998,
+      "step": 922
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808089222857655e-05,
+      "loss": 0.8325,
+      "step": 923
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8076245836102368e-05,
+      "loss": 0.8003,
+      "step": 924
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8071594424218998e-05,
+      "loss": 0.8848,
+      "step": 925
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8066937995817296e-05,
+      "loss": 0.7681,
+      "step": 926
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8062276553791255e-05,
+      "loss": 0.8381,
+      "step": 927
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8057610101037975e-05,
+      "loss": 0.8079,
+      "step": 928
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805293864045766e-05,
+      "loss": 0.8406,
+      "step": 929
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8048262174953643e-05,
+      "loss": 0.8303,
+      "step": 930
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8043580707432356e-05,
+      "loss": 0.8181,
+      "step": 931
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8038894240803347e-05,
+      "loss": 0.874,
+      "step": 932
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803420277797927e-05,
+      "loss": 0.8608,
+      "step": 933
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8029506321875875e-05,
+      "loss": 0.8132,
+      "step": 934
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8024804875412032e-05,
+      "loss": 0.8311,
+      "step": 935
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802009844150969e-05,
+      "loss": 0.8171,
+      "step": 936
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8015387023093924e-05,
+      "loss": 0.8059,
+      "step": 937
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8010670623092887e-05,
+      "loss": 0.7944,
+      "step": 938
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8005949244437836e-05,
+      "loss": 0.8328,
+      "step": 939
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8001222890063122e-05,
+      "loss": 0.8564,
+      "step": 940
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7996491562906186e-05,
+      "loss": 0.7524,
+      "step": 941
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799175526590756e-05,
+      "loss": 0.8269,
+      "step": 942
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7987014002010864e-05,
+      "loss": 0.8181,
+      "step": 943
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7982267774162813e-05,
+      "loss": 0.8306,
+      "step": 944
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7977516585313196e-05,
+      "loss": 0.8218,
+      "step": 945
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797276043841489e-05,
+      "loss": 0.8574,
+      "step": 946
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7967999336423857e-05,
+      "loss": 0.8313,
+      "step": 947
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7963233282299132e-05,
+      "loss": 0.8145,
+      "step": 948
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7958462279002834e-05,
+      "loss": 0.8167,
+      "step": 949
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795368632950015e-05,
+      "loss": 0.7971,
+      "step": 950
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7948905436759346e-05,
+      "loss": 0.833,
+      "step": 951
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7944119603751765e-05,
+      "loss": 0.8164,
+      "step": 952
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7939328833451814e-05,
+      "loss": 0.7822,
+      "step": 953
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793453312883697e-05,
+      "loss": 0.8132,
+      "step": 954
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7929732492887774e-05,
+      "loss": 0.8047,
+      "step": 955
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7924926928587834e-05,
+      "loss": 0.8176,
+      "step": 956
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7920116438923825e-05,
+      "loss": 0.8157,
+      "step": 957
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791530102688548e-05,
+      "loss": 0.8711,
+      "step": 958
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791048069546559e-05,
+      "loss": 0.8269,
+      "step": 959
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790565544766e-05,
+      "loss": 0.7471,
+      "step": 960
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7900825286467622e-05,
+      "loss": 0.7649,
+      "step": 961
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7895990214890408e-05,
+      "loss": 0.792,
+      "step": 962
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7891150235933376e-05,
+      "loss": 0.8254,
+      "step": 963
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788630535260458e-05,
+      "loss": 0.8225,
+      "step": 964
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788145556791513e-05,
+      "loss": 0.8093,
+      "step": 965
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787660088487918e-05,
+      "loss": 0.8389,
+      "step": 966
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787174130651393e-05,
+      "loss": 0.7656,
+      "step": 967
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7866876835839625e-05,
+      "loss": 0.8455,
+      "step": 968
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786200747587954e-05,
+      "loss": 0.7798,
+      "step": 969
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785713322966e-05,
+      "loss": 0.8152,
+      "step": 970
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7852254100210363e-05,
+      "loss": 0.8374,
+      "step": 971
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784737009056302e-05,
+      "loss": 0.7993,
+      "step": 972
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7842481203753395e-05,
+      "loss": 0.821,
+      "step": 973
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7837587442819946e-05,
+      "loss": 0.7593,
+      "step": 974
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7832688810804162e-05,
+      "loss": 0.7852,
+      "step": 975
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7827785310750552e-05,
+      "loss": 0.8196,
+      "step": 976
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7822876945706657e-05,
+      "loss": 0.801,
+      "step": 977
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781796371872304e-05,
+      "loss": 0.8162,
+      "step": 978
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7813045632853284e-05,
+      "loss": 0.8298,
+      "step": 979
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7808122691153996e-05,
+      "loss": 0.7639,
+      "step": 980
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780319489668479e-05,
+      "loss": 0.7432,
+      "step": 981
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7798262252508312e-05,
+      "loss": 0.8027,
+      "step": 982
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7793324761690217e-05,
+      "loss": 0.8335,
+      "step": 983
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7788382427299158e-05,
+      "loss": 0.803,
+      "step": 984
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7783435252406815e-05,
+      "loss": 0.8008,
+      "step": 985
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7778483240087873e-05,
+      "loss": 0.7844,
+      "step": 986
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7773526393420017e-05,
+      "loss": 0.814,
+      "step": 987
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7768564715483946e-05,
+      "loss": 0.8044,
+      "step": 988
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776359820936335e-05,
+      "loss": 0.8569,
+      "step": 989
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7758626878144932e-05,
+      "loss": 0.8184,
+      "step": 990
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7753650724918388e-05,
+      "loss": 0.832,
+      "step": 991
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774866975277641e-05,
+      "loss": 0.7705,
+      "step": 992
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7743683964814682e-05,
+      "loss": 0.8169,
+      "step": 993
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7738693364131893e-05,
+      "loss": 0.8477,
+      "step": 994
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773369795382971e-05,
+      "loss": 0.8652,
+      "step": 995
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7728697737012794e-05,
+      "loss": 0.8318,
+      "step": 996
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7723692716788793e-05,
+      "loss": 0.8538,
+      "step": 997
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771868289626834e-05,
+      "loss": 0.8335,
+      "step": 998
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7713668278565062e-05,
+      "loss": 0.8206,
+      "step": 999
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7708648866795547e-05,
+      "loss": 0.8379,
+      "step": 1000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770362466407938e-05,
+      "loss": 0.8049,
+      "step": 1001
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769859567353911e-05,
+      "loss": 0.8054,
+      "step": 1002
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7693561898300277e-05,
+      "loss": 0.8347,
+      "step": 1003
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768852334149138e-05,
+      "loss": 0.8296,
+      "step": 1004
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7683480006243896e-05,
+      "loss": 0.8416,
+      "step": 1005
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7678431895692278e-05,
+      "loss": 0.8193,
+      "step": 1006
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.767337901297393e-05,
+      "loss": 0.8328,
+      "step": 1007
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7668321361229248e-05,
+      "loss": 0.8435,
+      "step": 1008
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766325894360156e-05,
+      "loss": 0.8191,
+      "step": 1009
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7658191763237186e-05,
+      "loss": 0.7808,
+      "step": 1010
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7653119823285383e-05,
+      "loss": 0.7473,
+      "step": 1011
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764804312689838e-05,
+      "loss": 0.8181,
+      "step": 1012
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764296167723136e-05,
+      "loss": 0.8604,
+      "step": 1013
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763787547744245e-05,
+      "loss": 0.8479,
+      "step": 1014
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763278453069275e-05,
+      "loss": 0.8469,
+      "step": 1015
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7627688840146286e-05,
+      "loss": 0.8311,
+      "step": 1016
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7622588408970052e-05,
+      "loss": 0.8162,
+      "step": 1017
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7617483240333976e-05,
+      "loss": 0.8242,
+      "step": 1018
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761237333741093e-05,
+      "loss": 0.8301,
+      "step": 1019
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7607258703376744e-05,
+      "loss": 0.8052,
+      "step": 1020
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760213934141017e-05,
+      "loss": 0.7539,
+      "step": 1021
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7597015254692902e-05,
+      "loss": 0.7747,
+      "step": 1022
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759188644640958e-05,
+      "loss": 0.793,
+      "step": 1023
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7586752919747776e-05,
+      "loss": 0.8398,
+      "step": 1024
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7581614677897984e-05,
+      "loss": 0.8252,
+      "step": 1025
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7576471724053634e-05,
+      "loss": 0.8145,
+      "step": 1026
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7571324061411096e-05,
+      "loss": 0.8071,
+      "step": 1027
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7566171693169648e-05,
+      "loss": 0.8354,
+      "step": 1028
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75610146225315e-05,
+      "loss": 0.8293,
+      "step": 1029
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7555852852701794e-05,
+      "loss": 0.8247,
+      "step": 1030
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7550686386888583e-05,
+      "loss": 0.813,
+      "step": 1031
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754551522830283e-05,
+      "loss": 0.8464,
+      "step": 1032
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7540339380158435e-05,
+      "loss": 0.7344,
+      "step": 1033
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75351588456722e-05,
+      "loss": 0.8086,
+      "step": 1034
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7529973628063834e-05,
+      "loss": 0.8308,
+      "step": 1035
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752478373055597e-05,
+      "loss": 0.8325,
+      "step": 1036
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7519589156374145e-05,
+      "loss": 0.8125,
+      "step": 1037
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75143899087468e-05,
+      "loss": 0.7743,
+      "step": 1038
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7509185990905285e-05,
+      "loss": 0.8235,
+      "step": 1039
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7503977406083842e-05,
+      "loss": 0.8386,
+      "step": 1040
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7498764157519625e-05,
+      "loss": 0.8164,
+      "step": 1041
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7493546248452684e-05,
+      "loss": 0.8069,
+      "step": 1042
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748832368212596e-05,
+      "loss": 0.8064,
+      "step": 1043
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7483096461785295e-05,
+      "loss": 0.8274,
+      "step": 1044
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747786459067942e-05,
+      "loss": 0.8435,
+      "step": 1045
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7472628072059955e-05,
+      "loss": 0.811,
+      "step": 1046
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746738690918142e-05,
+      "loss": 0.8245,
+      "step": 1047
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7462141105301204e-05,
+      "loss": 0.8081,
+      "step": 1048
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7456890663679593e-05,
+      "loss": 0.856,
+      "step": 1049
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7451635587579745e-05,
+      "loss": 0.8096,
+      "step": 1050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744637588026771e-05,
+      "loss": 0.79,
+      "step": 1051
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7441111545012404e-05,
+      "loss": 0.7979,
+      "step": 1052
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7435842585085638e-05,
+      "loss": 0.7778,
+      "step": 1053
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7430569003762077e-05,
+      "loss": 0.7651,
+      "step": 1054
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7425290804319264e-05,
+      "loss": 0.7991,
+      "step": 1055
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742000799003762e-05,
+      "loss": 0.8372,
+      "step": 1056
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7414720564200426e-05,
+      "loss": 0.771,
+      "step": 1057
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740942853009383e-05,
+      "loss": 0.7661,
+      "step": 1058
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7404131891006852e-05,
+      "loss": 0.8098,
+      "step": 1059
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7398830650231366e-05,
+      "loss": 0.8281,
+      "step": 1060
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7393524811062105e-05,
+      "loss": 0.8025,
+      "step": 1061
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7388214376796663e-05,
+      "loss": 0.7878,
+      "step": 1062
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738289935073549e-05,
+      "loss": 0.8276,
+      "step": 1063
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737757973618189e-05,
+      "loss": 0.8345,
+      "step": 1064
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737225553644202e-05,
+      "loss": 0.7947,
+      "step": 1065
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736692675482488e-05,
+      "loss": 0.7754,
+      "step": 1066
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7361593394642327e-05,
+      "loss": 0.8032,
+      "step": 1067
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735625545920906e-05,
+      "loss": 0.7932,
+      "step": 1068
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735091295184262e-05,
+      "loss": 0.8347,
+      "step": 1069
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7345565875863384e-05,
+      "loss": 0.7986,
+      "step": 1070
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734021423459458e-05,
+      "loss": 0.7739,
+      "step": 1071
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733485803136227e-05,
+      "loss": 0.7603,
+      "step": 1072
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7329497269495342e-05,
+      "loss": 0.7771,
+      "step": 1073
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732413195232553e-05,
+      "loss": 0.802,
+      "step": 1074
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7318762083187394e-05,
+      "loss": 0.7854,
+      "step": 1075
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731338766541832e-05,
+      "loss": 0.7954,
+      "step": 1076
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7308008702358523e-05,
+      "loss": 0.8154,
+      "step": 1077
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7302625197351048e-05,
+      "loss": 0.7756,
+      "step": 1078
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7297237153741754e-05,
+      "loss": 0.8135,
+      "step": 1079
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7291844574879323e-05,
+      "loss": 0.8372,
+      "step": 1080
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7286447464115267e-05,
+      "loss": 0.8274,
+      "step": 1081
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728104582480389e-05,
+      "loss": 0.7996,
+      "step": 1082
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727563966030234e-05,
+      "loss": 0.8025,
+      "step": 1083
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727022897397056e-05,
+      "loss": 0.8281,
+      "step": 1084
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7264813769171304e-05,
+      "loss": 0.7871,
+      "step": 1085
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7259394049270138e-05,
+      "loss": 0.803,
+      "step": 1086
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725396981763543e-05,
+      "loss": 0.8225,
+      "step": 1087
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724854107763836e-05,
+      "loss": 0.7891,
+      "step": 1088
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.72431078326529e-05,
+      "loss": 0.7339,
+      "step": 1089
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7237670086055827e-05,
+      "loss": 0.833,
+      "step": 1090
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7232227841226724e-05,
+      "loss": 0.7866,
+      "step": 1091
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7226781101547955e-05,
+      "loss": 0.7673,
+      "step": 1092
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722132987040468e-05,
+      "loss": 0.8118,
+      "step": 1093
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721587415118486e-05,
+      "loss": 0.7903,
+      "step": 1094
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7210413947279235e-05,
+      "loss": 0.8005,
+      "step": 1095
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7204949262081344e-05,
+      "loss": 0.8762,
+      "step": 1096
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7199480098987498e-05,
+      "loss": 0.7908,
+      "step": 1097
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7194006461396798e-05,
+      "loss": 0.8582,
+      "step": 1098
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718852835271113e-05,
+      "loss": 0.8347,
+      "step": 1099
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7183045776335144e-05,
+      "loss": 0.8369,
+      "step": 1100
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7177558735676285e-05,
+      "loss": 0.7915,
+      "step": 1101
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717206723414476e-05,
+      "loss": 0.8369,
+      "step": 1102
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7166571275153554e-05,
+      "loss": 0.7598,
+      "step": 1103
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716107086211842e-05,
+      "loss": 0.7917,
+      "step": 1104
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7155565998457877e-05,
+      "loss": 0.7595,
+      "step": 1105
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7150056687593218e-05,
+      "loss": 0.8088,
+      "step": 1106
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7144542932948495e-05,
+      "loss": 0.812,
+      "step": 1107
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713902473795052e-05,
+      "loss": 0.8015,
+      "step": 1108
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7133502106028864e-05,
+      "loss": 0.8176,
+      "step": 1109
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712797504061586e-05,
+      "loss": 0.7949,
+      "step": 1110
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7122443545146598e-05,
+      "loss": 0.8093,
+      "step": 1111
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7116907623058914e-05,
+      "loss": 0.8193,
+      "step": 1112
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71113672777934e-05,
+      "loss": 0.8079,
+      "step": 1113
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7105822512793392e-05,
+      "loss": 0.8445,
+      "step": 1114
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7100273331504988e-05,
+      "loss": 0.8005,
+      "step": 1115
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7094719737377007e-05,
+      "loss": 0.8569,
+      "step": 1116
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7089161733861027e-05,
+      "loss": 0.7578,
+      "step": 1117
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7083599324411363e-05,
+      "loss": 0.7939,
+      "step": 1118
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7078032512485073e-05,
+      "loss": 0.7993,
+      "step": 1119
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707246130154194e-05,
+      "loss": 0.7927,
+      "step": 1120
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7066885695044482e-05,
+      "loss": 0.8052,
+      "step": 1121
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706130569645797e-05,
+      "loss": 0.8232,
+      "step": 1122
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7055721309250378e-05,
+      "loss": 0.8413,
+      "step": 1123
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7050132536892415e-05,
+      "loss": 0.7737,
+      "step": 1124
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704453938285753e-05,
+      "loss": 0.8594,
+      "step": 1125
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703894185062187e-05,
+      "loss": 0.8113,
+      "step": 1126
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7033339943664324e-05,
+      "loss": 0.8564,
+      "step": 1127
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702773366546649e-05,
+      "loss": 0.7952,
+      "step": 1128
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70221230195127e-05,
+      "loss": 0.8015,
+      "step": 1129
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.701650800928996e-05,
+      "loss": 0.8372,
+      "step": 1130
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7010888638288037e-05,
+      "loss": 0.8293,
+      "step": 1131
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700526490999937e-05,
+      "loss": 0.8555,
+      "step": 1132
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699963682791913e-05,
+      "loss": 0.8245,
+      "step": 1133
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699400439554519e-05,
+      "loss": 0.8513,
+      "step": 1134
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698836761637811e-05,
+      "loss": 0.8232,
+      "step": 1135
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6982726493921175e-05,
+      "loss": 0.8594,
+      "step": 1136
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6977081031680352e-05,
+      "loss": 0.8105,
+      "step": 1137
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697143123316431e-05,
+      "loss": 0.7979,
+      "step": 1138
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6965777101884423e-05,
+      "loss": 0.7771,
+      "step": 1139
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696011864135474e-05,
+      "loss": 0.8213,
+      "step": 1140
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6954455855092014e-05,
+      "loss": 0.791,
+      "step": 1141
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6948788746615684e-05,
+      "loss": 0.8152,
+      "step": 1142
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.694311731944787e-05,
+      "loss": 0.8438,
+      "step": 1143
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693744157711338e-05,
+      "loss": 0.8301,
+      "step": 1144
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6931761523139708e-05,
+      "loss": 0.7383,
+      "step": 1145
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692607716105702e-05,
+      "loss": 0.8054,
+      "step": 1146
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6920388494398165e-05,
+      "loss": 0.804,
+      "step": 1147
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6914695526698658e-05,
+      "loss": 0.7952,
+      "step": 1148
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6908998261496704e-05,
+      "loss": 0.8118,
+      "step": 1149
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6903296702333164e-05,
+      "loss": 0.8086,
+      "step": 1150
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689759085275157e-05,
+      "loss": 0.7827,
+      "step": 1151
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689188071629813e-05,
+      "loss": 0.7898,
+      "step": 1152
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6886166296521706e-05,
+      "loss": 0.751,
+      "step": 1153
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6880447596973826e-05,
+      "loss": 0.8108,
+      "step": 1154
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6874724621208677e-05,
+      "loss": 0.7715,
+      "step": 1155
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6868997372783106e-05,
+      "loss": 0.8098,
+      "step": 1156
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6863265855256613e-05,
+      "loss": 0.8333,
+      "step": 1157
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685753007219135e-05,
+      "loss": 0.8474,
+      "step": 1158
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685179002715213e-05,
+      "loss": 0.8425,
+      "step": 1159
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684604572370639e-05,
+      "loss": 0.7861,
+      "step": 1160
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6840297165424254e-05,
+      "loss": 0.8513,
+      "step": 1161
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683454435587845e-05,
+      "loss": 0.8318,
+      "step": 1162
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6828787298644364e-05,
+      "loss": 0.769,
+      "step": 1163
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6823025997300034e-05,
+      "loss": 0.8589,
+      "step": 1164
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6817260455426115e-05,
+      "loss": 0.772,
+      "step": 1165
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6811490676605916e-05,
+      "loss": 0.8052,
+      "step": 1166
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6805716664425367e-05,
+      "loss": 0.7993,
+      "step": 1167
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6799938422473026e-05,
+      "loss": 0.7976,
+      "step": 1168
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6794155954340098e-05,
+      "loss": 0.7876,
+      "step": 1169
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6788369263620396e-05,
+      "loss": 0.8013,
+      "step": 1170
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6782578353910363e-05,
+      "loss": 0.8403,
+      "step": 1171
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677678322880907e-05,
+      "loss": 0.8179,
+      "step": 1172
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6770983891918198e-05,
+      "loss": 0.8357,
+      "step": 1173
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6765180346842057e-05,
+      "loss": 0.8091,
+      "step": 1174
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6759372597187563e-05,
+      "loss": 0.8086,
+      "step": 1175
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675356064656425e-05,
+      "loss": 0.822,
+      "step": 1176
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674774449858426e-05,
+      "loss": 0.7825,
+      "step": 1177
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6741924156862345e-05,
+      "loss": 0.835,
+      "step": 1178
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6736099625015862e-05,
+      "loss": 0.8027,
+      "step": 1179
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6730270906664778e-05,
+      "loss": 0.8,
+      "step": 1180
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6724438005431656e-05,
+      "loss": 0.8057,
+      "step": 1181
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6718600924941658e-05,
+      "loss": 0.824,
+      "step": 1182
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671275966882255e-05,
+      "loss": 0.7834,
+      "step": 1183
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6706914240704682e-05,
+      "loss": 0.7505,
+      "step": 1184
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6701064644221015e-05,
+      "loss": 0.8057,
+      "step": 1185
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6695210883007077e-05,
+      "loss": 0.803,
+      "step": 1186
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6689352960701003e-05,
+      "loss": 0.7766,
+      "step": 1187
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6683490880943508e-05,
+      "loss": 0.7568,
+      "step": 1188
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6677624647377893e-05,
+      "loss": 0.8525,
+      "step": 1189
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6671754263650033e-05,
+      "loss": 0.79,
+      "step": 1190
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666587973340839e-05,
+      "loss": 0.7686,
+      "step": 1191
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6660001060304e-05,
+      "loss": 0.8152,
+      "step": 1192
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665411824799048e-05,
+      "loss": 0.7991,
+      "step": 1193
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664823130012401e-05,
+      "loss": 0.8223,
+      "step": 1194
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6642340220363344e-05,
+      "loss": 0.8245,
+      "step": 1195
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.66364450123698e-05,
+      "loss": 0.8433,
+      "step": 1196
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6630545679807274e-05,
+      "loss": 0.7637,
+      "step": 1197
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6624642226342215e-05,
+      "loss": 0.7761,
+      "step": 1198
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6618734655643637e-05,
+      "loss": 0.7869,
+      "step": 1199
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6612822971383108e-05,
+      "loss": 0.79,
+      "step": 1200
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6606907177234757e-05,
+      "loss": 0.7839,
+      "step": 1201
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660098727687527e-05,
+      "loss": 0.8354,
+      "step": 1202
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6595063273983874e-05,
+      "loss": 0.7856,
+      "step": 1203
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6589135172242363e-05,
+      "loss": 0.8025,
+      "step": 1204
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6583202975335064e-05,
+      "loss": 0.8157,
+      "step": 1205
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657726668694885e-05,
+      "loss": 0.8245,
+      "step": 1206
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657132631077315e-05,
+      "loss": 0.7856,
+      "step": 1207
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6565381850499914e-05,
+      "loss": 0.8333,
+      "step": 1208
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655943330982365e-05,
+      "loss": 0.8196,
+      "step": 1209
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6553480692441383e-05,
+      "loss": 0.832,
+      "step": 1210
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654752400205269e-05,
+      "loss": 0.801,
+      "step": 1211
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6541563242359663e-05,
+      "loss": 0.8369,
+      "step": 1212
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6535598417066934e-05,
+      "loss": 0.853,
+      "step": 1213
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652962952988166e-05,
+      "loss": 0.7627,
+      "step": 1214
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652365658451352e-05,
+      "loss": 0.833,
+      "step": 1215
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6517679584674716e-05,
+      "loss": 0.7634,
+      "step": 1216
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6511698534079973e-05,
+      "loss": 0.8162,
+      "step": 1217
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6505713436446522e-05,
+      "loss": 0.8193,
+      "step": 1218
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6499724295494125e-05,
+      "loss": 0.7749,
+      "step": 1219
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6493731114945047e-05,
+      "loss": 0.8074,
+      "step": 1220
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6487733898524068e-05,
+      "loss": 0.8003,
+      "step": 1221
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6481732649958476e-05,
+      "loss": 0.8445,
+      "step": 1222
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647572737297806e-05,
+      "loss": 0.769,
+      "step": 1223
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6469718071315114e-05,
+      "loss": 0.761,
+      "step": 1224
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6463704748704447e-05,
+      "loss": 0.7905,
+      "step": 1225
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.645768740888334e-05,
+      "loss": 0.8247,
+      "step": 1226
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64516660555916e-05,
+      "loss": 0.7883,
+      "step": 1227
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644564069257151e-05,
+      "loss": 0.7593,
+      "step": 1228
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6439611323567844e-05,
+      "loss": 0.8274,
+      "step": 1229
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643357795232788e-05,
+      "loss": 0.7949,
+      "step": 1230
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6427540582601368e-05,
+      "loss": 0.8208,
+      "step": 1231
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6421499218140558e-05,
+      "loss": 0.8357,
+      "step": 1232
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6415453862700167e-05,
+      "loss": 0.8381,
+      "step": 1233
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6409404520037406e-05,
+      "loss": 0.8035,
+      "step": 1234
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6403351193911956e-05,
+      "loss": 0.802,
+      "step": 1235
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6397293888085975e-05,
+      "loss": 0.8054,
+      "step": 1236
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6391232606324097e-05,
+      "loss": 0.8279,
+      "step": 1237
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6385167352393424e-05,
+      "loss": 0.7708,
+      "step": 1238
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637909813006353e-05,
+      "loss": 0.7014,
+      "step": 1239
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637302494310645e-05,
+      "loss": 0.8088,
+      "step": 1240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636694779529669e-05,
+      "loss": 0.8044,
+      "step": 1241
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636086669041121e-05,
+      "loss": 0.7703,
+      "step": 1242
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6354781632229443e-05,
+      "loss": 0.7998,
+      "step": 1243
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6348692624533258e-05,
+      "loss": 0.8513,
+      "step": 1244
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6342599671106995e-05,
+      "loss": 0.8127,
+      "step": 1245
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6336502775737443e-05,
+      "loss": 0.7864,
+      "step": 1246
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633040194221384e-05,
+      "loss": 0.7727,
+      "step": 1247
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632429717432787e-05,
+      "loss": 0.7971,
+      "step": 1248
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6318188475873657e-05,
+      "loss": 0.8176,
+      "step": 1249
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6312075850647783e-05,
+      "loss": 0.7617,
+      "step": 1250
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6305959302449255e-05,
+      "loss": 0.8022,
+      "step": 1251
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6299838835079534e-05,
+      "loss": 0.8337,
+      "step": 1252
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6293714452342493e-05,
+      "loss": 0.8389,
+      "step": 1253
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6287586158044467e-05,
+      "loss": 0.7649,
+      "step": 1254
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.62814539559942e-05,
+      "loss": 0.8184,
+      "step": 1255
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6275317850002875e-05,
+      "loss": 0.7961,
+      "step": 1256
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6269177843884096e-05,
+      "loss": 0.8105,
+      "step": 1257
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6263033941453895e-05,
+      "loss": 0.7893,
+      "step": 1258
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625688614653073e-05,
+      "loss": 0.8103,
+      "step": 1259
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6250734462935466e-05,
+      "loss": 0.7673,
+      "step": 1260
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624457889449139e-05,
+      "loss": 0.811,
+      "step": 1261
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623841944502421e-05,
+      "loss": 0.8013,
+      "step": 1262
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6232256118362043e-05,
+      "loss": 0.7673,
+      "step": 1263
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6226088918335406e-05,
+      "loss": 0.8064,
+      "step": 1264
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621991784877724e-05,
+      "loss": 0.7764,
+      "step": 1265
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621374291352287e-05,
+      "loss": 0.7822,
+      "step": 1266
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6207564116410047e-05,
+      "loss": 0.8303,
+      "step": 1267
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.62013814612789e-05,
+      "loss": 0.7866,
+      "step": 1268
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619519495197198e-05,
+      "loss": 0.7764,
+      "step": 1269
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6189004592334204e-05,
+      "loss": 0.8176,
+      "step": 1270
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618281038621291e-05,
+      "loss": 0.7947,
+      "step": 1271
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6176612337457807e-05,
+      "loss": 0.8123,
+      "step": 1272
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6170410449921002e-05,
+      "loss": 0.7798,
+      "step": 1273
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616420472745699e-05,
+      "loss": 0.7778,
+      "step": 1274
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6157995173922633e-05,
+      "loss": 0.8074,
+      "step": 1275
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6151781793177198e-05,
+      "loss": 0.8069,
+      "step": 1276
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6145564589082312e-05,
+      "loss": 0.823,
+      "step": 1277
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6139343565501985e-05,
+      "loss": 0.7651,
+      "step": 1278
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6133118726302603e-05,
+      "loss": 0.751,
+      "step": 1279
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6126890075352917e-05,
+      "loss": 0.7878,
+      "step": 1280
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6120657616524053e-05,
+      "loss": 0.7964,
+      "step": 1281
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6114421353689504e-05,
+      "loss": 0.7842,
+      "step": 1282
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610818129072512e-05,
+      "loss": 0.7576,
+      "step": 1283
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6101937431509118e-05,
+      "loss": 0.7881,
+      "step": 1284
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6095689779922075e-05,
+      "loss": 0.7869,
+      "step": 1285
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.608943833984693e-05,
+      "loss": 0.7908,
+      "step": 1286
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6083183115168958e-05,
+      "loss": 0.7654,
+      "step": 1287
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607692410977581e-05,
+      "loss": 0.8452,
+      "step": 1288
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6070661327557465e-05,
+      "loss": 0.843,
+      "step": 1289
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.606439477240627e-05,
+      "loss": 0.8103,
+      "step": 1290
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6058124448216897e-05,
+      "loss": 0.783,
+      "step": 1291
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6051850358886374e-05,
+      "loss": 0.8262,
+      "step": 1292
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6045572508314068e-05,
+      "loss": 0.8418,
+      "step": 1293
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6039290900401674e-05,
+      "loss": 0.7881,
+      "step": 1294
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6033005539053232e-05,
+      "loss": 0.8081,
+      "step": 1295
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602671642817511e-05,
+      "loss": 0.855,
+      "step": 1296
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6020423571676017e-05,
+      "loss": 0.8296,
+      "step": 1297
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6014126973466968e-05,
+      "loss": 0.7886,
+      "step": 1298
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6007826637461323e-05,
+      "loss": 0.8215,
+      "step": 1299
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6001522567574754e-05,
+      "loss": 0.8433,
+      "step": 1300
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5995214767725267e-05,
+      "loss": 0.8315,
+      "step": 1301
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598890324183317e-05,
+      "loss": 0.8435,
+      "step": 1302
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5982587993821087e-05,
+      "loss": 0.8386,
+      "step": 1303
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5976269027613975e-05,
+      "loss": 0.8564,
+      "step": 1304
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5969946347139083e-05,
+      "loss": 0.814,
+      "step": 1305
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5963619956325977e-05,
+      "loss": 0.8062,
+      "step": 1306
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5957289859106525e-05,
+      "loss": 0.7354,
+      "step": 1307
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5950956059414896e-05,
+      "loss": 0.7886,
+      "step": 1308
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5944618561187573e-05,
+      "loss": 0.7964,
+      "step": 1309
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593827736836332e-05,
+      "loss": 0.8486,
+      "step": 1310
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593193248488321e-05,
+      "loss": 0.8064,
+      "step": 1311
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5925583914690604e-05,
+      "loss": 0.7778,
+      "step": 1312
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591923166173116e-05,
+      "loss": 0.7886,
+      "step": 1313
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5912875729952817e-05,
+      "loss": 0.7661,
+      "step": 1314
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5906516123305803e-05,
+      "loss": 0.8149,
+      "step": 1315
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5900152845742638e-05,
+      "loss": 0.8135,
+      "step": 1316
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589378590121811e-05,
+      "loss": 0.8298,
+      "step": 1317
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5887415293689295e-05,
+      "loss": 0.7874,
+      "step": 1318
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.588104102711554e-05,
+      "loss": 0.8174,
+      "step": 1319
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5874663105458473e-05,
+      "loss": 0.8352,
+      "step": 1320
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5868281532681995e-05,
+      "loss": 0.8064,
+      "step": 1321
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586189631275226e-05,
+      "loss": 0.8274,
+      "step": 1322
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585550744963771e-05,
+      "loss": 0.7703,
+      "step": 1323
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5849114947309033e-05,
+      "loss": 0.8232,
+      "step": 1324
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584271880973919e-05,
+      "loss": 0.8147,
+      "step": 1325
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5836319040903402e-05,
+      "loss": 0.8091,
+      "step": 1326
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582991564477914e-05,
+      "loss": 0.8298,
+      "step": 1327
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5823508625346126e-05,
+      "loss": 0.7998,
+      "step": 1328
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5817097986586355e-05,
+      "loss": 0.8159,
+      "step": 1329
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581068373248404e-05,
+      "loss": 0.7766,
+      "step": 1330
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5804265867025674e-05,
+      "loss": 0.8074,
+      "step": 1331
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579784439419997e-05,
+      "loss": 0.783,
+      "step": 1332
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5791419317997884e-05,
+      "loss": 0.834,
+      "step": 1333
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5784990642412635e-05,
+      "loss": 0.8252,
+      "step": 1334
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5778558371439648e-05,
+      "loss": 0.8176,
+      "step": 1335
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5772122509076604e-05,
+      "loss": 0.7939,
+      "step": 1336
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576568305932341e-05,
+      "loss": 0.7922,
+      "step": 1337
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.57592400261822e-05,
+      "loss": 0.8433,
+      "step": 1338
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5752793413657336e-05,
+      "loss": 0.8064,
+      "step": 1339
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5746343225755412e-05,
+      "loss": 0.7676,
+      "step": 1340
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5739889466485227e-05,
+      "loss": 0.7349,
+      "step": 1341
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573343213985782e-05,
+      "loss": 0.7971,
+      "step": 1342
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572697124988643e-05,
+      "loss": 0.8123,
+      "step": 1343
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5720506800586522e-05,
+      "loss": 0.7935,
+      "step": 1344
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571403879597577e-05,
+      "loss": 0.7515,
+      "step": 1345
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5707567240074052e-05,
+      "loss": 0.804,
+      "step": 1346
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5701092136903465e-05,
+      "loss": 0.8113,
+      "step": 1347
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5694613490488295e-05,
+      "loss": 0.7632,
+      "step": 1348
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5688131304855047e-05,
+      "loss": 0.8037,
+      "step": 1349
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5681645584032414e-05,
+      "loss": 0.78,
+      "step": 1350
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5675156332051293e-05,
+      "loss": 0.7634,
+      "step": 1351
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566866355294476e-05,
+      "loss": 0.8208,
+      "step": 1352
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566216725074811e-05,
+      "loss": 0.8008,
+      "step": 1353
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5655667429498803e-05,
+      "loss": 0.8127,
+      "step": 1354
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5649164093236505e-05,
+      "loss": 0.8062,
+      "step": 1355
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5642657246003048e-05,
+      "loss": 0.7822,
+      "step": 1356
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563614689184246e-05,
+      "loss": 0.7971,
+      "step": 1357
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5629633034800945e-05,
+      "loss": 0.7986,
+      "step": 1358
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562311567892688e-05,
+      "loss": 0.77,
+      "step": 1359
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561659482827082e-05,
+      "loss": 0.7664,
+      "step": 1360
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5610070486885498e-05,
+      "loss": 0.8259,
+      "step": 1361
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5603542658825803e-05,
+      "loss": 0.865,
+      "step": 1362
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5597011348148803e-05,
+      "loss": 0.7776,
+      "step": 1363
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559047655891372e-05,
+      "loss": 0.8447,
+      "step": 1364
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.558393829518195e-05,
+      "loss": 0.7935,
+      "step": 1365
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5577396561017043e-05,
+      "loss": 0.7986,
+      "step": 1366
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55708513604847e-05,
+      "loss": 0.7334,
+      "step": 1367
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556430269765278e-05,
+      "loss": 0.7861,
+      "step": 1368
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5557750576591297e-05,
+      "loss": 0.804,
+      "step": 1369
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5551195001372418e-05,
+      "loss": 0.8352,
+      "step": 1370
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554463597607044e-05,
+      "loss": 0.8101,
+      "step": 1371
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5538073504761826e-05,
+      "loss": 0.7654,
+      "step": 1372
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5531507591525165e-05,
+      "loss": 0.8181,
+      "step": 1373
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552493824044119e-05,
+      "loss": 0.7917,
+      "step": 1374
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5518365455592776e-05,
+      "loss": 0.8218,
+      "step": 1375
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5511789241064922e-05,
+      "loss": 0.8062,
+      "step": 1376
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5505209600944763e-05,
+      "loss": 0.7903,
+      "step": 1377
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5498626539321563e-05,
+      "loss": 0.8054,
+      "step": 1378
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549204006028672e-05,
+      "loss": 0.8628,
+      "step": 1379
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548545016793374e-05,
+      "loss": 0.752,
+      "step": 1380
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5478856866358266e-05,
+      "loss": 0.7979,
+      "step": 1381
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547226015965805e-05,
+      "loss": 0.7856,
+      "step": 1382
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5465660051932966e-05,
+      "loss": 0.802,
+      "step": 1383
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5459056547284998e-05,
+      "loss": 0.7588,
+      "step": 1384
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545244964981824e-05,
+      "loss": 0.7842,
+      "step": 1385
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5445839363638905e-05,
+      "loss": 0.7986,
+      "step": 1386
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5439225692855293e-05,
+      "loss": 0.8484,
+      "step": 1387
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5432608641577825e-05,
+      "loss": 0.8225,
+      "step": 1388
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5425988213919016e-05,
+      "loss": 0.813,
+      "step": 1389
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5419364413993484e-05,
+      "loss": 0.8281,
+      "step": 1390
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5412737245917927e-05,
+      "loss": 0.7791,
+      "step": 1391
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5406106713811163e-05,
+      "loss": 0.8159,
+      "step": 1392
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.539947282179408e-05,
+      "loss": 0.7793,
+      "step": 1393
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5392835573989656e-05,
+      "loss": 0.7725,
+      "step": 1394
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5386194974522965e-05,
+      "loss": 0.8499,
+      "step": 1395
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5379551027521154e-05,
+      "loss": 0.8157,
+      "step": 1396
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.537290373711346e-05,
+      "loss": 0.7935,
+      "step": 1397
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5366253107431188e-05,
+      "loss": 0.7881,
+      "step": 1398
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5359599142607722e-05,
+      "loss": 0.748,
+      "step": 1399
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.535294184677852e-05,
+      "loss": 0.7644,
+      "step": 1400
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5346281224081113e-05,
+      "loss": 0.8162,
+      "step": 1401
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533961727865509e-05,
+      "loss": 0.8059,
+      "step": 1402
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5332950014642124e-05,
+      "loss": 0.76,
+      "step": 1403
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5326279436185923e-05,
+      "loss": 0.7915,
+      "step": 1404
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531960554743228e-05,
+      "loss": 0.8325,
+      "step": 1405
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531292835252903e-05,
+      "loss": 0.7666,
+      "step": 1406
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5306247855626074e-05,
+      "loss": 0.7917,
+      "step": 1407
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5299564060875357e-05,
+      "loss": 0.8127,
+      "step": 1408
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529287697243087e-05,
+      "loss": 0.8066,
+      "step": 1409
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5286186594448665e-05,
+      "loss": 0.7759,
+      "step": 1410
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527949293108683e-05,
+      "loss": 0.8135,
+      "step": 1411
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5272795986505487e-05,
+      "loss": 0.7839,
+      "step": 1412
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5266095764866813e-05,
+      "loss": 0.8198,
+      "step": 1413
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5259392270335017e-05,
+      "loss": 0.8191,
+      "step": 1414
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5252685507076331e-05,
+      "loss": 0.8162,
+      "step": 1415
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.524597547925903e-05,
+      "loss": 0.7439,
+      "step": 1416
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5239262191053419e-05,
+      "loss": 0.7739,
+      "step": 1417
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5232545646631821e-05,
+      "loss": 0.791,
+      "step": 1418
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.522582585016859e-05,
+      "loss": 0.7898,
+      "step": 1419
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5219102805840095e-05,
+      "loss": 0.78,
+      "step": 1420
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5212376517824727e-05,
+      "loss": 0.7883,
+      "step": 1421
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5205646990302898e-05,
+      "loss": 0.8196,
+      "step": 1422
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.519891422745702e-05,
+      "loss": 0.8154,
+      "step": 1423
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.519217823347153e-05,
+      "loss": 0.853,
+      "step": 1424
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5185439012532862e-05,
+      "loss": 0.772,
+      "step": 1425
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5178696568829465e-05,
+      "loss": 0.7812,
+      "step": 1426
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5171950906551787e-05,
+      "loss": 0.8152,
+      "step": 1427
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5165202029892273e-05,
+      "loss": 0.8022,
+      "step": 1428
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5158449943045361e-05,
+      "loss": 0.8047,
+      "step": 1429
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5151694650207506e-05,
+      "loss": 0.7703,
+      "step": 1430
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5144936155577136e-05,
+      "loss": 0.7949,
+      "step": 1431
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5138174463354668e-05,
+      "loss": 0.8586,
+      "step": 1432
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5131409577742523e-05,
+      "loss": 0.8213,
+      "step": 1433
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.512464150294509e-05,
+      "loss": 0.8181,
+      "step": 1434
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5117870243168744e-05,
+      "loss": 0.8232,
+      "step": 1435
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5111095802621844e-05,
+      "loss": 0.7917,
+      "step": 1436
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5104318185514728e-05,
+      "loss": 0.7703,
+      "step": 1437
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5097537396059698e-05,
+      "loss": 0.853,
+      "step": 1438
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5090753438471034e-05,
+      "loss": 0.8079,
+      "step": 1439
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5083966316964986e-05,
+      "loss": 0.822,
+      "step": 1440
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5077176035759772e-05,
+      "loss": 0.7515,
+      "step": 1441
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5070382599075568e-05,
+      "loss": 0.7695,
+      "step": 1442
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.506358601113451e-05,
+      "loss": 0.8069,
+      "step": 1443
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5056786276160703e-05,
+      "loss": 0.7876,
+      "step": 1444
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5049983398380195e-05,
+      "loss": 0.781,
+      "step": 1445
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5043177382020994e-05,
+      "loss": 0.8254,
+      "step": 1446
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5036368231313066e-05,
+      "loss": 0.8015,
+      "step": 1447
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5029555950488306e-05,
+      "loss": 0.782,
+      "step": 1448
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5022740543780572e-05,
+      "loss": 0.8101,
+      "step": 1449
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5015922015425657e-05,
+      "loss": 0.7656,
+      "step": 1450
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5009100369661294e-05,
+      "loss": 0.804,
+      "step": 1451
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5002275610727157e-05,
+      "loss": 0.7815,
+      "step": 1452
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4995447742864853e-05,
+      "loss": 0.7913,
+      "step": 1453
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4988616770317915e-05,
+      "loss": 0.8003,
+      "step": 1454
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4981782697331819e-05,
+      "loss": 0.7737,
+      "step": 1455
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4974945528153956e-05,
+      "loss": 0.7615,
+      "step": 1456
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4968105267033647e-05,
+      "loss": 0.7827,
+      "step": 1457
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4961261918222131e-05,
+      "loss": 0.802,
+      "step": 1458
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4954415485972573e-05,
+      "loss": 0.7849,
+      "step": 1459
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4947565974540046e-05,
+      "loss": 0.7598,
+      "step": 1460
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.494071338818154e-05,
+      "loss": 0.8481,
+      "step": 1461
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4933857731155957e-05,
+      "loss": 0.8015,
+      "step": 1462
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.492699900772411e-05,
+      "loss": 0.8372,
+      "step": 1463
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.492013722214871e-05,
+      "loss": 0.8418,
+      "step": 1464
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4913272378694377e-05,
+      "loss": 0.8115,
+      "step": 1465
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4906404481627633e-05,
+      "loss": 0.8145,
+      "step": 1466
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4899533535216891e-05,
+      "loss": 0.7998,
+      "step": 1467
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4892659543732463e-05,
+      "loss": 0.7871,
+      "step": 1468
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4885782511446562e-05,
+      "loss": 0.8083,
+      "step": 1469
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4878902442633273e-05,
+      "loss": 0.7708,
+      "step": 1470
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4872019341568585e-05,
+      "loss": 0.8474,
+      "step": 1471
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4865133212530358e-05,
+      "loss": 0.7756,
+      "step": 1472
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4858244059798344e-05,
+      "loss": 0.8308,
+      "step": 1473
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4851351887654168e-05,
+      "loss": 0.7754,
+      "step": 1474
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.484445670038134e-05,
+      "loss": 0.8137,
+      "step": 1475
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.483755850226523e-05,
+      "loss": 0.7546,
+      "step": 1476
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4830657297593094e-05,
+      "loss": 0.7734,
+      "step": 1477
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4823753090654047e-05,
+      "loss": 0.7913,
+      "step": 1478
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4816845885739071e-05,
+      "loss": 0.7957,
+      "step": 1479
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4809935687141017e-05,
+      "loss": 0.7786,
+      "step": 1480
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4803022499154589e-05,
+      "loss": 0.8013,
+      "step": 1481
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4796106326076353e-05,
+      "loss": 0.7932,
+      "step": 1482
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4789187172204732e-05,
+      "loss": 0.8074,
+      "step": 1483
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4782265041839999e-05,
+      "loss": 0.8096,
+      "step": 1484
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4775339939284275e-05,
+      "loss": 0.7725,
+      "step": 1485
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4768411868841534e-05,
+      "loss": 0.7874,
+      "step": 1486
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4761480834817585e-05,
+      "loss": 0.781,
+      "step": 1487
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4754546841520093e-05,
+      "loss": 0.8049,
+      "step": 1488
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4747609893258546e-05,
+      "loss": 0.772,
+      "step": 1489
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4740669994344284e-05,
+      "loss": 0.8176,
+      "step": 1490
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4733727149090467e-05,
+      "loss": 0.7964,
+      "step": 1491
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4726781361812094e-05,
+      "loss": 0.8035,
+      "step": 1492
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4719832636825995e-05,
+      "loss": 0.7568,
+      "step": 1493
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4712880978450818e-05,
+      "loss": 0.8223,
+      "step": 1494
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.470592639100704e-05,
+      "loss": 0.7852,
+      "step": 1495
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4698968878816952e-05,
+      "loss": 0.7322,
+      "step": 1496
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4692008446204676e-05,
+      "loss": 0.8525,
+      "step": 1497
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.468504509749613e-05,
+      "loss": 0.8013,
+      "step": 1498
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4678078837019064e-05,
+      "loss": 0.8486,
+      "step": 1499
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4671109669103021e-05,
+      "loss": 0.7656,
+      "step": 1500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.466413759807936e-05,
+      "loss": 0.7754,
+      "step": 1501
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4657162628281246e-05,
+      "loss": 0.8203,
+      "step": 1502
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4650184764043638e-05,
+      "loss": 0.7925,
+      "step": 1503
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4643204009703298e-05,
+      "loss": 0.8259,
+      "step": 1504
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4636220369598786e-05,
+      "loss": 0.8535,
+      "step": 1505
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4629233848070452e-05,
+      "loss": 0.7668,
+      "step": 1506
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4622244449460443e-05,
+      "loss": 0.7634,
+      "step": 1507
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4615252178112686e-05,
+      "loss": 0.8271,
+      "step": 1508
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4608257038372894e-05,
+      "loss": 0.812,
+      "step": 1509
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4601259034588573e-05,
+      "loss": 0.8113,
+      "step": 1510
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4594258171108997e-05,
+      "loss": 0.804,
+      "step": 1511
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.458725445228522e-05,
+      "loss": 0.7957,
+      "step": 1512
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.458024788247008e-05,
+      "loss": 0.7844,
+      "step": 1513
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4573238466018173e-05,
+      "loss": 0.8093,
+      "step": 1514
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4566226207285871e-05,
+      "loss": 0.7446,
+      "step": 1515
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4559211110631317e-05,
+      "loss": 0.7683,
+      "step": 1516
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.455219318041441e-05,
+      "loss": 0.7483,
+      "step": 1517
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4545172420996814e-05,
+      "loss": 0.7927,
+      "step": 1518
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.453814883674195e-05,
+      "loss": 0.7734,
+      "step": 1519
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4531122432014995e-05,
+      "loss": 0.7749,
+      "step": 1520
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4524093211182874e-05,
+      "loss": 0.8403,
+      "step": 1521
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.451706117861428e-05,
+      "loss": 0.7883,
+      "step": 1522
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4510026338679628e-05,
+      "loss": 0.825,
+      "step": 1523
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.45029886957511e-05,
+      "loss": 0.7769,
+      "step": 1524
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4495948254202601e-05,
+      "loss": 0.751,
+      "step": 1525
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4488905018409792e-05,
+      "loss": 0.7942,
+      "step": 1526
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4481858992750063e-05,
+      "loss": 0.7864,
+      "step": 1527
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4474810181602532e-05,
+      "loss": 0.8311,
+      "step": 1528
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4467758589348065e-05,
+      "loss": 0.8032,
+      "step": 1529
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.446070422036924e-05,
+      "loss": 0.7905,
+      "step": 1530
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4453647079050366e-05,
+      "loss": 0.7932,
+      "step": 1531
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4446587169777479e-05,
+      "loss": 0.7688,
+      "step": 1532
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4439524496938333e-05,
+      "loss": 0.8425,
+      "step": 1533
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4432459064922393e-05,
+      "loss": 0.7747,
+      "step": 1534
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.442539087812085e-05,
+      "loss": 0.7751,
+      "step": 1535
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4418319940926601e-05,
+      "loss": 0.8232,
+      "step": 1536
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4411246257734251e-05,
+      "loss": 0.8079,
+      "step": 1537
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4404169832940113e-05,
+      "loss": 0.811,
+      "step": 1538
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4397090670942208e-05,
+      "loss": 0.8142,
+      "step": 1539
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4390008776140257e-05,
+      "loss": 0.7886,
+      "step": 1540
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.438292415293567e-05,
+      "loss": 0.7632,
+      "step": 1541
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4375836805731562e-05,
+      "loss": 0.8037,
+      "step": 1542
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4368746738932742e-05,
+      "loss": 0.7954,
+      "step": 1543
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4361653956945708e-05,
+      "loss": 0.8059,
+      "step": 1544
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4354558464178634e-05,
+      "loss": 0.7527,
+      "step": 1545
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.43474602650414e-05,
+      "loss": 0.8064,
+      "step": 1546
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4340359363945548e-05,
+      "loss": 0.7598,
+      "step": 1547
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4333255765304313e-05,
+      "loss": 0.7966,
+      "step": 1548
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4326149473532598e-05,
+      "loss": 0.8521,
+      "step": 1549
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4319040493046983e-05,
+      "loss": 0.7781,
+      "step": 1550
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4311928828265724e-05,
+      "loss": 0.7764,
+      "step": 1551
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4304814483608735e-05,
+      "loss": 0.7864,
+      "step": 1552
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4297697463497603e-05,
+      "loss": 0.8052,
+      "step": 1553
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4290577772355575e-05,
+      "loss": 0.77,
+      "step": 1554
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4283455414607559e-05,
+      "loss": 0.811,
+      "step": 1555
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4276330394680122e-05,
+      "loss": 0.7764,
+      "step": 1556
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.426920271700148e-05,
+      "loss": 0.8044,
+      "step": 1557
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4262072386001508e-05,
+      "loss": 0.7871,
+      "step": 1558
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4254939406111723e-05,
+      "loss": 0.7825,
+      "step": 1559
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4247803781765295e-05,
+      "loss": 0.8047,
+      "step": 1560
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4240665517397031e-05,
+      "loss": 0.7891,
+      "step": 1561
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4233524617443389e-05,
+      "loss": 0.8599,
+      "step": 1562
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4226381086342449e-05,
+      "loss": 0.7661,
+      "step": 1563
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.421923492853394e-05,
+      "loss": 0.8447,
+      "step": 1564
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4212086148459218e-05,
+      "loss": 0.821,
+      "step": 1565
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4204934750561275e-05,
+      "loss": 0.8257,
+      "step": 1566
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4197780739284715e-05,
+      "loss": 0.7451,
+      "step": 1567
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4190624119075782e-05,
+      "loss": 0.781,
+      "step": 1568
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4183464894382335e-05,
+      "loss": 0.8228,
+      "step": 1569
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.417630306965385e-05,
+      "loss": 0.7891,
+      "step": 1570
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4169138649341422e-05,
+      "loss": 0.8088,
+      "step": 1571
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4161971637897758e-05,
+      "loss": 0.7725,
+      "step": 1572
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4154802039777176e-05,
+      "loss": 0.8254,
+      "step": 1573
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4147629859435601e-05,
+      "loss": 0.812,
+      "step": 1574
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4140455101330561e-05,
+      "loss": 0.8267,
+      "step": 1575
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4133277769921192e-05,
+      "loss": 0.8008,
+      "step": 1576
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4126097869668223e-05,
+      "loss": 0.7698,
+      "step": 1577
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.411891540503398e-05,
+      "loss": 0.8179,
+      "step": 1578
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4111730380482387e-05,
+      "loss": 0.8101,
+      "step": 1579
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4104542800478958e-05,
+      "loss": 0.8066,
+      "step": 1580
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4097352669490791e-05,
+      "loss": 0.8098,
+      "step": 1581
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4090159991986574e-05,
+      "loss": 0.793,
+      "step": 1582
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4082964772436575e-05,
+      "loss": 0.801,
+      "step": 1583
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4075767015312644e-05,
+      "loss": 0.8469,
+      "step": 1584
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4068566725088209e-05,
+      "loss": 0.8223,
+      "step": 1585
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4061363906238268e-05,
+      "loss": 0.7854,
+      "step": 1586
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.405415856323939e-05,
+      "loss": 0.7705,
+      "step": 1587
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4046950700569718e-05,
+      "loss": 0.8218,
+      "step": 1588
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4039740322708961e-05,
+      "loss": 0.7354,
+      "step": 1589
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4032527434138388e-05,
+      "loss": 0.8384,
+      "step": 1590
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4025312039340829e-05,
+      "loss": 0.8167,
+      "step": 1591
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4018094142800667e-05,
+      "loss": 0.7881,
+      "step": 1592
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4010873749003854e-05,
+      "loss": 0.8093,
+      "step": 1593
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4003650862437874e-05,
+      "loss": 0.8333,
+      "step": 1594
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3996425487591779e-05,
+      "loss": 0.7471,
+      "step": 1595
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.398919762895616e-05,
+      "loss": 0.7878,
+      "step": 1596
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3981967291023144e-05,
+      "loss": 0.7825,
+      "step": 1597
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3974734478286413e-05,
+      "loss": 0.7896,
+      "step": 1598
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3967499195241176e-05,
+      "loss": 0.7527,
+      "step": 1599
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3960261446384184e-05,
+      "loss": 0.803,
+      "step": 1600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.395302123621372e-05,
+      "loss": 0.7786,
+      "step": 1601
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.394577856922959e-05,
+      "loss": 0.802,
+      "step": 1602
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3938533449933131e-05,
+      "loss": 0.7783,
+      "step": 1603
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3931285882827208e-05,
+      "loss": 0.7791,
+      "step": 1604
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3924035872416204e-05,
+      "loss": 0.791,
+      "step": 1605
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.391678342320602e-05,
+      "loss": 0.7959,
+      "step": 1606
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3909528539704074e-05,
+      "loss": 0.7808,
+      "step": 1607
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3902271226419294e-05,
+      "loss": 0.7605,
+      "step": 1608
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3895011487862123e-05,
+      "loss": 0.7461,
+      "step": 1609
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3887749328544507e-05,
+      "loss": 0.8367,
+      "step": 1610
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3880484752979898e-05,
+      "loss": 0.7458,
+      "step": 1611
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3873217765683252e-05,
+      "loss": 0.7517,
+      "step": 1612
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3865948371171018e-05,
+      "loss": 0.7705,
+      "step": 1613
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3858676573961147e-05,
+      "loss": 0.813,
+      "step": 1614
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3851402378573083e-05,
+      "loss": 0.7903,
+      "step": 1615
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3844125789527755e-05,
+      "loss": 0.8047,
+      "step": 1616
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3836846811347583e-05,
+      "loss": 0.8176,
+      "step": 1617
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3829565448556478e-05,
+      "loss": 0.7654,
+      "step": 1618
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3822281705679826e-05,
+      "loss": 0.7998,
+      "step": 1619
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3814995587244487e-05,
+      "loss": 0.8091,
+      "step": 1620
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3807707097778806e-05,
+      "loss": 0.8062,
+      "step": 1621
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3800416241812602e-05,
+      "loss": 0.7703,
+      "step": 1622
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3793123023877164e-05,
+      "loss": 0.7854,
+      "step": 1623
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3785827448505243e-05,
+      "loss": 0.8499,
+      "step": 1624
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3778529520231056e-05,
+      "loss": 0.8203,
+      "step": 1625
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3771229243590296e-05,
+      "loss": 0.8416,
+      "step": 1626
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3763926623120096e-05,
+      "loss": 0.7787,
+      "step": 1627
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3756621663359062e-05,
+      "loss": 0.792,
+      "step": 1628
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3749314368847242e-05,
+      "loss": 0.7632,
+      "step": 1629
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3742004744126138e-05,
+      "loss": 0.7805,
+      "step": 1630
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3734692793738709e-05,
+      "loss": 0.8113,
+      "step": 1631
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3727378522229346e-05,
+      "loss": 0.7808,
+      "step": 1632
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3720061934143893e-05,
+      "loss": 0.7568,
+      "step": 1633
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3712743034029628e-05,
+      "loss": 0.8064,
+      "step": 1634
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3705421826435268e-05,
+      "loss": 0.7693,
+      "step": 1635
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3698098315910961e-05,
+      "loss": 0.7944,
+      "step": 1636
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3690772507008296e-05,
+      "loss": 0.7905,
+      "step": 1637
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3683444404280278e-05,
+      "loss": 0.7771,
+      "step": 1638
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3676114012281345e-05,
+      "loss": 0.8127,
+      "step": 1639
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3668781335567355e-05,
+      "loss": 0.7449,
+      "step": 1640
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3661446378695588e-05,
+      "loss": 0.7534,
+      "step": 1641
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3654109146224741e-05,
+      "loss": 0.7861,
+      "step": 1642
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3646769642714924e-05,
+      "loss": 0.8013,
+      "step": 1643
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3639427872727654e-05,
+      "loss": 0.7803,
+      "step": 1644
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3632083840825866e-05,
+      "loss": 0.7878,
+      "step": 1645
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3624737551573896e-05,
+      "loss": 0.8225,
+      "step": 1646
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3617389009537484e-05,
+      "loss": 0.7817,
+      "step": 1647
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3610038219283766e-05,
+      "loss": 0.7798,
+      "step": 1648
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3602685185381282e-05,
+      "loss": 0.7629,
+      "step": 1649
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.359532991239996e-05,
+      "loss": 0.8521,
+      "step": 1650
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.358797240491112e-05,
+      "loss": 0.7468,
+      "step": 1651
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3580612667487483e-05,
+      "loss": 0.7771,
+      "step": 1652
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.357325070470314e-05,
+      "loss": 0.7896,
+      "step": 1653
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3565886521133572e-05,
+      "loss": 0.7947,
+      "step": 1654
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3558520121355634e-05,
+      "loss": 0.8167,
+      "step": 1655
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3551151509947572e-05,
+      "loss": 0.7661,
+      "step": 1656
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3543780691488997e-05,
+      "loss": 0.7961,
+      "step": 1657
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3536407670560889e-05,
+      "loss": 0.7988,
+      "step": 1658
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3529032451745608e-05,
+      "loss": 0.8044,
+      "step": 1659
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3521655039626863e-05,
+      "loss": 0.8147,
+      "step": 1660
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3514275438789745e-05,
+      "loss": 0.7932,
+      "step": 1661
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3506893653820692e-05,
+      "loss": 0.7878,
+      "step": 1662
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3499509689307504e-05,
+      "loss": 0.7634,
+      "step": 1663
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3492123549839339e-05,
+      "loss": 0.8057,
+      "step": 1664
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3484735240006698e-05,
+      "loss": 0.8186,
+      "step": 1665
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3477344764401438e-05,
+      "loss": 0.7915,
+      "step": 1666
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3469952127616766e-05,
+      "loss": 0.7817,
+      "step": 1667
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3462557334247217e-05,
+      "loss": 0.803,
+      "step": 1668
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3455160388888678e-05,
+      "loss": 0.7993,
+      "step": 1669
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3447761296138373e-05,
+      "loss": 0.8196,
+      "step": 1670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.344036006059486e-05,
+      "loss": 0.8062,
+      "step": 1671
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3432956686858024e-05,
+      "loss": 0.7939,
+      "step": 1672
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3425551179529085e-05,
+      "loss": 0.781,
+      "step": 1673
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3418143543210585e-05,
+      "loss": 0.8394,
+      "step": 1674
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3410733782506391e-05,
+      "loss": 0.8035,
+      "step": 1675
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3403321902021691e-05,
+      "loss": 0.759,
+      "step": 1676
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3395907906362986e-05,
+      "loss": 0.7717,
+      "step": 1677
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3388491800138097e-05,
+      "loss": 0.8247,
+      "step": 1678
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3381073587956154e-05,
+      "loss": 0.7827,
+      "step": 1679
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3373653274427598e-05,
+      "loss": 0.7998,
+      "step": 1680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3366230864164173e-05,
+      "loss": 0.8623,
+      "step": 1681
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3358806361778929e-05,
+      "loss": 0.8052,
+      "step": 1682
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3351379771886212e-05,
+      "loss": 0.752,
+      "step": 1683
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3343951099101671e-05,
+      "loss": 0.7986,
+      "step": 1684
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3336520348042243e-05,
+      "loss": 0.8022,
+      "step": 1685
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3329087523326168e-05,
+      "loss": 0.7969,
+      "step": 1686
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.332165262957296e-05,
+      "loss": 0.812,
+      "step": 1687
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3314215671403425e-05,
+      "loss": 0.7832,
+      "step": 1688
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.330677665343966e-05,
+      "loss": 0.8442,
+      "step": 1689
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.329933558030503e-05,
+      "loss": 0.8218,
+      "step": 1690
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3291892456624185e-05,
+      "loss": 0.8711,
+      "step": 1691
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3284447287023046e-05,
+      "loss": 0.7822,
+      "step": 1692
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3277000076128807e-05,
+      "loss": 0.7761,
+      "step": 1693
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.326955082856993e-05,
+      "loss": 0.7754,
+      "step": 1694
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3262099548976145e-05,
+      "loss": 0.8088,
+      "step": 1695
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3254646241978439e-05,
+      "loss": 0.8083,
+      "step": 1696
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3247190912209065e-05,
+      "loss": 0.8306,
+      "step": 1697
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3239733564301533e-05,
+      "loss": 0.7305,
+      "step": 1698
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3232274202890602e-05,
+      "loss": 0.7705,
+      "step": 1699
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3224812832612288e-05,
+      "loss": 0.8057,
+      "step": 1700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3217349458103855e-05,
+      "loss": 0.8015,
+      "step": 1701
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3209884084003806e-05,
+      "loss": 0.761,
+      "step": 1702
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3202416714951896e-05,
+      "loss": 0.7861,
+      "step": 1703
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.319494735558911e-05,
+      "loss": 0.7048,
+      "step": 1704
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3187476010557684e-05,
+      "loss": 0.7815,
+      "step": 1705
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3180002684501075e-05,
+      "loss": 0.7722,
+      "step": 1706
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3172527382063974e-05,
+      "loss": 0.8489,
+      "step": 1707
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3165050107892306e-05,
+      "loss": 0.741,
+      "step": 1708
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3157570866633216e-05,
+      "loss": 0.7966,
+      "step": 1709
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.315008966293507e-05,
+      "loss": 0.8228,
+      "step": 1710
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3142606501447464e-05,
+      "loss": 0.8335,
+      "step": 1711
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3135121386821197e-05,
+      "loss": 0.8179,
+      "step": 1712
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3127634323708292e-05,
+      "loss": 0.7561,
+      "step": 1713
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3120145316761979e-05,
+      "loss": 0.8008,
+      "step": 1714
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3112654370636698e-05,
+      "loss": 0.8123,
+      "step": 1715
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3105161489988087e-05,
+      "loss": 0.7422,
+      "step": 1716
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3097666679473e-05,
+      "loss": 0.791,
+      "step": 1717
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3090169943749475e-05,
+      "loss": 0.7783,
+      "step": 1718
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.308267128747676e-05,
+      "loss": 0.824,
+      "step": 1719
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3075170715315291e-05,
+      "loss": 0.7881,
+      "step": 1720
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3067668231926685e-05,
+      "loss": 0.7739,
+      "step": 1721
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3060163841973768e-05,
+      "loss": 0.7776,
+      "step": 1722
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3052657550120534e-05,
+      "loss": 0.79,
+      "step": 1723
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3045149361032161e-05,
+      "loss": 0.7844,
+      "step": 1724
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3037639279375016e-05,
+      "loss": 0.7979,
+      "step": 1725
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3030127309816632e-05,
+      "loss": 0.7439,
+      "step": 1726
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3022613457025717e-05,
+      "loss": 0.7888,
+      "step": 1727
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3015097725672157e-05,
+      "loss": 0.7532,
+      "step": 1728
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3007580120426993e-05,
+      "loss": 0.7966,
+      "step": 1729
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3000060645962442e-05,
+      "loss": 0.7595,
+      "step": 1730
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2992539306951878e-05,
+      "loss": 0.8083,
+      "step": 1731
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2985016108069833e-05,
+      "loss": 0.8096,
+      "step": 1732
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2977491053991997e-05,
+      "loss": 0.7708,
+      "step": 1733
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2969964149395212e-05,
+      "loss": 0.8181,
+      "step": 1734
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2962435398957467e-05,
+      "loss": 0.7585,
+      "step": 1735
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2954904807357905e-05,
+      "loss": 0.7798,
+      "step": 1736
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2947372379276806e-05,
+      "loss": 0.73,
+      "step": 1737
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2939838119395601e-05,
+      "loss": 0.8218,
+      "step": 1738
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2932302032396849e-05,
+      "loss": 0.7239,
+      "step": 1739
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2924764122964244e-05,
+      "loss": 0.7239,
+      "step": 1740
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2917224395782627e-05,
+      "loss": 0.7668,
+      "step": 1741
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.290968285553796e-05,
+      "loss": 0.7698,
+      "step": 1742
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.290213950691732e-05,
+      "loss": 0.7883,
+      "step": 1743
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.289459435460893e-05,
+      "loss": 0.824,
+      "step": 1744
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2887047403302118e-05,
+      "loss": 0.7861,
+      "step": 1745
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2879498657687335e-05,
+      "loss": 0.7573,
+      "step": 1746
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2871948122456149e-05,
+      "loss": 0.7568,
+      "step": 1747
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2864395802301237e-05,
+      "loss": 0.7451,
+      "step": 1748
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2856841701916389e-05,
+      "loss": 0.8125,
+      "step": 1749
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2849285825996498e-05,
+      "loss": 0.8223,
+      "step": 1750
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.284172817923756e-05,
+      "loss": 0.7622,
+      "step": 1751
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2834168766336678e-05,
+      "loss": 0.7122,
+      "step": 1752
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2826607591992046e-05,
+      "loss": 0.7632,
+      "step": 1753
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2819044660902953e-05,
+      "loss": 0.7393,
+      "step": 1754
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2811479977769782e-05,
+      "loss": 0.7922,
+      "step": 1755
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.280391354729401e-05,
+      "loss": 0.7778,
+      "step": 1756
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2796345374178185e-05,
+      "loss": 0.76,
+      "step": 1757
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2788775463125957e-05,
+      "loss": 0.771,
+      "step": 1758
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2781203818842038e-05,
+      "loss": 0.7908,
+      "step": 1759
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2773630446032235e-05,
+      "loss": 0.7773,
+      "step": 1760
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2766055349403415e-05,
+      "loss": 0.7937,
+      "step": 1761
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2758478533663518e-05,
+      "loss": 0.7632,
+      "step": 1762
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2750900003521565e-05,
+      "loss": 0.7664,
+      "step": 1763
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2743319763687627e-05,
+      "loss": 0.8027,
+      "step": 1764
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2735737818872844e-05,
+      "loss": 0.7742,
+      "step": 1765
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.272815417378942e-05,
+      "loss": 0.8003,
+      "step": 1766
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2720568833150608e-05,
+      "loss": 0.8083,
+      "step": 1767
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.271298180167072e-05,
+      "loss": 0.8152,
+      "step": 1768
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2705393084065114e-05,
+      "loss": 0.7773,
+      "step": 1769
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2697802685050203e-05,
+      "loss": 0.8096,
+      "step": 1770
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2690210609343439e-05,
+      "loss": 0.7744,
+      "step": 1771
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2682616861663319e-05,
+      "loss": 0.834,
+      "step": 1772
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2675021446729377e-05,
+      "loss": 0.8296,
+      "step": 1773
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2667424369262186e-05,
+      "loss": 0.7427,
+      "step": 1774
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2659825633983351e-05,
+      "loss": 0.8,
+      "step": 1775
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2652225245615503e-05,
+      "loss": 0.7837,
+      "step": 1776
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2644623208882311e-05,
+      "loss": 0.8271,
+      "step": 1777
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2637019528508457e-05,
+      "loss": 0.8157,
+      "step": 1778
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.262941420921965e-05,
+      "loss": 0.8328,
+      "step": 1779
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2621807255742619e-05,
+      "loss": 0.7756,
+      "step": 1780
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2614198672805105e-05,
+      "loss": 0.771,
+      "step": 1781
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2606588465135865e-05,
+      "loss": 0.7896,
+      "step": 1782
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2598976637464663e-05,
+      "loss": 0.7859,
+      "step": 1783
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2591363194522268e-05,
+      "loss": 0.7803,
+      "step": 1784
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2583748141040461e-05,
+      "loss": 0.7688,
+      "step": 1785
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2576131481752018e-05,
+      "loss": 0.7668,
+      "step": 1786
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2568513221390713e-05,
+      "loss": 0.7947,
+      "step": 1787
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2560893364691314e-05,
+      "loss": 0.7791,
+      "step": 1788
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2553271916389587e-05,
+      "loss": 0.7434,
+      "step": 1789
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2545648881222277e-05,
+      "loss": 0.8218,
+      "step": 1790
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2538024263927128e-05,
+      "loss": 0.7761,
+      "step": 1791
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2530398069242859e-05,
+      "loss": 0.7451,
+      "step": 1792
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.252277030190917e-05,
+      "loss": 0.8164,
+      "step": 1793
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2515140966666742e-05,
+      "loss": 0.8145,
+      "step": 1794
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2507510068257223e-05,
+      "loss": 0.7881,
+      "step": 1795
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2499877611423241e-05,
+      "loss": 0.7842,
+      "step": 1796
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2492243600908391e-05,
+      "loss": 0.7542,
+      "step": 1797
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2484608041457227e-05,
+      "loss": 0.8054,
+      "step": 1798
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2476970937815274e-05,
+      "loss": 0.7427,
+      "step": 1799
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2469332294729012e-05,
+      "loss": 0.772,
+      "step": 1800
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2461692116945878e-05,
+      "loss": 0.7085,
+      "step": 1801
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2454050409214266e-05,
+      "loss": 0.7834,
+      "step": 1802
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2446407176283515e-05,
+      "loss": 0.7729,
+      "step": 1803
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2438762422903921e-05,
+      "loss": 0.7478,
+      "step": 1804
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2431116153826712e-05,
+      "loss": 0.7996,
+      "step": 1805
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.242346837380407e-05,
+      "loss": 0.7795,
+      "step": 1806
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2415819087589113e-05,
+      "loss": 0.7815,
+      "step": 1807
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.240816829993589e-05,
+      "loss": 0.7993,
+      "step": 1808
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2400516015599389e-05,
+      "loss": 0.8516,
+      "step": 1809
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2392862239335523e-05,
+      "loss": 0.762,
+      "step": 1810
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.238520697590114e-05,
+      "loss": 0.812,
+      "step": 1811
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2377550230054006e-05,
+      "loss": 0.7468,
+      "step": 1812
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2369892006552806e-05,
+      "loss": 0.7422,
+      "step": 1813
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2362232310157149e-05,
+      "loss": 0.814,
+      "step": 1814
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.235457114562756e-05,
+      "loss": 0.7739,
+      "step": 1815
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2346908517725472e-05,
+      "loss": 0.7622,
+      "step": 1816
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2339244431213229e-05,
+      "loss": 0.8379,
+      "step": 1817
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2331578890854085e-05,
+      "loss": 0.8008,
+      "step": 1818
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.232391190141219e-05,
+      "loss": 0.7349,
+      "step": 1819
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2316243467652603e-05,
+      "loss": 0.7869,
+      "step": 1820
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2308573594341275e-05,
+      "loss": 0.8059,
+      "step": 1821
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2300902286245055e-05,
+      "loss": 0.7886,
+      "step": 1822
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.229322954813168e-05,
+      "loss": 0.7708,
+      "step": 1823
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.228555538476978e-05,
+      "loss": 0.7834,
+      "step": 1824
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2277879800928867e-05,
+      "loss": 0.7366,
+      "step": 1825
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.227020280137934e-05,
+      "loss": 0.8384,
+      "step": 1826
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2262524390892473e-05,
+      "loss": 0.7485,
+      "step": 1827
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.225484457424042e-05,
+      "loss": 0.7917,
+      "step": 1828
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2247163356196212e-05,
+      "loss": 0.7395,
+      "step": 1829
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2239480741533744e-05,
+      "loss": 0.7405,
+      "step": 1830
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2231796735027781e-05,
+      "loss": 0.7556,
+      "step": 1831
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2224111341453959e-05,
+      "loss": 0.7861,
+      "step": 1832
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2216424565588767e-05,
+      "loss": 0.8101,
+      "step": 1833
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2208736412209557e-05,
+      "loss": 0.7974,
+      "step": 1834
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.220104688609454e-05,
+      "loss": 0.7771,
+      "step": 1835
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2193355992022778e-05,
+      "loss": 0.7427,
+      "step": 1836
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2185663734774185e-05,
+      "loss": 0.8162,
+      "step": 1837
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2177970119129516e-05,
+      "loss": 0.824,
+      "step": 1838
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2170275149870374e-05,
+      "loss": 0.8022,
+      "step": 1839
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2162578831779208e-05,
+      "loss": 0.7644,
+      "step": 1840
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.21548811696393e-05,
+      "loss": 0.7556,
+      "step": 1841
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2147182168234767e-05,
+      "loss": 0.7334,
+      "step": 1842
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2139481832350559e-05,
+      "loss": 0.8352,
+      "step": 1843
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2131780166772458e-05,
+      "loss": 0.7803,
+      "step": 1844
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.212407717628707e-05,
+      "loss": 0.7124,
+      "step": 1845
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2116372865681824e-05,
+      "loss": 0.7188,
+      "step": 1846
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2108667239744969e-05,
+      "loss": 0.7695,
+      "step": 1847
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2100960303265578e-05,
+      "loss": 0.7671,
+      "step": 1848
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2093252061033532e-05,
+      "loss": 0.7424,
+      "step": 1849
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2085542517839518e-05,
+      "loss": 0.7654,
+      "step": 1850
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2077831678475047e-05,
+      "loss": 0.7561,
+      "step": 1851
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2070119547732424e-05,
+      "loss": 0.7742,
+      "step": 1852
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2062406130404755e-05,
+      "loss": 0.7446,
+      "step": 1853
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2054691431285962e-05,
+      "loss": 0.7312,
+      "step": 1854
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.204697545517074e-05,
+      "loss": 0.8176,
+      "step": 1855
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2039258206854594e-05,
+      "loss": 0.748,
+      "step": 1856
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2031539691133818e-05,
+      "loss": 0.7698,
+      "step": 1857
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2023819912805486e-05,
+      "loss": 0.738,
+      "step": 1858
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2016098876667465e-05,
+      "loss": 0.7756,
+      "step": 1859
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.20083765875184e-05,
+      "loss": 0.7571,
+      "step": 1860
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2000653050157713e-05,
+      "loss": 0.7737,
+      "step": 1861
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1992928269385607e-05,
+      "loss": 0.8191,
+      "step": 1862
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1985202250003054e-05,
+      "loss": 0.8188,
+      "step": 1863
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1977474996811793e-05,
+      "loss": 0.7886,
+      "step": 1864
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1969746514614338e-05,
+      "loss": 0.7539,
+      "step": 1865
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1962016808213958e-05,
+      "loss": 0.7712,
+      "step": 1866
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.195428588241469e-05,
+      "loss": 0.8025,
+      "step": 1867
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1946553742021324e-05,
+      "loss": 0.7925,
+      "step": 1868
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1938820391839408e-05,
+      "loss": 0.7756,
+      "step": 1869
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1931085836675243e-05,
+      "loss": 0.8306,
+      "step": 1870
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1923350081335871e-05,
+      "loss": 0.8098,
+      "step": 1871
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1915613130629086e-05,
+      "loss": 0.7988,
+      "step": 1872
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1907874989363428e-05,
+      "loss": 0.7678,
+      "step": 1873
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1900135662348173e-05,
+      "loss": 0.7869,
+      "step": 1874
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1892395154393333e-05,
+      "loss": 0.7578,
+      "step": 1875
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1884653470309655e-05,
+      "loss": 0.7466,
+      "step": 1876
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1876910614908615e-05,
+      "loss": 0.7732,
+      "step": 1877
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1869166593002422e-05,
+      "loss": 0.8174,
+      "step": 1878
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1861421409404006e-05,
+      "loss": 0.8262,
+      "step": 1879
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.185367506892702e-05,
+      "loss": 0.7524,
+      "step": 1880
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1845927576385838e-05,
+      "loss": 0.7683,
+      "step": 1881
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1838178936595547e-05,
+      "loss": 0.772,
+      "step": 1882
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1830429154371945e-05,
+      "loss": 0.7637,
+      "step": 1883
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1822678234531549e-05,
+      "loss": 0.8406,
+      "step": 1884
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1814926181891573e-05,
+      "loss": 0.7852,
+      "step": 1885
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1807173001269938e-05,
+      "loss": 0.8049,
+      "step": 1886
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1799418697485272e-05,
+      "loss": 0.7454,
+      "step": 1887
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1791663275356892e-05,
+      "loss": 0.7915,
+      "step": 1888
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1783906739704819e-05,
+      "loss": 0.7588,
+      "step": 1889
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1776149095349757e-05,
+      "loss": 0.8176,
+      "step": 1890
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.17683903471131e-05,
+      "loss": 0.8254,
+      "step": 1891
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1760630499816945e-05,
+      "loss": 0.7949,
+      "step": 1892
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1752869558284048e-05,
+      "loss": 0.8184,
+      "step": 1893
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1745107527337855e-05,
+      "loss": 0.7937,
+      "step": 1894
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1737344411802496e-05,
+      "loss": 0.834,
+      "step": 1895
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1729580216502765e-05,
+      "loss": 0.7751,
+      "step": 1896
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.172181494626413e-05,
+      "loss": 0.7747,
+      "step": 1897
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1714048605912731e-05,
+      "loss": 0.77,
+      "step": 1898
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1706281200275368e-05,
+      "loss": 0.7976,
+      "step": 1899
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1698512734179503e-05,
+      "loss": 0.8262,
+      "step": 1900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1690743212453265e-05,
+      "loss": 0.8027,
+      "step": 1901
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.168297263992543e-05,
+      "loss": 0.7732,
+      "step": 1902
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1675201021425431e-05,
+      "loss": 0.7791,
+      "step": 1903
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.166742836178335e-05,
+      "loss": 0.8123,
+      "step": 1904
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1659654665829918e-05,
+      "loss": 0.7327,
+      "step": 1905
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1651879938396505e-05,
+      "loss": 0.7705,
+      "step": 1906
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1644104184315128e-05,
+      "loss": 0.7363,
+      "step": 1907
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1636327408418441e-05,
+      "loss": 0.8291,
+      "step": 1908
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.162854961553973e-05,
+      "loss": 0.7651,
+      "step": 1909
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1620770810512914e-05,
+      "loss": 0.7771,
+      "step": 1910
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1612990998172544e-05,
+      "loss": 0.7854,
+      "step": 1911
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1605210183353796e-05,
+      "loss": 0.7437,
+      "step": 1912
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1597428370892463e-05,
+      "loss": 0.8335,
+      "step": 1913
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1589645565624968e-05,
+      "loss": 0.7566,
+      "step": 1914
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1581861772388343e-05,
+      "loss": 0.8035,
+      "step": 1915
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1574076996020238e-05,
+      "loss": 0.731,
+      "step": 1916
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1566291241358916e-05,
+      "loss": 0.8135,
+      "step": 1917
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.155850451324324e-05,
+      "loss": 0.7695,
+      "step": 1918
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1550716816512686e-05,
+      "loss": 0.7778,
+      "step": 1919
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1542928156007332e-05,
+      "loss": 0.7852,
+      "step": 1920
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1535138536567845e-05,
+      "loss": 0.7891,
+      "step": 1921
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.15273479630355e-05,
+      "loss": 0.7749,
+      "step": 1922
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.151955644025216e-05,
+      "loss": 0.783,
+      "step": 1923
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1511763973060273e-05,
+      "loss": 0.7913,
+      "step": 1924
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.150397056630289e-05,
+      "loss": 0.7883,
+      "step": 1925
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1496176224823623e-05,
+      "loss": 0.8027,
+      "step": 1926
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1488380953466683e-05,
+      "loss": 0.7881,
+      "step": 1927
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1480584757076852e-05,
+      "loss": 0.8184,
+      "step": 1928
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1472787640499488e-05,
+      "loss": 0.8445,
+      "step": 1929
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1464989608580517e-05,
+      "loss": 0.7751,
+      "step": 1930
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1457190666166443e-05,
+      "loss": 0.7113,
+      "step": 1931
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1449390818104326e-05,
+      "loss": 0.8074,
+      "step": 1932
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1441590069241794e-05,
+      "loss": 0.8159,
+      "step": 1933
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1433788424427038e-05,
+      "loss": 0.7776,
+      "step": 1934
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1425985888508797e-05,
+      "loss": 0.749,
+      "step": 1935
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1418182466336372e-05,
+      "loss": 0.7849,
+      "step": 1936
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.141037816275961e-05,
+      "loss": 0.8154,
+      "step": 1937
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1402572982628905e-05,
+      "loss": 0.7532,
+      "step": 1938
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1394766930795202e-05,
+      "loss": 0.7456,
+      "step": 1939
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1386960012109983e-05,
+      "loss": 0.791,
+      "step": 1940
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1379152231425268e-05,
+      "loss": 0.7798,
+      "step": 1941
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1371343593593618e-05,
+      "loss": 0.7959,
+      "step": 1942
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1363534103468119e-05,
+      "loss": 0.7639,
+      "step": 1943
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1355723765902388e-05,
+      "loss": 0.7927,
+      "step": 1944
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1347912585750581e-05,
+      "loss": 0.7917,
+      "step": 1945
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1340100567867358e-05,
+      "loss": 0.8127,
+      "step": 1946
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.133228771710792e-05,
+      "loss": 0.8105,
+      "step": 1947
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1324474038327967e-05,
+      "loss": 0.72,
+      "step": 1948
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1316659536383723e-05,
+      "loss": 0.7456,
+      "step": 1949
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1308844216131928e-05,
+      "loss": 0.8293,
+      "step": 1950
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1301028082429823e-05,
+      "loss": 0.7822,
+      "step": 1951
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1293211140135152e-05,
+      "loss": 0.7605,
+      "step": 1952
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1285393394106173e-05,
+      "loss": 0.7898,
+      "step": 1953
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1277574849201632e-05,
+      "loss": 0.802,
+      "step": 1954
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1269755510280778e-05,
+      "loss": 0.7888,
+      "step": 1955
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1261935382203353e-05,
+      "loss": 0.7844,
+      "step": 1956
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1254114469829585e-05,
+      "loss": 0.8169,
+      "step": 1957
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1246292778020198e-05,
+      "loss": 0.7852,
+      "step": 1958
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1238470311636392e-05,
+      "loss": 0.7798,
+      "step": 1959
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1230647075539849e-05,
+      "loss": 0.8269,
+      "step": 1960
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1222823074592737e-05,
+      "loss": 0.8445,
+      "step": 1961
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1214998313657693e-05,
+      "loss": 0.804,
+      "step": 1962
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1207172797597823e-05,
+      "loss": 0.7578,
+      "step": 1963
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1199346531276714e-05,
+      "loss": 0.7842,
+      "step": 1964
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1191519519558407e-05,
+      "loss": 0.7693,
+      "step": 1965
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1183691767307412e-05,
+      "loss": 0.7852,
+      "step": 1966
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1175863279388703e-05,
+      "loss": 0.8013,
+      "step": 1967
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.11680340606677e-05,
+      "loss": 0.8276,
+      "step": 1968
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1160204116010291e-05,
+      "loss": 0.7297,
+      "step": 1969
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1152373450282808e-05,
+      "loss": 0.8254,
+      "step": 1970
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1144542068352026e-05,
+      "loss": 0.7771,
+      "step": 1971
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1136709975085173e-05,
+      "loss": 0.7783,
+      "step": 1972
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1128877175349916e-05,
+      "loss": 0.8062,
+      "step": 1973
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1121043674014367e-05,
+      "loss": 0.7769,
+      "step": 1974
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1113209475947064e-05,
+      "loss": 0.8113,
+      "step": 1975
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1105374586016984e-05,
+      "loss": 0.8167,
+      "step": 1976
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1097539009093527e-05,
+      "loss": 0.8096,
+      "step": 1977
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1089702750046537e-05,
+      "loss": 0.7759,
+      "step": 1978
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1081865813746259e-05,
+      "loss": 0.7305,
+      "step": 1979
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1074028205063383e-05,
+      "loss": 0.7861,
+      "step": 1980
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1066189928868996e-05,
+      "loss": 0.8066,
+      "step": 1981
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.105835099003461e-05,
+      "loss": 0.783,
+      "step": 1982
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1050511393432149e-05,
+      "loss": 0.7922,
+      "step": 1983
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1042671143933941e-05,
+      "loss": 0.7578,
+      "step": 1984
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1034830246412725e-05,
+      "loss": 0.7559,
+      "step": 1985
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.102698870574164e-05,
+      "loss": 0.7808,
+      "step": 1986
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1019146526794227e-05,
+      "loss": 0.7625,
+      "step": 1987
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.101130371444442e-05,
+      "loss": 0.8333,
+      "step": 1988
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1003460273566546e-05,
+      "loss": 0.7295,
+      "step": 1989
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0995616209035331e-05,
+      "loss": 0.7527,
+      "step": 1990
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0987771525725882e-05,
+      "loss": 0.7515,
+      "step": 1991
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0979926228513686e-05,
+      "loss": 0.7961,
+      "step": 1992
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0972080322274622e-05,
+      "loss": 0.7786,
+      "step": 1993
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0964233811884942e-05,
+      "loss": 0.7722,
+      "step": 1994
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0956386702221273e-05,
+      "loss": 0.7783,
+      "step": 1995
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0948538998160614e-05,
+      "loss": 0.7351,
+      "step": 1996
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.094069070458034e-05,
+      "loss": 0.8091,
+      "step": 1997
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0932841826358181e-05,
+      "loss": 0.7407,
+      "step": 1998
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0924992368372236e-05,
+      "loss": 0.7649,
+      "step": 1999
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0917142335500973e-05,
+      "loss": 0.7136,
+      "step": 2000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0909291732623202e-05,
+      "loss": 0.7878,
+      "step": 2001
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0901440564618095e-05,
+      "loss": 0.8083,
+      "step": 2002
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0893588836365181e-05,
+      "loss": 0.7483,
+      "step": 2003
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0885736552744325e-05,
+      "loss": 0.7739,
+      "step": 2004
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0877883718635744e-05,
+      "loss": 0.7915,
+      "step": 2005
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0870030338919995e-05,
+      "loss": 0.8628,
+      "step": 2006
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.086217641847798e-05,
+      "loss": 0.8193,
+      "step": 2007
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0854321962190925e-05,
+      "loss": 0.8564,
+      "step": 2008
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0846466974940403e-05,
+      "loss": 0.7793,
+      "step": 2009
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0838611461608308e-05,
+      "loss": 0.7297,
+      "step": 2010
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0830755427076865e-05,
+      "loss": 0.802,
+      "step": 2011
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0822898876228613e-05,
+      "loss": 0.748,
+      "step": 2012
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0815041813946434e-05,
+      "loss": 0.7512,
+      "step": 2013
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0807184245113505e-05,
+      "loss": 0.8123,
+      "step": 2014
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0799326174613324e-05,
+      "loss": 0.7708,
+      "step": 2015
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0791467607329713e-05,
+      "loss": 0.8274,
+      "step": 2016
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0783608548146787e-05,
+      "loss": 0.7537,
+      "step": 2017
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0775749001948974e-05,
+      "loss": 0.8113,
+      "step": 2018
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0767888973621006e-05,
+      "loss": 0.8174,
+      "step": 2019
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.076002846804791e-05,
+      "loss": 0.7717,
+      "step": 2020
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0752167490115012e-05,
+      "loss": 0.8186,
+      "step": 2021
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.074430604470793e-05,
+      "loss": 0.7527,
+      "step": 2022
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0736444136712578e-05,
+      "loss": 0.7671,
+      "step": 2023
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0728581771015154e-05,
+      "loss": 0.7576,
+      "step": 2024
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0720718952502139e-05,
+      "loss": 0.7776,
+      "step": 2025
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0712855686060293e-05,
+      "loss": 0.7681,
+      "step": 2026
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0704991976576662e-05,
+      "loss": 0.7793,
+      "step": 2027
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0697127828938564e-05,
+      "loss": 0.7598,
+      "step": 2028
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0689263248033581e-05,
+      "loss": 0.8052,
+      "step": 2029
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.068139823874958e-05,
+      "loss": 0.7737,
+      "step": 2030
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0673532805974687e-05,
+      "loss": 0.793,
+      "step": 2031
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.066566695459728e-05,
+      "loss": 0.7705,
+      "step": 2032
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.065780068950602e-05,
+      "loss": 0.7542,
+      "step": 2033
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0649934015589805e-05,
+      "loss": 0.7739,
+      "step": 2034
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0642066937737794e-05,
+      "loss": 0.8074,
+      "step": 2035
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0634199460839403e-05,
+      "loss": 0.7622,
+      "step": 2036
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0626331589784286e-05,
+      "loss": 0.7791,
+      "step": 2037
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0618463329462349e-05,
+      "loss": 0.8018,
+      "step": 2038
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0610594684763737e-05,
+      "loss": 0.7717,
+      "step": 2039
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0602725660578835e-05,
+      "loss": 0.8013,
+      "step": 2040
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0594856261798264e-05,
+      "loss": 0.7866,
+      "step": 2041
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0586986493312873e-05,
+      "loss": 0.8042,
+      "step": 2042
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.057911636001375e-05,
+      "loss": 0.8,
+      "step": 2043
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0571245866792204e-05,
+      "loss": 0.8218,
+      "step": 2044
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0563375018539763e-05,
+      "loss": 0.7761,
+      "step": 2045
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.055550382014819e-05,
+      "loss": 0.7942,
+      "step": 2046
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.054763227650945e-05,
+      "loss": 0.7686,
+      "step": 2047
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.053976039251573e-05,
+      "loss": 0.7615,
+      "step": 2048
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0531888173059428e-05,
+      "loss": 0.7881,
+      "step": 2049
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0524015623033155e-05,
+      "loss": 0.7568,
+      "step": 2050
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0516142747329715e-05,
+      "loss": 0.7886,
+      "step": 2051
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0508269550842125e-05,
+      "loss": 0.8149,
+      "step": 2052
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0500396038463603e-05,
+      "loss": 0.749,
+      "step": 2053
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.049252221508755e-05,
+      "loss": 0.7583,
+      "step": 2054
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0484648085607578e-05,
+      "loss": 0.7988,
+      "step": 2055
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0476773654917476e-05,
+      "loss": 0.7822,
+      "step": 2056
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0468898927911224e-05,
+      "loss": 0.7959,
+      "step": 2057
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0461023909482987e-05,
+      "loss": 0.7908,
+      "step": 2058
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0453148604527112e-05,
+      "loss": 0.7632,
+      "step": 2059
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0445273017938122e-05,
+      "loss": 0.7852,
+      "step": 2060
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0437397154610717e-05,
+      "loss": 0.7922,
+      "step": 2061
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0429521019439765e-05,
+      "loss": 0.7627,
+      "step": 2062
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0421644617320313e-05,
+      "loss": 0.8015,
+      "step": 2063
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0413767953147558e-05,
+      "loss": 0.804,
+      "step": 2064
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.040589103181687e-05,
+      "loss": 0.7622,
+      "step": 2065
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0398013858223784e-05,
+      "loss": 0.719,
+      "step": 2066
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.039013643726398e-05,
+      "loss": 0.7573,
+      "step": 2067
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0382258773833297e-05,
+      "loss": 0.7979,
+      "step": 2068
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0374380872827725e-05,
+      "loss": 0.7793,
+      "step": 2069
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0366502739143403e-05,
+      "loss": 0.772,
+      "step": 2070
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.035862437767661e-05,
+      "loss": 0.804,
+      "step": 2071
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0350745793323776e-05,
+      "loss": 0.7661,
+      "step": 2072
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0342866990981456e-05,
+      "loss": 0.772,
+      "step": 2073
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.033498797554635e-05,
+      "loss": 0.792,
+      "step": 2074
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0327108751915288e-05,
+      "loss": 0.7527,
+      "step": 2075
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0319229324985228e-05,
+      "loss": 0.7688,
+      "step": 2076
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0311349699653256e-05,
+      "loss": 0.7869,
+      "step": 2077
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0303469880816583e-05,
+      "loss": 0.8049,
+      "step": 2078
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.029558987337254e-05,
+      "loss": 0.7751,
+      "step": 2079
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.028770968221857e-05,
+      "loss": 0.7754,
+      "step": 2080
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.027982931225223e-05,
+      "loss": 0.7527,
+      "step": 2081
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.02719487683712e-05,
+      "loss": 0.8354,
+      "step": 2082
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0264068055473258e-05,
+      "loss": 0.7852,
+      "step": 2083
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.025618717845628e-05,
+      "loss": 0.7671,
+      "step": 2084
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0248306142218268e-05,
+      "loss": 0.781,
+      "step": 2085
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0240424951657295e-05,
+      "loss": 0.7871,
+      "step": 2086
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0232543611671548e-05,
+      "loss": 0.8184,
+      "step": 2087
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0224662127159302e-05,
+      "loss": 0.762,
+      "step": 2088
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0216780503018919e-05,
+      "loss": 0.7915,
+      "step": 2089
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0208898744148854e-05,
+      "loss": 0.8108,
+      "step": 2090
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.020101685544764e-05,
+      "loss": 0.8096,
+      "step": 2091
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0193134841813891e-05,
+      "loss": 0.7888,
+      "step": 2092
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0185252708146304e-05,
+      "loss": 0.7981,
+      "step": 2093
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0177370459343648e-05,
+      "loss": 0.7886,
+      "step": 2094
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0169488100304753e-05,
+      "loss": 0.7869,
+      "step": 2095
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0161605635928538e-05,
+      "loss": 0.7871,
+      "step": 2096
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.015372307111397e-05,
+      "loss": 0.7803,
+      "step": 2097
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0145840410760086e-05,
+      "loss": 0.782,
+      "step": 2098
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0137957659765985e-05,
+      "loss": 0.7698,
+      "step": 2099
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0130074823030814e-05,
+      "loss": 0.7483,
+      "step": 2100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.012219190545378e-05,
+      "loss": 0.7729,
+      "step": 2101
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0114308911934137e-05,
+      "loss": 0.769,
+      "step": 2102
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0106425847371188e-05,
+      "loss": 0.7832,
+      "step": 2103
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0098542716664282e-05,
+      "loss": 0.8083,
+      "step": 2104
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0090659524712802e-05,
+      "loss": 0.7561,
+      "step": 2105
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0082776276416177e-05,
+      "loss": 0.7703,
+      "step": 2106
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0074892976673866e-05,
+      "loss": 0.8037,
+      "step": 2107
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0067009630385362e-05,
+      "loss": 0.7886,
+      "step": 2108
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0059126242450187e-05,
+      "loss": 0.7925,
+      "step": 2109
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.005124281776789e-05,
+      "loss": 0.7793,
+      "step": 2110
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0043359361238036e-05,
+      "loss": 0.7769,
+      "step": 2111
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0035475877760222e-05,
+      "loss": 0.7869,
+      "step": 2112
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.002759237223405e-05,
+      "loss": 0.8108,
+      "step": 2113
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0019708849559142e-05,
+      "loss": 0.804,
+      "step": 2114
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0011825314635131e-05,
+      "loss": 0.7925,
+      "step": 2115
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0003941772361651e-05,
+      "loss": 0.7122,
+      "step": 2116
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.99605822763835e-06,
+      "loss": 0.7332,
+      "step": 2117
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.98817468536487e-06,
+      "loss": 0.8318,
+      "step": 2118
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.980291150440861e-06,
+      "loss": 0.8123,
+      "step": 2119
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.972407627765952e-06,
+      "loss": 0.8049,
+      "step": 2120
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.96452412223978e-06,
+      "loss": 0.7788,
+      "step": 2121
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.956640638761967e-06,
+      "loss": 0.7668,
+      "step": 2122
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.948757182232114e-06,
+      "loss": 0.7644,
+      "step": 2123
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.940873757549815e-06,
+      "loss": 0.7568,
+      "step": 2124
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.932990369614642e-06,
+      "loss": 0.7986,
+      "step": 2125
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.925107023326137e-06,
+      "loss": 0.76,
+      "step": 2126
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.917223723583825e-06,
+      "loss": 0.8167,
+      "step": 2127
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.909340475287201e-06,
+      "loss": 0.7739,
+      "step": 2128
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.90145728333572e-06,
+      "loss": 0.7498,
+      "step": 2129
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.893574152628815e-06,
+      "loss": 0.7698,
+      "step": 2130
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.885691088065866e-06,
+      "loss": 0.8118,
+      "step": 2131
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.877808094546223e-06,
+      "loss": 0.782,
+      "step": 2132
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.86992517696919e-06,
+      "loss": 0.7659,
+      "step": 2133
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.862042340234019e-06,
+      "loss": 0.8123,
+      "step": 2134
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.854159589239917e-06,
+      "loss": 0.8083,
+      "step": 2135
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.846276928886032e-06,
+      "loss": 0.8057,
+      "step": 2136
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.838394364071467e-06,
+      "loss": 0.7861,
+      "step": 2137
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.830511899695249e-06,
+      "loss": 0.7944,
+      "step": 2138
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.822629540656357e-06,
+      "loss": 0.7451,
+      "step": 2139
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.814747291853696e-06,
+      "loss": 0.7915,
+      "step": 2140
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.80686515818611e-06,
+      "loss": 0.7769,
+      "step": 2141
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.798983144552362e-06,
+      "loss": 0.8147,
+      "step": 2142
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.791101255851146e-06,
+      "loss": 0.8035,
+      "step": 2143
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.783219496981083e-06,
+      "loss": 0.7832,
+      "step": 2144
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.775337872840702e-06,
+      "loss": 0.7986,
+      "step": 2145
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.767456388328454e-06,
+      "loss": 0.7324,
+      "step": 2146
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.75957504834271e-06,
+      "loss": 0.7578,
+      "step": 2147
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.751693857781737e-06,
+      "loss": 0.7939,
+      "step": 2148
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.74381282154372e-06,
+      "loss": 0.7832,
+      "step": 2149
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.735931944526749e-06,
+      "loss": 0.7849,
+      "step": 2150
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.728051231628802e-06,
+      "loss": 0.7795,
+      "step": 2151
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.72017068774777e-06,
+      "loss": 0.7678,
+      "step": 2152
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.712290317781437e-06,
+      "loss": 0.7493,
+      "step": 2153
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.704410126627464e-06,
+      "loss": 0.7717,
+      "step": 2154
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.696530119183417e-06,
+      "loss": 0.8323,
+      "step": 2155
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.688650300346749e-06,
+      "loss": 0.7954,
+      "step": 2156
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.680770675014776e-06,
+      "loss": 0.7832,
+      "step": 2157
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.672891248084716e-06,
+      "loss": 0.7544,
+      "step": 2158
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.665012024453656e-06,
+      "loss": 0.7905,
+      "step": 2159
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.657133009018548e-06,
+      "loss": 0.7842,
+      "step": 2160
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.649254206676226e-06,
+      "loss": 0.7744,
+      "step": 2161
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.64137562232339e-06,
+      "loss": 0.7593,
+      "step": 2162
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.6334972608566e-06,
+      "loss": 0.7263,
+      "step": 2163
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.625619127172278e-06,
+      "loss": 0.7317,
+      "step": 2164
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.617741226166706e-06,
+      "loss": 0.7397,
+      "step": 2165
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.609863562736023e-06,
+      "loss": 0.8171,
+      "step": 2166
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.60198614177622e-06,
+      "loss": 0.7317,
+      "step": 2167
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.59410896818313e-06,
+      "loss": 0.7808,
+      "step": 2168
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.586232046852447e-06,
+      "loss": 0.7974,
+      "step": 2169
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.578355382679692e-06,
+      "loss": 0.7222,
+      "step": 2170
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.570478980560233e-06,
+      "loss": 0.7776,
+      "step": 2171
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.562602845389288e-06,
+      "loss": 0.7419,
+      "step": 2172
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.55472698206188e-06,
+      "loss": 0.822,
+      "step": 2173
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.54685139547289e-06,
+      "loss": 0.7852,
+      "step": 2174
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.538976090517018e-06,
+      "loss": 0.803,
+      "step": 2175
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.531101072088779e-06,
+      "loss": 0.7659,
+      "step": 2176
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.523226345082525e-06,
+      "loss": 0.7559,
+      "step": 2177
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.515351914392426e-06,
+      "loss": 0.7288,
+      "step": 2178
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.507477784912451e-06,
+      "loss": 0.729,
+      "step": 2179
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.4996039615364e-06,
+      "loss": 0.8035,
+      "step": 2180
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.491730449157878e-06,
+      "loss": 0.7485,
+      "step": 2181
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.483857252670289e-06,
+      "loss": 0.7566,
+      "step": 2182
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.47598437696685e-06,
+      "loss": 0.7874,
+      "step": 2183
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.468111826940572e-06,
+      "loss": 0.803,
+      "step": 2184
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.460239607484274e-06,
+      "loss": 0.791,
+      "step": 2185
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.452367723490553e-06,
+      "loss": 0.7927,
+      "step": 2186
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.444496179851812e-06,
+      "loss": 0.7637,
+      "step": 2187
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.43662498146024e-06,
+      "loss": 0.7666,
+      "step": 2188
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.428754133207799e-06,
+      "loss": 0.7815,
+      "step": 2189
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.42088363998625e-06,
+      "loss": 0.7803,
+      "step": 2190
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.41301350668713e-06,
+      "loss": 0.7439,
+      "step": 2191
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.40514373820174e-06,
+      "loss": 0.8262,
+      "step": 2192
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.397274339421168e-06,
+      "loss": 0.7605,
+      "step": 2193
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.389405315236266e-06,
+      "loss": 0.7688,
+      "step": 2194
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.381536670537653e-06,
+      "loss": 0.7754,
+      "step": 2195
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.373668410215717e-06,
+      "loss": 0.8125,
+      "step": 2196
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.3658005391606e-06,
+      "loss": 0.74,
+      "step": 2197
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.35793306226221e-06,
+      "loss": 0.7366,
+      "step": 2198
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.350065984410198e-06,
+      "loss": 0.7786,
+      "step": 2199
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.342199310493983e-06,
+      "loss": 0.7639,
+      "step": 2200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.334333045402721e-06,
+      "loss": 0.7366,
+      "step": 2201
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.326467194025317e-06,
+      "loss": 0.8132,
+      "step": 2202
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.318601761250421e-06,
+      "loss": 0.7539,
+      "step": 2203
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.310736751966422e-06,
+      "loss": 0.7812,
+      "step": 2204
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.30287217106144e-06,
+      "loss": 0.8003,
+      "step": 2205
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.29500802342334e-06,
+      "loss": 0.7537,
+      "step": 2206
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.28714431393971e-06,
+      "loss": 0.7783,
+      "step": 2207
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.279281047497863e-06,
+      "loss": 0.781,
+      "step": 2208
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.271418228984846e-06,
+      "loss": 0.7432,
+      "step": 2209
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.263555863287424e-06,
+      "loss": 0.7927,
+      "step": 2210
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.255693955292072e-06,
+      "loss": 0.7629,
+      "step": 2211
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.24783250988499e-06,
+      "loss": 0.729,
+      "step": 2212
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.239971531952095e-06,
+      "loss": 0.7854,
+      "step": 2213
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.232111026378997e-06,
+      "loss": 0.8369,
+      "step": 2214
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.224250998051028e-06,
+      "loss": 0.7224,
+      "step": 2215
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.216391451853218e-06,
+      "loss": 0.7471,
+      "step": 2216
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.208532392670289e-06,
+      "loss": 0.8032,
+      "step": 2217
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.200673825386676e-06,
+      "loss": 0.7869,
+      "step": 2218
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.192815754886502e-06,
+      "loss": 0.8347,
+      "step": 2219
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.18495818605357e-06,
+      "loss": 0.7544,
+      "step": 2220
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.177101123771387e-06,
+      "loss": 0.8003,
+      "step": 2221
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.169244572923142e-06,
+      "loss": 0.8037,
+      "step": 2222
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.161388538391694e-06,
+      "loss": 0.748,
+      "step": 2223
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.153533025059598e-06,
+      "loss": 0.739,
+      "step": 2224
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.145678037809078e-06,
+      "loss": 0.8181,
+      "step": 2225
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.137823581522024e-06,
+      "loss": 0.7891,
+      "step": 2226
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.129969661080009e-06,
+      "loss": 0.7842,
+      "step": 2227
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.122116281364258e-06,
+      "loss": 0.7756,
+      "step": 2228
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.114263447255677e-06,
+      "loss": 0.7522,
+      "step": 2229
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.106411163634822e-06,
+      "loss": 0.7712,
+      "step": 2230
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.098559435381903e-06,
+      "loss": 0.7703,
+      "step": 2231
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.0907082673768e-06,
+      "loss": 0.7693,
+      "step": 2232
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.08285766449903e-06,
+      "loss": 0.7964,
+      "step": 2233
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.075007631627764e-06,
+      "loss": 0.8335,
+      "step": 2234
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.067158173641826e-06,
+      "loss": 0.7781,
+      "step": 2235
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.059309295419665e-06,
+      "loss": 0.6824,
+      "step": 2236
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.051461001839386e-06,
+      "loss": 0.7769,
+      "step": 2237
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.043613297778732e-06,
+      "loss": 0.7754,
+      "step": 2238
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.03576618811506e-06,
+      "loss": 0.7498,
+      "step": 2239
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.027919677725378e-06,
+      "loss": 0.7783,
+      "step": 2240
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.020073771486319e-06,
+      "loss": 0.8025,
+      "step": 2241
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.012228474274123e-06,
+      "loss": 0.77,
+      "step": 2242
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.00438379096467e-06,
+      "loss": 0.8147,
+      "step": 2243
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.996539726433458e-06,
+      "loss": 0.7908,
+      "step": 2244
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.988696285555584e-06,
+      "loss": 0.8035,
+      "step": 2245
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.980853473205776e-06,
+      "loss": 0.7827,
+      "step": 2246
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.973011294258365e-06,
+      "loss": 0.7815,
+      "step": 2247
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.96516975358728e-06,
+      "loss": 0.7312,
+      "step": 2248
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.957328856066062e-06,
+      "loss": 0.7935,
+      "step": 2249
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.949488606567853e-06,
+      "loss": 0.8208,
+      "step": 2250
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.941649009965393e-06,
+      "loss": 0.7734,
+      "step": 2251
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.933810071131006e-06,
+      "loss": 0.8142,
+      "step": 2252
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.925971794936617e-06,
+      "loss": 0.7734,
+      "step": 2253
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.918134186253743e-06,
+      "loss": 0.728,
+      "step": 2254
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.910297249953467e-06,
+      "loss": 0.7949,
+      "step": 2255
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.902460990906474e-06,
+      "loss": 0.7559,
+      "step": 2256
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.894625413983023e-06,
+      "loss": 0.7563,
+      "step": 2257
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.886790524052939e-06,
+      "loss": 0.7327,
+      "step": 2258
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.878956325985636e-06,
+      "loss": 0.8191,
+      "step": 2259
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.871122824650085e-06,
+      "loss": 0.7793,
+      "step": 2260
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.863290024914828e-06,
+      "loss": 0.78,
+      "step": 2261
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.855457931647977e-06,
+      "loss": 0.7793,
+      "step": 2262
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.847626549717196e-06,
+      "loss": 0.7498,
+      "step": 2263
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.83979588398971e-06,
+      "loss": 0.7324,
+      "step": 2264
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.8319659393323e-06,
+      "loss": 0.7681,
+      "step": 2265
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.8241367206113e-06,
+      "loss": 0.7871,
+      "step": 2266
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.816308232692591e-06,
+      "loss": 0.7717,
+      "step": 2267
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.808480480441596e-06,
+      "loss": 0.7822,
+      "step": 2268
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.800653468723293e-06,
+      "loss": 0.708,
+      "step": 2269
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.79282720240218e-06,
+      "loss": 0.7632,
+      "step": 2270
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.78500168634231e-06,
+      "loss": 0.7837,
+      "step": 2271
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.777176925407263e-06,
+      "loss": 0.7561,
+      "step": 2272
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.769352924460154e-06,
+      "loss": 0.7617,
+      "step": 2273
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.761529688363612e-06,
+      "loss": 0.7629,
+      "step": 2274
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.753707221979803e-06,
+      "loss": 0.8301,
+      "step": 2275
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.745885530170418e-06,
+      "loss": 0.7458,
+      "step": 2276
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.738064617796652e-06,
+      "loss": 0.7971,
+      "step": 2277
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.730244489719224e-06,
+      "loss": 0.7815,
+      "step": 2278
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.722425150798373e-06,
+      "loss": 0.7646,
+      "step": 2279
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.714606605893832e-06,
+      "loss": 0.7834,
+      "step": 2280
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.70678885986485e-06,
+      "loss": 0.7966,
+      "step": 2281
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.698971917570183e-06,
+      "loss": 0.8206,
+      "step": 2282
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.691155783868073e-06,
+      "loss": 0.7268,
+      "step": 2283
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.683340463616277e-06,
+      "loss": 0.7646,
+      "step": 2284
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.675525961672038e-06,
+      "loss": 0.7839,
+      "step": 2285
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.667712282892084e-06,
+      "loss": 0.7544,
+      "step": 2286
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.65989943213264e-06,
+      "loss": 0.8137,
+      "step": 2287
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.652087414249424e-06,
+      "loss": 0.8323,
+      "step": 2288
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.644276234097613e-06,
+      "loss": 0.7803,
+      "step": 2289
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.636465896531884e-06,
+      "loss": 0.7581,
+      "step": 2290
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.628656406406388e-06,
+      "loss": 0.7634,
+      "step": 2291
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.620847768574734e-06,
+      "loss": 0.7683,
+      "step": 2292
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.613039987890019e-06,
+      "loss": 0.75,
+      "step": 2293
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.6052330692048e-06,
+      "loss": 0.7449,
+      "step": 2294
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.597427017371096e-06,
+      "loss": 0.7898,
+      "step": 2295
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.589621837240393e-06,
+      "loss": 0.8076,
+      "step": 2296
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.58181753366363e-06,
+      "loss": 0.7822,
+      "step": 2297
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.574014111491208e-06,
+      "loss": 0.7256,
+      "step": 2298
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.566211575572965e-06,
+      "loss": 0.7971,
+      "step": 2299
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.558409930758206e-06,
+      "loss": 0.7539,
+      "step": 2300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.550609181895679e-06,
+      "loss": 0.7532,
+      "step": 2301
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.54280933383356e-06,
+      "loss": 0.7367,
+      "step": 2302
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.535010391419484e-06,
+      "loss": 0.7883,
+      "step": 2303
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.527212359500518e-06,
+      "loss": 0.7678,
+      "step": 2304
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.519415242923152e-06,
+      "loss": 0.7429,
+      "step": 2305
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.511619046533319e-06,
+      "loss": 0.7957,
+      "step": 2306
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.503823775176382e-06,
+      "loss": 0.7949,
+      "step": 2307
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.496029433697116e-06,
+      "loss": 0.7661,
+      "step": 2308
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.488236026939727e-06,
+      "loss": 0.7705,
+      "step": 2309
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.480443559747846e-06,
+      "loss": 0.7871,
+      "step": 2310
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.472652036964504e-06,
+      "loss": 0.7368,
+      "step": 2311
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.464861463432158e-06,
+      "loss": 0.7332,
+      "step": 2312
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.457071843992675e-06,
+      "loss": 0.762,
+      "step": 2313
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.449283183487317e-06,
+      "loss": 0.7461,
+      "step": 2314
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.441495486756762e-06,
+      "loss": 0.7632,
+      "step": 2315
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.433708758641085e-06,
+      "loss": 0.8042,
+      "step": 2316
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.425923003979765e-06,
+      "loss": 0.781,
+      "step": 2317
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.418138227611658e-06,
+      "loss": 0.7573,
+      "step": 2318
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.410354434375034e-06,
+      "loss": 0.7449,
+      "step": 2319
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.40257162910754e-06,
+      "loss": 0.7703,
+      "step": 2320
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.394789816646207e-06,
+      "loss": 0.7417,
+      "step": 2321
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.387009001827457e-06,
+      "loss": 0.7808,
+      "step": 2322
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.37922918948709e-06,
+      "loss": 0.7747,
+      "step": 2323
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.371450384460272e-06,
+      "loss": 0.8462,
+      "step": 2324
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.363672591581562e-06,
+      "loss": 0.7949,
+      "step": 2325
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.355895815684875e-06,
+      "loss": 0.853,
+      "step": 2326
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.348120061603499e-06,
+      "loss": 0.7952,
+      "step": 2327
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.340345334170085e-06,
+      "loss": 0.7805,
+      "step": 2328
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.332571638216652e-06,
+      "loss": 0.7932,
+      "step": 2329
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.32479897857457e-06,
+      "loss": 0.7793,
+      "step": 2330
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.31702736007457e-06,
+      "loss": 0.7605,
+      "step": 2331
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.309256787546738e-06,
+      "loss": 0.7952,
+      "step": 2332
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.301487265820499e-06,
+      "loss": 0.8147,
+      "step": 2333
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.293718799724635e-06,
+      "loss": 0.7717,
+      "step": 2334
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.285951394087275e-06,
+      "loss": 0.7825,
+      "step": 2335
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.278185053735874e-06,
+      "loss": 0.7827,
+      "step": 2336
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.270419783497238e-06,
+      "loss": 0.7678,
+      "step": 2337
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.262655588197505e-06,
+      "loss": 0.7791,
+      "step": 2338
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.254892472662148e-06,
+      "loss": 0.7786,
+      "step": 2339
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.247130441715955e-06,
+      "loss": 0.7283,
+      "step": 2340
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.239369500183057e-06,
+      "loss": 0.7136,
+      "step": 2341
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.231609652886901e-06,
+      "loss": 0.7449,
+      "step": 2342
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.223850904650247e-06,
+      "loss": 0.8123,
+      "step": 2343
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.216093260295183e-06,
+      "loss": 0.77,
+      "step": 2344
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.208336724643113e-06,
+      "loss": 0.7695,
+      "step": 2345
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.200581302514733e-06,
+      "loss": 0.7119,
+      "step": 2346
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.192826998730063e-06,
+      "loss": 0.7986,
+      "step": 2347
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.185073818108434e-06,
+      "loss": 0.7466,
+      "step": 2348
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.177321765468456e-06,
+      "loss": 0.7048,
+      "step": 2349
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.169570845628057e-06,
+      "loss": 0.7527,
+      "step": 2350
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.161821063404458e-06,
+      "loss": 0.7534,
+      "step": 2351
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.154072423614164e-06,
+      "loss": 0.7656,
+      "step": 2352
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.14632493107298e-06,
+      "loss": 0.7505,
+      "step": 2353
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.138578590595998e-06,
+      "loss": 0.8428,
+      "step": 2354
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.13083340699758e-06,
+      "loss": 0.7737,
+      "step": 2355
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.123089385091388e-06,
+      "loss": 0.7815,
+      "step": 2356
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.11534652969035e-06,
+      "loss": 0.7402,
+      "step": 2357
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.107604845606668e-06,
+      "loss": 0.7993,
+      "step": 2358
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.099864337651829e-06,
+      "loss": 0.7625,
+      "step": 2359
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.09212501063657e-06,
+      "loss": 0.7668,
+      "step": 2360
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.084386869370917e-06,
+      "loss": 0.792,
+      "step": 2361
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.076649918664132e-06,
+      "loss": 0.8008,
+      "step": 2362
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.068914163324759e-06,
+      "loss": 0.8059,
+      "step": 2363
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.061179608160594e-06,
+      "loss": 0.7542,
+      "step": 2364
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.053446257978677e-06,
+      "loss": 0.7883,
+      "step": 2365
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.04571411758531e-06,
+      "loss": 0.7678,
+      "step": 2366
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.037983191786045e-06,
+      "loss": 0.7749,
+      "step": 2367
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.030253485385666e-06,
+      "loss": 0.7749,
+      "step": 2368
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.022525003188208e-06,
+      "loss": 0.7839,
+      "step": 2369
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.014797749996951e-06,
+      "loss": 0.7639,
+      "step": 2370
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.007071730614395e-06,
+      "loss": 0.7578,
+      "step": 2371
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.999346949842287e-06,
+      "loss": 0.6946,
+      "step": 2372
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.991623412481604e-06,
+      "loss": 0.7937,
+      "step": 2373
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.983901123332538e-06,
+      "loss": 0.7422,
+      "step": 2374
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.976180087194515e-06,
+      "loss": 0.7988,
+      "step": 2375
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.968460308866187e-06,
+      "loss": 0.7617,
+      "step": 2376
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.960741793145408e-06,
+      "loss": 0.7512,
+      "step": 2377
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.953024544829261e-06,
+      "loss": 0.7681,
+      "step": 2378
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.945308568714043e-06,
+      "loss": 0.7942,
+      "step": 2379
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.937593869595246e-06,
+      "loss": 0.7817,
+      "step": 2380
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.92988045226758e-06,
+      "loss": 0.7546,
+      "step": 2381
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.922168321524954e-06,
+      "loss": 0.8081,
+      "step": 2382
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.914457482160484e-06,
+      "loss": 0.7244,
+      "step": 2383
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.906747938966473e-06,
+      "loss": 0.7698,
+      "step": 2384
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.899039696734423e-06,
+      "loss": 0.7888,
+      "step": 2385
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.891332760255033e-06,
+      "loss": 0.7661,
+      "step": 2386
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.88362713431818e-06,
+      "loss": 0.8271,
+      "step": 2387
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.875922823712933e-06,
+      "loss": 0.7969,
+      "step": 2388
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.868219833227546e-06,
+      "loss": 0.7705,
+      "step": 2389
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.860518167649445e-06,
+      "loss": 0.7729,
+      "step": 2390
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.852817831765235e-06,
+      "loss": 0.7371,
+      "step": 2391
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.845118830360703e-06,
+      "loss": 0.7627,
+      "step": 2392
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.837421168220794e-06,
+      "loss": 0.7637,
+      "step": 2393
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.829724850129626e-06,
+      "loss": 0.7236,
+      "step": 2394
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.82202988087049e-06,
+      "loss": 0.762,
+      "step": 2395
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.814336265225819e-06,
+      "loss": 0.7681,
+      "step": 2396
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.806644007977222e-06,
+      "loss": 0.7798,
+      "step": 2397
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.798953113905465e-06,
+      "loss": 0.728,
+      "step": 2398
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.791263587790446e-06,
+      "loss": 0.7117,
+      "step": 2399
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.783575434411236e-06,
+      "loss": 0.7275,
+      "step": 2400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.775888658546046e-06,
+      "loss": 0.7444,
+      "step": 2401
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.768203264972222e-06,
+      "loss": 0.8096,
+      "step": 2402
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.760519258466258e-06,
+      "loss": 0.7556,
+      "step": 2403
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.752836643803787e-06,
+      "loss": 0.7693,
+      "step": 2404
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.745155425759581e-06,
+      "loss": 0.77,
+      "step": 2405
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.737475609107528e-06,
+      "loss": 0.7476,
+      "step": 2406
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.729797198620661e-06,
+      "loss": 0.7712,
+      "step": 2407
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.722120199071137e-06,
+      "loss": 0.7036,
+      "step": 2408
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.714444615230223e-06,
+      "loss": 0.769,
+      "step": 2409
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.706770451868322e-06,
+      "loss": 0.7886,
+      "step": 2410
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.69909771375495e-06,
+      "loss": 0.7529,
+      "step": 2411
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.691426405658727e-06,
+      "loss": 0.7607,
+      "step": 2412
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.683756532347397e-06,
+      "loss": 0.7756,
+      "step": 2413
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.676088098587813e-06,
+      "loss": 0.7925,
+      "step": 2414
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.668421109145918e-06,
+      "loss": 0.7405,
+      "step": 2415
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.660755568786771e-06,
+      "loss": 0.7117,
+      "step": 2416
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.653091482274531e-06,
+      "loss": 0.762,
+      "step": 2417
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.645428854372442e-06,
+      "loss": 0.7461,
+      "step": 2418
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.637767689842853e-06,
+      "loss": 0.7449,
+      "step": 2419
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.630107993447197e-06,
+      "loss": 0.7607,
+      "step": 2420
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6224497699459965e-06,
+      "loss": 0.7629,
+      "step": 2421
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.614793024098861e-06,
+      "loss": 0.7944,
+      "step": 2422
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.607137760664478e-06,
+      "loss": 0.7488,
+      "step": 2423
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.599483984400615e-06,
+      "loss": 0.731,
+      "step": 2424
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.591831700064112e-06,
+      "loss": 0.7686,
+      "step": 2425
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.584180912410888e-06,
+      "loss": 0.7773,
+      "step": 2426
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5765316261959335e-06,
+      "loss": 0.8203,
+      "step": 2427
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.568883846173291e-06,
+      "loss": 0.7734,
+      "step": 2428
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5612375770960814e-06,
+      "loss": 0.793,
+      "step": 2429
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.553592823716488e-06,
+      "loss": 0.7043,
+      "step": 2430
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.545949590785737e-06,
+      "loss": 0.825,
+      "step": 2431
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.538307883054122e-06,
+      "loss": 0.7449,
+      "step": 2432
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.530667705270992e-06,
+      "loss": 0.719,
+      "step": 2433
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.523029062184728e-06,
+      "loss": 0.7773,
+      "step": 2434
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5153919585427735e-06,
+      "loss": 0.8203,
+      "step": 2435
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.507756399091614e-06,
+      "loss": 0.8142,
+      "step": 2436
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.500122388576761e-06,
+      "loss": 0.8018,
+      "step": 2437
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.492489931742779e-06,
+      "loss": 0.7395,
+      "step": 2438
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4848590333332635e-06,
+      "loss": 0.7651,
+      "step": 2439
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.477229698090833e-06,
+      "loss": 0.7954,
+      "step": 2440
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.469601930757142e-06,
+      "loss": 0.7812,
+      "step": 2441
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.461975736072875e-06,
+      "loss": 0.771,
+      "step": 2442
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.454351118777725e-06,
+      "loss": 0.7529,
+      "step": 2443
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.446728083610417e-06,
+      "loss": 0.7512,
+      "step": 2444
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.43910663530869e-06,
+      "loss": 0.7441,
+      "step": 2445
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.431486778609291e-06,
+      "loss": 0.771,
+      "step": 2446
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.423868518247983e-06,
+      "loss": 0.8167,
+      "step": 2447
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.41625185895954e-06,
+      "loss": 0.8147,
+      "step": 2448
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.408636805477734e-06,
+      "loss": 0.7549,
+      "step": 2449
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.40102336253534e-06,
+      "loss": 0.783,
+      "step": 2450
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.393411534864139e-06,
+      "loss": 0.8228,
+      "step": 2451
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.385801327194897e-06,
+      "loss": 0.7625,
+      "step": 2452
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.378192744257384e-06,
+      "loss": 0.7856,
+      "step": 2453
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.370585790780352e-06,
+      "loss": 0.7253,
+      "step": 2454
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3629804714915456e-06,
+      "loss": 0.8079,
+      "step": 2455
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.355376791117692e-06,
+      "loss": 0.7729,
+      "step": 2456
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.347774754384498e-06,
+      "loss": 0.7422,
+      "step": 2457
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.340174366016653e-06,
+      "loss": 0.7537,
+      "step": 2458
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.332575630737817e-06,
+      "loss": 0.7324,
+      "step": 2459
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.324978553270624e-06,
+      "loss": 0.7759,
+      "step": 2460
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3173831383366855e-06,
+      "loss": 0.769,
+      "step": 2461
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.309789390656563e-06,
+      "loss": 0.748,
+      "step": 2462
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3021973149498e-06,
+      "loss": 0.7869,
+      "step": 2463
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.294606915934891e-06,
+      "loss": 0.7346,
+      "step": 2464
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.287018198329286e-06,
+      "loss": 0.7576,
+      "step": 2465
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.279431166849394e-06,
+      "loss": 0.781,
+      "step": 2466
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.271845826210586e-06,
+      "loss": 0.7529,
+      "step": 2467
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.26426218112716e-06,
+      "loss": 0.834,
+      "step": 2468
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.256680236312377e-06,
+      "loss": 0.7363,
+      "step": 2469
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2490999964784366e-06,
+      "loss": 0.7644,
+      "step": 2470
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.241521466336485e-06,
+      "loss": 0.7402,
+      "step": 2471
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.233944650596589e-06,
+      "loss": 0.7751,
+      "step": 2472
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.226369553967767e-06,
+      "loss": 0.7761,
+      "step": 2473
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.218796181157965e-06,
+      "loss": 0.812,
+      "step": 2474
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.211224536874046e-06,
+      "loss": 0.78,
+      "step": 2475
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.203654625821815e-06,
+      "loss": 0.7336,
+      "step": 2476
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.196086452705995e-06,
+      "loss": 0.7715,
+      "step": 2477
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.188520022230218e-06,
+      "loss": 0.7651,
+      "step": 2478
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.180955339097049e-06,
+      "loss": 0.7837,
+      "step": 2479
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.173392408007959e-06,
+      "loss": 0.7505,
+      "step": 2480
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.165831233663324e-06,
+      "loss": 0.7227,
+      "step": 2481
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.158271820762443e-06,
+      "loss": 0.7461,
+      "step": 2482
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.150714174003507e-06,
+      "loss": 0.7141,
+      "step": 2483
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.143158298083615e-06,
+      "loss": 0.7788,
+      "step": 2484
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.135604197698766e-06,
+      "loss": 0.7896,
+      "step": 2485
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1280518775438555e-06,
+      "loss": 0.7827,
+      "step": 2486
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.120501342312668e-06,
+      "loss": 0.7439,
+      "step": 2487
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1129525966978864e-06,
+      "loss": 0.7495,
+      "step": 2488
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.105405645391075e-06,
+      "loss": 0.7515,
+      "step": 2489
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.097860493082682e-06,
+      "loss": 0.7524,
+      "step": 2490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.090317144462045e-06,
+      "loss": 0.7266,
+      "step": 2491
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.082775604217371e-06,
+      "loss": 0.7891,
+      "step": 2492
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0752358770357576e-06,
+      "loss": 0.7485,
+      "step": 2493
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.067697967603155e-06,
+      "loss": 0.8105,
+      "step": 2494
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.060161880604401e-06,
+      "loss": 0.751,
+      "step": 2495
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.052627620723196e-06,
+      "loss": 0.7498,
+      "step": 2496
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.045095192642098e-06,
+      "loss": 0.7334,
+      "step": 2497
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.037564601042535e-06,
+      "loss": 0.7397,
+      "step": 2498
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.030035850604792e-06,
+      "loss": 0.7732,
+      "step": 2499
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0225089460080055e-06,
+      "loss": 0.7463,
+      "step": 2500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.014983891930167e-06,
+      "loss": 0.804,
+      "step": 2501
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.007460693048125e-06,
+      "loss": 0.8035,
+      "step": 2502
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.999939354037559e-06,
+      "loss": 0.7097,
+      "step": 2503
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9924198795730084e-06,
+      "loss": 0.7554,
+      "step": 2504
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9849022743278486e-06,
+      "loss": 0.7932,
+      "step": 2505
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.977386542974285e-06,
+      "loss": 0.77,
+      "step": 2506
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9698726901833706e-06,
+      "loss": 0.7427,
+      "step": 2507
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.962360720624989e-06,
+      "loss": 0.7483,
+      "step": 2508
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.95485063896784e-06,
+      "loss": 0.7625,
+      "step": 2509
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.947342449879468e-06,
+      "loss": 0.7812,
+      "step": 2510
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.939836158026236e-06,
+      "loss": 0.7739,
+      "step": 2511
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.932331768073317e-06,
+      "loss": 0.8018,
+      "step": 2512
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.924829284684714e-06,
+      "loss": 0.7834,
+      "step": 2513
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.917328712523242e-06,
+      "loss": 0.783,
+      "step": 2514
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.909830056250527e-06,
+      "loss": 0.7852,
+      "step": 2515
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9023333205270024e-06,
+      "loss": 0.781,
+      "step": 2516
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.894838510011914e-06,
+      "loss": 0.7581,
+      "step": 2517
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.887345629363306e-06,
+      "loss": 0.7593,
+      "step": 2518
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.879854683238023e-06,
+      "loss": 0.7729,
+      "step": 2519
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.872365676291709e-06,
+      "loss": 0.8159,
+      "step": 2520
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.864878613178805e-06,
+      "loss": 0.7651,
+      "step": 2521
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8573934985525405e-06,
+      "loss": 0.7317,
+      "step": 2522
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.849910337064931e-06,
+      "loss": 0.7729,
+      "step": 2523
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.84242913336679e-06,
+      "loss": 0.7175,
+      "step": 2524
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.834949892107698e-06,
+      "loss": 0.7839,
+      "step": 2525
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8274726179360285e-06,
+      "loss": 0.772,
+      "step": 2526
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.819997315498931e-06,
+      "loss": 0.7412,
+      "step": 2527
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.812523989442319e-06,
+      "loss": 0.7317,
+      "step": 2528
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.80505264441089e-06,
+      "loss": 0.8015,
+      "step": 2529
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.797583285048111e-06,
+      "loss": 0.7644,
+      "step": 2530
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.790115915996199e-06,
+      "loss": 0.7722,
+      "step": 2531
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.782650541896148e-06,
+      "loss": 0.7837,
+      "step": 2532
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7751871673877154e-06,
+      "loss": 0.7891,
+      "step": 2533
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.767725797109401e-06,
+      "loss": 0.7815,
+      "step": 2534
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.760266435698468e-06,
+      "loss": 0.7417,
+      "step": 2535
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.752809087790934e-06,
+      "loss": 0.7571,
+      "step": 2536
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.745353758021565e-06,
+      "loss": 0.7827,
+      "step": 2537
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.737900451023858e-06,
+      "loss": 0.7197,
+      "step": 2538
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7304491714300705e-06,
+      "loss": 0.7532,
+      "step": 2539
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.722999923871196e-06,
+      "loss": 0.7319,
+      "step": 2540
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.715552712976956e-06,
+      "loss": 0.7747,
+      "step": 2541
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.708107543375817e-06,
+      "loss": 0.7544,
+      "step": 2542
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7006644196949734e-06,
+      "loss": 0.7986,
+      "step": 2543
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.693223346560344e-06,
+      "loss": 0.7903,
+      "step": 2544
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.685784328596576e-06,
+      "loss": 0.7966,
+      "step": 2545
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.678347370427047e-06,
+      "loss": 0.7693,
+      "step": 2546
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.670912476673837e-06,
+      "loss": 0.7876,
+      "step": 2547
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.663479651957759e-06,
+      "loss": 0.7537,
+      "step": 2548
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.656048900898333e-06,
+      "loss": 0.7612,
+      "step": 2549
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.64862022811379e-06,
+      "loss": 0.6914,
+      "step": 2550
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.641193638221075e-06,
+      "loss": 0.7773,
+      "step": 2551
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.63376913583583e-06,
+      "loss": 0.7776,
+      "step": 2552
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.626346725572406e-06,
+      "loss": 0.761,
+      "step": 2553
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.618926412043847e-06,
+      "loss": 0.7456,
+      "step": 2554
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.611508199861907e-06,
+      "loss": 0.7695,
+      "step": 2555
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.604092093637018e-06,
+      "loss": 0.7598,
+      "step": 2556
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.596678097978313e-06,
+      "loss": 0.7363,
+      "step": 2557
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.58926621749361e-06,
+      "loss": 0.7559,
+      "step": 2558
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.581856456789419e-06,
+      "loss": 0.8037,
+      "step": 2559
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.574448820470918e-06,
+      "loss": 0.771,
+      "step": 2560
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.567043313141976e-06,
+      "loss": 0.7634,
+      "step": 2561
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.559639939405144e-06,
+      "loss": 0.7917,
+      "step": 2562
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.552238703861629e-06,
+      "loss": 0.771,
+      "step": 2563
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.544839611111324e-06,
+      "loss": 0.769,
+      "step": 2564
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.537442665752789e-06,
+      "loss": 0.7725,
+      "step": 2565
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5300478723832385e-06,
+      "loss": 0.7615,
+      "step": 2566
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.522655235598561e-06,
+      "loss": 0.7578,
+      "step": 2567
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.515264759993305e-06,
+      "loss": 0.791,
+      "step": 2568
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.507876450160664e-06,
+      "loss": 0.7275,
+      "step": 2569
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.5004903106924965e-06,
+      "loss": 0.75,
+      "step": 2570
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.493106346179312e-06,
+      "loss": 0.7676,
+      "step": 2571
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.485724561210258e-06,
+      "loss": 0.7354,
+      "step": 2572
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.478344960373137e-06,
+      "loss": 0.7644,
+      "step": 2573
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.470967548254399e-06,
+      "loss": 0.7573,
+      "step": 2574
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.463592329439114e-06,
+      "loss": 0.7314,
+      "step": 2575
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.456219308511005e-06,
+      "loss": 0.728,
+      "step": 2576
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.448848490052432e-06,
+      "loss": 0.7363,
+      "step": 2577
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.441479878644369e-06,
+      "loss": 0.7588,
+      "step": 2578
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4341134788664325e-06,
+      "loss": 0.7444,
+      "step": 2579
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4267492952968635e-06,
+      "loss": 0.7402,
+      "step": 2580
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4193873325125186e-06,
+      "loss": 0.7202,
+      "step": 2581
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.412027595088881e-06,
+      "loss": 0.7617,
+      "step": 2582
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4046700876000435e-06,
+      "loss": 0.7344,
+      "step": 2583
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.397314814618722e-06,
+      "loss": 0.7986,
+      "step": 2584
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.389961780716238e-06,
+      "loss": 0.7849,
+      "step": 2585
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.382610990462519e-06,
+      "loss": 0.7654,
+      "step": 2586
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.375262448426108e-06,
+      "loss": 0.7122,
+      "step": 2587
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.367916159174137e-06,
+      "loss": 0.7961,
+      "step": 2588
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.360572127272349e-06,
+      "loss": 0.7539,
+      "step": 2589
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.353230357285083e-06,
+      "loss": 0.7385,
+      "step": 2590
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3458908537752616e-06,
+      "loss": 0.7849,
+      "step": 2591
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.338553621304413e-06,
+      "loss": 0.7615,
+      "step": 2592
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.331218664432648e-06,
+      "loss": 0.8013,
+      "step": 2593
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.323885987718657e-06,
+      "loss": 0.7605,
+      "step": 2594
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.316555595719722e-06,
+      "loss": 0.7568,
+      "step": 2595
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.309227492991708e-06,
+      "loss": 0.7844,
+      "step": 2596
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3019016840890405e-06,
+      "loss": 0.7668,
+      "step": 2597
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.294578173564734e-06,
+      "loss": 0.7312,
+      "step": 2598
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.287256965970377e-06,
+      "loss": 0.8193,
+      "step": 2599
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.279938065856111e-06,
+      "loss": 0.7139,
+      "step": 2600
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.272621477770655e-06,
+      "loss": 0.7629,
+      "step": 2601
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.265307206261293e-06,
+      "loss": 0.7217,
+      "step": 2602
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.257995255873865e-06,
+      "loss": 0.79,
+      "step": 2603
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.250685631152762e-06,
+      "loss": 0.769,
+      "step": 2604
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.243378336640939e-06,
+      "loss": 0.751,
+      "step": 2605
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.236073376879907e-06,
+      "loss": 0.7729,
+      "step": 2606
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.228770756409706e-06,
+      "loss": 0.7703,
+      "step": 2607
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.221470479768943e-06,
+      "loss": 0.7917,
+      "step": 2608
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.214172551494764e-06,
+      "loss": 0.7959,
+      "step": 2609
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.206876976122839e-06,
+      "loss": 0.7607,
+      "step": 2610
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1995837581874e-06,
+      "loss": 0.8074,
+      "step": 2611
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1922929022211974e-06,
+      "loss": 0.7698,
+      "step": 2612
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.185004412755517e-06,
+      "loss": 0.7627,
+      "step": 2613
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.177718294320178e-06,
+      "loss": 0.7637,
+      "step": 2614
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1704345514435224e-06,
+      "loss": 0.8293,
+      "step": 2615
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1631531886524175e-06,
+      "loss": 0.751,
+      "step": 2616
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.155874210472248e-06,
+      "loss": 0.7537,
+      "step": 2617
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.148597621426922e-06,
+      "loss": 0.7483,
+      "step": 2618
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.141323426038857e-06,
+      "loss": 0.7834,
+      "step": 2619
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.134051628828985e-06,
+      "loss": 0.7183,
+      "step": 2620
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.126782234316752e-06,
+      "loss": 0.7573,
+      "step": 2621
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1195152470201056e-06,
+      "loss": 0.7607,
+      "step": 2622
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1122506714554956e-06,
+      "loss": 0.7454,
+      "step": 2623
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.104988512137877e-06,
+      "loss": 0.7563,
+      "step": 2624
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.097728773580709e-06,
+      "loss": 0.7773,
+      "step": 2625
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.090471460295928e-06,
+      "loss": 0.7588,
+      "step": 2626
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0832165767939795e-06,
+      "loss": 0.6846,
+      "step": 2627
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.075964127583799e-06,
+      "loss": 0.7615,
+      "step": 2628
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.068714117172793e-06,
+      "loss": 0.791,
+      "step": 2629
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.06146655006687e-06,
+      "loss": 0.7576,
+      "step": 2630
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.054221430770416e-06,
+      "loss": 0.741,
+      "step": 2631
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.046978763786285e-06,
+      "loss": 0.8105,
+      "step": 2632
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.039738553615816e-06,
+      "loss": 0.7866,
+      "step": 2633
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.032500804758827e-06,
+      "loss": 0.8093,
+      "step": 2634
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.02526552171359e-06,
+      "loss": 0.7871,
+      "step": 2635
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.018032708976857e-06,
+      "loss": 0.7439,
+      "step": 2636
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.010802371043845e-06,
+      "loss": 0.7751,
+      "step": 2637
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.003574512408224e-06,
+      "loss": 0.7285,
+      "step": 2638
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.996349137562128e-06,
+      "loss": 0.814,
+      "step": 2639
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.989126250996153e-06,
+      "loss": 0.7332,
+      "step": 2640
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.981905857199335e-06,
+      "loss": 0.7676,
+      "step": 2641
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.974687960659175e-06,
+      "loss": 0.7744,
+      "step": 2642
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.967472565861616e-06,
+      "loss": 0.7002,
+      "step": 2643
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.96025967729104e-06,
+      "loss": 0.7322,
+      "step": 2644
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.953049299430285e-06,
+      "loss": 0.7183,
+      "step": 2645
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.945841436760612e-06,
+      "loss": 0.7554,
+      "step": 2646
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.938636093761736e-06,
+      "loss": 0.8181,
+      "step": 2647
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.931433274911793e-06,
+      "loss": 0.7261,
+      "step": 2648
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.924232984687355e-06,
+      "loss": 0.7563,
+      "step": 2649
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.917035227563425e-06,
+      "loss": 0.804,
+      "step": 2650
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.909840008013428e-06,
+      "loss": 0.7693,
+      "step": 2651
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.90264733050921e-06,
+      "loss": 0.7771,
+      "step": 2652
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.895457199521046e-06,
+      "loss": 0.7241,
+      "step": 2653
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.888269619517615e-06,
+      "loss": 0.7275,
+      "step": 2654
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8810845949660226e-06,
+      "loss": 0.7817,
+      "step": 2655
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8739021303317825e-06,
+      "loss": 0.7366,
+      "step": 2656
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.866722230078812e-06,
+      "loss": 0.7693,
+      "step": 2657
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.85954489866944e-06,
+      "loss": 0.75,
+      "step": 2658
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.852370140564403e-06,
+      "loss": 0.7517,
+      "step": 2659
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.845197960222827e-06,
+      "loss": 0.718,
+      "step": 2660
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.838028362102246e-06,
+      "loss": 0.7332,
+      "step": 2661
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.830861350658581e-06,
+      "loss": 0.7678,
+      "step": 2662
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.823696930346151e-06,
+      "loss": 0.8191,
+      "step": 2663
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.816535105617669e-06,
+      "loss": 0.729,
+      "step": 2664
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.809375880924219e-06,
+      "loss": 0.7793,
+      "step": 2665
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.80221926071529e-06,
+      "loss": 0.7563,
+      "step": 2666
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.79506524943873e-06,
+      "loss": 0.739,
+      "step": 2667
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.787913851540783e-06,
+      "loss": 0.7793,
+      "step": 2668
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7807650714660644e-06,
+      "loss": 0.7375,
+      "step": 2669
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.773618913657555e-06,
+      "loss": 0.7458,
+      "step": 2670
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7664753825566155e-06,
+      "loss": 0.8054,
+      "step": 2671
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.759334482602973e-06,
+      "loss": 0.741,
+      "step": 2672
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.752196218234709e-06,
+      "loss": 0.7551,
+      "step": 2673
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.74506059388828e-06,
+      "loss": 0.7573,
+      "step": 2674
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7379276139984975e-06,
+      "loss": 0.7515,
+      "step": 2675
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7307972829985234e-06,
+      "loss": 0.7471,
+      "step": 2676
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.723669605319882e-06,
+      "loss": 0.7686,
+      "step": 2677
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.716544585392445e-06,
+      "loss": 0.7573,
+      "step": 2678
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.709422227644428e-06,
+      "loss": 0.729,
+      "step": 2679
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7023025365024e-06,
+      "loss": 0.7139,
+      "step": 2680
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.69518551639127e-06,
+      "loss": 0.7554,
+      "step": 2681
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.68807117173428e-06,
+      "loss": 0.7688,
+      "step": 2682
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.680959506953017e-06,
+      "loss": 0.748,
+      "step": 2683
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.673850526467406e-06,
+      "loss": 0.7559,
+      "step": 2684
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6667442346956915e-06,
+      "loss": 0.7913,
+      "step": 2685
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.659640636054453e-06,
+      "loss": 0.7244,
+      "step": 2686
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.652539734958607e-06,
+      "loss": 0.7646,
+      "step": 2687
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6454415358213685e-06,
+      "loss": 0.7561,
+      "step": 2688
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.638346043054297e-06,
+      "loss": 0.7808,
+      "step": 2689
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.631253261067257e-06,
+      "loss": 0.7488,
+      "step": 2690
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.624163194268441e-06,
+      "loss": 0.8118,
+      "step": 2691
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.617075847064334e-06,
+      "loss": 0.7446,
+      "step": 2692
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.609991223859746e-06,
+      "loss": 0.7495,
+      "step": 2693
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.602909329057794e-06,
+      "loss": 0.7473,
+      "step": 2694
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.595830167059888e-06,
+      "loss": 0.7947,
+      "step": 2695
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.588753742265751e-06,
+      "loss": 0.7676,
+      "step": 2696
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.581680059073405e-06,
+      "loss": 0.7512,
+      "step": 2697
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.574609121879153e-06,
+      "loss": 0.7773,
+      "step": 2698
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.567540935077609e-06,
+      "loss": 0.8008,
+      "step": 2699
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5604755030616735e-06,
+      "loss": 0.7625,
+      "step": 2700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.553412830222523e-06,
+      "loss": 0.8096,
+      "step": 2701
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.546352920949636e-06,
+      "loss": 0.7358,
+      "step": 2702
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.539295779630765e-06,
+      "loss": 0.8032,
+      "step": 2703
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.532241410651937e-06,
+      "loss": 0.751,
+      "step": 2704
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.525189818397469e-06,
+      "loss": 0.7231,
+      "step": 2705
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.518141007249944e-06,
+      "loss": 0.7363,
+      "step": 2706
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.511094981590212e-06,
+      "loss": 0.8127,
+      "step": 2707
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5040517457974e-06,
+      "loss": 0.7327,
+      "step": 2708
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.497011304248907e-06,
+      "loss": 0.7437,
+      "step": 2709
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.489973661320373e-06,
+      "loss": 0.7515,
+      "step": 2710
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.482938821385722e-06,
+      "loss": 0.7302,
+      "step": 2711
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4759067888171225e-06,
+      "loss": 0.7441,
+      "step": 2712
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.468877567985008e-06,
+      "loss": 0.6897,
+      "step": 2713
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.461851163258052e-06,
+      "loss": 0.772,
+      "step": 2714
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.454827579003185e-06,
+      "loss": 0.7405,
+      "step": 2715
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.44780681958559e-06,
+      "loss": 0.7883,
+      "step": 2716
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.440788889368682e-06,
+      "loss": 0.8032,
+      "step": 2717
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.433773792714127e-06,
+      "loss": 0.8225,
+      "step": 2718
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.426761533981829e-06,
+      "loss": 0.729,
+      "step": 2719
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.419752117529922e-06,
+      "loss": 0.7578,
+      "step": 2720
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.412745547714777e-06,
+      "loss": 0.7954,
+      "step": 2721
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.405741828891006e-06,
+      "loss": 0.7493,
+      "step": 2722
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.398740965411428e-06,
+      "loss": 0.7898,
+      "step": 2723
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.391742961627109e-06,
+      "loss": 0.7556,
+      "step": 2724
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.384747821887317e-06,
+      "loss": 0.771,
+      "step": 2725
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.377755550539559e-06,
+      "loss": 0.7207,
+      "step": 2726
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.370766151929549e-06,
+      "loss": 0.7524,
+      "step": 2727
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.363779630401216e-06,
+      "loss": 0.7839,
+      "step": 2728
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.356795990296708e-06,
+      "loss": 0.7483,
+      "step": 2729
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3498152359563674e-06,
+      "loss": 0.7715,
+      "step": 2730
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3428373717187575e-06,
+      "loss": 0.7612,
+      "step": 2731
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.335862401920644e-06,
+      "loss": 0.7942,
+      "step": 2732
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.328890330896983e-06,
+      "loss": 0.7363,
+      "step": 2733
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.32192116298094e-06,
+      "loss": 0.7578,
+      "step": 2734
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.314954902503874e-06,
+      "loss": 0.7532,
+      "step": 2735
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.307991553795328e-06,
+      "loss": 0.7573,
+      "step": 2736
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.301031121183049e-06,
+      "loss": 0.7363,
+      "step": 2737
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.294073608992966e-06,
+      "loss": 0.7441,
+      "step": 2738
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2871190215491865e-06,
+      "loss": 0.7454,
+      "step": 2739
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.280167363174008e-06,
+      "loss": 0.7605,
+      "step": 2740
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2732186381879115e-06,
+      "loss": 0.7249,
+      "step": 2741
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2662728509095375e-06,
+      "loss": 0.7681,
+      "step": 2742
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2593300056557205e-06,
+      "loss": 0.7903,
+      "step": 2743
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.252390106741458e-06,
+      "loss": 0.718,
+      "step": 2744
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.245453158479912e-06,
+      "loss": 0.7454,
+      "step": 2745
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.238519165182415e-06,
+      "loss": 0.7651,
+      "step": 2746
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.231588131158472e-06,
+      "loss": 0.7449,
+      "step": 2747
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.224660060715727e-06,
+      "loss": 0.7605,
+      "step": 2748
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.217734958160002e-06,
+      "loss": 0.8643,
+      "step": 2749
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.210812827795271e-06,
+      "loss": 0.7734,
+      "step": 2750
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.203893673923649e-06,
+      "loss": 0.7405,
+      "step": 2751
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.196977500845412e-06,
+      "loss": 0.7786,
+      "step": 2752
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.190064312858989e-06,
+      "loss": 0.802,
+      "step": 2753
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.183154114260932e-06,
+      "loss": 0.7629,
+      "step": 2754
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.176246909345956e-06,
+      "loss": 0.7935,
+      "step": 2755
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.169342702406907e-06,
+      "loss": 0.7534,
+      "step": 2756
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.162441497734772e-06,
+      "loss": 0.7959,
+      "step": 2757
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.155543299618664e-06,
+      "loss": 0.7544,
+      "step": 2758
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.148648112345831e-06,
+      "loss": 0.7622,
+      "step": 2759
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.14175594020166e-06,
+      "loss": 0.793,
+      "step": 2760
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1348667874696455e-06,
+      "loss": 0.7249,
+      "step": 2761
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.127980658431417e-06,
+      "loss": 0.7544,
+      "step": 2762
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.12109755736673e-06,
+      "loss": 0.7615,
+      "step": 2763
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.114217488553441e-06,
+      "loss": 0.7349,
+      "step": 2764
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.107340456267536e-06,
+      "loss": 0.7161,
+      "step": 2765
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.100466464783113e-06,
+      "loss": 0.7205,
+      "step": 2766
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.093595518372372e-06,
+      "loss": 0.7539,
+      "step": 2767
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.086727621305625e-06,
+      "loss": 0.7732,
+      "step": 2768
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0798627778512955e-06,
+      "loss": 0.7432,
+      "step": 2769
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.073000992275894e-06,
+      "loss": 0.7883,
+      "step": 2770
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.066142268844043e-06,
+      "loss": 0.7314,
+      "step": 2771
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.059286611818463e-06,
+      "loss": 0.7544,
+      "step": 2772
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.052434025459957e-06,
+      "loss": 0.7021,
+      "step": 2773
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.045584514027429e-06,
+      "loss": 0.7524,
+      "step": 2774
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.038738081777871e-06,
+      "loss": 0.7312,
+      "step": 2775
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0318947329663545e-06,
+      "loss": 0.7688,
+      "step": 2776
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0250544718460445e-06,
+      "loss": 0.7322,
+      "step": 2777
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.01821730266818e-06,
+      "loss": 0.7791,
+      "step": 2778
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.011383229682085e-06,
+      "loss": 0.7896,
+      "step": 2779
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0045522571351494e-06,
+      "loss": 0.7549,
+      "step": 2780
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.997724389272841e-06,
+      "loss": 0.7544,
+      "step": 2781
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.990899630338707e-06,
+      "loss": 0.7456,
+      "step": 2782
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9840779845743434e-06,
+      "loss": 0.7769,
+      "step": 2783
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.977259456219427e-06,
+      "loss": 0.7683,
+      "step": 2784
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.970444049511697e-06,
+      "loss": 0.7349,
+      "step": 2785
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.963631768686937e-06,
+      "loss": 0.7366,
+      "step": 2786
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.956822617979008e-06,
+      "loss": 0.7415,
+      "step": 2787
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.950016601619808e-06,
+      "loss": 0.7715,
+      "step": 2788
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9432137238393e-06,
+      "loss": 0.7388,
+      "step": 2789
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9364139888654925e-06,
+      "loss": 0.7302,
+      "step": 2790
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.929617400924435e-06,
+      "loss": 0.7437,
+      "step": 2791
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.922823964240228e-06,
+      "loss": 0.7485,
+      "step": 2792
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.916033683035015e-06,
+      "loss": 0.7754,
+      "step": 2793
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.909246561528967e-06,
+      "loss": 0.7944,
+      "step": 2794
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.902462603940307e-06,
+      "loss": 0.7524,
+      "step": 2795
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.895681814485277e-06,
+      "loss": 0.7483,
+      "step": 2796
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.888904197378157e-06,
+      "loss": 0.7344,
+      "step": 2797
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.882129756831262e-06,
+      "loss": 0.7183,
+      "step": 2798
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.875358497054915e-06,
+      "loss": 0.7893,
+      "step": 2799
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.86859042225748e-06,
+      "loss": 0.7717,
+      "step": 2800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.861825536645335e-06,
+      "loss": 0.7129,
+      "step": 2801
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8550638444228685e-06,
+      "loss": 0.7537,
+      "step": 2802
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.848305349792495e-06,
+      "loss": 0.7424,
+      "step": 2803
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.841550056954641e-06,
+      "loss": 0.7903,
+      "step": 2804
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.834797970107734e-06,
+      "loss": 0.6865,
+      "step": 2805
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.828049093448216e-06,
+      "loss": 0.7327,
+      "step": 2806
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8213034311705375e-06,
+      "loss": 0.7546,
+      "step": 2807
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.814560987467139e-06,
+      "loss": 0.7537,
+      "step": 2808
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.807821766528471e-06,
+      "loss": 0.7781,
+      "step": 2809
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.801085772542983e-06,
+      "loss": 0.7688,
+      "step": 2810
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.794353009697105e-06,
+      "loss": 0.7314,
+      "step": 2811
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.787623482175272e-06,
+      "loss": 0.7229,
+      "step": 2812
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.780897194159908e-06,
+      "loss": 0.7581,
+      "step": 2813
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.774174149831413e-06,
+      "loss": 0.7864,
+      "step": 2814
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.767454353368179e-06,
+      "loss": 0.7815,
+      "step": 2815
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.760737808946585e-06,
+      "loss": 0.7476,
+      "step": 2816
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.754024520740972e-06,
+      "loss": 0.7253,
+      "step": 2817
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.747314492923672e-06,
+      "loss": 0.7654,
+      "step": 2818
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.74060772966499e-06,
+      "loss": 0.8098,
+      "step": 2819
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.73390423513319e-06,
+      "loss": 0.7451,
+      "step": 2820
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.727204013494515e-06,
+      "loss": 0.7466,
+      "step": 2821
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.720507068913174e-06,
+      "loss": 0.7327,
+      "step": 2822
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.713813405551339e-06,
+      "loss": 0.7493,
+      "step": 2823
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.707123027569132e-06,
+      "loss": 0.7834,
+      "step": 2824
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.700435939124646e-06,
+      "loss": 0.7583,
+      "step": 2825
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.693752144373929e-06,
+      "loss": 0.7808,
+      "step": 2826
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.687071647470971e-06,
+      "loss": 0.7407,
+      "step": 2827
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.680394452567721e-06,
+      "loss": 0.7847,
+      "step": 2828
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.67372056381408e-06,
+      "loss": 0.7295,
+      "step": 2829
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.667049985357881e-06,
+      "loss": 0.7542,
+      "step": 2830
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.660382721344909e-06,
+      "loss": 0.7783,
+      "step": 2831
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.653718775918892e-06,
+      "loss": 0.7241,
+      "step": 2832
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.647058153221483e-06,
+      "loss": 0.7798,
+      "step": 2833
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.640400857392281e-06,
+      "loss": 0.7749,
+      "step": 2834
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.633746892568817e-06,
+      "loss": 0.7468,
+      "step": 2835
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.627096262886542e-06,
+      "loss": 0.7534,
+      "step": 2836
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.620448972478845e-06,
+      "loss": 0.7603,
+      "step": 2837
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6138050254770385e-06,
+      "loss": 0.7271,
+      "step": 2838
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.607164426010346e-06,
+      "loss": 0.7571,
+      "step": 2839
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6005271782059226e-06,
+      "loss": 0.738,
+      "step": 2840
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.59389328618884e-06,
+      "loss": 0.7385,
+      "step": 2841
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.587262754082073e-06,
+      "loss": 0.7419,
+      "step": 2842
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.580635586006519e-06,
+      "loss": 0.75,
+      "step": 2843
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.574011786080983e-06,
+      "loss": 0.7561,
+      "step": 2844
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.567391358422176e-06,
+      "loss": 0.8047,
+      "step": 2845
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5607743071447085e-06,
+      "loss": 0.749,
+      "step": 2846
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5541606363610955e-06,
+      "loss": 0.7554,
+      "step": 2847
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.54755035018176e-06,
+      "loss": 0.7253,
+      "step": 2848
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.540943452715002e-06,
+      "loss": 0.7744,
+      "step": 2849
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.534339948067032e-06,
+      "loss": 0.781,
+      "step": 2850
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.52773984034195e-06,
+      "loss": 0.8018,
+      "step": 2851
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5211431336417345e-06,
+      "loss": 0.7583,
+      "step": 2852
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.514549832066263e-06,
+      "loss": 0.7422,
+      "step": 2853
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.507959939713283e-06,
+      "loss": 0.7749,
+      "step": 2854
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5013734606784375e-06,
+      "loss": 0.8105,
+      "step": 2855
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.494790399055242e-06,
+      "loss": 0.7856,
+      "step": 2856
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.488210758935083e-06,
+      "loss": 0.6672,
+      "step": 2857
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.481634544407229e-06,
+      "loss": 0.7605,
+      "step": 2858
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.475061759558812e-06,
+      "loss": 0.7898,
+      "step": 2859
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.468492408474837e-06,
+      "loss": 0.7388,
+      "step": 2860
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.461926495238178e-06,
+      "loss": 0.7212,
+      "step": 2861
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.455364023929563e-06,
+      "loss": 0.7605,
+      "step": 2862
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.44880499862759e-06,
+      "loss": 0.7527,
+      "step": 2863
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.442249423408707e-06,
+      "loss": 0.7903,
+      "step": 2864
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.435697302347224e-06,
+      "loss": 0.8171,
+      "step": 2865
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.429148639515304e-06,
+      "loss": 0.7002,
+      "step": 2866
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.422603438982961e-06,
+      "loss": 0.7261,
+      "step": 2867
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.41606170481805e-06,
+      "loss": 0.7705,
+      "step": 2868
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4095234410862785e-06,
+      "loss": 0.7781,
+      "step": 2869
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.402988651851201e-06,
+      "loss": 0.7646,
+      "step": 2870
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3964573411741994e-06,
+      "loss": 0.7756,
+      "step": 2871
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3899295131145035e-06,
+      "loss": 0.7749,
+      "step": 2872
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.383405171729182e-06,
+      "loss": 0.7942,
+      "step": 2873
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.376884321073124e-06,
+      "loss": 0.7346,
+      "step": 2874
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.370366965199059e-06,
+      "loss": 0.7864,
+      "step": 2875
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.363853108157545e-06,
+      "loss": 0.7415,
+      "step": 2876
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.357342753996956e-06,
+      "loss": 0.7234,
+      "step": 2877
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.350835906763499e-06,
+      "loss": 0.7642,
+      "step": 2878
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3443325705012e-06,
+      "loss": 0.7407,
+      "step": 2879
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.337832749251893e-06,
+      "loss": 0.762,
+      "step": 2880
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.33133644705524e-06,
+      "loss": 0.7185,
+      "step": 2881
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.324843667948714e-06,
+      "loss": 0.7449,
+      "step": 2882
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.318354415967588e-06,
+      "loss": 0.7668,
+      "step": 2883
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3118686951449525e-06,
+      "loss": 0.7202,
+      "step": 2884
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.305386509511707e-06,
+      "loss": 0.7776,
+      "step": 2885
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.29890786309654e-06,
+      "loss": 0.7861,
+      "step": 2886
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.29243275992595e-06,
+      "loss": 0.7087,
+      "step": 2887
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.285961204024232e-06,
+      "loss": 0.7717,
+      "step": 2888
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.279493199413482e-06,
+      "loss": 0.7942,
+      "step": 2889
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2730287501135735e-06,
+      "loss": 0.7839,
+      "step": 2890
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.266567860142184e-06,
+      "loss": 0.7788,
+      "step": 2891
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.260110533514777e-06,
+      "loss": 0.7908,
+      "step": 2892
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2536567742445915e-06,
+      "loss": 0.7527,
+      "step": 2893
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.247206586342663e-06,
+      "loss": 0.7883,
+      "step": 2894
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.240759973817803e-06,
+      "loss": 0.7224,
+      "step": 2895
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.234316940676592e-06,
+      "loss": 0.7537,
+      "step": 2896
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.227877490923396e-06,
+      "loss": 0.7446,
+      "step": 2897
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.221441628560355e-06,
+      "loss": 0.7737,
+      "step": 2898
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2150093575873685e-06,
+      "loss": 0.7666,
+      "step": 2899
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.208580682002115e-06,
+      "loss": 0.7444,
+      "step": 2900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.202155605800035e-06,
+      "loss": 0.7815,
+      "step": 2901
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.195734132974328e-06,
+      "loss": 0.7148,
+      "step": 2902
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.189316267515957e-06,
+      "loss": 0.7429,
+      "step": 2903
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.182902013413649e-06,
+      "loss": 0.7134,
+      "step": 2904
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.176491374653874e-06,
+      "loss": 0.7437,
+      "step": 2905
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.170084355220862e-06,
+      "loss": 0.75,
+      "step": 2906
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.163680959096601e-06,
+      "loss": 0.7651,
+      "step": 2907
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1572811902608105e-06,
+      "loss": 0.7729,
+      "step": 2908
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.150885052690968e-06,
+      "loss": 0.7629,
+      "step": 2909
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1444925503622906e-06,
+      "loss": 0.7412,
+      "step": 2910
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.13810368724774e-06,
+      "loss": 0.7517,
+      "step": 2911
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.131718467318006e-06,
+      "loss": 0.7507,
+      "step": 2912
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.125336894541523e-06,
+      "loss": 0.7629,
+      "step": 2913
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1189589728844605e-06,
+      "loss": 0.7727,
+      "step": 2914
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.112584706310707e-06,
+      "loss": 0.7549,
+      "step": 2915
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1062140987818934e-06,
+      "loss": 0.7458,
+      "step": 2916
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.099847154257364e-06,
+      "loss": 0.7366,
+      "step": 2917
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.093483876694195e-06,
+      "loss": 0.7297,
+      "step": 2918
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.087124270047186e-06,
+      "loss": 0.7446,
+      "step": 2919
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.080768338268842e-06,
+      "loss": 0.7056,
+      "step": 2920
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.074416085309399e-06,
+      "loss": 0.7507,
+      "step": 2921
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.068067515116794e-06,
+      "loss": 0.6931,
+      "step": 2922
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.061722631636683e-06,
+      "loss": 0.7686,
+      "step": 2923
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.055381438812433e-06,
+      "loss": 0.748,
+      "step": 2924
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.049043940585107e-06,
+      "loss": 0.7642,
+      "step": 2925
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.042710140893482e-06,
+      "loss": 0.7581,
+      "step": 2926
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.036380043674028e-06,
+      "loss": 0.7737,
+      "step": 2927
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.030053652860919e-06,
+      "loss": 0.7266,
+      "step": 2928
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.02373097238603e-06,
+      "loss": 0.7375,
+      "step": 2929
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.017412006178917e-06,
+      "loss": 0.7686,
+      "step": 2930
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.011096758166837e-06,
+      "loss": 0.7666,
+      "step": 2931
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.004785232274736e-06,
+      "loss": 0.7507,
+      "step": 2932
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.998477432425247e-06,
+      "loss": 0.75,
+      "step": 2933
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.99217336253868e-06,
+      "loss": 0.6877,
+      "step": 2934
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.985873026533034e-06,
+      "loss": 0.7554,
+      "step": 2935
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.979576428323988e-06,
+      "loss": 0.7466,
+      "step": 2936
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9732835718248885e-06,
+      "loss": 0.7644,
+      "step": 2937
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9669944609467695e-06,
+      "loss": 0.7637,
+      "step": 2938
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.960709099598331e-06,
+      "loss": 0.74,
+      "step": 2939
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.954427491685937e-06,
+      "loss": 0.7844,
+      "step": 2940
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.948149641113628e-06,
+      "loss": 0.7217,
+      "step": 2941
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9418755517831085e-06,
+      "loss": 0.8184,
+      "step": 2942
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.935605227593735e-06,
+      "loss": 0.7278,
+      "step": 2943
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.929338672442536e-06,
+      "loss": 0.6917,
+      "step": 2944
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.923075890224196e-06,
+      "loss": 0.7485,
+      "step": 2945
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.916816884831044e-06,
+      "loss": 0.7358,
+      "step": 2946
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.910561660153074e-06,
+      "loss": 0.8025,
+      "step": 2947
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.904310220077927e-06,
+      "loss": 0.7561,
+      "step": 2948
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.898062568490885e-06,
+      "loss": 0.7744,
+      "step": 2949
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.891818709274883e-06,
+      "loss": 0.7439,
+      "step": 2950
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8855786463105015e-06,
+      "loss": 0.7644,
+      "step": 2951
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.879342383475949e-06,
+      "loss": 0.7502,
+      "step": 2952
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.873109924647085e-06,
+      "loss": 0.7314,
+      "step": 2953
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.866881273697398e-06,
+      "loss": 0.7261,
+      "step": 2954
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8606564344980176e-06,
+      "loss": 0.7314,
+      "step": 2955
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.854435410917689e-06,
+      "loss": 0.7148,
+      "step": 2956
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.848218206822803e-06,
+      "loss": 0.7771,
+      "step": 2957
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.842004826077369e-06,
+      "loss": 0.7786,
+      "step": 2958
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.835795272543014e-06,
+      "loss": 0.7698,
+      "step": 2959
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.829589550078997e-06,
+      "loss": 0.7122,
+      "step": 2960
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.823387662542196e-06,
+      "loss": 0.6931,
+      "step": 2961
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.817189613787093e-06,
+      "loss": 0.7446,
+      "step": 2962
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8109954076657963e-06,
+      "loss": 0.7742,
+      "step": 2963
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.804805048028025e-06,
+      "loss": 0.7095,
+      "step": 2964
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7986185387210995e-06,
+      "loss": 0.7732,
+      "step": 2965
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7924358835899556e-06,
+      "loss": 0.8184,
+      "step": 2966
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7862570864771333e-06,
+      "loss": 0.7637,
+      "step": 2967
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7800821512227648e-06,
+      "loss": 0.751,
+      "step": 2968
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.773911081664594e-06,
+      "loss": 0.7791,
+      "step": 2969
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.767743881637961e-06,
+      "loss": 0.7568,
+      "step": 2970
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.76158055497579e-06,
+      "loss": 0.7378,
+      "step": 2971
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7554211055086097e-06,
+      "loss": 0.7673,
+      "step": 2972
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7492655370645393e-06,
+      "loss": 0.7271,
+      "step": 2973
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7431138534692736e-06,
+      "loss": 0.7366,
+      "step": 2974
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7369660585461045e-06,
+      "loss": 0.717,
+      "step": 2975
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.730822156115904e-06,
+      "loss": 0.7573,
+      "step": 2976
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7246821499971285e-06,
+      "loss": 0.7874,
+      "step": 2977
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.718546044005802e-06,
+      "loss": 0.7737,
+      "step": 2978
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7124138419555357e-06,
+      "loss": 0.7717,
+      "step": 2979
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7062855476575065e-06,
+      "loss": 0.7336,
+      "step": 2980
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7001611649204685e-06,
+      "loss": 0.7207,
+      "step": 2981
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.694040697550745e-06,
+      "loss": 0.7415,
+      "step": 2982
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.687924149352219e-06,
+      "loss": 0.7188,
+      "step": 2983
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6818115241263463e-06,
+      "loss": 0.7671,
+      "step": 2984
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.675702825672135e-06,
+      "loss": 0.7754,
+      "step": 2985
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6695980577861614e-06,
+      "loss": 0.7463,
+      "step": 2986
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.66349722426256e-06,
+      "loss": 0.7676,
+      "step": 2987
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.657400328893007e-06,
+      "loss": 0.7864,
+      "step": 2988
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.651307375466745e-06,
+      "loss": 0.7544,
+      "step": 2989
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6452183677705633e-06,
+      "loss": 0.76,
+      "step": 2990
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6391333095887917e-06,
+      "loss": 0.7751,
+      "step": 2991
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6330522047033167e-06,
+      "loss": 0.7617,
+      "step": 2992
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6269750568935557e-06,
+      "loss": 0.7346,
+      "step": 2993
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6209018699364752e-06,
+      "loss": 0.7444,
+      "step": 2994
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.614832647606582e-06,
+      "loss": 0.7463,
+      "step": 2995
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6087673936759084e-06,
+      "loss": 0.7583,
+      "step": 2996
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6027061119140283e-06,
+      "loss": 0.7563,
+      "step": 2997
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5966488060880455e-06,
+      "loss": 0.7625,
+      "step": 2998
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5905954799625977e-06,
+      "loss": 0.7468,
+      "step": 2999
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5845461372998345e-06,
+      "loss": 0.7483,
+      "step": 3000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.578500781859444e-06,
+      "loss": 0.7544,
+      "step": 3001
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.572459417398635e-06,
+      "loss": 0.7717,
+      "step": 3002
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5664220476721243e-06,
+      "loss": 0.7332,
+      "step": 3003
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.560388676432158e-06,
+      "loss": 0.7539,
+      "step": 3004
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5543593074284975e-06,
+      "loss": 0.72,
+      "step": 3005
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.548333944408404e-06,
+      "loss": 0.7371,
+      "step": 3006
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.542312591116661e-06,
+      "loss": 0.7334,
+      "step": 3007
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5362952512955596e-06,
+      "loss": 0.7915,
+      "step": 3008
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5302819286848875e-06,
+      "loss": 0.7498,
+      "step": 3009
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.524272627021943e-06,
+      "loss": 0.7717,
+      "step": 3010
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.518267350041529e-06,
+      "loss": 0.7415,
+      "step": 3011
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5122661014759342e-06,
+      "loss": 0.7622,
+      "step": 3012
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.506268885054954e-06,
+      "loss": 0.7461,
+      "step": 3013
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.50027570450588e-06,
+      "loss": 0.7559,
+      "step": 3014
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4942865635534816e-06,
+      "loss": 0.7444,
+      "step": 3015
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.488301465920031e-06,
+      "loss": 0.7209,
+      "step": 3016
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4823204153252875e-06,
+      "loss": 0.7229,
+      "step": 3017
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.476343415486483e-06,
+      "loss": 0.7178,
+      "step": 3018
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4703704701183416e-06,
+      "loss": 0.7366,
+      "step": 3019
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4644015829330668e-06,
+      "loss": 0.7288,
+      "step": 3020
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.458436757640341e-06,
+      "loss": 0.7412,
+      "step": 3021
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4524759979473145e-06,
+      "loss": 0.7439,
+      "step": 3022
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.446519307558618e-06,
+      "loss": 0.7681,
+      "step": 3023
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.440566690176356e-06,
+      "loss": 0.7686,
+      "step": 3024
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4346181495000876e-06,
+      "loss": 0.7708,
+      "step": 3025
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.428673689226851e-06,
+      "loss": 0.7244,
+      "step": 3026
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.422733313051151e-06,
+      "loss": 0.7173,
+      "step": 3027
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4167970246649385e-06,
+      "loss": 0.804,
+      "step": 3028
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4108648277576373e-06,
+      "loss": 0.7178,
+      "step": 3029
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.404936726016127e-06,
+      "loss": 0.7729,
+      "step": 3030
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3990127231247337e-06,
+      "loss": 0.783,
+      "step": 3031
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3930928227652436e-06,
+      "loss": 0.7441,
+      "step": 3032
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.387177028616896e-06,
+      "loss": 0.7056,
+      "step": 3033
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.381265344356366e-06,
+      "loss": 0.7747,
+      "step": 3034
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.375357773657786e-06,
+      "loss": 0.7534,
+      "step": 3035
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3694543201927288e-06,
+      "loss": 0.7458,
+      "step": 3036
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.363554987630202e-06,
+      "loss": 0.7373,
+      "step": 3037
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3576597796366604e-06,
+      "loss": 0.7712,
+      "step": 3038
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3517686998759946e-06,
+      "loss": 0.7761,
+      "step": 3039
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.345881752009521e-06,
+      "loss": 0.7395,
+      "step": 3040
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3399989396959988e-06,
+      "loss": 0.7539,
+      "step": 3041
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3341202665916127e-06,
+      "loss": 0.7229,
+      "step": 3042
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3282457363499687e-06,
+      "loss": 0.7324,
+      "step": 3043
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.322375352622108e-06,
+      "loss": 0.7661,
+      "step": 3044
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3165091190564934e-06,
+      "loss": 0.7744,
+      "step": 3045
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3106470392989977e-06,
+      "loss": 0.7461,
+      "step": 3046
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.304789116992924e-06,
+      "loss": 0.7854,
+      "step": 3047
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2989353557789894e-06,
+      "loss": 0.7156,
+      "step": 3048
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2930857592953167e-06,
+      "loss": 0.7427,
+      "step": 3049
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2872403311774546e-06,
+      "loss": 0.7249,
+      "step": 3050
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2813990750583434e-06,
+      "loss": 0.7976,
+      "step": 3051
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.275561994568346e-06,
+      "loss": 0.7571,
+      "step": 3052
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2697290933352253e-06,
+      "loss": 0.7979,
+      "step": 3053
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.26390037498414e-06,
+      "loss": 0.752,
+      "step": 3054
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2580758431376623e-06,
+      "loss": 0.7341,
+      "step": 3055
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2522555014157454e-06,
+      "loss": 0.801,
+      "step": 3056
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2464393534357543e-06,
+      "loss": 0.7817,
+      "step": 3057
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.240627402812442e-06,
+      "loss": 0.7642,
+      "step": 3058
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2348196531579466e-06,
+      "loss": 0.7573,
+      "step": 3059
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.229016108081804e-06,
+      "loss": 0.7812,
+      "step": 3060
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2232167711909368e-06,
+      "loss": 0.7126,
+      "step": 3061
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2174216460896413e-06,
+      "loss": 0.7559,
+      "step": 3062
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2116307363796084e-06,
+      "loss": 0.8201,
+      "step": 3063
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2058440456599038e-06,
+      "loss": 0.7546,
+      "step": 3064
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.200061577526976e-06,
+      "loss": 0.7339,
+      "step": 3065
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.194283335574637e-06,
+      "loss": 0.7065,
+      "step": 3066
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1885093233940843e-06,
+      "loss": 0.7524,
+      "step": 3067
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1827395445738863e-06,
+      "loss": 0.783,
+      "step": 3068
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1769740026999696e-06,
+      "loss": 0.7178,
+      "step": 3069
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.171212701355637e-06,
+      "loss": 0.7544,
+      "step": 3070
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1654556441215577e-06,
+      "loss": 0.7783,
+      "step": 3071
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1597028345757506e-06,
+      "loss": 0.7585,
+      "step": 3072
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1539542762936082e-06,
+      "loss": 0.7732,
+      "step": 3073
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1482099728478765e-06,
+      "loss": 0.7422,
+      "step": 3074
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.142469927808651e-06,
+      "loss": 0.6631,
+      "step": 3075
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1367341447433897e-06,
+      "loss": 0.6917,
+      "step": 3076
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.131002627216898e-06,
+      "loss": 0.7749,
+      "step": 3077
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.125275378791326e-06,
+      "loss": 0.7913,
+      "step": 3078
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.119552403026177e-06,
+      "loss": 0.7266,
+      "step": 3079
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1138337034782985e-06,
+      "loss": 0.7898,
+      "step": 3080
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1081192837018736e-06,
+      "loss": 0.7649,
+      "step": 3081
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1024091472484318e-06,
+      "loss": 0.7983,
+      "step": 3082
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0967032976668423e-06,
+      "loss": 0.7876,
+      "step": 3083
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0910017385032997e-06,
+      "loss": 0.7373,
+      "step": 3084
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0853044733013438e-06,
+      "loss": 0.7568,
+      "step": 3085
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0796115056018383e-06,
+      "loss": 0.731,
+      "step": 3086
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.073922838942983e-06,
+      "loss": 0.6885,
+      "step": 3087
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0682384768602936e-06,
+      "loss": 0.769,
+      "step": 3088
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.062558422886619e-06,
+      "loss": 0.7654,
+      "step": 3089
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0568826805521335e-06,
+      "loss": 0.7419,
+      "step": 3090
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0512112533843185e-06,
+      "loss": 0.7798,
+      "step": 3091
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.045544144907986e-06,
+      "loss": 0.76,
+      "step": 3092
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0398813586452636e-06,
+      "loss": 0.7661,
+      "step": 3093
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0342228981155797e-06,
+      "loss": 0.7515,
+      "step": 3094
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0285687668356887e-06,
+      "loss": 0.7363,
+      "step": 3095
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0229189683196513e-06,
+      "loss": 0.7322,
+      "step": 3096
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.017273506078827e-06,
+      "loss": 0.7449,
+      "step": 3097
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0116323836218886e-06,
+      "loss": 0.7554,
+      "step": 3098
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0059956044548133e-06,
+      "loss": 0.7805,
+      "step": 3099
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.000363172080869e-06,
+      "loss": 0.7493,
+      "step": 3100
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.994735090000629e-06,
+      "loss": 0.7666,
+      "step": 3101
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9891113617119684e-06,
+      "loss": 0.7446,
+      "step": 3102
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.98349199071004e-06,
+      "loss": 0.7588,
+      "step": 3103
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9778769804873042e-06,
+      "loss": 0.7327,
+      "step": 3104
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.972266334533508e-06,
+      "loss": 0.7581,
+      "step": 3105
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9666600563356773e-06,
+      "loss": 0.6848,
+      "step": 3106
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.961058149378132e-06,
+      "loss": 0.7773,
+      "step": 3107
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.955460617142476e-06,
+      "loss": 0.7791,
+      "step": 3108
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.949867463107585e-06,
+      "loss": 0.6887,
+      "step": 3109
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9442786907496245e-06,
+      "loss": 0.7554,
+      "step": 3110
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.938694303542032e-06,
+      "loss": 0.7434,
+      "step": 3111
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.933114304955517e-06,
+      "loss": 0.7334,
+      "step": 3112
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.927538698458067e-06,
+      "loss": 0.771,
+      "step": 3113
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9219674875149317e-06,
+      "loss": 0.7878,
+      "step": 3114
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9164006755886385e-06,
+      "loss": 0.7151,
+      "step": 3115
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.910838266138979e-06,
+      "loss": 0.7551,
+      "step": 3116
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9052802626229992e-06,
+      "loss": 0.7217,
+      "step": 3117
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8997266684950197e-06,
+      "loss": 0.7258,
+      "step": 3118
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.89417748720661e-06,
+      "loss": 0.7461,
+      "step": 3119
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8886327222066037e-06,
+      "loss": 0.7146,
+      "step": 3120
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8830923769410903e-06,
+      "loss": 0.7439,
+      "step": 3121
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.877556454853405e-06,
+      "loss": 0.7271,
+      "step": 3122
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.872024959384141e-06,
+      "loss": 0.748,
+      "step": 3123
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8664978939711397e-06,
+      "loss": 0.7063,
+      "step": 3124
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8609752620494846e-06,
+      "loss": 0.801,
+      "step": 3125
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.855457067051507e-06,
+      "loss": 0.7334,
+      "step": 3126
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8499433124067855e-06,
+      "loss": 0.7429,
+      "step": 3127
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.844434001542126e-06,
+      "loss": 0.7258,
+      "step": 3128
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.838929137881584e-06,
+      "loss": 0.731,
+      "step": 3129
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8334287248464488e-06,
+      "loss": 0.7778,
+      "step": 3130
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8279327658552447e-06,
+      "loss": 0.7524,
+      "step": 3131
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8224412643237187e-06,
+      "loss": 0.7805,
+      "step": 3132
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8169542236648586e-06,
+      "loss": 0.7244,
+      "step": 3133
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8114716472888758e-06,
+      "loss": 0.7388,
+      "step": 3134
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8059935386032043e-06,
+      "loss": 0.7537,
+      "step": 3135
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.800519901012504e-06,
+      "loss": 0.7727,
+      "step": 3136
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.79505073791866e-06,
+      "loss": 0.7576,
+      "step": 3137
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7895860527207663e-06,
+      "loss": 0.7554,
+      "step": 3138
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7841258488151433e-06,
+      "loss": 0.7383,
+      "step": 3139
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7786701295953245e-06,
+      "loss": 0.698,
+      "step": 3140
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.77321889845205e-06,
+      "loss": 0.7549,
+      "step": 3141
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7677721587732776e-06,
+      "loss": 0.7471,
+      "step": 3142
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.762329913944174e-06,
+      "loss": 0.7764,
+      "step": 3143
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.756892167347104e-06,
+      "loss": 0.771,
+      "step": 3144
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.751458922361644e-06,
+      "loss": 0.7522,
+      "step": 3145
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.746030182364574e-06,
+      "loss": 0.7371,
+      "step": 3146
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7406059507298665e-06,
+      "loss": 0.7769,
+      "step": 3147
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.735186230828698e-06,
+      "loss": 0.7295,
+      "step": 3148
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.729771026029443e-06,
+      "loss": 0.7405,
+      "step": 3149
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7243603396976603e-06,
+      "loss": 0.7717,
+      "step": 3150
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.718954175196109e-06,
+      "loss": 0.7683,
+      "step": 3151
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.713552535884737e-06,
+      "loss": 0.7847,
+      "step": 3152
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7081554251206786e-06,
+      "loss": 0.7756,
+      "step": 3153
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.702762846258249e-06,
+      "loss": 0.7522,
+      "step": 3154
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.697374802648953e-06,
+      "loss": 0.7383,
+      "step": 3155
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6919912976414787e-06,
+      "loss": 0.7588,
+      "step": 3156
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.686612334581682e-06,
+      "loss": 0.791,
+      "step": 3157
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6812379168126067e-06,
+      "loss": 0.7544,
+      "step": 3158
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6758680476744715e-06,
+      "loss": 0.7151,
+      "step": 3159
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6705027305046593e-06,
+      "loss": 0.7419,
+      "step": 3160
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6651419686377335e-06,
+      "loss": 0.7024,
+      "step": 3161
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6597857654054237e-06,
+      "loss": 0.7703,
+      "step": 3162
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.654434124136619e-06,
+      "loss": 0.7141,
+      "step": 3163
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.649087048157384e-06,
+      "loss": 0.7324,
+      "step": 3164
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6437445407909433e-06,
+      "loss": 0.7722,
+      "step": 3165
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6384066053576742e-06,
+      "loss": 0.7212,
+      "step": 3166
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6330732451751208e-06,
+      "loss": 0.689,
+      "step": 3167
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.627744463557984e-06,
+      "loss": 0.7754,
+      "step": 3168
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.622420263818112e-06,
+      "loss": 0.7812,
+      "step": 3169
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.617100649264511e-06,
+      "loss": 0.7676,
+      "step": 3170
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.611785623203341e-06,
+      "loss": 0.7495,
+      "step": 3171
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.606475188937898e-06,
+      "loss": 0.7971,
+      "step": 3172
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6011693497686352e-06,
+      "loss": 0.7622,
+      "step": 3173
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.595868108993149e-06,
+      "loss": 0.7615,
+      "step": 3174
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.590571469906169e-06,
+      "loss": 0.7761,
+      "step": 3175
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.585279435799578e-06,
+      "loss": 0.7878,
+      "step": 3176
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.579992009962383e-06,
+      "loss": 0.7725,
+      "step": 3177
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.574709195680738e-06,
+      "loss": 0.7231,
+      "step": 3178
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5694309962379283e-06,
+      "loss": 0.7744,
+      "step": 3179
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5641574149143643e-06,
+      "loss": 0.739,
+      "step": 3180
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.558888454987598e-06,
+      "loss": 0.7542,
+      "step": 3181
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5536241197322955e-06,
+      "loss": 0.7737,
+      "step": 3182
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5483644124202576e-06,
+      "loss": 0.7595,
+      "step": 3183
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.543109336320414e-06,
+      "loss": 0.7664,
+      "step": 3184
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5378588946987993e-06,
+      "loss": 0.7461,
+      "step": 3185
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.532613090818582e-06,
+      "loss": 0.72,
+      "step": 3186
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5273719279400465e-06,
+      "loss": 0.8123,
+      "step": 3187
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.522135409320584e-06,
+      "loss": 0.6948,
+      "step": 3188
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5169035382147088e-06,
+      "loss": 0.7341,
+      "step": 3189
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5116763178740457e-06,
+      "loss": 0.7825,
+      "step": 3190
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.506453751547322e-06,
+      "loss": 0.7058,
+      "step": 3191
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5012358424803783e-06,
+      "loss": 0.7407,
+      "step": 3192
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.496022593916163e-06,
+      "loss": 0.7009,
+      "step": 3193
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4908140090947197e-06,
+      "loss": 0.7534,
+      "step": 3194
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4856100912532e-06,
+      "loss": 0.7417,
+      "step": 3195
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4804108436258545e-06,
+      "loss": 0.7688,
+      "step": 3196
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4752162694440317e-06,
+      "loss": 0.7773,
+      "step": 3197
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4700263719361693e-06,
+      "loss": 0.7693,
+      "step": 3198
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.464841154327804e-06,
+      "loss": 0.7271,
+      "step": 3199
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4596606198415686e-06,
+      "loss": 0.7422,
+      "step": 3200
     }
   ],
   "logging_steps": 1.0,
@@ -4814,7 +19214,7 @@
   "num_input_tokens_seen": 0,
   "num_train_epochs": 1,
   "save_steps": 800,
-  "total_flos": 1881161810051072.0,
+  "total_flos": 7556461979500544.0,
   "train_batch_size": 24,
   "trial_name": null,
   "trial_params": null